threads
listlengths 1
2.99k
|
|---|
[
{
"msg_contents": "I suppose there isn't any good reason that elog.c and exc.c refer to\nsys_nerr and sys_errlist directly rather than using strerror() (probably\nan artifact from before port/strerror.c was added?) and given that it is\nprovably unportable to do so (Cygwin) I'm inclined to make the obvious\nfix.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Thu, 28 Sep 2000 15:48:30 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "sys_nerr, sys_errlist"
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> I suppose there isn't any good reason that elog.c and exc.c refer to\n> sys_nerr and sys_errlist directly rather than using strerror()\n\nI don't see any direct references to sys_errlist. The references to\nsys_nerr are necessary to avoid coredump with an out-of-range errno\nvalue --- at least on machines where strerror() is a macro that expands\nto an unchecked array access.\n\n> an artifact from before port/strerror.c was added?) and given that it is\n> provably unportable to do so (Cygwin) I'm inclined to make the obvious\n> fix.\n\nI think the obvious fix is to supply a sys_nerr value on Cygwin.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 28 Sep 2000 11:12:25 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: sys_nerr, sys_errlist "
}
] |
[
{
"msg_contents": "anybody knows to use pqaccess ?\n\nI would like ->\n\ntable : ___________\nf1 : ___________\n\nexecute the query : SELECT f1, f2, f3,f4 FROM 'name_table' WHERE f1 LIKE\n'name_f1' ORDER BY f1 ;\n\n|--------------|----------|---------|-------|\n| f1 | f2 | f3 | f4 |\n|--------------|----------|---------|-------|\n| | | | |\n|--------------|----------|---------|-------|\n| | | | |\n|--------------|----------|---------|-------|\n| | | | |\n|--------------|----------|---------|-------|\n\nI think that I must use Form in pgaccess but...\n\nThanks in advance.\n\nJerome.\n",
"msg_date": "Thu, 28 Sep 2000 18:30:24 +0200",
"msg_from": "Jerome Raupach <[email protected]>",
"msg_from_op": true,
"msg_subject": "pgaccess"
}
] |
[
{
"msg_contents": "I just noticed the following peculiar behavior.\n\nBy default, a table's owner has all access rights for it:\n\n\tplay=> create table foo(f1 int);\n\tCREATE\n\tplay=> insert into foo values(1);\n\tINSERT 328713 1\n\nBut let him grant some rights to someone else:\n\n\tplay=> grant select on foo to public;\n\tCHANGE\n\nand all of a sudden he's not got all rights anymore.\n\n\tplay=> insert into foo values(1);\n\tERROR: foo: Permission denied.\n\nIt seems this is because the ACL code has two different ideas of\n\"default access rights\". If the table's pg_class row has a NULL acl\nfield, as it does initially, then the code uses a default acl\nsetting that is\n\tWORLD: no access\n\tOWNER: all access\nBut, when you issue a GRANT and the ChangeAcl routine goes to modify\nthe table's ACL, if the old value is NULL then it substitutes a\n*different* default as the starting point for the ACL change:\n\tWORLD: no access\nso what you end up with after a grant such as the above is\n\tWORLD: read\nand the poor owner has no rights except what he gave to WORLD.\n\nFortunately he's still the owner and can do \"GRANT ALL TO himself\".\nBut IMHO it's absolutely brain-dead that this command has to be\nissued explicitly as soon as one's done any other granting or\nrevoking.\n\nI think there should be just one default ACL and it should be\n\tWORLD: no access\n\tOWNER: all access\nIf you do any explicit granting then that should be the starting\npoint.\n\nComments?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 28 Sep 2000 21:31:40 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": true,
"msg_subject": "Bizarre behavior of default access permissions"
},
{
"msg_contents": "Tom Lane writes:\n\n> I think there should be just one default ACL and it should be\n> \tWORLD: no access\n> \tOWNER: all access\n> If you do any explicit granting then that should be the starting\n> point.\n\nDefinitely so.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Fri, 29 Sep 2000 14:37:10 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Bizarre behavior of default access permissions"
}
] |
[
{
"msg_contents": "Hi,\n\n we've suddenly started getting this error message out of postgres\n(7.0.2). Does anybody know where it comes from?\n\nERROR: UNLockBuffer: buffer 0 is not locked\n\nAny help appreciated,\n\nAdriaan\n\n",
"msg_date": "Fri, 29 Sep 2000 17:46:44 +0300",
"msg_from": "Adriaan Joubert <[email protected]>",
"msg_from_op": true,
"msg_subject": "Strange error message"
},
{
"msg_contents": "Adriaan Joubert <[email protected]> writes:\n> we've suddenly started getting this error message out of postgres\n> (7.0.2). Does anybody know where it comes from?\n\n> ERROR: UNLockBuffer: buffer 0 is not locked\n\nEvidently something is passing an invalid buffer number to LockBuffer\nin src/backend/storage/buffer/bufmgr.c. (0 is InvalidBuffer, but\nLockBuffer won't notice that unless you compiled with asserts enabled.)\nWhatever the bug is, it's not directly LockBuffer's fault.\n\nWhat exactly are you doing that provokes this message?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 29 Sep 2000 11:45:47 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Strange error message "
},
{
"msg_contents": "Tom Lane wrote:\n\n> > For a while I though it might be because we are using an alpha TAS in\n> > the spinlock rather than the old semaphore. I replaced our spinlock\n> > with the standard one and it made no difference. We have been running\n> > with our spinlock implementation for nearly 2 months on a production\n> > database now without a hitch, so I think it is ok. Did I ever submit\n> > any patches for the Alpha spinlock?\n>\n> Not that I recall. We did get some advice from some Alpha gurus at DEC\n> who seemed to think the existing TAS code is OK. What was it that you\n> felt needed to be improved?\n\nThe current code uses semaphores, which has the advantage that it works\nwell even on multi-processor machines, but the disadvantage that it is not\nthe fastest way possible. Writing a spinlock on Alpha for SMP machines is\nvery difficult, as you need to deal with memory barriers. A real mess. But\nthen one of the people at Compaq pointed out to us that there is a\nready-made routine on Alpha. We implemented it with the two patches below.\nI ran tests with lots of parallel back-ends and got around a 10% speed\nincrease. I include the two patches. Perhaps some of the other people\nrunning Tru64 can have a look at these as well.\n\nCheers,\n\nAdriaan",
"msg_date": "Mon, 02 Oct 2000 08:21:45 +0300",
"msg_from": "Adriaan Joubert <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Alpha spinlock"
},
{
"msg_contents": "Tom Lane wrote:\n\n> Adriaan Joubert <[email protected]> writes:\n> > we've suddenly started getting this error message out of postgres\n> > (7.0.2). Does anybody know where it comes from?\n>\n> > ERROR: UNLockBuffer: buffer 0 is not locked\n>\n> Evidently something is passing an invalid buffer number to LockBuffer\n> in src/backend/storage/buffer/bufmgr.c. (0 is InvalidBuffer, but\n> LockBuffer won't notice that unless you compiled with asserts enabled.)\n> Whatever the bug is, it's not directly LockBuffer's fault.\n\nRight, I'vebuilt a new database and everything seemed fine for a while and\nnow I've got this message back. It is due to the index on one of our\ntables getting messed up - at least, if we drop and recreate the index\neverything is fine. What should I do to track down what is happening?\nCompile with asserts, or run with specific logging? Any advice\nappreciated!\n\nAdriaan\n\n",
"msg_date": "Mon, 02 Oct 2000 08:42:33 +0300",
"msg_from": "Adriaan Joubert <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Strange error message"
},
{
"msg_contents": "Adriaan Joubert <[email protected]> writes:\n>>>> ERROR: UNLockBuffer: buffer 0 is not locked\n\n> Right, I'vebuilt a new database and everything seemed fine for a while and\n> now I've got this message back. It is due to the index on one of our\n> tables getting messed up - at least, if we drop and recreate the index\n> everything is fine. What should I do to track down what is happening?\n> Compile with asserts, or run with specific logging?\n\nCompile with asserts and -g, and get a backtrace from the ensuing\ncoredump. (LockBuffer *will* Assert when passed a zero. If we are\nreally lucky, we might see an earlier Assert failure that will give\nmore clue about where the zero comes from --- but if not, the backtrace\nfrom the bogus LockBuffer call might still be useful.)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 02 Oct 2000 02:22:49 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Strange error message "
},
{
"msg_contents": "Adriaan Joubert <[email protected]> writes:\n> Tom Lane wrote:\n>> Not that I recall. We did get some advice from some Alpha gurus at DEC\n>> who seemed to think the existing TAS code is OK. What was it that you\n>> felt needed to be improved?\n\n> The current code uses semaphores,\n\nOh, I'm sorry, I was thinking about the Linux-Alpha code, for which\nthere has been an inline-assembly version of tas() for a long time;\nand it was that code that we asked the DEC people about. I had\nforgotten that the OSF port uses a different semaphore method.\n\nIs there a reason we couldn't use the Linux-Alpha code on OSF too?\nI'd just as soon minimize the difference between the two ports ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 02 Oct 2000 02:26:40 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Alpha spinlock "
},
{
"msg_contents": "Tom, what do you want to do with this patch?\n\n> \n> \n> Tom Lane wrote:\n> \n> > > For a while I though it might be because we are using an alpha TAS in\n> > > the spinlock rather than the old semaphore. I replaced our spinlock\n> > > with the standard one and it made no difference. We have been running\n> > > with our spinlock implementation for nearly 2 months on a production\n> > > database now without a hitch, so I think it is ok. Did I ever submit\n> > > any patches for the Alpha spinlock?\n> >\n> > Not that I recall. We did get some advice from some Alpha gurus at DEC\n> > who seemed to think the existing TAS code is OK. What was it that you\n> > felt needed to be improved?\n> \n> The current code uses semaphores, which has the advantage that it works\n> well even on multi-processor machines, but the disadvantage that it is not\n> the fastest way possible. Writing a spinlock on Alpha for SMP machines is\n> very difficult, as you need to deal with memory barriers. A real mess. But\n> then one of the people at Compaq pointed out to us that there is a\n> ready-made routine on Alpha. We implemented it with the two patches below.\n> I ran tests with lots of parallel back-ends and got around a 10% speed\n> increase. I include the two patches. Perhaps some of the other people\n> running Tru64 can have a look at these as well.\n> \n> Cheers,\n> \n> Adriaan\n\n> *** src/include/storage/s_lock.h\tSat Sep 30 12:13:42 2000\n> --- src/include/storage/s_lock.h-\tSat Sep 30 12:13:42 2000\n> ***************\n> *** 252,266 ****\n> * Note that slock_t on the Alpha AXP is msemaphore instead of char\n> * (see storage/ipc.h).\n> */\n> - #include <alpha/builtins.h>\n> - #if 0\n> #define TAS(lock)\t (msem_lock((lock), MSEM_IF_NOWAIT) < 0)\n> #define S_UNLOCK(lock) msem_unlock((lock), 0)\n> #define S_INIT_LOCK(lock)\t\tmsem_init((lock), MSEM_UNLOCKED)\n> #define S_LOCK_FREE(lock)\t (!(lock)->msem_state)\n> - #else\n> - #define TAS(lock) (__INTERLOCKED_TESTBITSS_QUAD((lock),0))\n> - #endif\n> \n> #else /* i.e. not __osf__ */\n> \n> --- 252,261 ----\n\n> *** src/include/port/alpha.h\tSat Sep 30 12:13:21 2000\n> --- src/include/port/alpha.h-\tSat Sep 30 12:13:21 2000\n> ***************\n> *** 1,10 ****\n> #define USE_POSIX_TIME\n> #define DISABLE_XOPEN_NLS\n> #define HAS_TEST_AND_SET\n> ! /*#include <sys/mman.h>*/\t\t\t/* for msemaphore */\n> ! /*typedef msemaphore slock_t;*/\n> ! #include <alpha/builtins.h>\n> ! typedef volatile long slock_t;\n> \n> /* some platforms define __alpha, but not __alpha__ */\n> #if defined(__alpha) && !defined(__alpha__)\n> --- 1,8 ----\n> #define USE_POSIX_TIME\n> #define DISABLE_XOPEN_NLS\n> #define HAS_TEST_AND_SET\n> ! #include <sys/mman.h>\t\t\t/* for msemaphore */\n> ! typedef msemaphore slock_t;\n> \n> /* some platforms define __alpha, but not __alpha__ */\n> #if defined(__alpha) && !defined(__alpha__)\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 00:25:29 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: Alpha spinlock"
},
{
"msg_contents": "Bruce Momjian <[email protected]> writes:\n> Tom, what do you want to do with this patch?\n\nApply away ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 08 Oct 2000 00:34:30 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: Alpha spinlock "
},
{
"msg_contents": "Applied, but include/port/alpha.h is now include/port/osf.h.\n\n> \n> \n> Tom Lane wrote:\n> \n> > > For a while I though it might be because we are using an alpha TAS in\n> > > the spinlock rather than the old semaphore. I replaced our spinlock\n> > > with the standard one and it made no difference. We have been running\n> > > with our spinlock implementation for nearly 2 months on a production\n> > > database now without a hitch, so I think it is ok. Did I ever submit\n> > > any patches for the Alpha spinlock?\n> >\n> > Not that I recall. We did get some advice from some Alpha gurus at DEC\n> > who seemed to think the existing TAS code is OK. What was it that you\n> > felt needed to be improved?\n> \n> The current code uses semaphores, which has the advantage that it works\n> well even on multi-processor machines, but the disadvantage that it is not\n> the fastest way possible. Writing a spinlock on Alpha for SMP machines is\n> very difficult, as you need to deal with memory barriers. A real mess. But\n> then one of the people at Compaq pointed out to us that there is a\n> ready-made routine on Alpha. We implemented it with the two patches below.\n> I ran tests with lots of parallel back-ends and got around a 10% speed\n> increase. I include the two patches. Perhaps some of the other people\n> running Tru64 can have a look at these as well.\n> \n> Cheers,\n> \n> Adriaan\n\n> *** src/include/storage/s_lock.h\tSat Sep 30 12:13:42 2000\n> --- src/include/storage/s_lock.h-\tSat Sep 30 12:13:42 2000\n> ***************\n> *** 252,266 ****\n> * Note that slock_t on the Alpha AXP is msemaphore instead of char\n> * (see storage/ipc.h).\n> */\n> - #include <alpha/builtins.h>\n> - #if 0\n> #define TAS(lock)\t (msem_lock((lock), MSEM_IF_NOWAIT) < 0)\n> #define S_UNLOCK(lock) msem_unlock((lock), 0)\n> #define S_INIT_LOCK(lock)\t\tmsem_init((lock), MSEM_UNLOCKED)\n> #define S_LOCK_FREE(lock)\t (!(lock)->msem_state)\n> - #else\n> - #define TAS(lock) (__INTERLOCKED_TESTBITSS_QUAD((lock),0))\n> - #endif\n> \n> #else /* i.e. not __osf__ */\n> \n> --- 252,261 ----\n\n> *** src/include/port/alpha.h\tSat Sep 30 12:13:21 2000\n> --- src/include/port/alpha.h-\tSat Sep 30 12:13:21 2000\n> ***************\n> *** 1,10 ****\n> #define USE_POSIX_TIME\n> #define DISABLE_XOPEN_NLS\n> #define HAS_TEST_AND_SET\n> ! /*#include <sys/mman.h>*/\t\t\t/* for msemaphore */\n> ! /*typedef msemaphore slock_t;*/\n> ! #include <alpha/builtins.h>\n> ! typedef volatile long slock_t;\n> \n> /* some platforms define __alpha, but not __alpha__ */\n> #if defined(__alpha) && !defined(__alpha__)\n> --- 1,8 ----\n> #define USE_POSIX_TIME\n> #define DISABLE_XOPEN_NLS\n> #define HAS_TEST_AND_SET\n> ! #include <sys/mman.h>\t\t\t/* for msemaphore */\n> ! typedef msemaphore slock_t;\n> \n> /* some platforms define __alpha, but not __alpha__ */\n> #if defined(__alpha) && !defined(__alpha__)\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 00:38:34 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: Alpha spinlock"
}
] |
[
{
"msg_contents": "Hi,\n\nwe tried to switch the log on by lauching the postmaster\nwith -d 5 redirected to a file but this file is deseperatly empty,\nwhat it's wrong ? is the process writing to an other file or\ndo we made a mistake with the arguments ?\n\nthanks in advance\n\nregards,\n\njc",
"msg_date": "Fri, 29 Sep 2000 19:44:08 +0300",
"msg_from": "Jean-Christophe Pazzaglia <[email protected]>",
"msg_from_op": true,
"msg_subject": "Database log"
},
{
"msg_contents": "Jean-Christophe Pazzaglia writes:\n\n> we tried to switch the log on by lauching the postmaster\n> with -d 5 redirected to a file but this file is deseperatly empty,\n> what it's wrong ? is the process writing to an other file or\n> do we made a mistake with the arguments ?\n\nDon't use the -S option.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Fri, 29 Sep 2000 19:04:36 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Database log"
},
{
"msg_contents": "Jean-Christophe Pazzaglia <[email protected]> writes:\n> we tried to switch the log on by lauching the postmaster\n> with -d 5 redirected to a file but this file is deseperatly empty,\n\nIf you're using -S, stop doing so ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 29 Sep 2000 14:45:09 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Database log "
},
{
"msg_contents": "\ntry using the pg_options file itself, which, if you compile --with-syslog\nsupport, will allow you to write to syslog. the main benefit being, you\ncan rotate your log files, which you can't if you do a redirect to a file\n\n\n\nOn Fri, 29 Sep 2000, Jean-Christophe Pazzaglia wrote:\n\n> \n> \n> Hi,\n> \n> we tried to switch the log on by lauching the postmaster\n> with -d 5 redirected to a file but this file is deseperatly empty,\n> what it's wrong ? is the process writing to an other file or\n> do we made a mistake with the arguments ?\n> \n> thanks in advance\n> \n> regards,\n> \n> jc\n\nMarc G. Fournier ICQ#7615664 IRC Nick: Scrappy\nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sat, 7 Oct 2000 21:04:39 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Database log"
}
] |
[
{
"msg_contents": "\n> What I'm thinking about doing is eliminating the \"skipAcl\" RTE field\n> and instead adding an Oid field named something like \"checkAclAs\".\n> The semantics of this field would be \"if zero, check access \n> permissions\n> for this table using the current effective userID; but if not zero,\n> check access permissions as if you are this userID\". Then the rule\n> rewriter would do no access permission checks of its own, but would\n> set this field appropriately in RTEs that it adds to queries. All the\n> actual permissions checking would happen in one place in the executor.\n> \n> Comments? Is this a general enough mechanism, and does it fit well\n> with the various setUID tricks that people are thinking about?\n\nSounds good, and a step in the right direction.\n\nAndreas\n",
"msg_date": "Fri, 29 Sep 2000 19:19:42 +0200",
"msg_from": "Zeugswetter Andreas SB <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: Reimplementing permission checks for rules"
},
{
"msg_contents": "Zeugswetter Andreas SB <[email protected]> writes:\n>> What I'm thinking about doing is eliminating the \"skipAcl\" RTE field\n>> and instead adding an Oid field named something like \"checkAclAs\".\n>> \n>> Comments? Is this a general enough mechanism, and does it fit well\n>> with the various setUID tricks that people are thinking about?\n\n> Sounds good, and a step in the right direction.\n\nAfter looking at the rule rewriter some more, I realized that the only\nway to push all permissions checks to execution time is not only to keep\nskipAcl, but to generalize it. The problem is with checks on the view\nitself --- if you do INSERT INTO someView, which gets rewritten into\nan insert to someRealTable, then what you want the executor to check for\nis\n\tWrite access on someView by current user\n\tWrite access on someRealTable by owner of rule\nwhich is infeasible with the existing code because the executor only\nchecks for write access on the real target table (someRealTable).\n\nWhat I have now got, and hope to commit today, is the following fields\nin RangeTblEntry, replacing skipAcl:\n\n * checkForRead, checkForWrite, and checkAsUser control run-time access\n * permissions checks. A rel will be checked for read or write access\n * (or both, or neither) per checkForRead and checkForWrite. If\n * checkAsUser is not InvalidOid, then do the permissions checks using\n * the access rights of that user, not the current effective user ID.\n * (This allows rules to act as setuid gateways.)\n\n bool checkForRead; /* check rel for read access */\n bool checkForWrite; /* check rel for write access */\n Oid checkAsUser; /* if not zero, check access as this user */\n\nThe parser always sets checkAsUser = 0; it sets checkForRead if the rel\nis referenced explicitly anywhere in the query; and it sets\ncheckForWrite on the target table of INSERT/UPDATE/DELETE. It was a\nlittle tricky to get it to do the right thing when the same table is\nboth a source and target, eg in INSERT ... SELECT, but not too bad.\n\nWhen a rule is created, the stored parsetree has the rule creator's OID\nassigned into each RTE's checkAsUser field, except for the dummy *NEW*\nand *OLD* rtable entries.\n\nThe rewriter leaves the check fields on a view's RTE as they are, and\njust copies the check fields from the rule for the rtable entries it\nadds. No rewrite-time permission checks anywhere.\n\nThe executor just carries out the indicated checks. execMain's checking\ngot a *lot* simpler, too, because it doesn't have to carry around a lot\nof baggage to determine whether to check a given RTE for read, write,\nor both.\n\nSeems like a big win ...\n\nNOTE: there is a subtle change here. A rule used to be taken as\nexecuting setUID to the owner of the table the rule is attached to.\nNow it is executed as if setUID to the person who created the rule,\nwho could be a different user if the table owner gave away RULE rights.\nI think this is a more correct behavior, but I'm open to argument.\nIt'd be easy to make CREATE RULE store the table owner's OID instead,\nif anyone wants to argue that that was the right thing.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 29 Sep 2000 13:55:06 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: Reimplementing permission checks for rules "
}
] |
[
{
"msg_contents": "I've committed the announced unified regression test driver. (See Sep.20\nmessage for list of features.) You can invoke it via `make check', either\nin the top level directory, or directly in src/test/regress. To test\nagainst a running installation, use `make installcheck'.\n\nThe existing drivers still work unchanged, but should probably be\nsuperceded by this eventually.\n\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Fri, 29 Sep 2000 19:24:44 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "New unified regression test driver"
},
{
"msg_contents": "Peter Eisentraut wrote:\n> I've committed the announced unified regression test driver. (See Sep.20\n> message for list of features.) You can invoke it via `make check', either\n> in the top level directory, or directly in src/test/regress. To test\n> against a running installation, use `make installcheck'.\n\nFirst, Peter, thanks a million for all the work you've done on\nPostgreSQL.\n\nThat said, I have a request -- since the binary distributions might\npossibly be installed on machines with no make and no compiler, I would\nlike to see the regression tests be reduced to a simple script and\ndriver that could be simply invoked with no need for make. Binary-only\nusers might want to regress a little, too.\n\nUp until RH7, the postgresql-test RPM as shipped had a halfway built\nregression test suite, and the finishing touch was supplied by the user\nat run time who invoked regress.sh directly with the appropriate\narguments. However, this didn't allow the user to run the parallel\nregression suite.\n\nPeople are running PostgreSQL every day on machines with no compiler and\nno source code -- they need a regression suite, too.\n\nOf course, I can always work around the issue, as I have in the past....\n \n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Fri, 29 Sep 2000 13:37:51 -0400",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: New unified regression test driver"
},
{
"msg_contents": "Lamar Owen writes:\n\n> That said, I have a request -- since the binary distributions might\n> possibly be installed on machines with no make and no compiler, I would\n> like to see the regression tests be reduced to a simple script and\n> driver that could be simply invoked with no need for make. Binary-only\n> users might want to regress a little, too.\n\nThis was originally one of my implementation goals, but it's a bit tricky. \nSome of the SQL input files contains absolute paths as part of CREATE\nFUNCTION or COPY commands. We'd have to patch up these paths as part of\nthe installation run. That's terribly inelegant and it would probably\nbreak something else I'm thinking about. (Automatically generating\ninstall manifests by substituting something clever for the $(INSTALL) make\nvariable. You might be interested in this too. Stay tuned.)\n\nThat said, the new script should make things a bit easier for you because\nit is reasonably standalone, allows parallel and serial regression tests,\nand has command line options to find the input and output files. The ugly\ndetails about how to install the right files I would like to leave up to\nthe packager right now.\n\nOne thing I thought of was that perhaps the regression test driver could\nsubstitute the correct paths at run time rather than at build time. But\nthat would break backward compatibility with the old drivers, and as we\nhaven't even settled on the new one yet, it's probably an inopportune time\nto try it now.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Thu, 5 Oct 2000 19:42:39 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: New unified regression test driver"
},
{
"msg_contents": "This may sound stupid, but what's wrong with including a binary make\nexecutable in a binary PostgreSQL package? It could be stored in the\nregression test directory and then you could write a \"standard\" regression\ntest makefile instead of a different \"script\" for each platform. A compiler\nwould be out of the question due to its size, but a binary make shouldn't be\ntoo large.\n\nFred Reimer\nEclipsys Corporation\n\n> -----Original Message-----\n> From: [email protected]\n> [mailto:[email protected]]On\n> Behalf Of Peter Eisentraut\n> Sent: Thursday, October 05, 2000 1:43 PM\n> To: Lamar Owen\n> Cc: PostgreSQL Development\n> Subject: Re: [HACKERS] New unified regression test driver\n>\n>\n> Lamar Owen writes:\n>\n> > That said, I have a request -- since the binary distributions might\n> > possibly be installed on machines with no make and no\n> compiler, I would\n> > like to see the regression tests be reduced to a simple script and\n> > driver that could be simply invoked with no need for make.\n> Binary-only\n> > users might want to regress a little, too.\n>\n> This was originally one of my implementation goals, but it's\n> a bit tricky.\n> Some of the SQL input files contains absolute paths as part of CREATE\n> FUNCTION or COPY commands. We'd have to patch up these paths\n> as part of\n> the installation run. That's terribly inelegant and it would probably\n> break something else I'm thinking about. (Automatically generating\n> install manifests by substituting something clever for the\n> $(INSTALL) make\n> variable. You might be interested in this too. Stay tuned.)\n>\n> That said, the new script should make things a bit easier for\n> you because\n> it is reasonably standalone, allows parallel and serial\n> regression tests,\n> and has command line options to find the input and output\n> files. The ugly\n> details about how to install the right files I would like to\n> leave up to\n> the packager right now.\n>\n> One thing I thought of was that perhaps the regression test\n> driver could\n> substitute the correct paths at run time rather than at build\n> time. But\n> that would break backward compatibility with the old drivers,\n> and as we\n> haven't even settled on the new one yet, it's probably an\n> inopportune time\n> to try it now.\n>\n> --\n> Peter Eisentraut [email protected] http://yi.org/peter-e/\n>\n>\n\n",
"msg_date": "Thu, 5 Oct 2000 15:24:53 -0400",
"msg_from": "\"Frederick W. Reimer\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: New unified regression test driver"
},
{
"msg_contents": "Frederick W. Reimer writes:\n\n> This may sound stupid, but what's wrong with including a binary make\n> executable in a binary PostgreSQL package?\n\nWe don't need \"make\" for the regression test.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Thu, 5 Oct 2000 22:31:43 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: New unified regression test driver"
}
] |
[
{
"msg_contents": "\nI have a very serious need for replication for my Postgres\napplication. Rather than deciding to spend $40k on a commercial DBMS\nwith replication (minimum cost) I decided I would invest my money in\nhiring a contractor to add support for the feature I need into\nPostgres. The patch that gets written will be released under the same\nlicense as Postgres, so hopefully it will be added to a future\ndistribution.\n\nAnyway, I know that I'm not the first to want replication in Postgres,\nand I imagine someone else has probably done some work on it, or is\ncurrently working on it. If that's you, please contact me, because\nit's possible that we can avoid some duplication (replication?) of\neffort. Also, I want to make sure that this work is done in a way\nwhich would allow it to be integrated into a future release if at all\npossible.\n\nThanks,\n\ne\n\n-- \nThis message was my two cents worth. Please deposit two cents into my\ne-gold account by following this link:\nhttp://rootworks.com/twocentsworth.cgi?102861\n275A B627 1826 D627 ED35 B8DF 7DDE 4428 0F5C 4454\n",
"msg_date": "Fri, 29 Sep 2000 12:13:56 -0700 (PDT)",
"msg_from": "Erich <[email protected]>",
"msg_from_op": true,
"msg_subject": "IF YOU'RE WORKING ON REPLICATION PLEASE CONTACT ME"
},
{
"msg_contents": "* Erich <[email protected]> [000929 12:13] wrote:\n> \n> I have a very serious need for replication for my Postgres\n> application. Rather than deciding to spend $40k on a commercial DBMS\n> with replication (minimum cost) I decided I would invest my money in\n> hiring a contractor to add support for the feature I need into\n> Postgres. The patch that gets written will be released under the same\n> license as Postgres, so hopefully it will be added to a future\n> distribution.\n> \n> Anyway, I know that I'm not the first to want replication in Postgres,\n> and I imagine someone else has probably done some work on it, or is\n> currently working on it. If that's you, please contact me, because\n> it's possible that we can avoid some duplication (replication?) of\n> effort. Also, I want to make sure that this work is done in a way\n> which would allow it to be integrated into a future release if at all\n> possible.\n\nCheck this out, I'm sure your contribution can help realize the\nreplication server.\n\nhttp://www.pgsql.com/press/PR_5.html\n\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Fri, 29 Sep 2000 13:27:09 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: IF YOU'RE WORKING ON REPLICATION PLEASE CONTACT ME"
}
] |
[
{
"msg_contents": "Scrappy's late victory ... The documentation needs to know about the\ncorrect version number as well, but that doesn't work currently, without\ncreating a dependency on some file generated by config.status. So I'm\ninclined to make it a separate file that only contains the number.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Fri, 29 Sep 2000 22:46:46 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "We could use a separate VERSION file after all"
},
{
"msg_contents": "\nbetter late then never, eh? :)\n\n\nOn Fri, 29 Sep 2000, Peter Eisentraut wrote:\n\n> Scrappy's late victory ... The documentation needs to know about the\n> correct version number as well, but that doesn't work currently, without\n> creating a dependency on some file generated by config.status. So I'm\n> inclined to make it a separate file that only contains the number.\n> \n> -- \n> Peter Eisentraut [email protected] http://yi.org/peter-e/\n> \n\nMarc G. Fournier ICQ#7615664 IRC Nick: Scrappy\nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sat, 7 Oct 2000 21:05:58 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: We could use a separate VERSION file after all"
}
] |
[
{
"msg_contents": "The \"rule\" regression test gets stuck in an infinite loop in\noptimizer/prep/prepunion.c:find_inheritable_rt_entry. It seems to be a\ncoding mistake; from what I can tell the loop can't ever exit if there are\nno inherited tables.\n\nI also get failures in sanity_check and misc, which might be due to the\nchanged views implementation.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Sat, 30 Sep 2000 02:19:26 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Unruly rules"
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> The \"rule\" regression test gets stuck in an infinite loop in\n> optimizer/prep/prepunion.c:find_inheritable_rt_entry. It seems to be a\n> coding mistake; from what I can tell the loop can't ever exit if there are\n> no inherited tables.\n> I also get failures in sanity_check and misc, which might be due to the\n> changed views implementation.\n\n?? it all passes here ... platform dependency maybe? Or perhaps you\njust need a make clean and rebuild? I did a fair amount of struct-\ndeclaration-rearrangement...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 29 Sep 2000 23:21:50 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Unruly rules "
},
{
"msg_contents": "Tom Lane writes:\n\n> > The \"rule\" regression test gets stuck in an infinite loop in\n> > optimizer/prep/prepunion.c:find_inheritable_rt_entry.\n\n> ?? it all passes here ... platform dependency maybe? Or perhaps you\n> just need a make clean and rebuild? I did a fair amount of struct-\n> declaration-rearrangement...\n\nActually, it's still there (even after OS upgrade :). The \"misc\" and\n\"sanity_check\" failures were due to some test ordering problems, but the\n\"rules\" still hang.\n\nThe backend process hangs in this loop:\n\n| bool find_inheritable_rt_entry(List *rangetable,\n| Index *rt_index,\n| List **inheritors) {\n| Index count = 0;\n| List *temp;\n| \n| foreach(temp, rangetable)\n| {\n| RangeTblEntry *rt_entry = (RangeTblEntry *) lfirst(temp);\n| List *inhs;\n| \n| count++;\n| /* Ignore non-inheritable RT entries */\n| if (! rt_entry->inh)\n| continue;\n\nSomewhere a list pointer needs to be advanced there.\n\nThis happens at the query:\n\nDEBUG: query: insert into shoelace_ok select * from shoelace_arrive;\n\n",
"msg_date": "Sun, 1 Oct 2000 20:35:43 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Unruly rules "
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> The backend process hangs in this loop:\n\n> | foreach(temp, rangetable)\n\n> Somewhere a list pointer needs to be advanced there.\n\nWhat? foreach() does advance the pointer. The only way this can be\nan infinite loop is if the rangetable list is circular.\n\nI wouldn't rank that as out of the question, considering the enthusiasm\nwith which the rules code munges range lists, but I still don't see it\nhappening here. Is anyone else seeing regression failures with current\nsources?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 01 Oct 2000 19:43:51 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Unruly rules "
},
{
"msg_contents": "I don't know if you've noticed, but the Beos porters are also reporting\nthat the rule test hangs. Is there any way I can help debugging this?\n\nIt happens sporadically on my machine, which is potentially because I turn\nit off every night, thus the memory might be more uniformly initialized\nthan on long-running machines. (Happened before...)\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Sat, 7 Oct 2000 20:55:21 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Unruly rules "
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> I don't know if you've noticed, but the Beos porters are also reporting\n> that the rule test hangs. Is there any way I can help debugging this?\n\nYeah? Hmm, sounds more and more like a platform-specific bug then.\nI don't see it on HPUX, Linux/Intel or Linux/PPC. What are you running\non, and with what configure/build options?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 07 Oct 2000 22:56:04 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Unruly rules "
}
] |
[
{
"msg_contents": "This seems to be a bug:\n\n Create Table tab1 (f1 integer);\n Create Function tot_of_tab1() returns integer \n as 'select cast(sum(f1) as int4) from tab1' language 'sql';\n Alter Table tab1 add check(tot_of_tab1() > 0);\n \n zzz=# insert into tab1 values(1);\n INSERT 142380 1\n zzz=# insert into tab1 values(-10);\n INSERT 142381 1\n zzz=# select tot_of_tab1();\n tot_of_tab1\n -------------\n -9\n (1 row)\n \n zzz=# insert into tab1 values(-12);\n ERROR: ExecAppend: rejected due to CHECK constraint $1\n \nThe constraint should have failed on the second insert. Maybe the\nconstraint is evaluate before the insert? \n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Sat, 30 Sep 2000 11:24:42 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": true,
"msg_subject": "Bug in constraings (from GENERAL)"
}
] |
[
{
"msg_contents": "I have been away from the mailing lists for perhaps six weeks. I am\ndone my backlog of work from the book, and will start going through the\n5k messages I have accumulated. Please don't be upset with me for\nreplying to messages over one month old.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 29 Sep 2000 22:18:01 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "I am back"
}
] |
[
{
"msg_contents": "The core group has decided to delay 7.1 beta until November 1. We have\ndone this to enable the write-ahead log code (WAL) to be shipped with\n7.1.\n\nOf course, it also gives me time to catch up on my e-mail, which I am\ndoing now. :-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 29 Sep 2000 22:28:53 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "7.1 beta schedule"
},
{
"msg_contents": "On Fri, 29 Sep 2000, Bruce Momjian wrote:\n\n> The core group has decided to delay 7.1 beta until November 1. We have\n> done this to enable the write-ahead log code (WAL) to be shipped with\n> 7.1.\n> \n> Of course, it also gives me time to catch up on my e-mail, which I am\n> doing now. :-)\n\ntell me about it ... 7 days away and totally swamped again :)\n\n\n",
"msg_date": "Sat, 7 Oct 2000 21:06:30 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: 7.1 beta schedule"
}
] |
[
{
"msg_contents": "This brings up some good issues for the 7.2 release. Will large objects\nbecome just an API on top of toast, or should they remain as a separate\nphysical storage format?\n\n\n> At 08:30 PM 3/15/00 -0500, Bruce Momjian wrote:\n> \n> >Yes, we should keep it. I see now it is for purely binary data, while\n> >text is for null-terminated strings.\n> \n> donb=# create table foo (b bytea);\n> CREATE\n> donb=# insert into foo values('ab\\0cd');\n> INSERT 107497 1\n> donb=# select * from foo;\n> b \n> ----\n> ab\n> (1 row)\n> \n> donb=# \n> \n> Thus my comment \"maybe they should be made to work\" :)\n> \n> I don't know what's actually inside attr b, but the \"cd\" is at least\n> dropped on output.\n> \n> For the BLOB hack I did for our toolkit I did the equivalent of\n> uuencoding the input, which costs a predictable 4/3 expansion of\n> the binary data (this is a segmented type, all done outside PG\n> via SQL, triggers, and AOLserver driver magic but lets us stuff\n> binary data such as photos etc, and pg_dump/restore them).\n> \n> If TOAST weren't on the way, I'd sit down and do a proper BLOB,\n> as I explained to the folks on our web toolkit team lo is \n> tantilizingly close to being useful for folks like us, without\n> actually being useful.\n> \n> BLOBs should sit atop TOAST, though, and perhaps specialized I/O\n> routines for a BLOB type could be made. Those for bytea could\n> be changed, too, at risk of breaking existing code? But since\n> bytea really acts like text perhaps there is no real existing code\n> that exists that couldn't just operate on text instead, so there\n> could be freedom to change it?\n> \n> For real binary data, uuencoded strings are a better choice for\n> a printable output form that the text+\\nnn form (since a high\n> proportion of bytes will be emitted in the lengthy \\nnn form).\n> \n> But normally with BLOB one would like a way to just stuff a file\n> or data in a buffer into it, etc, much like current lo. The printable\n> dump of data is mostly useful for pg_dump, IMO - a binary backup would\n> remove the need for such a hack, too.\n> \n> Standard BLOBs provide a way to stuff segments into the db...\n> \n> BLOBs, as done by TOAST or my current segmented table hack used in\n> our toolkit, only require a single table (or a single table per\n> underlying user table in the case of TOAST) so don't clutter the\n> way lo does.\n> \n> But lo allows each binary object to be 2GB in length.\n> \n> So they kind of fit different needs. lo seems fine for those who\n> need really huge objects, and probably not a bazillion (since each\n> generates a file + index). My hack, or TOAST which will be similar\n> in table usage (both being segmented types in common tables), is\n> good for binary data of moderately large size not to exceed 2GB\n> in aggregate.\n> \n> Of course, with 64-bit systems on the horizon, the 2GB aggregate\n> limit will slowly begin to disappear, too. 'Til then, providing\n> a \"real BLOB\" while retaining lo for those who need single REALLY\n> huge data objects would seem best.\n> \n> \n> \n> \n> \n> - Don Baccus, Portland OR <[email protected]>\n> Nature photos, on-line guides, Pacific Northwest\n> Rare Bird Alert Service and other goodies at\n> http://donb.photo.net.\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 29 Sep 2000 22:35:45 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: bytea"
}
] |
[
{
"msg_contents": "Do other people have a comment on this. I am inclined to leave it\nalong. This is the first complaint I have heard, and elog(PGDEBUG) just\nlooks funny. We don't prefix NOTICE or ERROR.\n\n\n> In the file\n> \n> include/utils/elog.h\n> \n> there is a macro named\n> \n> DEBUG\n> \n> that conflicts with the perl5.6 macro DEBUG.\n> \n> PostgreSQL would probably \"play\" better with other products if\n> the DEBUG macro had a prefix, maybe PGSQLDEBUG or similar.\n> \n> Until there is some fix in this area, plperl will not build with\n> a version of perl that has debugging enabled.\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 29 Sep 2000 22:37:30 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Suggested change in include/utils/elog.h"
}
] |
[
{
"msg_contents": "If you do decide to prefix DEBUG, please consider prefixing at least ERROR\nas well. It produces a lot of warnings when compiling on Win32 (ERROR is\nalready defined as part of the standard windows headers). It's just\nwarnings, though.\n\n//Magnus\n\n\n> Do other people have a comment on this. I am inclined to leave it\n> along. This is the first complaint I have heard, and \n> elog(PGDEBUG) just\n> looks funny. We don't prefix NOTICE or ERROR.\n> \n> \n> > In the file\n> > \n> > include/utils/elog.h\n> > \n> > there is a macro named\n> > \n> > DEBUG\n> > \n> > that conflicts with the perl5.6 macro DEBUG.\n> > \n> > PostgreSQL would probably \"play\" better with other products if\n> > the DEBUG macro had a prefix, maybe PGSQLDEBUG or similar.\n> > \n> > Until there is some fix in this area, plperl will not build with\n> > a version of perl that has debugging enabled.\n> > \n> \n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> [email protected] | (610) 853-3000\n> + If your life is a hard drive, | 830 Blythe Avenue\n> + Christ can be your backup. | Drexel Hill, \n> Pennsylvania 19026\n> \n",
"msg_date": "Sat, 30 Sep 2000 10:28:53 +0200",
"msg_from": "Magnus Hagander <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: Suggested change in include/utils/elog.h"
},
{
"msg_contents": "Magnus Hagander wrote:\n\n> If you do decide to prefix DEBUG, please consider prefixing at least ERROR\n> as well. It produces a lot of warnings when compiling on Win32 (ERROR is\n> already defined as part of the standard windows headers). It's just\n> warnings, though.\n>\n> //Magnus\n>\n> > Do other people have a comment on this. I am inclined to leave it\n> > along. This is the first complaint I have heard, and\n> > elog(PGDEBUG) just\n> > looks funny. We don't prefix NOTICE or ERROR.\n> >\n> >\n> > > In the file\n> > >\n> > > include/utils/elog.h\n> > >\n> > > there is a macro named\n> > >\n> > > DEBUG\n> > >\n> > > that conflicts with the perl5.6 macro DEBUG.\n> > >\n> > > PostgreSQL would probably \"play\" better with other products if\n> > > the DEBUG macro had a prefix, maybe PGSQLDEBUG or similar.\n> > >\n> > > Until there is some fix in this area, plperl will not build with\n> > > a version of perl that has debugging enabled.\n> > >\n\nIt even got on my nerves (linux, ecpg) since I used to define a macro\n#define DEBUG(x) cout << x\nor\n#define DEBUG(x)\n\nDEBUG and ERROR are far too common to get defined for client programs.\n\nBut perhaps it is ecpg's fault for including \"elog.h\".\nIMHO these defines should never leave the database kernel.\n\nperhaps the common\n #ifdef _DBKERNEL_\n #endif\nwould do the trick.\n\nChristof\n\nPS: Having Datum unconditionally leaked to ecpg programs forced me to preced\na namespace to my own class.\n\n\n",
"msg_date": "Mon, 02 Oct 2000 23:54:46 +0200",
"msg_from": "Christof Petig <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Suggested change in include/utils/elog.h"
},
{
"msg_contents": "> > > > PostgreSQL would probably \"play\" better with other products if\n> > > > the DEBUG macro had a prefix, maybe PGSQLDEBUG or similar.\n> > > >\n> > > > Until there is some fix in this area, plperl will not build with\n> > > > a version of perl that has debugging enabled.\n> > > >\n> \n> It even got on my nerves (linux, ecpg) since I used to define a macro\n> #define DEBUG(x) cout << x\n> or\n> #define DEBUG(x)\n> \n> DEBUG and ERROR are far too common to get defined for client programs.\n> \n> But perhaps it is ecpg's fault for including \"elog.h\".\n> IMHO these defines should never leave the database kernel.\n> \n> perhaps the common\n> #ifdef _DBKERNEL_\n> #endif\n> would do the trick.\n> \n> Christof\n> \n> PS: Having Datum unconditionally leaked to ecpg programs forced me to preced\n> a namespace to my own class.\n\nYes, leaking into user programs is a bad practice. Is there a\nsolution/patch for that?\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 13:37:45 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Suggested change in include/utils/elog.h"
},
{
"msg_contents": "Bruce Momjian wrote:\n\n> > > > > PostgreSQL would probably \"play\" better with other products if\n> > > > > the DEBUG macro had a prefix, maybe PGSQLDEBUG or similar.\n> > > > >\n> > > > > Until there is some fix in this area, plperl will not build with\n> > > > > a version of perl that has debugging enabled.\n> > > > >\n> >\n> > It even got on my nerves (linux, ecpg) since I used to define a macro\n> > #define DEBUG(x) cout << x\n> > or\n> > #define DEBUG(x)\n> >\n> > DEBUG and ERROR are far too common to get defined for client programs.\n> >\n> > But perhaps it is ecpg's fault for including \"elog.h\".\n> > IMHO these defines should never leave the database kernel.\n> >\n> > perhaps the common\n> > #ifdef _DBKERNEL_\n> > #endif\n> > would do the trick.\n> >\n> > Christof\n> >\n> > PS: Having Datum unconditionally leaked to ecpg programs forced me to preced\n> > a namespace to my own class.\n>\n> Yes, leaking into user programs is a bad practice. Is there a\n> solution/patch for that?\n\nA solution would be a simple patch which is not available yet. But I\nplan on\ndoing one (some other things still have higher priority).\n\nChristof\n",
"msg_date": "Mon, 09 Oct 2000 12:50:26 +0200",
"msg_from": "Christof Petig <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Suggested change in include/utils/elog.h"
},
{
"msg_contents": "On Sun, Oct 08, 2000 at 01:37:45PM -0400, Bruce Momjian wrote:\n> > But perhaps it is ecpg's fault for including \"elog.h\".\n> > IMHO these defines should never leave the database kernel.\n> ... \n> Yes, leaking into user programs is a bad practice. Is there a\n> solution/patch for that?\n\nHmm, I haven't check carefully, but I don't think ecpg include elog.h\ndirectly. Maybe I just picked the wrong file to include.\n\nMichael\n-- \nMichael Meskes\[email protected]\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Tue, 10 Oct 2000 15:47:15 -0700",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Suggested change in include/utils/elog.h"
}
] |
[
{
"msg_contents": "Hi,\n\nThere seems to be a problem in the configure script for solaris.\n(Recent CVS copy)\n\n ./configure\n <snip>\n checking for POSIX signal interface... (cached) yes\n ./configure: syntax error at line 6907: `;' unexpected\n\n\nA \"for\" construct, traced back to config/tcl.m4, is upsetting\nthe shell interpreter.\n\nThe construct..\n\n for pgac_dir; do\n if test -r \"$pgac_dir/tclConfig.sh\"; then\n TCL_CONFIG_SH=$pgac_dir/tclConfig.sh\n break\n fi\n done\n\nfails, but..\n\n for pgac_dir\n do\n if test -r \"$pgac_dir/tclConfig.sh\"; then\n TCL_CONFIG_SH=$pgac_dir/tclConfig.sh\n break\n fi\n done\n\nis OK.\n\nI think either should be OK, indeed bash is happy with both.\n\nPerhaps we could change to the latter to keep Solaris happy?\n\nBTW: On the latest CVS, a couple of the changed regression tests\nfail because the solaris specific \"expected\" files don't match\nthe new results. Do you want me to supply diffs to bring them\nupto date?\n\nThanks,\nKeith.\n\n\n--\n",
"msg_date": "Sat, 30 Sep 2000 10:45:30 +0100",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "Solaris configure problems."
},
{
"msg_contents": "Keith Parks writes:\n\n> for pgac_dir; do\n> if test -r \"$pgac_dir/tclConfig.sh\"; then\n> TCL_CONFIG_SH=$pgac_dir/tclConfig.sh\n> break\n> fi\n> done\n> \n> fails\n\nThanks. The semicolon is actually only needed (allowed) with an `in'\nclause.\n\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Sat, 30 Sep 2000 12:50:16 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Solaris configure problems."
}
] |
[
{
"msg_contents": "I have a table. When I vacuum verbose it...\n\nmail=# vacuum verbose tblemaildomain ;\nNOTICE: --Relation tblemaildomain--\nNOTICE: Pages 199: Changed 0, reaped 0, Empty 0, New 0; Tup 27494: Vac 0,\nKeep/VTL 1/0, Crash 0, UnUsed 0, MinLen 40, MaxLen 80; Re-using:\nFree/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. CPU 0.00s/0.02u sec.\nNOTICE: Index tblemaildomain_pkey: Pages 70; Tuples 27494. CPU \n0.00s/0.03u sec.\nNOTICE: Index tblemaildomain_oid: Pages 70; Tuples 27494. CPU 0.00s/0.04u \nsec.\nVACUUM\n\nWhen I vacuum verbose analyze it...\n\nmail=# vacuum verbose analyze tblemaildomain ;\nNOTICE: --Relation tblemaildomain--\nFATAL 1: Memory exhausted in AllocSetAlloc()\npqReadData() -- backend closed the channel unexpectedly.\n This probably means the backend terminated abnormally\n before or while processing the request.\nThe connection to the server was lost. Attempting reset: Succeeded.\n\nThoughts/solutions? Thanks...\n\n- K\n\nKristofer Munn * KMI * 732-254-9305 * AIM KrMunn * http://www.munn.com/\n\n",
"msg_date": "Sat, 30 Sep 2000 13:39:42 -0400 (EDT)",
"msg_from": "Kristofer Munn <[email protected]>",
"msg_from_op": true,
"msg_subject": "Crash during vacuum analyze only..."
}
] |
[
{
"msg_contents": "Just for fun I changed the extension of my SPI function and trigger\nsource file from .c to .C and tried to compile it:\n\n> g++ -g -Wall -I /usr/include/postgresql -fPIC -pedantic -c -o bid_control.o bid_control.C\n> In file included from /usr/include/postgresql/nodes/relation.h:18,\n> from /usr/include/postgresql/executor/spi.h:18,\n> from bid_control.h:31,\n> from bid_control.C:21:\n> /usr/include/postgresql/nodes/parsenodes.h:871: parse error before `typename'\n\nIt seems the \"typename\" word used in the following structure is a\nreserved word in C++. (and is appears colored like a type in Vim)\n\n> /*\n> * ParamNo - specifies a parameter reference\n> */\n> typedef struct ParamNo\n> {\n> NodeTag type;\n> int number; /* the number of the parameter */\n> TypeName *typename; /* the typecast */\n ^^^^^^^^^\n> List *indirection; /* array references */\n> } ParamNo;\n\nIs it totally unreasonable try C++ in SPI programming?\n\nCheers,\n\n\n-- \nLouis-David Mitterrand - [email protected] - http://www.apartia.org\n\nIsn't vi that text editor with two modes... one that beeps and one\nthat corrupts your file?\" -- Dan Jocabson, on comp.os.linux.advocacy\n",
"msg_date": "Sun, 1 Oct 2000 10:11:02 +0200",
"msg_from": "Louis-David Mitterrand <[email protected]>",
"msg_from_op": true,
"msg_subject": "failed to compile a C++ SPI function"
},
{
"msg_contents": "I am willing to change it to typname. Any comments?\n\n> Just for fun I changed the extension of my SPI function and trigger\n> source file from .c to .C and tried to compile it:\n> \n> > g++ -g -Wall -I /usr/include/postgresql -fPIC -pedantic -c -o bid_control.o bid_control.C\n> > In file included from /usr/include/postgresql/nodes/relation.h:18,\n> > from /usr/include/postgresql/executor/spi.h:18,\n> > from bid_control.h:31,\n> > from bid_control.C:21:\n> > /usr/include/postgresql/nodes/parsenodes.h:871: parse error before `typename'\n> \n> It seems the \"typename\" word used in the following structure is a\n> reserved word in C++. (and is appears colored like a type in Vim)\n> \n> > /*\n> > * ParamNo - specifies a parameter reference\n> > */\n> > typedef struct ParamNo\n> > {\n> > NodeTag type;\n> > int number; /* the number of the parameter */\n> > TypeName *typename; /* the typecast */\n> ^^^^^^^^^\n> > List *indirection; /* array references */\n> > } ParamNo;\n> \n> Is it totally unreasonable try C++ in SPI programming?\n> \n> Cheers,\n> \n> \n> -- \n> Louis-David Mitterrand - [email protected] - http://www.apartia.org\n> \n> Isn't vi that text editor with two modes... one that beeps and one\n> that corrupts your file?\" -- Dan Jocabson, on comp.os.linux.advocacy\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sat, 7 Oct 2000 23:14:27 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: failed to compile a C++ SPI function"
},
{
"msg_contents": "Bruce Momjian <[email protected]> writes:\n> I am willing to change it to typname. Any comments?\n\nThat's hardly the only change you will have to make to get the headers\nto pass through a C++ compiler without complaint. How many existing\napplications do you want to risk breaking?\n\nParamNo probably isn't a widely used node type, but there are four\nstructs with fields named 'typename' and one with a field named 'class'\njust in parsenodes.h.\n\nI think that before we start making changes with C++ cleanness as a\ngoal, we should try to get a precise idea of the extent of the changes\nthat would be needed, and then estimate how much user code might\nreference the fields we'd need to rename.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 07 Oct 2000 23:57:12 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: failed to compile a C++ SPI function "
},
{
"msg_contents": "> Bruce Momjian <[email protected]> writes:\n> > I am willing to change it to typname. Any comments?\n> \n> That's hardly the only change you will have to make to get the headers\n> to pass through a C++ compiler without complaint. How many existing\n> applications do you want to risk breaking?\n> \n> ParamNo probably isn't a widely used node type, but there are four\n> structs with fields named 'typename' and one with a field named 'class'\n> just in parsenodes.h.\n> \n> I think that before we start making changes with C++ cleanness as a\n> goal, we should try to get a precise idea of the extent of the changes\n> that would be needed, and then estimate how much user code might\n> reference the fields we'd need to rename.\n\nAgreed.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 00:02:57 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: failed to compile a C++ SPI function"
}
] |
[
{
"msg_contents": "Hi,\n\nThe subject says it all.\n\nI want to mess up the backend so...\n\nDo I need lex or flex?\nDo I need autoconf (does it even exist for Windows)?\nDo I need Cygnus?\nDo I need any other aditional packages?\n\nDoes it compile with VC6?\nDoes nmake do or do I need gnu make?\n\nOr is it easier to play around with the source in Linux?\nWhat do I need in Linux (Cygnus, lex, flex...)?\n\n// Jarmo\n\n",
"msg_date": "Sun, 1 Oct 2000 14:07:28 +0200",
"msg_from": "\"Jarmo Paavilainen\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "What do I need to be able to compile the source in Win32"
},
{
"msg_contents": "\"Jarmo Paavilainen\" <[email protected]> writes:\n> Or is it easier to play around with the source in Linux?\n\nYes. You could maybe do it in a Cygwin environment, but I think\nit'd be an exercise in frustration.\n\n> What do I need in Linux (Cygnus, lex, flex...)?\n\nAs long as you install development tools (compilers etc) you'll\nprobably have all you need.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 01 Oct 2000 11:24:45 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: What do I need to be able to compile the source in Win32 "
}
] |
[
{
"msg_contents": "\nIt recently came to my attention that, when dumping a single table and\nusing the --blobs option, pg_dump will dump all BLOBs (not just those\nreferenced by the table). Similarly, restoring a single table from a backup\nwill restore all the BLOBs.\n\nThis is (at least) non-intuitive. \n\nUnfortunately, the task of restoring a single table with BLOBs is actually\nsomewhat complex: the OIDs will change, and the table needs to be updated\n(which is fine), but it is definitely not clear how to handle other tables\nin the database (since the current implementation allows multiple\nreferences to one blob, which is very non-standard). The simple solution of\nonly updating the OID fields in the restored table seems as unintuitive as\nthe current situation, and also likely to produce unexpected results.\n\nAlso, scanning through the backup file for BLOBs used in the table to be\nrestored will be slow, to say the least. It can be done in one pass by\nbuilding an indexed temporary table with all OID fields from the target\ntable, but it is definitely getting messy. Especially since Jan was talking\nabout a decent BLOB implementation RSN.\n\nMy inclination at this stage is to disallow --blobs when dumping or\nrestoring a single table, but I would be interested in hearing from anybody\nwith a better solution.\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Mon, 02 Oct 2000 00:41:05 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": true,
"msg_subject": "pg_dump, BLOBs and single-table dumps - RFC"
},
{
"msg_contents": "> Also, scanning through the backup file for BLOBs used in the table to be\n> restored will be slow, to say the least. It can be done in one pass by\n> building an indexed temporary table with all OID fields from the target\n> table, but it is definitely getting messy. Especially since Jan was talking\n> about a decent BLOB implementation RSN.\n> \n> My inclination at this stage is to disallow --blobs when dumping or\n> restoring a single table, but I would be interested in hearing from anybody\n> with a better solution.\n\nI agree.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 00:11:28 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump, BLOBs and single-table dumps - RFC"
}
] |
[
{
"msg_contents": "\nTrying to compile PostgreSQL latest-CVS on SuSE 6.4 produces the following\nerror (was talked about on the list this past week, IIRC):\n\nmake[3]: Entering directory `/tmp/pgsql/src/backend/parser'\nyacc -d gram.y\nyacc: f - maximum table size exceeded\nmake[3]: *** [parse.h] Error 2\n\nroot@server1:/tmp/pgsql > rpm -q -a | grep 'yacc'\nyacc-91.7.30-167\n\n\nJust a FYI on another distrib that might cause \"problems\" ...\n\n-- \nDominic J. Eidson\n \"Baruk Khazad! Khazad ai-menu!\" - Gimli\n-------------------------------------------------------------------------------\nhttp://www.the-infinite.org/ http://www.the-infinite.org/~dominic/\n\n",
"msg_date": "Sun, 1 Oct 2000 15:24:50 -0500 (CDT)",
"msg_from": "\"Dominic J. Eidson\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "On the subject of yacc..."
},
{
"msg_contents": "Dominic J. Eidson writes:\n\n> Trying to compile PostgreSQL latest-CVS on SuSE 6.4 produces the following\n> error (was talked about on the list this past week, IIRC):\n> \n> make[3]: Entering directory `/tmp/pgsql/src/backend/parser'\n> yacc -d gram.y\n> yacc: f - maximum table size exceeded\n> make[3]: *** [parse.h] Error 2\n\nYou need to install bison or get your yacc to increase its table size.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Sun, 1 Oct 2000 22:39:54 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: On the subject of yacc..."
