threads
listlengths 1
2.99k
|
|---|
[
{
"msg_contents": "JDBC works\npostgres works\n\nI is happy :)\n[wasn't posted so I figure I'd post]\n\nRuns very nicely under egcs-2.91.06 FWIW...\n[about 3-5 times faster than before ... though that could be postgres\nimprovements - and probably is]\n\nplatform : linux (I'm not posting kernel version! it doesn't matter!! :)\n\tegcs-2.91.06\t(gcc-2.8.0 with haifa scheduler + other updates)\n\tglibc-2.0.5c\tfrom RedHat-5.0 distrib - should be stable\n\t\t\t[but with full crypt, locale]\n\nBut : Here's output from regression tests:\nIs there anything wrong with the failed tests? (is it known?)\n\n=============== Notes... =================\npostmaster must already be running for the regression tests to succeed.\nThe time zone is now set to PST8PDT explicitly by this regression test\n client frontend. Please report any apparent problems to\n [email protected]\nSee regress/README for more information.\n\n=============== destroying old regression database... =================\nERROR: destroydb: database regression does not exist.\ndestroydb: database destroy failed on regression.\n=============== creating new regression database... =================\n=============== running regression queries... =================\nboolean .. ok\nchar .. ok\nchar2 .. ok\nchar4 .. ok\nchar8 .. ok\nchar16 .. ok\nvarchar .. ok\ntext .. ok\nstrings .. failed\nint2 .. failed\nint4 .. failed\noid .. ok\noidint2 .. failed\noidint4 .. failed\noidname .. ok\nfloat4 .. ok\nfloat8 .. failed\nnumerology .. ok\npoint .. ok\nlseg .. ok\nbox .. ok\npath .. ok\npolygon .. ok\ncircle .. ok\ngeometry .. failed\ntimespan .. failed\ndatetime .. failed\nreltime .. ok\nabstime .. ok\ntinterval .. ok\nhorology .. failed\ncomments .. ok\ncreate_function_1 .. ok\ncreate_type .. ok\ncreate_table .. ok\ncreate_function_2 .. ok\nconstraints .. ok\ntriggers .. failed\ncopy .. ok\ncreate_misc .. ok\ncreate_aggregate .. ok\ncreate_operator .. ok\ncreate_view .. ok\ncreate_index .. ok\nsanity_check .. ok\nerrors .. ok\nselect .. ok\nselect_into .. ok\nselect_distinct .. ok\nselect_distinct_on .. ok\naggregates .. ok\ntransactions .. ok\nrandom .. failed\nportals .. ok\nmisc .. ok\narrays .. ok\nbtree_index .. ok\nhash_index .. ok\nselect_views .. failed\nalter_table .. ok\nportals_p2 .. ok\n\n\n",
"msg_date": "Sun, 1 Feb 1998 17:16:45 -0700 (MST)",
"msg_from": "teunis <[email protected]>",
"msg_from_op": true,
"msg_subject": "(: JDBC+(Sun ~3:pm MST) CVS :) -also question about regression tests"
},
{
"msg_contents": "> \n> JDBC works\n> postgres works\n> \n> I is happy :)\n> [wasn't posted so I figure I'd post]\n> \n> Runs very nicely under egcs-2.91.06 FWIW...\n> [about 3-5 times faster than before ... though that could be postgres\n> improvements - and probably is]\n\nYou said 3-5 TIMES faster? Than 6.2.1, or last weeks source tree? Can\nyou give some figures. I am curious.\n\n> \n> platform : linux (I'm not posting kernel version! it doesn't matter!! :)\n> \tegcs-2.91.06\t(gcc-2.8.0 with haifa scheduler + other updates)\n> \tglibc-2.0.5c\tfrom RedHat-5.0 distrib - should be stable\n> \t\t\t[but with full crypt, locale]\n> \n> But : Here's output from regression tests:\n> Is there anything wrong with the failed tests? (is it known?)\n\nI get the same regression output. checkresults shows you the problems,\nand it mostly error message words or rounding.\n\n> \n> =============== Notes... =================\n> postmaster must already be running for the regression tests to succeed.\n> The time zone is now set to PST8PDT explicitly by this regression test\n> client frontend. Please report any apparent problems to\n> [email protected]\n> See regress/README for more information.\n> \n> =============== destroying old regression database... =================\n> ERROR: destroydb: database regression does not exist.\n> destroydb: database destroy failed on regression.\n> =============== creating new regression database... =================\n> =============== running regression queries... =================\n> boolean .. ok\n> char .. ok\n> char2 .. ok\n> char4 .. ok\n> char8 .. ok\n> char16 .. ok\n> varchar .. ok\n> text .. ok\n> strings .. failed\n> int2 .. failed\n> int4 .. failed\n> oid .. ok\n> oidint2 .. failed\n> oidint4 .. failed\n> oidname .. ok\n> float4 .. ok\n> float8 .. failed\n> numerology .. ok\n> point .. ok\n> lseg .. ok\n> box .. ok\n> path .. ok\n> polygon .. ok\n> circle .. ok\n> geometry .. failed\n> timespan .. failed\n> datetime .. failed\n> reltime .. ok\n> abstime .. ok\n> tinterval .. ok\n> horology .. failed\n> comments .. ok\n> create_function_1 .. ok\n> create_type .. ok\n> create_table .. ok\n> create_function_2 .. ok\n> constraints .. ok\n> triggers .. failed\n> copy .. ok\n> create_misc .. ok\n> create_aggregate .. ok\n> create_operator .. ok\n> create_view .. ok\n> create_index .. 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> aggregates .. ok\n> transactions .. ok\n> random .. failed\n> portals .. ok\n> misc .. ok\n> arrays .. ok\n> btree_index .. ok\n> hash_index .. ok\n> select_views .. failed\n> alter_table .. ok\n> portals_p2 .. ok\n> \n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sun, 1 Feb 1998 19:35:07 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "> > JDBC works\n> > postgres works\n> > platform : linux (I'm not posting kernel version! it doesn't matter!! :)\n> > egcs-2.91.06 (gcc-2.8.0 with haifa scheduler + other updates)\n> > glibc-2.0.5c from RedHat-5.0 distrib - should be stable\n> > [but with full crypt, locale]\n> >\n> > But : Here's output from regression tests:\n> > Is there anything wrong with the failed tests? (is it known?)\n>\n> I get the same regression output. checkresults shows you the problems,\n> and it mostly error message words or rounding.\n\nHmm. A linux box is used to generate the expected results, so we need to be\nmore careful here. I suspect that you have date/time trouble reported earlier\nby (Oliver?? can't find the e-mail, sorry). A few of the math functions in\nglibc2.0.x were misbehaving, leading to troubles like '3 hours 59 minutes 60\nseconds' rather than '4 hours' in timespan output.\n\nThat person submitted patches, but they were pretty specific to the glibc2\nproblems. Of course, I've already got some ugly code in there because Solaris\nhad some similar broken math, so perhaps we should figure out how to extract\nall of the busted code into the port-specific files?\n\n - Tom\n\n",
"msg_date": "Mon, 02 Feb 1998 01:54:12 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Mon, 2 Feb 1998, teunis wrote:\n\n> On Sun, 1 Feb 1998, Bruce Momjian wrote:\n> \n> > > \n> > > JDBC works\n> > > postgres works\n> > > \n> > > I is happy :)\n> > > [wasn't posted so I figure I'd post]\n> > > \n> > > Runs very nicely under egcs-2.91.06 FWIW...\n> > > [about 3-5 times faster than before ... though that could be postgres\n> > > improvements - and probably is]\n> > \n> > You said 3-5 TIMES faster? Than 6.2.1, or last weeks source tree? Can\n> > you give some figures. I am curious.\n> \n> unfortunately I can't give older times... I could make current times\n> though if I knew how. Postgres just _FLEW_ rebuilding the company\n> database *grin*.\n\n\tWell, that gives us something figurative to work with *grin* How\nbig is this company database? What do you mean by FLEW rebuilding?\nNumber of records? Disk space used up? Details man, details :)\n\n\n",
"msg_date": "Mon, 2 Feb 1998 11:53:00 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Sun, 1 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > JDBC works\n> > postgres works\n> > \n> > I is happy :)\n> > [wasn't posted so I figure I'd post]\n> > \n> > Runs very nicely under egcs-2.91.06 FWIW...\n> > [about 3-5 times faster than before ... though that could be postgres\n> > improvements - and probably is]\n> \n> You said 3-5 TIMES faster? Than 6.2.1, or last weeks source tree? Can\n> you give some figures. I am curious.\n\nunfortunately I can't give older times... I could make current times\nthough if I knew how. Postgres just _FLEW_ rebuilding the company\ndatabase *grin*.\nThis is Sunday CVS versus Friday CVS + egcs+haifa vs gcc-2.7.2.1\n\nsuspect it's an egcs thing mostly - though current postgres is faster than\n6.2.1.\n\nI've been following the CVS the entire time - except for about a 1 month\nblock ending last week :)\n\nI'm just happy it's still up, operational, and fast :)\n[egcs in the past usually crashed during the compile... and sometimes\nproduced unstable results. so I used gcc-2.7.2.1 as a rule. Just out of\ncuriousity I compiled using egcs+haifa which I'd just compiled+installed\nand postgres ran beautifully :]\n... No special command-line options though BTW.\nAnd I've been glibc-2 since last march.\n\nG'day, eh? :)\n\t- Teunis\n\n\n",
"msg_date": "Mon, 2 Feb 1998 09:54:59 -0700 (MST)",
"msg_from": "teunis <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Mon, 2 Feb 1998, Thomas G. Lockhart wrote:\n\n> > > JDBC works\n> > > postgres works\n> > > platform : linux (I'm not posting kernel version! it doesn't matter!! :)\n> > > egcs-2.91.06 (gcc-2.8.0 with haifa scheduler + other updates)\n> > > glibc-2.0.5c from RedHat-5.0 distrib - should be stable\n> > > [but with full crypt, locale]\n> > >\n> > > But : Here's output from regression tests:\n> > > Is there anything wrong with the failed tests? (is it known?)\n> >\n> > I get the same regression output. checkresults shows you the problems,\n> > and it mostly error message words or rounding.\n> \n> Hmm. A linux box is used to generate the expected results, so we need to be\n> more careful here. I suspect that you have date/time trouble reported earlier\n> by (Oliver?? can't find the e-mail, sorry). A few of the math functions in\n> glibc2.0.x were misbehaving, leading to troubles like '3 hours 59 minutes 60\n> seconds' rather than '4 hours' in timespan output.\n> \n> That person submitted patches, but they were pretty specific to the glibc2\n> problems. Of course, I've already got some ugly code in there because Solaris\n> had some similar broken math, so perhaps we should figure out how to extract\n> all of the busted code into the port-specific files?\n\nI'll say it again and again - glibc-2.0 is the _STANDARD_ (actually\nreference) platform for Unix. All Unix. Not just Linux.\nAdopted last year.\n\nIf there's any problems with glibc-2 - that is BAD for postgres. There\nHAVE been real bugs with glibc-2 : but they are very rare (thankfully :).\n\nIn other words - please fix it! :)\n[glibc-2 was adopted by Unix98 IIRC as the reference platform to base all\nlibc's on. Sure a nice change to have public/GNU software providing the\nspecs... :] (the last time was BSD's telnet/ftp varient - a LONG time ago \nIIRC)\n\nG'day, eh? :)\n\t- Teunis\n\n",
"msg_date": "Mon, 2 Feb 1998 09:59:10 -0700 (MST)",
"msg_from": "teunis <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Mon, 2 Feb 1998, teunis wrote:\n\n> On Mon, 2 Feb 1998, Thomas G. Lockhart wrote:\n> \n> > > > JDBC works\n> > > > postgres works\n> > > > platform : linux (I'm not posting kernel version! it doesn't matter!! :)\n> > > > egcs-2.91.06 (gcc-2.8.0 with haifa scheduler + other updates)\n> > > > glibc-2.0.5c from RedHat-5.0 distrib - should be stable\n> > > > [but with full crypt, locale]\n> > > >\n> > > > But : Here's output from regression tests:\n> > > > Is there anything wrong with the failed tests? (is it known?)\n> > >\n> > > I get the same regression output. checkresults shows you the problems,\n> > > and it mostly error message words or rounding.\n> > \n> > Hmm. A linux box is used to generate the expected results, so we need to be\n> > more careful here. I suspect that you have date/time trouble reported earlier\n> > by (Oliver?? can't find the e-mail, sorry). A few of the math functions in\n> > glibc2.0.x were misbehaving, leading to troubles like '3 hours 59 minutes 60\n> > seconds' rather than '4 hours' in timespan output.\n> > \n> > That person submitted patches, but they were pretty specific to the glibc2\n> > problems. Of course, I've already got some ugly code in there because Solaris\n> > had some similar broken math, so perhaps we should figure out how to extract\n> > all of the busted code into the port-specific files?\n> \n> I'll say it again and again - glibc-2.0 is the _STANDARD_ (actually\n> reference) platform for Unix. All Unix. Not just Linux.\n> Adopted last year.\n\n\tAnd...how many Unix (other then Linux) are *actually* using it?\nAny idea on how we can test whether it is being used or not?\n\n\tThe \"let's break all ports except Linux because the rest don't\nfollow a new standard\" argument just don't hold water for those not using\nLinux :)\n\n\n",
"msg_date": "Mon, 2 Feb 1998 11:59:54 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "> > > > JDBC works\n> > > > postgres works\n> > > > platform : linux (I'm not posting kernel version! it doesn't matter!! :)\n> > > > egcs-2.91.06 (gcc-2.8.0 with haifa scheduler + other updates)\n> > > > glibc-2.0.5c from RedHat-5.0 distrib - should be stable\n> > > > [but with full crypt, locale]\n\n> > Hmm. A linux box is used to generate the expected results, so we need to be\n> > more careful here. I suspect that you have date/time trouble reported earlier\n> > by (Oliver?? can't find the e-mail, sorry). A few of the math functions in\n> > glibc2.0.x were misbehaving, leading to troubles like '3 hours 59 minutes 60\n> > seconds' rather than '4 hours' in timespan output.\n> >\n> > That person submitted patches, but they were pretty specific to the glibc2\n> > problems. Of course, I've already got some ugly code in there because Solaris\n> > had some similar broken math, so perhaps we should figure out how to extract\n> > all of the busted code into the port-specific files?\n>\n> I'll say it again and again - glibc-2.0 is the _STANDARD_ (actually\n> reference) platform for Unix. All Unix. Not just Linux.\n> Adopted last year.\n>\n> If there's any problems with glibc-2 - that is BAD for postgres. There\n> HAVE been real bugs with glibc-2 : but they are very rare (thankfully :).\n\nGreat. Then when they get this bug out it will work on Postgres without changes.\nI'll wait. If you want to pursue it on the glibc2 side and give them a patch\nperhaps they will fix the bug faster.\n\nThe nature of the bug (if I recall; this has been going for a while now) is that\nthere are rounding behaviors in the math library which are at odds with other Unix\nsystems and with the *documented behaviors of the routines in question*. By my\nstandards, that counts as a bug. Happy to consider workarounds, but the patches\nshould be generic and testable, through configure, so that when glibc2 starts\nbehaving correctly there will be no need to hunt through the backend to find all\nthe kludge bug fixes.\n\n<Whew, I feel better now>\n\nThe other good possibility is for Oliver to develop a patch kit (he has done so\nalready) which we can include in the v6.3 distribution to be applied only to\nlinux/glibc2. When the beta settles down perhaps Oliver can generate a new patch\nkit which we can include?\n\nbtw, I'm running the same systems you are; RH5.0 is in the mail...\n\n - Tom\n\n> In other words - please fix it! :)\n> [glibc-2 was adopted by Unix98 IIRC as the reference platform to base all\n> libc's on. Sure a nice change to have public/GNU software providing the\n> specs... :] (the last time was BSD's telnet/ftp varient - a LONG time ago\n> IIRC)\n\n\n\n",
"msg_date": "Mon, 02 Feb 1998 17:42:08 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Mon, 2 Feb 1998, The Hermit Hacker wrote:\n\n> On Mon, 2 Feb 1998, teunis wrote:\n> \n> > > > Runs very nicely under egcs-2.91.06 FWIW...\n> > > > [about 3-5 times faster than before ... though that could be postgres\n> > > > improvements - and probably is]\n> > > \n> > > You said 3-5 TIMES faster? Than 6.2.1, or last weeks source tree? Can\n> > > you give some figures. I am curious.\n> > \n> > unfortunately I can't give older times... I could make current times\n> > though if I knew how. Postgres just _FLEW_ rebuilding the company\n> > database *grin*.\n> \n> \tWell, that gives us something figurative to work with *grin* How\n> big is this company database? What do you mean by FLEW rebuilding?\n> Number of records? Disk space used up? Details man, details :)\n\n*heh*\npsql simply flew past - it's not a very big database (just about no real\ndata - but about 40 complex tables)....\n\nI used to (last Friday, say :) be able to watch each table be created...\nnow it's almost instant. (~1/2 second per table creation before).\n\nDisk space? about 1M - not big. yet.\n\nOn the flip side, this database will be holding all of the company sales\nrecords / calls / customer accounting / etc.... as soon as the client\n(Java) software is stable *grin*...\n.. Also planning on selling the Java package - and doing installs of\nPostgres/Linux on each one *grin* when possible. But I'm also going to\nhave to be windows/some database compatible and I'm not looking foreward\nto that *deep sigh*.\n\nG'day, eh? :)\n\t- Teunis\n\n",
"msg_date": "Mon, 2 Feb 1998 12:23:16 -0700 (MST)",
"msg_from": "teunis <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "> > I'll say it again and again - glibc-2.0 is the _STANDARD_ (actually\n> > reference) platform for Unix. All Unix. Not just Linux.\n> > Adopted last year.\n> \n> \tAnd...how many Unix (other then Linux) are *actually* using it?\n> Any idea on how we can test whether it is being used or not?\n\nAFAIK some BSD's are now using it - but I am probably wrong. I gave up on\nBSD in '92.\n\n> \tThe \"let's break all ports except Linux because the rest don't\n> follow a new standard\" argument just don't hold water for those not using\n> Linux :)\n\nThere's no reason to break anything with this... This is a way of\ndetecting it (there are others IIRC).\n\n#if (__GLIBC__ >= 2)\n[glibc-2 stuff]\n#endif\n\nfor a complicated varient.\n\nMost of the minor nit-picks involve actually mean going into source and\nfixing it. (ie - not really port-related fixes).\n\nThat and #undef HAVE_INT_TIMEZONE in os/linux.h for glibc-2... as that's\nautodetected (and invalid). Detecting HAVE_SIGSETJMP would be nice to as\nit really IS a function in glibc (just remapped via a #define).\n\nAFAIK gnu's libc's are not used on just linux. So making it a linux-only\nthing would be almost as bad as not fixing it in the first place... *g*\n(not trying to be rude - personally I've found that the database works\nwell enough regardless of the regression tests :)\n\nG'day, eh? :)\n\t- Teunis\n\n",
"msg_date": "Mon, 2 Feb 1998 12:29:25 -0700 (MST)",
"msg_from": "teunis <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "> \n> Great. Then when they get this bug out it will work on Postgres without changes.\n> I'll wait. If you want to pursue it on the glibc2 side and give them a patch\n> perhaps they will fix the bug faster.\n\nOkay - I'll bite. Where's the bug?\nWhat part of the libs?\nWhat function?\nWhat behaviour?\n\nAnd you're right - the documentation rules where functions are defined.\n(And the documents are Unix98 - not any particular OS.... though BSD\nshould be a good reference for most :) [but not all]\n\nIf there's a real problem there's a real path to tell the people how to\nfix it :)\n[and I have full gcc/egcs/glibc/... source on my computer so can look it\nup...]\n\nG'day, eh? :)\n\t- Teunis\n\n\n",
"msg_date": "Mon, 2 Feb 1998 12:35:49 -0700 (MST)",
"msg_from": "teunis <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Mon, 2 Feb 1998, teunis wrote:\n\n> > > I'll say it again and again - glibc-2.0 is the _STANDARD_ (actually\n> > > reference) platform for Unix. All Unix. Not just Linux.\n> > > Adopted last year.\n> > \n> > \tAnd...how many Unix (other then Linux) are *actually* using it?\n> > Any idea on how we can test whether it is being used or not?\n> \n> AFAIK some BSD's are now using it - but I am probably wrong. I gave up on\n> BSD in '92.\n\n\tGeez, about the time I gave up on Linux and converted to *BSD\n*grin*\n\n> There's no reason to break anything with this... This is a way of\n> detecting it (there are others IIRC).\n> \n> #if (__GLIBC__ >= 2)\n> [glibc-2 stuff]\n> #endif\n> \n> for a complicated varient.\n\n\tI personally feel that this would be an acceptable way of fixing\nthe bugs...again, this is compiler defined, so its pretty transparent to\nthe 'end-user'...do you want to supply patches that does this? Thomas,\ndoes this work for you as well?\n\n> That and #undef HAVE_INT_TIMEZONE in os/linux.h for glibc-2... as that's\n> autodetected (and invalid). Detecting HAVE_SIGSETJMP would be nice to as\n> it really IS a function in glibc (just remapped via a #define).\n\n\tPatch for this? I'd rather see a patch of what you'd like to\nreview then to blindly go around and \"fix\" what I can't directly test. :)\n\n> AFAIK gnu's libc's are not used on just linux. So making it a linux-only\n> thing would be almost as bad as not fixing it in the first place... *g*\n> (not trying to be rude - personally I've found that the database works\n> well enough regardless of the regression tests :)\n\n\tIt would be worse, IMHO...just look at the 'wine' project...such\ngood potential, but they core developers feel that supporting Linux-only\nis the way to go...so they keep bringing in these really great features\n... that work *only* under Linux. Then, they release a new beta and find\nout that nobody else can use it anymore :(\n\n\tAt least 3 out of 4 of the core developers here are *BSD, so we\noffset the Linux-camp very well *grin* *wave to Thomas*\n\n\n",
"msg_date": "Mon, 2 Feb 1998 14:37:30 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Mon, 2 Feb 1998, The Hermit Hacker wrote:\n\n> On Mon, 2 Feb 1998, teunis wrote:\n> \n> > > > I'll say it again and again - glibc-2.0 is the _STANDARD_ (actually\n> > > > reference) platform for Unix. All Unix. Not just Linux.\n> > > > Adopted last year.\n> > > \n> > > \tAnd...how many Unix (other then Linux) are *actually* using it?\n> > > Any idea on how we can test whether it is being used or not?\n> > \n> > AFAIK some BSD's are now using it - but I am probably wrong. I gave up on\n> > BSD in '92.\n> \n> \tGeez, about the time I gave up on Linux and converted to *BSD\n> *grin*\n\nAwww - I only gave up on BSD 'cause it had no sound *grin*.... So I'm\nbiased. [and I've since grown to love Linux's many quirks so I'm never\ngoing back... but each to their own, yes? :]\n\n> > There's no reason to break anything with this... This is a way of\n> > detecting it (there are others IIRC).\n> > \n> > #if (__GLIBC__ >= 2)\n> > [glibc-2 stuff]\n> > #endif\n> > \n> > for a complicated varient.\n> \n> \tI personally feel that this would be an acceptable way of fixing\n> the bugs...again, this is compiler defined, so its pretty transparent to\n> the 'end-user'...do you want to supply patches that does this? Thomas,\n> does this work for you as well?\n\nI'll work on it (I've attached a modified os.h to the end - this is so it\nwill work at all). Don't know how much time I'll have though....\n\n[clipa]\n> > AFAIK gnu's libc's are not used on just linux. So making it a linux-only\n> > thing would be almost as bad as not fixing it in the first place... *g*\n> > (not trying to be rude - personally I've found that the database works\n> > well enough regardless of the regression tests :)\n> \n> \tIt would be worse, IMHO...just look at the 'wine' project...such\n> good potential, but they core developers feel that supporting Linux-only\n> is the way to go...so they keep bringing in these really great features\n> ... that work *only* under Linux. Then, they release a new beta and find\n> out that nobody else can use it anymore :(\n> \n> \tAt least 3 out of 4 of the core developers here are *BSD, so we\n> offset the Linux-camp very well *grin* *wave to Thomas*\n\n*grin* - as long as linux is supported at all I'm happy :)\nThere's no real special features there that would benefit anything other\nthan a really low-level system (eg. Wine, Dosemu) anyways...\n\n(aside: I also work on GGI (formerly linux-GGI) a [largely] system-neutral\ngraphics toolkit. Was originally modified kernel linux-only but now works\non SGI O/2, AIX, and others... under X, ascii (libaa), svgalib, and still\nsupporting the linux-native (KGI) drivers if under linux... :)\nThat crew is 100% linux-only but people had access to other hardware and\nported it.... soon people will try to port the kernel level to FreeBSD :)\n\n[linux.h - canna remember where it is... but here's an updated version.\nI'd prefer this stuff be in config.h or better handled by configure, but\nthat's life]\n\n/* __USE_POSIX, __USE_BSD, and __USE_BSD_SIGNAL used to be defined either\n here or with -D compile options, but __ macros should be set and used by C\n library macros, not Postgres code. __USE_POSIX is set by features.h,\n __USE_BSD is set by bsd/signal.h, and __USE_BSD_SIGNAL appears not to\n be used.\n*/\n#define JMP_BUF\n#define USE_POSIX_TIME\n#define USE_POSIX_SIGNALS\n#define NEED_I386_TAS_ASM\n#define HAS_TEST_AND_SET\n\n#if defined(PPC)\ntypedef unsigned int slock_t;\n#elif defined(__alpha__)\ntypedef long int slock_t;\n#else\ntypedef unsigned char slock_t;\n#endif\n\n#if (__GLIBC__ >= 2)\n#ifdef HAVE_INT_TIMEZONE\n#undef HAVE_INT_TIMEZONE\n#endif\n/* currently undefined as I ([email protected]) have not\n checked this yet */\n/* #define HAVE_SIGSETJMP 1 */\n#endif\n\n#if defined(PPC)\n#undef NEED_I386_TAS_ASM\n#undef HAVE_INT_TIMEZONE\n#endif\n\n#if defined(sparc)\n#undef NEED_I386_TAS_ASM\n#endif\n\n#if defined(__alpha__)\n#undef NEED_I386_TAS_ASM\n#endif\n\n",
"msg_date": "Mon, 2 Feb 1998 16:41:54 -0700 (MST)",