}
] |
[
{
"msg_contents": "> Yes. You could maybe do it in a Cygwin environment, but I think\n> it'd be an exercise in frustration.\nIt depends on your skills. Cygwin is a UNIX environment on NT under\ndevelopment. \n \n> > What do I need in Linux (Cygnus, lex, flex...)?\nCygwin 1.1.4 , CygIPC 1.05 or higher, the patch from Jason Tishler (posted\n27-9 in the ports list).\n\nThan follow the docs that come with PostgreSQL. In essence: make, make\ninstall.\n\nJoost Kraaijeveld\nAskesis B.V.\nMolukkenstraat 14\n6524NB Nijmegen\ntel: 024-3888063 / 06-51855277\nfax: 024-3608416\nemail: [email protected]\nweb: www.askesis.nl \n\n\n",
"msg_date": "Sun, 1 Oct 2000 23:10:38 +0200 ",
"msg_from": "Joost Kraaijeveld <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: What do I need to be able to compile the source in\n\tWin32"
}
] |
[
{
"msg_contents": "Sorry,\n\nPostgres is 6.5.3 on Linux 2.2.10\n\nRalf\n\n\n",
"msg_date": "Mon, 2 Oct 2000 00:59:48 +0200 (MET DST)",
"msg_from": "Ralf Tschiersch <[email protected]>",
"msg_from_op": true,
"msg_subject": "pg_log deleted, still help needed (2)"
}
] |
[
{
"msg_contents": "A couple glitches to work out in pg_regress:\n\n1. 'make installcheck' fails if PGPORT isn't set in environment. Should\nomit '-p $PGPORT' options from program invocations unless PGPORT is set.\nActually, I think you should omit -p period; the programs are perfectly\ncapable of reading PGPORT from the environment for themselves.\n\n2. I get this behavior:\n\ntest hash_index ... ok\ntest misc ... FAILED\ntest select_views ... ok\ntest alter_table ... ok\ntest portals_p2 ... ok\ntest rules ... ok\ntest foreign_key ... ok\ntest limit ... ok\ntest plpgsql ... ok\ntest temp ... ok\n\n=================================================\n 0 of 75 tests passed, 0 failed test(s) ignored.\n=================================================\n\n(The misc failure is from a bug in some other stuff I'm working on;\nI show it just to illustrate that pg_regress's helpful totals are\ncompletely incorrect.) I presume the output is more useful for you,\nwhich probably means that echo behaves differently on your platform.\nPossibly should be looking for '\\.\\.\\. *ok' and so on?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 02 Oct 2000 02:16:47 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": true,
"msg_subject": "pg_regress gripes"
}
] |
[
{
"msg_contents": "\n\n What happen with www.postgresql.org? My browser is always redirected \nto http://www.cz.postgresql.org/.\n\n But I haven't some terrible mirror (with last change \"July 1999\"), \nI want nice and native PG's web! :-)\n\n\t\t\t\tKarel\n\nPS. Exist some check what happen at PostgreSQL mirrors?\n The www.cz.postgresql.org is *discredit*... (sorry Czech friends).\n\n",
"msg_date": "Mon, 2 Oct 2000 09:21:18 +0200 (CEST)",
"msg_from": "Karel Zak <[email protected]>",
"msg_from_op": true,
"msg_subject": "www.postgresql.org"
},
{
"msg_contents": "The only fix is to do\n\n\thttp://www.cz.postgresql.org/index.html\n\nThis prevents mirror use. The old mirror should be fixed or removed.\n\n> \n> \n> What happen with www.postgresql.org? My browser is always redirected \n> to http://www.cz.postgresql.org/.\n> \n> But I haven't some terrible mirror (with last change \"July 1999\"), \n> I want nice and native PG's web! :-)\n> \n> \t\t\t\tKarel\n> \n> PS. Exist some check what happen at PostgreSQL mirrors?\n> The www.cz.postgresql.org is *discredit*... (sorry Czech friends).\n> \n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 2 Oct 2000 11:49:58 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: www.postgresql.org"
},
{
"msg_contents": "\n\nRemoved from redirect.\n\nVince.\n\n\nOn Mon, 2 Oct 2000, Karel Zak wrote:\n\n> \n> \n> What happen with www.postgresql.org? My browser is always redirected \n> to http://www.cz.postgresql.org/.\n> \n> But I haven't some terrible mirror (with last change \"July 1999\"), \n> I want nice and native PG's web! :-)\n> \n> \t\t\t\tKarel\n> \n> PS. Exist some check what happen at PostgreSQL mirrors?\n> The www.cz.postgresql.org is *discredit*... (sorry Czech friends).\n> \n> \n\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Mon, 2 Oct 2000 12:11:27 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: www.postgresql.org"
},
{
"msg_contents": "Thanks. Also, main server is backup now.\n\n> \n> \n> Removed from redirect.\n> \n> Vince.\n> \n> \n> On Mon, 2 Oct 2000, Karel Zak wrote:\n> \n> > \n> > \n> > What happen with www.postgresql.org? My browser is always redirected \n> > to http://www.cz.postgresql.org/.\n> > \n> > But I haven't some terrible mirror (with last change \"July 1999\"), \n> > I want nice and native PG's web! :-)\n> > \n> > \t\t\t\tKarel\n> > \n> > PS. Exist some check what happen at PostgreSQL mirrors?\n> > The www.cz.postgresql.org is *discredit*... (sorry Czech friends).\n> > \n> > \n> \n> -- \n> ==========================================================================\n> Vince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n> 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n> Online Campground Directory http://www.camping-usa.com\n> Online Giftshop Superstore http://www.cloudninegifts.com\n> ==========================================================================\n> \n> \n> \n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 2 Oct 2000 12:29:00 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: www.postgresql.org"
},
{
"msg_contents": "\nOn Mon, 2 Oct 2000, Bruce Momjian wrote:\n\n> The only fix is to do\n> \n> \thttp://www.cz.postgresql.org/index.html\n\nIt's returns still same a *bad* result.\n\n> \n> This prevents mirror use. The old mirror should be fixed or removed.\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nSure. I repost this mail to www.cz.postgresql.org admin at mendelu.cz...\n\n\t\t\t\t\tKarel\n\nPS. Thanks for redirect fix, www.postgresql.org is more nice :-)\n\n> \n> > \n> > \n> > What happen with www.postgresql.org? My browser is always redirected \n> > to http://www.cz.postgresql.org/.\n> > \n> > But I haven't some terrible mirror (with last change \"July 1999\"), \n> > I want nice and native PG's web! :-)\n> > \n> > \t\t\t\tKarel\n> > \n> > PS. Exist some check what happen at PostgreSQL mirrors?\n> > The www.cz.postgresql.org is *discredit*... (sorry Czech friends).\n> > \n\n",
"msg_date": "Mon, 2 Oct 2000 19:00:36 +0200 (CEST)",
"msg_from": "Karel Zak <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: www.postgresql.org"
},
{
"msg_contents": "On Mon, 2 Oct 2000, Karel Zak wrote:\n\n> \n> On Mon, 2 Oct 2000, Bruce Momjian wrote:\n> \n> > The only fix is to do\n> > \n> > \thttp://www.cz.postgresql.org/index.html\n> \n> It's returns still same a *bad* result.\n\nThis will bypass the redirect:\n\n http://www.postgresql.org/index.html\n\nVince.\n\n\n\n> \n> > \n> > This prevents mirror use. The old mirror should be fixed or removed.\n> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n> Sure. I repost this mail to www.cz.postgresql.org admin at mendelu.cz...\n> \n> \t\t\t\t\tKarel\n> \n> PS. Thanks for redirect fix, www.postgresql.org is more nice :-)\n> \n> > \n> > > \n> > > \n> > > What happen with www.postgresql.org? My browser is always redirected \n> > > to http://www.cz.postgresql.org/.\n> > > \n> > > But I haven't some terrible mirror (with last change \"July 1999\"), \n> > > I want nice and native PG's web! :-)\n> > > \n> > > \t\t\t\tKarel\n> > > \n> > > PS. Exist some check what happen at PostgreSQL mirrors?\n> > > The www.cz.postgresql.org is *discredit*... (sorry Czech friends).\n> > > \n> \n> \n\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Mon, 2 Oct 2000 13:07:46 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: www.postgresql.org"
},
{
"msg_contents": "> \n> On Mon, 2 Oct 2000, Bruce Momjian wrote:\n> \n> > The only fix is to do\n> > \n> > \thttp://www.cz.postgresql.org/index.html\n> \n> It's returns still same a *bad* result.\n\nNo, sorry, use:\n\n\thttp://www.postgresql.org/index.html\n\nThe index.html prevents the redirect.\n\n> \n> > \n> > This prevents mirror use. The old mirror should be fixed or removed.\n> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n> Sure. I repost this mail to www.cz.postgresql.org admin at mendelu.cz...\n> \n> \t\t\t\t\tKarel\n> \n> PS. Thanks for redirect fix, www.postgresql.org is more nice :-)\n> \n> > \n> > > \n> > > \n> > > What happen with www.postgresql.org? My browser is always redirected \n> > > to http://www.cz.postgresql.org/.\n> > > \n> > > But I haven't some terrible mirror (with last change \"July 1999\"), \n> > > I want nice and native PG's web! :-)\n> > > \n> > > \t\t\t\tKarel\n> > > \n> > > PS. Exist some check what happen at PostgreSQL mirrors?\n> > > The www.cz.postgresql.org is *discredit*... (sorry Czech friends).\n> > > \n> \n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 2 Oct 2000 13:23:57 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: www.postgresql.org"
},
{
"msg_contents": "Mon, Oct 02, 2000 ve 01:07:46PM -0400 Vince Vielhaber napsal:\n# On Mon, 2 Oct 2000, Karel Zak wrote:\n# \n# > \n# > On Mon, 2 Oct 2000, Bruce Momjian wrote:\n# > \n# > > The only fix is to do\n# > > \n# > > \thttp://www.cz.postgresql.org/index.html\n# > \n# > It's returns still same a *bad* result.\n# \n# This will bypass the redirect:\n# \n# http://www.postgresql.org/index.html\n# \n# Vince.\n# \n\nI set redirect permanent to http://www.postgresql.org/\nMy hard disk was broken :(\n\n--milan sorm\n",
"msg_date": "Wed, 4 Oct 2000 10:03:03 +0200",
"msg_from": "Milan Sorm <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: www.postgresql.org"
}
] |
[
{
"msg_contents": "The file \"postgres.h\" (or \"c.h\" or \"config.h\", whatever is used) needs to\nbe the very *first* file included by each source file. Next time you\ntouch a source file, please check that this is the case.\n\nThe obvious failure mode is that if config.h redefines const, volatile, or\ninline then it will cause confusion when some system headers are included\nbefore and some after that definition.\n\nThe slightly more esoteric problem I encountered is that when you compile\nwith CC='gcc -std=c99 -pedantic' on a glibc platform (i.e., \"Linux\") then\nyou need to define _SVID_SOURCE and _BSD_SOURCE before including any\nsystem header in order to get the full feature set from the headers.\n\n(Unfortunately, the flex output does not observe this rule either, so we\ncan't be 100% pedantic warning safe without doing surgery on those files.)\n\nOn a related note, does anyone know why the on_proc_exit and on_shmem_exit\nhooks use a second argument of type `caddr_t' rather than, say, void*,\nchar*, Datum, ...? This artifact is the cause of about two thirds of the\ncompile errors in the pedantic setting.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Mon, 2 Oct 2000 11:47:48 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Note about include files"
},
{
"msg_contents": "* Peter Eisentraut <[email protected]> [001002 02:51] wrote:\n> The file \"postgres.h\" (or \"c.h\" or \"config.h\", whatever is used) needs to\n> be the very *first* file included by each source file. Next time you\n> touch a source file, please check that this is the case.\n> \n> The obvious failure mode is that if config.h redefines const, volatile, or\n> inline then it will cause confusion when some system headers are included\n> before and some after that definition.\n> \n> The slightly more esoteric problem I encountered is that when you compile\n> with CC='gcc -std=c99 -pedantic' on a glibc platform (i.e., \"Linux\") then\n> you need to define _SVID_SOURCE and _BSD_SOURCE before including any\n> system header in order to get the full feature set from the headers.\n> \n> (Unfortunately, the flex output does not observe this rule either, so we\n> can't be 100% pedantic warning safe without doing surgery on those files.)\n\ngcc supports the '-include' directive which may be what you want.\n\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Mon, 2 Oct 2000 02:58:38 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Note about include files"
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> On a related note, does anyone know why the on_proc_exit and on_shmem_exit\n> hooks use a second argument of type `caddr_t' rather than, say, void*,\n> char*, Datum, ...? This artifact is the cause of about two thirds of the\n> compile errors in the pedantic setting.\n\nI was annoyed by that just the other day on an Alpha box (it provokes\nlots of \"integer cast to pointer of different size\" warnings there).\nI'm sure the use of caddr_t is strictly historical.\n\nIf you feel like doing something about it, changing the arguments of\nthese routines to be Datum and then adding the necessary\nto-and-from-Datum macros seems like the obvious solution path.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 02 Oct 2000 10:45:03 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Note about include files "
},
{
"msg_contents": "> The file \"postgres.h\" (or \"c.h\" or \"config.h\", whatever is used) needs to\n> be the very *first* file included by each source file. Next time you\n> touch a source file, please check that this is the case.\n> \n> The obvious failure mode is that if config.h redefines const, volatile, or\n> inline then it will cause confusion when some system headers are included\n> before and some after that definition.\n> \n> The slightly more esoteric problem I encountered is that when you compile\n> with CC='gcc -std=c99 -pedantic' on a glibc platform (i.e., \"Linux\") then\n> you need to define _SVID_SOURCE and _BSD_SOURCE before including any\n> system header in order to get the full feature set from the headers.\n> \n> (Unfortunately, the flex output does not observe this rule either, so we\n> can't be 100% pedantic warning safe without doing surgery on those files.)\n> \n> On a related note, does anyone know why the on_proc_exit and on_shmem_exit\n> hooks use a second argument of type `caddr_t' rather than, say, void*,\n> char*, Datum, ...? This artifact is the cause of about two thirds of the\n> compile errors in the pedantic setting.\n\nThat was me. caddr_t was I used for function pointers. Glad you\nchanged it.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 00:28:07 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Note about include files]"
}
] |
[
{
"msg_contents": "The current behaviour of libpq is to use Unix sockets whenever the host\nparameter (PGHOST or setdbLogin argument) is NULL/unset.\n\nCould we extend that to also use Unix sockets if the parameter is set but\nempty? That could avoid a bunch of shell contortions; e.g., you can't\nportably un-export variables, in some shells you don't even have \"unset\".\n\nOr would this be too incompatible?\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Mon, 2 Oct 2000 13:09:12 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "libpq PGHOST"
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> The current behaviour of libpq is to use Unix sockets whenever the host\n> parameter (PGHOST or setdbLogin argument) is NULL/unset.\n> Could we extend that to also use Unix sockets if the parameter is set but\n> empty?\n\nSeems reasonable, since the current behavior in that case is useless:\n\n$ PGHOST='' psql\npsql: connectDBStart() -- unknown hostname:\n$\n\nBackwards compatibility with that doesn't seem necessary ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 02 Oct 2000 10:48:15 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: libpq PGHOST "
},
{
"msg_contents": "Tom Lane writes:\n\n> > The current behaviour of libpq is to use Unix sockets whenever the host\n> > parameter (PGHOST or setdbLogin argument) is NULL/unset.\n> > Could we extend that to also use Unix sockets if the parameter is set but\n> > empty?\n> Seems reasonable, since the current behavior in that case is useless:\n> $ PGHOST='' psql\n> psql: connectDBStart() -- unknown hostname:\n> $\n> Backwards compatibility with that doesn't seem necessary ...\n\nAfter further investigation, there seems to be a larger unset/empty mess.\n \nWhen using PQconnectdb(), a NULL parameter (keyword was not given at all)\nmeans to use the environment variable, an explicit empty argument is used\nas is. When using PGsetdbLogin(), however both NULL and empty arguments\ncause the environment variable to be used. (An environment variable is\nalways used as is.)\n\nConsequently, if PGHOST is set in the environment and your application is\nusing PGsetdbLogin(), then it's just impossible to get a Unix socket\nconnection.\n\n(Note that this is independent of the proposed change, because PGHOST may\nbe set to some \"real\" string that you might wish to override.)\n\nIf we were to sort this out, then I think we'd need to change\nPQsetdbLogin() to take empty arguments uniformly \"as is\", and make the\nchange to make Unix sockets also with an empty host parameter, as\nproposed. Then you could use psql -h '' to request a Unix socket\nexplicitly.\n\n\nSomewhat related:\n\npeter=# create function \"\" () returns int as 'select 42' language 'sql';\nCREATE\npeter=# select \"\"();\n \n----\n 42\n(1 row)\n\nThat was probably not the plan.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Mon, 2 Oct 2000 21:07:58 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: libpq PGHOST "
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> If we were to sort this out, then I think we'd need to change\n> PQsetdbLogin() to take empty arguments uniformly \"as is\", and make the\n> change to make Unix sockets also with an empty host parameter, as\n> proposed. Then you could use psql -h '' to request a Unix socket\n> explicitly.\n\nSeems reasonable.\n\n> peter=# create function \"\" () returns int as 'select 42' language 'sql';\n> CREATE\n> peter=# select \"\"();\n \n> ----\n> 42\n> (1 row)\n\n> That was probably not the plan.\n\nI don't see any clear statement in SQL92 that <delimited identifiers>\ncan't have zero length, so I'm not convinced there's anything wrong here.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 02 Oct 2000 15:36:32 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: libpq PGHOST "
},
{
"msg_contents": "Tom Lane writes:\n\n> I don't see any clear statement in SQL92 that <delimited identifiers>\n> can't have zero length, so I'm not convinced there's anything wrong here.\n\nHmm, I do:\n\n <delimited identifier> ::=\n <double quote> <delimited identifier body> <double quote>\n \n <delimited identifier body> ::= <delimited identifier part>...\n \n <delimited identifier part> ::=\n <nondoublequote character>\n | <doublequote symbol>\n\nAs opposed to, say, an empty character string literal:\n\n <character string literal> ::=\n [ <introducer><character set specification> ]\n <quote> [ <character representation>... ] <quote>\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n [ { <separator>... <quote> [ <character representation>... ] <quote> }... ] \n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Tue, 3 Oct 2000 14:11:07 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: libpq PGHOST "
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> Tom Lane writes:\n>> I don't see any clear statement in SQL92 that <delimited identifiers>\n>> can't have zero length, so I'm not convinced there's anything wrong here.\n\n> Hmm, I do:\n\n> <delimited identifier> ::=\n> <double quote> <delimited identifier body> <double quote>\n \n> <delimited identifier body> ::= <delimited identifier part>...\n \n> <delimited identifier part> ::=\n> <nondoublequote character>\n> | <doublequote symbol>\n\n> As opposed to, say, an empty character string literal:\n\n> <character string literal> ::=\n> [ <introducer><character set specification> ]\n> <quote> [ <character representation>... ] <quote>\n> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n> [ { <separator>... <quote> [ <character representation>... ] <quote> }... ] \n\nHmm, you're right --- if they meant to allow zero-length identifiers\nthey'd have written something like\n\n <delimited identifier> ::=\n <double quote> [ <delimited identifier body> ] <double quote>\n\nI stand corrected.\n\nNext question is do we want to change it? I can't imagine any real good\nuse for a null-string name, but is there any risk of breaking existing\napplications?\n\nThe actual fix would be trivial --- just introduce a complaint for\nstrlen(literalbuf)==0 into parser/scan.l's <xd>{xdstop} rule.\nI'm just wondering if we should change or not.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 03 Oct 2000 10:26:43 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Delimited identifier length"
},
{
"msg_contents": "Tom Lane writes:\n\n> Next question is do we want to change it? I can't imagine any real good\n> use for a null-string name, but is there any risk of breaking existing\n> applications?\n\nI think the larger risk is breaking applications if we leave it this way. \nOne such broken application is the format_type function and thus pg_dump. \nThat can be fixed of course, but there might be more programs that use a\nsimilar simple-minded mechanism to check whether quoting is needed.\n\nAlso, an empty identifier might be more likely to be a client program bug\nthan really intended, especially in languages that handle variables\nloosely.\n\nOTOH, I'd suggest that we do not enforce the 128 maximum length mandated\nby SQL. :-)\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Wed, 4 Oct 2000 03:00:41 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Delimited identifier length"
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> Tom Lane writes:\n>> Next question is do we want to change it? I can't imagine any real good\n>> use for a null-string name, but is there any risk of breaking existing\n>> applications?\n\n> I think the larger risk is breaking applications if we leave it this way. \n\n[... cogitates ...] Yeah, you're probably right. I agree with changing\nit --- any other objectors out there?\n\n> OTOH, I'd suggest that we do not enforce the 128 maximum length mandated\n> by SQL. :-)\n\nCheck. Not that I want to set NAMEDATALEN to 129, mind you ... but\nI feel that *silent* truncation of overlength identifiers is good\nprogramming language style. At the moment I seem to be outvoted:\n\nregression=# create table z1234567890123456789012345678901234567890 (f1 int);\nNOTICE: identifier \"z1234567890123456789012345678901234567890\" will be truncated to \"z123456789012345678901234567890\"\nCREATE\n\nbut I will vote to remove that notice if it comes to a vote again.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 03 Oct 2000 23:14:49 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Delimited identifier length "
}
] |
[
{
"msg_contents": "Hi,\n\nI would like to know if postgres comes with some kind of backup application? \nI mean something that would do database backup and restore, something like \ninformix's ontape and logical logs.\n\nThanks\n\n-- \n\"And I'm happy, because you make me feel good, about me.\" - Melvin Udall\n-----------------------------------------------------------------\nMart�n Marqu�s\t\t\temail: \[email protected]\nSanta Fe - Argentina\t\thttp://math.unl.edu.ar/~martin/\nAdministrador de sistemas en math.unl.edu.ar\n-----------------------------------------------------------------\n",
"msg_date": "Mon, 2 Oct 2000 08:42:52 -0300",
"msg_from": "\"Martin A. Marques\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "backup and restore"
},
{
"msg_contents": "\"Martin A. Marques\" wrote:\n> \n> Hi,\n> \n> I would like to know if postgres comes with some kind of backup application?\n> I mean something that would do database backup and restore, something like\n> informix's ontape and logical logs.\n\nTake a look at pg_dump and pg_dumpall\n\n-------------\nHannu\n",
"msg_date": "Mon, 02 Oct 2000 14:45:35 +0300",
"msg_from": "Hannu Krosing <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: backup and restore"
},
{
"msg_contents": "On Mon, 02 Oct 2000, Hannu Krosing wrote:\n> \"Martin A. Marques\" wrote:\n> > Hi,\n> >\n> > I would like to know if postgres comes with some kind of backup\n> > application? I mean something that would do database backup and restore,\n> > something like informix's ontape and logical logs.\n>>\n> Take a look at pg_dump and pg_dumpall\n\nWell, that would only be part og what I'm looking for. The thing I like about \ninformix is that I can make a Level 0 backup of all the data (equal to the \npg_dumpall), and then leave the logical logs downloading continuosly, so that \nif in one moment the system breaks, I restore the Level 0 backup and then \napply the logical logs, which are the small changes that have been done to \nthe database in each transaction, administration, etc.\n\nCould this be added? I am willing to help with the coding.\n\nSaludos... :-)\n\n-- \n\"And I'm happy, because you make me feel good, about me.\" - Melvin Udall\n-----------------------------------------------------------------\nMart�n Marqu�s\t\t\temail: \[email protected]\nSanta Fe - Argentina\t\thttp://math.unl.edu.ar/~martin/\nAdministrador de sistemas en math.unl.edu.ar\n-----------------------------------------------------------------\n",
"msg_date": "Tue, 3 Oct 2000 10:26:04 -0300",
"msg_from": "\"Martin A. Marques\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: backup and restore"
},
{
"msg_contents": "This will be in 7.1 as WAL (write-ahead log).\n\n> On Mon, 02 Oct 2000, Hannu Krosing wrote:\n> > \"Martin A. Marques\" wrote:\n> > > Hi,\n> > >\n> > > I would like to know if postgres comes with some kind of backup\n> > > application? I mean something that would do database backup and restore,\n> > > something like informix's ontape and logical logs.\n> >>\n> > Take a look at pg_dump and pg_dumpall\n> \n> Well, that would only be part og what I'm looking for. The thing I like about \n> informix is that I can make a Level 0 backup of all the data (equal to the \n> pg_dumpall), and then leave the logical logs downloading continuosly, so that \n> if in one moment the system breaks, I restore the Level 0 backup and then \n> apply the logical logs, which are the small changes that have been done to \n> the database in each transaction, administration, etc.\n> \n> Could this be added? I am willing to help with the coding.\n> \n> Saludos... :-)\n> \n> -- \n> \"And I'm happy, because you make me feel good, about me.\" - Melvin Udall\n> -----------------------------------------------------------------\n> Mart�n Marqu�s\t\t\temail: \[email protected]\n> Santa Fe - Argentina\t\thttp://math.unl.edu.ar/~martin/\n> Administrador de sistemas en math.unl.edu.ar\n> -----------------------------------------------------------------\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 00:30:29 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: backup and restore"
},
{
"msg_contents": "On Sun, 08 Oct 2000, Bruce Momjian wrote:\n> This will be in 7.1 as WAL (write-ahead log).\n\nWhat does this exactly mean? Will this WAL be an equivalent of the logical \nlog of Informix? Where will they be kept? In the database, as a file, or as \neither?\nLooks pretty good. All that is needed after this is a good blob \nimplementation (for text and binary data).\n\nSaludos... ;-)\n\n-- \n\"And I'm happy, because you make me feel good, about me.\" - Melvin Udall\n-----------------------------------------------------------------\nMart�n Marqu�s\t\t\temail: \[email protected]\nSanta Fe - Argentina\t\thttp://math.unl.edu.ar/~martin/\nAdministrador de sistemas en math.unl.edu.ar\n-----------------------------------------------------------------\n",
"msg_date": "Mon, 9 Oct 2000 08:26:50 -0300",
"msg_from": "\"Martin A. Marques\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: backup and restore"
},
{
"msg_contents": "\"Martin A. Marques\" wrote:\n> \n> On Sun, 08 Oct 2000, Bruce Momjian wrote:\n> > This will be in 7.1 as WAL (write-ahead log).\n> \n> What does this exactly mean? Will this WAL be an equivalent of the logical\n> log of Informix? Where will they be kept? In the database, as a file, or as\n> either?\n> Looks pretty good. All that is needed after this is a good blob\n> implementation (for text and binary data).\n\nTOAST. Coming for 7.1.\n\nAnd we thought 7.0 had alot of new features relative to 6.5.....\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Mon, 09 Oct 2000 07:32:03 -0400",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: backup and restore"
},
{
"msg_contents": "> \"Martin A. Marques\" wrote:\n> > \n> > On Sun, 08 Oct 2000, Bruce Momjian wrote:\n> > > This will be in 7.1 as WAL (write-ahead log).\n> > \n> > What does this exactly mean? Will this WAL be an equivalent of the logical\n> > log of Informix? Where will they be kept? In the database, as a file, or as\n> > either?\n> > Looks pretty good. All that is needed after this is a good blob\n> > implementation (for text and binary data).\n> \n> TOAST. Coming for 7.1.\n> \n> And we thought 7.0 had alot of new features relative to 6.5.....\n\nI think the true feature-killer release was 6.5.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 9 Oct 2000 14:32:36 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: backup and restore"
},
{
"msg_contents": "> On Sun, 08 Oct 2000, Bruce Momjian wrote:\n> > This will be in 7.1 as WAL (write-ahead log).\n> \n> What does this exactly mean? Will this WAL be an equivalent of the logical \n> log of Informix? Where will they be kept? In the database, as a file, or as \n> either?\n> Looks pretty good. All that is needed after this is a good blob \n> implementation (for text and binary data).\n\nYes, I think it will be similar to Informix logical logs. Not sure\nwhere they will be kept, but assume you can put it anywhere you want. \nNot sure if tape logging will be enabled. It will be written as a file.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 9 Oct 2000 14:33:58 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: backup and restore"
},
{
"msg_contents": "Bruce Momjian wrote:\n> Lamar Owen Wrote:\n> > And we thought 7.0 had alot of new features relative to 6.5.....\n \n> I think the true feature-killer release was 6.5.\n\nWell, this is true. I skipped the whole 6.4 series for a reason :-). I\njust _knew_ 6.5 was going to be _it_... (in actuality, it's because\nRPM's for 6.4 weren't available until 6.5 was nearly ready to\nrelease...).\n\nIt was feature-shock to go from 6.3.2 to 6.5.... But the performance\ndifference was the real kicker, for multiuser concurrent processing.\n\nBut that changelog entry for 6.5.3->7.0 was _huge_. And there were\nsignificant changes to many files, not just a few -- the RPMset\ndifferences were substantial from the building/packaging side, the side\nI deal with.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Mon, 09 Oct 2000 15:24:50 -0400",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: backup and restore"
}
] |
[
{
"msg_contents": "This may seem a little silly, but I'd like to request a feature to\nimplement timed triggers.\n\nWith the number of cron jobs that run a perl script which runs a simple\nSELECT function(); that does various nightly cleanup or maintenance\n(billing run) it would be much nicer to have an actual trigger run every\n12 hours rather than be initiated by a cron job.\n\nLess places things could go wrong. Could be requested to run only when\na table isn't being used at the time (low priority maintenance). The\nsame thing could be implemented for a periodic vacuum.\n\nI'd like to take this up myself, but will need to learn alot about the\nway Postgres works first so if someone else sees the need as well please\nfeel free to implement (It'll take me atleast 6 months before I feel\ncomfortable with the results).\n\n-- \nRod Taylor\n--\n",
"msg_date": "Mon, 02 Oct 2000 09:26:23 -0400",
"msg_from": "Rod Taylor <[email protected]>",
"msg_from_op": true,
"msg_subject": "Timed Triggers?"
},
{
"msg_contents": "Rod Taylor <[email protected]> writes:\n> With the number of cron jobs that run a perl script which runs a simple\n> SELECT function(); that does various nightly cleanup or maintenance\n> (billing run) it would be much nicer to have an actual trigger run every\n> 12 hours rather than be initiated by a cron job.\n\nEr ... why? Seems like what you're describing is implementing the guts\nof cron in the postmaster. I don't see a real good reason to re-invent\nthe wheel like that.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 09 Oct 2000 18:33:33 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Timed Triggers? "
},
{
"msg_contents": "Tom Lane wrote:\n> \n> Rod Taylor <[email protected]> writes:\n> > With the number of cron jobs that run a perl script which runs a simple\n> > SELECT function(); that does various nightly cleanup or maintenance\n> > (billing run) it would be much nicer to have an actual trigger run every\n> > 12 hours rather than be initiated by a cron job.\n> \n> Er ... why? Seems like what you're describing is implementing the guts\n> of cron in the postmaster. I don't see a real good reason to re-invent\n> the wheel like that.\n\nMy current concern is the reliance on cron, perl, and the postmaster\nrunning properly in order to accomplish a job. When one fails, they\noperation fails. If the postmaster is down, how does cron know that\nsomething needs to be re-run? In the postmaster we could issue a\ntrigger of type 'once every 4 hours, and on startup' to ensure that the\nbilling run (or whatever) is executed properly as early as possibly. At\nthe moment it takes quite a bit of code outside of the database to\naccomplish what would be a trivial setting for the creator of the\ntrigger to tie to it.\n\nYour right though, it's too bad cron wasn't more library oriented so\nthat it could be used within programs rather than having to be\nre-implemented.\n\n-- \nRod Taylor\n\n\"People get annoyed when you try to debug them.\" -- Larry Wall, 2nd\nState of the Onion.\n\n\"A + B + C = Success if, A = Hard Work, B = Hard Play, C = Keeping your\nmouth shut.\" -- Albert Einstein\n\n--\n",
"msg_date": "Mon, 09 Oct 2000 19:02:16 -0400",
"msg_from": "Rod Taylor <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Timed Triggers?"
}
] |
[
{
"msg_contents": "Fresh from /var/log/postgres.log:\n\npostgres localhost www-data auction SELECT: ../iconv/skeleton.c:297: gconv: Asse\nrtion `outbufstart == ((void *)0)' failed.\nServer process (pid 27393) exited with status 6 at Mon Oct 2 09:01:43 2000\n\nStarting a new postmaster seems to cure the problem.\n\n-- \nLouis-David Mitterrand - [email protected] - http://www.apartia.org\n\nBill Gates to his broker: \"You idiot, I said $150 million on **SNAPPLE**!!!\"\n",
"msg_date": "Mon, 2 Oct 2000 16:03:04 +0200",
"msg_from": "Louis-David Mitterrand <[email protected]>",
"msg_from_op": true,
"msg_subject": "failed assertion error on PG-7.0.2"
},
{
"msg_contents": "Louis-David Mitterrand <[email protected]> writes:\n> Fresh from /var/log/postgres.log:\n> postgres localhost www-data auction SELECT: ../iconv/skeleton.c:297: gconv: Assertion `outbufstart == ((void *)0)' failed.\n\n?? Anyone recognize that?\n\nA few seconds with glimpse shows that there is no file named skeleton.c,\nindeed no directory named iconv, in the current sources; much less any\nroutine named gconv; nor any variable named outbufstart.\nSo I'm pretty confused... \n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 02 Oct 2000 11:01:12 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: failed assertion error on PG-7.0.2 "
},
{
"msg_contents": "Tom Lane <[email protected]> writes:\n\n> Louis-David Mitterrand <[email protected]> writes:\n> > Fresh from /var/log/postgres.log:\n> > postgres localhost www-data auction SELECT: ../iconv/skeleton.c:297: gconv: Assertion `outbufstart == ((void *)0)' failed.\n> \n> ?? Anyone recognize that?\n> \n> A few seconds with glimpse shows that there is no file named skeleton.c,\n> indeed no directory named iconv, in the current sources; much less any\n> routine named gconv; nor any variable named outbufstart.\n> So I'm pretty confused... \n\nglibc, related to i18n.\n\n-- \nTrond Eivind Glomsr�d\nRed Hat, Inc.\n",
"msg_date": "02 Oct 2000 11:35:18 -0400",
"msg_from": "[email protected] (Trond Eivind =?iso-8859-1?q?Glomsr=F8d?=)",
"msg_from_op": false,
"msg_subject": "Re: failed assertion error on PG-7.0.2"
}
] |
[
{
"msg_contents": "Hi,\n\nWhen two tables (table2 and table3) have foreign keys\nreferring to a common table(table1), I am unable to\nhave 2 concurrent transactions - one performing insert\non table1 and the other on table2, when the records\nbeing inserted have the same foreign key. \n\nIf I use JDBC, one of the transactions aborts.\nIf I open 2 psql sessions and try the same, one just\nwaits and does not show the prompt until the other\ntransaction has been committed or aborted.\n\nFor example,\ncreate table tmp1(idx int4, data int4);\ncreate table tmp2(idx2 int4, col2 int4, constraint\ntmpcon2 foreign key(col2) references tmp1(idx));\ncreate table tmp3(idx3 int4, col3 int4, constraint\ntmpcon3 foreign key(col3) references tmp1(idx));\ninsert into tmp1 values(1, 1);\n\nTransaction 1 :\nbegin work;\ninsert into tmp2 values(2, 1);\n\nTransaction2 :\nbegin work;\ninsert into tmp3 values(3,1);\n\nSince such transactions are common for me, for the\ntime-being I have dropped the foreign key constraint.\nAny ideas ?\n\nRini\n\n__________________________________________________\nDo You Yahoo!?\nYahoo! Photos - 35mm Quality Prints, Now Get 15 Free!\nhttp://photos.yahoo.com/\n",
"msg_date": "Mon, 2 Oct 2000 08:39:51 -0700 (PDT)",
"msg_from": "Rini Dutta <[email protected]>",
"msg_from_op": true,
"msg_subject": "foreign key introduces unnecessary locking ?"
},
{
"msg_contents": "On Mon, 2 Oct 2000, Rini Dutta wrote:\n\n> When two tables (table2 and table3) have foreign keys\n> referring to a common table(table1), I am unable to\n> have 2 concurrent transactions - one performing insert\n> on table1 and the other on table2, when the records\n> being inserted have the same foreign key. \n> \n> If I use JDBC, one of the transactions aborts.\n> If I open 2 psql sessions and try the same, one just\n> waits and does not show the prompt until the other\n> transaction has been committed or aborted.\n> \n> For example,\n> create table tmp1(idx int4, data int4);\n> create table tmp2(idx2 int4, col2 int4, constraint\n> tmpcon2 foreign key(col2) references tmp1(idx));\n> create table tmp3(idx3 int4, col3 int4, constraint\n> tmpcon3 foreign key(col3) references tmp1(idx));\n> insert into tmp1 values(1, 1);\n> \n> Transaction 1 :\n> begin work;\n> insert into tmp2 values(2, 1);\n> \n> Transaction2 :\n> begin work;\n> insert into tmp3 values(3,1);\n> \n> Since such transactions are common for me, for the\n> time-being I have dropped the foreign key constraint.\n> Any ideas ?\n\nEach is attempting to grab row locks on tmp1 to prevent \nthe rows from going away while we're testing the \nreferences. The second transaction is waiting for the row \nlock to go away so that it can do its row lock.\nI'm not sure why its failing in JDBC though.\n\n\n",
"msg_date": "Mon, 9 Oct 2000 16:27:32 -0700 (PDT)",
"msg_from": "Stephan Szabo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: foreign key introduces unnecessary locking ?"
}
] |
[
{
"msg_contents": "I've been looking at UNION/INTERSECT/EXCEPT with an eye to making them\nwork in views and subselect-in-FROM (same thing really ;-)). I had first\nthought that some marginal hacking on the parsetree representation might\nbe enough, but after study I am realizing just how broken this code really\nis. It turns out that it's not even very close to implementing the SQL\nspec. SQL92 7.10, general rule 1b says that if a row R has m duplicates\nin T1 and n duplicates in T2 (m >= 0, n >= 0) then:\n\n\tSet operation:\t\tcontains this many duplicates of R:\n\n\tT1 UNION T2\t\t1 if m > 0 or n > 0, else 0\n\n\tT1 INTERSECT T2\t\t1 if m > 0 and n > 0, else 0\n\n\tT1 EXCEPT T2\t\t1 if m > 0 and n = 0, else 0\n\n\tT1 UNION ALL T2\t\tm + n\n\n\tT1 INTERSECT ALL T2\tmin(m, n)\n\n\tT1 EXCEPT ALL T2\tmax(m - n, 0)\n\nWe are OK for UNION (which we do as append, sort, unique) and for\nUNION ALL (which we do as append). We are not OK for INTERSECT\nand EXCEPT, which the code presently tries to do as\n\nSELECT T1 INTERSECT SELECT T2 => SELECT T1 WHERE T1 IN (SELECT T2)\n\nSELECT T1 EXCEPT SELECT T2 => SELECT T1 WHERE T1 NOT IN (SELECT T2)\n\nThis will give the wrong number of duplicates when m > 1. It could be\nmade to give the right answer by adding a DISTINCT to the select, but\nthere's still no expansion path for implementing INTERSECT ALL and EXCEPT\nALL.\n\nThere are a bunch of internal problems too (which is why views didn't\nsupport UNION et al to begin with), mostly due to bad choices of data\nstructures. I have come to the conclusion that there's no point in\nhalf measures: throwing out this code and rewriting from scratch will\ntake less time than trying to patch it.\n\nHere's what I have in mind:\n\nParse-tree data structure (output of gram.y): remove\nunionClause/intersectClause from SelectStmt. Add a SetOperation node type\nthat indicates the operation type (UNION/INTERSECT/EXCEPT plus an \"all\"\nboolean) and links to two component SelectStmts or SetOperations. There\nwill be no loops in this data structure, unlike the present situation.\nThe top-level info (optional ORDER BY) added by the SelectStmt production\nwill be attached to the leftmost SelectStmt leaf in the tree.\n\nQuery-tree structure (output of parse_analyze): process the leaf\nSelectStmts into Queries individually, after removing the OrderBy etc\ninfo from the leftmost leaf. Create a top-level Query that contains the\nleaf Queries as subselects-in-FROM. In place of\nunionClause/intersectClause, Query nodes will have a Node *setOperations\nfield. In the top-level Query this will contain the SetOperation tree\nemitted by gram.y, but with the leaf SelectStmt nodes replaced by\nRangeTblRef references to the range table entries occupied by the leaf\nQueries. (Thus, still no loops or multiple links.) The top-level Query\nhas a dummy targetlist that exists mainly to show the union'd datatype\nof each output column, and it carries any sortClause, limitClause, etc\nneeded for the output of the entire operation. Note that we do not need\nto transform the datatypes of the leaf queries' targetlists, which\neliminates a large class of bugs that exist presently with\ncross-datatype UNIONs.\n\nRewriter: need pay no attention at all to setOperation tree.\n\nPlan-tree structure (output of planner): UNION/UNION ALL are handled\nsame as now, except that what we are appending together is not directly\nthe top-level plan of each leaf query, but a SubqueryScan plan scanning\nthe output of each leaf query. This gives us one extra level of\nprojection (targetlist evaluation) in which to put conversions to the\ncommon union datatype --- without breaking the semantics of GROUP BY,\nDISTINCT, and so forth in the leaf queries. INTERSECT and EXCEPT will\nbe handled by building SubqueryScan plans that emit the common-datatype\ncolumns plus a resjunk boolean column that shows whether the tuple is\ncoming from the left or right input. The outputs of these plans are\nthen appended together, sorted, and fed to a new plan node type that\nimplements INTERSECT(ALL) and EXCEPT(ALL). It will be a simple\ngeneralization of the Unique plan type: scan a set of successive tuples\nthat agree in all the non-resjunk columns, counting the number of tuples\nthat came from left and right sides. This gives us 'm' and 'n' for each\nset, from which the spec-defined behavior can be implemented immediately.\n\nExecutor: just need new plan-node-type implementation, which is easily\nderived from nodeUnique.\n\nI am not planning to try to implement SQL's \"CORRESPONDING\" option for\n7.1. Whenever we do get to it, it should be a fairly straightforward\nextension: add the corresponding-column lists to SetOperation nodes,\nand then in the plan tree, make the SubqueryScan nodes emit just these\ncolumns and not the entire targetlists of the leaf Queries. (This is\nanother reason why we need the extra level of targetlist...)\n\nComments?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 02 Oct 2000 13:04:29 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": true,
"msg_subject": "Reimplementing UNION/INTERSECT/EXCEPT"
},
{
"msg_contents": "> Plan-tree structure (output of planner): UNION/UNION ALL are handled\n> same as now, except that what we are appending together is not directly\n> the top-level plan of each leaf query, but a SubqueryScan plan scanning\n> the output of each leaf query. This gives us one extra level of\n> projection (targetlist evaluation) in which to put conversions to the\n> common union datatype --- without breaking the semantics of GROUP BY,\n> DISTINCT, and so forth in the leaf queries. INTERSECT and EXCEPT will\n> be handled by building SubqueryScan plans that emit the common-datatype\n> columns plus a resjunk boolean column that shows whether the tuple is\n> coming from the left or right input. The outputs of these plans are\n> then appended together, sorted, and fed to a new plan node type that\n> implements INTERSECT(ALL) and EXCEPT(ALL). It will be a simple\n> generalization of the Unique plan type: scan a set of successive tuples\n> that agree in all the non-resjunk columns, counting the number of tuples\n> that came from left and right sides. This gives us 'm' and 'n' for each\n> set, from which the spec-defined behavior can be implemented immediately.\n\nSounds good. We have always done these by applying our existing tools. \nNow that we have a better tool (subselects in FROM), we have a better\nsolution.\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 00:50:53 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Reimplementing UNION/INTERSECT/EXCEPT"
}
] |
[
{
"msg_contents": "Has anybody been getting pgsql-committers messages the last few days?\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Mon, 2 Oct 2000 21:39:46 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "What's happening with pgsql-committers?"
},
{
"msg_contents": "On Mon, 2 Oct 2000, Peter Eisentraut wrote:\n\n> Has anybody been getting pgsql-committers messages the last few days?\n\nI have.\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Mon, 2 Oct 2000 16:42:01 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: What's happening with pgsql-committers?"
},
{
"msg_contents": "Hi Peter,\n\nI'm getting COMMITTERS OK but last week I noticed I wasn't getting HACKERS\nany more.\n\nTry sending a \"show [email protected]\" in the body of a message to\[email protected] to see if you're still subscribed.\n\nKeith.\n\n--\n",
"msg_date": "Mon, 2 Oct 2000 22:32:07 +0100",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: What's happening with pgsql-committers?"
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> Has anybody been getting pgsql-committers messages the last few days?\n\nComing through fine for me (at least when hub.org isn't wedged\ncompletely, which it was several times over the weekend...)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 03 Oct 2000 00:20:42 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: What's happening with pgsql-committers? "
},
{
"msg_contents": "On Tue, 3 Oct 2000, Tom Lane wrote:\n\n> Peter Eisentraut <[email protected]> writes:\n> > Has anybody been getting pgsql-committers messages the last few days?\n> \n> Coming through fine for me (at least when hub.org isn't wedged\n> completely, which it was several times over the weekend...)\n\nWe hit a corruption of -general again this past week (I've really gotta\ndive into getting the postgresql backend working for majordomo instead of\nBerkeley DB :( ) ... it was causing a perl process of 900+Meg to run,\nwhich caused the load avg to climb unreasonably :( \n\n",
"msg_date": "Sat, 7 Oct 2000 22:13:02 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: What's happening with pgsql-committers? "
},
{
"msg_contents": "On Sat, 7 Oct 2000, The Hermit Hacker wrote:\n\n> On Tue, 3 Oct 2000, Tom Lane wrote:\n> \n> > Peter Eisentraut <[email protected]> writes:\n> > > Has anybody been getting pgsql-committers messages the last few days?\n> > \n> > Coming through fine for me (at least when hub.org isn't wedged\n> > completely, which it was several times over the weekend...)\n> \n> We hit a corruption of -general again this past week (I've really gotta\n> dive into getting the postgresql backend working for majordomo instead of\n> Berkeley DB :( ) ... it was causing a perl process of 900+Meg to run,\n> which caused the load avg to climb unreasonably :( \nOooo, ooo, how about instead writing a ndbm/gdbm/whateverdb wrapper for\npostgres? \n\nConfigurable by some file in /etc, where it would establish correspondence\nbetween \"file\" argument to dbopen() and a postgres view. This would rock\nfor converting applications which only support dbm mappings (such as\nsendmail) to use postgres transparently.\n\nThe only problem I see with it is db->fd() function which returns a\nfiledescriptor for the file, which some apps rely on to do flock/fcntl. On\nother hand, locking of that kind will be unnecessary for postgres, so it\ncan return fd of some file in /tmp...\n\nThere's an attempt to do this at http://www.is.kiruna.se/~goran/ldap/arkiv/\nbut its very raw and not configurable. \n\n-alex\n\n\n",
"msg_date": "Sat, 7 Oct 2000 21:42:54 -0400 (EDT)",
"msg_from": "Alex Pilosov <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: What's happening with pgsql-committers? "
}
] |
[
{
"msg_contents": "\ni need to query some oracle tables from a freebsd system.\n\nis there a lightweight method to do this, or do i have no choice but to\nput in the Oracle Linux stuff and use their API's?\n\n-- \n[ Jim Mercer [email protected] +1 416 410-5633 ]\n[ Reptilian Research -- Longer Life through Colder Blood ]\n[ Don't be fooled by cheap Finnish imitations; BSD is the One True Code. ]\n",
"msg_date": "Mon, 2 Oct 2000 15:43:59 -0400",
"msg_from": "Jim Mercer <[email protected]>",
"msg_from_op": true,
"msg_subject": "off-topic: (sorta) freebsd -> oracle, lightweight"
},
{
"msg_contents": "At 03:43 PM 10/2/00 -0400, Jim Mercer wrote:\n>\n>i need to query some oracle tables from a freebsd system.\n>\n>is there a lightweight method to do this, or do i have no choice but to\n>put in the Oracle Linux stuff and use their API's?\n\nOracle knows ODBC, no?\n\n\n\n- Don Baccus, Portland OR <[email protected]>\n Nature photos, on-line guides, Pacific Northwest\n Rare Bird Alert Service and other goodies at\n http://donb.photo.net.\n",
"msg_date": "Mon, 02 Oct 2000 12:57:50 -0700",
"msg_from": "Don Baccus <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: off-topic: (sorta) freebsd -> oracle, lightweight"
},
{
"msg_contents": "I think it's against the Oracle license to run it under any kind of\nemulation (which is what you would have to do with FreeBSD, run it under\nLinux emulation).. All that's void if they support FreeBSD natively now\n(which I don't think they do)..\n\nWouldn't this be a better question for an Oracle list since this has nothing\nto do with PostgreSQL? (Just a friendly suggestion) :-)\n\nGood luck!!\n\n-Mitch\n\n----- Original Message -----\nFrom: \"Jim Mercer\" <[email protected]>\nTo: <[email protected]>\nSent: Monday, October 02, 2000 12:43 PM\nSubject: [HACKERS] off-topic: (sorta) freebsd -> oracle, lightweight\n\n\n>\n> i need to query some oracle tables from a freebsd system.\n>\n> is there a lightweight method to do this, or do i have no choice but to\n> put in the Oracle Linux stuff and use their API's?\n>\n> --\n> [ Jim Mercer [email protected] +1 416\n410-5633 ]\n> [ Reptilian Research -- Longer Life through Colder\n ]\n> [ Don't be fooled by cheap Finnish imitations; BSD is the One True\node. ]\n>\n\n",
"msg_date": "Mon, 2 Oct 2000 13:15:25 -0700",
"msg_from": "\"Mitch Vincent\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: off-topic: (sorta) freebsd -> oracle, lightweight"
},
{
"msg_contents": "I have absolutely no idea how well it works, but I think Perl/DBI has a\n\"pass thru\" module that can do this sort of thing...\n\nnot sure if that helps much...\n\n-philip\n\nOn Mon, 2 Oct 2000, Jim Mercer wrote:\n\n> \n> i need to query some oracle tables from a freebsd system.\n> \n> is there a lightweight method to do this, or do i have no choice but to\n> put in the Oracle Linux stuff and use their API's?\n> \n> -- \n> [ Jim Mercer [email protected] +1 416 410-5633 ]\n> [ Reptilian Research -- Longer Life through Colder Blood ]\n> [ Don't be fooled by cheap Finnish imitations; BSD is the One True Code. ]\n> \n\n",
"msg_date": "Mon, 2 Oct 2000 13:35:26 -0700 (PDT)",
"msg_from": "Philip Hallstrom <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: off-topic: (sorta) freebsd -> oracle, lightweight"
},
{
"msg_contents": "The linux version of Oracle works pretty good on FreeBSD (not suggested\nif it's going to be a heavy hit database box), then again, I don't\nsuggest the Linux version of Oracle for that anyway.\n\nLinux compat from -stable is required.\n\nSend your emails off to Oracle requesting the FreeBSD native version (I\nunderstand theres one in creation) and in a few months (year?) we'll all\nbe happy. Problem (or solution) is that by that time Postgres will fill\nall my requirements so I probably wouldn't switch until load or brand\nname becomes a serious concern.\n\nFabrice Scemama wrote:\n> \n> He can use DBD::Proxy over DBI as he can use ODBC through\n> network; but at the end, he still has to have a box where\n> he can install oracle drivers (which are binaries, and\n> requier you to launch oracle java installer, as if you\n> were going to install the whole beast, then to choose\n> to only install a client machine).\n> \n> Well, Jim, you've got plenty of choices: NT, Linux,\n> Solaris, AIX, Sco... maybe a few others. But the only\n> free OS seems to be Linux. Maybe you should take a\n> cheap intel PC and install linux + oracle client library\n> on it.\n> \n> Regards\n> Fabrice Scemama\n> \n> Philip Hallstrom wrote:\n> >\n> > I have absolutely no idea how well it works, but I think Perl/DBI has a\n> > \"pass thru\" module that can do this sort of thing...\n> >\n> > not sure if that helps much...\n> >\n> > -philip\n> >\n> > On Mon, 2 Oct 2000, Jim Mercer wrote:\n> >\n> > >\n> > > i need to query some oracle tables from a freebsd system.\n> > >\n> > > is there a lightweight method to do this, or do i have no choice but to\n> > > put in the Oracle Linux stuff and use their API's?\n> > >\n> > > --\n> > > [ Jim Mercer [email protected] +1 416 410-5633 ]\n> > > [ Reptilian Research -- Longer Life through Colder Blood ]\n> > > [ Don't be fooled by cheap Finnish imitations; BSD is the One True Code. ]\n> > >\n> \n> --\n> \"La veritable et sincere amitie' verbale profondement superficielle\n> est celle sur laquelle on peut absolument compter quand on n'a\n> strictement besoin de rien.\" -- Pierre Dac\n\n-- \nRod Taylor\n\n\"People get annoyed when you try to debug them.\" -- Larry Wall, 2nd\nState of the Onion.\n\n\"A + B + C = Success if, A = Hard Work, B = Hard Play, C = Keeping your\nmouth shut.\" -- Albert Einstein\n\n--\n",
"msg_date": "Tue, 03 Oct 2000 06:55:28 -0400",
"msg_from": "Rod Taylor <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: off-topic: (sorta) freebsd -> oracle, lightweight"
},
{
"msg_contents": "He can use DBD::Proxy over DBI as he can use ODBC through\nnetwork; but at the end, he still has to have a box where\nhe can install oracle drivers (which are binaries, and\nrequier you to launch oracle java installer, as if you\nwere going to install the whole beast, then to choose\nto only install a client machine).\n\nWell, Jim, you've got plenty of choices: NT, Linux,\nSolaris, AIX, Sco... maybe a few others. But the only\nfree OS seems to be Linux. Maybe you should take a\ncheap intel PC and install linux + oracle client library\non it.\n\nRegards\nFabrice Scemama\n\nPhilip Hallstrom wrote:\n> \n> I have absolutely no idea how well it works, but I think Perl/DBI has a\n> \"pass thru\" module that can do this sort of thing...\n> \n> not sure if that helps much...\n> \n> -philip\n> \n> On Mon, 2 Oct 2000, Jim Mercer wrote:\n> \n> >\n> > i need to query some oracle tables from a freebsd system.\n> >\n> > is there a lightweight method to do this, or do i have no choice but to\n> > put in the Oracle Linux stuff and use their API's?\n> >\n> > --\n> > [ Jim Mercer [email protected] +1 416 410-5633 ]\n> > [ Reptilian Research -- Longer Life through Colder Blood ]\n> > [ Don't be fooled by cheap Finnish imitations; BSD is the One True Code. ]\n> >\n\n-- \n\"La veritable et sincere amitie' verbale profondement superficielle\nest celle sur laquelle on peut absolument compter quand on n'a\nstrictement besoin de rien.\" -- Pierre Dac\n",
"msg_date": "Tue, 03 Oct 2000 14:50:47 +0200",
"msg_from": "Fabrice Scemama <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: off-topic: (sorta) freebsd -> oracle, lightweight"
}
] |
[
{
"msg_contents": "I have noticed that some operating system distributors ship custom-patched\nversions of Autoconf. That is pretty brain-dead because some of the\nchanges make things work worse on other systems.\n\nTherefore I would recommend that everybody install the original\nautoconf-2.13.tar.gz from GNU and not use anything that comes with their\noperating system. That includes the installation on hub.org.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Mon, 2 Oct 2000 22:26:53 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Autoconf version discrepancies"
},
{
"msg_contents": "I always telnet into hub to run autoconf. I have a little script in my\n~momjian/bin directory called pgautoconf that does that, and cvs commits\nthe changes.\n\n\n> I have noticed that some operating system distributors ship custom-patched\n> versions of Autoconf. That is pretty brain-dead because some of the\n> changes make things work worse on other systems.\n> \n> Therefore I would recommend that everybody install the original\n> autoconf-2.13.tar.gz from GNU and not use anything that comes with their\n> operating system. That includes the installation on hub.org.\n> \n> -- \n> Peter Eisentraut [email protected] http://yi.org/peter-e/\n> \n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 2 Oct 2000 23:46:43 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Autoconf version discrepancies"
},
{
"msg_contents": "Bruce Momjian writes:\n\n> I always telnet into hub to run autoconf. I have a little script in my\n> ~momjian/bin directory called pgautoconf that does that, and cvs commits\n> the changes.\n\nThat's exactly the problem, the version on hub.org is custom-patched for\nFreeBSD and has significant loser potential on other platforms.\n\nAnd really, there's no need to do that either, if you install the original\nGNU tarball locally.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Tue, 3 Oct 2000 14:05:46 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Autoconf version discrepancies"
},
{
"msg_contents": "On Tue, 3 Oct 2000, Peter Eisentraut wrote:\n\n> Bruce Momjian writes:\n> \n> > I always telnet into hub to run autoconf. I have a little script in my\n> > ~momjian/bin directory called pgautoconf that does that, and cvs commits\n> > the changes.\n> \n> That's exactly the problem, the version on hub.org is custom-patched for\n> FreeBSD and has significant loser potential on other platforms.\n> \n> And really, there's no need to do that either, if you install the original\n> GNU tarball locally.\n\nAn upgrade must have replaced it then 'cuze I built 2.13 from original\nsources back in April '99 and Marc installed it on hub.\n\nMarc it's in my autoconf directory if you want to reinstall it. I just\nremade it so it's current to this version of the OS.\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Tue, 3 Oct 2000 09:19:10 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Autoconf version discrepancies"
},
{
"msg_contents": "> Bruce Momjian writes:\n> \n> > I always telnet into hub to run autoconf. I have a little script in my\n> > ~momjian/bin directory called pgautoconf that does that, and cvs commits\n> > the changes.\n> \n> That's exactly the problem, the version on hub.org is custom-patched for\n> FreeBSD and has significant loser potential on other platforms.\n> \n> And really, there's no need to do that either, if you install the original\n> GNU tarball locally.\n\nGood point. I think autoconf is run as part of the tarball creation, so\nwe had better have the right version on hub. Personally, I prefer to\nuse a version on hub because it is the official one for PostgreSQL. \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 3 Oct 2000 10:46:24 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Autoconf version discrepancies"
},
{
"msg_contents": "On Tue, 3 Oct 2000, Peter Eisentraut wrote:\n\n> Bruce Momjian writes:\n> \n> > I always telnet into hub to run autoconf. I have a little script in my\n> > ~momjian/bin directory called pgautoconf that does that, and cvs commits\n> > the changes.\n> \n> That's exactly the problem, the version on hub.org is custom-patched for\n> FreeBSD and has significant loser potential on other platforms.\n\nOkay, autoconf on hub.org is based on what is in ports ... the only\n\"custom patches\" are that which are in /usr/ports/devel/autoconf/patches,\nand I just went through them and there doesn't look like anything *odd* in\nthere ... can you look at those patches and tell me which one is screwing\nthings up for you, that I'm not seeing? :(\n\n\n",
"msg_date": "Sat, 7 Oct 2000 22:16:19 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Autoconf version discrepancies"
},
{
"msg_contents": "The Hermit Hacker writes:\n\n> Okay, autoconf on hub.org is based on what is in ports ... the only\n> \"custom patches\" are that which are in /usr/ports/devel/autoconf/patches,\n> and I just went through them and there doesn't look like anything *odd* in\n> there ... can you look at those patches and tell me which one is screwing\n> things up for you, that I'm not seeing? :(\n\nThe patches ad, ae, and af will cause configure to fail on machines\nwithout mktemp. It's not like things get \"screwed up\" for me, but the\npoint of Autoconf is portability to *all* machines, so FreeBSD-specific\nchanges/optimizations(?) seem misplaced.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Sun, 8 Oct 2000 21:51:37 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Autoconf version discrepancies"
},
{
"msg_contents": "> The Hermit Hacker writes:\n> \n> > Okay, autoconf on hub.org is based on what is in ports ... the only\n> > \"custom patches\" are that which are in /usr/ports/devel/autoconf/patches,\n> > and I just went through them and there doesn't look like anything *odd* in\n> > there ... can you look at those patches and tell me which one is screwing\n> > things up for you, that I'm not seeing? :(\n> \n> The patches ad, ae, and af will cause configure to fail on machines\n> without mktemp. It's not like things get \"screwed up\" for me, but the\n> point of Autoconf is portability to *all* machines, so FreeBSD-specific\n> changes/optimizations(?) seem misplaced.\n\nAre there any platforms that do not have mktemp? Hard to imagine.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 21:18:23 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Autoconf version discrepancies"
},
{
"msg_contents": "Bruce Momjian <[email protected]> writes:\n>> The patches ad, ae, and af will cause configure to fail on machines\n>> without mktemp. It's not like things get \"screwed up\" for me, but the\n>> point of Autoconf is portability to *all* machines, so FreeBSD-specific\n>> changes/optimizations(?) seem misplaced.\n\n> Are there any platforms that do not have mktemp? Hard to imagine.\n\nNot hard at all, considering that mktemp(1) is defined by no standard\naccording to the references I have handy.\n\nMore generally, I have to side with Peter on this: local patches to\nAutoconf are a fine example of Missing The Point. The output script\nhas to run everywhere, not only on your own platform.\n\nAlso, we not long ago went through the exercise of making sure that all\ncommitters were standardized on the same version of Autoconf, ie, 2.13.\nNow it emerges that hub.org is running a NON STANDARD version of\nAutoconf: 2.13 + unspecified BSD-originated hacks. So the output is\nlikely to change depending on who committed last and where they did it\nfrom.\n\nThis will not do. Please fix hub's copy of Autoconf.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 08 Oct 2000 22:20:05 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Autoconf version discrepancies "
},
{
"msg_contents": "On Sun, 8 Oct 2000, Bruce Momjian wrote:\n\n> Are there any platforms that do not have mktemp? Hard to imagine.\n\nmktemp(1) or mktemp(3)?\n\nThe latter is pretty much universal (and dangerous too).\n\nThe former is, AFAICS, available only on some Linux and BSD.\nBut it's under the BSD licence, and is not a long program.\nFor portability's sake, it may be easier just to include it\nin the distribution.\n\nMatthew.\n\n",
"msg_date": "Mon, 9 Oct 2000 12:45:12 +0100 (BST)",
"msg_from": "Matthew Kirkwood <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Autoconf version discrepancies"
},
{
"msg_contents": "On Sun, 8 Oct 2000, Peter Eisentraut wrote:\n\n> The Hermit Hacker writes:\n> \n> > Okay, autoconf on hub.org is based on what is in ports ... the only\n> > \"custom patches\" are that which are in /usr/ports/devel/autoconf/patches,\n> > and I just went through them and there doesn't look like anything *odd* in\n> > there ... can you look at those patches and tell me which one is screwing\n> > things up for you, that I'm not seeing? :(\n> \n> The patches ad, ae, and af will cause configure to fail on machines\n> without mktemp. It's not like things get \"screwed up\" for me, but the\n> point of Autoconf is portability to *all* machines, so FreeBSD-specific\n> changes/optimizations(?) seem misplaced.\n\nokay, I just looked through those patches, and all they are doing is\nchanging the use of $$ to use the process id as a temp file name to using\n'mktemp' ... have we had any reports of this breaking anything on any of\nthe platforms we support? If so, then I agree this is a bad thing, if\nnot, I'm curious as to why this has suddenly become a problem since we've\nbeen using this autoconf since March 8:\n\n-rwxr-xr-x 1 root wheel 4987 Mar 8 2000 /usr/local/bin/autoconf\n\n\n\n",
"msg_date": "Mon, 9 Oct 2000 16:04:33 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Autoconf version discrepancies"
},
{
"msg_contents": "On Sun, 8 Oct 2000, Tom Lane wrote:\n\n> Bruce Momjian <[email protected]> writes:\n> >> The patches ad, ae, and af will cause configure to fail on machines\n> >> without mktemp. It's not like things get \"screwed up\" for me, but the\n> >> point of Autoconf is portability to *all* machines, so FreeBSD-specific\n> >> changes/optimizations(?) seem misplaced.\n> \n> > Are there any platforms that do not have mktemp? Hard to imagine.\n> \n> Not hard at all, considering that mktemp(1) is defined by no standard\n> according to the references I have handy.\n> \n> More generally, I have to side with Peter on this: local patches to\n> Autoconf are a fine example of Missing The Point. The output script\n> has to run everywhere, not only on your own platform.\n> \n> Also, we not long ago went through the exercise of making sure that all\n> committers were standardized on the same version of Autoconf, ie, 2.13.\n> Now it emerges that hub.org is running a NON STANDARD version of\n> Autoconf: 2.13 + unspecified BSD-originated hacks. So the output is\n> likely to change depending on who committed last and where they did it\n> from.\n\nTake a look at the patches that I pointed Peter at for these \"Unspecified\"\npatches. The ones that Peter points out merely change from using .$$ to\nstipulate the temp file name to using 'mktemp' ...\n\nNow, what I'd be curious about is what platforms this does break things\non, cause if it does, it kinda makes using FreeBSD's autoconf difficult\nfor non-FreeBSD developments, about as difficult as some of the\nLinux-centric ones that we try to port to FreeBSD ...\n\nBasically, what benefits are there to using 'mktemp' over using the shell\nbased '.$$', and do those developing software under *BSD then break\nthemselves for other platforms as a result (or force other platforms to\nrun autoconf to build)?\n\nIf using mktemp doesn't break any platform, this is a moot point ... if it\ndoes, then I think it is something that *has* to be fix in the FreeBSD\nport itself so that it doesn't make us look FreeBSD-centric in our\ndevelopment efforts on any other package ...\n\n\n\n\n",
"msg_date": "Mon, 9 Oct 2000 16:11:20 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Autoconf version discrepancies "
},
{
"msg_contents": "On Mon, Oct 09, 2000 at 04:11:20PM -0300, The Hermit Hacker wrote:\n> On Sun, 8 Oct 2000, Tom Lane wrote:\n... \n> > Also, we not long ago went through the exercise of making sure that all\n> > committers were standardized on the same version of Autoconf, ie, 2.13.\n> > Now it emerges that hub.org is running a NON STANDARD version of\n> > Autoconf: 2.13 + unspecified BSD-originated hacks. So the output is\n> > likely to change depending on who committed last and where they did it\n> > from.\n> \n... \n> If using mktemp doesn't break any platform, this is a moot point ... if it\n> does, then I think it is something that *has* to be fix in the FreeBSD\n> port itself so that it doesn't make us look FreeBSD-centric in our\n> development efforts on any other package ...\n\nTo flog an already dead horse (then again my posts get stalled, so 8 Oct mail\nisn't that late :-) (whatever happenend to pgsql-loophole) )\n\nSECURITY CONSIDERATIONS\n The use of mktemp() should generally be avoided, as a hostile process can\n exploit a race condition in the time between the generation of a tempo-\n rary filename by mktemp() and the invoker's use of the temporary name. A\n link-time warning will be issued advising the use of mkstemp() or\n mkdtemp() instead.\n\nCheers,\n\nPatrick\n",
"msg_date": "Thu, 18 Jan 2001 15:34:22 +0000",
"msg_from": "Patrick Welche <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Autoconf version discrepancies"
}
] |
[
{
"msg_contents": "This time I'm pretty sure I caught the initial crash during an update:\n\nI disabled the vacuum analyze and still got table corruption with a crash:\n\ntwo crashdumps of 7.0.2+somepatches\n\n * $Header: /home/pgcvs/pgsql/src/backend/access/common/heaptuple.c,v 1.6\n2 2000/04/12 17:14:36 momjian Exp $\n\nProgram terminated with signal 11, Segmentation fault.\nReading symbols from /usr/lib/libcrypt.so.2...done.\nReading symbols from /usr/lib/libm.so.2...done.\nReading symbols from /usr/lib/libutil.so.3...done.\nReading symbols from /usr/lib/libreadline.so.4...done.\nReading symbols from /usr/lib/libncurses.so.5...done.\nReading symbols from /usr/lib/libc.so.4...done.\nReading symbols from /usr/libexec/ld-elf.so.1...done.\n#0 0x8063aa7 in nocachegetattr (tuple=0x84ae9fc, attnum=4, \n tupleDesc=0x84a6368, isnull=0x84afc20 \"\") at heaptuple.c:537\n537 off = att_addlength(off, att[i]->attlen, tp + off);\n(gdb) bt\n#0 0x8063aa7 in nocachegetattr (tuple=0x84ae9fc, attnum=4, \n tupleDesc=0x84a6368, isnull=0x84afc20 \"\") at heaptuple.c:537\n#1 0x80a027f in ExecEvalVar (variable=0x84974b0, econtext=0x84aedd8, \n isNull=0x84afc20 \"\") at execQual.c:314\n#2 0x80a0d97 in ExecEvalExpr (expression=0x84974b0, econtext=0x84aedd8, \n isNull=0x84afc20 \"\", isDone=0xbfbfe6db \"\\001��J\\b+�\\021\\b\\004�J\\b\\001\")\n at execQual.c:1214\n#3 0x80a090a in ExecEvalFuncArgs (fcache=0x84afc38, econtext=0x84aedd8, \n argList=0x84974d8, argV=0xbfbfe6dc, \n argIsDone=0xbfbfe6db \"\\001��J\\b+�\\021\\b\\004�J\\b\\001\") at execQual.c:635\n#4 0x80a09c1 in ExecMakeFunctionResult (node=0x8496a40, arguments=0x84974d8, \n econtext=0x84aedd8, isNull=0xbfbfe7db \"\", \n isDone=0xbfbfe75b \"\\b\\214翿\\027\\016\\n\\bHuI\\b��J\\b�翿X\\017B`\\004\")\n at execQual.c:711\n#5 0x80a0b37 in ExecEvalOper (opClause=0x8497548, econtext=0x84aedd8, \n isNull=0xbfbfe7db \"\") at execQual.c:902\n#6 0x80a0e17 in ExecEvalExpr (expression=0x8497548, econtext=0x84aedd8, \n isNull=0xbfbfe7db \"\", isDone=0xbfbfe7e0 \"\\001�\\016\\b\") at execQual.c:1249\n#7 0x80a1011 in ExecTargetList (targetlist=0x8497fd8, nodomains=6, \n targettype=0x84aefb0, values=0x84aee48, econtext=0x84aedd8, \n isDone=0xbfbfe90b \"\\001,鿿.K\\n\\bP�J\\b\\214H\\n\\b<鿿�A\\023\\b\\030�H\\b \")\n at execQual.c:1511\n#8 0x80a12af in ExecProject (projInfo=0x84aee20, \n isDone=0xbfbfe90b \"\\001,鿿.K\\n\\bP�J\\b\\214H\\n\\b<鿿�A\\023\\b\\030�H\\b \")\n at execQual.c:1721\n#9 0x80a1365 in ExecScan (node=0x84add50, accessMtd=0x80a488c <IndexNext>)\n at execScan.c:155\n#10 0x80a4b2e in ExecIndexScan (node=0x84add50) at nodeIndexscan.c:288\n#11 0x809fb6d in ExecProcNode (node=0x84add50, parent=0x84add50)\n at execProcnode.c:272\n#12 0x809ed59 in ExecutePlan (estate=0x84ae8a0, plan=0x84add50, \n operation=CMD_UPDATE, offsetTuples=0, numberTuples=0, \n direction=ForwardScanDirection, destfunc=0x84afaf0) at execMain.c:1052\n#13 0x809e2ba in ExecutorRun (queryDesc=0x84ae888, estate=0x84ae8a0, \n feature=3, limoffset=0x0, limcount=0x0) at execMain.c:327\n#14 0x80f92ca in ProcessQueryDesc (queryDesc=0x84ae888, limoffset=0x0, \n limcount=0x0) at pquery.c:310\n#15 0x80f9347 in ProcessQuery (parsetree=0x84965d0, plan=0x84add50, \n dest=Remote) at pquery.c:353\n#16 0x80f7ef0 in pg_exec_query_dest (\n query_string=0x81a9370 \"\\nUPDATE\\n webhit_details_formatted\\nSET\\n attr_hits = attr_hits + '1' \\nWHERE\\n counter_id = '11909'\\n AND attr_type = 'ATTR_OPERATINGSYS'\\n AND attr_name = 'win95'\\n AND attr_vers = '0'\\n;\", \n dest=Remote, aclOverride=0) at postgres.c:663\n#17 0x80f7db9 in pg_exec_query (\n query_string=0x81a9370 \"\\nUPDATE\\n webhit_details_formatted\\nSET\\n attr_hits = attr_hits + '1' \\nWHERE\\n counter_id = '11909'\\n AND attr_type = 'ATTR_OPERATINGSYS'\\n AND attr_name = 'win95'\\n AND attr_vers = '0'\\n;\")\n at postgres.c:562\n#18 0x80f8d1a in PostgresMain (argc=9, argv=0xbfbff0dc, real_argc=10, \n real_argv=0xbfbffb3c) at postgres.c:1590\n#19 0x80e1d06 in DoBackend (port=0x843f400) at postmaster.c:2009\n#20 0x80e1899 in BackendStartup (port=0x843f400) at postmaster.c:1776\n#21 0x80e0abd in ServerLoop () at postmaster.c:1037\n#22 0x80e04be in PostmasterMain (argc=10, argv=0xbfbffb3c) at postmaster.c:725\n#23 0x80aee43 in main (argc=10, argv=0xbfbffb3c) at main.c:93\n#24 0x80633c5 in _start ()\n(gdb) list\n532 \n533 if (usecache)\n534 att[i]->attcacheoff = off;\n535 }\n536 \n537 off = att_addlength(off, att[i]->attlen, tp + off);\n538 \n539 if (usecache &&\n540 att[i]->attlen == -1 && !VARLENA_FIXED_SIZE(att[i]))\n541 usecache = false;\n(gdb) print off\n$1 = 772814392\n(gdb) print att[i]->attlen\n$2 = -1\n(gdb) print off\n$3 = 772814392\n(gdb) print tp\n$4 = 0x5eab73d0 \"\\205.\"\n(gdb) print tp+off\n$7 = 0x8cbbaa08 <Address 0x8cbbaa08 out of bounds>\n(gdb) print usecache\n$8 = 0 '\\000'\n(gdb) print !VARLENA_FIXED_SIZE(att[i])\nNo symbol \"VARLENA_FIXED_SIZE\" in current context.\n(gdb) print att[i]\n$9 = 0x84a66c8\n(gdb) print *(att[i]) \n$10 = {attrelid = 3518994475, attname = {\n data = \"attr_vers\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 1043, \n attdisbursion = 0.125293151, attlen = -1, attnum = 4, attnelems = 0, \n attcacheoff = -1, atttypmod = 36, attbyval = 0 '\\000', attstorage = 112 'p', \n attisset = 0 '\\000', attalign = 105 'i', attnotnull = 0 '\\000', \n atthasdef = 0 '\\000'}\n(gdb) print i\n$11 = 3\n(gdb) print *(att[0])\n$12 = {attrelid = 3518994475, attname = {\n data = \"counter_id\", '\\000' <repeats 21 times>, \n alignmentDummy = 1853189987}, atttypid = 23, \n attdisbursion = 0.000228356235, attlen = 4, attnum = 1, attnelems = 0, \n attcacheoff = 0, atttypmod = -1, attbyval = 1 '\\001', attstorage = 112 'p', \n attisset = 0 '\\000', attalign = 105 'i', attnotnull = 0 '\\000', \n atthasdef = 0 '\\000'}\n(gdb) print *(att[1])\n$13 = {attrelid = 3518994475, attname = {\n data = \"attr_type\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 1043, \n attdisbursion = 0.0928893909, attlen = -1, attnum = 2, attnelems = 0, \n attcacheoff = 4, atttypmod = 36, attbyval = 0 '\\000', attstorage = 112 'p', \n attisset = 0 '\\000', attalign = 105 'i', attnotnull = 0 '\\000', \n atthasdef = 0 '\\000'}\n(gdb) print *(att[2])\n$14 = {attrelid = 3518994475, attname = {\n data = \"attr_name\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 1043, \n attdisbursion = 0.370779663, attlen = -1, attnum = 3, attnelems = 0, \n attcacheoff = -1, atttypmod = 36, attbyval = 0 '\\000', attstorage = 112 'p', \n attisset = 0 '\\000', attalign = 105 'i', attnotnull = 0 '\\000', \n atthasdef = 0 '\\000'}\n(gdb) print attnum\n$15 = 4\n(gdb) print *(att[4])\n$16 = {attrelid = 3518994475, attname = {\n data = \"attr_hits\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 20, attdisbursion = 0.0573871136, \n attlen = 8, attnum = 5, attnelems = 0, attcacheoff = -1, atttypmod = -1, \n attbyval = 0 '\\000', attstorage = 112 'p', attisset = 0 '\\000', \n attalign = 100 'd', attnotnull = 0 '\\000', atthasdef = 1 '\\001'}\n\n\n--------------------------------------------------\n\n\nI'm pretty sure this is a pg_dump that died when the fist crash\nhappened above:\n\n * $Header: /home/pgcvs/pgsql/src/backend/commands/copy.c,v 1.106.2.2 2000/06/28 06:13:01 tgl Exp $\n\n\nProgram terminated with signal 10, Bus error.\n#0 0x482a7d95 in ?? (?? )\n1 0x808c393 in CopyTo (rel=0x84e7890, binary=0 '\\000', oids=0 '\\000', \n fp=0x0, delim=0x8159fa9 \"\\t\", null_print=0x8159fab \"\\\\N\") at copy.c:508\n#2 0x808bf99 in DoCopy (relname=0x84930e8 \"\", binary=0 '\\000', oids=0 '\\000', \n from=0 '\\000', pipe=1 '\\001', filename=0x0, delim=0x8159fa9 \"\\t\", \n null_print=0x8159fab \"\\\\N\") at copy.c:374\n#3 0x80f98a3 in ProcessUtility (parsetree=0x8493110, dest=Remote)\n at utility.c:262\n#4 0x80f7e5e in pg_exec_query_dest (query_string=0x81a9370 \"\", dest=Remote, \n aclOverride=0) at postgres.c:617\n#5 0x80f7db9 in pg_exec_query (query_string=0x81a9370 \"\") at postgres.c:562\n#6 0x80f8d1a in PostgresMain (argc=9, argv=0xbfbff0bc, real_argc=10, \n real_argv=0xbfbffb1c) at postgres.c:1590\n#7 0x80e1d06 in DoBackend (port=0x843f000) at postmaster.c:2009\n#8 0x80e1899 in BackendStartup (port=0x843f000) at postmaster.c:1776\n#9 0x80e0abd in ServerLoop () at postmaster.c:1037\n#10 0x80e04be in PostmasterMain (argc=10, argv=0xbfbffb1c) at postmaster.c:725\n#11 0x80aee43 in main (argc=10, argv=0xbfbffb1c) at main.c:93\n#12 0x80633c5 in _start ()\n(gdb) up\n#1 0x808c393 in CopyTo (rel=0x84e7890, binary=0 '\\000', oids=0 '\\000', \n fp=0x0, delim=0x8159fa9 \"\\t\", null_print=0x8159fab \"\\\\N\") at copy.c:508\n508 string = (char *) (*fmgr_faddr(&out_functions[i]))\n(gdb) print out_functions[i]\n$1 = {fn_addr = 0, fn_plhandler = 0, fn_oid = 0, fn_nargs = 0}\n(gdb) print i\n$2 = 2\n(gdb) print isnull\n$3 = 0 '\\000'\n(gdb) print tupDesc\nNo symbol \"tupDesc\" in current context.\n(gdb) print tuple\n$4 = 0x8493268\n(gdb) print *tuple\n$5 = {t_len = 0, t_self = {ip_blkid = {bi_hi = 0, bi_lo = 0}, ip_posid = 0}, \n t_datamcxt = 0x0, t_data = 0x0}\n(gdb) print value\n$6 = 1072927316\n(gdb) print *value\nCannot access memory at address 0x3ff39254.\n(gdb) print oids\n$7 = 0 '\\000'\n(gdb) print binary\n$8 = 0 '\\000'\n(gdb) print string\n$9 = 0xfffffffc <Address 0xfffffffc out of bounds>\n\nNow I think I have the intial spot where it all goes to pot (the\ninitial traceback). I really appreciate the continued help and\npointers that I've been given and was wondering if someone could\nhelp me out a bit more.\n\nsorry for being such a pain and if any other info is needed please ask.\n\nthanks for you time,\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Mon, 2 Oct 2000 15:17:12 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": true,
"msg_subject": "more crashes"
}
] |
[
{