"msg_from": "teunis <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Mon, 2 Feb 1998, teunis wrote:\n\n> > \tGeez, about the time I gave up on Linux and converted to *BSD\n> > *grin*\n> \n> Awww - I only gave up on BSD 'cause it had no sound *grin*.... So I'm\n> biased. [and I've since grown to love Linux's many quirks so I'm never\n> going back... but each to their own, yes? :]\n\n\tAgreed...I'm just expected to maintain this anti-Linux\nattitude *grin*\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 2 Feb 1998 20:58:25 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "> > There's no reason to break anything with this... This is a way of\n> > detecting it (there are others IIRC).\n> >\n> > #if (__GLIBC__ >= 2)\n> > [glibc-2 stuff]\n> > #endif\n> >\n> > for a complicated variant.\n>\n> I personally feel that this would be an acceptable way of fixing\n> the bugs...again, this is compiler defined, so its pretty transparent to\n> the 'end-user'...do you want to supply patches that does this? Thomas,\n> does this work for you as well?\n\nThis is the nature of the patch Oliver developed which I'll bet works just\ngreat. However, the patch is on top of kludge code to get Solaris to work one\nor two releases ago. I'd rather back those changes out, and get #ifdef Solaris\nas well as #ifdef _GLIBC_ code blocks, or even better put these in the\nport-specific code.\n\nI'd also like to not have to watch this too closely for v6.3, since I should be\nswamped with docs, regression testing, and perhaps contributing to subselects.\n\nIf we do apply the glibc2 patches, then I'd like help from someone running\nSolaris to clean things up for v6.4.\n\n> That and #undef HAVE_INT_TIMEZONE in os/linux.h for glibc-2... as that's\n\n> > autodetected (and invalid). Detecting HAVE_SIGSETJMP would be nice to as\n> > it really IS a function in glibc (just remapped via a #define).\n\nSo, we kludge both HAVE_INT_TIMEZONE and HAVE_SIGSETJMP in config.h.in by\nbracketing them with #if (_GLIBC_ >= 2) ? I suspect that the RH5.0 distribution\nof Postgres has broken date and time behavior because HAVE_INT_TIMEZONE is\nwrong.\n\nThe HAVE_SIGSETJMP problem (which is present in RH4.x also since sigsetjmp() is\ndefined as a macro, which is explicitly disallowed by the configure test) can\nbe worked-around by checking for #ifdef sigsetjmp. btw, I've tried running\nRH4.2 compiled both ways and noticed no difference in behavior, but I'm not\ncertain what to look for.\n\nOh, I forgot; already fixed it because I was tired of the compiler warnings.\nHere is the code in bootstrap.c:\n\n/* The test for HAVE_SIGSETJMP fails on Linux 2.0.x because the test\n * explicitly disallows sigsetjmp being a #define, which is how it\n * is declared in Linux. So, to avoid compiler warnings about\n * sigsetjmp() being redefined, let's not redefine unless necessary.\n * - thomas 1997-12-27\n */\n\n#if !defined(HAVE_SIGSETJMP) && !defined(sigsetjmp)\nstatic jmp_buf Warn_restart;\n#define sigsetjmp(x,y) setjmp(x)\n#define siglongjmp longjmp\n\n#else\nstatic sigjmp_buf Warn_restart;\n#endif\n\n...\n\n> Patch for this? I'd rather see a patch of what you'd like to\n> review then to blindly go around and \"fix\" what I can't directly test. :)\n>\n> > AFAIK gnu's libc's are not used on just linux. So making it a linux-only\n> > thing would be almost as bad as not fixing it in the first place... *g*\n> > (not trying to be rude - personally I've found that the database works\n> > well enough regardless of the regression tests :)\n>\n> It would be worse, IMHO...just look at the 'wine' project...such\n> good potential, but they core developers feel that supporting Linux-only\n> is the way to go...so they keep bringing in these really great features\n> ... that work *only* under Linux. Then, they release a new beta and find\n> out that nobody else can use it anymore :(\n>\n> At least 3 out of 4 of the core developers here are *BSD, so we\n> offset the Linux-camp very well *grin* *wave to Thomas*\n\nSure, takes 3 BSD machines for an even fight *ducks and hits head on desk*\n\n - Tom\n\n",
"msg_date": "Tue, 03 Feb 1998 08:10:36 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Tue, 3 Feb 1998, Thomas G. Lockhart wrote:\n\n> I'd also like to not have to watch this too closely for v6.3, since I\n> should be swamped with docs, regression testing, and perhaps\n> contributing to subselects. \n\n\tOops, okay, sorry :( Agreed, let's look at this *after* v6.3 is\nreleased then\n\n> So, we kludge both HAVE_INT_TIMEZONE and HAVE_SIGSETJMP in config.h.in\n> by bracketing them with #if (_GLIBC_ >= 2) ? I suspect that the RH5.0\n> distribution of Postgres has broken date and time behavior because\n> HAVE_INT_TIMEZONE is wrong. \n\n\tI think that I can come up with \"cleaner\" tests for these in\nconfigure...\n\n> > At least 3 out of 4 of the core developers here are *BSD, so we\n> > offset the Linux-camp very well *grin* *wave to Thomas*\n> \n> Sure, takes 3 BSD machines for an even fight *ducks and hits head on desk*\n\n\tAs I said to a friend of mine the other day, I imagine that Linux\nconstitutes a good portion of our user base, especially now that RedHat\nincludes it as standard with their distributions *shrug* Its not\nMicrosloth, that's all that counts :)\n\n\n",
"msg_date": "Tue, 3 Feb 1998 07:26:42 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
}
] |
[
{
"msg_contents": "\nAfter over 3 months of intense development, the PostgreSQL Development\nGroup is proud to announce that the source code is no longer considered\nALPHA software. With a development freeze effective today (February 1st),\nwe now enter BETA test mode, with a release date set for March 1st.\n\nThere has been alot of work done on the various aspects of the software,\nincluding alot of new SQL related features.\n\nOne of the main priorities for the next month is to ensure that the ports\nwork, both as far as compiling and regression tests are concerned, which\nis where we need *everyones* help.\n\nFor those wishing to aid in debugging for your various operating system,\nyou can grab the source code at:\n\n\t\tftp.postgresql.org/pub/postgresql.snapshot.tar.gz\n\nThis gets regenerated every morning.\n\nOnce compiled and after having run the regression tests, there is now a\nfile created called 'regression.diffs', which is the diffs of all the\n'failed' tests. Please submit this file, via ftp, to\n\n\tftp.postgresql.org/pub/incoming/regression.<OS>-<Version>\n\nThis will be used to help debug various 'idiosyncracies'. Check the\nsrc/test/regress directory for a regression.<operating system> file to\ncompare against before submitting...see if there are any differences.\n\nFinally, we've also added in 'port specific' expected/*.out results, so\nthat we can take into consideration minor differences (but legit) between\nthe actually results an Operating System provides and that expected based\non our \"base system\" \n\nIncluded below is what has changed since v6.2.1 was released...\n\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n=======================================\nPostgreSQL 6.3 Sun Feb 1 14:57:30 EST 1998\n-------------------------------------------------------------\n\nA dump/restore is required for those wishing to migrate data from\nprevious releases of PostgreSQL.\n\nCHANGES IN THE 6.3 RELEASE\n--------------------------\n\nBug Fixes\n---------\nFix binary cursors broken by MOVE implementation(Vadim)\nFix for tcl library crash(Jan)\nFix for array handling, from Gerhard Hintermayer\nFix acl error, and remove duplicate pqtrace(Bruce)\nFix psql \\e for empty file(Bruce)\nFix for textcat on varchar() fields(Bruce)\nFix for DBT Sendproc (Zeugswetter Andres)\nFix vacuum analyze syntax problem(Bruce)\nFix for international identifiers(Tatsuo)\nFix aggregates on inherited tables(Bruce)\nFix substr() for out-of-bounds data\nFix for select 1=1 or 2=2, select 1=1 and 2=2, and select sum(2+2)(Bruce)\nFix notty output to show status result. -q option still turns it off(Bruce)\nFix for count(*), aggs with views and multiple tables and sum(3)(Bruce)\nFix cluster(Bruce)\nFix for PQtrace start/stop several times(Bruce)\nFix a variety of locking problems like newer lock waiters getting\n\tlock before older waiters, and having readlock people not share\n\tlocks if a writer is waiting for a lock, and waiting writers not\n\tgetting priority over waiting readers(Bruce)\nFix crashes in psql when executing queries from external files(James)\nFix problem with multiple order by columns, with the first one having\n\tNULL values(Jeroen)\nUse correct hash table support functions for float8 and int4(Thomas)\nRe-enable JOIN= option in CREATE OPERATOR statement (Thomas)\nChange precedence for boolean operators to match expected behavior(Thomas)\nGenerate elog(ERROR) on over-large integer(Bruce)\nAllow multiple-argument functions in constraint clauses(Thomas)\nCheck boolean input literals for 'true','false','yes','no','1','0'\n\tand throw elog(ERROR) if unrecognized(Thomas)\n\n\nEnhancements\n------------\nReal deadlock detection, no more timeouts(Bruce)\n<NOT DONE YET> Subselects with EXISTS, IN, ALL, ANY keywords (Vadim, Bruce, Thomas)\nAdd SQL92 \"constants\" CURRENT_DATE, CURRENT_TIME, CURRENT_TIMESTAMP, \n\tCURRENT_USER(Thomas)\nSpeedup by inlining some frequently-called functions\nModify constraint syntax to be SQL92-compliant(Thomas)\nImplement SQL92 PRIMARY KEY and UNIQUE clauses using indices(Thomas)\nRecognize SQL92 syntax for FOREIGN KEY. Throw elog notice(Thomas)\nAllow NOT NULL UNIQUE constraint clause (each allowed separately before)(Thomas)\nAllow Postgres-style casting (\"::\") of non-constants(Thomas)\nAdd support for SQL3 TRUE and FALSE boolean constants(Thomas)\nSupport SQL92 syntax for IS TRUE/IS FALSE/IS NOT TRUE/IS NOT FALSE(Thomas)\nAllow shorter strings for boolean literals (e.g. \"t\", \"tr\", \"tru\")(Thomas)\nAllow SQL92 delimited identifiers(Thomas)\nImplement SQL92 binary and hexadecimal string decoding (b'10' and x'1F')(Thomas)\nSupport SQL92 syntax for type coercion of literal strings\n\t(e.g. \"DATETIME 'now'\")(Thomas)\nAdd conversions for int2, int4, and OID types to and from text(Thomas)\nUse shared lock when building indices(Vadim)\nFree memory allocated for an user query inside transaction block after\n\tthis query is done, was turned off in <= 6.2.1(Vadim)\nNew SQL statement CREATE PROCEDURAL LANGUAGE(Jan)\nNew PostgreSQL Procedural Language (PL) backend interface(Jan)\nRename pg_dump -H option to -h(Bruce)\nAdd Java support for passwords, European dates(Peter)\nUse indices for LIKE and ~, !~ operations(Bruce)\nAdd hash functions for datetime and timespan(Thomas)\nTime Travel removed(Vadim, Bruce)\nAdd paging for \\d and \\z, and fix \\i(Bruce)\nAdd Unix domain socket support to backend and to frontend library(Goran)\nImplement CREATE DATABASE/WITH LOCATION and initlocation utility(Thomas)\nAllow more SQL92 and/or Postgres reserved words as column identifiers(Thomas)\nAugment support for SQL92 SET TIME ZONE...(Thomas)\nSET/SHOW/RESET TIME ZONE uses TZ backend environment variable(Thomas)\nImplement SET keyword = DEFAULT and SET TIME ZONE DEFAULT(Thomas)\nEnable SET TIME ZONE using TZ environment variable(Thomas)\nAdd PGDATESTYLE environment variable to frontend and backend initialization(Thomas)\nAdd PGTZ, PGCOSTHEAP, PGCOSTINDEX, PGRPLANS, PGGEQO\n\tfrontend library initialization environment variables(Thomas)\nRegression tests time zone automatically set with \"setenv PGTZ PST8PDT\"(Thomas)\nAdd pg_description table for info on tables, columns, operators, types, and\n\taggregates(Bruce)\nIncrease 16 char limit on system table/index names to 32 characters(Bruce)\nRename system indices(Bruce)\nAdd 'GERMAN' option to SET DATESTYLE(Thomas)\nDefine an \"ISO-style\" timespan output format with \"hh:mm:ss\" fields(Thomas)\nAllow fractional values for delta times (e.g. '2.5 days')(Thomas)\nValidate numeric input more carefully for delta times(Thomas)\nImplement day of year as possible input to date_part()(Thomas)\nDefine timespan_finite() and text_timespan() functions(Thomas)\nRemove archive stuff(Bruce)\nAllow for a pg_password authentication database that is separate from\n\tthe system password file(Todd)\nDump ACLs, GRANT, REVOKE permissions(Matt)\nDefine text, varchar, and bpchar string length functions(Thomas)\nFix Query handling for inheritance, and cost computations(Bruce)\nImplement CREATE TABLE/AS SELECT (alternative to SELECT/INTO)(Thomas)\nAllow NOT, IS NULL, IS NOT NULL in constraints(Thomas)\nImplement UNIONs for SELECT(Bruce)\nAdd UNION, GROUP, DISTINCT to INSERT(Bruce)\nvarchar() stores only necessary bytes on disk(Bruce)\nFix for BLOBs(Peter)\nMega-Patch for JDBC...see README_6.3 for list of changes(Peter)\nAllow installation data block size and max tuple size configuration(Darren)\nRemove unused \"option\" from PQconnectdb()\nNew LOCK command and lock manual page describing deadlocks(Bruce)\nAdd new psql \\da, \\dd, \\df, \\do, \\dS, and \\dT commands(Bruce)\nEnhance psql \\z to show sequences(Bruce)\nShow NOT NULL and DEFAULT in psql \\d table(Bruce)\nNew psql .psqlrc file startup(Andrew)\nModify sample startup script in contrib/linux to show syslog(Thomas)\nNew types for IP and MAC addresses in contrib/ip_and_mac(TomH)\nUnix system time conversions with date/time types in contrib/unixdate(Thomas)\nUpdate of contrib stuff(Massimo)\nAdd Unix socket support to DBD::Pg(Goran)\nNew python interface (PyGreSQL 2.0)(D'Arcy)\nNew frontend/backend protocol has a version number, network byte order(Phil)\nSecurity features in pg_hba.conf enhanced and documented, many cleanups(Phil)\nNew HTML and Postscript documentation(Thomas)\n\n\nSource Tree Changes\n-------------------\nAdd new html development tools, and flow chart in /tools/backend\nFix for SCO compiles\nStratus computer port \"Gillies, Robert\" <[email protected]>\nAdded support for shlib for BSD44_derived & i386_solaris\nMake configure more automated(Brook)\nAdd script to check regression test results\nBreak parser functions into smaller files, group together(Bruce)\nRename heap_create to heap_create_and_catalog, rename heap_creatr\n\tto heap_create()(Bruce)\nSparc/Linux patch for locking(TomS)\nRemove PORTNAME and reorganize port-specific stuff(Marc)\nAdd optimizer README file(Bruce)\nRemove some recursion in optimizer and clean up some code there(Bruce)\nFix for NetBSD locking(Henry)\nFix for libptcl make(Tatsuo)\nAIX patch(Darren)\nChange IS TRUE, IS FALSE, ... to expressions using \"=\" rather than\n\tfunction calls to istrue() or isfalse() to allow optimization(Thomas)\nVarious fixes NetBSD/Sparc related(TomH)\nAlpha linux locking(Travis,Ryan)\nChange elog(WARN) to elog(ERROR)(Bruce)\nFAQ for FreeBSD(Marc)\nBring in the PostODBC source tree as part of our standard distribution(Marc)\nA minor patch for HP/UX 10 vs 9(Stan)\nNew pg_attribute.atttypmod for type-specific info like varchar length(Bruce)\nUnixware patches(Billy)\nNew i386 'lock' for spin lock asm(Billy)\nSupport for multiplexed backends is removed\nStart an OpenBSD port\nStart an AUX port\nStart a Cygnus port\nAdd string functions to regression suite(Thomas)\nExpand a few function names formerly truncated to 16 characters(Thomas)\n\n\n\n",
"msg_date": "Sun, 1 Feb 1998 20:47:18 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "v6.3beta testing starting"
},
{
"msg_contents": "Hi,\n\ncompiling on sparc-solaris 2.5 with gcc.\n\ntas not defined error.\n\ngmake[3]: Entering directory `/amd/dolphin/export/database22/data-mining/postgr\ne\ns/pgsql/src/backend/storage/buffer'\ngcc -I../../../include -I../../../backend -I../../../backend/port/sparc_solari\ns\n -Wall -Wmissing-prototypes -I../.. -c buf_table.c -o buf_table.o\ngcc -I../../../include -I../../../backend -I../../../backend/port/sparc_solari\ns\n -Wall -Wmissing-prototypes -I../.. -c buf_init.c -o buf_init.o\n../../../include/storage/s_lock.h:123: warning: `tas' declared `static' but \nneve\nr defined\ngcc -I../../../include -I../../../backend -I../../../backend/port/sparc_solari\ns\n -Wall -Wmissing-prototypes -I../.. -c bufmgr.c -o bufmgr.o\n../../../include/storage/s_lock.h:123: warning: `tas' declared `static' but \nneve\nr defined\ngcc -I../../../include -I../../../backend -I../../../backend/port/sparc_solari\ns\n -Wall -Wmissing-prototypes -I../.. -c freelist.c -o freelist.o\ngcc -I../../../include -I../../../backend -I../../../backend/port/sparc_solari\ns\n -Wall -Wmissing-prototypes -I../.. -c localbuf.c -o localbuf.o\ngcc -I../../../include -I../../../backend -I../../../backend/port/sparc_solari\ns\n -Wall -Wmissing-prototypes -I../.. -c s_lock.c -o s_lock.o\n../../../include/storage/s_lock.h:123: warning: `tas' declared `static' but \nneve\nr defined\n\nlater the linking fails:\ngcc -o postgres access/SUBSYS.o bootstrap/SUBSYS.o catalog/SUBSYS.o \ncommands/SUB\nSYS.o executor/SUBSYS.o lib/SUBSYS.o libpq/SUBSYS.o main/SUBSYS.o \nnodes/SUBSYS.o\n optimizer/SUBSYS.o parser/SUBSYS.o port/SUBSYS.o postmaster/SUBSYS.o \nregex/SUBS\nYS.o rewrite/SUBSYS.o storage/SUBSYS.o tcop/SUBSYS.o utils/SUBSYS.o \n../utils/ver\nsion.o -lgen -lcrypt -lnsl -lsocket -ldl -lm -lreadline -ltermcap -lcurses\nUndefined first referenced\n symbol in file\ntas storage/SUBSYS.o\n\n--shiby\n\n\n",
"msg_date": "Mon, 02 Feb 1998 11:52:35 -0500",
"msg_from": "Shiby Thomas <[email protected]>",
"msg_from_op": false,
"msg_subject": "Compilation error on sparc-solaris"
},
{
"msg_contents": "Hi,\n\nIt looks like the compilation problem on sparc_solaris is due to some\ninconsistencies in the Makefiles. The -D$(PORTNAME) is not passed in \nCFLAGS. I made the foll. changes and it compiled fine.\n\nI am attaching the diffs that I made to the Makefiles. Please let me know\nif they are correct and if so fix them in the main source tree.\nsrc/Makefile.global.in :\n47,53d46\n< # Note that portname is defined here to be UNDEFINED to remind you\n< # to change it in Makefile.custom.\n< #\n< # make sure that you have no whitespaces after the PORTNAME setting\n< # or the makefiles can get confused\n< PORTNAME= @PORTNAME@\n< \n285,287d279\n< \n< # Globally pass PORTNAME\n< CFLAGS+= -D$(PORTNAME)\n\nsrc/backend/Makefile:\n71c71\n< $(MAKE) -C $(subst .dir,,$@) all PORTNAME=$(PORTNAME)\n---\n> $(MAKE) -C $(subst .dir,,$@) all \n107c107\n< for i in $(DIRS); do $(MAKE) -C $$i clean PORTNAME=$(PORTNAME); done\n---\n> for i in $(DIRS); do $(MAKE) -C $$i clean; done\n110c110\n< for i in $(DIRS); do $(MAKE) -C $$i $@ PORTNAME=$(PORTNAME); done\n---\n> for i in $(DIRS); do $(MAKE) -C $$i $@; done\n171c171\n< ./makeID $(PORTNAME)\n---\n> ./makeID \n\nsrc/backend/port/Makefile: (Here I think the fix should be in Makefile.in; I \njust edited the Makefile to try if it fixes problem)\n28c28\n< OBJS = $(PORTNAME)/SUBSYS.o dynloader.o inet_aton.o \n---\n> OBJS = dynloader.o inet_aton.o \n31c31\n< all: submake SUBSYS.o\n---\n> all: SUBSYS.o\n37,39d36\n< \n< submake:\n< $(MAKE) -C $(PORTNAME) SUBSYS.o\n\n--shiby\n\n\n\n",
"msg_date": "Mon, 02 Feb 1998 14:06:13 -0500",
"msg_from": "Shiby Thomas <[email protected]>",
"msg_from_op": false,
"msg_subject": "compilation on sparc_solaris"
},
{
"msg_contents": "On Mon, 2 Feb 1998, Shiby Thomas wrote:\n\n> Hi,\n> \n> It looks like the compilation problem on sparc_solaris is due to some\n> inconsistencies in the Makefiles. The -D$(PORTNAME) is not passed in \n> CFLAGS. I made the foll. changes and it compiled fine.\n\n\tI hate to ask, but what version is the port for? We already\nremoved all the PORTNAME stuff for v6.3 ;(\n\n\n\n> I am attaching the diffs that I made to the Makefiles. Please let me know\n> if they are correct and if so fix them in the main source tree.\n> src/Makefile.global.in :\n> 47,53d46\n> < # Note that portname is defined here to be UNDEFINED to remind you\n> < # to change it in Makefile.custom.\n> < #\n> < # make sure that you have no whitespaces after the PORTNAME setting\n> < # or the makefiles can get confused\n> < PORTNAME= @PORTNAME@\n> < \n> 285,287d279\n> < \n> < # Globally pass PORTNAME\n> < CFLAGS+= -D$(PORTNAME)\n> \n> src/backend/Makefile:\n> 71c71\n> < $(MAKE) -C $(subst .dir,,$@) all PORTNAME=$(PORTNAME)\n> ---\n> > $(MAKE) -C $(subst .dir,,$@) all \n> 107c107\n> < for i in $(DIRS); do $(MAKE) -C $$i clean PORTNAME=$(PORTNAME); done\n> ---\n> > for i in $(DIRS); do $(MAKE) -C $$i clean; done\n> 110c110\n> < for i in $(DIRS); do $(MAKE) -C $$i $@ PORTNAME=$(PORTNAME); done\n> ---\n> > for i in $(DIRS); do $(MAKE) -C $$i $@; done\n> 171c171\n> < ./makeID $(PORTNAME)\n> ---\n> > ./makeID \n> \n> src/backend/port/Makefile: (Here I think the fix should be in Makefile.in; I \n> just edited the Makefile to try if it fixes problem)\n> 28c28\n> < OBJS = $(PORTNAME)/SUBSYS.o dynloader.o inet_aton.o \n> ---\n> > OBJS = dynloader.o inet_aton.o \n> 31c31\n> < all: submake SUBSYS.o\n> ---\n> > all: SUBSYS.o\n> 37,39d36\n> < \n> < submake:\n> < $(MAKE) -C $(PORTNAME) SUBSYS.o\n> \n> --shiby\n> \n> \n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 2 Feb 1998 20:46:20 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: compilation on sparc_solaris"
},
{
"msg_contents": "=> On Mon, 2 Feb 1998, Shiby Thomas wrote:\n=> \n=> > Hi,\n=> > \n=> > It looks like the compilation problem on sparc_solaris is due to some\n=> > inconsistencies in the Makefiles. The -D$(PORTNAME) is not passed in \n=> > CFLAGS. I made the foll. changes and it compiled fine.\n=> \n=> \tI hate to ask, but what version is the port for? We already\n=> removed all the PORTNAME stuff for v6.3 ;(\n=> \nIts the latest snapshot on Feb 2. Without the PORTNAME, flag, it didn't compile\nwhen I tried. \ngcc -o postgres access/SUBSYS.o bootstrap/SUBSYS.o catalog/SUBSYS.o \ncommands/SUB\nSYS.o executor/SUBSYS.o lib/SUBSYS.o libpq/SUBSYS.o main/SUBSYS.o \nnodes/SUBSYS.o\n optimizer/SUBSYS.o parser/SUBSYS.o port/SUBSYS.o postmaster/SUBSYS.o \nregex/SUBS\nYS.o rewrite/SUBSYS.o storage/SUBSYS.o tcop/SUBSYS.o utils/SUBSYS.o \n../utils/ver\nsion.o -lgen -lcrypt -lnsl -lsocket -ldl -lm -lreadline -ltermcap -lcurses\nUndefined first referenced\n symbol in file\ntas storage/SUBSYS.o\nld: fatal: Symbol referencing errors. No output written to postgres\n\nIt also gave the foll. warning while compiling in the backend/storage dir\ngcc -I../../../include -I../../../backend -I../../../backend/port/sparc_solari\ns\n -Wall -Wmissing-prototypes -Dsparc_solaris -I../.. -c spin.c -o spin.o\n../../../include/storage/s_lock.h:123: warning: `tas' declared `static' but \nneve\nr defined\n\n--shiby\n\n\n",
"msg_date": "Mon, 02 Feb 1998 20:40:26 -0500",
"msg_from": "Shiby Thomas <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: compilation on sparc_solaris "
},
{
"msg_contents": "\nFixed..should be okay for tomorrow mornings snapshot\n\n\n\nOn Mon, 2 Feb 1998, Shiby Thomas wrote:\n\n> => On Mon, 2 Feb 1998, Shiby Thomas wrote:\n> => \n> => > Hi,\n> => > \n> => > It looks like the compilation problem on sparc_solaris is due to some\n> => > inconsistencies in the Makefiles. The -D$(PORTNAME) is not passed in \n> => > CFLAGS. I made the foll. changes and it compiled fine.\n> => \n> => \tI hate to ask, but what version is the port for? We already\n> => removed all the PORTNAME stuff for v6.3 ;(\n> => \n> Its the latest snapshot on Feb 2. Without the PORTNAME, flag, it didn't compile\n> when I tried. \n> gcc -o postgres access/SUBSYS.o bootstrap/SUBSYS.o catalog/SUBSYS.o \n> commands/SUB\n> SYS.o executor/SUBSYS.o lib/SUBSYS.o libpq/SUBSYS.o main/SUBSYS.o \n> nodes/SUBSYS.o\n> optimizer/SUBSYS.o parser/SUBSYS.o port/SUBSYS.o postmaster/SUBSYS.o \n> regex/SUBS\n> YS.o rewrite/SUBSYS.o storage/SUBSYS.o tcop/SUBSYS.o utils/SUBSYS.o \n> ../utils/ver\n> sion.o -lgen -lcrypt -lnsl -lsocket -ldl -lm -lreadline -ltermcap -lcurses\n> Undefined first referenced\n> symbol in file\n> tas storage/SUBSYS.o\n> ld: fatal: Symbol referencing errors. No output written to postgres\n> \n> It also gave the foll. warning while compiling in the backend/storage dir\n> gcc -I../../../include -I../../../backend -I../../../backend/port/sparc_solari\n> s\n> -Wall -Wmissing-prototypes -Dsparc_solaris -I../.. -c spin.c -o spin.o\n> ../../../include/storage/s_lock.h:123: warning: `tas' declared `static' but \n> neve\n> r defined\n> \n> --shiby\n> \n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 2 Feb 1998 22:06:30 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: compilation on sparc_solaris "
},
{
"msg_contents": "Hi,\n\nThe foll. query did not finish even after running for 20 HOURS. \nI remember some discussion about a self-join bug. Could it be because of that ?\nThis is with the snapshot on Feb 2. Foll. is the output of explain:\nHere data is a table with 2 attributes(tid, item) and about 1.1 million tuples.\nc2(item1, item2) has about 87,000 tuples. I am running on a 8 processor\nsun sparc with a total of 2GB memory and each processor is a\n248 MHz SUNW,UltraSPARC-II.\nI invoked postmaster as:\n postmaster -B 30000 -o \"-s -F -S 16384\"\nThe tables does not have any indices right now. Will it help with indices\nfor the hash join plan ?\n\nassoc=> explain select item1, item2, count(t1.tid) into table f2_temp from \ndata t1, data t2, c2 where t1.item = c2.item1 and t2.item = c2.item2 and \nt1.tid = t2.tid group by item1, item2;\nNOTICE: QUERY PLAN:\n\nAggregate (cost=211518.56 size=0 width=0)\n -> Group (cost=211518.56 size=0 width=0)\n -> Sort (cost=211518.56 size=0 width=0)\n -> Hash Join (cost=211518.56 size=14 width=24)\n -> Hash Join (cost=89881.04 size=1154369 width=16)\n -> Seq Scan on t2 (cost=45447.18 size=1154369 width=8)\n -> Hash (cost=0.00 size=0 width=0)\n -> Seq Scan on c2 (cost=3447.84 size=87571 width=8)\n -> Hash (cost=0.00 size=0 width=0)\n -> Seq Scan on t1 (cost=45447.18 size=1154369 width=8)\n\nEXPLAIN\n\n\n",
"msg_date": "Tue, 03 Feb 1998 10:58:32 -0500",
"msg_from": "Shiby Thomas <[email protected]>",
"msg_from_op": false,
"msg_subject": "Execution time."