"msg_contents": "Can someone comment on this?\n\n> Bruce Momjian wrote:\n> > \n> > Seems this was already done in 7.1, right?\n> [thread on puts() and fprintf(stderr and printf instead of tprintf or\n> replacement truncated]\n> \n> There are still scads of fprintf(stderr, \"some error message from\n> postmaster or backend\") lying around, in CURRENT as of this morning at\n> 1:00AM EDT. Some are things such as the usage message -- others are\n> obviously (IMHO) things that need to be sent to the logs. We're not\n> replacing the system fprintf , are we? (my assumption is that we are\n> NOT). The usage of puts(), OTOH, has been well nigh eradicated.\n> \n> Try a find -name '*.[hc]' -print -exec grep printf \\{\\} \\; with cwd=src\n> to see.... (of course, that also exposes where we are continuing to use\n> sprintf instead of snprintf, for whatever reason....)\n> \n> Comments?\n> \n> --\n> Lamar Owen\n> WGCR Internet Radio\n> 1 Peter 4:11\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 2 Oct 2000 23:43:02 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> Can someone comment on this?\n\nYou can tell we're getting close to beta when Bruce combs through the\narchives.... :-)\n\nThe situation has not changed much. There is still code printing to\nstderr and stdout that quite possibly should be using the standard\nlogging system that can timestamp and syslog. One of my favorites in\nthe heap am statistics printout.\n\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Tue, 03 Oct 2000 13:21:21 -0400",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "> Bruce Momjian wrote:\n> > \n> > Can someone comment on this?\n> \n> You can tell we're getting close to beta when Bruce combs through the\n> archives.... :-)\n\nMan, we have been around here too long. :-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 21:59:20 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "Can someone address this?\n\n> Bruce Momjian wrote:\n> > \n> > Seems this was already done in 7.1, right?\n> [thread on puts() and fprintf(stderr and printf instead of tprintf or\n> replacement truncated]\n> \n> There are still scads of fprintf(stderr, \"some error message from\n> postmaster or backend\") lying around, in CURRENT as of this morning at\n> 1:00AM EDT. Some are things such as the usage message -- others are\n> obviously (IMHO) things that need to be sent to the logs. We're not\n> replacing the system fprintf , are we? (my assumption is that we are\n> NOT). The usage of puts(), OTOH, has been well nigh eradicated.\n> \n> Try a find -name '*.[hc]' -print -exec grep printf \\{\\} \\; with cwd=src\n> to see.... (of course, that also exposes where we are continuing to use\n> sprintf instead of snprintf, for whatever reason....)\n> \n> Comments?\n> \n> --\n> Lamar Owen\n> WGCR Internet Radio\n> 1 Peter 4:11\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 9 Oct 2000 15:47:57 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> Can someone address this?\n> \n> > Bruce Momjian wrote:\n> > >\n> > > Seems this was already done in 7.1, right?\n> > [thread on puts() and fprintf(stderr and printf instead of tprintf or\n> > replacement truncated]\n> >\n> > There are still scads of fprintf(stderr, \"some error message from\n> > postmaster or backend\") lying around, in CURRENT as of this morning at\n> > 1:00AM EDT. Some are things such as the usage message -- others are\n> > obviously (IMHO) things that need to be sent to the logs. We're not\n> > replacing the system fprintf , are we? (my assumption is that we are\n> > NOT). The usage of puts(), OTOH, has been well nigh eradicated.\n\nWhere is elog() safe? (Going to Bruce 'comb through the archives' mode\nhere...)\n\nIf someone can educate me in that, I can tackle doing this. Don't know\nif I can do so before 7.1 release, but I'll certainly try.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Thu, 08 Feb 2001 14:17:26 -0500",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "Lamar Owen <[email protected]> writes:\n> There are still scads of fprintf(stderr, \"some error message from\n> postmaster or backend\") lying around, in CURRENT as of this morning at\n> 1:00AM EDT. Some are things such as the usage message -- others are\n> obviously (IMHO) things that need to be sent to the logs. We're not\n> replacing the system fprintf , are we? (my assumption is that we are\n> NOT). The usage of puts(), OTOH, has been well nigh eradicated.\n\n> Where is elog() safe? (Going to Bruce 'comb through the archives' mode\n> here...)\n\nMost places. Not in client communication failures, obviously,\nelse you'd have instant death by recursion since elog tries to\nreport to the client (if any) as well as the system log.\n\nThat case might (perhaps should be) dealt with by means of some\ninternal state flags inside elog.\n\nHowever, it's folly to imagine that we will ever get rid of stderr\noutput entirely. One glaring example is that on most platforms,\nif you have a failure while trying to load a dynamic-link library,\nthe dynamic linker will emit useful messages (like the names of\nunresolved symbols) on stderr. It will never be acceptable to throw\nthat info away, nor do we have a way to capture it and send it\nelsewhere than stderr.\n\nGiven these considerations, I'm not all that excited about mounting a\nholy war on stdout/stderr messages in the backend code. It'd be more\nprofitable to leave the code as-is and figure out a way to cause\nstdout/stderr output to be logged in a more admin-friendly manner.\nI like the idea of piping the output to a log-rotation program.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 08 Feb 2001 14:43:33 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs) "
},
{
"msg_contents": "> However, it's folly to imagine that we will ever get rid of stderr\n> output entirely. One glaring example is that on most platforms,\n> if you have a failure while trying to load a dynamic-link library,\n> the dynamic linker will emit useful messages (like the names of\n> unresolved symbols) on stderr. It will never be acceptable to throw\n> that info away, nor do we have a way to capture it and send it\n> elsewhere than stderr.\n> \n> Given these considerations, I'm not all that excited about mounting a\n> holy war on stdout/stderr messages in the backend code. It'd be more\n> profitable to leave the code as-is and figure out a way to cause\n> stdout/stderr output to be logged in a more admin-friendly manner.\n> I like the idea of piping the output to a log-rotation program.\n\nI am not out to eliminate it. I just want to be sure that we are using\nelog()/fprintf() in the proper places. If someone says we are, the item\nis closed.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Thu, 8 Feb 2001 14:49:55 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "Bruce Momjian wrote:\n> > Given these considerations, I'm not all that excited about mounting a\n> > holy war on stdout/stderr messages in the backend code. \n\nHoly War? Not quite -- just a desire for consistency. Not terribly\nimportant -- just cleaning out my over-stuffed inbox.\n\n> > It'd be more\n> > profitable to leave the code as-is and figure out a way to cause\n> > stdout/stderr output to be logged in a more admin-friendly manner.\n> > I like the idea of piping the output to a log-rotation program.\n \n> I am not out to eliminate it. I just want to be sure that we are using\n> elog()/fprintf() in the proper places.\n\nI _would_ like the output that is useful logging to be directable, as is\nthe case with elog(). It is nice to be able to runtime-configure\nlogging destinations -- syslog, stderr, both. If useful logging output\nis going to stderr when I'm looking in a syslog-managed file elsewhere\n(like on another hardened log bastion host running syslog with remote\nreception), it might as well not even go to stderr at all. And as far\nas dynamic linker output is concerned, that typically gets sent to\nsyslog as well, through other channels, at least in my experience. \nAOLserver is one example that successfully redirects dynamic linker\nmessages to it's own log.\n\nIs syslog not portable enough? Log rotation of syslog-generated\nlogfiles is stock fodder on most Unixoid systems. And PostgreSQL 7.1's\nsupport of syslog is much better than 7.0's.\n\nA syslogger of stderr would make a nice place to pipe the output :-).\n'postmaster .... 2>&1 | output-to-syslog-program -f facility.desired' or\nsome such. But that obviates the need to support syslog _at_all_ in the\nbackend, unless you want the stuff on stderr to go to a different\nfacility from the rest.\n\nBut the original complaint was that log messages were getting lost\nbecause they were going to stderr or stdout when the admin had\nspecifically configured for everything to go to syslog. \n\nWhen working in the total OS environment, and you already have generic\nlog analysis tools set up to work with the OS vendor's logrotate, etc,\nit pays to not reinvent the wheel but use the conventions and tools\nalready provided in the OS. Syslog is a standard way to do this.\n\nWhy even have syslog support otherwise? (Extremist? Maybe.)\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Thu, 08 Feb 2001 15:44:57 -0500",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "On Thu, 8 Feb 2001, Lamar Owen wrote:\n\n> A syslogger of stderr would make a nice place to pipe the output :-).\n> 'postmaster .... 2>&1 | output-to-syslog-program -f facility.desired' or\n> some such. But that obviates the need to support syslog _at_all_ in the\n\n2>&1 | logger -p facility.level\n\nmorannon:~> man logger\nLOGGER(1) System Reference Manual\nLOGGER(1)\n\nNAME \n logger - make entries in the system log\n\nSYNOPSIS\n logger [-is] [-f file] [-p pri] [-t tag] [-u socket] [message ...]\n\nDESCRIPTION\n Logger provides a shell command interface to the syslog(3) system log\n module.\n\n\n-- \nDominic J. Eidson\n \"Baruk Khazad! Khazad ai-menu!\" - Gimli\n-------------------------------------------------------------------------------\nhttp://www.the-infinite.org/ http://www.the-infinite.org/~dominic/\n\n",
"msg_date": "Thu, 8 Feb 2001 14:52:14 -0600 (CST)",
"msg_from": "\"Dominic J. Eidson\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "Lamar Owen <[email protected]> writes:\n> AOLserver is one example that successfully redirects dynamic linker\n> messages to it's own log.\n\nOh? How? Are you sure they're not just piping stderr to a program\nof their own devising? That's basically what I'm recommending.\n\n> Is syslog not portable enough?\n\nIt's got a lot of limitations and problems of its own, on many\nplatforms... but the fundamental point is that without an external\nredirector, we are never going to get everything of interest sent\nto syslog.\n\n> A syslogger of stderr would make a nice place to pipe the output :-).\n> 'postmaster .... 2>&1 | output-to-syslog-program -f facility.desired' or\n> some such. But that obviates the need to support syslog _at_all_ in the\n> backend,\n\nPrecisely my point. I think working hard on syslog support inside elog\nis misplaced effort.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 08 Feb 2001 15:56:02 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs) "
},
{
"msg_contents": "\"Dominic J. Eidson\" wrote:\n> On Thu, 8 Feb 2001, Lamar Owen wrote:\n> > A syslogger of stderr would make a nice place to pipe the output :-).\n> > 'postmaster .... 2>&1 | output-to-syslog-program -f facility.desired' or\n\n> 2>&1 | logger -p facility.level\n[snip]\n> Logger provides a shell command interface to the syslog(3) system log\n> module.\n\nGood. POSIX required, and part of the base system (basically, guaranteed\nto be there on any Linux box). Thanks for the pointer.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Thu, 08 Feb 2001 16:00:12 -0500",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "Tom Lane wrote:\n> Lamar Owen <[email protected]> writes:\n> > AOLserver is one example that successfully redirects dynamic linker\n> > messages to it's own log.\n \n> Oh? How? Are you sure they're not just piping stderr to a program\n> of their own devising? That's basically what I'm recommending.\n\nI'm not sure how they're going about it, but, I'll check.\n \n> > A syslogger of stderr would make a nice place to pipe the output :-).\n> > 'postmaster .... 2>&1 | output-to-syslog-program -f facility.desired' or\n> > some such. But that obviates the need to support syslog _at_all_ in the\n> > backend,\n \n> Precisely my point. I think working hard on syslog support inside elog\n> is misplaced effort.\n\nWell, I can think of a few things:\n1.)\tSome messages are more important than others. Syslog levels are\nuseful to segreggate debug, errors, informational, and critical\nmessages.\n2.)\tCritical messages might need to go to more than one place, while\ndebug messages might need to be dropped silently unless further\nconfiguration, etc, is performed.\n3.)\tSome messages need immediate attention -- syslog can go to the\nconsole for level 'crit' messages.\n\nI know that those three points are part of the same point -- but stderr\nis but a single stream, relegating all messages to the same priority. I\nmight want to keep critical messages far longer than debug messages. I\nmight want to keep FATAL, REALLYFATAL, and even ERROR messages longer\nthan logs of queries (a likely scenario).\n\nSyslog support in elog(), with proper errorlevel coding, allows the\nadmin to segregate messages as he sees fit. Using logger(1) means all\nmessages are the same.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Thu, 08 Feb 2001 16:08:04 -0500",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "On Thu, 8 Feb 2001, Dominic J. Eidson wrote:\n\n> On Thu, 8 Feb 2001, Lamar Owen wrote:\n>\n> > A syslogger of stderr would make a nice place to pipe the output :-).\n> > 'postmaster .... 2>&1 | output-to-syslog-program -f facility.desired' or\n> > some such. But that obviates the need to support syslog _at_all_ in the\n>\n> 2>&1 | logger -p facility.level\n\non linux box I got syslogd eats 45% of cpu.thats why I gave up and\nuse plain loging to file\n\n>\n> morannon:~> man logger\n> LOGGER(1) System Reference Manual\n> LOGGER(1)\n>\n> NAME\n> logger - make entries in the system log\n>\n> SYNOPSIS\n> logger [-is] [-f file] [-p pri] [-t tag] [-u socket] [message ...]\n>\n> DESCRIPTION\n> Logger provides a shell command interface to the syslog(3) system log\n> module.\n>\n>\n>\n\n\tRegards,\n\t\tOleg\n_____________________________________________________________\nOleg Bartunov, sci.researcher, hostmaster of AstroNet,\nSternberg Astronomical Institute, Moscow University (Russia)\nInternet: [email protected], http://www.sai.msu.su/~megera/\nphone: +007(095)939-16-83, +007(095)939-23-83\n\n",
"msg_date": "Fri, 9 Feb 2001 00:14:34 +0300 (GMT)",
"msg_from": "Oleg Bartunov <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "On Thu, Feb 08, 2001 at 02:43:33PM -0500, Tom Lane wrote:\n> Lamar Owen <[email protected]> writes:\n> > There are still scads of fprintf(stderr, \"some error message from\n> > postmaster or backend\") lying around, in CURRENT as of this morning at\n> > 1:00AM EDT. Some are things such as the usage message -- others are\n> > obviously (IMHO) things that need to be sent to the logs. We're not\n> > replacing the system fprintf , are we? (my assumption is that we are\n> > NOT). The usage of puts(), OTOH, has been well nigh eradicated.\n> \n> > Where is elog() safe? (Going to Bruce 'comb through the archives' mode\n> > here...)\n> \n> Most places. Not in client communication failures, obviously,\n> else you'd have instant death by recursion since elog tries to\n> report to the client (if any) as well as the system log.\n> \n> That case might (perhaps should be) dealt with by means of some\n> internal state flags inside elog.\n> \n> However, it's folly to imagine that we will ever get rid of stderr\n> output entirely. One glaring example is that on most platforms,\n> if you have a failure while trying to load a dynamic-link library,\n> the dynamic linker will emit useful messages (like the names of\n> unresolved symbols) on stderr. It will never be acceptable to throw\n> that info away, nor do we have a way to capture it and send it\n> elsewhere than stderr.\n> \n> Given these considerations, I'm not all that excited about mounting a\n> holy war on stdout/stderr messages in the backend code. It'd be more\n> profitable to leave the code as-is and figure out a way to cause\n> stdout/stderr output to be logged in a more admin-friendly manner.\n> I like the idea of piping the output to a log-rotation program.\n\nThere is a program called \"splogger\" for that purpose.\n\nSyslog's wire protocol stinks, but partly by design.\n\nGood logging is hard because you don't want to stall real work \nwaiting to get the messages onto disk or out to the net, but you\nalso don't want to lose messages. The best you can do under those\ncircumstances is to buffer log-traffic spikes and hope the averaged\ntraffic is supportable. Compression can help a lot. Ultimately you \nhave to decide whether you'd rather stall or would rather lose\nmessages, if the traffic gets too heavy, and then live with the \nresult.\n\nNathan Myers\[email protected]\n",
"msg_date": "Thu, 8 Feb 2001 13:57:31 -0800",
"msg_from": "[email protected] (Nathan Myers)",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "On Thu, Feb 08, 2001 at 04:00:12PM -0500, Lamar Owen wrote:\n> \"Dominic J. Eidson\" wrote:\n> > On Thu, 8 Feb 2001, Lamar Owen wrote:\n> > > A syslogger of stderr would make a nice place to pipe the output :-).\n> > > 'postmaster .... 2>&1 | output-to-syslog-program -f facility.desired' or\n> \n> > 2>&1 | logger -p facility.level\n> [snip]\n> > Logger provides a shell command interface to the syslog(3) system log\n> > module.\n> \n> Good. POSIX required, and part of the base system (basically, guaranteed\n> to be there on any Linux box). Thanks for the pointer.\n\nNot so fast... logger just writes its arguments to syslog. I don't\nsee any indication that it (portably) reads its standard input.\nIt's meant for use in shellscripts. You could write:\n\n ... 2>&1 | while read i; do logger -p local1.warning -t 'PG ' -- \"$i\"; done\n\nbut syslog is pretty high-overhead already without starting up logger\non every message. Maybe stderr messages are infrequent enough that\nit doesn't matter.\n\nNathan Myers\[email protected]\n",
"msg_date": "Thu, 8 Feb 2001 14:10:28 -0800",
"msg_from": "[email protected] (Nathan Myers)",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "[email protected] (Nathan Myers) writes:\n> Not so fast... logger just writes its arguments to syslog. I don't\n> see any indication that it (portably) reads its standard input.\n\nFWIW, the HPUX 10.20 man page for logger sez:\n\n A message can be given on the command line, which is logged\n immediately, or a file is read and each line is logged. If no file or\n message is specified, the contents of the standard input are logged.\n\nand they also claim\n\n STANDARDS CONFORMANCE\n logger: XPG4, POSIX.2\n\nThe fact that it's POSIX.2 rather than POSIX.1 might worry folks, but\nI suspect the majority of systems will have it if they have syslog.\n\n(Curiously, the HP man pages do not say that syslog(3) or syslogd(1m)\nconform to *any* standard ... hmm ... is logger more portable than\nsyslog?)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 08 Feb 2001 17:50:55 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs) "
},
{
"msg_contents": "> > Precisely my point. I think working hard on syslog support inside elog\n> > is misplaced effort.\n> \n> Well, I can think of a few things:\n> 1.)\tSome messages are more important than others. Syslog levels are\n> useful to segreggate debug, errors, informational, and critical\n> messages.\n> 2.)\tCritical messages might need to go to more than one place, while\n> debug messages might need to be dropped silently unless further\n> configuration, etc, is performed.\n> 3.)\tSome messages need immediate attention -- syslog can go to the\n> console for level 'crit' messages.\n> \n> I know that those three points are part of the same point -- but stderr\n> is but a single stream, relegating all messages to the same priority. I\n> might want to keep critical messages far longer than debug messages. I\n> might want to keep FATAL, REALLYFATAL, and even ERROR messages longer\n> than logs of queries (a likely scenario).\n\nGood point.\n\n> Syslog support in elog(), with proper errorlevel coding, allows the\n> admin to segregate messages as he sees fit. Using logger(1) means all\n> messages are the same.\n\nTotally agreed.\n\nThere are still many loggings using just plain fprintf(). They should\nbe replaced by elog(DEBUG) or elog(NOTICE), IMHO.\n--\nTatsuo Ishii\n",
"msg_date": "Fri, 09 Feb 2001 11:12:57 +0900",
"msg_from": "Tatsuo Ishii <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "Tatsuo Ishii <[email protected]> writes:\n> There are still many loggings using just plain fprintf(). They should\n> be replaced by elog(DEBUG) or elog(NOTICE), IMHO.\n\nI don't disagree with doing that where it's convenient and safe (which\nis not everywhere). I'm just pointing out that we need a solution that\ncovers stderr output as well.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 08 Feb 2001 21:25:45 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs) "
},
{
"msg_contents": "On Thu, Feb 08, 2001 at 05:50:55PM -0500, Tom Lane wrote:\n> [email protected] (Nathan Myers) writes:\n> > Not so fast... logger just writes its arguments to syslog. I don't\n> > see any indication that it (portably) reads its standard input.\n> \n> FWIW, the HPUX 10.20 man page for logger sez:\n> \n> A message can be given on the command line, which is logged\n> immediately, or a file is read and each line is logged. If no\n> file or message is specified, the contents of the standard input\n> are logged.\n\nRight, I missed where the Linux man page says:\n\n logger [-is] [-f file] [-p pri] [-t tag] [-u socket] [message ...]\n ...\n _message_ Write the message to log; if not specified, and the \n -f flag is not provided, standard input is logged.\n\nSo now the question is, why did they write splogger? splogger parses \nthe beginning of each message to assign a severity; if it finds \"alert:\" \nor \"warning:\" it assigns those, or \"info\" otherwise. To make splogger \nuseful you have to know it's listening.\n\n> and they also claim\n> \n> STANDARDS CONFORMANCE\n> logger: XPG4, POSIX.2\n> \n> The fact that it's POSIX.2 rather than POSIX.1 might worry folks, but\n> I suspect the majority of systems will have it if they have syslog.\n> \n> (Curiously, the HP man pages do not say that syslog(3) or syslogd(1m)\n> conform to *any* standard ... hmm ... is logger more portable than\n> syslog?)\n\nThe Linux page says just:\n\n HISTORY\n A syslog function call appeared in BSD 4.2.\n\nNormally if there's a standard they mention it.\n\nNathan Myers\[email protected]\n",
"msg_date": "Thu, 8 Feb 2001 19:38:47 -0800",
"msg_from": "[email protected] (Nathan Myers)",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "[email protected] (Nathan Myers) writes:\n> So now the question is, why did they write splogger? splogger parses \n> the beginning of each message to assign a severity; if it finds \"alert:\" \n> or \"warning:\" it assigns those, or \"info\" otherwise. To make splogger \n> useful you have to know it's listening.\n\nHowever, that answers Lamar's complaint about needing a way to control\nthe syslog level of messages. splogger might be more useful than logger\nfor our purposes --- even if we have to carry it along with us. What's\nits license? A slight tweak of splogger to recognize our ERROR/FATAL/\nDEBUG prefixes might be just the thing ...\n\n>> (Curiously, the HP man pages do not say that syslog(3) or syslogd(1m)\n>> conform to *any* standard ... hmm ... is logger more portable than\n>> syslog?)\n\n> The Linux page says just:\n> HISTORY\n> A syslog function call appeared in BSD 4.2.\n> Normally if there's a standard they mention it.\n\nYes, the HP man pages also trace it to BSD. I'm surprised syslog\n(apparently) hasn't made it into any formal standard.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 08 Feb 2001 22:52:34 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs) "
},
{
"msg_contents": "On Thu, 8 Feb 2001, Tom Lane wrote:\n\n> [email protected] (Nathan Myers) writes:\n> > The Linux page says just:\n> > HISTORY\n> > A syslog function call appeared in BSD 4.2.\n> > Normally if there's a standard they mention it.\n> \n> Yes, the HP man pages also trace it to BSD. I'm surprised syslog\n> (apparently) hasn't made it into any formal standard.\n\n\"man syslog\" shows the above, \"man logger\" however, shows this:\n\nSTANDARDS\n The logger command is expected to be IEEE Std1003.2 (``POSIX'') compati-\n ble.\n\n\n-- \nDominic J. Eidson\n \"Baruk Khazad! Khazad ai-menu!\" - Gimli\n-------------------------------------------------------------------------------\nhttp://www.the-infinite.org/ http://www.the-infinite.org/~dominic/\n\n",
"msg_date": "Thu, 8 Feb 2001 21:58:45 -0600 (CST)",
"msg_from": "\"Dominic J. Eidson\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs) "
},
{
"msg_contents": "[email protected] (Nathan Myers) writes:\n\n> So now the question is, why did they write splogger?\n\n``They'' didn't write splogger. Dan Bernstein did. It's part of the\nqmail distribution. Unfortunately, the license probably precludes\nincluding it with Postgres. Fortunately, it's only 72 lines long, and\nwould be trivial to recreate.\n\nIan\n",
"msg_date": "08 Feb 2001 20:08:39 -0800",
"msg_from": "Ian Lance Taylor <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "On 8 Feb 2001, Ian Lance Taylor wrote:\n\n> [email protected] (Nathan Myers) writes:\n>\n> > So now the question is, why did they write splogger?\n>\n> ``They'' didn't write splogger. Dan Bernstein did. It's part of the\n> qmail distribution. Unfortunately, the license probably precludes\n> including it with Postgres. Fortunately, it's only 72 lines long, and\n> would be trivial to recreate.\n\nI missed most of this, but has anyone actually ASKED Dan for permission?\nSome of his routines he released into the public domain, some kind of\narrangment may possibly be made. If everyone's afraid to talk to him,\nI will.\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Thu, 8 Feb 2001 23:36:38 -0500 (EST)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "On Thu, Feb 08, 2001 at 11:36:38PM -0500, Vince Vielhaber wrote:\n> On 8 Feb 2001, Ian Lance Taylor wrote:\n> >\n> > Unfortunately, the license [to splogger] probably precludes\n> > including it with Postgres. Fortunately, it's only 72 lines long, and\n> > would be trivial to recreate.\n> \n> I missed most of this, but has anyone actually ASKED Dan for permission?\n\nWhat's the point? I've attached an independent implementation.\nIt recognizes tags for all seven levels. It needs no command-line \narguments. Untagged messages end up logged as \"LOG_NOTICE\". \nUse it freely.\n\nNathan Myers\[email protected]\n\n--------------\n/* pglogger: stdin-to-syslog gateway for postgresql.\n *\n * Copyright 2001 by Nathan Myers <[email protected]>\n * Permission is granted to make copies for any purpose if\n * this copyright notice is retained unchanged.\n*/\n\n#include <stdio.h>\n#include <stddef.h>\n#include <syslog.h>\n#include <string.h>\n\nchar* levels[] =\n{\n \"\", \"emerg:\", \"alert:\", \"crit:\", \"err:\",\n \"warning:\", \"notice:\", \"info:\", \"debug:\" \n};\n\nint lengths[] = \n{\n 0, sizeof(\"emerg\"), sizeof(\"alert\"), sizeof(\"crit\"), sizeof(\"err\"),\n sizeof(\"warning\"), sizeof(\"notice\"), sizeof(\"info\"), sizeof(\"debug\")\n};\n\nint priorities[] = \n{\n LOG_NOTICE, LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR, \n LOG_WARNING, LOG_NOTICE, LOG_INFO, LOG_DEBUG\n};\n\nint main()\n{\n char buf[301];\n int c;\n char* pos = buf;\n int colon = 0;\n\n#ifndef DEBUG\n openlog(\"postgresql\", LOG_CONS, LOG_LOCAL1);\n#endif\n while ( (c = getchar()) != EOF) {\n if (c == '\\r') {\n continue;\n }\n if (c == '\\n') {\n int level = (colon ? sizeof(levels)/sizeof(*levels) : 1);\n char* bol;\n\n *pos = 0;\n while (--level) {\n if (pos - buf >= lengths[level] \n && strncmp(buf, levels[level], lengths[level]) == 0) {\n break; \n }\n }\n bol = buf + lengths[level];\n if (bol > buf && *bol == ' ') {\n ++bol;\n }\n if (pos - bol > 0) {\n#ifndef DEBUG\n syslog(priorities[level], \"%s\", bol);\n#else\n printf(\"%d/%s\\n\", priorities[level], bol);\n#endif\n }\n pos = buf;\n colon = 0;\n continue;\n }\n if (c == ':') {\n colon = 1;\n }\n if ((size_t)(pos - buf) < sizeof(buf)-1) {\n *pos++ = c;\n }\n }\n return 0;\n}\n",
"msg_date": "Fri, 9 Feb 2001 01:16:31 -0800",
"msg_from": "[email protected] (Nathan Myers)",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "Tom Lane wrote:\n> However, that answers Lamar's complaint about needing a way to control\n> the syslog level of messages. splogger might be more useful than logger\n> for our purposes --- even if we have to carry it along with us. What's\n> its license? A slight tweak of splogger to recognize our ERROR/FATAL/\n> DEBUG prefixes might be just the thing ...\n\nOk, I can work with that. Last night I dreamed the solution in shell --\nwhat a wicked pipeline..... Might want to change Nathan's pglogger\nutility to pg_logger for consistency, though. \n\nA quick and dirty solution that works is better than mucking around with\nthe core, particularly this close to release.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Fri, 09 Feb 2001 11:35:26 -0500",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "Tom Lane writes:\n\n> Tatsuo Ishii <[email protected]> writes:\n> > There are still many loggings using just plain fprintf(). They should\n> > be replaced by elog(DEBUG) or elog(NOTICE), IMHO.\n>\n> I don't disagree with doing that where it's convenient and safe (which\n> is not everywhere). I'm just pointing out that we need a solution that\n> covers stderr output as well.\n\nCouldn't we redirect stderr to do what we want? dup(), pipe(), fifo, who\nknows.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Fri, 9 Feb 2001 18:35:20 +0100 (CET)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs) "
},
{
"msg_contents": "Here's the latest version of the pg_logger utility. \nThe particular questions that come to my mind are:\n\n1. Do the prefixes it watches for match what PG produces?\n2. Should it log to LOG_LOCAL1 or to some other LOG_LOCALn?\n3. Is the ident string (\"postgresql\") right?\n4. Are the openlog() args right? (E.g. should it ask for LOG_PID too?)\n5. What am I failing to ask about?\n\nI'd like to turn it over to whoever can answer those questions.\n\nNathan Myers\[email protected]\n\n-----------------\n/* pg_logger: stdin-to-syslog gateway for postgresql.\n *\n * Copyright 2001 by Nathan Myers <[email protected]>\n * This software is distributed free of charge with no warranty of any kind.\n * You have permission to make copies for any purpose, provided that (1) \n * this copyright notice is retained unchanged, and (2) you agree to \n * absolve the author of all responsibility for all consequences arising \n * from any use. \n */\n\n#include <stdio.h>\n#include <stddef.h>\n#include <syslog.h>\n#include <string.h>\n\nstruct {\n char *name;\n int size;\n int priority;\n} tags[] = {\n { \"\", 0, LOG_NOTICE },\n { \"emerg:\", sizeof(\"emerg\"), LOG_EMERG },\n { \"alert:\", sizeof(\"alert\"), LOG_ALERT },\n { \"crit:\", sizeof(\"crit\"), LOG_CRIT },\n { \"err:\", sizeof(\"err\"), LOG_ERR },\n { \"error:\", sizeof(\"error\"), LOG_ERR },\n { \"warning:\", sizeof(\"warning\"), LOG_WARNING },\n { \"notice:\", sizeof(\"notice\"), LOG_NOTICE },\n { \"info:\", sizeof(\"info\"), LOG_INFO },\n { \"debug:\", sizeof(\"debug\"), LOG_DEBUG }\n};\n\nint main()\n{\n char buf[301];\n int c;\n char *pos = buf;\n const char *colon = 0;\n\n#ifndef DEBUG\n openlog(\"postgresql\", LOG_CONS, LOG_LOCAL1);\n#endif\n while ( (c = getchar()) != EOF) {\n if (c == '\\r') {\n continue;\n }\n if (c == '\\n') {\n int level = sizeof(tags)/sizeof(*tags);\n char *bol;\n\n if (colon == 0 || (size_t)(colon - buf) > sizeof(\"warning\")) {\n level = 1;\n }\n *pos = 0;\n while (--level) {\n if (pos - buf >= tags[level].size\n && strncmp(buf, tags[level].name, tags[level].size) == 0) {\n break; \n }\n }\n bol = buf + tags[level].size;\n if (bol > buf && *bol == ' ') {\n ++bol;\n }\n if (pos - bol > 0) {\n#ifndef DEBUG\n syslog(tags[level].priority, \"%s\", bol);\n#else\n printf(\"%d/%s\\n\", tags[level].priority, bol);\n#endif\n }\n pos = buf;\n colon = (char const *)0;\n continue;\n }\n if (c == ':' && !colon) {\n colon = pos;\n }\n if ((size_t)(pos - buf) < sizeof(buf)-1) {\n *pos++ = c;\n }\n }\n return 0;\n}\n",
"msg_date": "Fri, 9 Feb 2001 19:12:17 -0800",
"msg_from": "[email protected] (Nathan Myers)",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs)"
},
{
"msg_contents": "[email protected] (Nathan Myers) writes:\n> 4. Are the openlog() args right? (E.g. should it ask for LOG_PID too?)\n\nLOG_PID seems useless, since that would give you the PID of the logger\nprocess, not of the originating backend ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 09 Feb 2001 22:19:36 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Syslog and pg_options (for RPMs) "
}
] |
[
{
"msg_contents": "I think that pg_dump should not quote identifiers that only contain\nlower-case letters and underscores. That will make it easier to move\ndumps to other database products, because PostgreSQL is off the spec in\nmatters case folding in identifiers. But it's just an idea; if the\nperformance hit would be too high, forget it.\n\n(If you do this, then be careful about locale issues regarding what a\nlower-case letter is. Better to restrict it to ASCII perhaps.)\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Tue, 3 Oct 2000 19:55:46 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "pg_dump and quoting"
},
{
"msg_contents": "At 19:55 3/10/00 +0200, Peter Eisentraut wrote:\n>I think that pg_dump should not quote identifiers that only contain\n>lower-case letters and underscores. That will make it easier to move\n>dumps to other database products, because PostgreSQL is off the spec in\n>matters case folding in identifiers. But it's just an idea; if the\n>performance hit would be too high, forget it.\n\nIn something like pg_dump, I don't really think performance when dumping\nmetadata is an issue when you consider the cost of dumping actual data.\n\nBut to deal with the quoting issue, have you tried '-n' - I think it\nbasically turns off non-essential quoting.\n\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Wed, 04 Oct 2000 21:05:58 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and quoting"
}
] |
[
{
"msg_contents": "\nHi,\n\nI was wondering if anyone could offer me some pointers in extending the\nC++ classes used for client applications? I'm interested in developing a\nset of C++ classes which will interact directly with the postgres back-end\nwith the following enhancements to the current implementation:\n\nThese are preliminary musings and definitely open to change.\n\n1) Implement connections as a stream object. e.g. \n db = new pg_database(char * connection_params);\n db << sql_string;\n\n2) Implement results as a returned object from pg_database\n result = db.exec();\n result >> field1 >> field2 >> field3;\n and\n field = result[\"field\"];\n field = result[fieldnum];\n \n3) Implement asynchronous calls to the back-end permitting multiple\n queries to be executed concurrently\n\n4) Implement a standard set of exceptions which can be thrown \n ex. if connection to back-end is interrupted or cannot be established\n within 30 seconds of the first attempt, throw.\n\n5) I'm looking at making the libraries portable between FreeBSD, Solaris \n Linux and Win32.\n\nIf anyone has suggestions/additions to this list it would be greatly\nappreciated!\n\nCheers,\n\nRandy Jonasz\nSoftware Engineer\nClick2net Inc.\nWeb: http://www.click2net.com\nPhone: (905) 271-3550\n\n\"You cannot possibly pay a philosopher what he's worth,\nbut try your best\" -- Aristotle\n\n",
"msg_date": "Tue, 3 Oct 2000 14:47:45 -0400 (EDT)",
"msg_from": "Randy Jonasz <[email protected]>",
"msg_from_op": true,
"msg_subject": "C++ client libs"
},
{
"msg_contents": "Thanks for responding so quickly. Your help would be most welcome. I need\nto do some further research as well before laying down an API. The\ncompany I work for has offered to set up a CVS box open to the public for\ndeveloping this API. There's another person who has expressed interest in\nthis project as well. What I suggest is working together in developing\nconcrete/virtual classes with each of their member functions laid out on\npaper first. We can then parcel out the actual coding of the classes\nbetween us, using CVS to keep track of our code.\n\nCheers,\n\nRandy\n\nOn Tue, 3 Oct 2000, Adam Haberlach wrote:\n\n> On Tue, Oct 03, 2000 at 02:47:45PM -0400, Randy Jonasz wrote:\n> \n> > I was wondering if anyone could offer me some pointers in extending the\n> > C++ classes used for client applications? I'm interested in developing a\n> > set of C++ classes which will interact directly with the postgres back-end\n> > with the following enhancements to the current implementation:\n> > \n> > These are preliminary musings and definitely open to change.\n> > \n> > 1) Implement connections as a stream object. e.g. \n> > db = new pg_database(char * connection_params);\n> > db << sql_string;\n> \n> \tI don't really like heavy overloading of stream-type things.\n> \tIn this case I think it may make sense, but I prefer\n> \tthe db.Exec(sql_string) system for clarity...\n> \n> > 2) Implement results as a returned object from pg_database\n> > result = db.exec();\n> > result >> field1 >> field2 >> field3;\n> > and\n> > field = result[\"field\"];\n> > field = result[fieldnum];\n> \n> \tThe above system puts some state into the result object. Overloading\n> \tarray operators can be sketchy, too.\n> \t\n> \tThe current system definitely needs to decouple the result from the\n> \tconnection. I find that the current API causes most implementations\n> \tto create a lot of independent connections to the backend.\n> \n> > 3) Implement asynchronous calls to the back-end permitting multiple\n> > queries to be executed concurrently\n> \n> \tYes. I was going to do this in order to make use of the NOTIFY/LISTEN\n> \tfunctionality as well.\n> \n> > 4) Implement a standard set of exceptions which can be thrown \n> > ex. if connection to back-end is interrupted or cannot be established\n> > within 30 seconds of the first attempt, throw.\n> \n> \tIf you really want exceptions, sure. I don't like 'em myself, but I\n> \tcan understand their advantages.\n> \n> > 5) I'm looking at making the libraries portable between FreeBSD, Solaris \n> > Linux and Win32.\n> \n> \tPlease add BeOS that that list--I'll help keep it in line.\n> \n> > If anyone has suggestions/additions to this list it would be greatly\n> > appreciated!\n> \n> \tI'm very interested in helping with this (it is on my list of projects\n> \tand I've managed to stop adding new ones to the list for now). Should\n> \twe take this off-list, or keep it here?\n> \n> \tIn either case, I'll spend some time formally thinking about my\n> \tperfect API and we can get together and compare notes...\n> \n> -- \n> Adam Haberlach | A billion hours ago, human life appeared on\n> [email protected] | earth. A billion minutes ago, Christianity\n> http://www.newsnipple.com | emerged. A billion Coca-Colas ago was\n> '88 EX500 | yesterday morning. -1996 Coca-Cola Ann. Rpt.\n> \n> \n\nRandy Jonasz\nSoftware Engineer\nClick2net Inc.\nWeb: http://www.click2net.com\nPhone: (905) 271-3550\n\n\"You cannot possibly pay a philosopher what he's worth,\nbut try your best\" -- Aristotle\n\n",
"msg_date": "Tue, 3 Oct 2000 17:01:52 -0400 (EDT)",
"msg_from": "Randy Jonasz <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: C++ client libs"
},
{
"msg_contents": "On Tue, Oct 03, 2000 at 02:47:45PM -0400, Randy Jonasz wrote:\n\n> I was wondering if anyone could offer me some pointers in extending the\n> C++ classes used for client applications? I'm interested in developing a\n> set of C++ classes which will interact directly with the postgres back-end\n> with the following enhancements to the current implementation:\n> \n> These are preliminary musings and definitely open to change.\n> \n> 1) Implement connections as a stream object. e.g. \n> db = new pg_database(char * connection_params);\n> db << sql_string;\n\n\tI don't really like heavy overloading of stream-type things.\n\tIn this case I think it may make sense, but I prefer\n\tthe db.Exec(sql_string) system for clarity...\n\n> 2) Implement results as a returned object from pg_database\n> result = db.exec();\n> result >> field1 >> field2 >> field3;\n> and\n> field = result[\"field\"];\n> field = result[fieldnum];\n\n\tThe above system puts some state into the result object. Overloading\n\tarray operators can be sketchy, too.\n\t\n\tThe current system definitely needs to decouple the result from the\n\tconnection. I find that the current API causes most implementations\n\tto create a lot of independent connections to the backend.\n\n> 3) Implement asynchronous calls to the back-end permitting multiple\n> queries to be executed concurrently\n\n\tYes. I was going to do this in order to make use of the NOTIFY/LISTEN\n\tfunctionality as well.\n\n> 4) Implement a standard set of exceptions which can be thrown \n> ex. if connection to back-end is interrupted or cannot be established\n> within 30 seconds of the first attempt, throw.\n\n\tIf you really want exceptions, sure. I don't like 'em myself, but I\n\tcan understand their advantages.\n\n> 5) I'm looking at making the libraries portable between FreeBSD, Solaris \n> Linux and Win32.\n\n\tPlease add BeOS that that list--I'll help keep it in line.\n\n> If anyone has suggestions/additions to this list it would be greatly\n> appreciated!\n\n\tI'm very interested in helping with this (it is on my list of projects\n\tand I've managed to stop adding new ones to the list for now). Should\n\twe take this off-list, or keep it here?\n\n\tIn either case, I'll spend some time formally thinking about my\n\tperfect API and we can get together and compare notes...\n\n-- \nAdam Haberlach | A billion hours ago, human life appeared on\[email protected] | earth. A billion minutes ago, Christianity\nhttp://www.newsnipple.com | emerged. A billion Coca-Colas ago was\n'88 EX500 | yesterday morning. -1996 Coca-Cola Ann. Rpt.\n",
"msg_date": "Tue, 3 Oct 2000 14:43:13 -0700",
"msg_from": "Adam Haberlach <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: C++ client libs"
},
{
"msg_contents": "> > If anyone has suggestions/additions to this list it would be greatly\n> > appreciated!\n> \n> \tI'm very interested in helping with this (it is on my list of projects\n> \tand I've managed to stop adding new ones to the list for now). Should\n> \twe take this off-list, or keep it here?\n\nCertainly the PostgreSQL interfaces list is the best place to discuss,\nbut hackers is ok for a period.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sun, 8 Oct 2000 22:32:24 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: C++ client libs"
},
{
"msg_contents": "> > > 2) Implement results as a returned object from pg_database\n> > > result = db.exec();\n> > > result >> field1 >> field2 >> field3;\n> > > and\n> > > field = result[\"field\"];\n> > > field = result[fieldnum];\n\nRogue Wave Software has an SQL C++ interface in dbtools.h++. It is\nsimilar to what you describe. Here is a sample program:\n\n #include <rw/db/db.h>\n\n int main ()\n {\n 1. RWDBDatabase enterpriseDB = RWDBManager :: database\n\t\t\t (\"SYBASE\", \"SYBASESERVER\", \"ME\", \"PASSWORD\", \"DATA\");\n\n RWDBTable table1 = enterpriseDB.table(\"Actual_Sales\");\n\n 2. RWDBReader table1Reader = table1.reader();\n\n RWDBDatabase spreadsheet = RWDBManager :: database\n\t\t\t (\"ODBC\", \"MS_EXCEL\", \"ME\", \"PASSWORD\", \"C:\\mydir\");\n\n 3. spreadsheet.createTable (\"TMPSALES\", table1.schema());\n RWDBTable spreadsheetTable = spreadsheet.table (\"TMPSALES\");\n\n 4. RWDBInserter localInserter = spreadsheetTable.inserter();\n while (table1Reader()) {\n\t localInserter << table1Reader;\n\t localInserter.execute();\n }\n\n ...\n\n 5. /* Insert DDE code to plot your sales forecast, stored\n\t in the second spreadsheet, versus actual sales now\n\t stored in the TMPSALES spreadsheet. */\n }\n\t\t\t\t \n\nThere are white-papers describing the interface. It may be helpful for\nideas. You can see more at:\n\n\thttp://www.roguewave.com/products/xplatform/dbtools/\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 9 Oct 2000 03:36:52 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: C++ client libs"
}
] |
[
{
"msg_contents": "If you are looking for programs which tend to hold longstanding\ntransactions, this micro patch might be handy.\nWhether it is of general use is debatable. It will replace 'idle' by\n'trans' if the backend is idle but a transaction is pending.\n\nSimply use ps to view the backend's status.\n\nChristof",
"msg_date": "Tue, 03 Oct 2000 21:28:36 +0200",
"msg_from": "Christof Petig <[email protected]>",
"msg_from_op": true,
"msg_subject": "Small patch to replace 'idle' by 'trans' if transaction is still open"
},
{
"msg_contents": "I can't decide if this is of general use. My inclination is that\nsomeone in a transaction sitting a prompt should still show as idle.\n\n> If you are looking for programs which tend to hold longstanding\n> transactions, this micro patch might be handy.\n> Whether it is of general use is debatable. It will replace 'idle' by\n> 'trans' if the backend is idle but a transaction is pending.\n> \n> Simply use ps to view the backend's status.\n> \n> Christof\n\n[ application/x-gzip is not supported, skipping... ]\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 9 Oct 2000 20:16:16 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction\n\tis still open"
},
{
"msg_contents": "Bruce Momjian <[email protected]> writes:\n> I can't decide if this is of general use. My inclination is that\n> someone in a transaction sitting a prompt should still show as idle.\n\nThe idea seemed good to me, although I didn't look at the code to see\nif the implementation was any good ;-). I know we've frequently had\nquestions on the lists where it was interesting to know if any\ntransactions were being held open --- and right now there's no easy\nway to tell.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 09 Oct 2000 22:46:56 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction is still\n\topen"
},
{
"msg_contents": "Tom Lane wrote:\n\n> Bruce Momjian <[email protected]> writes:\n> > I can't decide if this is of general use. My inclination is that\n> > someone in a transaction sitting a prompt should still show as idle.\n>\n> The idea seemed good to me, although I didn't look at the code to see\n> if the implementation was any good ;-). I know we've frequently had\n> questions on the lists where it was interesting to know if any\n> transactions were being held open --- and right now there's no easy\n> way to tell.\n>\n> regards, tom lane\n\nI would even propose 'idle (transaction open)' if you agree. (Seemed too\nlong for me)\nThe patch is _trivial_ a string changed to a condifional operator and two\nstrings (two times). A two liner.\n\n Christof\n\n\n",
"msg_date": "Wed, 11 Oct 2000 08:54:29 +0200",
"msg_from": "Christof Petig <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction is\n\tstill open"
},
{
"msg_contents": "Christof Petig <[email protected]> writes:\n> I would even propose 'idle (transaction open)' if you agree. (Seemed too\n> long for me)\n\nNo, the ps strings have got to be *short*. A lot of platforms have a\ntight limit on how much command string ps will show. I thought 'trans'\nwas OK, though perhaps someone has an even better idea?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 11 Oct 2000 10:03:00 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction is still\n\topen"
},
{
"msg_contents": "Philip Warner <[email protected]> writes:\n> Any chance of showing if anything has been updated in the TX?\n\nHmm ... SharedBufferChanged would probably do as a proxy for that ...\nnot sure if it's a good idea though.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 11 Oct 2000 10:54:17 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction is still\n\topen"
},
{
"msg_contents": "> Christof Petig <[email protected]> writes:\n> > I would even propose 'idle (transaction open)' if you agree. (Seemed too\n> > long for me)\n> \n> No, the ps strings have got to be *short*. A lot of platforms have a\n> tight limit on how much command string ps will show. I thought 'trans'\n> was OK, though perhaps someone has an even better idea?\n\nidle/trans?\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Wed, 11 Oct 2000 11:14:56 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction\n\tis still open"
},
{
"msg_contents": "At 10:03 11/10/00 -0400, Tom Lane wrote:\n>Christof Petig <[email protected]> writes:\n>> I would even propose 'idle (transaction open)' if you agree. (Seemed too\n>> long for me)\n>\n>No, the ps strings have got to be *short*. A lot of platforms have a\n>tight limit on how much command string ps will show. I thought 'trans'\n>was OK, though perhaps someone has an even better idea?\n\nAny chance of showing if anything has been updated in the TX? Then the text\ncould be:\n\n RW TX\nor\n RO TX\n\nIf not, then how about 'TX Active' or 'TX Open' or just 'In TX'\n\n\n\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Thu, 12 Oct 2000 01:14:58 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if\n\ttransaction is still open"
},
{
"msg_contents": "Philip Warner <[email protected]> writes:\n> The idea was to indicate if the process is likely to be causing a locking\n> problem - but now I think about it, a more useful pice of information would\n> be if any tables/rows/whatever are locked in such a way that another user\n> might be prevented from doing an update. Is this possible?\n\nDon't see how you could convey a useful amount of info in the space\navailable. There are way too many levels of way too many locks that\nmight or might not conflict with someone else. (Even a read-only\ntransaction grabs read locks, so it's not like the presence or absence\nof any lock is sufficient info to be useful.)\n\nThere was another thread recently about adding a SHOW command or some\nsuch to dump out the state of the lock manager's table in a readable\nform. I think that'd be a more useful thing to work on than trying\nto cram one or two bits' worth of info into the ps display.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 11 Oct 2000 11:15:54 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction is still\n\topen"
},
{
"msg_contents": "At 10:54 11/10/00 -0400, Tom Lane wrote:\n>Philip Warner <[email protected]> writes:\n>> Any chance of showing if anything has been updated in the TX?\n>\n>Hmm ... SharedBufferChanged would probably do as a proxy for that ...\n>not sure if it's a good idea though.\n\nThe idea was to indicate if the process is likely to be causing a locking\nproblem - but now I think about it, a more useful pice of information would\nbe if any tables/rows/whatever are locked in such a way that another user\nmight be prevented from doing an update. Is this possible?\n\nI would maintain that if possible, it is useful to distinguish logical\n'read-only' transactions from TXs that might lock another user.\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Thu, 12 Oct 2000 02:12:47 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if\n\ttransaction is still open"
},
{
"msg_contents": "Tom Lane writes:\n\n> Christof Petig <[email protected]> writes:\n> > I would even propose 'idle (transaction open)' if you agree. (Seemed too\n> > long for me)\n> \n> No, the ps strings have got to be *short*. A lot of platforms have a\n> tight limit on how much command string ps will show. I thought 'trans'\n> was OK, though perhaps someone has an even better idea?\n\nSince nothing follows the status indicator in the ps display you can\nreally make it as long as you want; the rest will be truncated. So you\nmight want to put it \"transaction (idle)\", so that the essential\ninformation is first.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Wed, 11 Oct 2000 18:47:47 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction\n\tis still open"
},
{
"msg_contents": "> At 11:15 11/10/00 -0400, Tom Lane wrote:\n> >\n> >There was another thread recently about adding a SHOW command or some\n> >such to dump out the state of the lock manager's table in a readable\n> >form. I think that'd be a more useful thing to work on than trying\n> >to cram one or two bits' worth of info into the ps display.\n> >\n> \n> Absolutely. I assume from this that the nature and state of all locks are\n> stored in shared memory somewhere, and that the data is able to be\n> interpreted without reference to non-shared data. If so, this would seem to\n> be (a) a very useful thing to have and (b) not too hard. Is that right?\n> \n\nAdded to TODO:\n\n\t* Add SHOW command to display locks\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Thu, 12 Oct 2000 21:53:15 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction\n\tis still open"
},
{
"msg_contents": "At 11:15 11/10/00 -0400, Tom Lane wrote:\n>\n>There was another thread recently about adding a SHOW command or some\n>such to dump out the state of the lock manager's table in a readable\n>form. I think that'd be a more useful thing to work on than trying\n>to cram one or two bits' worth of info into the ps display.\n>\n\nAbsolutely. I assume from this that the nature and state of all locks are\nstored in shared memory somewhere, and that the data is able to be\ninterpreted without reference to non-shared data. If so, this would seem to\nbe (a) a very useful thing to have and (b) not too hard. Is that right?\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Fri, 13 Oct 2000 11:59:09 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if\n\ttransaction is still open"
},
{
"msg_contents": "Philip Warner <[email protected]> writes:\n>> There was another thread recently about adding a SHOW command or some\n>> such to dump out the state of the lock manager's table in a readable\n>> form. I think that'd be a more useful thing to work on than trying\n>> to cram one or two bits' worth of info into the ps display.\n\n> Absolutely. I assume from this that the nature and state of all locks are\n> stored in shared memory somewhere, and that the data is able to be\n> interpreted without reference to non-shared data. If so, this would seem to\n> be (a) a very useful thing to have and (b) not too hard. Is that right?\n\nThe information is theoretically available in shared memory, but I'm\nnot sure how hard it is to transform into a conveniently readable form\n(like \"process X holds locks A, B, and C and is waiting on lock D\").\nI know that the few times I've tried to grovel around in the lock table\nwith a debugger, it's been pretty excruciating to figure out what was\ngoing on --- there's just barely enough info there for the code to work,\nand no overhead to aid in interpretation. You might find that adding\nsome additional fields to locktable entries would be a good idea.\n(Or maybe not; I can tell you that it's no fun to try to follow it with\njust gdb print statements, but a piece of code might not have trouble.)\n\nThere already is some code in lock.c to print out the contents of\nthe table, but it's only conditionally compiled (ifdef LOCK_DEBUG),\nhas no way to be called except through hand intervention with a debugger,\nand furthermore is set up to dump to the backend's stdout, which is\nnot especially convenient in most scenarios. A first cut would be\nto transform that code into a user-callable statement that reports\nvia elog(NOTICE). Then we could look at the output and see whether\nit's useful for mere mortals or not...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 13 Oct 2000 20:51:27 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction is still\n\topen"
},
{
"msg_contents": "At 20:51 13/10/00 -0400, Tom Lane wrote:\n>Philip Warner <[email protected]> writes:\n>\n>The information is theoretically available in shared memory, but I'm\n>not sure how hard it is to transform into a conveniently readable form\n>(like \"process X holds locks A, B, and C and is waiting on lock D\").\n>I know that the few times I've tried to grovel around in the lock table\n>with a debugger, it's been pretty excruciating to figure out what was\n>going on --- there's just barely enough info there for the code to work,\n>and no overhead to aid in interpretation. You might find that adding\n>some additional fields to locktable entries would be a good idea.\n>(Or maybe not; I can tell you that it's no fun to try to follow it with\n>just gdb print statements, but a piece of code might not have trouble.)\n\nThis sounds good; I wrote a similar lock-tracer for Dec RDB before they\nintroduced a utility for this purpose. It's relatively easy, technically,\nso long at the lock manager has all the info. What I'd probably want to do\nis (a) dump current processes and their state as well as their lock request\nqueue, then (b) assuming a stalled lock is identified, allow the user/DBA\nto request a trace of the specific lock ID. Hope the terminology makes\nsense in terms of the PGSQL lock manager. It would help if the lock manager\nhad the date/time a request was made.\n\n\n>There already is some code in lock.c to print out the contents of\n>the table, but it's only conditionally compiled (ifdef LOCK_DEBUG),\n>has no way to be called except through hand intervention with a debugger,\n>and furthermore is set up to dump to the backend's stdout, which is\n>not especially convenient in most scenarios. A first cut would be\n>to transform that code into a user-callable statement that reports\n>via elog(NOTICE). Then we could look at the output and see whether\n>it's useful for mere mortals or not...\n\nSounds like a good start. But would it be possible/desirable for a utility\nprogram to map the lock manager tables directly? \n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Sat, 14 Oct 2000 22:27:41 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if\n\ttransaction is still open"
},
{
"msg_contents": "Philip Warner <[email protected]> writes:\n> Sounds like a good start. But would it be possible/desirable for a utility\n> program to map the lock manager tables directly? \n\nThe utility program would have to be able to attach to the shared memory\nsegments holding the locktable and associated spinlocks. And if you\ndid it that way, you couldn't inspect the locktable remotely without\nduplicating a bunch more postmaster/backend code. Offhand it seems to\nme that \"start an additional backend and use it to look at the\nlocktable\" is the best way to approach this.\n\nThis does raise an interesting point though: do we need any security on\nthis new SHOW command? Like, say, only allow it to superusers?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 14 Oct 2000 12:27:50 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction is still\n\topen"
},
{
"msg_contents": "\nSeems we decided against this. Sorry.\n\n> Tom Lane wrote:\n> \n> > Bruce Momjian <[email protected]> writes:\n> > > I can't decide if this is of general use. My inclination is that\n> > > someone in a transaction sitting a prompt should still show as idle.\n> >\n> > The idea seemed good to me, although I didn't look at the code to see\n> > if the implementation was any good ;-). I know we've frequently had\n> > questions on the lists where it was interesting to know if any\n> > transactions were being held open --- and right now there's no easy\n> > way to tell.\n> >\n> > regards, tom lane\n> \n> I would even propose 'idle (transaction open)' if you agree. (Seemed too\n> long for me)\n> The patch is _trivial_ a string changed to a condifional operator and two\n> strings (two times). A two liner.\n> \n> Christof\n> \n> \n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 19 Jan 2001 12:25:18 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction\n\tis still open"
},
{
"msg_contents": "On Fri, 19 Jan 2001, Bruce Momjian wrote:\n\n>\n> Seems we decided against this. Sorry.\n\nHuh? from reading Tom's response, sounds like it would be something\nuseful? I know I find having as much information about state in the ps\nlisting helps me alot, and knowing if its 'idle' vs 'idle (in\ntransaction)' provides at lesat more detailed information then just 'idle'\n...\n\nWho was the 'we' in the above decision? Tom seemed in favor of it, I know\nI'm in favor of it .. and you are not in favor of it ...\n\n > > > Tom Lane wrote:\n> >\n> > > Bruce Momjian <[email protected]> writes:\n> > > > I can't decide if this is of general use. My inclination is that\n> > > > someone in a transaction sitting a prompt should still show as idle.\n> > >\n> > > The idea seemed good to me, although I didn't look at the code to see\n> > > if the implementation was any good ;-). I know we've frequently had\n> > > questions on the lists where it was interesting to know if any\n> > > transactions were being held open --- and right now there's no easy\n> > > way to tell.\n> > >\n> > > regards, tom lane\n> >\n> > I would even propose 'idle (transaction open)' if you agree. (Seemed too\n> > long for me)\n> > The patch is _trivial_ a string changed to a condifional operator and two\n> > strings (two times). A two liner.\n> >\n> > Christof\n> >\n> >\n> >\n>\n>\n> --\n> Bruce Momjian | http://candle.pha.pa.us\n> [email protected] | (610) 853-3000\n> + If your life is a hard drive, | 830 Blythe Avenue\n> + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n>\n\nMarc G. Fournier ICQ#7615664 IRC Nick: Scrappy\nSystems Administrator @ hub.org\nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org\n\n",
"msg_date": "Fri, 19 Jan 2001 14:21:00 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction\n\tis still open"
},
{
"msg_contents": "> On Fri, 19 Jan 2001, Bruce Momjian wrote:\n> \n> >\n> > Seems we decided against this. Sorry.\n> \n> Huh? from reading Tom's response, sounds like it would be something\n> useful? I know I find having as much information about state in the ps\n> listing helps me alot, and knowing if its 'idle' vs 'idle (in\n> transaction)' provides at lesat more detailed information then just 'idle'\n\n> Who was the 'we' in the above decision? Tom seemed in favor of it, I know\n> I'm in favor of it .. and you are not in favor of it ...\n\nThere must have been some discussion about it. I don't see it in the\ncode, and I remember it was rejected for some reason. Check the archives.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 19 Jan 2001 15:59:41 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Small patch to replace 'idle' by 'trans' if transaction\n\tis still open"
},
{
"msg_contents": "If you don't know what is all about read the bottom (I'll requote my posting)\n\nBruce Momjian wrote:\n\n> > On Fri, 19 Jan 2001, Bruce Momjian wrote:\n> > > Seems we decided against this. Sorry.\n> >\n> > Huh? from reading Tom's response, sounds like it would be something\n> > useful? I know I find having as much information about state in the ps\n> > listing helps me alot, and knowing if its 'idle' vs 'idle (in\n> > transaction)' provides at lesat more detailed information then just 'idle'\n>\n> > Who was the 'we' in the above decision? Tom seemed in favor of it, I know\n> > I'm in favor of it .. and you are not in favor of it ...\n>\n> There must have been some discussion about it. I don't see it in the\n> code, and I remember it was rejected for some reason. Check the archives.\n\nThe thing which comes most close to a rejection was the 'I can't decide' mail\nby you (answered by Tom). The conclusion sounded like 'since we're not clear on\nthis subject we won't touch this, yet'. And there was some unsettled discussion\nabout the best wording to show in 'ps'.\n\n'trans' seemed too short (and too much unlike 'idle') (as indicated by Bruce)\nand 'idle (open transaction)' might give difficulties on platforms which limit\nthe length of the string (as indicated by Tom)\n\nI'll CC Hackers (where this discussion belongs)\n\n Christof\n\n---------------------\n\nQuoting:\n\nSubject: Re: [PATCHES] Small patch to replace 'idle' by 'trans' if transaction\nis still open\nDate: Mon, 09 Oct 2000 22:46:56 -0400\nFrom: Tom Lane <[email protected]>\n\nBruce Momjian <[email protected]> writes:\n> I can't decide if this is of general use. My inclination is that\n> someone in a transaction sitting a prompt should still show as idle.\n\nThe idea seemed good to me, although I didn't look at the code to see\nif the implementation was any good ;-). I know we've frequently had\nquestions on the lists where it was interesting to know if any\ntransactions were being held open --- and right now there's no easy\nway to tell.\n\n regards, tom lane\n\n--------------\n\nSubject: Small patch to replace 'idle' by 'trans' if transaction is still open\nDate: Tue, 03 Oct 2000 21:28:36 +0200\nFrom: Christof Petig <[email protected]>\n\nIf you are looking for programs which tend to hold longstanding\ntransactions, this micro patch might be handy.\nWhether it is of general use is debatable. It will replace 'idle' by\n'trans' if the backend is idle but a transaction is pending.\n\nSimply use ps to view the backend's status.\n\nChristof\n\n--- src/backend/commands/async.c~ Sun May 14 05:18:35 2000\n+++ src/backend/commands/async.c Tue Oct 3 10:31:54 2000\n@@ -818,7 +818,7 @@\n */\n pq_flush();\n\n- PS_SET_STATUS(\"idle\");\n+ PS_SET_STATUS(IsTransactionBlock()?\"trans\":\"idle\");\n TPRINTF(TRACE_NOTIFY, \"ProcessIncomingNotify: done\");\n }\n\n--- src/backend/tcop/postgres.c~ Thu Aug 31 09:18:57 2000\n+++ src/backend/tcop/postgres.c Tue Oct 3 10:32:23 2000\n@@ -1496,7 +1496,7 @@\n\n for (;;)\n {\n- PS_SET_STATUS(\"idle\");\n+ PS_SET_STATUS(IsTransactionBlock()?\"trans\":\"idle\");\n\n /* ----------------\n * (1) tell the frontend we're ready for a new query.\n\n\n",
"msg_date": "Mon, 22 Jan 2001 08:51:33 +0100",
"msg_from": "Christof Petig <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Small patch to replace 'idle' by 'trans' if\n\ttransactionis still open"
},
{
"msg_contents": "\nWell, I'm still for this ... how about something simiple like 'idle in\ntrans'? show, and easy to figure out what it means?\n\nOn Mon, 22 Jan 2001, Christof Petig wrote:\n\n> If you don't know what is all about read the bottom (I'll requote my posting)\n>\n> Bruce Momjian wrote:\n>\n> > > On Fri, 19 Jan 2001, Bruce Momjian wrote:\n> > > > Seems we decided against this. Sorry.\n> > >\n> > > Huh? from reading Tom's response, sounds like it would be something\n> > > useful? I know I find having as much information about state in the ps\n> > > listing helps me alot, and knowing if its 'idle' vs 'idle (in\n> > > transaction)' provides at lesat more detailed information then just 'idle'\n> >\n> > > Who was the 'we' in the above decision? Tom seemed in favor of it, I know\n> > > I'm in favor of it .. and you are not in favor of it ...\n> >\n> > There must have been some discussion about it. I don't see it in the\n> > code, and I remember it was rejected for some reason. Check the archives.\n>\n> The thing which comes most close to a rejection was the 'I can't decide' mail\n> by you (answered by Tom). The conclusion sounded like 'since we're not clear on\n> this subject we won't touch this, yet'. And there was some unsettled discussion\n> about the best wording to show in 'ps'.\n>\n> 'trans' seemed too short (and too much unlike 'idle') (as indicated by Bruce)\n> and 'idle (open transaction)' might give difficulties on platforms which limit\n> the length of the string (as indicated by Tom)\n>\n> I'll CC Hackers (where this discussion belongs)\n>\n> Christof\n>\n> ---------------------\n>\n> Quoting:\n>\n> Subject: Re: [PATCHES] Small patch to replace 'idle' by 'trans' if transaction\n> is still open\n> Date: Mon, 09 Oct 2000 22:46:56 -0400\n> From: Tom Lane <[email protected]>\n>\n> Bruce Momjian <[email protected]> writes:\n> > I can't decide if this is of general use. My inclination is that\n> > someone in a transaction sitting a prompt should still show as idle.\n>\n> The idea seemed good to me, although I didn't look at the code to see\n> if the implementation was any good ;-). I know we've frequently had\n> questions on the lists where it was interesting to know if any\n> transactions were being held open --- and right now there's no easy\n> way to tell.\n>\n> regards, tom lane\n>\n> --------------\n>\n> Subject: Small patch to replace 'idle' by 'trans' if transaction is still open\n> Date: Tue, 03 Oct 2000 21:28:36 +0200\n> From: Christof Petig <[email protected]>\n>\n> If you are looking for programs which tend to hold longstanding\n> transactions, this micro patch might be handy.\n> Whether it is of general use is debatable. It will replace 'idle' by\n> 'trans' if the backend is idle but a transaction is pending.\n>\n> Simply use ps to view the backend's status.\n>\n> Christof\n>\n> --- src/backend/commands/async.c~ Sun May 14 05:18:35 2000\n> +++ src/backend/commands/async.c Tue Oct 3 10:31:54 2000\n> @@ -818,7 +818,7 @@\n> */\n> pq_flush();\n>\n> - PS_SET_STATUS(\"idle\");\n> + PS_SET_STATUS(IsTransactionBlock()?\"trans\":\"idle\");\n> TPRINTF(TRACE_NOTIFY, \"ProcessIncomingNotify: done\");\n> }\n>\n> --- src/backend/tcop/postgres.c~ Thu Aug 31 09:18:57 2000\n> +++ src/backend/tcop/postgres.c Tue Oct 3 10:32:23 2000\n> @@ -1496,7 +1496,7 @@\n>\n> for (;;)\n> {\n> - PS_SET_STATUS(\"idle\");\n> + PS_SET_STATUS(IsTransactionBlock()?\"trans\":\"idle\");\n>\n> /* ----------------\n> * (1) tell the frontend we're ready for a new query.\n>\n>\n>\n\nMarc G. Fournier ICQ#7615664 IRC Nick: Scrappy\nSystems Administrator @ hub.org\nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org\n\n",
"msg_date": "Mon, 22 Jan 2001 09:00:49 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Small patch to replace 'idle' by 'trans' if\n\ttransactionis still open"
},
{
"msg_contents": "\nAny solid consensus on this?\n\n> \n> Well, I'm still for this ... how about something simiple like 'idle in\n> trans'? show, and easy to figure out what it means?\n> \n> On Mon, 22 Jan 2001, Christof Petig wrote:\n> \n> > If you don't know what is all about read the bottom (I'll requote my posting)\n> >\n> > Bruce Momjian wrote:\n> >\n> > > > On Fri, 19 Jan 2001, Bruce Momjian wrote:\n> > > > > Seems we decided against this. Sorry.\n> > > >\n> > > > Huh? from reading Tom's response, sounds like it would be something\n> > > > useful? I know I find having as much information about state in the ps\n> > > > listing helps me alot, and knowing if its 'idle' vs 'idle (in\n> > > > transaction)' provides at lesat more detailed information then just 'idle'\n> > >\n> > > > Who was the 'we' in the above decision? Tom seemed in favor of it, I know\n> > > > I'm in favor of it .. and you are not in favor of it ...\n> > >\n> > > There must have been some discussion about it. I don't see it in the\n> > > code, and I remember it was rejected for some reason. Check the archives.\n> >\n> > The thing which comes most close to a rejection was the 'I can't decide' mail\n> > by you (answered by Tom). The conclusion sounded like 'since we're not clear on\n> > this subject we won't touch this, yet'. And there was some unsettled discussion\n> > about the best wording to show in 'ps'.\n> >\n> > 'trans' seemed too short (and too much unlike 'idle') (as indicated by Bruce)\n> > and 'idle (open transaction)' might give difficulties on platforms which limit\n> > the length of the string (as indicated by Tom)\n> >\n> > I'll CC Hackers (where this discussion belongs)\n> >\n> > Christof\n> >\n> > ---------------------\n> >\n> > Quoting:\n> >\n> > Subject: Re: [PATCHES] Small patch to replace 'idle' by 'trans' if transaction\n> > is still open\n> > Date: Mon, 09 Oct 2000 22:46:56 -0400\n> > From: Tom Lane <[email protected]>\n> >\n> > Bruce Momjian <[email protected]> writes:\n> > > I can't decide if this is of general use. My inclination is that\n> > > someone in a transaction sitting a prompt should still show as idle.\n> >\n> > The idea seemed good to me, although I didn't look at the code to see\n> > if the implementation was any good ;-). I know we've frequently had\n> > questions on the lists where it was interesting to know if any\n> > transactions were being held open --- and right now there's no easy\n> > way to tell.\n> >\n> > regards, tom lane\n> >\n> > --------------\n> >\n> > Subject: Small patch to replace 'idle' by 'trans' if transaction is still open\n> > Date: Tue, 03 Oct 2000 21:28:36 +0200\n> > From: Christof Petig <[email protected]>\n> >\n> > If you are looking for programs which tend to hold longstanding\n> > transactions, this micro patch might be handy.\n> > Whether it is of general use is debatable. It will replace 'idle' by\n> > 'trans' if the backend is idle but a transaction is pending.\n> >\n> > Simply use ps to view the backend's status.\n> >\n> > Christof\n> >\n> > --- src/backend/commands/async.c~ Sun May 14 05:18:35 2000\n> > +++ src/backend/commands/async.c Tue Oct 3 10:31:54 2000\n> > @@ -818,7 +818,7 @@\n> > */\n> > pq_flush();\n> >\n> > - PS_SET_STATUS(\"idle\");\n> > + PS_SET_STATUS(IsTransactionBlock()?\"trans\":\"idle\");\n> > TPRINTF(TRACE_NOTIFY, \"ProcessIncomingNotify: done\");\n> > }\n> >\n> > --- src/backend/tcop/postgres.c~ Thu Aug 31 09:18:57 2000\n> > +++ src/backend/tcop/postgres.c Tue Oct 3 10:32:23 2000\n> > @@ -1496,7 +1496,7 @@\n> >\n> > for (;;)\n> > {\n> > - PS_SET_STATUS(\"idle\");\n> > + PS_SET_STATUS(IsTransactionBlock()?\"trans\":\"idle\");\n> >\n> > /* ----------------\n> > * (1) tell the frontend we're ready for a new query.\n> >\n> >\n> >\n> \n> Marc G. Fournier ICQ#7615664 IRC Nick: Scrappy\n> Systems Administrator @ hub.org\n> primary: [email protected] secondary: scrappy@{freebsd|postgresql}.org\n> \n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 23 Jan 2001 22:46:51 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Small patch to replace 'idle' by 'trans' if\n\ttransactionis still open"
},
{
"msg_contents": "Bruce Momjian <[email protected]> writes:\n> Any solid consensus on this?\n\nI'm for it (given a short status string --- \"idle in tx\" or \"idle in trans\"\nseem fine). Marc's for it. Who's against it?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 23 Jan 2001 23:41:10 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Small patch to replace 'idle' by 'trans' if\n\ttransactionis still open"
},
{
"msg_contents": "I liked the 'idle in trans' because people can search for just the first\nword if they want.\n\n> Bruce Momjian <[email protected]> writes:\n> > Any solid consensus on this?\n> \n> I'm for it (given a short status string --- \"idle in tx\" or \"idle in trans\"\n> seem fine). Marc's for it. Who's against it?\n> \n> \t\t\tregards, tom lane\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Wed, 24 Jan 2001 00:07:39 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Small patch to replace 'idle' by 'trans' if\n\ttransactionis still open"
},
{
"msg_contents": "Tom Lane writes:\n\n> Bruce Momjian <[email protected]> writes:\n> > Any solid consensus on this?\n>\n> I'm for it (given a short status string --- \"idle in tx\" or \"idle in trans\"\n> seem fine). Marc's for it. Who's against it?\n\n(I'm not.) If you're going to do \"idle in trans\" you might as well spell\nit out. The string will be truncated to the allowed length with no penalty.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Wed, 24 Jan 2001 16:53:05 +0100 (CET)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Small patch to replace 'idle' by 'trans' if\n\ttransactionis still open"
},
{
"msg_contents": "> Tom Lane writes:\n> \n> > Bruce Momjian <[email protected]> writes:\n> > > Any solid consensus on this?\n> >\n> > I'm for it (given a short status string --- \"idle in tx\" or \"idle in trans\"\n> > seem fine). Marc's for it. Who's against it?\n> \n> (I'm not.) If you're going to do \"idle in trans\" you might as well spell\n> it out. The string will be truncated to the allowed length with no penalty.\n\nOK, patch applied:\n\t\n\t#$ ps -ax|grep post\n\t27085 ?? Ss 0:00.05 ./bin/postmaster -B 384 -i -d0 -o -F -d0\n\t27108 ?? S 0:00.04 postgres test [local] idle (postmaster)\n\t27150 p7 S+ 0:00.01 grep post\n\t#$ ps -ax|grep post\n\t27085 ?? Is 0:00.05 ./bin/postmaster -B 384 -i -d0 -o -F -d0\n\t27108 ?? S 0:00.04 postgres test [local] idle in transaction (postmaster\n\t27185 p7 S+ 0:00.02 grep post\n\t#$ \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n\n? config.log\n? config.cache\n? config.status\n? GNUmakefile\n? src/Makefile.custom\n? src/GNUmakefile\n? src/Makefile.global\n? src/log\n? src/crtags\n? src/backend/postgres\n? src/backend/catalog/global.description\n? src/backend/catalog/global.bki\n? src/backend/catalog/template1.bki\n? src/backend/catalog/template1.description\n? src/backend/port/Makefile\n? src/bin/initdb/initdb\n? src/bin/initlocation/initlocation\n? src/bin/ipcclean/ipcclean\n? src/bin/pg_config/pg_config\n? src/bin/pg_ctl/pg_ctl\n? src/bin/pg_dump/pg_dump\n? src/bin/pg_dump/pg_restore\n? src/bin/pg_dump/pg_dumpall\n? src/bin/pg_id/pg_id\n? src/bin/pg_passwd/pg_passwd\n? src/bin/pgaccess/pgaccess\n? src/bin/pgtclsh/Makefile.tkdefs\n? src/bin/pgtclsh/Makefile.tcldefs\n? src/bin/pgtclsh/pgtclsh\n? src/bin/pgtclsh/pgtksh\n? src/bin/psql/psql\n? src/bin/scripts/createlang\n? src/include/config.h\n? src/include/stamp-h\n? src/interfaces/ecpg/lib/libecpg.so.3.2.0\n? src/interfaces/ecpg/preproc/ecpg\n? src/interfaces/libpgeasy/libpgeasy.so.2.1\n? src/interfaces/libpgtcl/libpgtcl.so.2.1\n? src/interfaces/libpq/libpq.so.2.1\n? src/interfaces/perl5/blib\n? src/interfaces/perl5/Makefile\n? src/interfaces/perl5/pm_to_blib\n? src/interfaces/perl5/Pg.c\n? src/interfaces/perl5/Pg.bs\n? src/pl/plperl/blib\n? src/pl/plperl/Makefile\n? src/pl/plperl/pm_to_blib\n? src/pl/plperl/SPI.c\n? src/pl/plperl/plperl.bs\n? src/pl/plpgsql/src/libplpgsql.so.1.0\n? src/pl/tcl/Makefile.tcldefs\nIndex: src/backend/tcop/postgres.c\n===================================================================\nRCS file: /home/projects/pgsql/cvsroot/pgsql/src/backend/tcop/postgres.c,v\nretrieving revision 1.204\ndiff -c -r1.204 postgres.c\n*** src/backend/tcop/postgres.c\t2001/01/24 14:32:32\t1.204\n--- src/backend/tcop/postgres.c\t2001/01/24 15:52:18\n***************\n*** 1791,1797 ****\n \n \t\tEnableNotifyInterrupt();\n \n! \t\tset_ps_display(\"idle\");\n \n \t\t/* Allow \"die\" interrupt to be processed while waiting */\n \t\tImmediateInterruptOK = true;\n--- 1791,1799 ----\n \n \t\tEnableNotifyInterrupt();\n \n! \t\tif (!IsTransactionBlock())\n! \t\t\tset_ps_display(\"idle\");\n! \t\telse\tset_ps_display(\"idle in transaction\");\n \n \t\t/* Allow \"die\" interrupt to be processed while waiting */\n \t\tImmediateInterruptOK = true;\nIndex: src/backend/utils/misc/postgresql.conf.sample\n===================================================================\nRCS file: /home/projects/pgsql/cvsroot/pgsql/src/backend/utils/misc/postgresql.conf.sample,v\nretrieving revision 1.2\ndiff -c -r1.2 postgresql.conf.sample\n*** src/backend/utils/misc/postgresql.conf.sample\t2000/12/03 21:12:19\t1.2\n--- src/backend/utils/misc/postgresql.conf.sample\t2001/01/24 15:52:18\n***************\n*** 18,20 ****\n--- 18,196 ----\n # Any option can also be given as a command line switch to the\n # postmaster, e.g., 'postmaster -c log_connections=on'. Some options\n # can be set at run-time with the 'SET' SQL command.\n+ \n+ \n+ \tenable_seqscan = true\n+ \tenable_indexscan = true\n+ \tenable_tidscan = true\n+ \tenable_sort = true\n+ \tenable_nestloop = true\n+ \tenable_mergejoin = true\n+ \tenable_hashjoin = true\n+ \n+ \tksqo = izer, false\n+ \tgeqo = true\n+ \n+ \ttcpip_socket = false\n+ \tssl = false\n+ \tfsync = true\n+ \tsilent_mode = false\n+ \n+ \tlog_connections = false\n+ \tlog_timestamp = false\n+ \tlog_pid = false\n+ \n+ #ifdef USE_ASSERT_CHECKING\n+ \tdebug_assertions = true\n+ #endif\n+ \n+ \tdebug_print_query = false\n+ \tdebug_print_parse = false\n+ \tdebug_print_rewritten = false\n+ \tdebug_print_plan = false\n+ \tdebug_pretty_print = false\n+ \n+ \tshow_parser_stats = false\n+ \tshow_planner_stats = false\n+ \tshow_executor_stats = false\n+ \tshow_query_stats = false\n+ #ifdef BTREE_BUILD_STATS\n+ \tshow_btree_build_stats = false\n+ #endif\n+ \n+ \ttrace_notify = false\n+ \n+ #ifdef LOCK_DEBUG\n+ \ttrace_locks = false\n+ \ttrace_userlocks = false\n+ \ttrace_spinlocks = false\n+ \tdebug_deadlocks = false\n+ #endif\n+ \n+ \thostname_lookup = false\n+ \tshow_source_port = false\n+ \n+ \tsql_inheritance = true\n+ \n+ \t{NULL, 0, NULL, false}\n+ };\n+ \n+ \n+ static struct config_int\n+ ConfigureNamesInt[] =\n+ {\n+ \tgeqo_threshold = PGC_USERSET, &geqo_rels,\n+ \t DEFAULT_GEQO_RELS, 2, INT_MAX\n+ \tgeqo_pool_size = PGC_USERSET, &Geqo_pool_size,\n+ \t DEFAULT_GEQO_POOL_SIZE, 0, MAX_GEQO_POOL_SIZE\n+ \tgeqo_effort = PGC_USERSET, &Geqo_effort,\n+ \t 1, 1, INT_MAX\n+ \tgeqo_generations = PGC_USERSET, &Geqo_generations,\n+ \t 0, 0, INT_MAX\n+ \tgeqo_random_seed = PGC_USERSET, &Geqo_random_seed,\n+ \t -1, INT_MIN, INT_MAX\n+ \n+ \tdeadlock_timeout = PGC_POSTMASTER, &DeadlockTimeout,\n+ \t 1000, 0, INT_MAX\n+ \n+ #ifdef ENABLE_SYSLOG\n+ \tsyslog = PGC_SIGHUP, &Use_syslog,\n+ \t 0, 0, 2\n+ #endif\n+ \n+ \t/*\n+ \t * Note: There is some postprocessing done in PostmasterMain() to\n+ \t * make sure the buffers are at least twice the number of\n+ \t * backends, so the constraints here are partially unused.\n+ \t */\n+ \tmax_connections = PGC_POSTMASTER, &MaxBackends,\n+ \t DEF_MAXBACKENDS, 1, MAXBACKENDS\n+ \tshared_buffers = PGC_POSTMASTER, &NBuffers,\n+ \t DEF_NBUFFERS, 16, INT_MAX\n+ \tport = PGC_POSTMASTER, &PostPortNumber,\n+ \t DEF_PGPORT, 1, 65535\n+ \n+ \tsort_mem = PGC_USERSET, &SortMem,\n+ \t 512, 1, INT_MAX\n+ \n+ \tdebug_level = PGC_USERSET, &DebugLvl,\n+ \t 0, 0, 16\n+ \n+ #ifdef LOCK_DEBUG\n+ \ttrace_lock_oidmin = PGC_SUSET, &Trace_lock_oidmin,\n+ \t BootstrapObjectIdData, 1, INT_MAX\n+ \ttrace_lock_table = PGC_SUSET, &Trace_lock_table,\n+ \t 0, 0, INT_MAX\n+ #endif\n+ \tmax_expr_depth = PGC_USERSET, &max_expr_depth,\n+ \t DEFAULT_MAX_EXPR_DEPTH, 10, INT_MAX\n+ \n+ \tunix_socket_permissions = PGC_POSTMASTER, &Unix_socket_permissions,\n+ \t 0777, 0000, 0777\n+ \n+ \tcheckpoint_timeout\",\tPGC_POSTMASTER,\t\t\t&CheckPointTimeout,\n+ \t 300, 30, 1800\n+ \n+ \twal_buffers\",\t\t\tPGC_POSTMASTER,\t\t\t&XLOGbuffers,\n+ \t 8, 4, INT_MAX\n+ \n+ \twal_files\",\t\t\tPGC_POSTMASTER,\t\t\t&XLOGfiles,\n+ \t 0, 0, 64\n+ \n+ \twal_debug\",\t\t\tPGC_SUSET,\t\t\t\t&XLOG_DEBUG,\n+ \t 0, 0, 16\n+ \n+ \tcommit_delay\",\t\tPGC_USERSET,\t\t\t&CommitDelay,\n+ \t 5, 0, 1000\n+ \n+ {NULL, 0, NULL, 0, 0, 0}\n+ };\n+ \n+ \n+ static struct config_real\n+ ConfigureNamesReal[] =\n+ {\n+ effective_cache_size = PGC_USERSET, &effective_cache_size,\n+ DEFAULT_EFFECTIVE_CACHE_SIZE, 0, DBL_MAX\n+ random_page_cost = PGC_USERSET, &random_page_cost,\n+ DEFAULT_RANDOM_PAGE_COST, 0, DBL_MAX\n+ cpu_tuple_cost = PGC_USERSET, &cpu_tuple_cost,\n+ DEFAULT_CPU_TUPLE_COST, 0, DBL_MAX\n+ cpu_index_tuple_cost = PGC_USERSET, &cpu_index_tuple_cost,\n+ DEFAULT_CPU_INDEX_TUPLE_COST, 0, DBL_MAX\n+ cpu_operator_cost = PGC_USERSET, &cpu_operator_cost,\n+ DEFAULT_CPU_OPERATOR_COST, 0, DBL_MAX\n+ \n+ geqo_selection_bias = PGC_USERSET, &Geqo_selection_bias,\n+ DEFAULT_GEQO_SELECTION_BIAS, MIN_GEQO_SELECTION_BIAS, MAX_GEQO_SELECTION_BIAS\n+ \n+ {NULL, 0, NULL, 0.0, 0.0, 0.0}\n+ };\n+ \n+ \n+ static struct config_string\n+ ConfigureNamesString[] =\n+ {\n+ \tkrb_server_keyfile = PGC_POSTMASTER, &pg_krb_server_keyfile,\n+ \t PG_KRB_SRVTAB, NULL\n+ \n+ \tunix_socket_group = PGC_POSTMASTER, &Unix_socket_group,\n+ \t \" = NULL\n+ \n+ #ifdef ENABLE_SYSLOG\n+ \tsyslog_facility = PGC_POSTMASTER,\t &Syslog_facility, \n+ \t\"LOCAL0 = check_facility\t \n+ \tsyslog_ident = PGC_POSTMASTER,\t &Syslog_ident, \n+ \t\"postgres = NULL\t \n+ #endif\n+ \n+ \tunix_socket_directory\",\t PGC_POSTMASTER, &UnixSocketDir,\n+ \t \" = NULL\n+ \n+ \tvirtual_host\",\t\t\t PGC_POSTMASTER,\t\t&VirtualHost,\n+ \t \" = NULL\n+ \n+ \t{NULL, 0, NULL, NULL, NULL}\n+ };\n+ \n+",
"msg_date": "Wed, 24 Jan 2001 10:53:43 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Small patch to replace 'idle' by 'trans' if\n\ttransactionis still open"
}
] |
[
{
"msg_contents": "\nI am trying to compile an app (compiles on *BSD, Linux) but for some\nreason I cant get it to compile on AIX 4.3.3.3. It seems not to like the\n\nlibcrypt ??\n\nusing:\ngcc version 2.95.2 19991024 (release)\nPostgreSQL 7.02\nGNU Make version 3.79.1\nld ??\nFlags: -I/usr/include/pgsql/ -lrts -lcrypt -L/usr/lib/psql/ -static\n-lecpg -lpq -lm\n\ngr_server.c:184: warning: passing arg 3 of `naccept' from incompatible\npointer type\nld: 0711-224 WARNING: Duplicate symbol: .connect\nld: 0711-345 Use the -bloadmap or -bnoquiet option to obtain more\ninformation.\nld: 0711-317 ERROR: Undefined symbol: __crypt_r\nld: 0711-317 ERROR: Undefined symbol: __setkey_r\nld: 0711-317 ERROR: Undefined symbol: __encrypt_r\ncollect2: ld returned 8 exit status\n\nany help would be great !!\n\n\nthank you\n\n\nnathan\n\n\n\n",
"msg_date": "Tue, 03 Oct 2000 13:36:45 -0700",
"msg_from": "Nathan Boeger <[email protected]>",
"msg_from_op": true,
"msg_subject": "AIX compile ?? libcrypt HELP !!"