},
{
"msg_contents": "Shiby Thomas wrote:\n> \n> Hi,\n> \n> The foll. query did not finish even after running for 20 HOURS.\n> I remember some discussion about a self-join bug. Could it be because of that ?\n\nNo. Self-join is problem of old optimizer, not executor. You got output\nfrom EXPLAIN ==> optimizer' phase is done.\n\n> This is with the snapshot on Feb 2. Foll. is the output of explain:\n> Here data is a table with 2 attributes(tid, item) and about 1.1 million tuples.\n> c2(item1, item2) has about 87,000 tuples. I am running on a 8 processor\n> sun sparc with a total of 2GB memory and each processor is a\n> 248 MHz SUNW,UltraSPARC-II.\n> I invoked postmaster as:\n> postmaster -B 30000 -o \"-s -F -S 16384\"\n ^^^^^\n240M of shared memory ?!\n\n> The tables does not have any indices right now. Will it help with indices\n> for the hash join plan ?\n ^^^^^^^^^^^^^^\nNo. But indices could be used in nestloop plan...\n\n> \n> assoc=> explain select item1, item2, count(t1.tid) into table f2_temp from\n> data t1, data t2, c2 where t1.item = c2.item1 and t2.item = c2.item2 and\n> t1.tid = t2.tid group by item1, item2;\n> NOTICE: QUERY PLAN:\n> \n> Aggregate (cost=211518.56 size=0 width=0)\n> -> Group (cost=211518.56 size=0 width=0)\n> -> Sort (cost=211518.56 size=0 width=0)\n> -> Hash Join (cost=211518.56 size=14 width=24)\n> -> Hash Join (cost=89881.04 size=1154369 width=16)\n> -> Seq Scan on t2 (cost=45447.18 size=1154369 width=8)\n> -> Hash (cost=0.00 size=0 width=0)\n> -> Seq Scan on c2 (cost=3447.84 size=87571 width=8)\n> -> Hash (cost=0.00 size=0 width=0)\n> -> Seq Scan on t1 (cost=45447.18 size=1154369 width=8)\n ^^^^^^^\nYour big table data will be entirely in memory (this is caused\nby -B 30000).\n\nOk. First, try to create indices on both tables.\nDoes it help ?\nEXPLAIN ?\n\nVadim\n",
"msg_date": "Wed, 04 Feb 1998 10:09:59 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Execution time."
},
{
"msg_contents": "\n=> > I invoked postmaster as:\n=> > postmaster -B 30000 -o \"-s -F -S 16384\"\n=> ^^^^^\n=> 240M of shared memory ?!\n=> \nIs the size of the hash bucket dependent on the number of buffers available ?\nIf so, with -B 30000, it might create huge hash buckets and hence the\nhash join could degenerate to a nested loops join. Is that possible -- just\nchecking. What could be good values for -B and -S. The largest table I am\njoining is about 60M\n\nThanks\n--shiby\n\n\n\n",
"msg_date": "Wed, 04 Feb 1998 11:26:11 -0500",
"msg_from": "Shiby Thomas <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Execution time. "
},
{
"msg_contents": "On Wed, 4 Feb 1998, Shiby Thomas wrote:\n\n> \n> => > I invoked postmaster as:\n> => > postmaster -B 30000 -o \"-s -F -S 16384\"\n> => ^^^^^\n> => 240M of shared memory ?!\n> => \n> Is the size of the hash bucket dependent on the number of buffers available ?\n> If so, with -B 30000, it might create huge hash buckets and hence the\n> hash join could degenerate to a nested loops join. Is that possible -- just\n> checking. What could be good values for -B and -S. The largest table I am\n> joining is about 60M\n\n\tMy production server runs with a -B 256, and a -S 10240\n\n\n",
"msg_date": "Wed, 4 Feb 1998 11:38:04 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Execution time. "
},
{
"msg_contents": "Hi,\n\nassoc=> explain insert into f2_temp select p.item, q.item, count(*) as cnt \nfrom t1 p, t1 q where p.tid = q.tid and p.item < q.item group by p.item, \nq.item;\nERROR: The field being grouped by must appear in the target list\n\nHowever, it works this way:\nassoc=> explain select p.item, q.item, count(*) as cnt into table f2_temp from \nt1 p, t1 q where p.tid = q.tid and p.item < q.item group by p.item, q.item;\nNOTICE: QUERY PLAN:\n\n--shiby\n\n\n",
"msg_date": "Wed, 04 Feb 1998 12:56:53 -0500",
"msg_from": "Shiby Thomas <[email protected]>",
"msg_from_op": false,
"msg_subject": "Is this a parser error ?"
},
{
"msg_contents": "\n=> > Is the size of the hash bucket dependent on the number of buffers available ?\n=> > If so, with -B 30000, it might create huge hash buckets and hence the\n=> > hash join could degenerate to a nested loops join. Is that possible -- just\n=> > checking. What could be good values for -B and -S. The largest table I am\n=> > joining is about 60M\n=> \n=> \tMy production server runs with a -B 256, and a -S 10240\n=> \n=> \nWith less buffers, it is giving \"hash table out of memory\" error.\nI ran with -B 512\n\nAGG :c=435416.5312 :s=0 :w=0 \n l: GROUP :c=435416.5312 :s=0 :w=0 \n l: SORT :c=435416.5312 :s=0 :w=0 \n l: HASHJOIN :c=435416.5312 :s=305715 :w=16 \n l: SEQSCAN :c=36107.7500 :s=917144 :w=8 ( t1 )\n r: HASH :c=0.0000 :s=0 :w=0 \n l: SEQSCAN :c=36107.7500 :s=917144 :w=8 ( t1 )\nERROR: hash table out of memory. Use -B parameter to increase buffers.\n\n\n",
"msg_date": "Wed, 04 Feb 1998 13:07:01 -0500",
"msg_from": "Shiby Thomas <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Execution time. "
},
{
"msg_contents": "Hi,\n\nI get the foll. error in a select statement:\n\nassoc=> select * from f1_temp where cnt > 1000;\nERROR: fmgr_info: function 0: cache lookup failed\n\nWithout the where clause its fine.\n\nThis is with today's (Feb 4) snapshot.\n\n--shiby\n\n\n\n",
"msg_date": "Wed, 04 Feb 1998 14:15:10 -0500",
"msg_from": "Shiby Thomas <[email protected]>",
"msg_from_op": false,
"msg_subject": "Error in select"
},
{
"msg_contents": "> \n> Hi,\n> \n> I get the foll. error in a select statement:\n> \n> assoc=> select * from f1_temp where cnt > 1000;\n> ERROR: fmgr_info: function 0: cache lookup failed\n> \n> Without the where clause its fine.\n> \n> This is with today's (Feb 4) snapshot.\n> \n\nYep, we know the cause, and will fix it soon. Sorry.\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 4 Feb 1998 14:35:03 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Error in select"
},
{
"msg_contents": "Shiby Thomas wrote:\n> \n> => > Is the size of the hash bucket dependent on the number of buffers available ?\n> => > If so, with -B 30000, it might create huge hash buckets and hence the\n> => > hash join could degenerate to a nested loops join. Is that possible -- just\n> => > checking. What could be good values for -B and -S. The largest table I am\n> => > joining is about 60M\n> =>\n> => My production server runs with a -B 256, and a -S 10240\n> =>\n> =>\n> With less buffers, it is giving \"hash table out of memory\" error.\n\nKnown bug...\n\n> I ran with -B 512\n> \n> AGG :c=435416.5312 :s=0 :w=0\n> l: GROUP :c=435416.5312 :s=0 :w=0\n> l: SORT :c=435416.5312 :s=0 :w=0\n> l: HASHJOIN :c=435416.5312 :s=305715 :w=16\n> l: SEQSCAN :c=36107.7500 :s=917144 :w=8 ( t1 )\n> r: HASH :c=0.0000 :s=0 :w=0\n> l: SEQSCAN :c=36107.7500 :s=917144 :w=8 ( t1 )\n> ERROR: hash table out of memory. Use -B parameter to increase buffers.\n\nTry to create indices...\n\nVadim\n",
"msg_date": "Thu, 05 Feb 1998 08:51:25 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Execution time."
},
{
"msg_contents": "> > I get the foll. error in a select statement:\n> >\n> > assoc=> select * from f1_temp where cnt > 1000;\n> > ERROR: fmgr_info: function 0: cache lookup failed\n> >\n> > Without the where clause its fine.\n> >\n> > This is with today's (Feb 4) snapshot.\n> >\n>\n> Yep, we know the cause, and will fix it soon. Sorry.\n\nTry doing a \"make clean; make install\" since possibly the problem is due\nto my changing a few keywords (got rid of two!) in the parser. It seems\nthis quite often requires a clean install since I don't know which\nplaces are affected and it affects more than just the parser directory.\n\nDon't need to reload the database though.\n\n - Tom\n\n",
"msg_date": "Thu, 05 Feb 1998 02:25:02 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Error in select"
}
] |
[
{
"msg_contents": "\nI have the variable block size working, but haven't changed\nthe varchar/char limits to vary with the chosen size. (Also\nnot positive that I've found everywhere that should be check\nthe block size).\n\nWithout the increased char sizes, there really isn't too much\nbenefit to changing the disk block size.\n\nI'll do that tomorrow evening though. Damn bean-counters here\nat work turn the thermostats back to 55F on the weekends. Not\nbad in the day as my cubicle faces east, but come 5:30 and the\njoint freezes up again.\n\nAlso grep'd around the src for increasing the packet size to\n32768 (or _32769_ for _libpq_).\n\nWhy oh why does libpq have a max size of 8193? Is the extra 1\na hack? Drove me nuts trying to figure out where the max packet\nsize begin set at 8k...finally grep'd for 819 and there it was.\n\nLooks like this variable size stuff will be the first new feature\nfor 6.4 then, eh? :)\n\nTime to split, my hands are almost too cold to type any more...\n\ndarrenk\n",
"msg_date": "Sun, 1 Feb 1998 20:21:27 -0500",
"msg_from": "[email protected] (Darren King)",
"msg_from_op": true,
"msg_subject": "Variable block size..."
},
{
"msg_contents": "On Sun, 1 Feb 1998, Darren King wrote:\n\n> Looks like this variable size stuff will be the first new feature\n> for 6.4 then, eh? :)\n\n\tWhat is the chance of having this ready by Friday? Still would\ngive us 3weeks of debugging...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sun, 1 Feb 1998 21:58:08 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Variable block size..."
},
{
"msg_contents": "> > Looks like this variable size stuff will be the first new feature\n> > for 6.4 then, eh? :)\n>\n> What is the chance of having this ready by Friday? Still would\n> give us 3weeks of debugging...\n\nWell, it's OK to leave _one_ feature for v6.4. Bruce is probably tired of typing\nthe feature list :)\n\nAlso, we still have a few loose ends for v6.3. But, I'll bet Darren's stuff can\ngo in and not affect anything if we run using the current block size.\n\n - Tom\n\n",
"msg_date": "Mon, 02 Feb 1998 02:40:12 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Variable block size..."
},
{
"msg_contents": "> \n> > > Looks like this variable size stuff will be the first new feature\n> > > for 6.4 then, eh? :)\n> >\n> > What is the chance of having this ready by Friday? Still would\n> > give us 3weeks of debugging...\n> \n> Well, it's OK to leave _one_ feature for v6.4. Bruce is probably tired of typing\n> the feature list :)\n\nAlready on the HISTORY, because some of the changes were already made:\n\n Allow installation data block size and max tuple size configuration(Darren)\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 11:45:06 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Variable block size..."
}
] |
[
{
"msg_contents": "\n*** expected/float8.out Wed Jan 28 22:05:29 1998\n--- results/float8.out Sun Feb 1 21:30:39 1998\n***************\n*** 9,17 ****\n QUERY: INSERT INTO FLOAT8_TBL(f1) VALUES ('-10e400');\n ERROR: Bad float8 input format '-10e400'\n QUERY: INSERT INTO FLOAT8_TBL(f1) VALUES ('10e-400');\n- ERROR: Bad float8 input format '10e-400'\n QUERY: INSERT INTO FLOAT8_TBL(f1) VALUES ('-10e-400');\n- ERROR: Bad float8 input format '-10e-400'\n QUERY: SELECT '' AS five, FLOAT8_TBL.*;\n five|f1 \n ----+--------------------\n\nIf I try this manually, teh 10e-400 and -10e-400 produce a value of\n\"0\"...are we considering this to be correct? Or is this a bug? \n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sun, 1 Feb 1998 22:17:31 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "float8 regression test"
},
{
"msg_contents": "The Hermit Hacker wrote:\n\n> *** expected/float8.out Wed Jan 28 22:05:29 1998\n> --- results/float8.out Sun Feb 1 21:30:39 1998\n> ***************\n> *** 9,17 ****\n> QUERY: INSERT INTO FLOAT8_TBL(f1) VALUES ('-10e400');\n> ERROR: Bad float8 input format '-10e400'\n> QUERY: INSERT INTO FLOAT8_TBL(f1) VALUES ('10e-400');\n> - ERROR: Bad float8 input format '10e-400'\n> QUERY: INSERT INTO FLOAT8_TBL(f1) VALUES ('-10e-400');\n> - ERROR: Bad float8 input format '-10e-400'\n> QUERY: SELECT '' AS five, FLOAT8_TBL.*;\n> five|f1\n> ----+--------------------\n>\n> If I try this manually, teh 10e-400 and -10e-400 produce a value of\n> \"0\"...are we considering this to be correct? Or is this a bug?\n\nBug. On my 980112 snapshot, I get\n\npostgres=> select '-10e400'::float8;\nERROR: Bad float8 input format '-10e400'\n\nHope it hasn't changed.\n\n - Tom\n\n",
"msg_date": "Mon, 02 Feb 1998 05:48:46 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] float8 regression test"
}
] |
[
{
"msg_contents": "\ngeometry regression test:\n\n\nQUERY: SELECT '' AS thirty, p.f1, l.s, p.f1 ## l.s AS closest\n FROM LSEG_TBL l, POINT_TBL p;\nthirty|f1 |s |closest\n\n------+----------+-----------------------------+--------------------------------\n----\n |(0,0) |[(1,2),(3,4)] |(1,2)\n\n |(-10,0) |[(1,2),(3,4)] |(1,2)\n\n |(-3,4) |[(1,2),(3,4)] |(1,2)\n\n |(5.1,34.5)|[(1,2),(3,4)] |(3,4)\n\n |(-5,-12) |[(1,2),(3,4)] |(1,2)\n\n |(10,10) |[(1,2),(3,4)] |(3,4)\n\n |(0,0) |[(0,0),(6,6)] |(-0,0)\n\nThis last line, in the current results, provides a value of (0,0) for\nclosest, but I can't find what ## means in order to determine if this is\nas expected...\n\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sun, 1 Feb 1998 22:24:04 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "What is ##?"
},
{
"msg_contents": "> geometry regression test:\n>\n> QUERY: SELECT '' AS thirty, p.f1, l.s, p.f1 ## l.s AS closest\n> FROM LSEG_TBL l, POINT_TBL p;\n> thirty|f1 |s |closest\n>\n> ------+----------+-----------------------------+--------------------------------\n> ----\n> |(0,0) |[(1,2),(3,4)] |(1,2)\n>\n> |(-10,0) |[(1,2),(3,4)] |(1,2)\n>\n> |(-3,4) |[(1,2),(3,4)] |(1,2)\n>\n> |(5.1,34.5)|[(1,2),(3,4)] |(3,4)\n>\n> |(-5,-12) |[(1,2),(3,4)] |(1,2)\n>\n> |(10,10) |[(1,2),(3,4)] |(3,4)\n>\n> |(0,0) |[(0,0),(6,6)] |(-0,0)\n>\n> This last line, in the current results, provides a value of (0,0) for\n> closest, but I can't find what ## means in order to determine if this is\n> as expected...\n\nIt is the \"closest point on\" operator. I've been working on the geometric operators\ntoday, and will submit some fix-up patches soon (hopefully by Tuesday). Will\naugment the regression test for geometry also. (I've been adding entries to the\n\"DESCR\" macros in pg_proc.h and pg_operator.h, so there will be a description\navailable.)\n\nI assume that the \"-0\" is just a rounding artifact on your machine.\n\nThere were two regression tests whose CVS expected results should probably be\nbacked out. I was getting different results and had convinced myself that they were\nreasonable, but a week or two later someone fixed the backend problem and the\nbehavior reverted to the original. Look for the last two cvs updates by thomas in\nthe expected subdirectory.\n\n - Tom\n\n",
"msg_date": "Mon, 02 Feb 1998 05:56:43 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] What is ##?"
}
] |
[
{
"msg_contents": "\nHere is what I've been able to get so far. I don't see anything obvious.\nBut then, I don't really know what I'm looking for. One thing I do see\nis that \"pg_statistic\" shows up in both situations (\"\\d city\" and\n\"\\dS\"). \n\nIf anyone can make sense of this and point me in the right direction...\n\n-James\n\nBTW: This is the 1-31 cvs tree\n\n\n#0 0x400a9369 in __kill ()\n#1 0x810a0a6 in elog (lev=-1, \n fmt=0x813f1c8 \"fmgr_info: function %d: cache lookup failed\\n\")\n at elog.c:180\n#2 0x810ac89 in fmgr_info (procedureId=0, finfo=0xbfffd5f4) at fmgr.c:179\n#3 0x810ae0d in fmgr (procedureId=0) at fmgr.c:272\n#4 0x806c080 in heapgettup (relation=0x81c6be8, tid=0x0, dir=1, b=0x827250c, \n seeself=0 '\\000', nkeys=3, key=0x8272550) at heapam.c:422\n\n\nThe code turns into a string of macros after this and I have problems\ntracing it. I dumped as much information as I could from this frame... \n\n\n\n(gdb) i arg\nrelation = 0x81c6be8\ntid = 0x0\ndir = 1\nb = (Buffer *) 0x827250c\nseeself = 0 '\\000'\nnkeys = 3\nkey = 0x8272550\n\n\n(gdb) i loc\n__isnull = 0 '\\000'\n__atp = 0\n__test = 0\n__cur_nkeys = 2\n__cur_keys = 0x8272550\n_tuple = 0x4017ac78\n_res = 1 '\\001'\ntid = 0x0\nseeself = 0 '\\000'\nlpp = 0x40178cd0\ndp = 0x40178cc8 \"4\"\npage = 0\npages = 1\nlines = 1075293304\nrtup = 0x0\nlineoff = 0\nlinesleft = 10\n\n(gdb) p *__cur_keys\n$165 = {sk_flags = 0, sk_attno = 1, sk_procedure = 184, sk_func = {\n fn_addr = 0, fn_plhandler = 0, fn_oid = 0, fn_nargs = 0}, sk_nargs =\n0, sk_argument = 1249}\n\n(gdb) p *_tuple\n$166 = {t_len = 76, t_oid = 39712, t_cmin = 0, t_cmax = 0, t_xmin = \n23647, t_xmax = 0, t_ctid = {ip_blkid = {bi_hi = 0, bi_lo = 0}, ip_posid =\n1}, t_natts = 5, t_infomask = 2050, t_hoff = 40 '(', t_bits =\n\"\\000\\000\\000\"}\n\n\n(gdb) p *lpp\n$167 = {lp_off = 8112, lp_flags = 1, lp_len = 76}\n\n\n(gdb) p *relation\n$159 = {rd_fd = 16, rd_nblocks = 1, rd_refcnt = 44, rd_islocal = 0 '\\000', \n rd_isnailed = 0 '\\000', rd_istemp = 0 '\\000', rd_tmpunlinked = 0 '\\000', \n rd_am = 0x0, rd_rel = 0x81c6ae8, rd_id = 16577, lockInfo = 0x81c6d40 \"\\001\", \n rd_att = 0x81c6b40, rd_rules = 0x0, rd_istrat = 0x0, rd_support = 0x0, \n trigdesc = 0x0}\n\n\n(gdb) p *relation.rd_rel\n$160 = {relname = {data = \"pg_statistic\", '\\000' <repeats 19 times>}, \n reltype = 0, relowner = 406, relam = 0, relpages = 0, reltuples = 0, \n relhasindex = 0 '\\000', relisshared = 0 '\\000', relkind = 114 'r', \n relnatts = 5, relchecks = 0, reltriggers = 0, relhasrules = 0 '\\000', \n relacl = {2}}\n\n\n(gdb) p *relation.rd_att\n$161 = {natts = 5, attrs = 0x81c6a68, constr = 0x0}\n\n\n(gdb) p **relation.rd_att.attrs\n$163 = {attrelid = 16577, attname = {\n data = \"starelid\", '\\000' <repeats 23 times>}, atttypid = 26, \n attdisbursion = 0, attlen = 4, attnum = 1, attnelems = 0, attcacheoff = 0, \n atttypmod = 0, attbyval = 1 '\\001', attisset = 0 '\\000', \n attalign = 0 '\\000', attnotnull = 0 '\\000', atthasdef = 0 '\\000'}\n\n\n(gdb) p *key\n$164 = {sk_flags = 0, sk_attno = 1, sk_procedure = 184, sk_func = {\n fn_addr = 0, fn_plhandler = 0, fn_oid = 0, fn_nargs = 0}, sk_nargs = 0, \n sk_argument = 1249}\n\n\n\n#5 0x806c841 in heap_getnext (scandesc=0x82724f8, backw=0, b=0x0)\n at heapam.c:968\n#6 0x8101285 in gethilokey (relid=1249, attnum=6, opid=520, high=0xbfffd720, \n low=0xbfffd724) at selfuncs.c:353\n#7 0x8100eee in intltsel (opid=520, relid=1249, attno=6, value=0, flag=3)\n at selfuncs.c:113\n#8 0x8101030 in intgtsel (opid=520, relid=1249, attno=6, value=0, flag=3)\n at selfuncs.c:176\n#9 0x810ab38 in fmgr_c (finfo=0xbfffd798, values=0xbfffd7a8, \n isNull=0xbfffd797 \"\") at fmgr.c:115\n#10 0x810ae53 in fmgr (procedureId=104) at fmgr.c:285\n#11 0x80bbe38 in restriction_selectivity (functionObjectId=104, \n operatorObjectId=520, relationObjectId=1249, attributeNumber=6, \n constValue=0x0, constFlag=3) at plancat.c:344\n#12 0x80b37e9 in compute_selec (root=0x826b980, clauses=0x8272478, \n or_selectivities=0x0) at clausesel.c:319\n#13 0x80b36ab in compute_clause_selec (root=0x826b980, clause=0x826fef8, \n or_selectivities=0x0) at clausesel.c:199\n#14 0x80b7ddc in add_clause_to_rels (root=0x826b980, clause=0x826fef8)\n at initsplan.c:214\n#15 0x80b7d2a in initialize_base_rels_jinfo (root=0x826b980, clauses=0x82700b0)\n at initsplan.c:154\n#16 0x80b82cb in subplanner (root=0x826b980, flat_tlist=0x8270190, \n qual=0x82700b0) at planmain.c:249\n#17 0x80b820b in query_planner (root=0x826b980, command_type=1, \n tlist=0x826dd90, qual=0x826fe98) at planmain.c:168\n#18 0x80b8842 in planner (parse=0x826b980) at planner.c:110\n#19 0x80e5db4 in pg_parse_and_plan (\n query_string=0xbfffda08 \"SELECT a.attnum, a.attname, t.typname, a.attlen, a.atttypmod, a.attnotnull, a.atthasdef FROM pg_class c, pg_attribute a, pg_type t WHERE c.relname = 'city' and a.attnum > 0 and a.attrelid = c.o\"..., \n typev=0x0, nargs=0, queryListP=0xbfffd9a8, dest=Remote) at postgres.c:534\n#20 0x80e5eb2 in pg_exec_query_dest (\n query_string=0xbfffda08 \"SELECT a.attnum, a.attname, t.typname, a.attlen, a.atttypmod, a.attnotnull, a.atthasdef FROM pg_class c, pg_attribute a, pg_type t WHERE c.relname = 'city' and a.attnum > 0 and a.attrelid = c.o\"..., \n argv=0x0, typev=0x0, nargs=0, dest=Remote) at postgres.c:615\n#21 0x80e5e80 in pg_exec_query (\n query_string=0xbfffda08 \"SELECT a.attnum, a.attname, t.typname, a.attlen, a.atttypmod, a.attnotnull, a.atthasdef FROM pg_class c, pg_attribute a, pg_type t WHERE c.relname = 'city' and a.attnum > 0 and a.attrelid = c.o\"..., \n argv=0x0, typev=0x0, nargs=0) at postgres.c:597\n#22 0x80e6c81 in PostgresMain (argc=9, argv=0xbffffa48) at postgres.c:1373\n#23 0x80a91ed in main (argc=9, argv=0xbffffa48) at main.c:79\n#24 0x806014e in _start ()\n\n\n",
"msg_date": "Sun, 1 Feb 1998 23:54:21 -0500 (EST)",
"msg_from": "James Hughes <[email protected]>",
"msg_from_op": true,
"msg_subject": "VACUUM ANALYZE Problem (debugging output)"
},
{
"msg_contents": "I assume you have run initdb after doing the sup. Prior to the beta\nrelease today, things could have changed in the database format.\n\n> \n> \n> Here is what I've been able to get so far. I don't see anything obvious.\n> But then, I don't really know what I'm looking for. One thing I do see\n> is that \"pg_statistic\" shows up in both situations (\"\\d city\" and\n> \"\\dS\"). \n> \n> If anyone can make sense of this and point me in the right direction...\n> \n> -James\n> \n> BTW: This is the 1-31 cvs tree\n> \n> \n> #0 0x400a9369 in __kill ()\n> #1 0x810a0a6 in elog (lev=-1, \n> fmt=0x813f1c8 \"fmgr_info: function %d: cache lookup failed\\n\")\n> at elog.c:180\n> #2 0x810ac89 in fmgr_info (procedureId=0, finfo=0xbfffd5f4) at fmgr.c:179\n> #3 0x810ae0d in fmgr (procedureId=0) at fmgr.c:272\n> #4 0x806c080 in heapgettup (relation=0x81c6be8, tid=0x0, dir=1, b=0x827250c, \n> seeself=0 '\\000', nkeys=3, key=0x8272550) at heapam.c:422\n> \n> \n> The code turns into a string of macros after this and I have problems\n> tracing it. I dumped as much information as I could from this frame... \n> \n> \n> \n> (gdb) i arg\n> relation = 0x81c6be8\n> tid = 0x0\n> dir = 1\n> b = (Buffer *) 0x827250c\n> seeself = 0 '\\000'\n> nkeys = 3\n> key = 0x8272550\n> \n> \n> (gdb) i loc\n> __isnull = 0 '\\000'\n> __atp = 0\n> __test = 0\n> __cur_nkeys = 2\n> __cur_keys = 0x8272550\n> _tuple = 0x4017ac78\n> _res = 1 '\\001'\n> tid = 0x0\n> seeself = 0 '\\000'\n> lpp = 0x40178cd0\n> dp = 0x40178cc8 \"4\"\n> page = 0\n> pages = 1\n> lines = 1075293304\n> rtup = 0x0\n> lineoff = 0\n> linesleft = 10\n> \n> (gdb) p *__cur_keys\n> $165 = {sk_flags = 0, sk_attno = 1, sk_procedure = 184, sk_func = {\n> fn_addr = 0, fn_plhandler = 0, fn_oid = 0, fn_nargs = 0}, sk_nargs =\n> 0, sk_argument = 1249}\n> \n> (gdb) p *_tuple\n> $166 = {t_len = 76, t_oid = 39712, t_cmin = 0, t_cmax = 0, t_xmin = \n> 23647, t_xmax = 0, t_ctid = {ip_blkid = {bi_hi = 0, bi_lo = 0}, ip_posid =\n> 1}, t_natts = 5, t_infomask = 2050, t_hoff = 40 '(', t_bits =\n> \"\\000\\000\\000\"}\n> \n> \n> (gdb) p *lpp\n> $167 = {lp_off = 8112, lp_flags = 1, lp_len = 76}\n> \n> \n> (gdb) p *relation\n> $159 = {rd_fd = 16, rd_nblocks = 1, rd_refcnt = 44, rd_islocal = 0 '\\000', \n> rd_isnailed = 0 '\\000', rd_istemp = 0 '\\000', rd_tmpunlinked = 0 '\\000', \n> rd_am = 0x0, rd_rel = 0x81c6ae8, rd_id = 16577, lockInfo = 0x81c6d40 \"\\001\", \n> rd_att = 0x81c6b40, rd_rules = 0x0, rd_istrat = 0x0, rd_support = 0x0, \n> trigdesc = 0x0}\n> \n> \n> (gdb) p *relation.rd_rel\n> $160 = {relname = {data = \"pg_statistic\", '\\000' <repeats 19 times>}, \n> reltype = 0, relowner = 406, relam = 0, relpages = 0, reltuples = 0, \n> relhasindex = 0 '\\000', relisshared = 0 '\\000', relkind = 114 'r', \n> relnatts = 5, relchecks = 0, reltriggers = 0, relhasrules = 0 '\\000', \n> relacl = {2}}\n> \n> \n> (gdb) p *relation.rd_att\n> $161 = {natts = 5, attrs = 0x81c6a68, constr = 0x0}\n> \n> \n> (gdb) p **relation.rd_att.attrs\n> $163 = {attrelid = 16577, attname = {\n> data = \"starelid\", '\\000' <repeats 23 times>}, atttypid = 26, \n> attdisbursion = 0, attlen = 4, attnum = 1, attnelems = 0, attcacheoff = 0, \n> atttypmod = 0, attbyval = 1 '\\001', attisset = 0 '\\000', \n> attalign = 0 '\\000', attnotnull = 0 '\\000', atthasdef = 0 '\\000'}\n> \n> \n> (gdb) p *key\n> $164 = {sk_flags = 0, sk_attno = 1, sk_procedure = 184, sk_func = {\n> fn_addr = 0, fn_plhandler = 0, fn_oid = 0, fn_nargs = 0}, sk_nargs = 0, \n> sk_argument = 