},
{
"msg_contents": "Nathan Boeger <[email protected]> writes:\n> I am trying to compile an app (compiles on *BSD, Linux) but for some\n> reason I cant get it to compile on AIX 4.3.3.3. It seems not to like the\n> libcrypt ??\n\n> ld: 0711-317 ERROR: Undefined symbol: __crypt_r\n> ld: 0711-317 ERROR: Undefined symbol: __setkey_r\n> ld: 0711-317 ERROR: Undefined symbol: __encrypt_r\n\nI'll bet AIX doesn't have the \"reentrant\" versions crypt_r(), setkey_r(),\netc. Did you check its man page for the crypt functions?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 03 Oct 2000 19:34:53 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AIX compile ?? libcrypt HELP !! "
},
{
"msg_contents": "Tom Lane wrote:\n\n> Nathan Boeger <[email protected]> writes:\n> > I am trying to compile an app (compiles on *BSD, Linux) but for some\n> > reason I cant get it to compile on AIX 4.3.3.3. It seems not to like the\n> > libcrypt ??\n>\n> > ld: 0711-317 ERROR: Undefined symbol: __crypt_r\n> > ld: 0711-317 ERROR: Undefined symbol: __setkey_r\n> > ld: 0711-317 ERROR: Undefined symbol: __encrypt_r\n>\n> I'll bet AIX doesn't have the \"reentrant\" versions crypt_r(), setkey_r(),\n> etc. Did you check its man page for the crypt functions?\n>\n> regards, tom lane\n\nwell, no\n\nbut... I did however link agianst the ssl libcrypto (-L/usr/local/ssl/lib\n-lcrypto ) and it works...\n\nwhy ?\n\nthank you for the response !!\n\nnathan\n\n",
"msg_date": "Tue, 03 Oct 2000 22:20:55 -0700",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: AIX compile ?? libcrypt HELP !!"
}
] |
[
{
"msg_contents": ">> If anyone has suggestions/additions to this list it would be greatly\n>> appreciated!\n\n>\tI'm very interested in helping with this (it is on my list of\nprojects\n>\tand I've managed to stop adding new ones to the list for now).\nShould\n>\twe take this off-list, or keep it here?\n\nWe could move it to pgsql-interfaces or setup a list on one of the free,\nweb-based list engines.\n\nJoel\n",
"msg_date": "Tue, 3 Oct 2000 17:15:29 -0400 ",
"msg_from": "\"Clark, Joel\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: C++ client libs"
}
] |
[
{
"msg_contents": "\nHmm, no response on pgsql-general. Anyone up on\noracle's implementation?\n\nelein\n==================\nElein Mustain\[email protected]\n(510)420-3120 \n\n-----Original Message-----\nFrom: elein [mailto:[email protected]]\nSent: Wednesday, September 27, 2000 6:41 PM\nTo: [email protected]\nCc: elein\nSubject: oracle ate\n\n\n\nDoes anyone have a writeup or deconstruction of the\nimplementation of Oracle's extensibility capabilities?\n\nI am a long time ordbms proponent (illustra, informix and\nnow postgresql). And I have heard mixed information\nregarding Oracle's extensibility. Do they use a unified \ntype system? Are cartriges separate processes? Do the \nseparate processes (if they are) share memory with the \nserver process or is all the communication through pipish \nthings. If I don't use their standard cartridges (image, \ntext, video) how easy is it to write my own. Any GIS \nspecific pro or con info w/Oracle? Am I asking the right\nquestions?\n\nI would like to put the questions to rest and get on with it.\n\nFeel free to point me to the right audience or source if\nthis is not a good question for this group.\n\nthanks\n\[email protected]\n",
"msg_date": "Tue, 3 Oct 2000 17:47:55 -0700",
"msg_from": "\"elein\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "FW: oracle ate"
},
{
"msg_contents": "At 05:47 PM 10/3/00 -0700, elein wrote:\n>\n>Hmm, no response on pgsql-general. Anyone up on\n>oracle's implementation?\n\nYou can subscribe to Oracle's Technet on the web for free (start\nat www.oracle.com) and read all their manuals on-line.\n\nAFAIK, Oracle themselves are somewhat up on Oracle's implementation\nof various things, so reading their FREE on-line documentation would\nseem like a good place to start.\n\nIf you can't find what you want there, FREE discussion forums not\nterribly unlike this one are filled with Oracle drones eager to ask\nyour questions.\n\nIf you still can't get the answer you need, download a FREE Oracle\ndeveloper's copy, install it, and test yourself.\n\nAs a fervent Postgres supporter, I happen to believe in availing myself\nof such FREE Oracle services as much as possible, partly because I use it\nin one sphere of my professional (paid) life, but also because they are\nsuch a fertile source of information on Oracle.\n\nTrust me - you'll learn more about Oracle from Oracle docs, Oracle \nforums, and free copies of Oracle than you'll learn here.\n\nJust as clearly, you'll learn a lot more about Postgres here...\n\n\n\n- Don Baccus, Portland OR <[email protected]>\n Nature photos, on-line guides, Pacific Northwest\n Rare Bird Alert Service and other goodies at\n http://donb.photo.net.\n",
"msg_date": "Tue, 03 Oct 2000 20:13:52 -0700",
"msg_from": "Don Baccus <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: FW: oracle ate"
}
] |
[
{
"msg_contents": "Hi all,\n\nJust wondering if anyone has already created a version data type where\n1.100 > 1.10 > 1.9 > 1.01 etc...\n\nI don't want to reinvent the wheel...\n\nI was thinking that an array with a '.' as the delimeter would work\nbest, but I don't know how ORDER BY 's work on arrays....\n\nThank You\n\nAshley Cambrell\n\n",
"msg_date": "Wed, 04 Oct 2000 12:22:58 +1100",
"msg_from": "Ashley Cambrell <[email protected]>",
"msg_from_op": true,
"msg_subject": "Version data type."
},
{
"msg_contents": "Hello all,\n\nI thought I would give it a try anyways...\n\nThis data type is based on Garrett A. Wollman isbn / issn code\n\nIt can be downloaded from:\n\nftp://ftp.freaky-namuh.com/pub/devel/Postgresql/version_number/version_number-0.2.tar.gz\n\nThe README file explains most of it.\n\nIt's not perfect, but it's still useful... (and I don't know enough\nabout Postgres's internals to know if it's \"done right\" (tm).)\n\n\nAshley Cambrell\n\n\nAshley Cambrell wrote:\n\n> Hi all,\n>\n> Just wondering if anyone has already created a version data type where\n> 1.100 > 1.10 > 1.9 > 1.01 etc...\n>\n> I don't want to reinvent the wheel...\n>\n> I was thinking that an array with a '.' as the delimeter would work\n> best, but I don't know how ORDER BY 's work on arrays....\n>\n> Thank You\n>\n> Ashley Cambrell\n\n",
"msg_date": "Thu, 05 Oct 2000 02:30:41 +1100",
"msg_from": "Ashley Cambrell <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Version data type. - DONE"
}
] |
[
{
"msg_contents": ">I was wondering if anyone could offer me some pointers in extending the\n>C++ classes used for client applications? I'm interested in developing a\n>set of C++ classes which will interact directly with the postgres\n>back-end with the following enhancements to the current implementation:\n\n>These are preliminary musings and definitely open to change.\n\n>1) Implement connections as a stream object. e.g. \n> db = new pg_database(char * connection_params);\n> db << sql_string;\n\nI wouldn't have thought of using the stream operators this way, but I\nguess it sounds reasonable. One tricky bit, though, how do you distinguish\nsynchronous and asynchronous use this way? In the synchronous case, do you\nexpect some sort of return value from this process? Where do you get it \nfrom? In the asynchronous case, how do you prevent a new query being sent\ndown the socket while a previous one is still working away and\ntransmitting results?\n\n>2) Implement results as a returned object from pg_database\n> result = db.exec();\n> result >> field1 >> field2 >> field3;\n> and\n> field = result[\"field\"];\n> field = result[fieldnum];\n\nWhat type is field? Is it always string, or can it be int, double, time_t\netc? And presumably you want to specify the row in there somewhere (maybe\nit defaults to zero); or is field an array of some sort?\n\n>3) Implement asynchronous calls to the back-end permitting multiple\n> queries to be executed concurrently\n\nImplement async calls, yes, definitely a good thing to do. But this by\nitself doesn't buy you multiple queries concurrently; for that you need\nmultiple connections to the database, which would also be worth doing in\nthis library.\n\nBTW, for async calls, you would want to create some convenient way of\nregistering your interest in the socket connection, and some standard\nmethods that read the result from it. Presuming the normal use for this is\nvia a select-driven event loop, you want to be able to register the socket\nwith whatever code drives the select, and register a callback that looks\nat the socket whenever select goes off. Given that there are umpteen dozen\nimplementations of a select-driven event loop, it will be a challenge\ndeveloping something that works nicely with all of them.\n\nIt'd also be good to find out where Chris Bitmead is up to with his work\non a streaming version of libpq, since that would be a bit tricky to\nhandle here, but buys you a lot of useful functionality and much more\nefficient memory usage, so it would be worthwhile.\n\n>4) Implement a standard set of exceptions which can be thrown \n> ex. if connection to back-end is interrupted or cannot be established\n> within 30 seconds of the first attempt, throw.\n\nFair enough, but I'd suggest you make this an optional feature, since not\neveryone wants to use exceptions. Make sure there's some way to switch\nexceptions off (or better, some way to not switch exceptions on).\n\n>5) I'm looking at making the libraries portable between FreeBSD, Solaris \n> Linux and Win32.\n\n>If anyone has suggestions/additions to this list it would be greatly\n>appreciated!\n\nI've developed something vaguely similar for our internal use, which\naddresses some of the above (particularly async calls); unfortunately it's\nheavily tied to other libraries we use in-house, so not of much use\ngenerally. I did have a plan at the back of my mind to one of these days\nproduce a more independent version of our library, and go about persuading\nfolks here that we should contribute it to PostgreSQL, but if your effort\ngoes ahead successfully then that won't be necessary.\n\nBTW, I'd urge you to keep the discussion public. The interfaces list might\nbe a good place, rather than hackers, but don't go hiding away in private\nemail or some other obscure place. A public discussion will get you useful\ninput from a much wider range of people. And it would make a welcome\nchange from the usual \"Help! Help! MS Access has made all my fields\nread-only!\" and \"Why doesn't the JDBC driver implement this?\" questions on\nthe interfaces list :-).\n\n>Cheers,\n>\n>Randy Jonasz\n\n\nTim Allen\n\n--\n-----------------------------------------------\nTim Allen [email protected]\nProximity Pty Ltd http://www.proximity.com.au/\n http://www4.tpg.com.au/users/rita_tim/\n\n",
"msg_date": "Wed, 4 Oct 2000 12:38:35 +1100 (EST)",
"msg_from": "Tim Allen <[email protected]>",
"msg_from_op": true,
"msg_subject": "C++ client libs"
},
{
"msg_contents": "Tim Allen <[email protected]> writes:\n> BTW, I'd urge you to keep the discussion public. The interfaces list might\n> be a good place, rather than hackers, but don't go hiding away in private\n> email or some other obscure place.\n\n100% agreement here. pgsql-interfaces seems like the appropriate\ndiscussion forum. When and if the C++ discussion threatens to overwhelm\nother traffic there, I'm sure Marc would be amenable to opening up a\nnew list --- but for now, interfaces is the right place.\n\nAs for places to host the code, I believe both pgsql.com and\ngreatbridge.com have plans to open up CVS servers for Postgres-related\nprojects. I don't think either is quite open for business yet, but\ntry pounding on the door and see what happens...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 03 Oct 2000 23:33:25 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [INTERFACES] C++ client libs "
},
{
"msg_contents": "On Wed, Oct 04, 2000 at 12:38:35PM +1100, Tim Allen wrote:\n\n> >3) Implement asynchronous calls to the back-end permitting multiple\n> > queries to be executed concurrently\n> \n> Implement async calls, yes, definitely a good thing to do. But this by\n> itself doesn't buy you multiple queries concurrently; for that you need\n> multiple connections to the database, which would also be worth doing in\n> this library.\n> \n> BTW, for async calls, you would want to create some convenient way of\n> registering your interest in the socket connection, and some standard\n> methods that read the result from it. Presuming the normal use for this is\n> via a select-driven event loop, you want to be able to register the socket\n> with whatever code drives the select, and register a callback that looks\n> at the socket whenever select goes off. Given that there are umpteen dozen\n> implementations of a select-driven event loop, it will be a challenge\n> developing something that works nicely with all of them.\n\n\tMy preference would be an easily-overridable function so I could\nsubclass the parent object and drop in a function to send a BMessage\nto an arbitrary BLooper. Your messaging system may vary...\n\n\n-- \nAdam Haberlach | A billion hours ago, human life appeared on\[email protected] | earth. A billion minutes ago, Christianity\nhttp://www.newsnipple.com | emerged. A billion Coca-Colas ago was\n'88 EX500 | yesterday morning. -1996 Coca-Cola Ann. Rpt.\n",
"msg_date": "Tue, 3 Oct 2000 20:40:40 -0700",
"msg_from": "Adam Haberlach <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: C++ client libs"
}
] |
[
{
"msg_contents": "Hi,\n\nI'm in a situation where I urgently need to debug PostgreSQL 7.0.2 \nfor deadlocks that it does not notice/timeout\n\nWhere can I find info about running several concurrent backends \nunder a debugger ?\n\n-----------\nHannu\n",
"msg_date": "Wed, 04 Oct 2000 08:42:05 +0300",
"msg_from": "Hannu Krosing <[email protected]>",
"msg_from_op": true,
"msg_subject": "I want tips for debugging deadlocks"
},
{
"msg_contents": "Hannu Krosing <[email protected]> writes:\n> I'm in a situation where I urgently need to debug PostgreSQL 7.0.2 \n> for deadlocks that it does not notice/timeout\n\nThe most likely bet is that you are seeing deadlocks that involve a\nbuffer spinlock (LockBuffer() in bufmgr.c) --- there's no timeout or\ndeadlock detection check in that code. I have been suspicious for\nsome time that there are deadlocks possible there, but haven't had\nany luck getting a reproducible example to study. (If you can present\na reproducible way to make the problem happen, please post it!)\n\n> Where can I find info about running several concurrent backends \n> under a debugger ?\n\nJust fire up N backends and attach to each one with N instances of gdb.\nIt's a little confusing but I've done it ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 04 Oct 2000 11:08:13 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: I want tips for debugging deadlocks "
},
{
"msg_contents": "By the way, we finally understood that our main problem,\nthe one that was making our Pg hang forever, comes from\na deadlock problem. Same as Hannu's one.\n\nThere are no deadlock detection, indeed. Good DBAs, or\nDBAs working with good coders, will never come across\nthe problem :) but we did :(\n\nI think a trace log being sent to the DBA would be\na great thing when a deadlock is detected, with if\npossible the query that cannot be executed. Oracle\ndoes a good job, there.\n\nFabrice\n\nTom Lane wrote:\n> \n> Hannu Krosing <[email protected]> writes:\n> > I'm in a situation where I urgently need to debug PostgreSQL 7.0.2\n> > for deadlocks that it does not notice/timeout\n> \n> The most likely bet is that you are seeing deadlocks that involve a\n> buffer spinlock (LockBuffer() in bufmgr.c) --- there's no timeout or\n> deadlock detection check in that code. I have been suspicious for\n> some time that there are deadlocks possible there, but haven't had\n> any luck getting a reproducible example to study. (If you can present\n> a reproducible way to make the problem happen, please post it!)\n> \n> > Where can I find info about running several concurrent backends\n> > under a debugger ?\n> \n> Just fire up N backends and attach to each one with N instances of gdb.\n> It's a little confusing but I've done it ...\n> \n> regards, tom lane\n",
"msg_date": "Wed, 04 Oct 2000 17:48:03 +0200",
"msg_from": "Fabrice Scemama <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: I want tips for debugging deadlocks"
},
{
"msg_contents": "Fabrice Scemama <[email protected]> writes:\n> By the way, we finally understood that our main problem,\n> the one that was making our Pg hang forever, comes from\n> a deadlock problem. Same as Hannu's one.\n\n> There are no deadlock detection, indeed. Good DBAs, or\n> DBAs working with good coders, will never come across\n> the problem :) but we did :(\n\nThe reason LockBuffer() has no deadlock detection is that it's not\nsupposed to be possible for a deadlock to occur there, user mistake\nor no. So I consider this a bug --- either there is an actual logic\nerror somewhere, or else the supposition is wrong and we need to add\ndeadlock handling.\n\nIf you can describe what you were doing so that the behavior can be\nreproduced, it'd be very helpful.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 04 Oct 2000 12:08:27 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: I want tips for debugging deadlocks "
}
] |
[
{
"msg_contents": "I just wanted to repost this one more time in case developers didn't\ncatch it. I have a reliable way to make postgresql crash after a\ncouple of hours over here and a backtrace that looks like a good\ncatch.\n\nMy apologies if this one time to many, I won't be posting it again.\n\nthanks for your time,\n-Alfred\n\n----- Forwarded message from Alfred Perlstein <[email protected]> -----\n\nFrom: Alfred Perlstein <[email protected]>\nTo: [email protected]\nCc: Tom Lane <[email protected]>\nSubject: [HACKERS] more crashes\nDate: Mon, 2 Oct 2000 15:17:12 -0700\nMessage-ID: <[email protected]>\nUser-Agent: Mutt/1.2.4i\nX-Mailing-List: [email protected]\nSender: [email protected]\n\nThis time I'm pretty sure I caught the initial crash during an update:\n\nI disabled the vacuum analyze and still got table corruption with a crash:\n\ntwo crashdumps of 7.0.2+somepatches\n\n * $Header: /home/pgcvs/pgsql/src/backend/access/common/heaptuple.c,v 1.6\n2 2000/04/12 17:14:36 momjian Exp $\n\nProgram terminated with signal 11, Segmentation fault.\nReading symbols from /usr/lib/libcrypt.so.2...done.\nReading symbols from /usr/lib/libm.so.2...done.\nReading symbols from /usr/lib/libutil.so.3...done.\nReading symbols from /usr/lib/libreadline.so.4...done.\nReading symbols from /usr/lib/libncurses.so.5...done.\nReading symbols from /usr/lib/libc.so.4...done.\nReading symbols from /usr/libexec/ld-elf.so.1...done.\n#0 0x8063aa7 in nocachegetattr (tuple=0x84ae9fc, attnum=4, \n tupleDesc=0x84a6368, isnull=0x84afc20 \"\") at heaptuple.c:537\n537 off = att_addlength(off, att[i]->attlen, tp + off);\n(gdb) bt\n#0 0x8063aa7 in nocachegetattr (tuple=0x84ae9fc, attnum=4, \n tupleDesc=0x84a6368, isnull=0x84afc20 \"\") at heaptuple.c:537\n#1 0x80a027f in ExecEvalVar (variable=0x84974b0, econtext=0x84aedd8, \n isNull=0x84afc20 \"\") at execQual.c:314\n#2 0x80a0d97 in ExecEvalExpr (expression=0x84974b0, econtext=0x84aedd8, \n isNull=0x84afc20 \"\", isDone=0xbfbfe6db \"\\001��J\\b+�\\021\\b\\004�J\\b\\001\")\n at execQual.c:1214\n#3 0x80a090a in ExecEvalFuncArgs (fcache=0x84afc38, econtext=0x84aedd8, \n argList=0x84974d8, argV=0xbfbfe6dc, \n argIsDone=0xbfbfe6db \"\\001��J\\b+�\\021\\b\\004�J\\b\\001\") at execQual.c:635\n#4 0x80a09c1 in ExecMakeFunctionResult (node=0x8496a40, arguments=0x84974d8, \n econtext=0x84aedd8, isNull=0xbfbfe7db \"\", \n isDone=0xbfbfe75b \"\\b\\214翿\\027\\016\\n\\bHuI\\b��J\\b�翿X\\017B`\\004\")\n at execQual.c:711\n#5 0x80a0b37 in ExecEvalOper (opClause=0x8497548, econtext=0x84aedd8, \n isNull=0xbfbfe7db \"\") at execQual.c:902\n#6 0x80a0e17 in ExecEvalExpr (expression=0x8497548, econtext=0x84aedd8, \n isNull=0xbfbfe7db \"\", isDone=0xbfbfe7e0 \"\\001�\\016\\b\") at execQual.c:1249\n#7 0x80a1011 in ExecTargetList (targetlist=0x8497fd8, nodomains=6, \n targettype=0x84aefb0, values=0x84aee48, econtext=0x84aedd8, \n isDone=0xbfbfe90b \"\\001,鿿.K\\n\\bP�J\\b\\214H\\n\\b<鿿�A\\023\\b\\030�H\\b \")\n at execQual.c:1511\n#8 0x80a12af in ExecProject (projInfo=0x84aee20, \n isDone=0xbfbfe90b \"\\001,鿿.K\\n\\bP�J\\b\\214H\\n\\b<鿿�A\\023\\b\\030�H\\b \")\n at execQual.c:1721\n#9 0x80a1365 in ExecScan (node=0x84add50, accessMtd=0x80a488c <IndexNext>)\n at execScan.c:155\n#10 0x80a4b2e in ExecIndexScan (node=0x84add50) at nodeIndexscan.c:288\n#11 0x809fb6d in ExecProcNode (node=0x84add50, parent=0x84add50)\n at execProcnode.c:272\n#12 0x809ed59 in ExecutePlan (estate=0x84ae8a0, plan=0x84add50, \n operation=CMD_UPDATE, offsetTuples=0, numberTuples=0, \n direction=ForwardScanDirection, destfunc=0x84afaf0) at execMain.c:1052\n#13 0x809e2ba in ExecutorRun (queryDesc=0x84ae888, estate=0x84ae8a0, \n feature=3, limoffset=0x0, limcount=0x0) at execMain.c:327\n#14 0x80f92ca in ProcessQueryDesc (queryDesc=0x84ae888, limoffset=0x0, \n limcount=0x0) at pquery.c:310\n#15 0x80f9347 in ProcessQuery (parsetree=0x84965d0, plan=0x84add50, \n dest=Remote) at pquery.c:353\n#16 0x80f7ef0 in pg_exec_query_dest (\n query_string=0x81a9370 \"\\nUPDATE\\n webhit_details_formatted\\nSET\\n attr_hits = attr_hits + '1' \\nWHERE\\n counter_id = '11909'\\n AND attr_type = 'ATTR_OPERATINGSYS'\\n AND attr_name = 'win95'\\n AND attr_vers = '0'\\n;\", \n dest=Remote, aclOverride=0) at postgres.c:663\n#17 0x80f7db9 in pg_exec_query (\n query_string=0x81a9370 \"\\nUPDATE\\n webhit_details_formatted\\nSET\\n attr_hits = attr_hits + '1' \\nWHERE\\n counter_id = '11909'\\n AND attr_type = 'ATTR_OPERATINGSYS'\\n AND attr_name = 'win95'\\n AND attr_vers = '0'\\n;\")\n at postgres.c:562\n#18 0x80f8d1a in PostgresMain (argc=9, argv=0xbfbff0dc, real_argc=10, \n real_argv=0xbfbffb3c) at postgres.c:1590\n#19 0x80e1d06 in DoBackend (port=0x843f400) at postmaster.c:2009\n#20 0x80e1899 in BackendStartup (port=0x843f400) at postmaster.c:1776\n#21 0x80e0abd in ServerLoop () at postmaster.c:1037\n#22 0x80e04be in PostmasterMain (argc=10, argv=0xbfbffb3c) at postmaster.c:725\n#23 0x80aee43 in main (argc=10, argv=0xbfbffb3c) at main.c:93\n#24 0x80633c5 in _start ()\n(gdb) list\n532 \n533 if (usecache)\n534 att[i]->attcacheoff = off;\n535 }\n536 \n537 off = att_addlength(off, att[i]->attlen, tp + off);\n538 \n539 if (usecache &&\n540 att[i]->attlen == -1 && !VARLENA_FIXED_SIZE(att[i]))\n541 usecache = false;\n(gdb) print off\n$1 = 772814392\n(gdb) print att[i]->attlen\n$2 = -1\n(gdb) print off\n$3 = 772814392\n(gdb) print tp\n$4 = 0x5eab73d0 \"\\205.\"\n(gdb) print tp+off\n$7 = 0x8cbbaa08 <Address 0x8cbbaa08 out of bounds>\n(gdb) print usecache\n$8 = 0 '\\000'\n(gdb) print !VARLENA_FIXED_SIZE(att[i])\nNo symbol \"VARLENA_FIXED_SIZE\" in current context.\n(gdb) print att[i]\n$9 = 0x84a66c8\n(gdb) print *(att[i]) \n$10 = {attrelid = 3518994475, attname = {\n data = \"attr_vers\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 1043, \n attdisbursion = 0.125293151, attlen = -1, attnum = 4, attnelems = 0, \n attcacheoff = -1, atttypmod = 36, attbyval = 0 '\\000', attstorage = 112 'p', \n attisset = 0 '\\000', attalign = 105 'i', attnotnull = 0 '\\000', \n atthasdef = 0 '\\000'}\n(gdb) print i\n$11 = 3\n(gdb) print *(att[0])\n$12 = {attrelid = 3518994475, attname = {\n data = \"counter_id\", '\\000' <repeats 21 times>, \n alignmentDummy = 1853189987}, atttypid = 23, \n attdisbursion = 0.000228356235, attlen = 4, attnum = 1, attnelems = 0, \n attcacheoff = 0, atttypmod = -1, attbyval = 1 '\\001', attstorage = 112 'p', \n attisset = 0 '\\000', attalign = 105 'i', attnotnull = 0 '\\000', \n atthasdef = 0 '\\000'}\n(gdb) print *(att[1])\n$13 = {attrelid = 3518994475, attname = {\n data = \"attr_type\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 1043, \n attdisbursion = 0.0928893909, attlen = -1, attnum = 2, attnelems = 0, \n attcacheoff = 4, atttypmod = 36, attbyval = 0 '\\000', attstorage = 112 'p', \n attisset = 0 '\\000', attalign = 105 'i', attnotnull = 0 '\\000', \n atthasdef = 0 '\\000'}\n(gdb) print *(att[2])\n$14 = {attrelid = 3518994475, attname = {\n data = \"attr_name\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 1043, \n attdisbursion = 0.370779663, attlen = -1, attnum = 3, attnelems = 0, \n attcacheoff = -1, atttypmod = 36, attbyval = 0 '\\000', attstorage = 112 'p', \n attisset = 0 '\\000', attalign = 105 'i', attnotnull = 0 '\\000', \n atthasdef = 0 '\\000'}\n(gdb) print attnum\n$15 = 4\n(gdb) print *(att[4])\n$16 = {attrelid = 3518994475, attname = {\n data = \"attr_hits\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 20, attdisbursion = 0.0573871136, \n attlen = 8, attnum = 5, attnelems = 0, attcacheoff = -1, atttypmod = -1, \n attbyval = 0 '\\000', attstorage = 112 'p', attisset = 0 '\\000', \n attalign = 100 'd', attnotnull = 0 '\\000', atthasdef = 1 '\\001'}\n\n\n--------------------------------------------------\n\n\nI'm pretty sure this is a pg_dump that died when the fist crash\nhappened above:\n\n * $Header: /home/pgcvs/pgsql/src/backend/commands/copy.c,v 1.106.2.2 2000/06/28 06:13:01 tgl Exp $\n\n\nProgram terminated with signal 10, Bus error.\n#0 0x482a7d95 in ?? (?? )\n1 0x808c393 in CopyTo (rel=0x84e7890, binary=0 '\\000', oids=0 '\\000', \n fp=0x0, delim=0x8159fa9 \"\\t\", null_print=0x8159fab \"\\\\N\") at copy.c:508\n#2 0x808bf99 in DoCopy (relname=0x84930e8 \"\", binary=0 '\\000', oids=0 '\\000', \n from=0 '\\000', pipe=1 '\\001', filename=0x0, delim=0x8159fa9 \"\\t\", \n null_print=0x8159fab \"\\\\N\") at copy.c:374\n#3 0x80f98a3 in ProcessUtility (parsetree=0x8493110, dest=Remote)\n at utility.c:262\n#4 0x80f7e5e in pg_exec_query_dest (query_string=0x81a9370 \"\", dest=Remote, \n aclOverride=0) at postgres.c:617\n#5 0x80f7db9 in pg_exec_query (query_string=0x81a9370 \"\") at postgres.c:562\n#6 0x80f8d1a in PostgresMain (argc=9, argv=0xbfbff0bc, real_argc=10, \n real_argv=0xbfbffb1c) at postgres.c:1590\n#7 0x80e1d06 in DoBackend (port=0x843f000) at postmaster.c:2009\n#8 0x80e1899 in BackendStartup (port=0x843f000) at postmaster.c:1776\n#9 0x80e0abd in ServerLoop () at postmaster.c:1037\n#10 0x80e04be in PostmasterMain (argc=10, argv=0xbfbffb1c) at postmaster.c:725\n#11 0x80aee43 in main (argc=10, argv=0xbfbffb1c) at main.c:93\n#12 0x80633c5 in _start ()\n(gdb) up\n#1 0x808c393 in CopyTo (rel=0x84e7890, binary=0 '\\000', oids=0 '\\000', \n fp=0x0, delim=0x8159fa9 \"\\t\", null_print=0x8159fab \"\\\\N\") at copy.c:508\n508 string = (char *) (*fmgr_faddr(&out_functions[i]))\n(gdb) print out_functions[i]\n$1 = {fn_addr = 0, fn_plhandler = 0, fn_oid = 0, fn_nargs = 0}\n(gdb) print i\n$2 = 2\n(gdb) print isnull\n$3 = 0 '\\000'\n(gdb) print tupDesc\nNo symbol \"tupDesc\" in current context.\n(gdb) print tuple\n$4 = 0x8493268\n(gdb) print *tuple\n$5 = {t_len = 0, t_self = {ip_blkid = {bi_hi = 0, bi_lo = 0}, ip_posid = 0}, \n t_datamcxt = 0x0, t_data = 0x0}\n(gdb) print value\n$6 = 1072927316\n(gdb) print *value\nCannot access memory at address 0x3ff39254.\n(gdb) print oids\n$7 = 0 '\\000'\n(gdb) print binary\n$8 = 0 '\\000'\n(gdb) print string\n$9 = 0xfffffffc <Address 0xfffffffc out of bounds>\n\nNow I think I have the intial spot where it all goes to pot (the\ninitial traceback). I really appreciate the continued help and\npointers that I've been given and was wondering if someone could\nhelp me out a bit more.\n\nsorry for being such a pain and if any other info is needed please ask.\n\nthanks for you time,\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n\n----- End forwarded message -----\n\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Wed, 4 Oct 2000 00:06:25 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": true,
"msg_subject": "(forw) more crashes"
},
{
"msg_contents": "Alfred Perlstein <[email protected]> writes:\n> I have a reliable way to make postgresql crash after a\n> couple of hours over here and a backtrace that looks like a good\n> catch.\n\nI'm interested in pursuing this, but the backtrace doesn't give enough\ninfo to debug it. It looks like the backend is crashing because of\na previously-corrupted tuple, so what we'll need to do is work backwards\nto find where the data corruption is occurring.\n\nCan you boil down the test sequence to something that could be\nreproduced by other people? The most convenient way to work on it\nwould be to see it happen here...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 04 Oct 2000 12:54:11 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: (forw) more crashes "
},
{
"msg_contents": "* Tom Lane <[email protected]> [001004 09:56] wrote:\n> Alfred Perlstein <[email protected]> writes:\n> > I have a reliable way to make postgresql crash after a\n> > couple of hours over here and a backtrace that looks like a good\n> > catch.\n> \n> I'm interested in pursuing this, but the backtrace doesn't give enough\n> info to debug it. It looks like the backend is crashing because of\n> a previously-corrupted tuple, so what we'll need to do is work backwards\n> to find where the data corruption is occurring.\n> \n> Can you boil down the test sequence to something that could be\n> reproduced by other people? The most convenient way to work on it\n> would be to see it happen here...\n\nI just wanted to note on the list that these crashes seem to have\nstopped with the latest 7.0.2-patches (as of 11:30ish PM EST Oct,\n4th), it's been over 24 hours since the upgrade (previously I\ncouldn't go for more than 20 without a crash).\n\nMy only concern is that I didn't notice anything on the cvs list\nthat referenced a fix for crashes.\n\nWell anyhow I'll post an update in a couple of days if all is well\nor not.\n\nthanks for you time,\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Fri, 6 Oct 2000 15:59:43 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: (forw) more crashes"
},
{
"msg_contents": "* Alfred Perlstein <[email protected]> [001006 16:02] wrote:\n> * Tom Lane <[email protected]> [001004 09:56] wrote:\n> > Alfred Perlstein <[email protected]> writes:\n> > > I have a reliable way to make postgresql crash after a\n> > > couple of hours over here and a backtrace that looks like a good\n> > > catch.\n> > \n> > I'm interested in pursuing this, but the backtrace doesn't give enough\n> > info to debug it. It looks like the backend is crashing because of\n> > a previously-corrupted tuple, so what we'll need to do is work backwards\n> > to find where the data corruption is occurring.\n> > \n> > Can you boil down the test sequence to something that could be\n> > reproduced by other people? The most convenient way to work on it\n> > would be to see it happen here...\n> \n> I just wanted to note on the list that these crashes seem to have\n> stopped with the latest 7.0.2-patches (as of 11:30ish PM EST Oct,\n> 4th), it's been over 24 hours since the upgrade (previously I\n> couldn't go for more than 20 without a crash).\n> \n> My only concern is that I didn't notice anything on the cvs list\n> that referenced a fix for crashes.\n> \n> Well anyhow I'll post an update in a couple of days if all is well\n> or not.\n\nUnfortunatly I'm still getting crashes, this one looks like it's\nduring a vacuum, previously I got a crash while doing an UPDATE, but\nin exactly the same spot, it took quite a bit longer to provoke this\ntime:\n\n-rw------- 1 pgsql pgsql 277561344 Oct 8 02:56 postgres.core\n\n\n#0 0x8063c8b in nocachegetattr (tuple=0xbfbfe974, attnum=3,\n tupleDesc=0x84ca368, isnull=0xbfbfe7fb \"\") at heaptuple.c:537\n537 off = att_addlength(off, att[i]->attlen, tp + off);\n(gdb) bt\n#0 0x8063c8b in nocachegetattr (tuple=0xbfbfe974, attnum=3, \n tupleDesc=0x84ca368, isnull=0xbfbfe7fb \"\") at heaptuple.c:537\n#1 0x8075851 in GetIndexValue (tuple=0xbfbfe974, hTupDesc=0x84ca368, \n attOff=3, attrNums=0x8508240, fInfo=0x0, attNull=0xbfbfe7fb \"\")\n at indexam.c:445\n#2 0x80903be in FormIndexDatum (numberOfAttributes=4, \n attributeNumber=0x8508240, heapTuple=0xbfbfe974, heapDescriptor=0x84ca368, \n datum=0x8508018, nullv=0x84ba170 \" \", fInfo=0x0) at index.c:1256\n#3 0x80a05e6 in vc_repair_frag (vacrelstats=0x84ba290, onerel=0x84c6788, \n vacuum_pages=0xbfbfea1c, fraged_pages=0xbfbfea0c, nindices=1, \n Irel=0x84ba118) at vacuum.c:1634\n#4 0x809e3b9 in vc_vacone (relid=1315147913, analyze=0, va_cols=0x0)\n at vacuum.c:640\n#5 0x809d9ac in vc_vacuum (VacRelP=0xbfbfeaac, analyze=0 '\\000', va_cols=0x0)\n at vacuum.c:299\n#6 0x809d934 in vacuum (vacrel=0x84ba0e8 \"\\030\", verbose=1, analyze=0 '\\000', \n va_spec=0x0) at vacuum.c:223\n#7 0x810ca8c in ProcessUtility (parsetree=0x84ba110, dest=Remote)\n at utility.c:694\n#8 0x810a44e in pg_exec_query_dest (\n query_string=0x81cd370 \"VACUUM verbose webhit_details_formatted;\", \n dest=Remote, aclOverride=0) at postgres.c:617\n#9 0x810a3a9 in pg_exec_query (\n query_string=0x81cd370 \"VACUUM verbose webhit_details_formatted;\")\n at postgres.c:562\n#10 0x810b336 in PostgresMain (argc=7, argv=0xbfbff12c, real_argc=10, \n real_argv=0xbfbffb8c) at postgres.c:1588\n#11 0x80f0742 in DoBackend (port=0x8464000) at postmaster.c:2009\n#12 0x80f02d5 in BackendStartup (port=0x8464000) at postmaster.c:1776\n#13 0x80ef4f9 in ServerLoop () at postmaster.c:1037\n#14 0x80eeede in PostmasterMain (argc=10, argv=0xbfbffb8c) at postmaster.c:725\n#15 0x80bf3eb in main (argc=10, argv=0xbfbffb8c) at main.c:93\n#16 0x8063495 in _start ()\nst\n532 \n533 if (usecache)\n534 att[i]->attcacheoff = off;\n535 }\n536 \n537 off = att_addlength(off, att[i]->attlen, tp + off);\n538 \n539 if (usecache &&\n540 att[i]->attlen == -1 && !VARLENA_FIXED_SIZE(att[i]))\n541 usecache = false;\n\nit looks like it's dieing in the same place as the previous coredumps\nhowever this looks like it's during a vacuum rather than an update:\n\n(gdb) print off\n$1 = -838833616\n(gdb) print att[i]\n$2 = 0x84ca640\n(gdb) print *(att[i])\n$3 = {attrelid = 1315147913, attname = {\n data = \"attr_name\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 1043, attdisbursion = 0, \n attlen = -1, attnum = 3, attnelems = 0, attcacheoff = -1, atttypmod = 36, \n attbyval = 0 '\\000', attstorage = 112 'p', attisset = 0 '\\000', \n attalign = 105 'i', attnotnull = 0 '\\000', atthasdef = 0 '\\000'}\n(gdb) print i \n$4 = 2\n(gdb) print tp\n$5 = 0x5808eba5 \"Yj\"\n(gdb) print tp+off \n$6 = 0x260955d5 <Address 0x260955d5 out of bounds>\n\nack!\n\n(gdb) print usecache\n$7 = 0 '\\000'\n(gdb) print attnum\n$8 = 3\n(gdb) print slow\n$9 = 139159376\n(gdb) print *slow\n$10 = 139241024\n(gdb) print (char *) tup + tup->t_hoff\n$11 = 0x5808eba5 \"Yj\"\n(gdb) print tup\n$12 = 0x5808eba0\n(gdb) print *tup\n$13 = {t_oid = 0, t_cmin = 6969654, t_cmax = 6958161, t_xmin = 1742, \n t_xmax = 6955895, t_ctid = {ip_blkid = {bi_hi = 0, bi_lo = 639}, \n ip_posid = 84}, t_natts = 737, t_infomask = 32846, t_hoff = 5 '\\005', \n t_bits = \"\\000\\002� \"}\n(gdb) print *tupleDesc \n$14 = {natts = 1358981721, attrs = 0xce006a2c, constr = 0x77000006}\n(gdb) print *(att[0])\n$15 = {attrelid = 1315147913, attname = {\n data = \"counter_id\", '\\000' <repeats 21 times>, \n alignmentDummy = 1853189987}, atttypid = 23, attdisbursion = 0, \n attlen = 4, attnum = 1, attnelems = 0, attcacheoff = 0, atttypmod = -1, \n attbyval = 1 '\\001', attstorage = 112 'p', attisset = 0 '\\000', \n attalign = 105 'i', attnotnull = 0 '\\000', atthasdef = 0 '\\000'}\n(gdb) print *(att[1])\n$16 = {attrelid = 1315147913, attname = {\n data = \"attr_type\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 1043, attdisbursion = 0, \n attlen = -1, attnum = 2, attnelems = 0, attcacheoff = 4, atttypmod = 36, \n attbyval = 0 '\\000', attstorage = 112 'p', attisset = 0 '\\000', \n attalign = 105 'i', attnotnull = 0 '\\000', atthasdef = 0 '\\000'}\n(gdb) print *(att[2])\n$17 = {attrelid = 1315147913, attname = {\n data = \"attr_name\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 1043, attdisbursion = 0, \n attlen = -1, attnum = 3, attnelems = 0, attcacheoff = -1, atttypmod = 36, \n attbyval = 0 '\\000', attstorage = 112 'p', attisset = 0 '\\000', \n attalign = 105 'i', attnotnull = 0 '\\000', atthasdef = 0 '\\000'}\n(gdb) print *(att[3])\n$18 = {attrelid = 1315147913, attname = {\n data = \"attr_vers\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 1043, attdisbursion = 0, \n attlen = -1, attnum = 4, attnelems = 0, attcacheoff = -1, atttypmod = 36, \n attbyval = 0 '\\000', attstorage = 112 'p', attisset = 0 '\\000', \n attalign = 105 'i', attnotnull = 0 '\\000', atthasdef = 0 '\\000'}\n(gdb) print *(att[4])\n$19 = {attrelid = 1315147913, attname = {\n data = \"attr_hits\", '\\000' <repeats 22 times>, \n alignmentDummy = 1920234593}, atttypid = 20, attdisbursion = 0, \n attlen = 8, attnum = 5, attnelems = 0, attcacheoff = -1, atttypmod = -1, \n attbyval = 0 '\\000', attstorage = 112 'p', attisset = 0 '\\000', \n attalign = 100 'd', attnotnull = 0 '\\000', atthasdef = 1 '\\001'}\n(gdb) print *tuple\n$20 = {t_len = 80, t_self = {ip_blkid = {bi_hi = 0, bi_lo = 640}, \n ip_posid = 5}, t_datamcxt = 0x0, t_data = 0x5808eba0}\n\n\n\nthanks,\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Sun, 8 Oct 2000 03:48:34 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": true,
"msg_subject": "Still crashing with latest 7.0.2 (Re: (forw) more crashes)"
},
{
"msg_contents": "* Alfred Perlstein <[email protected]> [001008 03:52] wrote:\n> \n> Unfortunatly I'm still getting crashes, this one looks like it's\n> during a vacuum, previously I got a crash while doing an UPDATE, but\n> in exactly the same spot, it took quite a bit longer to provoke this\n> time:\n\nI should have also said that since this is pretty reproducable I'd\nbe happy to run with patches for stronger assertion checks so\nthat perhaps backends would check to see if they might be writing\nout these incorrect blocks.\n\n(Tom Lane said that this looks like it's caused by previous corruption)\n\n-Alfred\n",
"msg_date": "Sun, 8 Oct 2000 11:02:40 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Still crashing with latest 7.0.2 (Re: (forw) more crashes)"
},
{
"msg_contents": "Alfred Perlstein <[email protected]> writes:\n> I should have also said that since this is pretty reproducable I'd\n> be happy to run with patches for stronger assertion checks \n\nIf you're not running with --enable-cassert configured, I'd certainly\nrecommend trying that.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 08 Oct 2000 21:54:40 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Still crashing with latest 7.0.2 (Re: (forw) more crashes) "
},
{
"msg_contents": "* Tom Lane <[email protected]> [001008 18:57] wrote:\n> Alfred Perlstein <[email protected]> writes:\n> > I should have also said that since this is pretty reproducable I'd\n> > be happy to run with patches for stronger assertion checks \n> \n> If you're not running with --enable-cassert configured, I'd certainly\n> recommend trying that.\n\nI am:\n\n#! /bin/sh\n# Generated automatically by configure.\n# Run this file to recreate the current configuration.\n# This directory was configured as follows,\n# on host abacus.paycounter.com:\n#\n# ./configure --with-maxbackends=128 --enable-debug --enable-cassert --with-perl\n...\n\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Sun, 8 Oct 2000 19:54:59 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Still crashing with latest 7.0.2 (Re: (forw) more crashes)"
}
] |
[
{
"msg_contents": "Hello all,\n\nPlease help me with\ncreate function days_in_month(int4, int4, int4) returns int4\nas '/tmp/days.so' language 'c';\n\nCan I write this function days_in_month in C++ ?\nI've did the following:\n\nextern \"C\" {\nint days_in_month(int year, int mo, int day); \n} \n\nextern \"C\" { \nint days_in_month(int year, int mo, int day) { \nО©╫О©╫О©╫ return 13;\n} \n} \n\nCompile with\n $ g++ -I/usr/local/qt/include -o days.so -shared days.cpp\n\nthen (as user postgres)\n my=> create function days_in_month(int4, int4, int4) returns int4\n my=> as '/tmp/days.so' language 'c';\n CREATE\n \nI've tried to use new function\n my=> select days_in_month(3, 3, 3);\n ERROR:О©╫ Can't find function days_in_month in file /tmp/days.so\n my=>\n \nWhat is wrong ?\nI'm using postgres 6.4 under RedHat.\nAny help will be appreciated.\n\n-- \nBest regards,\n Vladimir mailto:[email protected]\n\n\n",
"msg_date": "Wed, 4 Oct 2000 12:59:06 +0400",
"msg_from": "\"Vladimir V. Zolotych\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "postgres functions and C++"
},
{
"msg_contents": "\"Vladimir V. Zolotych\" <[email protected]> writes:\n> my=> select days_in_month(3, 3, 3);\n> ERROR: Can't find function days_in_month in file /tmp/days.so\n\nTry using 'nm' to see what symbol name is actually being exported\nfrom the .so file. I suspect that despite your use of extern \"C\",\nyour C++ compiler is being uncooperative and is naming the function\nin some strange fashion at the link level.\n\nThere is an option in CREATE FUNCTION to specify the link symbol\nname separately from the SQL name of the function, so if you can't\nget your compiler to play nice you could still get it to work by\nquoting whatever nm tells you...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 04 Oct 2000 11:29:48 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgres functions and C++ "
},
{
"msg_contents": "\"Vladimir V. Zolotych\" wrote:\n\n> Hello all,\n>\n> Compile with\n> $ g++ -I/usr/local/qt/include -o days.so -shared days.cpp\n\nIIRC you have to compile as days.o and then link to .so:\ng++ .... -o days.o\nld -shared -o days.so days.o\n\nI don't know if the compiler gets it right in one line. It might.\n\nChristof\n\n\n",
"msg_date": "Sat, 07 Oct 2000 23:19:23 +0200",
"msg_from": "Christof Petig <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgres functions and C++"
}
] |
[
{
"msg_contents": "Hey,\n\nI got this error when attempting to setup a table with a trigger that ran a\nplpgsql program upon insert:\n\n\"DBD::Pg::st execute failed: ERROR: Load of file\n/usr/local/pgsql/lib/plpgsql.so failed: ld.so.1:\n/usr/local/pgsql/bin/postmaster: fatal: relocation error: file\n/usr/local/pgsql/lib/plpgsql.so: symbol CurrentMemoryContext: referenced\nsymbol not found\n\"\n\nOS is Solaris 7 /w recommended and security patches as of about a month or\ntwo ago. PostgreSQL is 7.0. I can do a 'nm plpgsql.so | grep\nCurrentMemoryContext' in the /usr/local/pgsql/lib directory and I get:\n\n\"\tU CurrentMemoryContext\"\n\nso it looks like the function is in the library. Any clues as to what may\nbe causing this?\n\nPlease CC me at work ([email protected]) as I only get the list at\nhome ([email protected]).\n\nThanks,\n\nFred Reimer\nEclipsys Corporation\n\n",
"msg_date": "Wed, 4 Oct 2000 09:14:29 -0400 ",
"msg_from": "\"Reimer, Fred\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "ld.so error when inserting with a plpgsql trigger function"
},
{
"msg_contents": "The mailing list is a bit sluggish, your message has taken a week to\nget through to me.\n\nIf you have not sorted it out already, look at my message \"Re: [PORTS]\npkg of 7.0.2 for SunOS 5.6?\" of 4 Sep 2000 08:48:23 +0100 (BST).\nBasically, libld.so is not needed AFAIK. It is for accessing COFF\nfiles and all Solaris objects should be ELF. I found that I needed to\nhack the configure script to disable -lld properly.\n-- \nPete Forman -./\\.- Disclaimer: This post is originated\nWestern Geophysical -./\\.- by myself and does not represent\[email protected] -./\\.- the opinion of Baker Hughes or\nhttp://www.crosswinds.net/~petef -./\\.- its divisions.\n",
"msg_date": "Tue, 10 Oct 2000 09:21:37 +0100 (BST)",
"msg_from": "Pete Forman <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: ld.so error when inserting with a plpgsql trigger function"
}
] |
[
{
"msg_contents": "I would like to allow variables in ECPG at all positions the backend allows\nconstants. But I wonder how to get this done in the parser. \n\n1) I need to allow :variable:indicator in AexprConst since this is the only\nposition where NULL is a possible constant.\n2) :variable should be allowed whereever Iconst, FCONST and Sconst are used.\n\nThis would finally allow us to use something like \"exec sql move :number ...\".\n\nAnyway, the problem is that some rules expand to either Iconst, FCONST or\nSconst. So do I have to change all these rules? \n\nJust changing the rule for Iconst and Sconst e.g doesn't work since\nAexprConst expands to the variable in two different ways. And bison\ncertainly does not like that. \n\nBut just adding a combined Const rule won't do either, since there are also\nrules just asking for a special constant like in the aformentioned MOVE\ncommand.\n\nFinally bison cannot distinguish between character and numerical variables.\nI can implement such a check but is it a good idea? How about I want to\nstore my numerical passwords in an int since it's calculated? etc.\n\nAny ideas anyone?\n\nMichael\n\n-- \nMichael Meskes\[email protected]\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Wed, 4 Oct 2000 17:16:08 +0200",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": true,
"msg_subject": "yacc guru needed"
},
{
"msg_contents": "Michael Meskes <[email protected]> writes:\n> Anyway, the problem is that some rules expand to either Iconst, FCONST or\n> Sconst. So do I have to change all these rules? \n\n> Just changing the rule for Iconst and Sconst e.g doesn't work since\n> AexprConst expands to the variable in two different ways. And bison\n> certainly does not like that. \n\nIt looks to me like you ought to try to clean up the not-very-consistent\ntreatment of constants in the grammar. Some rules use raw ICONST, some\nuse Iconst, some use IntegerOnly --- ugh. There's no need for all that\nvariation IMHO.\n\nI'd think about making a small number of productions like\n\nAnyConst: ICONST | FCONST | SCONST\n\nIntegerConst: ICONST | - ICONST\n\nStringConst: SCONST\n\nand trying to make *all* the grammar's uses of constants go through one\nof these productions. For instance AexprConst would produce either\nAnyConst or one of the cast-decorated variants. Then, ecpg's grammar\nwould differ from the backend's grammar by adding \":variable\" as an\nalternative to each of this small group of productions.\n\nThe trick is to choose a good set of gateway productions; the above is\nprobably not quite right...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 04 Oct 2000 11:49:30 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: yacc guru needed "
},
{
"msg_contents": "On Wed, Oct 04, 2000 at 11:49:30AM -0400, Tom Lane wrote:\n> It looks to me like you ought to try to clean up the not-very-consistent\n> treatment of constants in the grammar. Some rules use raw ICONST, some\n> use Iconst, some use IntegerOnly --- ugh. There's no need for all that\n> variation IMHO.\n\nI agree.\n\n> I'd think about making a small number of productions like\n> ...\n\nYes, I thought about this too.\n \nIf noone else is working on it I will as soon as I find time.\n\nMichael\n-- \nMichael Meskes\[email protected]\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Wed, 4 Oct 2000 20:14:15 +0200",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: yacc guru needed"
}
] |
[
{
"msg_contents": "A client just got this error. Anyone know why I suddenly get this?\nThe program has been working flawlessly for months before this.\n\nWaitOnLock: error on wakeup - Aborting this transaction\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 425 1212 (DoD#0082) (eNTP) | what's for dinner.\n",
"msg_date": "Wed, 4 Oct 2000 11:42:04 -0400 (EDT)",
"msg_from": "[email protected] (D'Arcy J.M. Cain)",
"msg_from_op": true,
"msg_subject": "WaitOnLock"
},
{
"msg_contents": "[email protected] (D'Arcy J.M. Cain) writes:\n> WaitOnLock: error on wakeup - Aborting this transaction\n\nWhat happens to NOTICEs in your application?\n\nFor some odd reason, deadlock reports come out as a NOTICE followed by\nthis error message; if you are bit-bucketing NOTICEs then you may well\nbe confused.\n\n\nWindow 1:\n\nregression=# begin;\nBEGIN\nregression=# lock int4_tbl;\nLOCK TABLE\n\nWindow 2:\n\nregression=# begin;\nBEGIN\nregression=# lock int8_tbl;\nLOCK TABLE\nregression=# lock int4_tbl;\n<< waits >>\n\nWindow 1:\n\nregression=# lock int8_tbl;\nNOTICE: Deadlock detected -- See the lock(l) manual page for a possible cause.\nERROR: WaitOnLock: error on wakeup - Aborting this transaction\n\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 04 Oct 2000 11:57:20 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: WaitOnLock "
}
] |
[
{
"msg_contents": "Just a little question :\nWhen I drop a database \"xxx\", and create another with the same name\n\"xxx\", old database sequences are always.\nWhy ?\n\nOlivier\n\n",
"msg_date": "Wed, 04 Oct 2000 18:35:40 +0200",
"msg_from": "Olivier Detour <[email protected]>",
"msg_from_op": true,
"msg_subject": "Sequence"
}
] |
[
{
"msg_contents": "I have a problem with inheritance and references.\n\nTABLE parent_table (\nid_parent SERIAL PRIMARY KEY,\n....\n)\n\nNo primary key in child\nTABLE child_parent (\n.....\n) INHERITS (parent_table);\n\nAnd another table :\nTABLE other_table (\nref_child_table INTEGER REFERENCES parent_table\n);\n\nReference must be take on parent_table, as child_parent inherit\nparent_table's primary key. On child_table, it doesn't work (No primary\nkey in child_table).\n\nSo, it's ok for creating, but I can't do INSERT into other_table because\npostgresql search primary key into parent_table instead into\nchild_table.\n\nHow can I do this ?\nPlease Help...\n",
"msg_date": "Wed, 04 Oct 2000 19:59:31 +0200",
"msg_from": "Olivier Detour <[email protected]>",
"msg_from_op": true,
"msg_subject": "(no subject)"
}
] |
[
{
"msg_contents": "Greetings,\n\nI'm not sure if this is a bug, but I thought someone might like to know \nabout it. If you create a numeric field with the precision and scale set \nthe same, you can't insert anything into the field... pg-7.0.2 on \nFreeBSD-4.0-Rel.\n\nMatthew\n\n\nequinox=# create table test2 ( d numeric(8,2) );\nCREATE\nequinox=# insert into test2 values ('08.00');\nINSERT 167730 1\nequinox=# select * from test2;\n d\n------\n 8.00\n(1 row)\n\n\nequinox=# create table test2 ( d numeric(3,3) );\nCREATE\nequinox=# insert into test2 values ('8');\nERROR: overflow on numeric ABS(value) >= 10^0 for field with precision 3 \nscale 3\nequinox=# drop table test2;\nDROP\nequinox=# create table test2 ( d numeric(4,4) );\nCREATE\nequinox=# insert into test2 values ('8');\nERROR: overflow on numeric ABS(value) >= 10^0 for field with precision 4 \nscale 4\n\nequinox=# create table test2 ( d numeric(3,2) );\nCREATE\nequinox=# insert into test2 values ('8');\nINSERT 167742 1\n\n",
"msg_date": "Wed, 04 Oct 2000 23:41:14 -0400",
"msg_from": "Matthew Hagerty <[email protected]>",
"msg_from_op": true,
"msg_subject": "Numeric field quirk"
}
] |
[
{
"msg_contents": "Announce: First public release of PgSQL Version 1.0\n===================================================\n\nPgSQL v1.0 has been released. This is the first public release of PgSQL.\nIt is available at http://sourceforge.net/projects/pgsql.\n\nPgSQL is a package of two (2) modules that provide a Python DB-API 2.0\ncompliant interface to PostgreSQL databases. The first module, libpq,\nexports the PostgreSQL C API to Python. This module is written in C and\ncan be compiled into Python or can be dynamically loaded on demand. The\nsecond module, PgSQL, provides the DB-API 2.0 compliant interface and\nsupport for various PostgreSQL data types, such as INT8, NUMERIC, MONEY,\nBOOL, ARRAYS, etc. This module is written in Python and works with\nPostgreSQL 6.5.2 or later and Python 1.5.2 or later.\n\nPostgreSQL is a sophisticated Object-Relational DBMS, supporting almost all\nSQL constructs, including sub-selects, transactions, and user-defined types\nand functions. It is the most advanced open-source database available\nanywhere More information about PostgreSQL can be found at the PostgreSQL\nhome page at http://www.postgresql.org.\n\nPython is an interpreted, interactive, object-oriented programming lang-\nuage. It combines remarkable power with very clear syntax. It has\nmodules, classes, exceptions, very high level dynamic data types, and\ndynamic typing. There are interfaces to many system calls and libraries,\nas well as to various windowing systems (X11, Motif, Tk, Mac, MFC). New\nbuiltin modules are easily written in C or C++. Python is also usable as\nan extension language for applications that need a programmable interface. \nPython is copyrighted but freely usable and distributable, even for\ncommercial use. More information about Python can be found on the Python\nhome page at http://www.python.org.\n\n-- \n____ | Billy G. Allie | Domain....: [email protected]\n| /| | 7436 Hartwell | Compuserve: 76337,2061\n|-/-|----- | Dearborn, MI 48126| MSN.......: [email protected]\n|/ |LLIE | (313) 582-1540 |",
"msg_date": "Wed, 04 Oct 2000 23:47:42 -0400",
"msg_from": "\"Billy G. Allie\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Announcing PgSQL - a Python DB-API 2.0 compliant interface to\n\tPostgreSQL"
},
{
"msg_contents": "Billy G. Allie writes:\n\n> PgSQL v1.0 has been released. This is the first public release of PgSQL.\n> It is available at http://sourceforge.net/projects/pgsql.\n\nSounds interesting, but isn't \"pgsql\" an extremely unfortunate choice of\nname, given that it's already used as an abbreviation for \"PostgreSQL\"?\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Mon, 9 Oct 2000 18:27:21 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [INTERFACES] Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "I need to know how this is different than our current python interface,\nPyGreSQL.\n\n-- Start of PGP signed section.\n> Announce: First public release of PgSQL Version 1.0\n> ===================================================\n> \n> PgSQL v1.0 has been released. This is the first public release of PgSQL.\n> It is available at http://sourceforge.net/projects/pgsql.\n> \n> PgSQL is a package of two (2) modules that provide a Python DB-API 2.0\n> compliant interface to PostgreSQL databases. The first module, libpq,\n> exports the PostgreSQL C API to Python. This module is written in C and\n> can be compiled into Python or can be dynamically loaded on demand. The\n> second module, PgSQL, provides the DB-API 2.0 compliant interface and\n> support for various PostgreSQL data types, such as INT8, NUMERIC, MONEY,\n> BOOL, ARRAYS, etc. This module is written in Python and works with\n> PostgreSQL 6.5.2 or later and Python 1.5.2 or later.\n> \n> PostgreSQL is a sophisticated Object-Relational DBMS, supporting almost all\n> SQL constructs, including sub-selects, transactions, and user-defined types\n> and functions. It is the most advanced open-source database available\n> anywhere More information about PostgreSQL can be found at the PostgreSQL\n> home page at http://www.postgresql.org.\n> \n> Python is an interpreted, interactive, object-oriented programming lang-\n> uage. It combines remarkable power with very clear syntax. It has\n> modules, classes, exceptions, very high level dynamic data types, and\n> dynamic typing. There are interfaces to many system calls and libraries,\n> as well as to various windowing systems (X11, Motif, Tk, Mac, MFC). New\n> builtin modules are easily written in C or C++. Python is also usable as\n> an extension language for applications that need a programmable interface. \n> Python is copyrighted but freely usable and distributable, even for\n> commercial use. More information about Python can be found on the Python\n> home page at http://www.python.org.\n> \n> -- \n> ____ | Billy G. Allie | Domain....: [email protected]\n> | /| | 7436 Hartwell | Compuserve: 76337,2061\n> |-/-|----- | Dearborn, MI 48126| MSN.......: [email protected]\n> |/ |LLIE | (313) 582-1540 | \n> \n> \n-- End of PGP section.\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 9 Oct 2000 19:17:35 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Announcing PgSQL - a Python DB-API 2.0 compliant interface\n\tto PostgreSQL"
},
{
"msg_contents": "On Mon, 9 Oct 2000, Bruce Momjian wrote:\n\n> I need to know how this is different than our current python interface,\n> PyGreSQL.\n\nIs this a product of pgsql.com?\n\nVince.\n\n> \n> -- Start of PGP signed section.\n> > Announce: First public release of PgSQL Version 1.0\n> > ===================================================\n> > \n> > PgSQL v1.0 has been released. This is the first public release of PgSQL.\n> > It is available at http://sourceforge.net/projects/pgsql.\n> > \n> > PgSQL is a package of two (2) modules that provide a Python DB-API 2.0\n> > compliant interface to PostgreSQL databases. The first module, libpq,\n> > exports the PostgreSQL C API to Python. This module is written in C and\n> > can be compiled into Python or can be dynamically loaded on demand. The\n> > second module, PgSQL, provides the DB-API 2.0 compliant interface and\n> > support for various PostgreSQL data types, such as INT8, NUMERIC, MONEY,\n> > BOOL, ARRAYS, etc. This module is written in Python and works with\n> > PostgreSQL 6.5.2 or later and Python 1.5.2 or later.\n> > \n> > PostgreSQL is a sophisticated Object-Relational DBMS, supporting almost all\n> > SQL constructs, including sub-selects, transactions, and user-defined types\n> > and functions. It is the most advanced open-source database available\n> > anywhere More information about PostgreSQL can be found at the PostgreSQL\n> > home page at http://www.postgresql.org.\n> > \n> > Python is an interpreted, interactive, object-oriented programming lang-\n> > uage. It combines remarkable power with very clear syntax. It has\n> > modules, classes, exceptions, very high level dynamic data types, and\n> > dynamic typing. There are interfaces to many system calls and libraries,\n> > as well as to various windowing systems (X11, Motif, Tk, Mac, MFC). New\n> > builtin modules are easily written in C or C++. Python is also usable as\n> > an extension language for applications that need a programmable interface. \n> > Python is copyrighted but freely usable and distributable, even for\n> > commercial use. More information about Python can be found on the Python\n> > home page at http://www.python.org.\n> > \n> > -- \n> > ____ | Billy G. Allie | Domain....: [email protected]\n> > | /| | 7436 Hartwell | Compuserve: 76337,2061\n> > |-/-|----- | Dearborn, MI 48126| MSN.......: [email protected]\n> > |/ |LLIE | (313) 582-1540 | \n> > \n> > \n> -- End of PGP section.\n> \n> \n> \n\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Mon, 9 Oct 2000 19:29:33 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "Nope, neigher PyGreSQL nor \"\"PGSQL\"\" are products of\nPgsql.com (aka PostgreSQL Inc)On Mon, 9 Oct 2000, Vince Vielhaber wrote:\n\n> On Mon, 9 Oct 2000, Bruce Momjian wrote:\n> \n> > I need to know how this is different than our current python interface,\n> > PyGreSQL.\n> \n> Is this a product of pgsql.com?\n> \n> Vince.\n> \n\nJeff MacDonald,\n\n-----------------------------------------------------\nPostgreSQL Inc\t\t| Hub.Org Networking Services\[email protected]\t\t| [email protected]\nwww.pgsql.com\t\t| www.hub.org\n1-902-542-0713\t\t| 1-902-542-3657\n-----------------------------------------------------\nFacsimile : 1 902 542 5386\nIRC Nick : bignose\n\n",
"msg_date": "Tue, 10 Oct 2000 08:58:05 -0300 (ADT)",
"msg_from": "Jeff MacDonald <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "On Tue, 10 Oct 2000, Jeff MacDonald wrote:\n\n> Nope, neigher PyGreSQL nor \"\"PGSQL\"\" are products of\n> Pgsql.com (aka PostgreSQL Inc)On Mon, 9 Oct 2000, Vince Vielhaber wrote:\n\nI knew PyGreSQL wasn't but PgSQL appeared a bit misleading. Thanks\nfor clearing that up. I guess on the website I should call it\nPgSQL - NA \n\nVince.\n\n\n> \n> > On Mon, 9 Oct 2000, Bruce Momjian wrote:\n> > \n> > > I need to know how this is different than our current python interface,\n> > > PyGreSQL.\n> > \n> > Is this a product of pgsql.com?\n> > \n> > Vince.\n> > \n> \n> Jeff MacDonald,\n> \n> -----------------------------------------------------\n> PostgreSQL Inc\t\t| Hub.Org Networking Services\n> [email protected]\t\t| [email protected]\n> www.pgsql.com\t\t| www.hub.org\n> 1-902-542-0713\t\t| 1-902-542-3657\n> -----------------------------------------------------\n> Facsimile : 1 902 542 5386\n> IRC Nick : bignose\n> \n> \n\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Tue, 10 Oct 2000 09:47:33 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "Bruce Momjian writes:\n\n> I need to know how this is different than our current python interface,\n> PyGreSQL.\n\n> > PgSQL is a package of two (2) modules that provide a Python DB-API 2.0\n> > compliant interface to PostgreSQL databases.\n\nDB-API is the standard database interface for Python. Kind of like\nDBD/DBI for Perl.\n\nThe existing one is hand-crafted, kind of like the Pg Perl module.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Tue, 10 Oct 2000 17:54:07 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [INTERFACES] Re: Announcing PgSQL - a Python DB-API\n\t2.0 compliant interface to PostgreSQL"
},
{
"msg_contents": "Peter Eisentraut wrote:\n> Billy G. Allie writes:\n> \n> > PgSQL v1.0 has been released. This is the first public release of PgSQL.\n> > It is available at http://sourceforge.net/projects/pgsql.\n> \n> Sounds interesting, but isn't \"pgsql\" an extremely unfortunate choice of\n> name, given that it's already used as an abbreviation for \"PostgreSQL\"?\n> \n> -- \n> Peter Eisentraut [email protected] http://yi.org/peter-e/\n> \n\nPgSQL is the name of the module you import in Python to access a PostgreSQL\ndatabase from Python using the DB-API 2.0. The project name on SourceForge\nis \"Python Interface to PostgreSQL\". Of course, it there is too much heart\nburn with the modules name, I can change it.\n___________________________________________________________________________\n____ | Billy G. Allie | Domain....: [email protected]\n| /| | 7436 Hartwell | Compuserve: 76337,2061\n|-/-|----- | Dearborn, MI 48126| MSN.......: [email protected]\n|/ |LLIE | (313) 582-1540 | \n",
"msg_date": "Tue, 10 Oct 2000 16:45:43 -0400",
"msg_from": "\"Billy G. Allie\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [INTERFACES] Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "\"Billy G. Allie\" <[email protected]> writes:\n> Peter Eisentraut wrote:\n>> Sounds interesting, but isn't \"pgsql\" an extremely unfortunate choice of\n>> name, given that it's already used as an abbreviation for \"PostgreSQL\"?\n\n> PgSQL is the name of the module you import in Python to access a PostgreSQL\n> database from Python using the DB-API 2.0. The project name on SourceForge\n> is \"Python Interface to PostgreSQL\". Of course, it there is too much heart\n> burn with the modules name, I can change it.\n\nFWIW, my initial reaction was the same as Peter's: that name is certain\nto cause confusion. I don't want to try to force you to change it, but\nI think it's not a good choice.\n\nDon't have a better alternative to offer offhand, however.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 10 Oct 2000 17:06:40 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [INTERFACES] Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "Tom Lane wrote:\n> \"Billy G. Allie\" <[email protected]> writes:\n> > Peter Eisentraut wrote:\n> >> Sounds interesting, but isn't \"pgsql\" an extremely unfortunate choice of\n> >> name, given that it's already used as an abbreviation for \"PostgreSQL\"?\n> \n> > PgSQL is the name of the module you import in Python to access a PostgreSQL\n> > database from Python using the DB-API 2.0. The project name on SourceForge\n> > is \"Python Interface to PostgreSQL\". Of course, it there is too much heart\n> > burn with the modules name, I can change it.\n> \n> FWIW, my initial reaction was the same as Peter's: that name is certain\n> to cause confusion. I don't want to try to force you to change it, but\n> I think it's not a good choice.\n> \n> Don't have a better alternative to offer offhand, however.\n> \n> \t\t\tregards, tom lane\n\nI couldn't think of a better alternative, either. That's why the module\nname is PgSQL. Of course, seeing 'import PgSQL' in the python code seems\nto imply loading code to access PostgreSQL. It's kind of self-documenting,\ndon'y you think? :-)\n\n\nI am, however, open to suggestions to a better alternative if one can be\noffered.\n\n\nI am, however, open to suggestions to a better alternative if one can be\noffered.\n___________________________________________________________________________\n____ | Billy G. Allie | Domain....: [email protected]\n| /| | 7436 Hartwell | Compuserve: 76337,2061\n|-/-|----- | Dearborn, MI 48126| MSN.......: [email protected]\n|/ |LLIE | (313) 582-1540 | \n",
"msg_date": "Tue, 10 Oct 2000 17:42:42 -0400",
"msg_from": "\"Billy G. Allie\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [INTERFACES] Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "On Tue, 10 Oct 2000, Billy G. Allie wrote:\n\n> Tom Lane wrote:\n> > \"Billy G. Allie\" <[email protected]> writes:\n> > > Peter Eisentraut wrote:\n> > >> Sounds interesting, but isn't \"pgsql\" an extremely unfortunate choice of\n> > >> name, given that it's already used as an abbreviation for \"PostgreSQL\"?\n> > \n> > > PgSQL is the name of the module you import in Python to access a PostgreSQL\n> > > database from Python using the DB-API 2.0. The project name on SourceForge\n> > > is \"Python Interface to PostgreSQL\". Of course, it there is too much heart\n> > > burn with the modules name, I can change it.\n> > \n> > FWIW, my initial reaction was the same as Peter's: that name is certain\n> > to cause confusion. I don't want to try to force you to change it, but\n> > I think it's not a good choice.\n> > \n> > Don't have a better alternative to offer offhand, however.\n> > \n> > \t\t\tregards, tom lane\n> \n> I couldn't think of a better alternative, either. That's why the module\n> name is PgSQL. Of course, seeing 'import PgSQL' in the python code seems\n> to imply loading code to access PostgreSQL. It's kind of self-documenting,\n> don'y you think? :-)\n\nD'Arcy's is PyGreSQL ... \n> \n> \n> I am, however, open to suggestions to a better alternative if one can be\n> offered.\n> \n> \n> I am, however, open to suggestions to a better alternative if one can be\n> offered.\n> ___________________________________________________________________________\n> ____ | Billy G. Allie | Domain....: [email protected]\n> | /| | 7436 Hartwell | Compuserve: 76337,2061\n> |-/-|----- | Dearborn, MI 48126| MSN.......: [email protected]\n> |/ |LLIE | (313) 582-1540 | \n> \n\nMarc G. Fournier ICQ#7615664 IRC Nick: Scrappy\nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Tue, 10 Oct 2000 19:58:08 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [INTERFACES] Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
}
] |
[
{
"msg_contents": "Greetings,\n\nWell, it seems that the numeric issue I was having has nothing to do with \nthe precision and scale being set the same, it has to do with the input \ndata. The precision has to be at least 2 greater than the biggest number \nyou need to enter, i.e.\n\nequinox=# create table test ( d numeric(4,2) );\nCREATE\nequinox=# insert into test values(10);\nINSERT 167844 1\nequinox=# insert into test values(101);\nERROR: overflow on numeric ABS(value) >= 10^2 for field with precision 4 \nscale 2\n\n\nMatthew\n\n",
"msg_date": "Wed, 04 Oct 2000 23:57:45 -0400",
"msg_from": "Matthew Hagerty <[email protected]>",
"msg_from_op": true,
"msg_subject": "Numeric field quirk [Again]"
},
{
"msg_contents": "Matthew Hagerty <[email protected]> writes:\n> Well, it seems that the numeric issue I was having has nothing to do with \n> the precision and scale being set the same, it has to do with the input \n> data. The precision has to be at least 2 greater than the biggest number \n> you need to enter, i.e.\n\n> equinox=# create table test ( d numeric(4,2) );\n> CREATE\n> equinox=# insert into test values(10);\n> INSERT 167844 1\n> equinox=# insert into test values(101);\n> ERROR: overflow on numeric ABS(value) >= 10^2 for field with precision 4 \n> scale 2\n\nPrecision is the total number of digits, scale is the number of digits\nto the right of the decimal point. So the above (4,2) means numbers\nlike XX.XX --- and obviously 10 fits, 101 doesn't. Your previous\nexample of (3,3) means numbers like .XXX --- no integer will fit,\nbut fractions will. I see no bug here...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 05 Oct 2000 02:02:42 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Numeric field quirk [Again] "
},
{
"msg_contents": "Matthew Hagerty wrote:\n> Greetings,\n>\n> Well, it seems that the numeric issue I was having has nothing to do with\n> the precision and scale being set the same, it has to do with the input\n> data. The precision has to be at least 2 greater than the biggest number\n> you need to enter, i.e.\n>\n> equinox=# create table test ( d numeric(4,2) );\n> CREATE\n> equinox=# insert into test values(10);\n> INSERT 167844 1\n> equinox=# insert into test values(101);\n> ERROR: overflow on numeric ABS(value) >= 10^2 for field with precision 4\n> scale 2\n\n The scale digits aren't added, they are taken from the\n precision specified digits. Numeric(4.2) is 99.99,\n numeric(8.2) is 999999.99 and numeric(4.4) is .9999 - so the\n above looks correct to me.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#================================================== [email protected] #\n\n\n",
"msg_date": "Thu, 5 Oct 2000 03:44:08 -0500 (EST)",
"msg_from": "Jan Wieck <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Numeric field quirk [Again]"
},
{
"msg_contents": "> Greetings,\n> \n> Well, it seems that the numeric issue I was having has nothing to do with \n> the precision and scale being set the same, it has to do with the input \n> data. The precision has to be at least 2 greater than the biggest number \n> you need to enter, i.e.\n> \n> equinox=# create table test ( d numeric(4,2) );\n> CREATE\n> equinox=# insert into test values(10);\n> INSERT 167844 1\n> equinox=# insert into test values(101);\n> ERROR: overflow on numeric ABS(value) >= 10^2 for field with precision 4 \n> scale 2\n\nThe variable above is 4 digits, with 2 digits to the right of the\ndecimal place. Doing numberic(3,3) allows insertion of 0.4, but not\n4.0. This is expected.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 9 Oct 2000 13:02:36 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Numeric field quirk [Again]"
}
] |
[
{
"msg_contents": "\nProbably [email protected] isnt work from my network :(\n\n--------------------------------------------------------------\nThe reboots are for hardware upgrades!\n\"http://www.nmmm.nu\"; < [email protected] >\n\n\n\n-----Original Message-----\nFrom: Nikolay Mijaylov [mailto:[email protected]]\nSent: Thursday, October 05, 2000 3:21 PM\nTo: [email protected]\nSubject: FW: URGENT: pgsql on the web server - memory problems....\n\n\n\n\n\n--------------------------------------------------------------\nThe reboots are for hardware upgrades!\n\"http://www.nmmm.nu\"; < [email protected] >\n\n\n\n-----Original Message-----\nFrom: Nikolay Mijaylov [mailto:[email protected]]\nSent: Thursday, October 05, 2000 1:28 PM\nTo: [email protected]\nSubject: URGENT: pgsql on the web server - memory problems....\n\n\nHi all\nI;m not sure how to explain this\n\nwe have a web server with several clients which uses PostgreSQL. Currently\nversion 7.0.2\n\nthe server is linux machine with 64MB RAM and 200 MB swap on the disk. One\nHDD partition with free disk space about 1.5 GB.\n\ni started the server with:\npostmaster -N 64 -B 256\n\nUnfortunately one of our clients runs wery big database and very big\n(fuckin' strange complex) queries using PHP.\n\nThe problem is that PHP is auto killing after a while (1-2 min), but the\npostgres backend process stay (and works about 2 or many more hours). Soon\nall Linux is filled with 64 postgres processes, free memory is 0 and Linux,\nfreeze :)\n\nThese queries freeze our computer and it stops responding. The only way we\ncan do is to press \"reset\" button.\n\nMy question is : is there a way to limit the memory (and CPU time) used by\npostgres processes by our clients. Can we make auto kill like in PHP?\n\nI play with ulimit and think it is not a such a good idea....\nFor a moment I stop this client and do\n\npostmaster -N 24 -B 64\n\n...\n\nAny Suggestions?\n\nThank you\n\n\n--------------------------------------------------------------\nThe reboots are for hardware upgrades!\n\"http://www.nmmm.nu\"; < [email protected] >\n\n\n\n",
"msg_date": "Thu, 5 Oct 2000 15:36:47 +0300",
"msg_from": "\"Nikolay Mijaylov\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "FW: URGENT: pgsql on the web server - memory problems...."
}
] |
[
{
"msg_contents": "Hi there\n\nI have been developing a web application using an Oracle database on the backend, and Java beans and JSP for the frontend.\n \nAt a late stage in the project I have been asked to look into the viability of\nhosting the application with postgreSQL as the backend database.\n\nNow my major concern is that we developed a whole bunch of scripts for table definitions, \ntriggers, sequences and Stored Procedures using Oracle standard PLSQL and datatypes.\n\nShould I decide to change to PostGRESQL, what course of action would you recommend I take?\n\nRegards,\n\nJono\n\n\n\n\n\n\n\n \nHi there\n \nI have been developing a web application using an \nOracle database on the backend, and Java beans and JSP for the \nfrontend.\n \nAt a late stage in the project I have been asked to \nlook into the viability of\nhosting the application with postgreSQL as the \nbackend database.\n \nNow my major concern is that we \ndeveloped a whole bunch of scripts \nfor table definitions, \ntriggers, sequences and Stored \nProcedures using Oracle standard PLSQL and \ndatatypes.\n \nShould I decide to change to PostGRESQL, \nwhat course of \naction would you recommend I take?\n \nRegards,\n \nJono",
"msg_date": "Thu, 5 Oct 2000 15:03:04 +0200",
"msg_from": "\"Jono\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Oracle to POSTGRESQL"
}
] |
[
{
"msg_contents": "Hi,\n\nIm still trying to figure out how postgre does things. And Im clueless.\n\nSomewhere inside the backend there must occur a compare between \"column\" and\n'string'. But where (and how)?\n\nIve been messing around with the source code in Linux, but with no success.\n\n// Jarmo\n\nPS. dbinit failed when I was trying out my own build. But I told it to keep\nthe files it created, and the postmaster worked as it should (could create\ntable, select and insert without problem. Is this a known problem with\ndbinit? or should I try to figure out what went wrong. DS.\n\n",
"msg_date": "Thu, 5 Oct 2000 19:09:03 +0200",
"msg_from": "\"Jarmo Paavilainen\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "... WHERE \"column\" = 'string'"
}
] |
[
{
"msg_contents": "\nI sent an email on this the other day, but didn't see it pop up in the list\nand there have been no responses, so I don't know if it got through. Sorry\nif this is a dup.\n\nI'm getting the following error when I add a row to a table that has a\ntrigger set for a plpgsql function. I did a nm on the plpgsql.so file and\nit lists CurrentMemoryContext as being in there, so I don't know why it\ncan't find it. Anyone see anything similar? What do you suggest, a\nrecompile and install?\n\n\"DBD::Pg::st execute failed: ERROR: Load of file\n/usr/local/pgsql/lib/plpgsql.so failed: ld.so.1:\n/usr/local/pgsql/bin/postmaster: fatal: relocation error: file\n/usr/local/pgsql/lib/plpgsql.so: symbol CurrentMemoryContext: referenced\nsymbol not found\"\n\nThanks,\n\nFred Reimer\nEclipsys Corporation\n\n\n",
"msg_date": "Thu, 5 Oct 2000 15:21:20 -0400",
"msg_from": "\"Frederick W. Reimer\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "symbol not found in plpgsql.so"
},
{
"msg_contents": "Oh, forgot to say that this is v7.0 on Solaris 7...\n\nFred\n\n> -----Original Message-----\n> From: [email protected] [mailto:[email protected]]On\n> Behalf Of Frederick W. Reimer\n> Sent: Thursday, October 05, 2000 3:21 PM\n> To: [email protected]; [email protected]\n> Subject: [PORTS] symbol not found in plpgsql.so\n> \n> \n> \n> I sent an email on this the other day, but didn't see it pop \n> up in the list\n> and there have been no responses, so I don't know if it got \n> through. Sorry\n> if this is a dup.\n> \n> I'm getting the following error when I add a row to a table that has a\n> trigger set for a plpgsql function. I did a nm on the \n> plpgsql.so file and\n> it lists CurrentMemoryContext as being in there, so I don't \n> know why it\n> can't find it. Anyone see anything similar? What do you suggest, a\n> recompile and install?\n> \n> \"DBD::Pg::st execute failed: ERROR: Load of file\n> /usr/local/pgsql/lib/plpgsql.so failed: ld.so.1:\n> /usr/local/pgsql/bin/postmaster: fatal: relocation error: file\n> /usr/local/pgsql/lib/plpgsql.so: symbol CurrentMemoryContext: \n> referenced\n> symbol not found\"\n> \n> Thanks,\n> \n> Fred Reimer\n> Eclipsys Corporation\n> \n> \n> \n",
"msg_date": "Thu, 5 Oct 2000 15:32:35 -0400",
"msg_from": "\"Frederick W. Reimer\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [PORTS] symbol not found in plpgsql.so"
},
{
"msg_contents": "\"Frederick W. Reimer\" <[email protected]> writes:\n> I'm getting the following error when I add a row to a table that has a\n> trigger set for a plpgsql function. I did a nm on the plpgsql.so file and\n> it lists CurrentMemoryContext as being in there, so I don't know why it\n> can't find it.\n> \"DBD::Pg::st execute failed: ERROR: Load of file\n> /usr/local/pgsql/lib/plpgsql.so failed: ld.so.1:\n> /usr/local/pgsql/bin/postmaster: fatal: relocation error: file\n> /usr/local/pgsql/lib/plpgsql.so: symbol CurrentMemoryContext: referenced\n> symbol not found\"\n\nNo, you're interpreting this backward: the linker is complaining because\nit can't find a CurrentMemoryContext symbol in the main program to bind\nthe .so's reference to.\n\nYou didn't say what platform you're on, but on lots of platforms it's\nnecessary to do something special to ensure that all symbols in a main\nprogram are exported so that dynamically loaded shared libs can see\nthem. (That's not normally the default behavior because the symbol\ntable space would be wasted in most programs, which don't do dynamic\nloading of random .so files.) On HPUX, for example, the loader has to\nbe given a -E switch when linking the postgres executable. Dunno what\nthe incantation is where you live...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 05 Oct 2000 16:26:31 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: symbol not found in plpgsql.so "
},
{
"msg_contents": "Tom Lane writes:\n\n> On HPUX, for example, the loader has to be given a -E switch when\n> linking the postgres executable.\n\nBtw., something that has struck me as odd is that the hpux template with\nthe native compiler sets CFLAGS='-Wl,-E -Ae'. I know what the -Ae is for,\nbut the -Wl,-E should probably not be in every compilation command,\nno? (in particular since LDFLAGS already contains -Wl,-E)\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Thu, 5 Oct 2000 23:25:44 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: symbol not found in plpgsql.so "
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> Btw., something that has struck me as odd is that the hpux template with\n> the native compiler sets CFLAGS='-Wl,-E -Ae'. I know what the -Ae is for,\n> but the -Wl,-E should probably not be in every compilation command,\n> no? (in particular since LDFLAGS already contains -Wl,-E)\n\nOh, I hadn't noticed that. Yes, that's a waste of time if LDFLAGS has\nit.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 05 Oct 2000 17:26:12 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: symbol not found in plpgsql.so "
}
] |
[
{
"msg_contents": "Hello,\n\nI've run into a really strange problem that's taken a while for \nme to track down, but I think I have enough information here\nfor people to verify what I'm saying and hopefully enough information\nto point someone in the right direction for fixing.\n\nMy basic problem is that several queries are returning less results\nthan expected, more results than expected, or no results at all. I've\nbeen able to create a test case that causes the failure which \nproduces no results. All of the failures involve this basic type of\nquery:\n\n\tselect * from foo where \n\t\texists (select * from bar1..barN where join1..joinN)\n\nI have attached a test database with enough for me to reproduce a \nfailure all of the time. I'm running RedHat-7.0/i686 and PG-7.0.2\nI've also tested this on RedHat-6.2\n\nTo reproduce the failure:\n\n\t1) create a test db\n\t2) load database from attachment\n\t3) ---> VACUUM ANALYZE <----\n\t4) check plan; run query --> you get 0 rows\n\t5) drop index c_key \n\t6) check plan; run query --> you get 4 rows\n\nYou MUST VACUUM before running the query otherwise a different plan\nwill be choosen and the problem won't be observed.\n\nQUERY that fails:\n\tselect g.id from g where exists ( select * from a, b, c, c_c, j\n\twhere a.v = 2 and a.id = g.id and substr(b.v,1,4) = '1990' and\n \tb.id = g.id and c.v = c_c.id and c_c.v = 1 and c.v = 2 and\n \tc.id = g.id and j.v = 2 and j.id = g.id);\n\n\nThis is the plan that is known to fail on my machine:\n\nSeq Scan on g (cost=0.00..22619.63 rows=2402 width=4)\n SubPlan\n -> Nested Loop (cost=0.00..9.40 rows=1 width=52)\n -> Nested Loop (cost=0.00..7.37 rows=1 width=44)\n -> Index Scan using c_key on c (cost=0.00..2.02 rows=1 width=8)\n -> Materialize (cost=5.35..5.35 rows=1 width=36)\n -> Nested Loop (cost=0.00..5.35 rows=1 width=36)\n -> Nested Loop (cost=0.00..3.78 rows=1 width=28)\n -> Seq Scan on j (cost=0.00..2.12 rows=1 width=12)\n -> Seq Scan on b (cost=0.00..1.65 rows=1 width=16)\n -> Seq Scan on a (cost=0.00..1.55 rows=1 width=8)\n -> Index Scan using c_c_key on c_c (cost=0.00..2.02 rows=1 width=8)\n\nOther plans not using Materialize seem to work okay.\n\nPlease contact me if I can help someone solve this problem or supply more \ninformation. I want to help out since I rely heavily on postgres!\n\n\n--Brian Hirt\n\n-- \nThe world's most ambitious and comprehensive PC game database project.\n\n http://www.mobygames.com",
"msg_date": "Thu, 5 Oct 2000 16:58:56 -0500",
"msg_from": "Brian Hirt <[email protected]>",
"msg_from_op": true,
"msg_subject": "query returns incorrect results."
},
{
"msg_contents": "Brian Hirt <[email protected]> writes:\n> I've run into a really strange problem that's taken a while for \n> me to track down, but I think I have enough information here\n> for people to verify what I'm saying and hopefully enough information\n> to point someone in the right direction for fixing.\n\nNasty. Looks like the problem is that Materialize isn't paying\nattention to chgParam (boo hiss), so when the subplan is re-executed\nwith successive g.id values, it keeps handing back the same rows\nit computed on the first time through. You only see a failure if\none of the subplans under the Materialize pays attention to g.id\n(else the result *should* be the same each time), which is why the\nfailure comes and goes with different plans. \n\nThis actually works correctly in current sources, because I had\nrewritten nodeMaterial awhile back and put in the same ReScan\nlogic that I saw in other plan node types --- I just did that on\ngeneral principles, not thinking that it was a live bug fix.\nSo it hadn't occurred to me to try to back-patch that change,\nbut clearly we need to do something about it for 7.0.3.\n\nI'll see what I can do about it...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 05 Oct 2000 20:46:19 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: query returns incorrect results. "
},
{
"msg_contents": "Try the attached patch against 7.0.2 --- not well tested but it\nseems to work...\n\n\t\t\tregards, tom lane\n\n\n*** src/backend/executor/nodeMaterial.c.orig\tWed Mar 1 23:06:39 2000\n--- src/backend/executor/nodeMaterial.c\tThu Oct 5 21:28:47 2000\n***************\n*** 8,14 ****\n *\n *\n * IDENTIFICATION\n! *\t $Header: /home/projects/pgsql/cvsroot/pgsql/src/backend/executor/nodeMaterial.c,v 1.30 2000/03/02 04:06:39 tgl Exp $\n *\n *-------------------------------------------------------------------------\n */\n--- 8,14 ----\n *\n *\n * IDENTIFICATION\n! *\t $Header: /home/projects/pgsql/cvsroot/pgsql/src/backend/executor/nodeMaterial.c,v 1.30.2.2 2000/10/06 01:28:47 tgl Exp $\n *\n *-------------------------------------------------------------------------\n */\n***************\n*** 39,50 ****\n *\t\tcalls to ExecMaterial return successive tuples from the temp\n *\t\trelation.\n *\n- *\t\tInitial State:\n- *\n- *\t\tExecMaterial assumes the temporary relation has been\n- *\t\tcreated and opened by ExecInitMaterial during the prior\n- *\t\tInitPlan() phase.\n- *\n * ----------------------------------------------------------------\n */\n TupleTableSlot *\t\t\t\t/* result tuple from subplan */\n--- 39,44 ----\n***************\n*** 78,96 ****\n \n \tif (matstate->mat_Flag == false)\n \t{\n \t\t/* ----------------\n! \t\t *\tset all relations to be scanned in the forward direction\n! \t\t *\twhile creating the temporary relation.\n \t\t * ----------------\n \t\t */\n! \t\testate->es_direction = ForwardScanDirection;\n \n \t\t/* ----------------\n \t\t *\t if we couldn't create the temp relation then\n \t\t *\t we print a warning and return NULL.\n \t\t * ----------------\n \t\t */\n- \t\ttempRelation = matstate->mat_TempRelation;\n \t\tif (tempRelation == NULL)\n \t\t{\n \t\t\telog(DEBUG, \"ExecMaterial: temp relation is NULL! aborting...\");\n--- 72,105 ----\n \n \tif (matstate->mat_Flag == false)\n \t{\n+ \t\tTupleDesc\ttupType;\n+ \n \t\t/* ----------------\n! \t\t *\tget type information needed for ExecCreatR\n \t\t * ----------------\n \t\t */\n! \t\ttupType = ExecGetScanType(&matstate->csstate);\n! \n! \t\t/* ----------------\n! \t\t *\tExecCreatR wants its second argument to be an object id of\n! \t\t *\ta relation in the range table or a _NONAME_RELATION_ID\n! \t\t *\tindicating that the relation is not in the range table.\n! \t\t *\n! \t\t *\tIn the second case ExecCreatR creates a temp relation.\n! \t\t *\t(currently this is the only case we support -cim 10/16/89)\n! \t\t * ----------------\n! \t\t */\n! \t\t/* ----------------\n! \t\t *\tcreate the temporary relation\n! \t\t * ----------------\n! \t\t */\n! \t\ttempRelation = ExecCreatR(tupType, _NONAME_RELATION_ID_);\n \n \t\t/* ----------------\n \t\t *\t if we couldn't create the temp relation then\n \t\t *\t we print a warning and return NULL.\n \t\t * ----------------\n \t\t */\n \t\tif (tempRelation == NULL)\n \t\t{\n \t\t\telog(DEBUG, \"ExecMaterial: temp relation is NULL! aborting...\");\n***************\n*** 98,103 ****\n--- 107,126 ----\n \t\t}\n \n \t\t/* ----------------\n+ \t\t *\tsave the relation descriptor in the sortstate\n+ \t\t * ----------------\n+ \t\t */\n+ \t\tmatstate->mat_TempRelation = tempRelation;\n+ \t\tmatstate->csstate.css_currentRelation = NULL;\n+ \n+ \t\t/* ----------------\n+ \t\t *\tset all relations to be scanned in the forward direction\n+ \t\t *\twhile creating the temporary relation.\n+ \t\t * ----------------\n+ \t\t */\n+ \t\testate->es_direction = ForwardScanDirection;\n+ \n+ \t\t/* ----------------\n \t\t *\t retrieve tuples from the subplan and\n \t\t *\t insert them in the temporary relation\n \t\t * ----------------\n***************\n*** 135,143 ****\n \t\tmatstate->csstate.css_currentRelation = currentRelation;\n \t\tmatstate->csstate.css_currentScanDesc = currentScanDesc;\n \n- \t\tExecAssignScanType(&matstate->csstate,\n- \t\t\t\t\t\t RelationGetDescr(currentRelation));\n- \n \t\t/* ----------------\n \t\t *\tfinally set the sorted flag to true\n \t\t * ----------------\n--- 158,163 ----\n***************\n*** 178,187 ****\n {\n \tMaterialState *matstate;\n \tPlan\t *outerPlan;\n- \tTupleDesc\ttupType;\n- \tRelation\ttempDesc;\n- \n- \t/* int\t\t\t\t\t\tlen; */\n \n \t/* ----------------\n \t *\tassign the node's execution state\n--- 198,203 ----\n***************\n*** 226,237 ****\n \tExecInitNode(outerPlan, estate, (Plan *) node);\n \n \t/* ----------------\n- \t *\tinitialize matstate information\n- \t * ----------------\n- \t */\n- \tmatstate->mat_Flag = false;\n- \n- \t/* ----------------\n \t *\tinitialize tuple type.\tno need to initialize projection\n \t *\tinfo because this node doesn't do projections.\n \t * ----------------\n--- 242,247 ----\n***************\n*** 239,277 ****\n \tExecAssignScanTypeFromOuterPlan((Plan *) node, &matstate->csstate);\n \tmatstate->csstate.cstate.cs_ProjInfo = NULL;\n \n- \t/* ----------------\n- \t *\tget type information needed for ExecCreatR\n- \t * ----------------\n- \t */\n- \ttupType = ExecGetScanType(&matstate->csstate);\n- \n- \t/* ----------------\n- \t *\tExecCreatR wants its second argument to be an object id of\n- \t *\ta relation in the range table or a _NONAME_RELATION_ID\n- \t *\tindicating that the relation is not in the range table.\n- \t *\n- \t *\tIn the second case ExecCreatR creates a temp relation.\n- \t *\t(currently this is the only case we support -cim 10/16/89)\n- \t * ----------------\n- \t */\n- \t/* ----------------\n- \t *\tcreate the temporary relation\n- \t * ----------------\n- \t */\n- \ttempDesc = ExecCreatR(tupType, _NONAME_RELATION_ID_);\n- \n- \t/* ----------------\n- \t *\tsave the relation descriptor in the sortstate\n- \t * ----------------\n- \t */\n- \tmatstate->mat_TempRelation = tempDesc;\n- \tmatstate->csstate.css_currentRelation = NULL;\n- \n- \t/* ----------------\n- \t *\treturn relation oid of temporary relation in a list\n- \t *\t(someday -- for now we return LispTrue... cim 10/12/89)\n- \t * ----------------\n- \t */\n \treturn TRUE;\n }\n \n--- 249,254 ----\n***************\n*** 343,386 ****\n {\n \tMaterialState *matstate = node->matstate;\n \n \tif (matstate->mat_Flag == false)\n \t\treturn;\n \n! \tmatstate->csstate.css_currentScanDesc = ExecReScanR(matstate->csstate.css_currentRelation,\n! \t\t\t\t\t\t\t\t matstate->csstate.css_currentScanDesc,\n! \t\t\t\t\t\t\t\tnode->plan.state->es_direction, 0, NULL);\n \n }\n \n- #ifdef NOT_USED\t\t\t\t\t/* not used */\n /* ----------------------------------------------------------------\n *\t\tExecMaterialMarkPos\n * ----------------------------------------------------------------\n */\n! List\t\t\t\t\t\t\t/* nothing of interest */\n! ExecMaterialMarkPos(Material node)\n {\n! \tMaterialState matstate;\n \tHeapScanDesc scan;\n \n \t/* ----------------\n! \t *\tif we haven't materialized yet, just return NIL.\n \t * ----------------\n \t */\n! \tmatstate = get_matstate(node);\n! \tif (get_mat_Flag(matstate) == false)\n! \t\treturn NIL;\n \n \t/* ----------------\n! \t *\tXXX access methods don't return positions yet so\n! \t *\t\tfor now we return NIL.\tIt's possible that\n! \t *\t\tthey will never return positions for all I know -cim 10/16/89\n \t * ----------------\n \t */\n! \tscan = get_css_currentScanDesc((CommonScanState) matstate);\n \theap_markpos(scan);\n- \n- \treturn NIL;\n }\n \n /* ----------------------------------------------------------------\n--- 320,384 ----\n {\n \tMaterialState *matstate = node->matstate;\n \n+ \t/*\n+ \t * If we haven't materialized yet, just return. If outerplan' chgParam is\n+ \t * not NULL then it will be re-scanned by ExecProcNode, else - no\n+ \t * reason to re-scan it at all.\n+ \t */\n \tif (matstate->mat_Flag == false)\n \t\treturn;\n \n! \t/*\n! \t * If subnode is to be rescanned then we forget previous stored results;\n! \t * we have to re-read the subplan and re-store.\n! \t *\n! \t * Otherwise we can just rewind and rescan the stored output.\n! \t */\n! \tif (((Plan *) node)->lefttree->chgParam != NULL)\n! \t{\n! \t\tRelation\ttempRelation = matstate->mat_TempRelation;\n \n+ \t\tmatstate->csstate.css_currentRelation = NULL;\n+ \t\tExecCloseR((Plan *) node);\n+ \t\tExecClearTuple(matstate->csstate.css_ScanTupleSlot);\n+ \t\tif (tempRelation != NULL)\n+ \t\t\theap_drop(tempRelation);\n+ \t\tmatstate->mat_TempRelation = NULL;\n+ \t\tmatstate->mat_Flag = false;\n+ \t}\n+ \telse\n+ \t{\n+ \t\tmatstate->csstate.css_currentScanDesc =\n+ \t\t\tExecReScanR(matstate->csstate.css_currentRelation,\n+ \t\t\t\t\t\tmatstate->csstate.css_currentScanDesc,\n+ \t\t\t\t\t\tnode->plan.state->es_direction, 0, NULL);\n+ \t}\n }\n \n /* ----------------------------------------------------------------\n *\t\tExecMaterialMarkPos\n * ----------------------------------------------------------------\n */\n! void\n! ExecMaterialMarkPos(Material *node)\n {\n! \tMaterialState *matstate;\n \tHeapScanDesc scan;\n \n \t/* ----------------\n! \t *\tif we haven't materialized yet, just return.\n \t * ----------------\n \t */\n! \tmatstate = node->matstate;\n! \tif (matstate->mat_Flag == false)\n! \t\treturn;\n \n \t/* ----------------\n! \t *\tmark the scan position\n \t * ----------------\n \t */\n! \tscan = matstate->csstate.css_currentScanDesc;\n \theap_markpos(scan);\n }\n \n /* ----------------------------------------------------------------\n***************\n*** 388,412 ****\n * ----------------------------------------------------------------\n */\n void\n! ExecMaterialRestrPos(Material node)\n {\n! \tMaterialState matstate;\n \tHeapScanDesc scan;\n \n \t/* ----------------\n \t *\tif we haven't materialized yet, just return.\n \t * ----------------\n \t */\n! \tmatstate = get_matstate(node);\n! \tif (get_mat_Flag(matstate) == false)\n \t\treturn;\n \n \t/* ----------------\n \t *\trestore the scan to the previously marked position\n \t * ----------------\n \t */\n! \tscan = get_css_currentScanDesc((CommonScanState) matstate);\n \theap_restrpos(scan);\n }\n- \n- #endif\n--- 386,408 ----\n * ----------------------------------------------------------------\n */\n void\n! ExecMaterialRestrPos(Material *node)\n {\n! \tMaterialState *matstate;\n \tHeapScanDesc scan;\n \n \t/* ----------------\n \t *\tif we haven't materialized yet, just return.\n \t * ----------------\n \t */\n! \tmatstate = node->matstate;\n! \tif (matstate->mat_Flag == false)\n \t\treturn;\n \n \t/* ----------------\n \t *\trestore the scan to the previously marked position\n \t * ----------------\n \t */\n! \tscan = matstate->csstate.css_currentScanDesc;\n \theap_restrpos(scan);\n }\n",
"msg_date": "Thu, 05 Oct 2000 21:35:02 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: query returns incorrect results. "
}
] |
[
{
"msg_contents": "I got a message in my syslog stating:\n\nServer process (pid 26969) exited with status 139 at Thu Oct 5 03:02:21\n2000\n\nIt disconnected all other active processes, etc.\n\nAnyone have any idea what status 139 is?\n\n- K\n\nKristofer Munn * KMI * 732-254-9305 * AIM KrMunn * http://www.munn.com/\n\n",
"msg_date": "Fri, 6 Oct 2000 01:06:08 -0400 (EDT)",
"msg_from": "Kristofer Munn <[email protected]>",
"msg_from_op": true,
"msg_subject": "Exited with status 139"
},
{
"msg_contents": "Kristofer Munn <[email protected]> writes:\n> Anyone have any idea what status 139 is?\n\nSIGSEGV, ie, bad pointer dereference. There should be a core dump\nfile --- can you provide a stack backtrace from it?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 06 Oct 2000 01:29:20 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Exited with status 139 "
},
{
"msg_contents": "On Fri, 6 Oct 2000, Tom Lane wrote:\n\n> SIGSEGV, ie, bad pointer dereference. There should be a core dump\n> file --- can you provide a stack backtrace from it?\n\nI can try:\n\n# gdb bin/postgres core\nGNU gdb 4.17.0.11 with Linux support\nCopyright 1998 Free Software Foundation, Inc.\nGDB is free software, covered by the GNU General Public License, and you\nare\nwelcome to change it and/or distribute copies of it under certain\nconditions.\nType \"show copying\" to see the conditions.\nThere is absolutely no warranty for GDB. Type \"show warranty\" for\ndetails.\nThis GDB was configured as \"i386-redhat-linux\"...\n\nwarning: core file may not match specified executable file.\nCore was generated by `/db/pgsql/bin/postgres localhost f'.\nProgram terminated with signal 11, Segmentation fault.\nReading symbols from /lib/libcrypt.so.1...done.\nReading symbols from /lib/libnsl.so.1...done.\nReading symbols from /lib/libdl.so.2...done.\nReading symbols from /lib/libm.so.6...done.\nReading symbols from /lib/libutil.so.1...done.\nReading symbols from /usr/lib/libreadline.so.3...done.\nReading symbols from /lib/libtermcap.so.2...done.\nReading symbols from /usr/lib/libncurses.so.4...done.\nReading symbols from /lib/libc.so.6...done.\nReading symbols from /lib/ld-linux.so.2...done.\n#0 0x4013d223 in strncpy (s1=0x8975d60 \"\", s2=0x402c2d68 \"\",\nn=4294967293)\n at ../sysdeps/generic/strncpy.c:82\n../sysdeps/generic/strncpy.c:82: No such file or directory.\n(gdb) bt\n#0 0x4013d223 in strncpy (s1=0x8975d60 \"\", s2=0x402c2d68 \"\",\nn=4294967293)\n at ../sysdeps/generic/strncpy.c:82\n#1 0x8110fe5 in varcharout ()\n#2 0x808b208 in CopyTo ()\n#3 0x808ae4f in DoCopy ()\n#4 0x80ec6a8 in ProcessUtility ()\n#5 0x80eac88 in pg_exec_query_dest ()\n#6 0x80eabe7 in pg_exec_query ()\n#7 0x80ebad2 in PostgresMain ()\n#8 0x80d61c2 in DoBackend ()\n#9 0x80d5da1 in BackendStartup ()\n#10 0x80d515a in ServerLoop ()\n#11 0x80d4be4 in PostmasterMain ()\n#12 0x80ab716 in main ()\n#13 0x400f7cb3 in __libc_start_main (main=0x80ab6b0 <main>, argc=5,\n argv=0xbffffd14, init=0x8064018 <_init>, fini=0x812925c <_fini>,\n rtld_fini=0x4000a350 <_dl_fini>, stack_end=0xbffffd0c)\n at ../sysdeps/generic/libc-start.c:78\n(gdb)\n\nDid I do that right?\n\n- K\n\nKristofer Munn * KMI * 732-254-9305 * AIM KrMunn * http://www.munn.com/\n\n",
"msg_date": "Fri, 6 Oct 2000 01:38:17 -0400 (EDT)",
"msg_from": "Kristofer Munn <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Exited with status 139 "
},
{
"msg_contents": "* Kristofer Munn <[email protected]> [001005 22:41] wrote:\n> On Fri, 6 Oct 2000, Tom Lane wrote:\n> \n> > SIGSEGV, ie, bad pointer dereference. There should be a core dump\n> > file --- can you provide a stack backtrace from it?\n> \n> I can try:\n> \n> # gdb bin/postgres core\n> GNU gdb 4.17.0.11 with Linux support\n> (gdb)\n> \n> Did I do that right?\n\nHalf-way. :)\n\nYou need to do a 'up' to get out of strcpy, then do a 'list', then\nuse the 'print' command to print out any nearby variables like so:\n\nprint foo\n\nyou may also want to try 'print *foo'\n\nbest of luck,\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Thu, 5 Oct 2000 23:32:34 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Exited with status 139"
}
] |
[
{
"msg_contents": "Hi,\n\n\nWell, I discovered a strange behavior (ie 85/0.0085 <> 10000 )\nIs my SQL wrong or is it postgres ?\nWe are running 7.02 (but also 6.x) on a alpha (alphaev6) box.\n\n\nThanks in advance.\n\njc\n\nHere is a snapshot \n\njctest=> create table eigthyfive ( cs float4, csbytenthousand float4,\ncsbyhundred float);\nCREATE \n\njctest=> insert into eigthyfive VALUES (88,0.0088,0.88);\nINSERT 7973003 1\njctest=> insert into eigthyfive VALUES (86,0.0086,0.86);\nINSERT 7973004 1\njctest=> insert into eigthyfive VALUES (85,0.0085,0.85);\nINSERT 7973005 1\njctest=> insert into eigthyfive VALUES (84,0.0084,0.84);\nINSERT 7973006 1\n\njctest=> select * from eigthyfive ;\n cs | csbytenthousand | csbyhundred\n----+-----------------+-------------\n 88 | 0.0088 | 0.88\n 86 | 0.0086 | 0.86\n 85 | 0.0085 | 0.85\n 84 | 0.0084 | 0.84\n(4 rows)\n\n jctest=> select * from eigthyfive where (cs/csbytenthousand=10000);\n cs | csbytenthousand | csbyhundred\n----+-----------------+-------------\n 88 | 0.0088 | 0.88\n 86 | 0.0086 | 0.86\n 84 | 0.0084 | 0.84\n(3 rows)\n \n*** oh oh 85 disappeared ! ***\n\nBUT\n\njctest=> select (85/0.0085=10000);\n ?column?\n----------\n t\n(1 row)\n\n\njctest=> select * from eigthyfive where (cs/csbyhundred=100);\n cs | csbytenthousand | csbyhundred\n----+-----------------+-------------\n 88 | 0.0088 | 0.88\n 86 | 0.0086 | 0.86\n 85 | 0.0085 | 0.85\n 84 | 0.0084 | 0.84\n(4 rows)\n\n** 85 is back ** \n\njctest=> insert into eigthyfive VALUES (85,0.00085);\nINSERT 7973007 1\njctest=> select * from eigthyfive where (cs/csbytenthousand=100000);\n cs | csbytenthousand | csbyhundred\n----+-----------------+-------------\n 85 | 0.00085 |\n(1 row)",
"msg_date": "Fri, 06 Oct 2000 15:28:18 +0300",
"msg_from": "Jean-Christophe Pazzaglia <[email protected]>",
"msg_from_op": true,
"msg_subject": "The 85/0.0085 mistery ?"
},
{
"msg_contents": "\nNothing is wrong, that's just how computers work. Fractional numbers are\nstored in a variety of formats on different platforms. Most support the\nIEEE formats, but some use their own formats. In all cases, that I'm aware\nof, the numbers are stored as a mantissa and exponent. The numbers are\n\"base 2\", so certain base 10 numbers don't have an exact representation in\nbase 2, no matter how many bits are used in the mantissa. Other base 10\nnumbers have problems with the \"short\" mantissa formats, but are represented\nexactly with the \"long\" mantissa formats.\n\nIn general, it's recommended you don't check for exact equivalence when\ndoing calculations on real numbers and instead check for \"almost exactness.\"\nIn other words, depending on the format used on your machine, 85/0.0085 may\nbe equal to 1000.000000000000001.\n\nOf course, this may all be wrong, but that's my understanding. I think they\nmention this in the docs for the regression tests. Check them out!\n\nFred Reimer\nEclipsys Corporation\n\n\n\n> -----Original Message-----\n> From: [email protected]\n> [mailto:[email protected]]On\n> Behalf Of Jean-Christophe Pazzaglia\n> Sent: Friday, October 06, 2000 8:28 AM\n> To: PostgreSQL-development\n> Subject: [HACKERS] The 85/0.0085 mistery ?\n>\n>\n> Hi,\n>\n>\n> Well, I discovered a strange behavior (ie 85/0.0085 <> 10000 )\n> Is my SQL wrong or is it postgres ?\n> We are running 7.02 (but also 6.x) on a alpha (alphaev6) box.\n>\n>\n> Thanks in advance.\n>\n> jc\n>\n> Here is a snapshot\n>\n> jctest=> create table eigthyfive ( cs float4, csbytenthousand float4,\n> csbyhundred float);\n> CREATE\n>\n> jctest=> insert into eigthyfive VALUES (88,0.0088,0.88);\n> INSERT 7973003 1\n> jctest=> insert into eigthyfive VALUES (86,0.0086,0.86);\n> INSERT 7973004 1\n> jctest=> insert into eigthyfive VALUES (85,0.0085,0.85);\n> INSERT 7973005 1\n> jctest=> insert into eigthyfive VALUES (84,0.0084,0.84);\n> INSERT 7973006 1\n>\n> jctest=> select * from eigthyfive ;\n> cs | csbytenthousand | csbyhundred\n> ----+-----------------+-------------\n> 88 | 0.0088 | 0.88\n> 86 | 0.0086 | 0.86\n> 85 | 0.0085 | 0.85\n> 84 | 0.0084 | 0.84\n> (4 rows)\n>\n> jctest=> select * from eigthyfive where (cs/csbytenthousand=10000);\n> cs | csbytenthousand | csbyhundred\n> ----+-----------------+-------------\n> 88 | 0.0088 | 0.88\n> 86 | 0.0086 | 0.86\n> 84 | 0.0084 | 0.84\n> (3 rows)\n>\n> *** oh oh 85 disappeared ! ***\n>\n> BUT\n>\n> jctest=> select (85/0.0085=10000);\n> ?column?\n> ----------\n> t\n> (1 row)\n>\n>\n> jctest=> select * from eigthyfive where (cs/csbyhundred=100);\n> cs | csbytenthousand | csbyhundred\n> ----+-----------------+-------------\n> 88 | 0.0088 | 0.88\n> 86 | 0.0086 | 0.86\n> 85 | 0.0085 | 0.85\n> 84 | 0.0084 | 0.84\n> (4 rows)\n>\n> ** 85 is back **\n>\n> jctest=> insert into eigthyfive VALUES (85,0.00085);\n> INSERT 7973007 1\n> jctest=> select * from eigthyfive where (cs/csbytenthousand=100000);\n> cs | csbytenthousand | csbyhundred\n> ----+-----------------+-------------\n> 85 | 0.00085 |\n> (1 row)\n\n",
"msg_date": "Fri, 6 Oct 2000 08:47:23 -0400",
"msg_from": "\"Frederick W. Reimer\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: The 85/0.0085 mistery ?"
},
{
"msg_contents": "\"Frederick W. Reimer\" wrote:\n> \n> Nothing is wrong, that's just how computers work. Fractional numbers are\n> stored in a variety of formats on different platforms. Most support the\n> IEEE formats, but some use their own formats. In all cases, that I'm aware\n> of, the numbers are stored as a mantissa and exponent. The numbers are\n> \"base 2\", so certain base 10 numbers don't have an exact representation in\n> base 2, no matter how many bits are used in the mantissa. Other base 10\n> numbers have problems with the \"short\" mantissa formats, but are represented\n> exactly with the \"long\" mantissa formats.\n\nthank you for the lesson, \nI probably have done too much database this time :o\n\nwhat I was really affraid was the BUT clause ...\n\n> > jctest=> select * from eigthyfive where (cs/csbytenthousand=10000);\n> > cs | csbytenthousand | csbyhundred\n> > ----+-----------------+-------------\n> > 88 | 0.0088 | 0.88\n> > 86 | 0.0086 | 0.86\n> > 84 | 0.0084 | 0.84\n> > (3 rows)\n> >\n> > *** oh oh 85 disappeared ! ***\n> >\n> > BUT\n> >\n> > jctest=> select (85/0.0085=10000);\n> > ?column?\n> > ----------\n> > t\n> > (1 row)\n\n... It seems that this is due to the single precision\n(float4) instead of double precision (float)\n(and even with a test done in base 3 :P) \nBTW there is no round(float4) function\n\nthanks\n\njc",
"msg_date": "Fri, 06 Oct 2000 17:07:45 +0300",
"msg_from": "Jean-Christophe Pazzaglia <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: The 85/0.0085 mistery ?"
},
{
"msg_contents": "Tom Lane wrote:\n> \n> \"Frederick W. Reimer\" <[email protected]> writes:\n> > In general, it's recommended you don't check for exact equivalence when\n> > doing calculations on real numbers and instead check for \"almost exactness.\"\n> \n\n\nperhaps \"where trunc(85/0.0085) = 10000\" works better.\n\n\n> Or use datatype NUMERIC, which behaves in a more intuitive way (at least\n> for people who are accustomed to thinking in decimal). But yes, exact\n> equality checks on float values are usually foolish.\n\nBut ...\n\nhannu=# create table ntest(n numeric(3), n1 numeric(6,4));\nCREATE\nhannu=# insert into ntest values(80, 80/10000);\nINSERT 311338 1\nhannu=# insert into ntest values(81, 81/10000);\nINSERT 311339 1\nhannu=# select count(*) from ntest where n/n1 = 10000;\nERROR: division by zero on numeric\nhannu=# select * from ntest;\n n | n1 \n----+--------\n 80 | 0.0000\n 81 | 0.0000\n\neek !\n\nI understand why this is so, but should it be ?\n\nCan't we assume that 80/10000 is numeric and back-propagate that \nknowledge to constituents so that 80 and 10000 are also considered \nnumetics. \n\nOr even better assume that 80/10000 is a rational number ;)\n\n-------------------\nHannu\n",
"msg_date": "Fri, 06 Oct 2000 17:23:19 +0300",
"msg_from": "Hannu Krosing <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: The 85/0.0085 mistery ?"
},
{
"msg_contents": "Hannu Krosing wrote:\n> \n> Tom Lane wrote:\n> >\n> > \"Frederick W. Reimer\" <[email protected]> writes:\n> > > In general, it's recommended you don't check for exact equivalence when\n> > > doing calculations on real numbers and instead check for \"almost exactness.\"\n> >\n> \n> perhaps \"where trunc(85/0.0085) = 10000\" works better.\n\nok I tested it and it does not work ;(\n\nhannu=# select trunc(85.0::float4/0.0085::float4) = '10000';\nERROR: Function 'trunc(float4)' does not exist\n Unable to identify a function that satisfies the given argument types\n You may need to add explicit typecasts\nhannu=# select trunc(float8(85.0::float4/0.0085::float4));\n trunc \n-------\n 9999\n(1 row)\n\nBut this does work \n\nhannu=# select text(85.0::float4/0.0085::float4) = '10000';\n ?column? \n----------\n t\n(1 row)\n \nAs does this\n\nhannu=# select round(float8(85.0::float4/0.0085::float4));\n round \n-------\n 10000\n(1 row)\n\n--------------\nHannu\n",
"msg_date": "Fri, 06 Oct 2000 17:36:30 +0300",
"msg_from": "Hannu Krosing <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: The 85/0.0085 mistery ?"