1249}\n> \n> \n> \n> #5 0x806c841 in heap_getnext (scandesc=0x82724f8, backw=0, b=0x0)\n> at heapam.c:968\n> #6 0x8101285 in gethilokey (relid=1249, attnum=6, opid=520, high=0xbfffd720, \n> low=0xbfffd724) at selfuncs.c:353\n> #7 0x8100eee in intltsel (opid=520, relid=1249, attno=6, value=0, flag=3)\n> at selfuncs.c:113\n> #8 0x8101030 in intgtsel (opid=520, relid=1249, attno=6, value=0, flag=3)\n> at selfuncs.c:176\n> #9 0x810ab38 in fmgr_c (finfo=0xbfffd798, values=0xbfffd7a8, \n> isNull=0xbfffd797 \"\") at fmgr.c:115\n> #10 0x810ae53 in fmgr (procedureId=104) at fmgr.c:285\n> #11 0x80bbe38 in restriction_selectivity (functionObjectId=104, \n> operatorObjectId=520, relationObjectId=1249, attributeNumber=6, \n> constValue=0x0, constFlag=3) at plancat.c:344\n> #12 0x80b37e9 in compute_selec (root=0x826b980, clauses=0x8272478, \n> or_selectivities=0x0) at clausesel.c:319\n> #13 0x80b36ab in compute_clause_selec (root=0x826b980, clause=0x826fef8, \n> or_selectivities=0x0) at clausesel.c:199\n> #14 0x80b7ddc in add_clause_to_rels (root=0x826b980, clause=0x826fef8)\n> at initsplan.c:214\n> #15 0x80b7d2a in initialize_base_rels_jinfo (root=0x826b980, clauses=0x82700b0)\n> at initsplan.c:154\n> #16 0x80b82cb in subplanner (root=0x826b980, flat_tlist=0x8270190, \n> qual=0x82700b0) at planmain.c:249\n> #17 0x80b820b in query_planner (root=0x826b980, command_type=1, \n> tlist=0x826dd90, qual=0x826fe98) at planmain.c:168\n> #18 0x80b8842 in planner (parse=0x826b980) at planner.c:110\n> #19 0x80e5db4 in pg_parse_and_plan (\n> query_string=0xbfffda08 \"SELECT a.attnum, a.attname, t.typname, a.attlen, a.atttypmod, a.attnotnull, a.atthasdef FROM pg_class c, pg_attribute a, pg_type t WHERE c.relname = 'city' and a.attnum > 0 and a.attrelid = c.o\"..., \n> typev=0x0, nargs=0, queryListP=0xbfffd9a8, dest=Remote) at postgres.c:534\n> #20 0x80e5eb2 in pg_exec_query_dest (\n> query_string=0xbfffda08 \"SELECT a.attnum, a.attname, t.typname, a.attlen, a.atttypmod, a.attnotnull, a.atthasdef FROM pg_class c, pg_attribute a, pg_type t WHERE c.relname = 'city' and a.attnum > 0 and a.attrelid = c.o\"..., \n> argv=0x0, typev=0x0, nargs=0, dest=Remote) at postgres.c:615\n> #21 0x80e5e80 in pg_exec_query (\n> query_string=0xbfffda08 \"SELECT a.attnum, a.attname, t.typname, a.attlen, a.atttypmod, a.attnotnull, a.atthasdef FROM pg_class c, pg_attribute a, pg_type t WHERE c.relname = 'city' and a.attnum > 0 and a.attrelid = c.o\"..., \n> argv=0x0, typev=0x0, nargs=0) at postgres.c:597\n> #22 0x80e6c81 in PostgresMain (argc=9, argv=0xbffffa48) at postgres.c:1373\n> #23 0x80a91ed in main (argc=9, argv=0xbffffa48) at main.c:79\n> #24 0x806014e in _start ()\n> \n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 00:08:37 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] VACUUM ANALYZE Problem (debugging output)"
},
{
"msg_contents": "\n\nOn Mon, 2 Feb 1998, Bruce Momjian wrote:\n\n: I assume you have run initdb after doing the sup. Prior to the beta\n: release today, things could have changed in the database format.\n: \n\nYes. I always do a \"make distclean\", cvsup the latest code, build and\ninstall everything, delete the data directory and start from scratch.\n\nEverything works as expected until >>AFTER<< running \"vacuum analyze\"...\n\nvtest=> \\d\n\nDatabase = vtest\n +------------------+----------------------------------+----------+\n | Owner | Relation | Type |\n +------------------+----------------------------------+----------+\n | jamesh | city | table |\n | jamesh | county | table |\n | jamesh | i_city | index |\n | jamesh | i_county | index |\n | jamesh | i_state | index |\n | jamesh | s_city | sequence |\n | jamesh | s_county | sequence |\n | jamesh | s_state | sequence |\n | jamesh | state | table |\n +------------------+----------------------------------+----------+\nvtest=> \\d city\n\nTable = city\n+----------------------+-----------------------------+-------+\n| Field | Type | Length|\n+----------------------+-----------------------------+-------+\n| serial | int4 |4 |\n| county | int4 |4 |\n| state | int4 |4 |\n| addr1 | text |var |\n| addr2 | text |var |\n| name | text |var |\n| zip | text |var |\n| title | text |var |\n| lname | text |var |\n| fname | text |var |\n| phone | text |var |\n| fax | text |var |\n| notes | text |var |\n+----------------------+-----------------------------+-------+\nvtest=> \nvtest=> vacuum ;\nVACUUM\nvtest=> \\d city\n\nTable = city\n+----------------------+-----------------------------+-------+\n| Field | Type |Length |\n+----------------------+-----------------------------+-------+\n| serial | int4 |4 |\n| county | int4 |4 |\n| state | int4 |4 |\n| addr1 | text |var |\n| addr2 | text |var |\n| name | text |var |\n| zip | text |var |\n| title | text |var |\n| lname | text |var |\n| fname | text |var |\n| phone | text |var |\n| fax | text |var |\n| notes | text |var |\n+----------------------+-----------------------------+-------+\nvtest=> vacuum analyze;\nVACUUM\nvtest=> select * from city;\nserial|county|state|addr1|addr2|name|zip|title|lname|fname|phone|fax|notes\n------+------+-----+-----+-----+----+---+-----+-----+-----+-----+---+-----\n(0 rows)\n\n<< Exit & Run Perl Script To Populate Database >>\n\nvtest=> select * from city;\n\n...\n10000565|10000098|10000000| | |Yadkinville | | |\n| | | | \n10000566|10000099|10000000| | |Burnsville | | |\n| | | | \n(567 rows)\n\nvtest=> vacuum; \nVACUUM\n\nvtest=> \\d city\n\nTable = city\n+-----------------------+---------------------------+-------+\n| Field | Type |Length |\n+-----------------------+---------------------------+-------+\n| serial | int4 |4 |\n| county | int4 |4 |\n| state | int4 |4 |\n| addr1 | text ...\n ... |var |\n+-----------------------+---------------------------+-------+\nvtest=> vacuum analyze;\nVACUUM\nvtest=> \\d city\nERROR: fmgr_info: function 0: cache lookup failed\n\nvtest=> \\dS\nERROR: fmgr_info: function 0: cache lookup failed\n\n\n\n-James\n\n",
"msg_date": "Mon, 2 Feb 1998 06:45:57 -0500 (EST)",
"msg_from": "James Hughes <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] VACUUM ANALYZE Problem (debugging output)"
}
] |
[
{
"msg_contents": "Hi!\n\nvac=> create table x (y int, z int);\nCREATE\nvac=> insert into x values (1,1);\nINSERT 18168 1\nvac=> insert into x values (1,2);\nINSERT 18169 1\nvac=> insert into x values (2,1);\nINSERT 18170 1\nvac=> insert into x values (2,2);\nINSERT 18171 1\nvac=> select * from x where y = (select max(y) from x);\nERROR: parser: parse error at or near \"select\"\nvac=> select * from x where y <> (select max(y) from x);\nERROR: parser: parse error at or near \"select\"\nvac=> select * from x where y < (select max(y) from x);\nERROR: parser: parse error at or near \"select\"\nvac=> select * from x where (y,z) = (select max(y), max(z) from x);\nERROR: parser: parse error at or near \"=\"\nvac=> select * from x where (y,z) = ANY (select min(y), max(z) from x);\nERROR: parser: parse error at or near \"=\"\nvac=> select * from x where (y,z) <> (select max(y), max(z) from x);\ny|z\n-+-\n1|1\n1|2\n2|1\n(3 rows)\n\nTom, Bruce - could you take care about this ?\n(BTW, I fixed parse_expr.c broken for EXISTS...)\nI'm going home now and will be here ~ 2 Feb 20:00 (PST). Hope to include \nsubselect code into CVS in the next 24 hrs (from now)...\n\nAlso, could someone take care about data/queries for regression tests ?\n(May be by using \"big boys\"...)\nThis would be very helpful!\n\nTIA,\n\tVadim\n",
"msg_date": "Mon, 02 Feb 1998 17:54:01 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "subquery syntax broken"
},
{
"msg_contents": "> Tom, Bruce - could you take care about this ?\n\nBruce? Let me know if you want me to look at it. I didn't include this\nsyntax originally since I thought singleton results like aggregates would\ncertainly not be implemented in the first cut. Forgot about Vadim's\nprodigious talents :)\n\n> (BTW, I fixed parse_expr.c broken for EXISTS...)\n> I'm going home now and will be here ~ 2 Feb 20:00 (PST). Hope to include\n> subselect code into CVS in the next 24 hrs (from now)...\n>\n> Also, could someone take care about data/queries for regression tests ?\n> (May be by using \"big boys\"...)\n\nYes, we should add a \"subselect.sql\" regression test. Anyone taking a first\ncut?\n\n - Tom\n\n",
"msg_date": "Mon, 02 Feb 1998 15:41:22 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
},
{
"msg_contents": "> vac=> select * from x where y = (select max(y) from x);\n> ERROR: parser: parse error at or near \"select\"\n> vac=> select * from x where y <> (select max(y) from x);\n> ERROR: parser: parse error at or near \"select\"\n\nSorry I missed that most obvious option. Here is the patch. I will\napply it today.\n\n---------------------------------------------------------------------------\n\n*** ./backend/parser/gram.y.orig\tMon Feb 2 11:51:05 1998\n--- ./backend/parser/gram.y\tMon Feb 2 11:59:12 1998\n***************\n*** 3330,3335 ****\n--- 3330,3345 ----\n \t\t\t\t\tn->subselect = $5;\n \t\t\t\t\t$$ = (Node *)n;\n \t\t\t\t}\n+ \t\t| a_expr Op '(' SubSelect ')'\n+ \t\t\t\t{\n+ \t\t\t\t\tSubLink *n = makeNode(SubLink);\n+ \t\t\t\t\tn->lefthand = lcons($1, NULL);\n+ \t\t\t\t\tn->oper = lcons($2,NIL);\n+ \t\t\t\t\tn->useor = false;\n+ \t\t\t\t\tn->subLinkType = ALL_SUBLINK;\n+ \t\t\t\t\tn->subselect = $4;\n+ \t\t\t\t\t$$ = (Node *)n;\n+ \t\t\t\t}\n \t\t| a_expr AND a_expr\n \t\t\t\t{\t$$ = makeA_Expr(AND, NULL, $1, $3); }\n \t\t| a_expr OR a_expr\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 12:00:14 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
},
{
"msg_contents": "> \n> > Tom, Bruce - could you take care about this ?\n> \n> Bruce? Let me know if you want me to look at it. I didn't include this\n> syntax originally since I thought singleton results like aggregates would\n> certainly not be implemented in the first cut. Forgot about Vadim's\n> prodigious talents :)\n\nIt's not the aggregates, it is the whole '= (subquery)' that is missing\nfrom gram.y. I am adding it now.\n\n\ttest=> select * from pg_user where usesysid = (select usesysid from\n\tpg_user);\n\tERROR: parser: syntax error at or near \"select\"\n\n\n> \n> > (BTW, I fixed parse_expr.c broken for EXISTS...)\n> > I'm going home now and will be here ~ 2 Feb 20:00 (PST). Hope to include\n> > subselect code into CVS in the next 24 hrs (from now)...\n> >\n> > Also, could someone take care about data/queries for regression tests ?\n> > (May be by using \"big boys\"...)\n> \n> Yes, we should add a \"subselect.sql\" regression test. Anyone taking a first\n> cut?\n> \n> - Tom\n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 12:12:01 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
},
{
"msg_contents": "> It's not the aggregates, it is the whole '= (subquery)' that is missing\n> from gram.y. I am adding it now.\n\nRight, and aggregates are the only way in general to get a singleton result from\na subselect. OK, I forgot about \"where y = (select 1)\". Well, only _useful_\nway?? I'll bet I'm forgetting another one too...\n\n - Tom\n\n",