},
{
"msg_contents": "\"Frederick W. Reimer\" <[email protected]> writes:\n> In general, it's recommended you don't check for exact equivalence when\n> doing calculations on real numbers and instead check for \"almost exactness.\"\n\nOr use datatype NUMERIC, which behaves in a more intuitive way (at least\nfor people who are accustomed to thinking in decimal). But yes, exact\nequality checks on float values are usually foolish.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 06 Oct 2000 10:58:58 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: The 85/0.0085 mistery ? "
},
{
"msg_contents": "Hannu Krosing <[email protected]> writes:\n> Can't we assume that 80/10000 is numeric\n\nI don't think so, unless you want to break a lot of existing\napplications that assume integer/integer means integer division.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 06 Oct 2000 11:44:22 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: The 85/0.0085 mistery ? "
}
] |
[
{
"msg_contents": "Ned Lilly of Great Bridge is finishing up the third day of an on-line\ninterview accessible through\n<URL:http://forums.itworld.com/[email protected]^[email protected]>.\n\nI'm the moderator of the interview, and have a professional stake in\nseeing it attract plenty of readers. More than that, though, I want\n\"outsiders\" to glimpse how serious/important/lively/... PostGres is. \nNed's handling questions gracefully. This is a good opportunity to\nbring up whatever might be on your mind.\n\n\nCameron Laird http://www.phaseit.net\nPhaseit, Inc. +1 713 763 8366 pager\[email protected] +1 281 996 8546 FAX\n",
"msg_date": "Fri, 6 Oct 2000 09:20:41 -0500 (CDT)",
"msg_from": "Cameron Laird <[email protected]>",
"msg_from_op": true,
"msg_subject": "Ned Lilly available for questioning"
}
] |
[
{
"msg_contents": "Here's some news for you:\nwww.royal.com, website for Royal Corp (you know, the cash\nregister/typewriter people) is being backed by OpenACS, which runs on\ntop of AOLserver + PostgreSQL. See the results as www.royal.com.\n\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Fri, 06 Oct 2000 13:01:22 -0400",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": true,
"msg_subject": "PostgreSQL backing www.royal.com"
}
] |
[
{
"msg_contents": "Thanks for the tip. The site is indeed impressive, fast.\nNice product lineup, too.\n\nLooked at AOLserver, but found something missing at OpenACS. Guess I better \nlook again.\n\n>===== Original Message From Lamar Owen <[email protected]> =====\n>Here's some news for you:\n>www.royal.com, website for Royal Corp (you know, the cash\n>register/typewriter people) is being backed by OpenACS, which runs on\n>top of AOLserver + PostgreSQL. See the results as www.royal.com.\n>\n>--\n>Lamar Owen\n>WGCR Internet Radio\n>1 Peter 4:11\n\n",
"msg_date": "Fri, 6 Oct 2000 13:37:59 -0400",
"msg_from": "Bill Barnes <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: PostgreSQL backing www.royal.com"
},
{
"msg_contents": "Bill Barnes wrote:\n> Thanks for the tip. The site is indeed impressive, fast.\n> Nice product lineup, too.\n \n> Looked at AOLserver, but found something missing at OpenACS. Guess I better\n> look again.\n\nwww.openacs.org\n\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Fri, 06 Oct 2000 14:46:15 -0400",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: PostgreSQL backing www.royal.com"
}
] |
[
{
"msg_contents": "Applied. Thanks.\n\n[ Charset ISO-8859-1 unsupported, converting... ]\n> \n> \tThe beos port in the source tree doesn't even compile. and even \n> after that dynamic loading isn't working and shared memory handling is \n> broken.\n> \n> \tAttached with this message, there is a Zip file which contain :\n> \n> \t* beos.diff = patch file generated with difforig\n> \t* beos = folder with beos support files which need to be moved in /\n> src/backend/port\t\n> \t* expected = foler with three file for message and precision \n> difference in regression test\n> \t* regression.diff = rule problem (need to kill the backend manualy)\t\n> \t* dynloader = dynloader files (they are also in the pacth files, \n> but there is so much modification that I have join full files)\n> \n> \tEverything works except a problem in 'rules' Is there some problems \n> with rules in the current tree ? It used to works with last week tree.\t\n> \n> \n> \t\tThanks\n> \n> \t\tcyril\n> \n> \n> \n> \tregression test results\n> \n> \n> =============== dropping old regression database... =================\n> ERROR: DROP DATABASE: Database \"regression\" does not exist\n> dropdb: database removal failed\n> =============== creating new regression database... =================\n> CREATE DATABASE\n> =============== installing languages... =================\n> installing PL/pgSQL .. ok\n> =============== running regression queries... =================\n> boolean .. ok\n> char .. ok\n> name .. ok\n> varchar .. ok\n> text .. ok\n> int2 .. ok\n> int4 .. ok\n> int8 .. ok\n> oid .. ok\n> float4 .. ok\n> float8 .. ok\n> numeric .. ok\n> strings .. ok\n> numerology .. ok\n> point .. ok\n> lseg .. ok\n> box .. ok\n> path .. ok\n> polygon .. ok\n> circle .. ok\n> date .. ok\n> time .. ok\n> timestamp .. ok\n> interval .. ok\n> abstime .. ok\n> reltime .. ok\n> tinterval .. ok\n> inet .. ok\n> comments .. ok\n> oidjoins .. ok\n> type_sanity .. ok\n> opr_sanity .. ok\n> geometry .. ok\n> horology .. ok\n> create_function_1 .. ok\n> create_type .. ok\n> create_table .. ok\n> create_function_2 .. ok\n> copy .. ok\n> constraints .. ok\n> triggers .. ok\n> create_misc .. ok\n> create_aggregate .. ok\n> create_operator .. ok\n> create_index .. ok\n> inherit .. ok\n> create_view .. ok\n> sanity_check .. ok\n> errors .. ok\n> select .. ok\n> select_into .. ok\n> select_distinct .. ok\n> select_distinct_on .. ok\n> select_implicit .. ok\n> select_having .. ok\n> subselect .. ok\n> union .. ok\n> case .. ok\n> join .. ok\n> aggregates .. ok\n> transactions .. ok\n> random .. ok\n> portals .. ok\n> arrays .. ok\n> btree_index .. ok\n> hash_index .. ok\n> misc .. ok\n> select_views .. ok\n> alter_table .. ok\n> portals_p2 .. ok\n> rules .. failed\n> foreign_key .. ok\n> limit .. ok\n> plpgsql .. ok\n> temp .. ok\n> ACTUAL RESULTS OF REGRESSION TEST ARE NOW IN FILE regress.out\n> \n> To run the optional big test(s) too, type 'make bigtest'\n> These big tests can take over an hour to complete\n> These actually are: numeric_big \n> \n> \n> \n> \n> \tdetails on modifications :\n> \n> \t\t/src/template/beos : use the new format, added some path to \n> GeekGadget in order to be able to use readline ....\n> \t\n> \t\tit now builds\n> \t\t\n> \t\t/src/bin/pg_id/pg_id.c : cheat about user (without that initdb \n> won't run at all)\n> \t\t/src/backend/utils/error/elog.c : fixed a bug in the beos \n> modification which made postgres crash in elog\n> \n> \t\tinitdb now works, postgres in interactive mode works\n> \n> \t\t/src/include/config.h.in : add beos to the platefrom without \n> unix sockets\n> \n> \t\tpostmaster and psql now runs\n> \n> \t\t* the shared memory support is completly broken (the memory is \n> not shared between backends but is in copy on write mode (due to \n> another beos fork problem)). I propose another SEM and SHM \n> implementation which is less intrusive in postgres code (I emulate all \n> the sys V functions) and which works properly.\n> \t\t* the shared library support doesn't works\n> \n> \t\t/src/backend/storage/ipc/ipc.c : remove all beos code\n> \t\t/src/backend/storage/ipc/lmgr/proc.c : remove all beos code\n> \t\t/src/backend/main/main.c : add a call to a specific init \n> function\n> \t\t/src/backend/port/Makefile.in : add beos port target\n> \t\t/src/backend/tcop/postgres.c : add a call to a specific beos \n> init function\n> \n> \t\t/src/backend/port/beos : add various support files (dynamic \n> loading, SYSV IPC support)\n> \t\t/src/backend/port/dynloader/beos.c : dynamic loading\n> \t\t/src/backend/port/dynloader/beos.h : dynamic loading\n> \n> \t\t/src/Makefile.shlib : added support for building shared lib on \n> beos\n> \t\t/src/Makefile/Makefile.beos : add target for support files in \n> include/port/beos needed for the dynamic loader and fixed the shared \n> lib building\n> \t\t/src/configure.in : added the beos template declaration (--with\n> -template might be omitted in configure)\n> \t\t/src/test/regress/regress.sh : add to the platefroms chich need \n> internet sockets\n> \t\t/src/test/regress/resultmap : add beos file mapping for message \n> end precision difference\n> \t\t/src/test/regress/expected : three files for message and \n> precision differences\n> \n> \n\n[ application/zip is not supported, skipping... ]\n\n[ application/x-be_attribute is not supported, skipping... ]\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sat, 7 Oct 2000 10:35:29 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re : BeOS port"
}
] |
[
{
"msg_contents": "There is a nice article about optimization on Freshmeat:\n\n\thttp://freshmeat.net/news/2000/10/07/970977599.html\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sat, 7 Oct 2000 16:04:01 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "optimization"
}
] |
[
{
"msg_contents": "I'm evaluating a couple of different databases for use as\nthe back-end to a web-based publishing system that's currently\nbeing developed in Java and Perl.\n\nI want to keep _all_ of the data in the database, to\nsimplify future replication and data management. That\nincludes such data as GIF images, large HTML files,\neven multi-megabyte downloadable software archives.\n\nI've been using MySQL for initial development; it has pretty\nclean and easy-to-use BLOB support. You just declare a BLOB\ncolumn type, then read and write arbitrarily large chunks of data.\nIn Perl, BLOB columns work just like varchar columns; in JDBC,\nthe getBinaryStream()/setBinaryStream() functions provide support\nfor streaming large data objects.\n\nHow well-supported is this functionality in PostgreSQL?\nI did some early experimenting with PG, but couldn't\nfind any column type that would accept binary data\n(apparently PG's parser chokes on null characters?).\n\nI've heard about TOAST, but have no idea what it really\nis, how to use it, or how well it performs. I'm leery\nof database-specific APIs.\n\n\t\t\t- Tim Kientzle\n",
"msg_date": "Sat, 07 Oct 2000 15:52:34 -0700",
"msg_from": "Tim Kientzle <[email protected]>",
"msg_from_op": true,
"msg_subject": "Using BLOBs with PostgreSQL"
},
{
"msg_contents": "On Sat, 07 Oct 2000, Tim Kientzle wrote:\n>\n> I've been using MySQL for initial development; it has pretty\n> clean and easy-to-use BLOB support. You just declare a BLOB\n> column type, then read and write arbitrarily large chunks of data.\n> In Perl, BLOB columns work just like varchar columns; in JDBC,\n> the getBinaryStream()/setBinaryStream() functions provide support\n> for streaming large data objects.\n\nIf you're talking about BLOB texts, just declare the column as text and thats \nall. In the case of binary data, I don't have an idea. I only work we text \ndata.\n\n> How well-supported is this functionality in PostgreSQL?\n> I did some early experimenting with PG, but couldn't\n> find any column type that would accept binary data\n> (apparently PG's parser chokes on null characters?).\n>\n> I've heard about TOAST, but have no idea what it really\n> is, how to use it, or how well it performs. I'm leery\n> of database-specific APIs.\n\nAs far as I have listen, it looks like a nice way to optimize searches in \nblobs. Don't know anything else.\n\nSaludos... :-)\n\n-- \n\"And I'm happy, because you make me feel good, about me.\" - Melvin Udall\n-----------------------------------------------------------------\nMart�n Marqu�s\t\t\temail: \[email protected]\nSanta Fe - Argentina\t\thttp://math.unl.edu.ar/~martin/\nAdministrador de sistemas en math.unl.edu.ar\n-----------------------------------------------------------------\n",
"msg_date": "Sat, 7 Oct 2000 20:11:03 -0300",
"msg_from": "\"Martin A. Marques\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Using BLOBs with PostgreSQL"
},
{
"msg_contents": "> On Sat, 07 Oct 2000, Tim Kientzle wrote:\n> >\n> > I've been using MySQL for initial development; it has pretty\n> > clean and easy-to-use BLOB support. You just declare a BLOB\n> > column type, then read and write arbitrarily large chunks of data.\n> > In Perl, BLOB columns work just like varchar columns; in JDBC,\n> > the getBinaryStream()/setBinaryStream() functions provide support\n> > for streaming large data objects.\n> \n> If you're talking about BLOB texts, just declare the column as text and thats \n> all. In the case of binary data, I don't have an idea. I only work we text \n> data.\n\n [General CC removed, hackers CC added.]\n\nYes, this opens a good question. We have long text fields, and bytea for\nbinary input with escapes. As far as I know, we can store binary values\nin text fields, but we have no way of getting them in there via SQL\nqueries (except for bytea using escapes for the binary values).\n\nShould be look at extending along some type of large-object style API to\nget binary values into these fields? Do we create a BLOB type that can\ntake a file name like large objects? Seems we shouldn't use TEXT for\nbinary if we can help it because it is confusing, and a BLOB type would\nallow selects on BLOB type to return nothing rather than blasting binary\ndata into the user's terminal.\n\nWhen we had the 8k limit, no one would really wanted to store binary\ndata in there because most binary values are >8k, but now that we can\nstore them, should we make a way for users to get them into TOAST\nvalues.\n\nI have on the TODO list already:\n\n\t* Add non-large-object binary field\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 9 Oct 2000 14:21:03 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Using BLOBs with PostgreSQL"
},
{
"msg_contents": "Bruce Momjian wrote:\n> > On Sat, 07 Oct 2000, Tim Kientzle wrote:\n> > > I've been using MySQL for initial development; it has pretty\n> > > clean and easy-to-use BLOB support. You just declare a BLOB\n> > > column type, then read and write arbitrarily large chunks of data.\n> >\n> > If you're talking about BLOB texts, just declare the column as text and thats\n> > all. In the case of binary data, I don't have an idea. I only work we text\n> > data.\n> \n> Yes, this opens a good question. We have long text fields, and bytea for\n> binary input with escapes. As far as I know, we can store binary values\n> in text fields, but we have no way of getting them in there via SQL\n> queries (except for bytea using escapes for the binary values).\n\n\"bytea for binary data with escapes\"??\n\nWhat would those escapes look like? I could not find any escape\nthat would put a NULL character into a bytea field.\n\n\t\t\t\t- Tim Kientzle\n",
"msg_date": "Mon, 09 Oct 2000 17:06:37 -0700",
"msg_from": "Tim Kientzle <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [GENERAL] Using BLOBs with PostgreSQL"
},
{
"msg_contents": "> > Yes, this opens a good question. We have long text fields, and bytea for\n> > binary input with escapes. As far as I know, we can store binary values\n> > in text fields, but we have no way of getting them in there via SQL\n> > queries (except for bytea using escapes for the binary values).\n> \n> \"bytea for binary data with escapes\"??\n> \n> What would those escapes look like? I could not find any escape\n> that would put a NULL character into a bytea field.\n\n'a\\\\0b' put a, NULL, b. You need double-backslash to get one backslash\nto the backend.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 9 Oct 2000 20:17:14 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Using BLOBs with PostgreSQL"
},
{
"msg_contents": "Bruce Momjian <[email protected]> writes:\n>> \"bytea for binary data with escapes\"??\n>> \n>> What would those escapes look like? I could not find any escape\n>> that would put a NULL character into a bytea field.\n\n> 'a\\\\0b' put a, NULL, b. You need double-backslash to get one backslash\n> to the backend.\n\nLooking at byteain, it's clear that you must write 3 octal digits,\nso the correct incantation is\n\ta\\000b\nin COPY input data, or\n\ta\\\\000b\nin a quoted literal in an SQL INSERT statement (since the parser will\neat one level of backslash before the type-specific converter gets it).\n\nThis strikes me as not too unreasonable for ASCII conversion of bytea.\n\nI believe that it should be possible to retrieve an unconverted form\nof a bytea value via a BINARY CURSOR, though I haven't tried it myself.\n\nYou could also do I/O without conversion using COPY BINARY, but only if\nyou are superuser, and even then the surrounding file syntax is weird\nand undocumented. COPY BINARY across stdin/stdout (which'd be needed\nfor non-superuser usage) doesn't work.\n\nI don't think there's anything particularly wrong with toasted bytea\nas a storage mechanism for non-textual BLOBs, but we'd certainly need\nto do something about the COPY situation to make storing random binary\ndata in a bytea anything except an exercise in masochism. Also it'd\nbe nice if bytea wasn't quite so impoverished operator-wise.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 09 Oct 2000 22:40:45 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [GENERAL] Using BLOBs with PostgreSQL "
}
] |
[
{
"msg_contents": "Suppose I have table a and b (b inherits a).\n\nThen I do select * from a*;\n\nNow, I'd like to know which table this particular row came from (a or b).\nIs this possible? Or do I have to have a column where I'd store what kind\nof object this is?\n\n-alex\n\n",
"msg_date": "Sat, 7 Oct 2000 19:51:38 -0400 (EDT)",
"msg_from": "Alex Pilosov <[email protected]>",
"msg_from_op": true,
"msg_subject": "inheritance/oid questions"
},
{
"msg_contents": "In the up and coming version of postgres there will be a special column\ncalled \"tableoid\" which gives the oid from pg_class of the table it came\nfrom. If you are not using the current CVS sources, you will have to\nkeep a column updated yourself.\n\nAlex Pilosov wrote:\n> \n> Suppose I have table a and b (b inherits a).\n> \n> Then I do select * from a*;\n> \n> Now, I'd like to know which table this particular row came from (a or b).\n> Is this possible? Or do I have to have a column where I'd store what kind\n> of object this is?\n> \n> -alex\n\n-- \nChris Bitmead\nmailto:[email protected]\nhttp://www.techphoto.org - Photography News, Stuff that Matters\n",
"msg_date": "Sun, 08 Oct 2000 21:54:21 +1100",
"msg_from": "Chris <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: inheritance/oid questions"
}
] |
[
{
"msg_contents": "Can I do following?\n\ncreate table foo (\nx int4 references bar*\n)\n\nOr, since 7.1 will have bar* as default for bar, will using 'references\nbar' do what I want?\n\n-alex\n\n",
"msg_date": "Sat, 7 Oct 2000 19:55:06 -0400 (EDT)",
"msg_from": "Alex Pilosov <[email protected]>",
"msg_from_op": true,
"msg_subject": "inheritance question 2/ref integrity"
},
{
"msg_contents": "\nOn Sat, 7 Oct 2000, Alex Pilosov wrote:\n\n> Can I do following?\n> \n> create table foo (\n> x int4 references bar*\n> )\n> \n> Or, since 7.1 will have bar* as default for bar, will using 'references\n> bar' do what I want?\n\nNo, and not really. Parts of it may sort of work, but referential\nactions will definately fail (this includes the default no action)\nbecause I don't believe triggers inherit in an automatic fashion.\n\nAlso, at least under 7.0 and earlier, constraints don't inherit and\nthey would have to for the above to be meaningful (the values in bar*'s\nprimary key have to be unique through all of the inheritance tree or\nmost of the references stuff simply doesn't work right).\n\n",
"msg_date": "Sat, 7 Oct 2000 17:43:20 -0700 (PDT)",
"msg_from": "Stephan Szabo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: inheritance question 2/ref integrity"
},
{
"msg_contents": "\nThe referential integrity doesn't work properly with inheritance I\nthink. It would be nice if it was fixed to work with inheritance but\nthat would probably also need inheritable indexes to work (?? I don't\nknow how much ref integrity relies on having an index in place).\n\nAlex Pilosov wrote:\n> \n> Can I do following?\n> \n> create table foo (\n> x int4 references bar*\n> )\n> \n> Or, since 7.1 will have bar* as default for bar, will using 'references\n> bar' do what I want?\n> \n> -alex\n\n-- \nChris Bitmead\nmailto:[email protected]\nhttp://www.techphoto.org - Photography News, Stuff that Matters\n",
"msg_date": "Sun, 08 Oct 2000 21:56:40 +1100",
"msg_from": "Chris <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: inheritance question 2/ref integrity"
}
] |
[
{
"msg_contents": "When I execute the following two queries, the results differ -- with the\nonly change being that another table is joined (a 1-1 join that should not\naffect the results -- I reduced down a much larger query that was\nexhibiting the behavior to what appears to be the cause). I know that\nviews have some limitations, and two of the relations used are views, so I\nbelive that that may be the problem, but I want to be sure...\n\nThanks\n\nMichael Fork - CCNA - MCP - A+\nNetwork Support - Toledo Internet Access - Toledo Ohio\n\nfootball=# SELECT play.play_id as play_id, year.correct_picks\nas ytd_correct_picks FROM tblweek_correct correct, tblplayer play,\ntblyear_correct year WHERE play.play_id = year.play_id AND play.play_id =\ncorrect.play_id AND correct.game_week = 6 AND play.play_id = 4;\n play_id | ytd_correct_picks \n---------+-------------------\n 4 | 141\n(1 row)\n\nfootball=# SELECT play.play_id as play_id, year.correct_picks as\nytd_correct_picks FROM tblplayer play, tblyear_correct year WHERE\nplay.play_id = year.play_id AND play.play_id = 4;\n play_id | ytd_correct_picks \n---------+-------------------\n 4 | 47\n(1 row)\n\nfootball=# EXPLAIN SELECT play.play_id as play_id, year.correct_picks as\nytd_correct_picks FROM tblweek_correct correct, tblplayer play,\ntblyear_correct year WHERE play.play_id = year.play_id AND play.play_id =\ncorrect.play_id AND correct.game_week = 6 AND play.play_id = 4;\nNOTICE: QUERY PLAN:\n\nAggregate (cost=108.40..108.41 rows=0 width=64)\n -> Group (cost=108.40..108.40 rows=1 width=64)\n -> Sort (cost=108.40..108.40 rows=1 width=64)\n -> Nested Loop (cost=0.00..108.39 rows=1 width=64)\n -> Nested Loop (cost=0.00..106.36 rows=1 width=56)\n -> Nested Loop (cost=0.00..104.33 rows=1\nwidth=52)\n -> Nested Loop (cost=0.00..16.54 rows=1\nwidth=40)\n -> Nested Loop (cost=0.00..14.75\nrows=1 width=36)\n -> Nested Loop\n(cost=0.00..13.30 rows=1 width=32)\n -> Nested Loop\n(cost=0.00..11.52 rows=1 width=28)\n -> Nested Loop\n(cost=0.00..5.11 rows=1 width=16)\n -> Seq Scan\non tblgame game (cost=0.00..2.08 rows=1 width=8)\n -> Index\nScan using tblgame_winner_pkey on tblgame_winner winner (cost=0.00..2.01\nrows=1 width=8)\n -> Index Scan\nusing tblpick_gameid_playid on tblpick pick (cost=0.00..5.05 rows=3\nwidth=12)\n -> Seq Scan on\ntblplayer play (cost=0.00..1.35 rows=35 width=4)\n -> Seq Scan on tblplayer play\n(cost=0.00..1.44 rows=1 width=4)\n -> Seq Scan on tblplayer play\n(cost=0.00..1.35 rows=35 width=4)\n -> Seq Scan on tblpick pick\n(cost=0.00..51.24 rows=2924 width=12)\n -> Index Scan using tblgame_pkey on tblgame\ngame (cost=0.00..2.01 rows=1 width=4)\n -> Index Scan using tblgame_winner_pkey on\ntblgame_winner winner (cost=0.00..2.01 rows=1 width=8)\n\nEXPLAIN\nfootball=# EXPLAIN SELECT play.play_id as play_id, year.correct_picks as\nytd_correct_picks FROM tblplayer play, tblyear_correct year WHERE\nplay.play_id = year.play_id AND play.play_id = 4;\nNOTICE: QUERY PLAN:\n\nAggregate (cost=101.12..101.12 rows=0 width=32)\n -> Group (cost=101.12..101.12 rows=1 width=32)\n -> Sort (cost=101.12..101.12 rows=1 width=32)\n -> Hash Join (cost=96.27..101.11 rows=1 width=32)\n -> Seq Scan on tblgame_winner winner\n(cost=0.00..1.72 rows=72 width=8)\n -> Hash (cost=96.26..96.26 rows=4 width=24)\n -> Hash Join (cost=3.40..96.26 rows=4\nwidth=24)\n -> Nested Loop (cost=0.00..91.02 rows=19\nwidth=20)\n -> Nested Loop\n(cost=0.00..3.23 rows=1 width=8)\n -> Seq Scan on tblplayer play\n(cost=0.00..1.44 rows=1 width=4)\n -> Seq Scan on tblplayer play\n(cost=0.00..1.35 rows=35 width=4)\n -> Seq Scan on tblpick pick\n(cost=0.00..51.24 rows=2924 width=12)\n -> Hash (cost=1.86..1.86 rows=86\nwidth=4)\n -> Seq Scan on tblgame game\n(cost=0.00..1.86 rows=86 width=4)\n\nEXPLAIN\nfootball=# \\d tblweek_correct\n View \"tblweek_correct\"\n Attribute | Type | Modifier \n---------------+---------+----------\n play_id | integer | \n game_week | integer | \n correct_picks | integer | \nView definition: SELECT play.play_id, game.game_week, count(*) AS\ncorrect_picks FROM tblgame game, tblpick pick, tblgame_winner winner,\ntblplayer play WHERE ((((pick.play_id = play.play_id) AND (game.game_id =\npick.game_id)) AND (pick.team_id = winner.team_id)) AND (game.game_id =\nwinner.game_id)) GROUP BY play.play_id, game.game_week;\n\nfootball=# \\d tblyear_correct\n View \"tblyear_correct\"\n Attribute | Type | Modifier \n---------------+---------+----------\n play_id | integer | \n correct_picks | integer | \nView definition: SELECT play.play_id, count(*) AS correct_picks FROM\ntblgame game, tblpick pick, tblgame_winner winner, tblplayer play WHERE\n((((pick.play_id = play.play_id) AND (game.game_id =\npick.game_id)) AND (pick.team_id = winner.team_id)) AND (game.game_id =\nwinner.game_id)) GROUP BY play.play_id;\n\nfootball=# \\d tblplayer play\n Table \"tblplayer\"\n Attribute | Type | Modifier \n---------------+-------------+---------------------------------------------------------\n play_id | integer | not null default\nnextval('tblplayer_play_id_seq'::text)\n play_name | varchar(30) | not null\n play_username | varchar(16) | not null\n play_password | varchar(16) | not null\n play_online | boolean | default 'f'\nIndices: idx_play_username,\n tblplayer_pkey\n\n\\d: extra argument 'play' ignored\nfootball=# select version();\n version \n--------------------------------------------------------\n PostgreSQL 7.0.2 on i386-pc-bsdi3.1, compiled by gcc2 \n(1 row)\n\n\n",
"msg_date": "Sun, 8 Oct 2000 16:56:44 -0400 (EDT)",
"msg_from": "Michael Fork <[email protected]>",
"msg_from_op": true,
"msg_subject": "Query returning incorrect results"
},
{
"msg_contents": "Michael Fork <[email protected]> writes:\n> When I execute the following two queries, the results differ -- with the\n> only change being that another table is joined (a 1-1 join that should not\n> affect the results -- I reduced down a much larger query that was\n> exhibiting the behavior to what appears to be the cause). I know that\n> views have some limitations, and two of the relations used are views, so I\n> belive that that may be the problem, but I want to be sure...\n\nGrouped views don't really work in 7.0.* or prior releases, except in\nthe very simplest cases. The query rewriter basically just does \"macro\nexpansion\" of the view into your query, so unless you can write out the\nequivalent query as valid SQL without using views, it won't work. In\nparticular, selecting simultaneously from two views with different\ngrouping requirements cannot work, because there's only one GROUP BY\nphase available. Unfortunately, the rewriter is too stupid to notice\nthat it can't produce a correct translation :-(\n\nThis is fixed in current sources --- a grouped view will be done as\na true sub-query.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 08 Oct 2000 22:41:15 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Query returning incorrect results "
}
] |
[
{
"msg_contents": "Savepoints will be implemented after WAL.\n\nVadim\n\n> I was wondering, is it a possibility/projected to have nested \n> transactions in postgres? Would MVCC support that?\n> \n> What I would like is following:\n> begin tran a;\n> do some work;\n> begin tran b;\n> do some other work\n> rollback b;\n> <at this point, only changes by transaction b are rolled back>\n> ... \n> commit a;\n> \n> Alternatively, is there a way to trap an exception in \n> plpgsql? [I would assume that depends on SPI ability to trap\nexceptions...]\n> \n> These features are really needed for people who do serious\n> database-backend scripting, to compete with sybase (who has multilevel\n> transactions) and oracle (who has very good exception handling).\n> \n> Thank you\n> \n> -alex\n> \n",
"msg_date": "Sun, 8 Oct 2000 20:30:43 -0700 ",
"msg_from": "\"Mikheev, Vadim\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: newbieish exception/multilevel transactions"
}
] |
[
{
"msg_contents": "Hi,\n\nI was wondering, is it a possibility/projected to have nested transactions\nin postgres? Would MVCC support that?\n\nWhat I would like is following:\nbegin tran a;\n do some work;\n begin tran b;\n do some other work\n rollback b;\n<at this point, only changes by transaction b are rolled back>\n... \ncommit a;\n\nAlternatively, is there a way to trap an exception in plpgsql? [I would\nassume that depends on SPI ability to trap exceptions...]\n\nThese features are really needed for people who do serious\ndatabase-backend scripting, to compete with sybase (who has multilevel\ntransactions) and oracle (who has very good exception handling).\n\nThank you\n\n -alex\n\n",
"msg_date": "Sun, 8 Oct 2000 23:36:23 -0400 (EDT)",
"msg_from": "Alex Pilosov <[email protected]>",
"msg_from_op": true,
"msg_subject": "newbieish exception/multilevel transactions"
}
] |
[
{
"msg_contents": "I have compiled and tested PostgreSQL 7.1devel on UnixWare 7.1. During the \nprocess, the need for changes to the FAQ_SCO document was uncovered. The \nattach patch file implements thost changes.\n\n\n\n____ | Billy G. Allie | Domain....: [email protected]\n| /| | 7436 Hartwell | Compuserve: 76337,2061\n|-/-|----- | Dearborn, MI 48126| MSN.......: [email protected]\n|/ |LLIE | (313) 582-1540 |",
"msg_date": "Sun, 08 Oct 2000 23:41:38 -0400",
"msg_from": "\"Billy G. Allie\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Patch to FAQ_SCO for PostgreSQL 7.1"
}
] |
[
{
"msg_contents": "Hello,\n\nI have found this problem running Postgre 7.0.2. I am running pg as DB\nserver\non one internet portal, so I need to solve this problem really fast. I\nintend to find\nproblem in Linux kernel (because of reported problems with VMM allocating\nshared\nmemory), but I would like to ask here, if some has some similar experience\nand maybe some solution...\n\nThis messages (still repeating) I received from backend after trying to\nrestart postmaster... :\nDEBUG: Data Base System is starting up at Wed Oct 4 18:30:42 2000\nDEBUG: Data Base System was interrupted being in production at Wed Oct 4\n18:30:40\n2000\nDEBUG: Data Base System is in production state at Wed Oct 4 18:30:42\n2000\nTRAP: Failed Assertion(\"!(metaopaque->btpo_flags & (1 << 3)):\", File:\n\"nbtpage.c\", Line: 181)!(metaopaque->btpo_flags & (1 << 3)) (0) [File or\ndirectory not found] Server process (pid 22129) exited with status 134 at\nWed Oct 4 18:30:44\n\nthanx much,\npeter\n\nps: sorry for my english... please reply to my priv. address (this one I\nwrite from),\nI didn't found any suggestiuon, how to subsribe these lists. Maybe I am\nblind... :-)))\n",
"msg_date": "Mon, 9 Oct 2000 16:06:43 +0200 ",
"msg_from": "\"VARGA, Peter\" <[email protected]>",
"msg_from_op": true,
"msg_subject": ""
}
] |
[
{
"msg_contents": "Yes, this is true. The API is not isolated for SPI, so you see\neverything, including C pollution into the C++ namespace. Not sure\nabout a solution, except renaming some of the identifiers in the\nbackend. We have discussed that, and are waiting for more reports of\nproblems before taking it on. \n\n[ Charset KOI8-R unsupported, converting... ]\n> Bruce Momjian wrote:\n> \n> > Sounds like you want to use the SPI interface.\n> \n> Thanks, I've read carefully about SPI and tried some things.\n> No problems with using C with SPI.\n> But when I've tried to compile C++ code with SPI,\n> I've found that some SPI headers use the following names:\n> class, typename. So this can't be compiled with g++.\n> Declarations like\n> typedef struct foo {/*...*/} *foo;\n> also used, and caused compilation errors.\n> \n> It is much desired to me to compile with g++.\n> Old code uses libpq++, libpq supports C++, SPI look like NOT.\n> Is it true ?\n> \n> \n> --\n> Vladimir Zolotych [email protected]\n> \n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Mon, 9 Oct 2000 14:27:21 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [ADMIN] postgres functions(.so)"
}
] |
[
{
"msg_contents": "Try this for both FK tables:\n\ncreate table tmp2(idx2 int4, col2 int4, constraint\ntmpcon2 foreign key(col2) references tmp1(idx) INITIALLY DEFERRED);\n\nThis will defer constraint checks till transaction commit...\nthough constraint triggers should use SnapshotDirty instead of\nSELECT FOR UPDATE anyway.\n\nDid you consider this, Jan?\n\nVadim\n\n> When two tables (table2 and table3) have foreign keys\n> referring to a common table(table1), I am unable to\n> have 2 concurrent transactions - one performing insert\n> on table1 and the other on table2, when the records\n> being inserted have the same foreign key. \n> \n> If I use JDBC, one of the transactions aborts.\n> If I open 2 psql sessions and try the same, one just\n> waits and does not show the prompt until the other\n> transaction has been committed or aborted.\n> \n> For example,\n> create table tmp1(idx int4, data int4);\n> create table tmp2(idx2 int4, col2 int4, constraint\n> tmpcon2 foreign key(col2) references tmp1(idx));\n> create table tmp3(idx3 int4, col3 int4, constraint\n> tmpcon3 foreign key(col3) references tmp1(idx));\n> insert into tmp1 values(1, 1);\n> \n> Transaction 1 :\n> begin work;\n> insert into tmp2 values(2, 1);\n> \n> Transaction2 :\n> begin work;\n> insert into tmp3 values(3,1);\n> \n> Since such transactions are common for me, for the\n> time-being I have dropped the foreign key constraint.\n> Any ideas ?\n> \n> Rini\n> \n> __________________________________________________\n> Do You Yahoo!?\n> Yahoo! Photos - 35mm Quality Prints, Now Get 15 Free!\n> http://photos.yahoo.com/\n> \n",
"msg_date": "Mon, 9 Oct 2000 16:11:04 -0700 ",
"msg_from": "\"Mikheev, Vadim\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: foreign key introduces unnecessary locking ?"
},
{
"msg_contents": "Mikheev, Vadim wrote:\n> Try this for both FK tables:\n>\n> create table tmp2(idx2 int4, col2 int4, constraint\n> tmpcon2 foreign key(col2) references tmp1(idx) INITIALLY DEFERRED);\n>\n> This will defer constraint checks till transaction commit...\n> though constraint triggers should use SnapshotDirty instead of\n> SELECT FOR UPDATE anyway.\n>\n> Did you consider this, Jan?\n>\n> Vadim\n\n Whenever the checks are done, the transaction inserting a new\n reference to the key must ensure that this key cannot get\n deleted until it is done and it's newly inserted reference is\n visible to others. Otherwise a referential action,\n preventing referenced key deletion (or other action) wouldn't\n see those and it would be possible to violate the constraint.\n\n I don't see any other way doing it than obtaining a lock.\n Using SnapshotDirty would mean, that one transaction could\n DELETE a reference, then another transaction removes the\n primary key (because using Dirty the DELETE is already\n visible), but now the first transaction rolls back. Voila,\n constraint violated.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#================================================== [email protected] #\n\n\n",
"msg_date": "Sun, 22 Oct 2000 07:26:12 -0500 (EST)",
"msg_from": "Jan Wieck <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: foreign key introduces unnecessary locking ?"
},
{
"msg_contents": "> > though constraint triggers should use SnapshotDirty instead of\n> > SELECT FOR UPDATE anyway.\n> >\n> > Did you consider this, Jan?\n>\n> Whenever the checks are done, the transaction inserting a new\n> reference to the key must ensure that this key cannot get\n> deleted until it is done and it's newly inserted reference is\n> visible to others. Otherwise a referential action,\n> preventing referenced key deletion (or other action) wouldn't\n> see those and it would be possible to violate the constraint.\n>\n> I don't see any other way doing it than obtaining a lock.\n> Using SnapshotDirty would mean, that one transaction could\n> DELETE a reference, then another transaction removes the\n> primary key (because using Dirty the DELETE is already\n> visible), but now the first transaction rolls back. Voila,\n> constraint violated.\n\nUsing Dirty transaction removing/updating PK could see that concurrent\nxaction attempts to update/insert FK and so would wait for its commit/abort.\nJust like now same row writers wait for each other.\n\nHaving this, we could insert FK without holding locks over PK. At the moment\nof constarint check we would see and wait concurrent PK deletion.\n\nIf two xactons will wait for each other then one of them will be aborted.\n\nThis behaviour is more natural for MVCC system postulated that\nonly same-row-writers wait for each other. Why two same FK inserters\nshould wait for each other if we can avoid this?\n\nAlso, is there any way to get deferrable PK/UK constraints? I wonder\nwhy unique index is used for them.\n\nVadim\n\n\n",
"msg_date": "Mon, 23 Oct 2000 11:22:28 -0700",
"msg_from": "\"Vadim Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: foreign key introduces unnecessary locking ?"
},
{
"msg_contents": "> Using Dirty transaction removing/updating PK could see that concurrent\n> xaction attempts to update/insert FK and so would wait for its\ncommit/abort.\n\n^^^^^^^^\nOf course this will require some function that would take tid as one of\narguments, fetch row and check if someone is updating it.\n\n> Just like now same row writers wait for each other.\n\nVadim\n\n\n",
"msg_date": "Mon, 23 Oct 2000 14:29:02 -0700",
"msg_from": "\"Vadim Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: foreign key introduces unnecessary locking ?"
}
] |
[
{
"msg_contents": "Would it be possible to implement \"generic triggers\", that is a trigger function for a whole data base and not only for a single table? \n\nAny elegant way to implement this with existing facilities (other than writing a trigger function for each of my 250+ tables) ?\n\nHorst\n\n",
"msg_date": "Tue, 10 Oct 2000 10:32:40 +1000",
"msg_from": "\"Horst Herb\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "generic triggers - implementation request"
}
] |
[
{
"msg_contents": "\naltho I'm still investigating, does anyone have an idea why the following\n'CREATE TABLE' will cause this erro:\n\nNOTICE: CREATE TABLE/PRIMARY KEY will create implicit index 'playlist_info_pkey\n' for table 'playlist_info'\nFindExec: found \"/pgsql/bin/postgres\" using argv[0]\nTRAP: Failed Assertion(\"!(metaopaque->btpo_flags & (1 << 3)):\", File: \"nbtpage.c\n\", Line: 181)\n\n!(metaopaque->btpo_flags & (1 << 3)) (0) [Inappropriate ioctl for device]\n\nCREATE TABLE \"song_info\" (\n \"song_id\" int4 NOT NULL,\n \"song_name\" character varying,\n \"genre\" character varying,\n \"duration\" character varying,\n \"artist\" character varying,\n \"album\" character varying,\n \"url\" character varying,\n \"song_seq\" int4,\n PRIMARY KEY (\"song_id\")\n);\n\n\nMarc G. Fournier [email protected]\nSystems Administrator @ hub.org \nscrappy@{postgresql|isc}.org ICQ#7615664\n\n",
"msg_date": "Mon, 9 Oct 2000 23:48:03 -0400 (EDT)",
"msg_from": "\"Marc G. Fournier\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "[7.0.2] cannot create an index ..."
},
{
"msg_contents": "\"Marc G. Fournier\" <[email protected]> writes:\n> TRAP: Failed Assertion(\"!(metaopaque->btpo_flags & (1 << 3)):\", File: \"nbtpage.c\n> \", Line: 181)\n\nHmm. That suggests a clobbered index (as in, first page zeroed out).\nThe Assert report doesn't tell us which index. Possibly a backtrace\nof the corefile would help.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 10 Oct 2000 00:25:02 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [7.0.2] cannot create an index ... "
},
{
"msg_contents": "On Tue, 10 Oct 2000, Tom Lane wrote:\n\n> \"Marc G. Fournier\" <[email protected]> writes:\n> > TRAP: Failed Assertion(\"!(metaopaque->btpo_flags & (1 << 3)):\", File: \"nbtpage.c\n> > \", Line: 181)\n> \n> Hmm. That suggests a clobbered index (as in, first page zeroed out).\n> The Assert report doesn't tell us which index. Possibly a backtrace\n> of the corefile would help.\n\nI've rebuilt the whole data structure, as I fear it is a residual of the\ncrash we had last week, but, if it means anytyhing, we had done a 'drop\ndatabase/create database' before we tried to rebuild teh database ... is\nit possible that the pg_log got corrupted and could affect this? or maybe\ntemplate1, so that when the create database was run, it was pulling in\ncorrupted system tables?\n\n\n",
"msg_date": "Tue, 10 Oct 2000 00:32:12 -0400 (EDT)",
"msg_from": "\"Marc G. Fournier\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [7.0.2] cannot create an index ... "
},
{
"msg_contents": "\"Marc G. Fournier\" <[email protected]> writes:\n>> Hmm. That suggests a clobbered index (as in, first page zeroed out).\n\n> I've rebuilt the whole data structure, as I fear it is a residual of the\n> crash we had last week, but, if it means anytyhing, we had done a 'drop\n> database/create database' before we tried to rebuild teh database ... is\n> it possible that the pg_log got corrupted and could affect this? or maybe\n> template1, so that when the create database was run, it was pulling in\n> corrupted system tables?\n\npg_log has no indexes, so it couldn't provoke this. But dropped bits in\nan index in template1 could, since createdb just does 'cp -r' to\nreplicate template1 into a new database.\n\nIf you suspect template1 is hosed, you could try doing initdb in a\nscratch directory and then do a diff of the resulting template1 against\nyour working copy.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 10 Oct 2000 00:36:17 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [7.0.2] cannot create an index ... "
}
] |
[
{
"msg_contents": "My guess is that FRONTEND is not getting defined anymore, and libpq is\npicking up the asserts from dllist.c. Now how did that change. Let me\nCc hackers.\n\n> > ../../../src/interfaces/libpq/libpq.so: undefined reference to `assert_enabled'\n> \n> Offhand it looks like someone put an Assert() call into frontend code\n> (either libpq or pg_dump itself). Bad move. I haven't updated from\n> CVS for a few days... what's changed lately?\n> \n> \t\t\tregards, tom lane\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 00:36:49 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: assert problem"
}
] |
[
{
"msg_contents": "As many of you know, several businesses are involved in providing\nPostgreSQL support.\n\nAfter careful consideration, I have decided to accept a job with Great\nBridge. There will be a press announcement tomorrow (Tuesday) with more\ndetails. I will post a URL here when I have it.\n\nInterestingly, I am the last core member to become officially attached\nto a PostgreSQL company. I have been assisting Great Bridge for some\ntime, but always in an unofficial capacity.\n\nI realize my signing on with any company will make some of you\nuncomfortable. I hope my value to the group does not suffer, and that I\ncontinue to be a positive influence for all involved. I expect my new\njob will give me even more time to continue doing the things I have done\nin the past with PostgreSQL.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 01:06:42 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "My new job"
},
{
"msg_contents": "* Bruce Momjian <[email protected]> [001009 22:11] wrote:\n> As many of you know, several businesses are involved in providing\n> PostgreSQL support.\n> \n> After careful consideration, I have decided to accept a job with Great\n> Bridge. There will be a press announcement tomorrow (Tuesday) with more\n> details. I will post a URL here when I have it.\n> \n> Interestingly, I am the last core member to become officially attached\n> to a PostgreSQL company. I have been assisting Great Bridge for some\n> time, but always in an unofficial capacity.\n> \n> I realize my signing on with any company will make some of you\n> uncomfortable. I hope my value to the group does not suffer, and that I\n> continue to be a positive influence for all involved. I expect my new\n> job will give me even more time to continue doing the things I have done\n> in the past with PostgreSQL.\n\nBruce, this is a great thing to hear, congratulations, and I hope\nyou enjoy your new position at what is looking to be a great company.\n\nAnyone with any reservations about Great Bridge need only speak to\nthe people there and the Postgresql developers already employed by\nthem to realize that this is a good thing.\n\nI've sat on the sidelines while commercial funding and support has\npoured into FreeBSD (companies hiring up our developers to work\nfull time on FreeBSD) and so far things are going great for us.\n\nbest wishes,\n-Alfred\n",
"msg_date": "Mon, 9 Oct 2000 22:21:13 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: My new job"
},
{
"msg_contents": "Bruce Momjian <[email protected]> el d�a Tue, 10 Oct 2000 01:06:42 \n-0400 (EDT), escribi�:\n\n[...]\n>After careful consideration, I have decided to accept a job with Great\n>Bridge. There will be a press announcement tomorrow (Tuesday) with more\n>details. I will post a URL here when I have it.\n\nfirst off: congratulations.\n\nsecond: what about vadim ? is he working for a postgresql company ?\n\n",
"msg_date": "Tue, 10 Oct 2000 09:56:11 -0300",
"msg_from": "\"Sergio A. Kessler\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: My new job"
},
{
"msg_contents": "[ Charset ISO-8859-1 unsupported, converting... ]\n> Bruce Momjian <[email protected]> el d?a Tue, 10 Oct 2000 01:06:42 \n> -0400 (EDT), escribi?:\n> \n> [...]\n> >After careful consideration, I have decided to accept a job with Great\n> >Bridge. There will be a press announcement tomorrow (Tuesday) with more\n> >details. I will post a URL here when I have it.\n> \n> first off: congratulations.\n> \n> second: what about vadim ? is he working for a postgresql company ?\n\nVadim is a contributor to PostgreSQL, Inc, according to the PostgreSQL,\nInc web page:\n\n\thttp://www.pgsql.com/bio/\n\n [Marc, the contributors/advisors column in my browser is not lined up\nwith the names.]\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 09:10:53 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: My new job"
},
{
"msg_contents": "On Tue, 10 Oct 2000, Sergio A. Kessler wrote:\n\n> Bruce Momjian <[email protected]> el d�a Tue, 10 Oct 2000 01:06:42 \n> -0400 (EDT), escribi�:\n> \n> [...]\n> >After careful consideration, I have decided to accept a job with Great\n> >Bridge. There will be a press announcement tomorrow (Tuesday) with more\n> >details. I will post a URL here when I have it.\n> \n> first off: congratulations.\n> \n> second: what about vadim ? is he working for a postgresql company ?\n\nvadim works for EnTrust Solutions, a partner of PostgreSQL, Inc ...\n\n\n",
"msg_date": "Tue, 10 Oct 2000 12:09:34 -0400 (EDT)",
"msg_from": "\"Marc G. Fournier\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: My new job"
},
{
"msg_contents": "Bruce Momjian writes:\n\n> After careful consideration, I have decided to accept a job with Great\n> Bridge.\n\nWhatever happened to this:\n\nDate: Tue, 09 May 2000 15:19:48 -0400\nFrom: Tom Lane <[email protected]>\nTo: Ross J. Reedstrom <[email protected]>\nCc: PostgreSQL-general <[email protected]>\nSubject: Re: [GENERAL] Steering committee responce to Great Bridge LLC\n\n: One thing we have agreed to is that there must not be an unseemly fraction\n: of core members working for the same company. With six people on core,\n: probably about two working at the same company would be a reasonable\n: limit.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Tue, 10 Oct 2000 18:46:37 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] My new job"
},
{
"msg_contents": "> Bruce Momjian writes:\n> \n> > After careful consideration, I have decided to accept a job with Great\n> > Bridge.\n> \n> Whatever happened to this:\n> \n> Date: Tue, 09 May 2000 15:19:48 -0400\n> From: Tom Lane <[email protected]>\n> To: Ross J. Reedstrom <[email protected]>\n> Cc: PostgreSQL-general <[email protected]>\n> Subject: Re: [GENERAL] Steering committee responce to Great Bridge LLC\n> \n> : One thing we have agreed to is that there must not be an unseemly fraction\n> : of core members working for the same company. With six people on core,\n> : probably about two working at the same company would be a reasonable\n> : limit.\n\nExcellent question. I suggested leaving core, but that would still mean\nmore than 1/3 of core people would be in one company. Our short-term\nsolution is to keep going until we see some problems. Our long-term\nstrategy is to increase the size of the core group.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 12:55:55 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] My new job"
},
{
"msg_contents": "Peter Eisentraut <[email protected]> writes:\n> Bruce Momjian writes:\n>> After careful consideration, I have decided to accept a job with Great\n>> Bridge.\n\n> Whatever happened to this:\n\n> From: Tom Lane <[email protected]>\n> : One thing we have agreed to is that there must not be an unseemly fraction\n> : of core members working for the same company. With six people on core,\n> : probably about two working at the same company would be a reasonable\n> : limit.\n\nI knew someone was going to bring that up ;-).\n\nThere's already been discussion of this point among core. What we\nnow have is three core members employed by Great Bridge and the\nother three either fully or partly employed by PostgreSQL Inc.\nIn one sense that's a stable situation, but on the other hand it does\nnot agree with our original informal goal of keeping any one company\nto a minority position of the core membership.\n\nNone of the core members are interested in giving up their new\npositions. En masse resignation from the core committee would preserve\nour high moral standards, perhaps, but it wouldn't do the project any\ngood that I can see. So it seems like the choices are to accept the\nstatus quo, or to appoint some more core committee members to bring\nthe numbers back where we said they should be.\n\nWhile I can think of a number of well-qualified candidates for core\nmembership, I don't much like the notion of appointing core members\njust to meet some kind of numerical quota. Also, suppose we do appoint\nmore members, and then some of them accept positions with GB or PgSQL\nInc; do we repeat the exercise indefinitely? (This is not an unlikely\nscenario, since the sort of people who'd be asked to join core are\nexactly the sort of people whom both companies would love to hire.)\n\nBottom line is we're not sure what to do now. Opinions from the \nfloor, anyone?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 10 Oct 2000 13:02:52 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] My new job "
},
{
"msg_contents": "> Peter Eisentraut <[email protected]> writes:\n> > Bruce Momjian writes:\n> >> After careful consideration, I have decided to accept a job with Great\n> >> Bridge.\n> \n> > Whatever happened to this:\n> \n> > From: Tom Lane <[email protected]>\n> > : One thing we have agreed to is that there must not be an unseemly fraction\n> > : of core members working for the same company. With six people on core,\n> > : probably about two working at the same company would be a reasonable\n> > : limit.\n> \n> I knew someone was going to bring that up ;-).\n> \n> There's already been discussion of this point among core. What we\n> now have is three core members employed by Great Bridge and the\n> other three either fully or partly employed by PostgreSQL Inc.\n\nI should mention that the Great Bridge hires are full-time employment,\nwhile not all the PostgreSQL Inc.'s are, so the Great Bridge group is\nmore in voliation of the original plan.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 13:09:24 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] My new job"
},
{
"msg_contents": "* Peter Eisentraut <[email protected]> [001010 09:47] wrote:\n> Bruce Momjian writes:\n> \n> > After careful consideration, I have decided to accept a job with Great\n> > Bridge.\n> \n> Whatever happened to this:\n> \n> Date: Tue, 09 May 2000 15:19:48 -0400\n> From: Tom Lane <[email protected]>\n> To: Ross J. Reedstrom <[email protected]>\n> Cc: PostgreSQL-general <[email protected]>\n> Subject: Re: [GENERAL] Steering committee responce to Great Bridge LLC\n> \n> : One thing we have agreed to is that there must not be an unseemly fraction\n> : of core members working for the same company. With six people on core,\n> : probably about two working at the same company would be a reasonable\n> : limit.\n\nI think Great Bridge makes a shining example of an exception to\nthat rule, the impression I got from the developers already there\nas well as the managment was very good.\n\nAnd although I loath to speak for others, you wouldn't think that\nBruce would take this position if it somehow compromised the\nintegrity of the project somehow, would you?\n\nThis is Bruce's choice, but if this was somehow put up to a vote,\nhow many of you would like to have him working full time on Postgresql\nalongside several other highly skilled developers and compensated\nfor his work rather than trying to squeeze it into his everyday\nlife like so many other opensource authors with \"real jobs\" on the\nside?\n\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Tue, 10 Oct 2000 10:09:45 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "> I think Great Bridge makes a shining example of an exception to\n> that rule, the impression I got from the developers already there\n> as well as the managment was very good.\n> \n> And although I loath to speak for others, you wouldn't think that\n> Bruce would take this position if it somehow compromised the\n> integrity of the project somehow, would you?\n> \n> This is Bruce's choice, but if this was somehow put up to a vote,\n> how many of you would like to have him working full time on Postgresql\n> alongside several other highly skilled developers and compensated\n> for his work rather than trying to squeeze it into his everyday\n> life like so many other opensource authors with \"real jobs\" on the\n> side?\n\nActually, I have written a draft article that outlines some of the\ndynamics of companies supporting open-source software. It is attached.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n\n\n\nThe Behavior of Open-Source Support Companies\n\n\n\n\n\n\n\n\n\n\n\nThe Behavior of Open-Source \nSupport Companies\nBruce Momjian\n\nDRAFT\n\n\n\nIntroduction\n\n\nOpen-source support companies face unique challenges. This article explores\na few of them:\n\n\n\nThe unusual inter-company issues faced by open-source support companies.\nThe paradox of why, sometimes, doing the best for your company may be the wrong\nthing to do.\nWhy companies should work together to preserve their shared open-source asset.\n\n\n\nOpen-Source As a Shared Resource\n\n\nTraditional companies have their own marketing, sales, research/development,\nand manufacturing departments. They are self-contained organizations that share\nvery little with other companies in the same market.\n\n\nOpen-source support companies are different. The open-source software they support\nis a shared resource. All support companies rely on the health of that\nshared resource for their livelihood, and because they rely on it, companies\ntake actions to maximize the value they derive from that shared resource. However,\nthese actions can make things worse.\n\n\n\nSelf-interest and Non-optimal Results -- The Prisoner's Dilemma\n\n\nHow can acting in your self-interest actually be counter to your self-interest?\nThe prisoner's dilemma illustrates such a case. Two prisoners are captured\nby police and placed in separate cells. The police have enough evidence to convict\neach of a minor crime that will result in a one-year prison term. However, the\npolice know the prisoners have committed a more serious crime. Each prisoner\nis told that if he confesses and the other prisoner does not confess, he will\ngo free, and the other will receive a twenty-year prison term. If they both\nconfess, they will each receive a ten years prison term. \n\n\nFor each prisoner, the decision in their self-interest is to confess.\nEach prisoner does not know what the other will do. However, confessing produces\nbetter results no matter what the other prisoner does:\n\n\n\n``Suppose the other prisoner confesses. If I confess too, I get ten years instead\nof twenty.''\n``Suppose the other prisoner does not confess. If I confess, I go free\ninstead of serving one year in jail.''\nThe interesting effect of the prisoner's dilemma is that each prisoner, acting\nin their own self-interest, produces a worse result, two ten-year jail terms,\nthan if both had not confessed and gotten only one-year jail terms.\n\n\nThe prisoner's dilemma, first formulated by Albert W. Tucker in the 1950s, has\nbeen applied to many fields, including economics, foreign policy. and philosophy[Blumen].\nThe prisoner's dilemma even applies to open-source support companies. Each company\nis like a prisoner in a cell. Each wants to dominate the open-source community,\nand fears other companies will do the same. Unfortunately, domination by multiple\ncompanies only diminishes the health of the open-source community, yielding\na worse result than if they had not acted.\n\n\nPerhaps dominate is too strong a word, but companies do position themselves\nto receive maximum benefit. When all companies do that, they can destroy the\nshared resource they rely upon. In prisoner's dilemma terms, they receive ten\nyears in jail instead of one. They reason, ``If I dominate the shared resource,\nand the other companies don't, I win. If they do, and I don't, my business suffers.''\nUnfortunately, if they both do, the community suffers, and the companies along\nwith them.\n\n\n\nCompany Behavior\n\n\nThe good news that the prisoner's dilemma is not played just once. It is played\nby open-source companies over and over again, in the little and big things they\ndo that affect their shared resource. And with repetition, there is hope. When\ncompanies realize how their actions to control the shared resource cause other\ncompanies to do the same, an arms race occurs. And once they realize\nthat, they can start to seek a truce, where companies respect the shared resource,\nrather than dominate it at every opportunity. With such cooperation, companies\nget the maximum benefit, because the shared resource remains healthy and vibrant,\nand all companies prosper.\n\n\n\nConclusion \n\n\nThis paper describes how open-source software companies can analyze their actions\nin cases the affect the open-source community. Restraint is often the best practice.\nFortunately, 99% of a company actions have no affect on the open-source community,\nso they can behave just like normal companies, seeking to grow and prosper.\n\n\n\n\n\nThe Behavior of Open-Source \nSupport Companies\nBruce Momjian\n\nDRAFT\n\n\n\nIntroduction\n\n\nOpen-source support companies face unique challenges. This article explores\na few of them:\n\n\n\nThe unusual inter-company issues faced by open-source support companies.\nThe paradox of why, sometimes, doing the best for your company may be the wrong\nthing to do.\nWhy companies should work together to preserve their shared open-source asset.\n\n\n\nOpen-Source As a Shared Resource\n\n\nTraditional companies have their own marketing, sales, research/development,\nand manufacturing departments. They are self-contained organizations that share\nvery little with other companies in the same market.\n\n\nOpen-source support companies are different. The open-source software they support\nis a shared resource. All support companies rely on the health of that\nshared resource for their livelihood, and because they rely on it, companies\ntake actions to maximize the value they derive from that shared resource. However,\nthese actions can make things worse.\n\n\n\nSelf-interest and Non-optimal Results -- The Prisoner's Dilemma\n\n\nHow can acting in your self-interest actually be counter to your self-interest?\nThe prisoner's dilemma illustrates such a case. Two prisoners are captured\nby police and placed in separate cells. The police have enough evidence to convict\neach of a minor crime that will result in a one-year prison term. However, the\npolice know the prisoners have committed a more serious crime. Each prisoner\nis told that if he confesses and the other prisoner does not confess, he will\ngo free, and the other will receive a twenty-year prison term. If they both\nconfess, they will each receive a ten years prison term. \n\n\nFor each prisoner, the decision in their self-interest is to confess.\nEach prisoner does not know what the other will do. However, confessing produces\nbetter results no matter what the other prisoner does:\n\n\n\n``Suppose the other prisoner confesses. If I confess too, I get ten years instead\nof twenty.''\n``Suppose the other prisoner does not confess. If I confess, I go free\ninstead of serving one year in jail.''\nThe interesting effect of the prisoner's dilemma is that each prisoner, acting\nin their own self-interest, produces a worse result, two ten-year jail terms,\nthan if both had not confessed and gotten only one-year jail terms.\n\n\nThe prisoner's dilemma, first formulated by Albert W. Tucker in the 1950s, has\nbeen applied to many fields, including economics, foreign policy. and philosophy[Blumen].\nThe prisoner's dilemma even applies to open-source support companies. Each company\nis like a prisoner in a cell. Each wants to dominate the open-source community,\nand fears other companies will do the same. Unfortunately, domination by multiple\ncompanies only diminishes the health of the open-source community, yielding\na worse result than if they had not acted.\n\n\nPerhaps dominate is too strong a word, but companies do position themselves\nto receive maximum benefit. When all companies do that, they can destroy the\nshared resource they rely upon. In prisoner's dilemma terms, they receive ten\nyears in jail instead of one. They reason, ``If I dominate the shared resource,\nand the other companies don't, I win. If they do, and I don't, my business suffers.''\nUnfortunately, if they both do, the community suffers, and the companies along\nwith them.\n\n\n\nCompany Behavior\n\n\nThe good news that the prisoner's dilemma is not played just once. It is played\nby open-source companies over and over again, in the little and big things they\ndo that affect their shared resource. And with repetition, there is hope. When\ncompanies realize how their actions to control the shared resource cause other\ncompanies to do the same, an arms race occurs. And once they realize\nthat, they can start to seek a truce, where companies respect the shared resource,\nrather than dominate it at every opportunity. With such cooperation, companies\nget the maximum benefit, because the shared resource remains healthy and vibrant,\nand all companies prosper.\n\n\n\nConclusion \n\n\nThis paper describes how open-source software companies can analyze their actions\nin cases the affect the open-source community. Restraint is often the best practice.\nFortunately, 99% of a company actions have no affect on the open-source community,\nso they can behave just like normal companies, seeking to grow and prosper.\n\n\nBibliography\n\nAxelrod\nAxelrod, Robert, The Complexity of Cooperation: Agent-Based Models of\nCompetition and Collaboration, Princeton University Press, 1997.\nBlumen\nBlumen, Jonathan, The Ethical Spectacle, http://www.spectacle.org/995/.\nBritanica\nEncyclopedia Britanica, <#54#>``The prisoners' dilemma,'' http://www.britannica.com/bcom/eb/article/5/0,5716,117275+10+109420,00.html?query=prisoners\nHelfrich\nHelfrich, Serge, The Prisoner's Dilemma, http://www.xs4all.nl/ helfrich/prisoner/.\nRawles\nRawls, John, A Theory of Justice, Belknap Press, 1999.",
"msg_date": "Tue, 10 Oct 2000 13:12:32 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "What is the main concern? That Great Bridge or PostgreSQL Inc will try to\ninfluence development? This is just my lowly opinion but it seems to me\nthat this could be a storm brewing in a tea cup, it just doesn't seem to be\nthat threatening a situation at a glance.\n\nCongrats to everyone on their new positions. *hats off*\n\n-Mitch\n\n----- Original Message -----\nFrom: \"Tom Lane\" <[email protected]>\nTo: \"Peter Eisentraut\" <[email protected]>\nCc: \"PostgreSQL-general\" <[email protected]>;\n\"PostgreSQL-development\" <[email protected]>\nSent: Tuesday, October 10, 2000 10:02 AM\nSubject: Re: [HACKERS] My new job\n\n\n> Peter Eisentraut <[email protected]> writes:\n> > Bruce Momjian writes:\n> >> After careful consideration, I have decided to accept a job with Great\n> >> Bridge.\n>\n> > Whatever happened to this:\n>\n> > From: Tom Lane <[email protected]>\n> > : One thing we have agreed to is that there must not be an unseemly\nfraction\n> > : of core members working for the same company. With six people on\ncore,\n> > : probably about two working at the same company would be a reasonable\n> > : limit.\n>\n> I knew someone was going to bring that up ;-).\n>\n> There's already been discussion of this point among core. What we\n> now have is three core members employed by Great Bridge and the\n> other three either fully or partly employed by PostgreSQL Inc.\n> In one sense that's a stable situation, but on the other hand it does\n> not agree with our original informal goal of keeping any one company\n> to a minority position of the core membership.\n>\n> None of the core members are interested in giving up their new\n> positions. En masse resignation from the core committee would preserve\n> our high moral standards, perhaps, but it wouldn't do the project any\n> good that I can see. So it seems like the choices are to accept the\n> status quo, or to appoint some more core committee members to bring\n> the numbers back where we said they should be.\n>\n> While I can think of a number of well-qualified candidates for core\n> membership, I don't much like the notion of appointing core members\n> just to meet some kind of numerical quota. Also, suppose we do appoint\n> more members, and then some of them accept positions with GB or PgSQL\n> Inc; do we repeat the exercise indefinitely? (This is not an unlikely\n> scenario, since the sort of people who'd be asked to join core are\n> exactly the sort of people whom both companies would love to hire.)\n>\n> Bottom line is we're not sure what to do now. Opinions from the\n> floor, anyone?\n>\n> regards, tom lane\n>\n\n",
"msg_date": "Tue, 10 Oct 2000 10:14:57 -0700",
"msg_from": "\"Mitch Vincent\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] My new job "
},
{
"msg_contents": "* Tom Lane <[email protected]> [001010 10:03] wrote:\n\n> > From: Tom Lane <[email protected]>\n> > : One thing we have agreed to is that there must not be an unseemly fraction\n> > : of core members working for the same company. With six people on core,\n> > : probably about two working at the same company would be a reasonable\n> > : limit.\n> \n> I knew someone was going to bring that up ;-).\n> \n> There's already been discussion of this point among core. What we\n> now have is three core members employed by Great Bridge and the\n> other three either fully or partly employed by PostgreSQL Inc.\n> In one sense that's a stable situation, but on the other hand it does\n> not agree with our original informal goal of keeping any one company\n> to a minority position of the core membership.\n> \n> None of the core members are interested in giving up their new\n> positions. En masse resignation from the core committee would preserve\n> our high moral standards, perhaps, but it wouldn't do the project any\n> good that I can see. So it seems like the choices are to accept the\n> status quo, or to appoint some more core committee members to bring\n> the numbers back where we said they should be.\n> \n> While I can think of a number of well-qualified candidates for core\n> membership, I don't much like the notion of appointing core members\n> just to meet some kind of numerical quota. Also, suppose we do appoint\n> more members, and then some of them accept positions with GB or PgSQL\n> Inc; do we repeat the exercise indefinitely? (This is not an unlikely\n> scenario, since the sort of people who'd be asked to join core are\n> exactly the sort of people whom both companies would love to hire.)\n> \n> Bottom line is we're not sure what to do now. Opinions from the \n> floor, anyone?\n\nI think anyone with doubts should take a good look at the initial\ncompanies backing Linux, (Redhat, VA, Debian) to see what a boon\nthis can be to project.\n\nIt is open source, so if you guys do happen to piss us off too much\nwe can always fork off our own version no? :)\n\nSo instead of panicing, it makes much more sense to ride it out and\nget a feel for where things are going, there's never going to be\nanything terribly binding that will come out of this because it is\nan opensource project.\n\nIt's much more important to continue on with the rapid pace of\ndevelopement than to fear black helicopters that haven't even\nshown up as blips on the radar.\n\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Tue, 10 Oct 2000 10:15:03 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] My new job"
},
{
"msg_contents": "I think it comes down to more of an issue of \"conflict of interest\". Worry\nthat core members will have more loyalty to the project in view of their\nemployers as opposed to the view of the project itself. That risk is\nincreased ten-fold when multiple members are in the same company.\n\nIt is tough what to say because there are basically two camps: make a rule\nnow to prevent possible issues later on, or not worry too much about it and\ndeal with it if an issue develops.\n\nAdam Lang\nSystems Engineer\nRutgers Casualty Insurance Company\n----- Original Message -----\nFrom: \"Alfred Perlstein\" <[email protected]>\nTo: \"Peter Eisentraut\" <[email protected]>\nCc: \"Bruce Momjian\" <[email protected]>; \"PostgreSQL-general\"\n<[email protected]>\nSent: Tuesday, October 10, 2000 1:09 PM\nSubject: Re: [GENERAL] Re: [HACKERS] My new job\n\n\n> * Peter Eisentraut <[email protected]> [001010 09:47] wrote:\n> > Bruce Momjian writes:\n> >\n> > > After careful consideration, I have decided to accept a job with Great\n> > > Bridge.\n> >\n> > Whatever happened to this:\n> >\n> > Date: Tue, 09 May 2000 15:19:48 -0400\n> > From: Tom Lane <[email protected]>\n> > To: Ross J. Reedstrom <[email protected]>\n> > Cc: PostgreSQL-general <[email protected]>\n> > Subject: Re: [GENERAL] Steering committee responce to Great Bridge LLC\n> >\n> > : One thing we have agreed to is that there must not be an unseemly\nfraction\n> > : of core members working for the same company. With six people on\ncore,\n> > : probably about two working at the same company would be a reasonable\n> > : limit.\n>\n> I think Great Bridge makes a shining example of an exception to\n> that rule, the impression I got from the developers already there\n> as well as the managment was very good.\n>\n> And although I loath to speak for others, you wouldn't think that\n> Bruce would take this position if it somehow compromised the\n> integrity of the project somehow, would you?\n>\n> This is Bruce's choice, but if this was somehow put up to a vote,\n> how many of you would like to have him working full time on Postgresql\n> alongside several other highly skilled developers and compensated\n> for his work rather than trying to squeeze it into his everyday\n> life like so many other opensource authors with \"real jobs\" on the\n> side?\n>\n> --\n> -Alfred Perlstein - [[email protected]|[email protected]]\n> \"I have the heart of a child; I keep it in a jar on my desk.\"\n\n",
"msg_date": "Tue, 10 Oct 2000 13:37:33 -0400",
"msg_from": "\"Adam Lang\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "At 01:02 PM 10/10/00 -0400, Tom Lane wrote:\n\n>Bottom line is we're not sure what to do now. Opinions from the \n>floor, anyone?\n\nYeah, quit worrying and work your collective butts off on 7.1 and 7.2 :)\n\nSeriously...the core group is obviously committed to PG, and appear to \nbe folks of integrity. We all will benefit by your working on PG full\ntime while being paid enough so you can eat, drink, and be merry, too.\n\n\n\n- Don Baccus, Portland OR <[email protected]>\n Nature photos, on-line guides, Pacific Northwest\n Rare Bird Alert Service and other goodies at\n http://donb.photo.net.\n",
"msg_date": "Tue, 10 Oct 2000 10:42:42 -0700",
"msg_from": "Don Baccus <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] My new job "
},
{
"msg_contents": "> At 01:02 PM 10/10/00 -0400, Tom Lane wrote:\n> \n> >Bottom line is we're not sure what to do now. Opinions from the \n> >floor, anyone?\n> \n> Yeah, quit worrying and work your collective butts off on 7.1 and 7.2 :)\n\nC'mon, Don. Stop beating around the bush. Tell us what you really\nthink. :-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 13:51:24 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] My new job"
},
{
"msg_contents": "On Tue, Oct 10, 2000 at 01:37:33PM -0400, Adam Lang wrote:\n> that core members will have more loyalty to the project in view of their\n> employers as opposed to the view of the project itself. That risk is\n> increased ten-fold when multiple members are in the same company.\n\nHey, I don't think this discussion makes a lot of sense. Yes, I do see the\nrisks for the project, but Bruce has to do whatever is best for his life. If\nyou are offered your dream job, would you really back off just because there\nare too many of your friends working their too?\n\nAfter all IF GreatBridge were to take over PostgreSQL the core members are\nfree to keep on working on a free version in their spare time. So that's\nabout the same situation as it is now.\n\nOkay they may stop liking to do that, but that can happen regardless of what\njob they are in. After all working on free software takes away\nquite a lot of your spare time. So many become tired of this once their real\nlife takes its toll.\n\nMichael\n-- \nMichael Meskes\[email protected]\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Tue, 10 Oct 2000 20:51:55 +0200",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "Alfred Perlstein <[email protected]> writes:\n> I think anyone with doubts should take a good look at the initial\n> companies backing Linux, (Redhat, VA, Debian) to see what a boon\n> this can be to project.\n\nI'll just clarify that Debian is not a company, it is an non-profit\nall-volunteer effort.\n\nI agree with everything else that Alfred says.\n\nMike.\n",
"msg_date": "10 Oct 2000 14:52:53 -0400",
"msg_from": "Michael Alan Dorman <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] My new job"
},
{
"msg_contents": "On Tue, Oct 10, 2000 at 10:15:03AM -0700, Alfred Perlstein wrote:\n> I think anyone with doubts should take a good look at the initial\n> companies backing Linux, (Redhat, VA, Debian) to see what a boon\n\nI certainly get your point but I have to correct this as Debian is not,\nnever has been and never will be a company. Sorry, couldn't resist. :-)\n\nMichael\n-- \nMichael Meskes\[email protected]\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Tue, 10 Oct 2000 20:54:47 +0200",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] My new job"
},
{
"msg_contents": "Is the following expected behavior for a UNION query with ORDER BY:\n\nexecuting this query:\n\n+++++++++++++++++++++++++++++++++++++++++++\n\nSELECT a.attnum as number, \n a.attname as attribute,\n CASE WHEN t.typname = 'varchar' THEN \n t.typname || '(' || a.atttypmod - 4 || ')' \n ELSE \n t.typname \n END as type,\n CASE WHEN a.attnotnull = 't' THEN \n 'not null '::text ELSE ''::text \n END || 'default ' || \n CASE WHEN a.atthasdef = 't' THEN \n substring(d.adsrc for 128)::text \n ELSE ''::text END as modifier\nFROM pg_class c, \n pg_attribute a, \n pg_type t,\n pg_attrdef d\nWHERE c.relname = 'tblplayer' AND \n a.attnum > 0 AND \n a.attrelid = c.oid AND \n a.atttypid = t.oid AND\n c.oid = d.adrelid AND\n d.adnum = a.attnum\nUNION ALL \nSELECT a.attnum as number, \n a.attname as attribute, \n CASE WHEN t.typname = 'varchar' THEN \n t.typname || '(' || a.atttypmod - 4 || ')' \n ELSE \n t.typname \n END as type,\n CASE WHEN a.attnotnull = 't' THEN \n 'not null '::text \n ELSE \n ''::text \n END as modifier\nFROM pg_class c, \n pg_attribute a, \n pg_type t\nWHERE c.relname = 'tblplayer' AND \n a.attnum > 0 AND \n a.attrelid = c.oid AND \n a.atttypid = t.oid AND\n a.attname NOT IN (SELECT a.attname \n FROM pg_class c, \n pg_attribute a, \n pg_attrdef d\n WHERE c.relname = 'tblplayer' AND \n a.attnum > 0 AND \n a.attrelid = c.oid AND \n a.atttypid = t.oid AND \n c.oid = d.adrelid AND \n d.adnum = a.attnum)\nORDER BY a.attnum;\n\n+++++++++++++++++++++++++++++++++++++++\n\nyields\n\n number | attribute | type | modifier \n--------+---------------+-------------+--------------------------------\n 1 | play_id | int4 | not null default nextval('...\n 2 | play_name | varchar(30) | not null \n 3 | play_username | varchar(16) | not null \n 4 | play_password | varchar(16) | not null \n 5 | play_online | bool | default 'f'\n\n\nHowever, if I execute the same query and drop \"a.attnum as number\" from\nthe select part, it returns the following:\n\n attribute | type | modifier \n---------------+-------------+--------------------------------\n play_id | int4 | not null default nextval('...\n play_online | bool | default 'f'\n play_name | varchar(30) | not null \n play_username | varchar(16) | not null \n play_password | varchar(16) | not null \n\nwhich is incorrect accoring to the initial query. It appears to be\nordering the individual selects and then appending the second query to\nthe first -- is this correct?\n\nThanks\n\nMichael Fork - CCNA - MCP - A+\nNetwork Support - Toledo Internet Access - Toledo Ohio\n\n\n",
"msg_date": "Tue, 10 Oct 2000 15:56:56 -0400 (EDT)",
"msg_from": "Michael Fork <[email protected]>",
"msg_from_op": false,
"msg_subject": "ORDER BY and UNION"
},
{
"msg_contents": "Bruce Momjian writes:\n\n> Excellent question. I suggested leaving core, but that would still mean\n> more than 1/3 of core people would be in one company. Our short-term\n> solution is to keep going until we see some problems. Our long-term\n> strategy is to increase the size of the core group.\n\nIn the end, PostgreSQL is still controlled by those who contribute the\nwork, so it doesn't matter who's in core and who's not. Your leaving core\nwould certainly have been the worse \"fix\". But I'm pleased that this\nagreement was at least remembered.\n\nThose who hang around GNU toolchain/build tools development lists may\nrecall occasional annoyances that design or implementation decisions are\napparently made on Cygnus-internal mailing lists. Even if those cases\nmight have been unintended in hindsight, this is the sort of stuff that\none needs to be aware of.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Tue, 10 Oct 2000 22:11:50 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] My new job"
},
{
"msg_contents": "I wasn't judging. I was mentioning to others what the concerns probably\nwere. Also, it isn't a concern of \"Company B\" taking over. It is of the\npossibility of development put in the direction that best benefits of\nCompany B as opposed to the project itself. And again, yes, the other core\nmembers can tell them to \"blow it out their arse\" but then you have a\nsituation of them either going on their own and doing a \"splinter\" or just\nquitting in general and the direction is then even more put in the direction\nof Company B.\n\nIt is merely a conflict of interest issue. Same issue in law as having\nattorney's from the same firm on the side of defendant and plaintiff. If\nthe plaintiff is a multi-million dollar client and the defendant is\npro-bono... there is concern about bias.\n\nCongratulations is in order for Mr. Momjian. I'm not saying he should or\nshouldn't work for them. I'm just playing devil's advocate.\n\nAdam Lang\nSystems Engineer\nRutgers Casualty Insurance Company\n----- Original Message -----\nFrom: \"Michael Meskes\" <[email protected]>\nTo: \"PostgreSQL-general\" <[email protected]>\nSent: Tuesday, October 10, 2000 2:51 PM\nSubject: Re: [GENERAL] Re: [HACKERS] My new job\n\n\n> On Tue, Oct 10, 2000 at 01:37:33PM -0400, Adam Lang wrote:\n> > that core members will have more loyalty to the project in view of their\n> > employers as opposed to the view of the project itself. That risk is\n> > increased ten-fold when multiple members are in the same company.\n>\n> Hey, I don't think this discussion makes a lot of sense. Yes, I do see the\n> risks for the project, but Bruce has to do whatever is best for his life.\nIf\n> you are offered your dream job, would you really back off just because\nthere\n> are too many of your friends working their too?\n>\n> After all IF GreatBridge were to take over PostgreSQL the core members are\n> free to keep on working on a free version in their spare time. So that's\n> about the same situation as it is now.\n>\n> Okay they may stop liking to do that, but that can happen regardless of\nwhat\n> job they are in. After all working on free software takes away\n> quite a lot of your spare time. So many become tired of this once their\nreal\n> life takes its toll.\n>\n> Michael\n> --\n> Michael Meskes\n> [email protected]\n> Go SF 49ers! Go Rhein Fire!\n> Use Debian GNU/Linux! Use PostgreSQL!\n\n",
"msg_date": "Tue, 10 Oct 2000 16:25:00 -0400",
"msg_from": "\"Adam Lang\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "Michael Fork <[email protected]> writes:\n> However, if I execute the same query and drop \"a.attnum as number\" from\n> the select part, it returns the following:\n> ...\n> which is incorrect accoring to the initial query. It appears to be\n> ordering the individual selects and then appending the second query to\n> the first -- is this correct?\n\nI believe that this query should not be considered valid --- and, in\nfact, current sources will return an error if you try to ORDER a UNION\nresult by something that's not one of the output columns of the UNION.\n\nThe issue is that if you are union'ing arbitrary queries together,\nhow do you decide what the ORDER BY expression means in the context\nof each component query? Consider\n\n\tselect a, b from tab1\n\tUNION\n\tselect c, d from tab2\n\tORDER BY z;\n\nEven assuming that there are columns named z in both tables, the ORDER\nBY would be exceeding its authority to assume that those columns are\nwhat is meant. Furthermore, since we're doing a UNION here, the result\nwill be reduced to just the unique output rows, meaning that there might\nbe more than one possible z value for each output row; so the sort order\nwouldn't be well-defined.\n\nIt seems to me that for UNION-type queries we need to stick to the\nletter of the SQL standard and only allow ORDER BY an output column\nname. In my example you'd be allowed to do \"ORDER BY a\" or equivalently\n\"ORDER BY 1\", but not \"ORDER BY z\".\n\nExisting releases fail to defend against this situation, and produce a\nplan that does who-knows-what. In current sources you'll get an error:\n\nregression=# select q2 from int8_tbl union all select q2 from int8_tbl\nregression-# order by int8_tbl.q1;\nERROR: ORDER BY on a UNION/INTERSECT/EXCEPT result must be on one of the result columns\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 10 Oct 2000 16:44:30 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: ORDER BY and UNION "
},
{
"msg_contents": "> It is merely a conflict of interest issue. Same issue in law as having\n> attorney's from the same firm on the side of defendant and plaintiff. If\n> the plaintiff is a multi-million dollar client and the defendant is\n> pro-bono... there is concern about bias.\n> \n> Congratulations is in order for Mr. Momjian. I'm not saying he should or\n> shouldn't work for them. I'm just playing devil's advocate.\n\nThe interesting issue here is that law firms by nature are adviserial. \nMy paper showed that most issues are actually ones of companies managing\na shared resource, so the analogy is not quite accurate.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 16:49:02 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "Bruce Momjian <[email protected]> writes:\n\n> > It is merely a conflict of interest issue. Same issue in law as having\n> > attorney's from the same firm on the side of defendant and plaintiff. If\n> > the plaintiff is a multi-million dollar client and the defendant is\n> > pro-bono... there is concern about bias.\n> > \n> > Congratulations is in order for Mr. Momjian. I'm not saying he should or\n> > shouldn't work for them. I'm just playing devil's advocate.\n> \n> The interesting issue here is that law firms by nature are adviserial. \n> My paper showed that most issues are actually ones of companies managing\n> a shared resource, so the analogy is not quite accurate.\n\nOTOH, Red Hat, Inc. employs most of the gcc developers and by far the\nmost GDB developers(80-90% of commits). There are still mechanisms in\nplace to hinder that we have any sort of control over these...\n\n-- \nTrond Eivind Glomsr�d\nRed Hat, Inc.\n",
"msg_date": "10 Oct 2000 16:57:36 -0400",
"msg_from": "[email protected] (Trond Eivind =?iso-8859-1?q?Glomsr=F8d?=)",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "[ Charset ISO-8859-1 unsupported, converting... ]\n> Bruce Momjian <[email protected]> writes:\n> \n> > > It is merely a conflict of interest issue. Same issue in law as having\n> > > attorney's from the same firm on the side of defendant and plaintiff. If\n> > > the plaintiff is a multi-million dollar client and the defendant is\n> > > pro-bono... there is concern about bias.\n> > > \n> > > Congratulations is in order for Mr. Momjian. I'm not saying he should or\n> > > shouldn't work for them. I'm just playing devil's advocate.\n> > \n> > The interesting issue here is that law firms by nature are adviserial. \n> > My paper showed that most issues are actually ones of companies managing\n> > a shared resource, so the analogy is not quite accurate.\n> \n> OTOH, Red Hat, Inc. employs most of the gcc developers and by far the\n> most GDB developers(80-90% of commits). There are still mechanisms in\n> place to hinder that we have any sort of control over these...\n\nYes, as I said in my paper, controlling it is actually against the best\ninterests of the company.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 17:02:51 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "I agree. My analogy was more in describing the conflict of interest, not\nnecessarily the shared resource issue.\n\nAgain, I'm not trying to come off on the wrong end. Just many people seemed\nto take the situation a little too easy and wondered why the issue even\nwould have come up. In some instances it really is an issue that groups\nhave to concern themselves with.\n\nAdam Lang\nSystems Engineer\nRutgers Casualty Insurance Company\n----- Original Message -----\nFrom: \"Bruce Momjian\" <[email protected]>\nTo: \"Adam Lang\" <[email protected]>\nCc: \"PostgreSQL-general\" <[email protected]>\nSent: Tuesday, October 10, 2000 4:49 PM\nSubject: Re: [GENERAL] Re: [HACKERS] My new job\n\n\n> > It is merely a conflict of interest issue. Same issue in law as having\n> > attorney's from the same firm on the side of defendant and plaintiff.\nIf\n> > the plaintiff is a multi-million dollar client and the defendant is\n> > pro-bono... there is concern about bias.\n> >\n> > Congratulations is in order for Mr. Momjian. I'm not saying he should\nor\n> > shouldn't work for them. I'm just playing devil's advocate.\n>\n> The interesting issue here is that law firms by nature are adviserial.\n> My paper showed that most issues are actually ones of companies managing\n> a shared resource, so the analogy is not quite accurate.\n>\n> --\n> Bruce Momjian | http://candle.pha.pa.us\n> [email protected] | (610) 853-3000\n> + If your life is a hard drive, | 830 Blythe Avenue\n> + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n\n",
"msg_date": "Tue, 10 Oct 2000 17:03:22 -0400",
"msg_from": "\"Adam Lang\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "Adam Lang wrote:\n> \n> I wasn't judging. I was mentioning to others what the concerns probably\n> were. Also, it isn't a concern of \"Company B\" taking over. It is of the\n> possibility of development put in the direction that best benefits of\n> Company B as opposed to the project itself. And again, yes, the other core\n> members can tell them to \"blow it out their arse\" but then you have a\n> situation of them either going on their own and doing a \"splinter\" or just\n> quitting in general and the direction is then even more put in the direction\n> of Company B.\n> \n> It is merely a conflict of interest issue. Same issue in law as having\n> attorney's from the same firm on the side of defendant and plaintiff. If\n> the plaintiff is a multi-million dollar client and the defendant is\n> pro-bono... there is concern about bias.\n> \n> Congratulations is in order for Mr. Momjian. I'm not saying he should or\n> shouldn't work for them. I'm just playing devil's advocate.\n> \n> Adam Lang\n> Systems Engineer\n> Rutgers Casualty Insurance Company\n>\n\nWell to calm any fears of Great Bridge taking over what exactly are the\nterms of employment? Are the developers merely continuing on with what\nthey were working on and now getting paid for it, or is Great Bridge\nsaying here are the projects we want done so do it. \n-- \nDave Smith\nCandata Systems Ltd.\n(416) 493-9020\[email protected]\n",
"msg_date": "Tue, 10 Oct 2000 17:06:37 -0400",
"msg_from": "Dave Smith <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "[email protected] (Trond Eivind =?iso-8859-1?q?Glomsr=F8d?=) writes:\n> OTOH, Red Hat, Inc. employs most of the gcc developers and by far the\n> most GDB developers(80-90% of commits). There are still mechanisms in\n> place to hinder that we have any sort of control over these...\n\nWhat sort of mechanisms? Perhaps we need to borrow some ideas from\nyour situation.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 10 Oct 2000 17:10:53 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "\"Adam Lang\" <[email protected]> writes:\n> I wasn't judging. I was mentioning to others what the concerns probably\n> were. Also, it isn't a concern of \"Company B\" taking over. It is of the\n> possibility of development put in the direction that best benefits of\n> Company B as opposed to the project itself.\n> ...\n> It is merely a conflict of interest issue.\n\nRight, exactly. That was why we originally suggested putting a limit on\nthe number of core members employed by any one company: to reduce both\nthe actual and perceived potential for core decisions being taken in a\nway that is more for the benefit of some company than for the project as\na whole.\n\nI am not sure that the *real* potential for bad choices is all that\nhigh. I think all the core members understand very well that we are\nstewards of a shared resource, and in the long run decisions counter\nto the community-wide best interest will also not be in the best\ninterest of our companies. But it's also important that the rest of\nthe PG community *perceive* that core decisions are well-founded.\n\nHowever, given recent events the original two-of-six idea isn't feasible\nany more --- and certainly none of us were going to tell Bruce that he\ncouldn't take that job because that'd make three GB employees on core.\nSo the question is, what do we do now?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 10 Oct 2000 17:25:08 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "Tom Lane <[email protected]> writes:\n\n> [email protected] (Trond Eivind =?iso-8859-1?q?Glomsr=F8d?=) writes:\n> > OTOH, Red Hat, Inc. employs most of the gcc developers and by far the\n> > most GDB developers(80-90% of commits). There are still mechanisms in\n> > place to hinder that we have any sort of control over these...\n> \n> What sort of mechanisms? Perhaps we need to borrow some ideas from\n> your situation.\n\nSteering committees, where one entitity (university, company etc) are\nrestricted to a maximum quota of members.\n\nhttp://gcc.gnu.org/steering.html\n\n-- \nTrond Eivind Glomsr�d\nRed Hat, Inc.\n",
"msg_date": "10 Oct 2000 17:29:11 -0400",
"msg_from": "[email protected] (Trond Eivind =?iso-8859-1?q?Glomsr=F8d?=)",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "> I am not sure that the *real* potential for bad choices is all that\n> high. I think all the core members understand very well that we are\n> stewards of a shared resource, and in the long run decisions counter\n> to the community-wide best interest will also not be in the best\n> interest of our companies. But it's also important that the rest of\n> the PG community *perceive* that core decisions are well-founded.\n\nI have already had to reject a Great Bridge request, saying, \"This is\nthe first of many disappointments I will deliver.\" When I explained\nthat I was better as someone objective than as a \"yes\" man, they got the\npoint.\n\nFortunately, since I wrote the article, I think things are clearer now. \nThey understand the trade-off of open-source control.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 17:39:48 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "On Tue, 10 Oct 2000, Tom Lane wrote:\n\n> \"Adam Lang\" <[email protected]> writes:\n> > I wasn't judging. I was mentioning to others what the concerns probably\n> > were. Also, it isn't a concern of \"Company B\" taking over. It is of the\n> > possibility of development put in the direction that best benefits of\n> > Company B as opposed to the project itself.\n> > ...\n> > It is merely a conflict of interest issue.\n> \n> Right, exactly. That was why we originally suggested putting a limit on\n> the number of core members employed by any one company: to reduce both\n> the actual and perceived potential for core decisions being taken in a\n> way that is more for the benefit of some company than for the project as\n> a whole.\n> \n> I am not sure that the *real* potential for bad choices is all that\n> high. I think all the core members understand very well that we are\n> stewards of a shared resource, and in the long run decisions counter\n> to the community-wide best interest will also not be in the best\n> interest of our companies. But it's also important that the rest of\n> the PG community *perceive* that core decisions are well-founded.\n> \n> However, given recent events the original two-of-six idea isn't feasible\n> any more --- and certainly none of us were going to tell Bruce that he\n> couldn't take that job because that'd make three GB employees on core.\n> So the question is, what do we do now?\n\nFor now, nothing. If it never becomes a problem then it's moot - besides,\nnow you're out of core members to go anywhere :) Yeah I know, someone can\nalways move from pgsql to gb, but the same can go in the other direction.\nMake all decisions cautiously and in the best interest of PostgreSQL and\nthere should be no problems. \n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Tue, 10 Oct 2000 17:48:59 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "Dave Smith <[email protected]> writes:\n> Well to calm any fears of Great Bridge taking over what exactly are the\n> terms of employment? Are the developers merely continuing on with what\n> they were working on and now getting paid for it, or is Great Bridge\n> saying here are the projects we want done so do it. \n\nFWIW, I've been employed by Great Bridge since 1 August, and so far\nthey haven't said word one about what I should be working on; \"do what\nyou think is needed\" are the sum total of my orders.\n\nThis happy state of affairs may not last forever --- in particular,\nonce GB has actual customers I will become one of their last-resort tech\nsupport people, and so some amount of time will go into responding to\ncustomer bug reports. Of course, I do a lot of bug-fix work anyway.\nWhat I foresee is that I'll put higher priority on fixing bugs reported\nby paying customers than on fixing those reported via the mailing lists.\nBut, hey, those people are paying for something; and it's not like I\nplan to stop reading the lists.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 10 Oct 2000 18:10:25 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "On Tue, 10 Oct 2000, Michael Meskes wrote:\n\n> On Tue, Oct 10, 2000 at 10:15:03AM -0700, Alfred Perlstein wrote:\n> > I think anyone with doubts should take a good look at the initial\n> > companies backing Linux, (Redhat, VA, Debian) to see what a boon\n> \n> I certainly get your point but I have to correct this as Debian is not,\n> never has been and never will be a company. Sorry, couldn't resist. :-)\n\nand last I heard, RedHat doesn't necessarily have the best name ...\n\n\n",
"msg_date": "Tue, 10 Oct 2000 19:54:14 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "The Hermit Hacker <[email protected]> writes:\n\n> On Tue, 10 Oct 2000, Michael Meskes wrote:\n> \n> > On Tue, Oct 10, 2000 at 10:15:03AM -0700, Alfred Perlstein wrote:\n> > > I think anyone with doubts should take a good look at the initial\n> > > companies backing Linux, (Redhat, VA, Debian) to see what a boon\n> > \n> > I certainly get your point but I have to correct this as Debian is not,\n> > never has been and never will be a company. Sorry, couldn't resist. :-)\n> \n> and last I heard, RedHat doesn't necessarily have the best name ...\n\nRed Hat-bashing doesn't change the fact that Red Hat employees is by\nfar the largest corporate (or other single entity) contributor open\nsource projects. Project on which we contribute a lot include gcc, gdb\n(through former Cygnus and other employees, we are by far the biggest\nthere), rpm, XFree86, glibc, gtk+, gnome, the Linux kernel and apache.\n\nWe also try hard to feed patches back to the original authors when we fix\nsomething generic.\n\nAnyway, flamewars never serve any particular purpose - followups\nshould go to /dev/null\n-- \nTrond Eivind Glomsr�d\nRed Hat, Inc.\n",
"msg_date": "10 Oct 2000 19:08:29 -0400",
"msg_from": "[email protected] (Trond Eivind =?iso-8859-1?q?Glomsr=F8d?=)",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "> Dave Smith <[email protected]> writes:\n> > Well to calm any fears of Great Bridge taking over what exactly are the\n> > terms of employment? Are the developers merely continuing on with what\n> > they were working on and now getting paid for it, or is Great Bridge\n> > saying here are the projects we want done so do it. \n> \n> FWIW, I've been employed by Great Bridge since 1 August, and so far\n> they haven't said word one about what I should be working on; \"do what\n> you think is needed\" are the sum total of my orders.\n> \n> This happy state of affairs may not last forever --- in particular,\n> once GB has actual customers I will become one of their last-resort tech\n> support people, and so some amount of time will go into responding to\n\nHe is my first-resort bug fixer, but of course, he was before anyway. :-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 19:31:41 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "\"Adam Lang\" <[email protected]> writes:\n\n> I wasn't judging. I was mentioning to others what the concerns probably\n> were. Also, it isn't a concern of \"Company B\" taking over. It is of the\n> possibility of development put in the direction that best benefits of\n> Company B as opposed to the project itself. And again, yes, the other core\n\nWho decides what is in the best interest of the project itself ? A\ncommunity is so diverse that there is alot of conflicting interests.\n\nmvh, \n\n\tGunnar\n",
"msg_date": "11 Oct 2000 01:27:15 +0100",
"msg_from": "Gunnar R|nning <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "Bruce Momjian <[email protected]> writes:\n\n> more in voliation of the original plan.\n> \n\nviolation ? Or is this just another gap in my knowledge of the English\nlanguage ? \n\n\tGunnar\n",
"msg_date": "11 Oct 2000 01:31:05 +0100",
"msg_from": "Gunnar R|nning <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "As promised, the press release is at:\n\n\thttp://www.greatbridge.com/news/p_101020001.html\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 20:48:36 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "Adam Haberlach <[email protected]> writes:\n\n> \tAll this whinging about \"corperate direction\" is really meaningless\n> unless you are prepared to jump ship or split off in a clone of the\n> original one. The code is what you do with it. We are all lucky that\n> it is as good and useful as it is right now.\n\n\nHallejuja. OK, I'm not Christian, but I agree very much in this sentiment.\n\nRegards, \n\n\tGunnar\n",
"msg_date": "11 Oct 2000 02:03:02 +0100",
"msg_from": "Gunnar R|nning <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> As promised, the press release is at:\n> \n> http://www.greatbridge.com/news/p_101020001.html\n\nWow. Nice. The combined qualifications of the six steering committee\nmembers are staggering.\n\nMe, I'm just a lowly broadcast engineer with ten years experience and a\nmeasly Bachelor's degree in Electronics Engineering Technology. Oh\nwell. Even I have a job to do!\n\n--\nLamar Owen\nWGCR Intermet Radio\n1 Peter 4:11\n",
"msg_date": "Tue, 10 Oct 2000 21:05:34 -0400",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "On Tue, Oct 10, 2000 at 05:06:37PM -0400, Dave Smith wrote:\n> Adam Lang wrote:\n\n> Well to calm any fears of Great Bridge taking over what exactly are the\n> terms of employment? Are the developers merely continuing on with what\n> they were working on and now getting paid for it, or is Great Bridge\n> saying here are the projects we want done so do it. \n\n\"merely?\"\n\n\tI've told many people that postgres is one of the best-managed\n(open-source or otherwise) projects I've seen. The core group knows the\ncode, the deadlines, the bugs, and the solutions.\n\n\tMy feeling:\n\tThe source is open and you are free to do whatever you want with it.\nIf Great Bridge decides that they want to make postgresql into the best\ndamn pinball simulator they can, that is their perogative. If the\ncore developers decide they want to get paid to write a pinball\nsimulator, theat that is their gig and there isn't a damn thing we\ncan do about it, except branch off and decide not to integrate their\npatches. I would feel very sorry if this happened--and Bruce, Tom,\nand the other guys who I can't remember names of all understand this.\n\n\tAll this whinging about \"corperate direction\" is really meaningless\nunless you are prepared to jump ship or split off in a clone of the\noriginal one. The code is what you do with it. We are all lucky that\nit is as good and useful as it is right now.\n\n\n-- \nAdam Haberlach | A billion hours ago, human life appeared on\[email protected] | earth. A billion minutes ago, Christianity\nhttp://www.newsnipple.com | emerged. A billion Coca-Colas ago was\n'88 EX500 | yesterday morning. -1996 Coca-Cola Ann. Rpt.\n",
"msg_date": "Tue, 10 Oct 2000 19:06:47 -0700",
"msg_from": "Adam Haberlach <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "At 17:25 10/10/00 -0400, Tom Lane wrote:\n>\n>So the question is, what do we do now?\n>\n\nThere seem to be several concerns (in no particular order):\n\n1. Conscious design/development choices based partly/solely on the needs of\none or more companies as opposed to the interest of the open source project. \n\neg. if changes to the core of pgsql to support erServer are actually\ndetrimental to the maintainability and reliability of the open source project.\n\nThis can not be completely avoided, but the existing core team review\nsystem will presumably help. Expanding the core to include non-company\npeople is a good idea.\n\n\n2. Subconscious design/development choices based on the interests of one or\nmore companies. \n\nCan't really avoid this, but one hopes such subconscious decisions will be\nfar less significant than the conscious ones. Again, expanding the core to\ninclude non-company people is a good idea.\n\n\n3. Loss of core members to wholly private development.\n\nCan't avoid this. Always was and will be a risk.\n\n\nIn answer to \"What do we do now\", it seems a first step would be to ensure\ntransparency in decision making (something that I think Peter E mentioned).\nThe fact we have two companies, who in theory will compete, is a good thing\n(let's hope there are no strategic alliances announced in the near future).\n\nWhat about setting up some kind of committee consisting of an expanded core\nas well as some rotated members (possibly) selected randomly from the\nnon-core developers or users?\n\nIn reality, any suggestions of 'what to do' has to come from the core. It\nhas to be something you are happy to follow and which is not painful, but\nwhich also satisfies the concerns already raised. \n\nThe first attempt at self-regulation failed, probably because the sights\nwere set unreasonably high. What is needed now is an agreed and reasonable\nset of guidelines or principles.\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Wed, 11 Oct 2000 12:44:31 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "Gunnar R|nning <[email protected]> writes:\n> Bruce Momjian <[email protected]> writes:\n>> more in voliation of the original plan.\n\n> violation ? Or is this just another gap in my knowledge of the English\n> language ? \n\nYou're right, he's wrong. We native English speakers are notoriously\npoor spellers of our own language ;-)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 10 Oct 2000 23:08:01 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "> Gunnar R|nning <[email protected]> writes:\n> > Bruce Momjian <[email protected]> writes:\n> >> more in voliation of the original plan.\n> \n> > violation ? Or is this just another gap in my knowledge of the English\n> > language ? \n> \n> You're right, he's wrong. We native English speakers are notoriously\n> poor spellers of our own language ;-)\n\nOh, sorry. I didn't even see the spelling error. Yes, violation.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 23:20:42 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "> \n> Bottom line is we're not sure what to do now. Opinions from the \n> floor, anyone?\n> \n\n From the lowly end of the floor... for what I am concerned, I'm not \nworried about the involvment of the core team. Instead, I'm happy \nthat companies like GB and Postgres Inc have been founded.\n\nI'm not an active member of open source community (if not for \nadvocating it), but just for the lack of skill. I know RS's and ESR's \nworks, I think I got the ideas, but I think that \"commercial support\" \nis useful for the quality of the projects, and not detrimental.\n\nI really think that there is no possibility that a commercial company \nbased on a open source project could steer away from the good of \nthe project. The equation is simple: the more the \"product\" is good, \nthe more the company would penetrate the market.\n\nWe all know that marketing and FUD approaches are incompatable \nwith open source projects, just the quality of the product can give \npeople the reason to adopt it. \n\nWhat should we fear? That GB will purpusedly put some limitations \nor bugs in tha code, so they could gain more on supporting it\n(ya 'now, somebody says that some guy have earned billions \nfollowing this strategy ;-))?\nBut this is simply not feasible. They don't sell the product, so they \ncould not gain on \"new realeses\" and \"service packs\". And who \ncould hide bugs in an open source project and call them \"features\"?\n\nAt the most, as Tom said, they will be more focused to hunt bugs \nand add features basing on requests made by paying customers. \nWell, those are nonetheless bugs that will be corrected and new \nfeatures that will be added, and we all will benefit for them. There's \ngood chance that they are the same bugs and same features that \nsome of OUR customers (I'm meaning \"we\" as in \"independent \nconsultants and developers that use open source projects as \ntools\") will ask for. \nAnd this way the people that are working on that will be also well \npayed (er, I don't know the payrolls, I'm just hoping that they are \ngood...), and I can't see anything bad in that!\n\nNo, as I said, commercial companies investing in open source \ndevelopment can only do good. \n\njust my 0.02 Euro ;-)\n\nand good luck to all core members for their new jobs!\n\n\n/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\\/\n\nFabrizio Ermini Alternate E-mail:\nC.so Umberto, 7 [email protected]\nloc. Meleto Valdarno Mail on GSM: (keep it short!)\n52020 Cavriglia (AR) [email protected]\n",
"msg_date": "Wed, 11 Oct 2000 11:02:24 +0200",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "Since it seems that core is rather mature and \"loves\" postgresql, not too\nmuch concern is probably neccessary. Maybe \"modify\" the rules and say no\nmore than 50% can be in one company. Also, maybe inform GB and others of\nthe policy so they don't actively pursue core to make it an issue.\n\nOther than that, just wait and see I guess.\n\nAlso... if you are concerned about the community seeing core as \"founded\" or\nnot, maybe on one of the postgresl sites, disclose where core works if it is\na \"conflict\" issue... but that is a bit of an privacy infringement I'd\nthink.\n\nEither way... it seems like core has the backing of this list at least... if\nthe backing is universal... I say it would be upto core to decide what they\nfeel is safe.\n\nAdam Lang\nSystems Engineer\nRutgers Casualty Insurance Company\n----- Original Message -----\nFrom: \"Tom Lane\" <[email protected]>\nTo: \"Adam Lang\" <[email protected]>\nCc: \"PostgreSQL-general\" <[email protected]>\nSent: Tuesday, October 10, 2000 5:25 PM\nSubject: Re: [GENERAL] Re: [HACKERS] My new job\n\n\n> \"Adam Lang\" <[email protected]> writes:\n> > I wasn't judging. I was mentioning to others what the concerns probably\n> > were. Also, it isn't a concern of \"Company B\" taking over. It is of\nthe\n> > possibility of development put in the direction that best benefits of\n> > Company B as opposed to the project itself.\n> > ...\n> > It is merely a conflict of interest issue.\n>\n> Right, exactly. That was why we originally suggested putting a limit on\n> the number of core members employed by any one company: to reduce both\n> the actual and perceived potential for core decisions being taken in a\n> way that is more for the benefit of some company than for the project as\n> a whole.\n>\n> I am not sure that the *real* potential for bad choices is all that\n> high. I think all the core members understand very well that we are\n> stewards of a shared resource, and in the long run decisions counter\n> to the community-wide best interest will also not be in the best\n> interest of our companies. But it's also important that the rest of\n> the PG community *perceive* that core decisions are well-founded.\n>\n> However, given recent events the original two-of-six idea isn't feasible\n> any more --- and certainly none of us were going to tell Bruce that he\n> couldn't take that job because that'd make three GB employees on core.\n> So the question is, what do we do now?\n>\n> regards, tom lane\n\n",
"msg_date": "Wed, 11 Oct 2000 09:10:29 -0400",
"msg_from": "\"Adam Lang\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "Probably the best solution is to make the solution the least complicated and\nimposing. :)\n\nAdam Lang\nSystems Engineer\nRutgers Casualty Insurance Company\n----- Original Message -----\n> At 17:25 10/10/00 -0400, Tom Lane wrote:\n> >\n> >So the question is, what do we do now?\n> >\n\n\n",
"msg_date": "Wed, 11 Oct 2000 09:37:52 -0400",
"msg_from": "\"Adam Lang\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "On Tue, Oct 10, 2000 at 01:02:52PM -0400, Tom Lane wrote:\n> \n> Bottom line is we're not sure what to do now. Opinions from the \n> floor, anyone?\n> \n\nFirst, I include my voice with those congratulating Bruce, and wishing\nhim the best of luck in his new position.\n\nI concur that no one who has paid any attention at all to how the core\ndevelopers interact with the user base on the mailing lists can have any\nthing but the highest regard for all of your integrity and devotion to\nthe project. Given that, there is little fear of overt actions by Great\nBridge that could harm the project, short of firing you all and leaving\nyou destitute on the street (which wouldn't last long, I'm sure ;-)\n\nAs I mentioned in thread that followed Ned Lilly's first post here, the\nreal threat to code quality will be management pressures, such as schedule\npressure: when Management wants a new release so that Marketing can better\nsell against a competitor with a higher release number, what do you do?\n\nThere are a dozen scenarios I could spin, each more fantastic than the\nlast, but all grounded in someones real world experience. All the core\nmembers have more experience than I in the world of corporate coding,\nso can probably spin worse ones. How each of you handles those kind\nof pressures is up to your own internal compass: just let me remind\nyou that, unlike most situtations where the individual is alone, the\nuser community here can be a _personal_ resource, standing up for you,\nand providing an outside voice, if needed.\n\nEnough with worst case scenarios. As you said, Tom, the real problem is\nabout the preception of the community. How to avoid misunderstandings?\nI think Peter's point about transparency of development _process_ is\ncrucial. As it is, there have been in the past occasional back channel\ncommunications where design decisions get made, via IRC or phone calls.\nThis in and of itself is not a problem: some problems are just easier to\nthrash out that way. The problem comes when the decision is presented as\na fait accompli, without a clear public statement of the reasoning behind\nthe decision. Sometimes it's easy to forget if a particular point got\nmade on te phone, in a public email list, or a private, core list. This\ncould easily spin out of control, if decisions get made over the water\ncooler, as it were.\n\nTo date, the core developers have served as steering committee, as\nwell. This is only natural on an all volunteer project: in that case,\nno one can order anyone to do anything they don't want to, so only the\ndevelopers can direct the project. The Debian project runs into this\nall the time: Herding kittens, it's called. ;-)\n\nNow, the problem is that it is perceived that some one _can_ order\nthe developers: analogous to the criticisms of electing John Kennedy\nas U.S. President, since he was Catholic, and therefore preceived to be\nunder the Pope's control. That's, of course, extreme, but Tom himself has\nsaid that'd he'd work on bug fixes for paying customers over mailing list\nsubmissions. That's his right, and no different than a volunteer developer\ndeciding that work or school assignments take precedence. It's happened\nto me, enough. But it's the perception that matter here, not the fact.\n\nWhat to do? Make as much communication as possible public. When in doubt,\nerr on the public side. Develop in the fish bowl. If you feel there still\na need for private channels, perhaps include some outside representitive,\ntrusted by the community, who can serve as a witness of record, if you\nwill, vouching for the intent of the communications, without having to\nreveal the content.\n\nWell, there's my nickel. Do with it what you will.\n\nRoss\n\nRoss J. Reedstrom, <[email protected]> NSBRI Research Scientist/Programmer\nComputer and Information Technology Institute Rice University, 6100\nS. Main St., Houston, TX 77005\n\n-- \nOpen source code is like a natural resource, it's the result of providing\nfood and sunshine to programmers, and then staying out of their way.\n[...] [It] is not going away because it has utility for both the developers \nand users independent of economic motivations. Jim Flynn, Sunnyvale, Calif.\n",
"msg_date": "Fri, 13 Oct 2000 12:00:28 -0500",
"msg_from": "\"Ross J. Reedstrom\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] My new job"
},
{
"msg_contents": "> What to do? Make as much communication as possible public. When in doubt,\n> err on the public side. Develop in the fish bowl. If you feel there still\n> a need for private channels, perhaps include some outside representative,\n> trusted by the community, who can serve as a witness of record, if you\n> will, vouching for the intent of the communications, without having to\n> reveal the content.\n> \n\nWe were talking about this today down here at Great Bridge, and I\nmentioned that there is very little that happens in the core group. Up\nuntil Great Bridge arrived, and we had to secretly communicate with\nthem, there really wasn't much going on in core. Occasionally they will\nget upset with me about accepting too many patches, but other than that,\nmonths go by with nothing happening in core at all.\n\nSo, I am really saying that core doesn't do much. You non-core folks\naren't missing anything.\n\n--\n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 13 Oct 2000 16:10:48 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "On Fri, Oct 13, 2000 at 04:10:48PM -0400, Bruce Momjian wrote:\n>\n> We were talking about this today down here at Great Bridge, and I\n> mentioned that there is very little that happens in the core group. Up\n> until Great Bridge arrived, and we had to secretly communicate with\n> them, there really wasn't much going on in core. Occasionally they will\n> get upset with me about accepting too many patches, but other than that,\n> months go by with nothing happening in core at all.\n>\n> So, I am really saying that core doesn't do much. You non-core folks\n> aren't missing anything.\n\nYeah, that's what you say in public ... There is no cabal!\n\nSeriously though, we're talking about perceptions. The once or twice over\nthe last two years I've noticed sign of out-of-band decision making, it's\nusually been a mention of IRC or a phone call. No great shakes, just if\nsomeone's already (irrationally) upset about their 'great' design idea not\ngetting in, and it's not clear why, but something happened in person (or\nvia IRC or phone or core) ... I'd rather _not_ see the slashdot thread.\n\nRoss\n-- \nOpen source code is like a natural resource, it's the result of providing\nfood and sunshine to programmers, and then staying out of their way.\n[...] [It] is not going away because it has utility for both the developers \nand users independent of economic motivations. Jim Flynn, Sunnyvale, Calif.\n",
"msg_date": "Fri, 13 Oct 2000 15:59:14 -0500",
"msg_from": "\"Ross J. Reedstrom\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "> > So, I am really saying that core doesn't do much. You non-core folks\n> > aren't missing anything.\n> \n> Yeah, that's what you say in public ... There is no cabal!\n> \n> Seriously though, we're talking about perceptions. The once or twice over\n> the last two years I've noticed sign of out-of-band decision making, it's\n> usually been a mention of IRC or a phone call. No great shakes, just if\n> someone's already (irrationally) upset about their 'great' design idea not\n> getting in, and it's not clear why, but something happened in person (or\n> via IRC or phone or core) ... I'd rather _not_ see the slashdot thread.\n\nYou should know that phone calls are my secret weapon.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 13 Oct 2000 22:10:36 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "> At 16:10 13/10/00 -0400, Bruce Momjian wrote:\n> >\n> >So, I am really saying that core doesn't do much. You non-core folks\n> >aren't missing anything.\n> >\n> \n> The issue here is transparency: doing the right thing as well as being seen\n> to do the right thing.\n> \n> When someone devotes hours of time to PGSQL for no recompense, their\n> motives are generally not questioned. So when someone makes a design\n> decision, the motive is assumed to be because it is best in the long term\n> for the project. As soon as someone is paid to do work, their motive is (at\n> least partly) to get paid. As Tom has already said, this has the potential\n> to distort scheduling priorities.\n\nThe majority of core discussions are closed because either we need to\ndecide on a central direction for the project (release date) or we need\nto discuss something that would embarrass someone if it were publically\nknown.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 13 Oct 2000 22:12:40 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "> At 22:12 13/10/00 -0400, Bruce Momjian wrote:\n> >\n> >The majority of core discussions are closed because either we need to\n> >decide on a central direction for the project (release date) \n> \n> These are the things that you should consider making more transparent.\n> \n> >or we need\n> >to discuss something that would embarrass someone if it were publically\n> >known.\n> \n> Personal opinions are of course private. Can you think of an example of a\n> secret embarrassing item that has affected the direction of the project?\n> I'd be fascinated!\n\nYou know, we have to take people aside once and a while and get them\nback on course. Of course, we do that for core members too.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 13 Oct 2000 22:22:09 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "On Sat, 14 Oct 2000, Philip Warner wrote:\n\n> The fear is that this may distort other priorities - hence why\n> increased transparency in decision making is important. If Bruce, Tom\n> & Jan make a design decision, then chances are it's going to be pretty\n> good. The problem is it will/may be seen as a GB decision.\n\nI don't know ... the recent discussion in -hackers on the whole ALTER\nTABLE DROP COLUMN tends to show that even those \"in the pay of\" don't\nnecessarily agree :) \n\n\n",
"msg_date": "Fri, 13 Oct 2000 23:31:46 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "\"Ross J. Reedstrom\" <[email protected]> writes:\n>> So, I am really saying that core doesn't do much. You non-core folks\n>> aren't missing anything.\n\n> Yeah, that's what you say in public ... There is no cabal!\n\nIt's true that very little goes on on the private core mailing list,\nand we try to keep it that way. I think that most of the power that\ncore has (such as it is) is that people on pghackers are willing to\ndefer to us on decisions like what the release schedule should be.\nThere are a dozen or more non-core people with CVS write access,\nso it's not like core is tightly controlling what happens to the code.\n\nI think ideally our role is one of cat herders, as you put it ---\nmaking the kinds of decisions that a group of dozens or hundreds\ncan't make effectively. But the long-term direction of the project\nis largely determined by what the individual CVS committers choose to\nwork on. In that sense, a core member has no more power than any\nnon-core committer. (Case in point: Peter E. has had more influence\non what 7.1 will look like than most of core ;-).)\n\nWhen you look at it from that point of view, power comes from having\ntime to work on the code. In that sense, now that Great Bridge is\npaying me to work full-time on Postgres, I personally may be the\nmost dangerous loose cannon on the deck. (Jan is less dangerous\nright at the moment only because he's distracted by moving concerns.\nOnce he's settled again in Norfolk, look out...) Outer joins will be\nin 7.1 because *I* decided that would be a good thing to work on ---\nthis wasn't a core decision, nor one imposed on me by Great Bridge.\nI doubt anyone will complain too hard about that particular choice,\nbut further down the road I might make more debatable choices about\nhow to spend my time.\n\nI agree 100% with your comments that openness of decision-making\nis a critical element in keeping the trust of the community. But\nlooking at it as just an issue of core vs non-core is missing some part\nof the problem. Everyone who contributes code has a responsibility,\nproportionate to how much work they're doing, to ensure that the rest\nof the community understands and approves of what they're doing.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 13 Oct 2000 22:34:58 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "> > At 22:12 13/10/00 -0400, Bruce Momjian wrote:\n> > >\n> > >The majority of core discussions are closed because either we need to\n> > >decide on a central direction for the project (release date) \n> > \n> > These are the things that you should consider making more transparent.\n> > \n> > >or we need\n> > >to discuss something that would embarrass someone if it were publically\n> > >known.\n> > \n> > Personal opinions are of course private. Can you think of an example of a\n> > secret embarrassing item that has affected the direction of the project?\n> > I'd be fascinated!\n> \n> You know, we have to take people aside once and a while and get them\n> back on course. Of course, we do that for core members too.\n\nIn fact, we built a shed outside especially for Jan, who is in Poland\ngiving a speech and can't possibly respond in a timely manner. :-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 13 Oct 2000 22:53:28 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "> On Sat, 14 Oct 2000, Philip Warner wrote:\n> \n> > The fear is that this may distort other priorities - hence why\n> > increased transparency in decision making is important. If Bruce, Tom\n> > & Jan make a design decision, then chances are it's going to be pretty\n> > good. The problem is it will/may be seen as a GB decision.\n> \n> I don't know ... the recent discussion in -hackers on the whole ALTER\n> TABLE DROP COLUMN tends to show that even those \"in the pay of\" don't\n> necessarily agree :) \n\nMan, Tom, our cover is working perfectly. Let's disagree on something\nagain. That will really convince them. :-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 13 Oct 2000 22:54:31 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "At 16:10 13/10/00 -0400, Bruce Momjian wrote:\n>\n>So, I am really saying that core doesn't do much. You non-core folks\n>aren't missing anything.\n>\n\nThe issue here is transparency: doing the right thing as well as being seen\nto do the right thing.\n\nWhen someone devotes hours of time to PGSQL for no recompense, their\nmotives are generally not questioned. So when someone makes a design\ndecision, the motive is assumed to be because it is best in the long term\nfor the project. As soon as someone is paid to do work, their motive is (at\nleast partly) to get paid. As Tom has already said, this has the potential\nto distort scheduling priorities.\n\nThe fear is that this may distort other priorities - hence why increased\ntransparency in decision making is important. If Bruce, Tom & Jan make a\ndesign decision, then chances are it's going to be pretty good. The problem\nis it will/may be seen as a GB decision.\n\nAs I said in an earlier post, any and all drive for increased transparency\nhas to come from GB; they must now be aware of the issues and potential\nrisks. Certainly planning to contribute *all* source is a very good sign\nsince it will drastically reduce the chances of making compromising design\ndecisions.\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Sat, 14 Oct 2000 12:55:33 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "> I think ideally our role is one of cat herders, as you put it ---\n> making the kinds of decisions that a group of dozens or hundreds\n> can't make effectively. But the long-term direction of the project\n> is largely determined by what the individual CVS committers choose to\n> work on. In that sense, a core member has no more power than any\n> non-core committer. (Case in point: Peter E. has had more influence\n> on what 7.1 will look like than most of core ;-).)\n\nJan says that if I start coding more, GB will have to hire more\ndevelopers to clean up after me. Now, is that supposed to make me feel\nvalued? :-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 13 Oct 2000 22:56:25 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "At 22:12 13/10/00 -0400, Bruce Momjian wrote:\n>\n>The majority of core discussions are closed because either we need to\n>decide on a central direction for the project (release date) \n\nThese are the things that you should consider making more transparent.\n\n>or we need\n>to discuss something that would embarrass someone if it were publically\n>known.\n\nPersonal opinions are of course private. Can you think of an example of a\nsecret embarrassing item that has affected the direction of the project?\nI'd be fascinated!\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Sat, 14 Oct 2000 13:19:56 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "Philip Warner <[email protected]> writes:\n>> or we need to discuss something that would embarrass someone if it\n>> were publically known.\n\n> Personal opinions are of course private. Can you think of an example of a\n> secret embarrassing item that has affected the direction of the project?\n> I'd be fascinated!\n\nThere have been a couple of cases where core has decided that a committer\nneeded to be admonished (\"yo, mon, why you committing new features\nduring beta freeze?\" or something like that). Marc has generally done\nthe admonishing with a cc to core, but we don't embarrass people in\npublic. I don't propose to name names here for obvious reasons.\n\nAnother class of properly-private discussions have been reports of\nsecurity-related bugs; that sort of thing seems best not mentioned\ntoo widely on the public lists until a fix is available. (BTW, if\nyou ever have a security bug report that you don't think ought to be\nmentioned in the public archives, send it to pgsql-core.)\n\nDunno about \"affecting the course of the project\". I don't think that\ncore as core has all that much influence on the course of the project.\nTiming (release schedule) yes, because people allow us to decree that,\nbut direction no.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 13 Oct 2000 23:31:22 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "> > I think ideally our role is one of cat herders, as you put it ---\n> > making the kinds of decisions that a group of dozens or hundreds\n> > can't make effectively. But the long-term direction of the project\n> > is largely determined by what the individual CVS committers choose to\n> > work on. In that sense, a core member has no more power than any\n> > non-core committer. (Case in point: Peter E. has had more influence\n> > on what 7.1 will look like than most of core ;-).)\n> \n> Jan says that if I start coding more, GB will have to hire more\n> developers to clean up after me. Now, is that supposed to make me feel\n> valued? :-)\n\nOK, why is no one laughing at my crafty jokes? Did core tell you guys\nnot to laugh? Is that what they are doing? :-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 13 Oct 2000 23:38:41 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "Philip Warner <[email protected]> writes:\n> When someone devotes hours of time to PGSQL for no recompense, their\n> motives are generally not questioned. So when someone makes a design\n> decision, the motive is assumed to be because it is best in the long term\n> for the project. As soon as someone is paid to do work, their motive is (at\n> least partly) to get paid. As Tom has already said, this has the potential\n> to distort scheduling priorities.\n\nA side comment here: generally committers' motives are not questioned,\nbut what makes you think they're doing it for no recompense? I know\nthat when I first started getting involved with PGSQL, the first fixes/\nchanges I sent in were directly related to problems my then company\nwas having. Since most uses for databases seem to be business-related,\nI suspect that most people who are involved with PGSQL have at least\nsome connection to a business need.\n\nThe real issue is how much control does any one entity exert, and if\nit's a lot, is that entity driving things in a direction that other\npeople don't like?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 13 Oct 2000 23:41:11 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job "
},
{
"msg_contents": "On Fri, Oct 13, 2000 at 10:54:31PM -0400, Bruce Momjian wrote:\n> \n> Man, Tom, our cover is working perfectly. Let's disagree on something\n> again. That will really convince them. :-)\n> \n\nThis was supposed to go to -core, right? (I see Bruce got my joke. I\nwasn't so sure Tom did ...)\n\nRoss\n-- \nOpen source code is like a natural resource, it's the result of providing\nfood and sunshine to programmers, and then staying out of their way.\n[...] [It] is not going away because it has utility for both the developers \nand users independent of economic motivations. Jim Flynn, Sunnyvale, Calif.\n",
"msg_date": "Fri, 13 Oct 2000 22:58:51 -0500",
"msg_from": "\"Ross J. Reedstrom\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "> On Fri, Oct 13, 2000 at 10:54:31PM -0400, Bruce Momjian wrote:\n> > \n> > Man, Tom, our cover is working perfectly. Let's disagree on something\n> > again. That will really convince them. :-)\n> > \n> \n> This was supposed to go to -core, right? (I see Bruce got my joke. I\n> wasn't so sure Tom did ...)\n\nTom chuckles, even if he doesn't get the joke.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Fri, 13 Oct 2000 23:59:59 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
},
{
"msg_contents": "Peter Eisentraut wrote:\n> Bruce Momjian writes:\n>\n> > After careful consideration, I have decided to accept a job with Great\n> > Bridge.\n>\n> Whatever happened to this:\n>\n> Date: Tue, 09 May 2000 15:19:48 -0400\n> From: Tom Lane <[email protected]>\n> To: Ross J. Reedstrom <[email protected]>\n> Cc: PostgreSQL-general <[email protected]>\n> Subject: Re: [GENERAL] Steering committee responce to Great Bridge LLC\n>\n> : One thing we have agreed to is that there must not be an unseemly fraction\n> : of core members working for the same company. With six people on core,\n> : probably about two working at the same company would be a reasonable\n> : limit.\n\n Been in Poland for a week, so pardon for the delay.\n\n Initially it was (if I recall correct) Vadim's and my idea.\n The main reason behind it wasn't to avoid influence from\n commercial entities into core. We've all been working\n together for years as a group with great honour and trusting,\n so the aims of all core members where never questioned. We\n just decided this \"fraction\" to avoid any hireing to look\n like a takeover.\n\n This world is spinning a little fast at the moment. Let me\n repeat what I said to a person I met last week in Poland,\n during a PosrgreSQL conference in Wierzba. We have a BSD\n license and now I know a good reason why we kept it all the\n time. With that license in place, there's absolutely no\n reason to panic right now. Nothing can be taken away, and if\n things go wrong in the future, those left in the \"OpenSource\"\n corner can start from our last official release again - be\n sure I'll be somewhere in that corner, even if it might take\n some time before I can surface again. And I'm sure I'll meet\n all those I loved to work with together in that corner again.\n\n Never underestimate the power of Open Source.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#================================================== [email protected] #\n\n\n",
"msg_date": "Sun, 22 Oct 2000 12:41:36 -0500 (EST)",
"msg_from": "Jan Wieck <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re: [HACKERS] My new job"
}
] |
[
{
"msg_contents": "*Sigh*, still not having much fun over here:\n\n# gdb /usr/local/pgsql/bin/postgres postgres.36561.core \n#0 0x8063b0f in nocachegetattr (tuple=0xbfbfe928, attnum=2, \n tupleDesc=0x84ca368, isnull=0xbfbfe7af \"\") at heaptuple.c:494\n494 off = att_addlength(off, att[j]->attlen, tp + off);\n(gdb) list\n489 */\n490 off = att_align(off, att[j]->attlen, att[j]->attalign);\n491 \n492 att[j]->attcacheoff = off;\n493 \n494 off = att_addlength(off, att[j]->attlen, tp + off);\n495 }\n496 \n497 return (Datum) fetchatt(&(att[attnum]), tp + att[attnum]->attcacheoff);\n498 }\n(gdb) bt\n#0 0x8063b0f in nocachegetattr (tuple=0xbfbfe928, attnum=2, \n tupleDesc=0x84ca368, isnull=0xbfbfe7af \"\") at heaptuple.c:494\n#1 0x8075851 in GetIndexValue (tuple=0xbfbfe928, hTupDesc=0x84ca368, \n attOff=2, attrNums=0x84e9768, fInfo=0x0, attNull=0xbfbfe7af \"\")\n at indexam.c:445\n#2 0x80903be in FormIndexDatum (numberOfAttributes=4, \n attributeNumber=0x84e9768, heapTuple=0xbfbfe928, heapDescriptor=0x84ca368, \n datum=0x84e9518, nullv=0x84ba170 \" \", fInfo=0x0) at index.c:1256\n#3 0x80a05e6 in vc_repair_frag (vacrelstats=0x84ba290, onerel=0x84c6788, \n vacuum_pages=0xbfbfe9d0, fraged_pages=0xbfbfe9c0, nindices=1, \n Irel=0x84ba118) at vacuum.c:1634\n#4 0x809e3b9 in vc_vacone (relid=1315147913, analyze=0, va_cols=0x0)\n at vacuum.c:640\n#5 0x809d9ac in vc_vacuum (VacRelP=0xbfbfea60, analyze=0 '\\000', va_cols=0x0)\n at vacuum.c:299\n#6 0x809d934 in vacuum (vacrel=0x84ba0e8 \"\\030\", verbose=1, analyze=0 '\\000', \n va_spec=0x0) at vacuum.c:223\n#7 0x810ca8c in ProcessUtility (parsetree=0x84ba110, dest=Remote)\n at utility.c:694\n#8 0x810a44e in pg_exec_query_dest (\n query_string=0x81cd370 \"VACUUM verbose webhit_details_formatted;\", \n dest=Remote, aclOverride=0) at postgres.c:617\n#9 0x810a3a9 in pg_exec_query (\n query_string=0x81cd370 \"VACUUM verbose webhit_details_formatted;\")\n at postgres.c:562\n#10 0x810b336 in PostgresMain (argc=9, argv=0xbfbff0e0, real_argc=10, \n real_argv=0xbfbffb40) at postgres.c:1588\n#11 0x80f0742 in DoBackend (port=0x8464000) at postmaster.c:2009\n#12 0x80f02d5 in BackendStartup (port=0x8464000) at postmaster.c:1776\n#13 0x80ef4f9 in ServerLoop () at postmaster.c:1037\n#14 0x80eeede in PostmasterMain (argc=10, argv=0xbfbffb40) at postmaster.c:725\n#15 0x80bf3eb in main (argc=10, argv=0xbfbffb40) at main.c:93\n#16 0x8063495 in _start ()\n(gdb) \n\nIsn't there something someone can offer to help track down why this\nis happening? It only seems to happen when we enable a particular\ndatabase script, and it always seems to happen while a pg_dump is\ngoing on in the background while performing other updates, or a\nvacuum.\n\nAll my crashes seem to happen in this file in 'nocachegetattr()',\nis there anything I can do to provide more comprehensive error\nreporting or something?\n\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Mon, 9 Oct 2000 22:42:56 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": true,
"msg_subject": "more crashes with 7.0.2 from oct 4th-PATCHES"
}
] |
[
{
"msg_contents": "Hello,\n\nI have quite strange behavior of the following SQL:\n\ninsert into address (cid,email) select distinct '49'::int,member.email from \nmember imit 1 ;\n\nIt should insert just 1 record.\nBut it insert all recodrs which will be selected by subselect...\nWhat's wrong with this SQL? Or this is a bug? If it is a bug...\nHow to fix it (patch, workaround...)\n\nThanks in advance.\n\n-- \nSincerely Yours,\nDenis Perchine\n\n----------------------------------\nE-Mail: [email protected]\nHomePage: http://www.perchine.com/dyp/\nFidoNet: 2:5000/120.5\n----------------------------------\n",
"msg_date": "Tue, 10 Oct 2000 14:02:22 +0700",
"msg_from": "Denis Perchine <[email protected]>",
"msg_from_op": true,
"msg_subject": "Problem specifying limit in select inside insert."
},
{
"msg_contents": "Denis Perchine <[email protected]> writes:\n> insert into address (cid,email) select distinct '49'::int,member.email from \n> member imit 1 ;\n\nINSERT ... SELECT ... LIMIT doesn't work. In existing releases the\nLIMIT clause is just dropped on the floor by the parser :-(. In current\nCVS sources you get\n\tERROR: LIMIT is not supported in subselects\nwhich at least lets you know you have a problem. I have a to-do item\nto see if this can be made to work right, but it's not very high\npriority. I think the correct fix is to pull LIMIT handling out of\nExecutorRun and make it be handled by a new plan node type, but that'll\nprobably take more time than I can spare before 7.1 ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 10 Oct 2000 12:21:34 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Problem specifying limit in select inside insert. "
}
] |
[
{
"msg_contents": "\n\nPhilip Warner wrote:\n\n> I get the following when I build from CVS:\n>\n> gcc -I../../../src/include -I../../../src/interfaces/libpq -O2 -g -Wall\n> -Wmissing-prototypes -Wmissing-declarations -o pg_dump pg_dump.o common.o\n> pg_backup_archiver.o pg_backup_db.o pg_backup_custom.o pg_backup_files.o\n> pg_backup_null.o pg_backup_tar.o -L../../..\n> /src/interfaces/libpq -lpq -lz -lcrypt -lnsl -ldl -lm -lreadline -lncurses\n> -lz\n> ../../../src/interfaces/libpq/libpq.so: undefined reference to\n> `FailedAssertion'\n> ../../../src/interfaces/libpq/libpq.so: undefined reference to\n> `assert_enabled'\n> ../../../src/interfaces/libpq/libpq.so: undefined reference to\n> `ExceptionalCondition'\n>\n> Does anyone else have this problem?\n>\n\nI have already encountered the problem.\n-DFRONTEND isn't set in the above command line.\n\nFor example the following line\n CFLAGS+= -DFRONTEND -I$(srcdir)\nin makefiles doesn't work currently.\n\nI found the recent change in Makefile.global.in.\nifeq ($(GCC), yes)\n override CFLAGS += -Wall -Wmissing-prototypes -Wmissing-declarations\nendif\n\nDo we have to override CFLAGS every time ?\n\nRegards.\n\nHiroshi Inoue\n\n",
"msg_date": "Tue, 10 Oct 2000 19:56:40 +0900",
"msg_from": "Hiroshi Inoue <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: CVS broken?"
},
{
"msg_contents": "\nI get the following when I build from CVS:\n\ngcc -I../../../src/include -I../../../src/interfaces/libpq -O2 -g -Wall\n-Wmissing-prototypes -Wmissing-declarations -o pg_dump pg_dump.o common.o\npg_backup_archiver.o pg_backup_db.o pg_backup_custom.o pg_backup_files.o\npg_backup_null.o pg_backup_tar.o -L../../..\n/src/interfaces/libpq -lpq -lz -lcrypt -lnsl -ldl -lm -lreadline -lncurses\n-lz\n../../../src/interfaces/libpq/libpq.so: undefined reference to\n`FailedAssertion'\n../../../src/interfaces/libpq/libpq.so: undefined reference to\n`assert_enabled'\n../../../src/interfaces/libpq/libpq.so: undefined reference to\n`ExceptionalCondition'\n\nDoes anyone else have this problem?\n\nIt's building under Linux i586 and there's some more context is below. The\nrest of the log looks fine.\n\n\nmake[3]: Entering directory\n`/home/pjw/work/postgresql-cvs/pgsql/src/bin/pg_dump'\nmake -C ../../../src/interfaces/libpq all\nmake[4]: Entering directory\n`/home/pjw/work/postgresql-cvs/pgsql/src/interfaces/libpq'\nmake[4]: Nothing to be done for `all'.\nmake[4]: Leaving directory\n`/home/pjw/work/postgresql-cvs/pgsql/src/interfaces/libpq'\ngcc -c -I../../../src/interfaces/libpq -I../../../src/include\n-I../../../src/interfaces/libpq -O2 -g -Wall -Wmissing-prototypes -W\nmissing-declarations pg_dump.c -o pg_dump.o\ngcc -c -I../../../src/interfaces/libpq -I../../../src/include\n-I../../../src/interfaces/libpq -O2 -g -Wall -Wmissing-prototypes -W\nmissing-declarations common.c -o common.o\ngcc -c -I../../../src/interfaces/libpq -I../../../src/include\n-I../../../src/interfaces/libpq -O2 -g -Wall -Wmissing-prototypes -W\nmissing-declarations pg_backup_archiver.c -o pg_backup_archiver.o\ngcc -c -I../../../src/interfaces/libpq -I../../../src/include\n-I../../../src/interfaces/libpq -O2 -g -Wall -Wmissing-prototypes -W\nmissing-declarations pg_backup_db.c -o pg_backup_db.o\ngcc -c -I../../../src/interfaces/libpq -I../../../src/include\n-I../../../src/interfaces/libpq -O2 -g -Wall -Wmissing-prototypes -W\nmissing-declarations pg_backup_custom.c -o pg_backup_custom.o\ngcc -c -I../../../src/interfaces/libpq -I../../../src/include\n-I../../../src/interfaces/libpq -O2 -g -Wall -Wmissing-prototypes -W\nmissing-declarations pg_backup_files.c -o pg_backup_files.o\ngcc -c -I../../../src/interfaces/libpq -I../../../src/include\n-I../../../src/interfaces/libpq -O2 -g -Wall -Wmissing-prototypes -W\nmissing-declarations pg_backup_null.c -o pg_backup_null.o\ngcc -c -I../../../src/interfaces/libpq -I../../../src/include\n-I../../../src/interfaces/libpq -O2 -g -Wall -Wmissing-prototypes -W\nmissing-declarations pg_backup_tar.c -o pg_backup_tar.o\ngcc -I../../../src/include -I../../../src/interfaces/libpq -O2 -g -Wall\n-Wmissing-prototypes -Wmissing-declarations -o pg_dump pg_dump.o common.o\npg_backup_archiver.o pg_backup_db.o pg_backup_custom.o pg_backup_files.o\npg_backup_null.o pg_backup_tar.o -L../../..\n/src/interfaces/libpq -lpq -lz -lcrypt -lnsl -ldl -lm -lreadline -lncurses\n-lz\n../../../src/interfaces/libpq/libpq.so: undefined reference to\n`FailedAssertion'\n../../../src/interfaces/libpq/libpq.so: undefined reference to\n`assert_enabled'\n../../../src/interfaces/libpq/libpq.so: undefined reference to\n`ExceptionalCondition'\ncollect2: ld returned 1 exit status\nmake[3]: *** [pg_dump] Error 1\nmake[3]: Leaving directory\n`/home/pjw/work/postgresql-cvs/pgsql/src/bin/pg_dump'\nmake[2]: *** [all] Error 2\nmake[2]: Leaving directory `/home/pjw/work/postgresql-cvs/pgsql/src/bin'\nmake[1]: *** [all] Error 2\nmake[1]: Leaving directory `/home/pjw/work/postgresql-cvs/pgsql/src'\nmake: *** [all] Error 2\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n",
"msg_date": "Tue, 10 Oct 2000 21:37:53 +1000",
"msg_from": "Philip Warner <[email protected]>",
"msg_from_op": false,
"msg_subject": "CVS broken?"
},
{
"msg_contents": "> > -Wmissing-prototypes -Wmissing-declarations -o pg_dump pg_dump.o common.o\n> > pg_backup_archiver.o pg_backup_db.o pg_backup_custom.o pg_backup_files.o\n> > pg_backup_null.o pg_backup_tar.o -L../../..\n> > /src/interfaces/libpq -lpq -lz -lcrypt -lnsl -ldl -lm -lreadline -lncurses\n> > -lz\n> > ../../../src/interfaces/libpq/libpq.so: undefined reference to\n> > `FailedAssertion'\n> > ../../../src/interfaces/libpq/libpq.so: undefined reference to\n> > `assert_enabled'\n> > ../../../src/interfaces/libpq/libpq.so: undefined reference to\n> > `ExceptionalCondition'\n> >\n> > Does anyone else have this problem?\n> >\n> \n> I have already encountered the problem.\n> -DFRONTEND isn't set in the above command line.\n> \n> For example the following line\n> CFLAGS+= -DFRONTEND -I$(srcdir)\n> in makefiles doesn't work currently.\n> \n> I found the recent change in Makefile.global.in.\n> ifeq ($(GCC), yes)\n> override CFLAGS += -Wall -Wmissing-prototypes -Wmissing-declarations\n> endif\n\nOK, I have removed 'override' from CFLAGS, and committed. Seems like it\ncame in with:\n\ndate: 2000/10/08 21:13:27; author: petere; state: Exp; lines: +165 -137\nAppend \"/postgresql\" to (certain) installation subdirectories when\ninstalling into a shared location. Also Makefile.global organizational\ncleanup.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Tue, 10 Oct 2000 09:06:18 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: CVS broken?"
},
{
"msg_contents": "Hiroshi Inoue writes:\n\n> For example the following line\n> CFLAGS+= -DFRONTEND -I$(srcdir)\n> in makefiles doesn't work currently.\n> \n> I found the recent change in Makefile.global.in.\n> ifeq ($(GCC), yes)\n> override CFLAGS += -Wall -Wmissing-prototypes -Wmissing-declarations\n> endif\n\nSeems like a problem in make. What versions are you using?\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Tue, 10 Oct 2000 18:49:16 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: CVS broken?"
},
{
"msg_contents": "\n\nPeter Eisentraut wrote:\n\n> Hiroshi Inoue writes:\n>\n> > For example the following line\n> > CFLAGS+= -DFRONTEND -I$(srcdir)\n> > in makefiles doesn't work currently.\n> >\n> > I found the recent change in Makefile.global.in.\n> > ifeq ($(GCC), yes)\n> > override CFLAGS += -Wall -Wmissing-prototypes -Wmissing-declarations\n> > endif\n>\n> Seems like a problem in make. What versions are you using?\n>\n\nI tried the following makefile in cygwin and linux 2.0.36.\n\n[makefile]\n#\noverride CFLAGS += -DDEFINE1\nCFLAGS += -DDEFINE2\n\nall: a.o\n\n[cygwin]\n$ make -v\nGNU Make version 3.79.1, by Richard Stallman and Roland McGrath.\nBuilt for i686-pc-cygwin\nCopyright (C) 1988, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 2000\n Free Software Foundation, Inc.\n...\n$ make\ngcc -DDEFINE1 -c -o a.o a.c\na.c:1: parse error before `{'\nmake: *** [a.o] Error 1\n\n[linux]\n$ make -v\nGNU Make version 3.76.1, by Richard Stallman and Roland McGrath.\nCopyright (C) 1988, 89, 90, 91, 92, 93, 94, 95, 96, 97\n Free Software Foundation, Inc.\n...\n$ make\ncc -DDEFINE1 -c a.c -o a.o\na.c:1: parse error before `{'\nmake: *** [a.o] Error 1\n\n-DDEFINE2 is expanded in neither case.\n\nRegards.\n\nHiroshi Inoue\n\n",
"msg_date": "Wed, 11 Oct 2000 08:35:57 +0900",
"msg_from": "Hiroshi Inoue <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: CVS broken?"
}
] |
[
{
"msg_contents": "As a long time (and mostly silent) PostgreSQL user, I\nwould like to say that I agree with Alfred. The more\ntime core members can spend working on PostgreSQL the\nbetter, and if you can finagle it so that you get paid\nactual money for doing it, then more power to you. \nYou certainly deserve it.\n\nWorking on an open source project is drastically\ndifferent then working on a closed source one. The\nsource is out there, and it can be forked if all of\nyou decide to sell your souls to the devil. Remember,\nyou yourselves didn't start the work on PostgreSQL,\nand you have put a lot of work into documenting what\nyou have done. A fork would probably easier for the\nnext batch of hackers than it was for you. Not only\nthat, but if your employer pushes you to do something\nto PostgreSQL that you don't agree with you can always\nwalk away and start your own PostgreSQL based company.\n If PostgreSQL was a closed-source project your\nemployer would own your work, but it isn't, and they\ndon't, so why worry?\n\nGreat Bridge and PostgreSQL Inc. can't buy your souls,\nand they can't steal your work. All they can do is\nfeed your families and help you sell PostgreSQL to the\nworld. Great Bridge and PostgreSQL Inc. aren't even\npaying for your code (they could get that for free),\nthey are paying for your expertise. Like the rest of\nyour users they realize that your talents are worth\ninvesting in.\n\nHowever, you might want to commit your changes to the\npublic CVS server a little more often :).\n\nThanks for the hard work,\nJason Earl\n\n--- Alfred Perlstein <[email protected]> wrote:\n> * Tom Lane <[email protected]> [001010 10:03] wrote:\n> \n> > > From: Tom Lane <[email protected]>\n> > > : One thing we have agreed to is that there must\n> not be an unseemly fraction\n> > > : of core members working for the same company. \n> With six people on core,\n> > > : probably about two working at the same company\n> would be a reasonable\n> > > : limit.\n> > \n> > I knew someone was going to bring that up ;-).\n> > \n> > There's already been discussion of this point\n> among core. What we\n> > now have is three core members employed by Great\n> Bridge and the\n> > other three either fully or partly employed by\n> PostgreSQL Inc.\n> > In one sense that's a stable situation, but on the\n> other hand it does\n> > not agree with our original informal goal of\n> keeping any one company\n> > to a minority position of the core membership.\n> > \n> > None of the core members are interested in giving\n> up their new\n> > positions. En masse resignation from the core\n> committee would preserve\n> > our high moral standards, perhaps, but it wouldn't\n> do the project any\n> > good that I can see. So it seems like the choices\n> are to accept the\n> > status quo, or to appoint some more core committee\n> members to bring\n> > the numbers back where we said they should be.\n> > \n> > While I can think of a number of well-qualified\n> candidates for core\n> > membership, I don't much like the notion of\n> appointing core members\n> > just to meet some kind of numerical quota. Also,\n> suppose we do appoint\n> > more members, and then some of them accept\n> positions with GB or PgSQL\n> > Inc; do we repeat the exercise indefinitely? \n> (This is not an unlikely\n> > scenario, since the sort of people who'd be asked\n> to join core are\n> > exactly the sort of people whom both companies\n> would love to hire.)\n> > \n> > Bottom line is we're not sure what to do now. \n> Opinions from the \n> > floor, anyone?\n> \n> I think anyone with doubts should take a good look\n> at the initial\n> companies backing Linux, (Redhat, VA, Debian) to see\n> what a boon\n> this can be to project.\n> \n> It is open source, so if you guys do happen to piss\n> us off too much\n> we can always fork off our own version no? :)\n> \n> So instead of panicing, it makes much more sense to\n> ride it out and\n> get a feel for where things are going, there's never\n> going to be\n> anything terribly binding that will come out of this\n> because it is\n> an opensource project.\n> \n> It's much more important to continue on with the\n> rapid pace of\n> developement than to fear black helicopters that\n> haven't even\n> shown up as blips on the radar.\n> \n> -- \n> -Alfred Perlstein -\n> [[email protected]|[email protected]]\n> \"I have the heart of a child; I keep it in a jar on\n> my desk.\"\n\n\n__________________________________________________\nDo You Yahoo!?\nGet Yahoo! Mail - Free email you can access from anywhere!\nhttp://mail.yahoo.com/\n",
"msg_date": "Tue, 10 Oct 2000 12:45:40 -0700 (PDT)",
"msg_from": "Jason Earl <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Re: [HACKERS] My new job"
}
] |
[
{
"msg_contents": "Actually with version 3.0 of PyGreSQL there is a\nDB-API v2.0 compliant wrapper pgdb.py.\n\n--- Peter Eisentraut <[email protected]> wrote:\n> Bruce Momjian writes:\n> \n> > I need to know how this is different than our\n> current python interface,\n> > PyGreSQL.\n> \n> > > PgSQL is a package of two (2) modules that\n> provide a Python DB-API 2.0\n> > > compliant interface to PostgreSQL databases.\n> \n> DB-API is the standard database interface for\n> Python. Kind of like\n> DBD/DBI for Perl.\n> \n> The existing one is hand-crafted, kind of like the\n> Pg Perl module.\n> \n> -- \n> Peter Eisentraut [email protected] \n> http://yi.org/peter-e/\n> \n\n\n__________________________________________________\nDo You Yahoo!?\nGet Yahoo! Mail - Free email you can access from anywhere!\nhttp://mail.yahoo.com/\n",
"msg_date": "Tue, 10 Oct 2000 13:03:35 -0700 (PDT)",
"msg_from": "Jason Earl <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [INTERFACES] Re: Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
}
] |
[
{
"msg_contents": "> > create table t1\n> > (\n> > f1 integer,\n> > f2 integer\n> > );\n> > \n> > create table t2\n> > (\n> > f1 integer references t1(f1),\n> > f2 integer\n> > );\n> \n> > begin transaction;\n> > insert into t1(f1,f2) values(1,1);\n> > delete from t1 where f1=1;\n> \n> > ERROR: triggered data change violation on relation \"t1\"\n> \n> You cannot change data twice within a transaction if there's a RI\n> constraint on the table. This is per SQL, nothing we can do about it.\n ^^^^^^^^^^^^^^^\nIs it true?! *Any reasons* for this? DELETE doesn't break integrity rules.\nJust tested it in Oracle - deletion is allowed!\nBut yes, I know that Oracle doesn't always follow standards -:)\nCan someone test it under Informix, others?\n\nVadim\n",
"msg_date": "Tue, 10 Oct 2000 13:07:13 -0700",
"msg_from": "\"Mikheev, Vadim\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: POSTGRES BUG - FIX IT PLEASE"
},
{
"msg_contents": "Mikheev, Vadim writes:\n\n> > You cannot change data twice within a transaction if there's a RI\n> > constraint on the table. This is per SQL, nothing we can do about it.\n> ^^^^^^^^^^^^^^^\n> Is it true?!\n\nActually, it's not. Jan Wieck first explained this on July 23 in \"Re:\n[GENERAL] failed Delete after Insert in a transaction\", and we've been\nbelieving it ever since, but I just found out that it's wrong.\n\nThe standard reads\n\n 11.8 <referential constraint definition>\n\n 9) If any attempt is made within an SQL-statement to update some\n site to a value that is distinct from the value to which that\n site was previously updated within the same SQL-statement,\n then an exception condition is raised: triggered data change\n violation.\n \n 10) If a site in an object row is an <object column> of an <update\n statement: positioned> or <update statement: searched>, and\n there is any attempt within the same SQL-statement to delete the\n row containing that site, then an exception condition is raised:\n triggered data change violation. \n\n(also 11.8 GR 8 b) i) 2), but it's too boring to quote...)\n\nNote that it talks about \"statements\", not \"transactions\".\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Tue, 10 Oct 2000 23:05:23 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] RE: POSTGRES BUG - FIX IT PLEASE"
}
] |
[
{
"msg_contents": "Maybe I'm just overlooking something really simple but this has me a bit\nconfused.\n\nWhat I'm trying to do is get the amount of time from A to B -- I thought\nage() would do just that but it seems to be about a day off sometimes.\n\nhhs=# SELECT age('Sun Dec 03 08:00:00 2000 EST','Tue Oct 10 08:00:00 2000\nEDT') as esec;\n esec\n------------------------\n @ 1 mon 24 days 1 hour\n(1 row)\n\nOk, but if I turn right around and add that value back , I get :\n\n\nhhs=# SELECT ('Tue Oct 10 08:00:00 2000 EDT'::timestamp + '1 mon 24 days 1\nhour'::interval);\n ?column?\n------------------------------\n Mon Dec 04 08:00:00 2000 EST\n(1 row)\n\nLike I said, perhaps I'm blind and can't see what's happening here but\nshouldn't that be Sunday the 3rd of December?\n\n\nI should point out that it works as I expected it to on other values..\n\nhhs=# SELECT age('Sun Nov 05 08:00:00 2000 EST','Tue Oct 10 08:00:00 2000\nEDT') as esec;\n esec\n------------------\n @ 26 days 1 hour\n(1 row)\n\nhhs=# SELECT ('Tue Oct 10 08:00:00 2000 EDT'::timestamp + '1 mon 24 days 1\nhour'::interval);\n ?column?\n------------------------------\n Mon Dec 04 08:00:00 2000 EST\n(1 row)\n\n*shrug*\n\nThanks!\n\n-Mitch\n\n",
"msg_date": "Tue, 10 Oct 2000 14:13:42 -0700",
"msg_from": "\"Mitch Vincent\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Possible age() bug? "
},
{
"msg_contents": "> I should point out that it works as I expected it to on other values..\n>\n> hhs=# SELECT age('Sun Nov 05 08:00:00 2000 EST','Tue Oct 10 08:00:00 2000\n> EDT') as esec;\n> esec\n> ------------------\n> @ 26 days 1 hour\n> (1 row)\n>\n> hhs=# SELECT ('Tue Oct 10 08:00:00 2000 EDT'::timestamp + '1 mon 24 days 1\n> hour'::interval);\n> ?column?\n> ------------------------------\n> Mon Dec 04 08:00:00 2000 EST\n> (1 row)\n\n\nI was obviously on drugs when I put this in the email, I apologize. It does\nindeed work as I expected it to on other values.\n\nIn an old 6.4 database I get what I expected ::\n\nhhs=> SELECT age('Sun Dec 03 08:00:00 2000 EST','Tue Oct 10 08:00:00 2000\nEDT') as esec;\nesec\n----------------------\n@ 1 mon 24 days 1 hour\n(1 row)\n\nhhs=> SELECT ('Tue Oct 10 08:00:00 2000 EDT'::timestamp + '1 mon 24 days 1\nhour'::interval);\n?column?\n----------------------------\nSun Dec 03 07:00:00 2000 CST\n(1 row)\n\n(Timezone differences are there and expected, these boxes are all across the\ncountry).\n\n\n-Mitch\n\n\n",
"msg_date": "Tue, 10 Oct 2000 14:23:08 -0700",
"msg_from": "\"Mitch Vincent\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Possible age() bug? "
}
] |
[
{
"msg_contents": "Would it be politically incorrect to combine named templates (including\nport makefiles, etc.) into one, since they all do the same thing anyway?\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Tue, 10 Oct 2000 23:40:17 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "FreeBSD, OpenBSD, NetBSD templates"
},
{
"msg_contents": "On Tue, Oct 10, 2000 at 11:40:17PM +0200, Peter Eisentraut wrote:\n> Would it be politically incorrect to combine named templates (including\n> port makefiles, etc.) into one, since they all do the same thing anyway?\n\ni'm not sure about openbsd, but freebsd likes to install \"packages\" into\n/usr/local and netbsd likes to install them in /usr/pkg.\n\nthe postgresql bundling might be different, but i thought i'd comment.\n\n-- \n[ Jim Mercer [email protected] +1 416 410-5633 ]\n[ Reptilian Research -- Longer Life through Colder Blood ]\n[ Don't be fooled by cheap Finnish imitations; BSD is the One True Code. ]\n",
"msg_date": "Tue, 10 Oct 2000 17:50:30 -0400",
"msg_from": "Jim Mercer <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: FreeBSD, OpenBSD, NetBSD templates"
},
{
"msg_contents": "On Tue, 10 Oct 2000, Peter Eisentraut wrote:\n\n> Would it be politically incorrect to combine named templates (including\n> port makefiles, etc.) into one, since they all do the same thing anyway?\n\nDon't forget bsdi. Lately they've all been referring to the group of\nthem as bsd. \n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Tue, 10 Oct 2000 18:01:26 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: FreeBSD, OpenBSD, NetBSD templates"
}
] |
[
{
"msg_contents": "> The project name on SourceForge is \"Python Interface to PostgreSQL\".\n\nHow about PIPgSQL or piPgSQL?\n\nRegards,\n//Dave\n",
"msg_date": "Tue, 10 Oct 2000 15:15:50 -0700",
"msg_from": "[email protected]",
"msg_from_op": true,
"msg_subject": "RE: Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "On Wed, 11 Oct 2000, [email protected] wrote:\n> > The project name on SourceForge is \"Python Interface to PostgreSQL\".\n> \n> How about PIPgSQL or piPgSQL?\n\nPerhaps Pi2PgSQL, or PySQL_ba\n\n-- \nSincerely etc.,\n\n NAME Christopher Sawtell\n CELL PHONE 021 257 4451\n ICQ UIN 45863470\n EMAIL csawtell @ xtra . co . nz\n CNOTES ftp://ftp.funet.fi/pub/languages/C/tutorials/sawtell_C.tar.gz\n\n -->> Please refrain from using HTML or WORD attachments in e-mails to me <<--\n\n",
"msg_date": "Wed, 11 Oct 2000 13:36:15 +1300",
"msg_from": "Christopher Sawtell <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: Announcing PgSQL - a Python DB-API 2.0 compliant interface to\n\tPostgreSQL"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Christopher Sawtell wrote:\n\n> On Wed, 11 Oct 2000, [email protected] wrote:\n> > > The project name on SourceForge is \"Python Interface to PostgreSQL\".\n> > \n> > How about PIPgSQL or piPgSQL?\n> \n> Perhaps Pi2PgSQL, or PySQL_ba\n\nPySQL_ba? *grin*\n\n\n",
"msg_date": "Tue, 10 Oct 2000 22:18:24 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [INTERFACES] Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "On Tue, 10 Oct 2000, The Hermit Hacker wrote:\n\n> On Wed, 11 Oct 2000, Christopher Sawtell wrote:\n> \n> > On Wed, 11 Oct 2000, [email protected] wrote:\n> > > > The project name on SourceForge is \"Python Interface to PostgreSQL\".\n> > > \n> > > How about PIPgSQL or piPgSQL?\n> > \n> > Perhaps Pi2PgSQL, or PySQL_ba\n> \n> PySQL_ba? *grin*\n\nPyg ??\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Tue, 10 Oct 2000 21:43:26 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: RE: [INTERFACES] Announcing PgSQL - a Python DB-API\n\t2.0 compliant interface to PostgreSQL"
},
{
"msg_contents": "Christopher Sawtell wrote:\n> On Wed, 11 Oct 2000, [email protected] wrote:\n> > > The project name on SourceForge is \"Python Interface to PostgreSQL\".\n> > How about PIPgSQL or piPgSQL?\n \n> Perhaps Pi2PgSQL, or PySQL_ba\n\nPyGSQyl.\n\nAnagram fun on \"PostgreSQL\":\nPostgreSQL=\nEg: Ports SQL\nGets Pro SQL\nGot reps, SQL?\n\n\"Great Bridge\" = Bigger Trade :-)\n\"Tuple TOASTer\" = Alert! Pest out! (The pest being the 8K limit).\n\"Outer Joins\" = Join so true.\n\"My new job\" = Enjoy BMW.\n\"Open Source\" = Our one spec.\n\"Linux versus FreeBSD\" = Feud reruns vex bliss.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Tue, 10 Oct 2000 23:17:56 -0400",
"msg_from": "Lamar Owen <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [INTERFACES] Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "Two things.\n\nFirstly, when dealing with a large ResultSet (about 120000 rows), I get a\nnull pointer exception on the line:\n wasNullFlag = (this_row[columnIndex - 1] == null);\nWhenever I call getString(), has anyone else had this? And does anybody\nhave a solution?\n\nSecondly, I've not seen it mentioned on here but the jdbc driver will\nsometimes throw a bad time stamp exception when you use getTimeStamp() on\ntimes which have are accurate to more than a second, this is the patch we\nuse to fix it.\n\nHope it is of some use to somebody,\n\nMichael Stephenson\n\n*** src/interfaces/jdbc/org/postgresql/jdbc2/ResultSet.java\tFri May 12\n20:54:22 2000\n--- src/interfaces/jdbc/org/postgresql/jdbc2/ResultSetPatch.java\nMon Sep 25 15:36:46 2000\n***************\n*** 439,445 ****\n if(s==null)\n \treturn null;\n \n! SimpleDateFormat df = new SimpleDateFormat(\"yyyy-MM-dd\nHH:mm:sszzz\");\n \n try {\n \treturn new Timestamp(df.parse(s).getTime());\n--- 439,456 ----\n if(s==null)\n \treturn null;\n \n! SimpleDateFormat df = null;\n! if (s.length()>21 && s.indexOf('.') != -1) {\n! df = new SimpleDateFormat(\"yyyy-MM-dd HH:mm:ss.SSzzz\");\n! } else if (s.length()>19 && s.indexOf('.') == -1) {\n! df = new SimpleDateFormat(\"yyyy-MM-dd HH:MM:sszzz\");\n! } else if (s.length()>19 && s.indexOf('.') != -1) {\n! df = new SimpleDateFormat(\"yyyy-MM-dd HH:MM:ss.SS\");\n! } else if (s.length()>10 && s.length()<=18) {\n! df = new SimpleDateFormat(\"yyyy-MM-dd HH:MM:ss\");\n! } else {\n! df = new SimpleDateFormat(\"yyyy-MM-dd\");\n! } \n \n try {\n \treturn new Timestamp(df.parse(s).getTime());\n\n\n",
"msg_date": "Wed, 11 Oct 2000 16:52:07 +0100 (BST)",
"msg_from": "Michael Stephenson <[email protected]>",
"msg_from_op": false,
"msg_subject": "JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Michael Stephenson wrote:\n\n> Two things.\n> \n> Firstly, when dealing with a large ResultSet (about 120000 rows), I get a\n> null pointer exception on the line:\n> wasNullFlag = (this_row[columnIndex - 1] == null);\n> Whenever I call getString(), has anyone else had this? And does anybody\n> have a solution?\n\nAre you getting any out of memory errors at all?\n\nThe problem with the current implementation is that it reads the entire\nresult into memory, so 120000 rows may be filling up your VM's memory\n(defaults to about 16Mb).\n\nDoes it occur if you add the -mx argument to java, ie:\n\n\tjava -mx 64m uk.org.retep.sql.RetepSQL\n\nI'm in the design stage of implementing a version of ResultSet that will\nuse cursors, to limit how much is loaded in memory at a time.\n\n> Secondly, I've not seen it mentioned on here but the jdbc driver will\n> sometimes throw a bad time stamp exception when you use getTimeStamp() on\n> times which have are accurate to more than a second, this is the patch we\n> use to fix it.\n\nThis was fixed a few weeks ago and should be in the current CVS already.\n\npeter\n\n-- \nPeter T Mount [email protected] http://www.retep.org.uk\nPostgreSQL JDBC Driver http://www.retep.org.uk/postgres/\nJava PDF Generator http://www.retep.org.uk/pdf/\n\n\n",
"msg_date": "Wed, 11 Oct 2000 17:11:41 +0100 (BST)",
"msg_from": "Peter Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "Peter Mount wrote:\n> \n> On Wed, 11 Oct 2000, Michael Stephenson wrote:\n> \n> > Two things.\n> >\n> > Firstly, when dealing with a large ResultSet (about 120000 rows), I get a\n> > null pointer exception on the line:\n> > wasNullFlag = (this_row[columnIndex - 1] == null);\n> > Whenever I call getString(), has anyone else had this? And does anybody\n> > have a solution?\n> \n> Are you getting any out of memory errors at all?\n> \n> The problem with the current implementation is that it reads the entire\n> result into memory, so 120000 rows may be filling up your VM's memory\n> (defaults to about 16Mb).\n> \n> Does it occur if you add the -mx argument to java, ie:\n> \n> java -mx 64m uk.org.retep.sql.RetepSQL\n> \n> I'm in the design stage of implementing a version of ResultSet that will\n> use cursors, to limit how much is loaded in memory at a time.\n> \n\nThe problem with that is the same problem I ran into with locking.\ncursors need to be in a BEGIN END block, and other calls from different\nStatement objects using the same db connectioni will interfere. Make\nsure it is clearly documented that that will not be thread safe (unless\npostgres gets named locks by then).\n",
"msg_date": "Wed, 11 Oct 2000 13:50:06 -0400",
"msg_from": "Joseph Shraibman <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Joseph Shraibman wrote:\n\n> Peter Mount wrote:\n> > \n> > On Wed, 11 Oct 2000, Michael Stephenson wrote:\n> > \n> > > Two things.\n> > >\n> > > Firstly, when dealing with a large ResultSet (about 120000 rows), I get a\n> > > null pointer exception on the line:\n> > > wasNullFlag = (this_row[columnIndex - 1] == null);\n> > > Whenever I call getString(), has anyone else had this? And does anybody\n> > > have a solution?\n> > \n> > Are you getting any out of memory errors at all?\n> > \n> > The problem with the current implementation is that it reads the entire\n> > result into memory, so 120000 rows may be filling up your VM's memory\n> > (defaults to about 16Mb).\n> > \n> > Does it occur if you add the -mx argument to java, ie:\n> > \n> > java -mx 64m uk.org.retep.sql.RetepSQL\n> > \n> > I'm in the design stage of implementing a version of ResultSet that will\n> > use cursors, to limit how much is loaded in memory at a time.\n> > \n> \n> The problem with that is the same problem I ran into with locking.\n> cursors need to be in a BEGIN END block, and other calls from different\n> Statement objects using the same db connectioni will interfere. Make\n> sure it is clearly documented that that will not be thread safe (unless\n> postgres gets named locks by then).\n\nYes, there is a problem with multiple statements and transactions on the\nsame connection when it comes to thread safety.\n\nThe problem as I see it is two fold.\n\n1: How to deal with two statements within one transaction. Related to this\nis the metadata methods that issue queries, how to deal with them while\nwithin a transaction.\n\n2: Currently the JDBC Specs only have transactions supported at the\nConnection level, so I can't see how they thought that Statements could\npossibly run within their own transactions, unless they thought that a\nworkaround of this is the use of batches.\n\nPeter\n\n-- \nPeter T Mount [email protected] http://www.retep.org.uk\nPostgreSQL JDBC Driver http://www.retep.org.uk/postgres/\nJava PDF Generator http://www.retep.org.uk/pdf/\n\n\n",
"msg_date": "Wed, 11 Oct 2000 19:08:15 +0100 (BST)",
"msg_from": "Peter Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "Peter Mount wrote:\n...\n> Yes, there is a problem with multiple statements and transactions on the\n> same connection when it comes to thread safety.\n> \n> The problem as I see it is two fold.\n> \n> 1: How to deal with two statements within one transaction. Related to this\n> is the metadata methods that issue queries, how to deal with them while\n> within a transaction.\n> \n> 2: Currently the JDBC Specs only have transactions supported at the\n> Connection level, so I can't see how they thought that Statements could\n> possibly run within their own transactions, unless they thought that a\n> workaround of this is the use of batches.\n\nAh, that probably explains why I've seen \"tuple arrived before metadata\"\nmessages when I've got several apps talking through CORBA to a java app\nthat connects to postgres. Do I need to synchronize both inserts and\nqueries at the java app level to prevent this? (I was hoping that\nthe BEGIN/END block in a transaction would be sufficient, but this makes\nit sound as though it isn't.)\n\n--\nSteve Wampler- SOLIS Project, National Solar Observatory\[email protected]\n",
"msg_date": "Wed, 11 Oct 2000 11:27:40 -0700",
"msg_from": "Steve Wampler <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "Steve Wampler wrote:\n> \n> Peter Mount wrote:\n> ...\n> > Yes, there is a problem with multiple statements and transactions on the\n> > same connection when it comes to thread safety.\n> >\n> > The problem as I see it is two fold.\n> >\n> > 1: How to deal with two statements within one transaction. Related to this\n> > is the metadata methods that issue queries, how to deal with them while\n> > within a transaction.\n> >\n> > 2: Currently the JDBC Specs only have transactions supported at the\n> > Connection level, so I can't see how they thought that Statements could\n> > possibly run within their own transactions, unless they thought that a\n> > workaround of this is the use of batches.\n> \n> Ah, that probably explains why I've seen \"tuple arrived before metadata\"\n> messages when I've got several apps talking through CORBA to a java app\n> that connects to postgres. Do I need to synchronize both inserts and\n> queries at the java app level to prevent this? (I was hoping that\n> the BEGIN/END block in a transaction would be sufficient, but this makes\n> it sound as though it isn't.)\n> \nIt isn't. I wish there were online mail archives. But anyway the\nreason it isn't is that if two statements use the same connection, when\none of them calls enters a transaction the other one does too. So if\nyou do:\n1) BEGIN;\n2) SELECT blah FROM tablea FOR UPDATE;\n3) UPDATE tablea SET blah = newblah;\n\nsince both statements are using the same connection they ARE NOT LOCKED\nFROM EACH OTHER, and when one calls and END; or COMIT; both of them exit\ntheir transactions. That is why I'm using a connection pool now when I\nhave to do locking. And if Peter wants to use cursors behind the scenes\nit will be even worse, because the cursors themselves need to be in a\nBEGIN; END;, and what happens if they user thinks he is in a transaction\nbut the cursor ended it for him a while ago? Named transactions would\nhelp with this, but the real answer would be to be able to have more\nthen one connection to a backend (maybe tunnelled over on TCP/IP link).\nRight now each new connection requires forking off another backend,\nwhich makes it impractical to connect whenever you have a new\ntransaction to do.\n",
"msg_date": "Wed, 11 Oct 2000 14:56:41 -0400",
"msg_from": "Joseph Shraibman <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Steve Wampler wrote:\n\n> Peter Mount wrote:\n> ...\n> > Yes, there is a problem with multiple statements and transactions on the\n> > same connection when it comes to thread safety.\n> > \n> > The problem as I see it is two fold.\n> > \n> > 1: How to deal with two statements within one transaction. Related to this\n> > is the metadata methods that issue queries, how to deal with them while\n> > within a transaction.\n> > \n> > 2: Currently the JDBC Specs only have transactions supported at the\n> > Connection level, so I can't see how they thought that Statements could\n> > possibly run within their own transactions, unless they thought that a\n> > workaround of this is the use of batches.\n> \n> Ah, that probably explains why I've seen \"tuple arrived before metadata\"\n> messages when I've got several apps talking through CORBA to a java app\n> that connects to postgres. Do I need to synchronize both inserts and\n> queries at the java app level to prevent this? (I was hoping that\n> the BEGIN/END block in a transaction would be sufficient, but this makes\n> it sound as though it isn't.)\n\nI think you may need to, although the existing thread locking in the\ndriver should prevent this. BEGIN/END is protecting the tables, but the\n\"tuple arrived before metadata\" message is from the network protocol\n(someone correct me at any point if I'm wrong).\n\nWhat happens at the moment is that when a query is issued by JDBC, a lock\nis made against the network connection, and then the query is issued. Once\neverything has been read, the lock is released. This mechanism should\nprevent any one thread using the same network connection as another which\nis already using it.\n\nIs your corba app under heavy load when this happens, or can it happen\nwith say 2-3 apps running?\n\nPeter\n\n-- \nPeter T Mount [email protected] http://www.retep.org.uk\nPostgreSQL JDBC Driver http://www.retep.org.uk/postgres/\nJava PDF Generator http://www.retep.org.uk/pdf/\n\n\n",
"msg_date": "Thu, 12 Oct 2000 09:05:50 +0100 (BST)",
"msg_from": "Peter Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Joseph Shraibman wrote:\n\n[snip]\n\n> It isn't. I wish there were online mail archives. But anyway the\n> reason it isn't is that if two statements use the same connection, when\n> one of them calls enters a transaction the other one does too. So if\n> you do:\n> 1) BEGIN;\n> 2) SELECT blah FROM tablea FOR UPDATE;\n> 3) UPDATE tablea SET blah = newblah;\n> \n> since both statements are using the same connection they ARE NOT LOCKED\n> FROM EACH OTHER, and when one calls and END; or COMIT; both of them exit\n> their transactions. That is why I'm using a connection pool now when I\n> have to do locking. And if Peter wants to use cursors behind the scenes\n> it will be even worse, because the cursors themselves need to be in a\n> BEGIN; END;, and what happens if they user thinks he is in a transaction\n> but the cursor ended it for him a while ago?\n\nWe already have this problem with the two MetaData interfaces. Some of\nthose methods issue their own queries, and if they fail while within the\nclient's app's transaction all hell can break loose. This is one of the\nreasons why I'm intending that the cursor based ResultSet only gets used\nif the client has requested one.\n\n> Named transactions would help with this, but the real answer would be\n> to be able to have more then one connection to a backend (maybe\n> tunnelled over on TCP/IP link).\n\nNamed transactions would help with both of these problems. Not knowing\nenough of how the existing transaction mechanism works internally, I'd say\nit's easer to do than tunnelling.\n\n> Right now each new connection requires forking off another backend,\n> which makes it impractical to connect whenever you have a new\n> transaction to do.\n\nYes, that's the last thing we want, especially while we are trying to\noptimise things...\n\nPeter\n\n-- \nPeter T Mount [email protected] http://www.retep.org.uk\nPostgreSQL JDBC Driver http://www.retep.org.uk/postgres/\nJava PDF Generator http://www.retep.org.uk/pdf/\n\n\n",
"msg_date": "Thu, 12 Oct 2000 09:28:58 +0100 (BST)",
"msg_from": "Peter Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "Peter Mount wrote:\n> \n> On Wed, 11 Oct 2000, Steve Wampler wrote:\n> \n> > Ah, that probably explains why I've seen \"tuple arrived before metadata\"\n> > messages when I've got several apps talking through CORBA to a java app\n> > that connects to postgres. Do I need to synchronize both inserts and\n> > queries at the java app level to prevent this? (I was hoping that\n> > the BEGIN/END block in a transaction would be sufficient, but this makes\n> > it sound as though it isn't.)\n> \n> I think you may need to, although the existing thread locking in the\n> driver should prevent this. BEGIN/END is protecting the tables, but the\n> \"tuple arrived before metadata\" message is from the network protocol\n> (someone correct me at any point if I'm wrong).\n> \n> What happens at the moment is that when a query is issued by JDBC, a lock\n> is made against the network connection, and then the query is issued. Once\n> everything has been read, the lock is released. This mechanism should\n> prevent any one thread using the same network connection as another which\n> is already using it.\n> \n> Is your corba app under heavy load when this happens, or can it happen\n> with say 2-3 apps running?\n\nI'm not sure how to define heavy load, but I'd say yes - there were about\n10 processes (spread across 3 machines) all talking corba to the app with\nthe jdbc app to postgres. Two apps was doing block inserts while another 8\nwere doing queries. I think there were around 100000 entries added in a\n20-25minute time span, and there would have been queries accessing most\nof those during the same period (the DB acts both as an archive and as\na cache between an instrument and the processes that analyze the instrument's\ndata).\n\n\n--\nSteve Wampler- SOLIS Project, National Solar Observatory\[email protected]\n",
"msg_date": "Thu, 12 Oct 2000 06:36:03 -0700",
"msg_from": "Steve Wampler <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "On Thu, 12 Oct 2000, Steve Wampler wrote:\n\n> Peter Mount wrote:\n> > \n> > On Wed, 11 Oct 2000, Steve Wampler wrote:\n> > \n> > > Ah, that probably explains why I've seen \"tuple arrived before metadata\"\n> > > messages when I've got several apps talking through CORBA to a java app\n> > > that connects to postgres. Do I need to synchronize both inserts and\n> > > queries at the java app level to prevent this? (I was hoping that\n> > > the BEGIN/END block in a transaction would be sufficient, but this makes\n> > > it sound as though it isn't.)\n> > \n> > I think you may need to, although the existing thread locking in the\n> > driver should prevent this. BEGIN/END is protecting the tables, but the\n> > \"tuple arrived before metadata\" message is from the network protocol\n> > (someone correct me at any point if I'm wrong).\n> > \n> > What happens at the moment is that when a query is issued by JDBC, a lock\n> > is made against the network connection, and then the query is issued. Once\n> > everything has been read, the lock is released. This mechanism should\n> > prevent any one thread using the same network connection as another which\n> > is already using it.\n> > \n> > Is your corba app under heavy load when this happens, or can it happen\n> > with say 2-3 apps running?\n> \n> I'm not sure how to define heavy load, but I'd say yes - there were about\n> 10 processes (spread across 3 machines) all talking corba to the app with\n> the jdbc app to postgres. Two apps was doing block inserts while another 8\n> were doing queries. I think there were around 100000 entries added in a\n> 20-25minute time span, and there would have been queries accessing most\n> of those during the same period (the DB acts both as an archive and as\n> a cache between an instrument and the processes that analyze the instrument's\n> data).\n\nHmmm, I think you may want to look at using a connection pool, especially\nwith 100k entries. I've just looked through my Corba books, and they all\nseem to use some form of pool, so perhaps that's the assumed best way to\ndo it.\n\nPeter\n\n-- \nPeter T Mount [email protected] http://www.retep.org.uk\nPostgreSQL JDBC Driver http://www.retep.org.uk/postgres/\nJava PDF Generator http://www.retep.org.uk/pdf/\n\n\n",
"msg_date": "Thu, 12 Oct 2000 15:15:53 +0100 (BST)",
"msg_from": "Peter Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "Peter Mount wrote:\n> \n> On Wed, 11 Oct 2000, Joseph Shraibman wrote:\n> \n> [snip]\n> \n> > It isn't. I wish there were online mail archives. But anyway the\n> > reason it isn't is that if two statements use the same connection, when\n> > one of them calls enters a transaction the other one does too. So if\n> > you do:\n> > 1) BEGIN;\n> > 2) SELECT blah FROM tablea FOR UPDATE;\n> > 3) UPDATE tablea SET blah = newblah;\n> >\n> > since both statements are using the same connection they ARE NOT LOCKED\n> > FROM EACH OTHER, and when one calls and END; or COMIT; both of them exit\n> > their transactions. That is why I'm using a connection pool now when I\n> > have to do locking. And if Peter wants to use cursors behind the scenes\n> > it will be even worse, because the cursors themselves need to be in a\n> > BEGIN; END;, and what happens if they user thinks he is in a transaction\n> > but the cursor ended it for him a while ago?\n> \n> We already have this problem with the two MetaData interfaces. Some of\n> those methods issue their own queries, and if they fail while within the\n> client's app's transaction all hell can break loose. This is one of the\n> reasons why I'm intending that the cursor based ResultSet only gets used\n> if the client has requested one.\n> \n> > Named transactions would help with this, but the real answer would be\n> > to be able to have more then one connection to a backend (maybe\n> > tunnelled over on TCP/IP link).\n> \n> Named transactions would help with both of these problems. Not knowing\n> enough of how the existing transaction mechanism works internally, I'd say\n> it's easer to do than tunnelling.\n> \n\nI don't think tunnelling would be that hard, just add a number to say\nwhich connection this query is for. The real reason it won't be done is\nthat the coders who coded the backend don't want to share a backends\nwith more than one connection to keep a segfault (or other error) from\none connection taking down all the others. I see the logic in that, but\nthere is a compromise solution of allowing one client to have more than\none connection (with locks for each connection being seperate) over the\nsame TCP/IP socket so only one client is talking with one backend\nprocess.\n\n> > Right now each new connection requires forking off another backend,\n> > which makes it impractical to connect whenever you have a new\n> > transaction to do.\n> \n> Yes, that's the last thing we want, especially while we are trying to\n> optimise things...\n> \n> Peter\n> \n> --\n> Peter T Mount [email protected] http://www.retep.org.uk\n> PostgreSQL JDBC Driver http://www.retep.org.uk/postgres/\n> Java PDF Generator http://www.retep.org.uk/pdf/\n",
"msg_date": "Thu, 12 Oct 2000 16:17:28 -0400",
"msg_from": "Joseph Shraibman <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [INTERFACES] JDBC Large ResultSet problem + BadTimeStamp Patch"
},
{
"msg_contents": "\n> > > The project name on SourceForge is \"Python Interface to PostgreSQL\".\n> > \n> > How about PIPgSQL or piPgSQL?\n> \n> Perhaps Pi2PgSQL, or PySQL_ba\n\nor PyPgSQL?\n\nDo we get a reward if you choose our name? ;)\n\n\n",
"msg_date": "Fri, 13 Oct 2000 09:24:34 +0200 (IST)",
"msg_from": "Cedar Cox <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: Announcing PgSQL - a Python DB-API 2.0 compliant\n\tinterface to PostgreSQL"
},
{
"msg_contents": "I'm sorry to bother the list with this question - I know people are\nalways asking it. I thought I understood how to do this, but my code\ndoesn't work.\n\nI am trying to save a java object in a database using the setBytes()\nmethod of PreparedStatement - I don't want to use the large object manager\nbecause I want this to be somewhat portable.\n\nThe code below gives me the error: \nException: FastPath protocol error: Z :: FastPath protocol error: Z\nAs soon as I call setBytes()\n\nI thought this only came when you forget to call \"setAutoCommit(false)\".\nCan anyone please tell me what I'm doing wrong. \n\nThanks a lot, Rob\n\n\n-------------------- PSQLTest.java -----------------------\n\nimport java.sql.* ; \nimport java.io.* ; \nimport java.util.* ;\n\npublic class PSQLTest {\n \n public PSQLTest() { }\n \n public static void main(String[] args) {\n\n\tVector vec = new Vector() ;\n\tfor (int i=0; i<10; ++i) \n\t vec.addElement(new Integer(i+5)) ;\n\n\tConnection conn = null ;\n\ttry {\n\t Class.forName(\"postgresql.Driver\") ;\n\t conn = DriverManager.getConnection\n\t\t (\"jdbc:postgresql://127.0.0.1:5432/rob\", \"rob\", \"\") ;\n\t conn.setAutoCommit(false);\n\n\t byte[] bytes ;\n\t ByteArrayOutputStream out = new ByteArrayOutputStream() ;\n\t ObjectOutputStream objOut = new ObjectOutputStream(out) ;\n\t objOut.writeObject(vec) ;\n\t objOut.flush() ;\n\t bytes = out.toByteArray() ;\n\t objOut.close() ;\n\n\t PreparedStatement ps = conn.prepareStatement \n\t\t(\"insert into vectors (name, id) values ( ?, ?)\") ; \n\n\t ps.setString(1, \"Vector name\") ;\n\t ps.setBytes(2, bytes) ; \n\t ps.executeUpdate() ; \n\t ps.close() ; \n\n\t conn.commit() ;\n\t} catch (Exception e) { \n\t System.out.println(\"Exception: \"+e+\" :: \"+e.getMessage());\n\t e.printStackTrace() ;\n\t}\n\n\t\n\n\t\n }\n \n}\n\n",
"msg_date": "Mon, 16 Oct 2000 20:57:02 -0500 (CDT)",
"msg_from": "Rob Judd <[email protected]>",
"msg_from_op": false,
"msg_subject": "Save java objects using JDBC"
},
{
"msg_contents": "Vince Vielhaber wrote:\n> On Tue, 10 Oct 2000, The Hermit Hacker wrote:\n> \n> > On Wed, 11 Oct 2000, Christopher Sawtell wrote:\n> > \n> > > On Wed, 11 Oct 2000, [email protected] wrote:\n> > > > > The project name on SourceForge is \"Python Interface to PostgreSQL\".\n> > > > \n> > > > How about PIPgSQL or piPgSQL?\n> > > \n> > > Perhaps Pi2PgSQL, or PySQL_ba\n> > \n> > PySQL_ba? *grin*\n> \n> Pyg ??\n\n PyLephant\n\n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#================================================== [email protected] #\n\n",
"msg_date": "Mon, 23 Oct 2000 04:48:35 -0500 (EST)",
"msg_from": "Jan Wieck <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] RE: Announcing PgSQL - a Python DB-API 2.0\n\tcompliant interface to PostgreSQLL"
},
{
"msg_contents": "On Mon, 23 Oct 2000, Jan Wieck wrote:\n\n> Vince Vielhaber wrote:\n> > On Tue, 10 Oct 2000, The Hermit Hacker wrote:\n> > \n> > > On Wed, 11 Oct 2000, Christopher Sawtell wrote:\n> > > \n> > > > On Wed, 11 Oct 2000, [email protected] wrote:\n> > > > > > The project name on SourceForge is \"Python Interface to PostgreSQL\".\n> > > > > \n> > > > > How about PIPgSQL or piPgSQL?\n> > > > \n> > > > Perhaps Pi2PgSQL, or PySQL_ba\n> > > \n> > > PySQL_ba? *grin*\n> > \n> > Pyg ??\n> \n> PyLephant\n> \n\nI like that!\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Mon, 23 Oct 2000 11:50:43 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] RE: Announcing PgSQL - a Python DB-API\n\t2.0 compliant interface to PostgreSQLL"
}
] |
[
{
"msg_contents": "If anyone was following my request to have a large object api for TOAST,\nthis is of interest.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n\nTom Lane wrote:\n> My own feeling is that the current LO setup is fundamentally flawed\n> by its reliance on specific OID values, and that the right answer is\n> to find a way to avoid that. contrib/lo might provide some food for\n> thought here (although it's clearly not the whole answer either).\n\n I have some ideas to replace the entire LO handling by\n something completely different. More compatible with the\n Oracle way of handling large objects. That is, that on\n INSERT/UPDATE someone uses a function to place an LO\n reference into the tuple. Then having a new interface in\n libpq, that works like file I/O on the references that appear\n in a result set. To open them for writing, the user must have\n selected them for update, otherwise he can open them for\n reading. The file I/O itself can be based on the fastpath\n interface.\n\n The LO's follow Oracles copy semantic, so if someone does\n INSERT ... SELECT, the LO gets duplicated.\n\n All LO's for one base table can be stored in one big,\n segmented external heap (more or less like toast values). The\n system would automatically know when objects are obsolete.\n\n An INSERT operation might then return a result set as if\n someone did a SELECT FOR UPDATE of exactly what he just\n inserted. So he immediately has access to the LO references\n to place the values.\n\n Don't know yet how to interface it from psql - but let me\n sleep another night or so over it.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#================================================== [email protected] #",
"msg_date": "Tue, 10 Oct 2000 21:08:39 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: heap_create with OID? (fwd)"
}
] |
[
{
"msg_contents": "This relates to large objects too.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n\nAt 14:41 4/07/00 +0200, Jan Wieck wrote:\n>Tom Lane wrote:\n>\n> I have some ideas to replace the entire LO handling by\n> something completely different. More compatible with the\n> Oracle way of handling large objects. That is, that on\n> INSERT/UPDATE someone uses a function to place an LO\n> reference into the tuple. Then having a new interface in\n> libpq, that works like file I/O on the references that appear\n> in a result set. To open them for writing, the user must have\n> selected them for update, otherwise he can open them for\n> reading. \n\nIt might be worth looking at the SQL standard and it's 'locators' for BLOB\ndata. They sound a lot like this, but may have some requirements that\naffect your design.\n\n\n> The file I/O itself can be based on the fastpath\n> interface.\n\nWhat's the 'fastpath interface'?\n\n\n> The LO's follow Oracles copy semantic, so if someone does\n> INSERT ... SELECT, the LO gets duplicated.\n\nGreat. The standard seems (by default) to treat them as just another\n'value', which this is consistent with. It may be relevant to mention that\nthe standard also allows substr, ||, etc to work on BLOBs.\n\n\n> An INSERT operation might then return a result set as if\n> someone did a SELECT FOR UPDATE of exactly what he just\n> inserted. So he immediately has access to the LO references\n> to place the values.\n>\n> Don't know yet how to interface it from psql - but let me\n> sleep another night or so over it.\n\nFWIW, Dec/Rdb uses a kind of cursor for the BLOB data. The row insert must\nbe done in a normal insert cursor, then another cursor is opened to write\nthe BLOB data. It's very cumbersome. But if you use an insert cursor for\nyour table rows, then at least you do have a context available for BLOB\ninsertion.\n\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.C.N. 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/",
"msg_date": "Tue, 10 Oct 2000 21:08:51 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: heap_create with OID? (fwd)"
}
] |
[
{
"msg_contents": "At times I need to call PQendcopy, how to I determine that it won't\nblock me waiting for output from the backend?\n\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Tue, 10 Oct 2000 19:23:17 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": true,
"msg_subject": "calling PQendcopy() without blocking."
}
] |
[
{
"msg_contents": "I just got this question in my mailbox. Anyone with more knowledge than I\ncan probably answer it. :-)\n\n> I need \"generic\" trigger functions, that is trigger functions that have to know\n> about the table they have been triggerd from.\n> \n> Example: \n> a trigger function that generates a crc sum on all fields of any table.\n> \n> In pltcl I can get the field names of the triggering table, but I am not able\n> to call my \"C\" function from within pltcl (I can call my \"C\" fuinction\n> perfectly well from the psql command line or from withing a plpgsql function.\n> \n> Any helpful ideas? Appreciate any hint.\n\nMichael\n-- \nMichael Meskes\[email protected]\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Wed, 11 Oct 2000 17:04:37 +0200",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": true,
"msg_subject": "Trigger question"
}
] |
[
{
"msg_contents": "Does anyone know what backend/tioga is for? It's not Spanish for TOAST,\nis it? :-) It looks like it wasn't compilable for quite a while.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Wed, 11 Oct 2000 20:01:02 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": true,
"msg_subject": "TIOGA"
},
{
"msg_contents": "> Does anyone know what backend/tioga is for? It's not Spanish for TOAST,\n> is it? :-) It looks like it wasn't compilable for quite a while.\n\nIt was some db visualization tool. There is a web page about it\nsomewhere at Berkeley. I say remove the directory. It is junk now.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Wed, 11 Oct 2000 14:12:57 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: TIOGA"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Bruce Momjian wrote:\n\n> > Does anyone know what backend/tioga is for? It's not Spanish for TOAST,\n> > is it? :-) It looks like it wasn't compilable for quite a while.\n> \n> It was some db visualization tool. There is a web page about it\n> somewhere at Berkeley. I say remove the directory. It is junk now.\n\nthat is a bit rash considering that you tend to imply that you don't even\nknow what it is?\n\nPeter, according to google, the primary Berkeley site that comes up is:\n\n\thttp://datasplash.cs.berkeley.edu/\n\n\"DataSplash combines a sophisticated navigation model with a paint program\ninterface with which users can create custom visualizations. A beta\nversion of the software is now available.\"\n\n\tAnd, from the overview:\n\n\"Tioga DataSplash (formerly known as Tioga-2) allows users to\ninteractively create and browse visualizations of database tables.\nDataSplash was developed by members of the Tioga Project at the University\nof California at Berkeley. DataSplash software is available on Alpha/OSF,\nHP/UX, Intel/Linux, and Sparc /Solaris platforms. We are porting to other\nUNIX platforms as well.\"\n\n\tThe \"Quick Tour\" looks really cool, but I'm not sure how tied into\nthe whole backend/tioga directory this whole thing is. My first feel is\nthat 'tioga datasplash' is a userland application, but someone else might\nbe able to make a better determination?\n\t\n\tCheck out this, which explains what 'Splashes' are ... it almost\ngives me the feel of Visual Statistical Modeling ... ?\n\n\thttp://datasplash.cs.berkeley.edu/splashes.html\n\n\tIf it isn't breaking anything, please leave it in there, as I'd\nlike to look into this a bit further ...\n\n\n\n",
"msg_date": "Wed, 11 Oct 2000 15:28:06 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: TIOGA"
},
{
"msg_contents": "Considering no one has used it in 4 years that I remember, my guess is\nthat it is an abandonded Berkeley project.\n\n> On Wed, 11 Oct 2000, Bruce Momjian wrote:\n> \n> > > Does anyone know what backend/tioga is for? It's not Spanish for TOAST,\n> > > is it? :-) It looks like it wasn't compilable for quite a while.\n> > \n> > It was some db visualization tool. There is a web page about it\n> > somewhere at Berkeley. I say remove the directory. It is junk now.\n> \n> that is a bit rash considering that you tend to imply that you don't even\n> know what it is?\n> \n> Peter, according to google, the primary Berkeley site that comes up is:\n> \n> \thttp://datasplash.cs.berkeley.edu/\n> \n> \"DataSplash combines a sophisticated navigation model with a paint program\n> interface with which users can create custom visualizations. A beta\n> version of the software is now available.\"\n> \n> \tAnd, from the overview:\n> \n> \"Tioga DataSplash (formerly known as Tioga-2) allows users to\n> interactively create and browse visualizations of database tables.\n> DataSplash was developed by members of the Tioga Project at the University\n> of California at Berkeley. DataSplash software is available on Alpha/OSF,\n> HP/UX, Intel/Linux, and Sparc /Solaris platforms. We are porting to other\n> UNIX platforms as well.\"\n> \n> \tThe \"Quick Tour\" looks really cool, but I'm not sure how tied into\n> the whole backend/tioga directory this whole thing is. My first feel is\n> that 'tioga datasplash' is a userland application, but someone else might\n> be able to make a better determination?\n> \t\n> \tCheck out this, which explains what 'Splashes' are ... it almost\n> gives me the feel of Visual Statistical Modeling ... ?\n> \n> \thttp://datasplash.cs.berkeley.edu/splashes.html\n> \n> \tIf it isn't breaking anything, please leave it in there, as I'd\n> like to look into this a bit further ...\n> \n> \n> \n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Wed, 11 Oct 2000 14:31:33 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: TIOGA"
},
{
"msg_contents": "> On Wed, 11 Oct 2000, Bruce Momjian wrote:\n> \n> > > Does anyone know what backend/tioga is for? It's not Spanish for TOAST,\n> > > is it? :-) It looks like it wasn't compilable for quite a while.\n> > \n> > It was some db visualization tool. There is a web page about it\n> > somewhere at Berkeley. I say remove the directory. It is junk now.\n> \n> that is a bit rash considering that you tend to imply that you don't even\n\nI was born with a rash..., oh you meant something else. :-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Wed, 11 Oct 2000 14:36:04 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: TIOGA"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Bruce Momjian wrote:\n\n> Considering no one has used it in 4 years that I remember, my guess is\n> that it is an abandonded Berkeley project.\n\nIf it does not break anything in our build process, please let it in there\nwhile I do further research on what it is ... I personally think what I've\nseen on the web site so far is interesting and bears further investigation\nbefore we just yank it out ...\n\n \n> > > On Wed, 11 Oct 2000, Bruce Momjian wrote:\n> > \n> > > > Does anyone know what backend/tioga is for? It's not Spanish for TOAST,\n> > > > is it? :-) It looks like it wasn't compilable for quite a while.\n> > > \n> > > It was some db visualization tool. There is a web page about it\n> > > somewhere at Berkeley. I say remove the directory. It is junk now.\n> > \n> > that is a bit rash considering that you tend to imply that you don't even\n> > know what it is?\n> > \n> > Peter, according to google, the primary Berkeley site that comes up is:\n> > \n> > \thttp://datasplash.cs.berkeley.edu/\n> > \n> > \"DataSplash combines a sophisticated navigation model with a paint program\n> > interface with which users can create custom visualizations. A beta\n> > version of the software is now available.\"\n> > \n> > \tAnd, from the overview:\n> > \n> > \"Tioga DataSplash (formerly known as Tioga-2) allows users to\n> > interactively create and browse visualizations of database tables.\n> > DataSplash was developed by members of the Tioga Project at the University\n> > of California at Berkeley. DataSplash software is available on Alpha/OSF,\n> > HP/UX, Intel/Linux, and Sparc /Solaris platforms. We are porting to other\n> > UNIX platforms as well.\"\n> > \n> > \tThe \"Quick Tour\" looks really cool, but I'm not sure how tied into\n> > the whole backend/tioga directory this whole thing is. My first feel is\n> > that 'tioga datasplash' is a userland application, but someone else might\n> > be able to make a better determination?\n> > \t\n> > \tCheck out this, which explains what 'Splashes' are ... it almost\n> > gives me the feel of Visual Statistical Modeling ... ?\n> > \n> > \thttp://datasplash.cs.berkeley.edu/splashes.html\n> > \n> > \tIf it isn't breaking anything, please leave it in there, as I'd\n> > like to look into this a bit further ...\n> > \n> > \n> > \n> > \n> \n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> [email protected] | (610) 853-3000\n> + If your life is a hard drive, | 830 Blythe Avenue\n> + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n> \n\nMarc G. Fournier ICQ#7615664 IRC Nick: Scrappy\nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Wed, 11 Oct 2000 15:38:16 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: TIOGA"
},
{
"msg_contents": "> On Wed, 11 Oct 2000, Bruce Momjian wrote:\n> \n> > Considering no one has used it in 4 years that I remember, my guess is\n> > that it is an abandonded Berkeley project.\n> \n> If it does not break anything in our build process, please let it in there\n> while I do further research on what it is ... I personally think what I've\n> seen on the web site so far is interesting and bears further investigation\n> before we just yank it out ...\n\nAgreed. It looked quite nifty.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Wed, 11 Oct 2000 14:40:47 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: TIOGA"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Bruce Momjian wrote:\n> Considering no one has used it in 4 years that I remember, my guess is\n> that it is an abandonded Berkeley project.\n\nIt appeard to work the last time I tried it (2 years ago or so.)\n\n-- \n| Matthew N. Dodd | '78 Datsun 280Z | '75 Volvo 164E | FreeBSD/NetBSD |\n| [email protected] | 2 x '84 Volvo 245DL | ix86,sparc,pmax |\n| http://www.jurai.net/~winter | This Space For Rent | ISO8802.5 4ever |\n\n",
"msg_date": "Thu, 12 Oct 2000 02:53:10 -0400 (EDT)",
"msg_from": "\"Matthew N. Dodd\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: TIOGA"
},
{
"msg_contents": "On Wed, Oct 11, 2000 at 02:40:47PM -0400, Bruce Momjian wrote:\n> > On Wed, 11 Oct 2000, Bruce Momjian wrote:\n> > \n> > > Considering no one has used it in 4 years that I remember, my guess is\n> > > that it is an abandonded Berkeley project.\n> > \n> > If it does not break anything in our build process, please let it in there\n> > while I do further research on what it is ... I personally think what I've\n> > seen on the web site so far is interesting and bears further investigation\n> > before we just yank it out ...\n> \n> Agreed. It looked quite nifty.\n> \n\nWhite papers refering to Tioga, stretching back to '92 (latest from '96)\ncan be found at:\n\nhttp://s2k-ftp.cs.berkeley.edu:8000/postgres/papers/\n\nRoss\n-- \nOpen source code is like a natural resource, it's the result of providing\nfood and sunshine to programmers, and then staying out of their way.\n[...] [It] is not going away because it has utility for both the developers \nand users independent of economic motivations. Jim Flynn, Sunnyvale, Calif.\n",
"msg_date": "Fri, 13 Oct 2000 14:11:19 -0500",
"msg_from": "\"Ross J. Reedstrom\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: TIOGA"
}
] |
[
{
"msg_contents": "How far are we from seeing the version 7.1 out?\n\n-- \n\"And I'm happy, because you make me feel good, about me.\" - Melvin Udall\n-----------------------------------------------------------------\nMart�n Marqu�s\t\t\temail: \[email protected]\nSanta Fe - Argentina\t\thttp://math.unl.edu.ar/~martin/\nAdministrador de sistemas en math.unl.edu.ar\n-----------------------------------------------------------------\n",
"msg_date": "Wed, 11 Oct 2000 22:22:50 -0300",
"msg_from": "\"Martin A. Marques\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "postgresql 7.1"
},
{
"msg_contents": "What;s the new in Postgresql 7.1 ?\n\n\nAsidha Luhwidyanto\n\n\"Martin A. Marques\" wrote:\n\n> How far are we from seeing the version 7.1 out?\n>\n> --\n> \"And I'm happy, because you make me feel good, about me.\" - Melvin Udall\n> -----------------------------------------------------------------\n> Mart�n Marqu�s email: [email protected]\n> Santa Fe - Argentina http://math.unl.edu.ar/~martin/\n> Administrador de sistemas en math.unl.edu.ar\n> -----------------------------------------------------------------\n\n",
"msg_date": "Thu, 12 Oct 2000 08:45:34 +0700",
"msg_from": "Asidha Luhwidyanto <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgresql 7.1"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Martin A. Marques wrote:\n\n> How far are we from seeing the version 7.1 out?\n\nbeta starts ~Nov 1st, release in January ...\n\n\n",
"msg_date": "Wed, 11 Oct 2000 22:46:06 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgresql 7.1"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Asidha Luhwidyanto wrote:\n> What;s the new in Postgresql 7.1 ?\n\nTOAST and WAL!!!!\n\nI think these are big changes that will make me think about postgresql as a \nbig time competitor against the big commercial database engines.\n\nWAL is a backup system.\nTOAST is a system for working with rows that have to use more then the 8K \nlimitation.\nAFAIK!\n\n-- \n\"And I'm happy, because you make me feel good, about me.\" - Melvin Udall\n-----------------------------------------------------------------\nMart�n Marqu�s\t\t\temail: \[email protected]\nSanta Fe - Argentina\t\thttp://math.unl.edu.ar/~martin/\nAdministrador de sistemas en math.unl.edu.ar\n-----------------------------------------------------------------\n",
"msg_date": "Wed, 11 Oct 2000 23:04:02 -0300",
"msg_from": "\"Martin A. Marques\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: postgresql 7.1"
},
{
"msg_contents": "* Martin A. Marques <[email protected]> [001011 19:10] wrote:\n> On Wed, 11 Oct 2000, Asidha Luhwidyanto wrote:\n> > What;s the new in Postgresql 7.1 ?\n> \n> TOAST and WAL!!!!\n> \n> I think these are big changes that will make me think about postgresql as a \n> big time competitor against the big commercial database engines.\n> \n> WAL is a backup system.\n\nafaik WAL means the end of the dreaded 'vacuum', it allows the system\nto reuse free space without explicitly scanning the datafiles, it should\nalso do it in a manner that still allows transactional dumping.\n\n-- \n-Alfred Perlstein - [[email protected]|[email protected]]\n\"I have the heart of a child; I keep it in a jar on my desk.\"\n",
"msg_date": "Wed, 11 Oct 2000 19:32:48 -0700",
"msg_from": "Alfred Perlstein <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgresql 7.1"
},
{
"msg_contents": "On Wed, 11 Oct 2000, Tim Uckun wrote:\n\n> At 11:04 PM 10/11/2000 -0300, Martin A. Marques wrote:\n> >On Wed, 11 Oct 2000, Asidha Luhwidyanto wrote:\n> > > What;s the new in Postgresql 7.1 ?\n> >\n> >TOAST and WAL!!!!\n> >\n> >I think these are big changes that will make me think about postgresql as a\n> >big time competitor against the big commercial database engines.\n> >\n> >WAL is a backup system.\n> >TOAST is a system for working with rows that have to use more then the 8K\n> >limitation.\n> >AFAIK!\n> \n> What happened to outer joins? Don't you need outer joins to compete with \n> the big boys?\n\ntom lane has been working aggressively on that :)\n\n\n",
"msg_date": "Thu, 12 Oct 2000 02:14:07 -0300 (ADT)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgresql 7.1"
},
{
"msg_contents": "At 11:04 PM 10/11/2000 -0300, Martin A. Marques wrote:\n>On Wed, 11 Oct 2000, Asidha Luhwidyanto wrote:\n> > What;s the new in Postgresql 7.1 ?\n>\n>TOAST and WAL!!!!\n>\n>I think these are big changes that will make me think about postgresql as a\n>big time competitor against the big commercial database engines.\n>\n>WAL is a backup system.\n>TOAST is a system for working with rows that have to use more then the 8K\n>limitation.\n>AFAIK!\n\nWhat happened to outer joins? Don't you need outer joins to compete with \nthe big boys?\n----------------------------------------------\n Tim Uckun\n Mobile Intelligence Unit.\n----------------------------------------------\n \"There are some who call me TIM?\"\n----------------------------------------------\n",
"msg_date": "Wed, 11 Oct 2000 22:21:55 -0700",
"msg_from": "Tim Uckun <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgresql 7.1"
},
{
"msg_contents": "Tim Uckun <[email protected]> writes:\n> What happened to outer joins? Don't you need outer joins to compete with \n> the big boys?\n\nThey're done too ;-)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 12 Oct 2000 01:37:03 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgresql 7.1 "
},
{
"msg_contents": "At 01:37 AM 10/12/2000 -0400, Tom Lane wrote:\n>Tim Uckun <[email protected]> writes:\n> > What happened to outer joins? Don't you need outer joins to compete with\n> > the big boys?\n>\n>They're done too ;-)\n\nWooo Hooo time to break out the champagne!\n----------------------------------------------\n Tim Uckun\n Mobile Intelligence Unit.\n----------------------------------------------\n \"There are some who call me TIM?\"\n----------------------------------------------\n",
"msg_date": "Thu, 12 Oct 2000 02:43:24 -0700",
"msg_from": "Tim Uckun <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgresql 7.1 "
},
{
"msg_contents": "So a new version will be out soon. How often Bruce's book will be revised?\nWith each new version?\n\n----- Original Message -----\nFrom: \"Tim Uckun\" <[email protected]>\nTo: <[email protected]>\nSent: Thursday, October 12, 2000 5:43 AM\nSubject: Re: [GENERAL] postgresql 7.1\n\n\n> At 01:37 AM 10/12/2000 -0400, Tom Lane wrote:\n> >Tim Uckun <[email protected]> writes:\n> > > What happened to outer joins? Don't you need outer joins to compete\nwith\n> > > the big boys?\n> >\n> >They're done too ;-)\n>\n> Wooo Hooo time to break out the champagne!\n> ----------------------------------------------\n> Tim Uckun\n> Mobile Intelligence Unit.\n> ----------------------------------------------\n> \"There are some who call me TIM?\"\n> ----------------------------------------------\n>\n",
"msg_date": "Thu, 12 Oct 2000 09:12:27 -0400",
"msg_from": "\"Efrain Caro\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgresql 7.1"
},
{
"msg_contents": "Martin A. Marques writes:\n\n> How far are we from seeing the version 7.1 out?\n\nBeta is supposed to be November 1st. Personally, I'm putting my chips on\na February 1st release.\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Thu, 12 Oct 2000 16:43:04 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] postgresql 7.1"
},
{
"msg_contents": "The Hermit Hacker writes:\n\n> > How far are we from seeing the version 7.1 out?\n> \n> beta starts ~Nov 1st, release in January ...\n\nJust wondering, WAL is going to be integrated when, and that gives how\nmuch time to test it before releasing the beta?\n\n-- \nPeter Eisentraut [email protected] http://yi.org/peter-e/\n\n",
"msg_date": "Thu, 12 Oct 2000 16:51:57 +0200 (CEST)",
"msg_from": "Peter Eisentraut <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: postgresql 7.1"
},
{
"msg_contents": "[ Charset ISO-8859-1 unsupported, converting... ]\n> So a new version will be out soon. How often Bruce's book will be revised?\n> With each new version?\n\nGood question. I know the initial printing run was reduced, perhaps so\nnew editions could be made.\n\nI do the entire book here, and deliver camera-ready copy, so the process\nof generating a new edition is pretty easy.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Thu, 12 Oct 2000 11:29:50 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgresql 7.1"
},
{
"msg_contents": "This sort of thing had crossed my mind because of sendmail (among many other\nopen source projects). Tring to get up to date info on running it (8.10\nversion) and the \"BEST\" book out there is 8.x version, which really is not\ntoo relevant for 8.10 features.\n\nVersioning changes often as well as new features and revisions of old ones.\nAlso, compilation issues always change from release to release. For a\npublishing company it would be expensive to constantly put out a book for\neach version that entails everything. Plus, consumers really don't want to\nkeep buying the same material over and over again.\n\nFor prohjects such as this that have commercial documentation, why don't\nthey have \"patches\" for printed books also?\n\nExample, Software 1.0, author releases a full fledged book called Software\n1.0 Bible. Contains compiling issues, how to use the software, etc. All\nthat you could want from a 40-60 dollar book. Software 1.1 is release 3\nmonths later. After that author writes a short and concise book titled\nSoftware Bible from 1.0 to 1.1. Doesn't cover the same info as 1.0 Bible.\nGoes into as much detail as the first book did, but only involving new info\npertaining to 1.1 (and mayhaps a chapter on upgrading from 1.0 to 1.1). So\ninstead of another 400 page book, maybe it is only 100, depending on the new\nfeature set from version to version. Book assumes you HAVE read 1.0 Bible or\nknow enough about 1.0 where you don't need it.\n\nSoftware keeps doing releases and the same production release for books\nfollows. After two years, Software 2.0 comes out (implying a significant\nversion difference from 1.0). A new book is written (re-written) and it\nstarts from scratch and called Software 2.0 Bible. Same book \"patch\"\nschedule is followed as previously, etc.\n\nIt would be an interesting documentation project that would really keep\ninformation organized and relatively accessible ('cause sometimes digging\nthrough webpages and email groups is too time consuming).\n\n\nAdam Lang\nSystems Engineer\nRutgers Casualty Insurance Company\n----- Original Message -----\nFrom: \"Bruce Momjian\" <[email protected]>\nTo: \"Efrain Caro\" <[email protected]>\nCc: <[email protected]>\nSent: Thursday, October 12, 2000 11:29 AM\nSubject: Re: [GENERAL] postgresql 7.1\n\n\n> [ Charset ISO-8859-1 unsupported, converting... ]\n> > So a new version will be out soon. How often Bruce's book will be\nrevised?\n> > With each new version?\n>\n> Good question. I know the initial printing run was reduced, perhaps so\n> new editions could be made.\n>\n> I do the entire book here, and deliver camera-ready copy, so the process\n> of generating a new edition is pretty easy.\n>\n> --\n> Bruce Momjian | http://candle.pha.pa.us\n> [email protected] | (610) 853-3000\n> + If your life is a hard drive, | 830 Blythe Avenue\n> + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n\n",
"msg_date": "Thu, 12 Oct 2000 11:54:21 -0400",
"msg_from": "\"Adam Lang\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: postgresql 7.1"
}
] |
[
{
"msg_contents": "Added to TODO:\n\n\t* add XML interface capability\n\n> Hello Bruce,\n> I was browsing in your postgres page TODO list and saw no reference\n> at all about XML. Although I am aware of the importance the other tasks\n> in the list have, I strongly believe XML support should be somewhere in\n> the plan. XML is an excellent way to define Object-Relational\n> structures (which i believe is a goal of the postgress project) and an\n> open standard for exchange of information between applications,\n> especially e-business ones. Some recommended reading is the IEEE Data\n> engineering workshop proceedings (2000) where you can read about the\n> work of researchers of IBM, Oracle, Informix and Microsoft on the\n> subject.\n> I cannot volunteer for the project at this time because i am\n> initiating another open source project, i just wanted to share this\n> thought.\n> \n> Keep up the excellent work on postgres,\n> Congratulations on the benchmarking results reported on Apachetoday.\n> \n> Titos Dragonas.\n> \n> \n> \n> __________________________________________________\n> Do You Yahoo!?\n> Talk to your friends online with Yahoo! Messenger.\n> http://im.yahoo.com\n> \n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Thu, 12 Oct 2000 00:19:32 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Postgress & XML"
}
] |
[
{
"msg_contents": "I just spent a fruitless half hour rooting around the website for\nour elephant-in-crystal logo. I need a moderately large, clean copy\nfor the title slide of a presentation. Where's the original hiding?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 12 Oct 2000 01:12:20 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": true,
"msg_subject": "Where's the PostgreSQL logo these days?"
},
{
"msg_contents": "On Thu, 12 Oct 2000, Tom Lane wrote:\n\n> I just spent a fruitless half hour rooting around the website for\n> our elephant-in-crystal logo. I need a moderately large, clean copy\n> for the title slide of a presentation. Where's the original hiding?\n\nWe never really had a large one. There was a gold colored one on the\nprevious website but it wasn't clean since it was a blowup of the size\nwe have on the site now. There are some others on the pgsql site in\nthe propaganda section or perhaps we can get Jan to generate a large\nversion.\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] http://www.pop4.net\n 128K ISDN from $22.00/mo - 56K Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n",
"msg_date": "Thu, 12 Oct 2000 07:05:34 -0400 (EDT)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Where's the PostgreSQL logo these days?"
},
{
"msg_contents": "Vince Vielhaber <[email protected]> writes:\n> We never really had a large one.\n\nDrat. Guess I'll have to use the cartoon elephant instead.\n\nDoes anyone know how to convert a \".ai\" file into EPS? I think that's\nsome kind of vector-art format, and it'd be nice to preserve scalability.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 12 Oct 2000 10:40:59 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Where's the PostgreSQL logo these days? "
},
{
"msg_contents": "Tom Lane wrote:\n> \n> Vince Vielhaber <[email protected]> writes:\n> > We never really had a large one.\n> \n> Drat. Guess I'll have to use the cartoon elephant instead.\n> \n> Does anyone know how to convert a \".ai\" file into EPS? I think that's\n> some kind of vector-art format, and it'd be nice to preserve scalability.\n> \n> regards, tom lane\n\ni know you can open them in paint shop pro, but i don't know if it\nrasterizes them or not. you might try ghostscript, too, because i think\nit's just a variant of postscript. if you're in a bind, we have a copy\nof adobe illustrator here that i can install and see what the options\nare.\n\n-- \n\nJeff Hoffmann\nPropertyKey.com\n",
"msg_date": "Thu, 12 Oct 2000 10:06:19 -0500",
"msg_from": "Jeff Hoffmann <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Where's the PostgreSQL logo these days?"
},
{
"msg_contents": "On Thu, 12 Oct 2000, Tom Lane wrote:\n\n> Vince Vielhaber <[email protected]> writes:\n> > We never really had a large one.\n> \n> Drat. Guess I'll have to use the cartoon elephant instead.\n> \n> Does anyone know how to convert a \".ai\" file into EPS? I think that's\n> some kind of vector-art format, and it'd be nice to preserve scalability.\n\nJust checked Photoshop, and it can handle .ai although I've only got 5.5,\nso it would get changed into a bitmap :-(\n\nIt lists it under \"Generic EPS (*.eps, *.AI, ...)\" so its similar to eps\nanyhow.\n\nPeter\n\n-- \nPeter T Mount [email protected] http://www.retep.org.uk\nPostgreSQL JDBC Driver http://www.retep.org.uk/postgres/\nJava PDF Generator http://www.retep.org.uk/pdf/\n\n\n",
"msg_date": "Thu, 12 Oct 2000 16:24:14 +0100 (BST)",
"msg_from": "Peter Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Where's the PostgreSQL logo these days? "
},
{
"msg_contents": "I looked for that myself. It is on the web site in slices so it can be\nrendered faster in a browser.\n\n> I just spent a fruitless half hour rooting around the website for\n> our elephant-in-crystal logo. I need a moderately large, clean copy\n> for the title slide of a presentation. Where's the original hiding?\n> \n> \t\t\tregards, tom lane\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Thu, 12 Oct 2000 11:47:45 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Where's the PostgreSQL logo these days?"
},
{
"msg_contents": "Peter Mount <[email protected]> writes:\n> It lists it under \"Generic EPS (*.eps, *.AI, ...)\" so its similar to eps\n> anyhow.\n\nYes, I found that I just had to comment out some stuff at the end\n(%%PageTrailer and showpage) to get the file to work as plain EPS.\nNow I've got a pretty sharp-looking title page, and it's pure scalable\nPostscript which means I'm not limited to a particular resolution.\nExcellent...\n\nI'd still like an EPS copy of the crystal elephant someday, though.\nJan?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 12 Oct 2000 12:39:37 -0400",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Where's the PostgreSQL logo these days? "
},
{
"msg_contents": "> Peter Mount <[email protected]> writes:\n> > It lists it under \"Generic EPS (*.eps, *.AI, ...)\" so its similar to eps\n> > anyhow.\n> \n> Yes, I found that I just had to comment out some stuff at the end\n> (%%PageTrailer and showpage) to get the file to work as plain EPS.\n> Now I've got a pretty sharp-looking title page, and it's pure scalable\n> Postscript which means I'm not limited to a particular resolution.\n> Excellent...\n> \n> I'd still like an EPS copy of the crystal elephant someday, though.\n> Jan?\n\nHe is in Poland until Sunday for a speech.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Thu, 12 Oct 2000 12:44:50 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Where's the PostgreSQL logo these days?u"
}
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.