"msg_date": "Mon, 02 Feb 1998 17:53:35 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
},
{
"msg_contents": "> \n> > It's not the aggregates, it is the whole '= (subquery)' that is missing\n> > from gram.y. I am adding it now.\n> \n> Right, and aggregates are the only way in general to get a singleton result from\n> a subselect. OK, I forgot about \"where y = (select 1)\". Well, only _useful_\n> way?? I'll bet I'm forgetting another one too...\n> \n\nSometimes you have = (subselect) with one row, often with correlated\nsubqueries, but most often with aggregates.\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 13:29:33 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> > It's not the aggregates, it is the whole '= (subquery)' that is missing\n> > from gram.y. I am adding it now.\n> \n> Right, and aggregates are the only way in general to get a singleton result from\n> a subselect. OK, I forgot about \"where y = (select 1)\". Well, only _useful_\n> way?? I'll bet I'm forgetting another one too...\n\nHow about \"where y = (select distinct foo from bar where n=5)\" ?\n\nOcie\n",
"msg_date": "Mon, 2 Feb 1998 10:52:10 -0800 (PST)",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> > vac=> select * from x where y = (select max(y) from x);\n> > ERROR: parser: parse error at or near \"select\"\n> > vac=> select * from x where y <> (select max(y) from x);\n> > ERROR: parser: parse error at or near \"select\"\n> \n> Sorry I missed that most obvious option. Here is the patch. I will\n> apply it today.\n\nIt doesn't work for =, >, etc:\n\nvac=> select * from x where y = (select max(y) from x);\nERROR: parser: parse error at or near \"select\"\n\nbut work for others:\n\nvac=> select * from x where y @ (select max(y) from x);\nERROR: There is no operator '@' for types 'int4' and 'int4'\n You will either have to retype this query using an explicit cast,\n or you will have to define the operator using CREATE OPERATOR\n\nAlso:\n\n> + | a_expr Op '(' SubSelect ')'\n> + {\n> + SubLink *n = makeNode(SubLink);\n> + n->lefthand = lcons($1, NULL);\n> + n->oper = lcons($2,NIL);\n> + n->useor = false;\n> + n->subLinkType = ALL_SUBLINK;\n ^^^^^^^^^^^\n should be EXPR_SUBLINK\n> + n->subselect = $4;\n> + $$ = (Node *)n;\n> + }\n\nVadim\n",
"msg_date": "Tue, 03 Feb 1998 09:50:17 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
},
{
"msg_contents": "\nI believe it has to do with the fact that '=' has right precedence. \nThomas, can you comment. Maybe we need to %right 'Op' at that point so\nit doesn't shift too early?\n\n> \n> Bruce Momjian wrote:\n> > \n> > > vac=> select * from x where y = (select max(y) from x);\n> > > ERROR: parser: parse error at or near \"select\"\n> > > vac=> select * from x where y <> (select max(y) from x);\n> > > ERROR: parser: parse error at or near \"select\"\n> > \n> > Sorry I missed that most obvious option. Here is the patch. I will\n> > apply it today.\n> \n> It doesn't work for =, >, etc:\n> \n> vac=> select * from x where y = (select max(y) from x);\n> ERROR: parser: parse error at or near \"select\"\n> \n> but work for others:\n> \n> vac=> select * from x where y @ (select max(y) from x);\n> ERROR: There is no operator '@' for types 'int4' and 'int4'\n> You will either have to retype this query using an explicit cast,\n> or you will have to define the operator using CREATE OPERATOR\n> \n> Also:\n> \n> > + | a_expr Op '(' SubSelect ')'\n> > + {\n> > + SubLink *n = makeNode(SubLink);\n> > + n->lefthand = lcons($1, NULL);\n> > + n->oper = lcons($2,NIL);\n> > + n->useor = false;\n> > + n->subLinkType = ALL_SUBLINK;\n> ^^^^^^^^^^^\n> should be EXPR_SUBLINK\n> > + n->subselect = $4;\n> > + $$ = (Node *)n;\n> > + }\n> \n> Vadim\n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 22:17:41 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
},
{
"msg_contents": "> I believe it has to do with the fact that '=' has right precedence.\n> Thomas, can you comment. Maybe we need to %right 'Op' at that point so\n> it doesn't shift too early?\n\nNo, the single-character operators each need their own code in the parser.\nCheck near line 2980 in the parser for examples from the \"a_expr\" syntax. You\njust need to replicate the subselect \"Op\" definition blocks and substitute\neach of '=', '<', and '>' in the copies. The existing \"Op\" code handles the\nmulti-character operators such as '<=' and '>='...\n\n - Tom\n\n> > It doesn't work for =, >, etc:\n> >\n> > vac=> select * from x where y = (select max(y) from x);\n> > ERROR: parser: parse error at or near \"select\"\n> >\n> > but work for others:\n> >\n> > vac=> select * from x where y @ (select max(y) from x);\n> > ERROR: There is no operator '@' for types 'int4' and 'int4'\n> > You will either have to retype this query using an explicit cast,\n> > or you will have to define the operator using CREATE OPERATOR\n> >\n> > Also:\n> >\n> > > + | a_expr Op '(' SubSelect ')'\n> > > + {\n> > > + SubLink *n = makeNode(SubLink);\n> > > + n->lefthand = lcons($1, NULL);\n> > > + n->oper = lcons($2,NIL);\n> > > + n->useor = false;\n> > > + n->subLinkType = ALL_SUBLINK;\n> > ^^^^^^^^^^^\n> > should be EXPR_SUBLINK\n> > > + n->subselect = $4;\n> > > + $$ = (Node *)n;\n> > > + }\n> >\n> > Vadim\n> >\n>\n> --\n> Bruce Momjian\n> [email protected]\n\n\n\n",
"msg_date": "Tue, 03 Feb 1998 08:18:37 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> > I believe it has to do with the fact that '=' has right precedence.\n> > Thomas, can you comment. Maybe we need to %right 'Op' at that point so\n> > it doesn't shift too early?\n> \n> No, the single-character operators each need their own code in the parser.\n> Check near line 2980 in the parser for examples from the \"a_expr\" syntax. You\n> just need to replicate the subselect \"Op\" definition blocks and substitute\n> each of '=', '<', and '>' in the copies. The existing \"Op\" code handles the\n\nDon't forget about ALL/ANY modifiers, too...\n\n> multi-character operators such as '<=' and '>='...\n\nVadim\n",
"msg_date": "Tue, 03 Feb 1998 15:34:58 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
},
{
"msg_contents": "Done. Patch applied. Makes the grammar bigger, though.\n\n> \n> > I believe it has to do with the fact that '=' has right precedence.\n> > Thomas, can you comment. Maybe we need to %right 'Op' at that point so\n> > it doesn't shift too early?\n> \n> No, the single-character operators each need their own code in the parser.\n> Check near line 2980 in the parser for examples from the \"a_expr\" syntax. You\n> just need to replicate the subselect \"Op\" definition blocks and substitute\n> each of '=', '<', and '>' in the copies. The existing \"Op\" code handles the\n> multi-character operators such as '<=' and '>='...\n> \n> - Tom\n> \n> > > It doesn't work for =, >, etc:\n> > >\n> > > vac=> select * from x where y = (select max(y) from x);\n> > > ERROR: parser: parse error at or near \"select\"\n> > >\n> > > but work for others:\n> > >\n> > > vac=> select * from x where y @ (select max(y) from x);\n> > > ERROR: There is no operator '@' for types 'int4' and 'int4'\n> > > You will either have to retype this query using an explicit cast,\n> > > or you will have to define the operator using CREATE OPERATOR\n> > >\n> > > Also:\n> > >\n> > > > + | a_expr Op '(' SubSelect ')'\n> > > > + {\n> > > > + SubLink *n = makeNode(SubLink);\n> > > > + n->lefthand = lcons($1, NULL);\n> > > > + n->oper = lcons($2,NIL);\n> > > > + n->useor = false;\n> > > > + n->subLinkType = ALL_SUBLINK;\n> > > ^^^^^^^^^^^\n> > > should be EXPR_SUBLINK\n> > > > + n->subselect = $4;\n> > > > + $$ = (Node *)n;\n> > > > + }\n> > >\n> > > Vadim\n> > >\n> >\n> > --\n> > Bruce Momjian\n> > [email protected]\n> \n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 3 Feb 1998 14:26:40 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] subquery syntax broken"
}
] |
[
{
"msg_contents": "I noticed a bug in parse_expr.c line 104:\n\ntemplate1=> select 'Andreas'::char(16);\nPQexec() -- Request was sent to backend, but backend closed the channel\nbefore responding.\n This probably means the backend terminated abnormally before or\nwhile processing the request.\ntemplate1=> \\c template1\nconnecting to new database: template1\ntemplate1=> select 'Andreas'::varchar(16);\nPQexec() -- Request was sent to backend, but backend closed the channel\nbefore responding.\n This probably means the backend terminated abnormally before or\nwhile processing the request.\ntemplate1=> \\c template1\ntemplate1=> select 'this works'::char16;\n?column?\n--------\nthis works\n(1 row)\nsnapshot 020298 on aix_gcc and on linux_elf same behavior\n\ndbx sais:\n(dbx) where\n[stipped some lines]\npalloc(0x0), line 69 in \"palloc.c\"\nbpcharin(0x2007cb48, 0x12, 0x0), line 78 in \"varchar.c\"\nfmgr_c(0x2ff20520, 0x2ff204f8, 0x2ff20530), line 107 in \"fmgr.c\"\nfmgr(), line 285 in \"fmgr.c\"\nstringTypeString(0x2008a108, 0x2007cb48, 0x0), line 146 in\n\"parse_type.c\"\nparser_typecast(0x20088ab0, 0x20088ae8, 0x0), line 434 in \"parse_expr.c\"\nunnamed block $b768, line 104 in \"parse_expr.c\"\ntransformExpr(0x20088c88, 0x20088aa8, 0x1), line 104 in \"parse_expr.c\"\n[stripped some lines]\n\nNotice: \tthe atttypmod is 0 instead of -1 (or maybe it should be\n16, but I would \n\tignore the char(16) argument in the cast and use -1)\n\nline should be:\n*** parse_expr.c.ori Mon Feb 2 09:00:20 1998\n--- parse_expr.c Mon Feb 2 14:20:14 1998\n***************\n*** 103,105 ****\n if (con->typename != NULL)\n! result = parser_typecast(val,\ncon->typename, 0);\n else\n--- 103,105 ----\n if (con->typename != NULL)\n! result = parser_typecast(val,\ncon->typename, -1);\n else\n\nseems to work fine !\n\nAndreas\n",
"msg_date": "Mon, 2 Feb 1998 14:35:52 +0100",
"msg_from": "Zeugswetter Andreas DBT <[email protected]>",
"msg_from_op": true,
"msg_subject": "bug fix for select 'some Text'::char(16) (newest snapshot)"
},
{
"msg_contents": "Yes, I am inclined to agree with you. I will apply the patch.\n\n---------------------------------------------------------------------------\n\n\n> *** parse_expr.c.ori Mon Feb 2 09:00:20 1998\n> --- parse_expr.c Mon Feb 2 14:20:14 1998\n> ***************\n> *** 103,105 ****\n> if (con->typename != NULL)\n> ! result = parser_typecast(val,\n> con->typename, 0);\n> else\n> --- 103,105 ----\n> if (con->typename != NULL)\n> ! result = parser_typecast(val,\n> con->typename, -1);\n> else\n> \n> seems to work fine !\n> \n> Andreas\n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 12:05:17 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] bug fix for select 'some Text'::char(16) (newest\n\tsnapshot)"
}
] |
[
{
"msg_contents": "Has anyone confirmed these problems under Irix?\n\n> Hi,\n> \n> i have just grabbed the 6.3 snapshot and tried to compile it on irix 6.2 -\n> without success.\n> \n> The following errors occurred:\n> \n> 1) configure fails\n> \n> % ./configure \n> [...]\n> linking ./backend/port/tas/i386_solaris.s to backend/port/tas.s\n> linking ./backend/port/dynloader/irix5.c to backend/port/dynloader.c\n> configure: error: ./backend/port/dynloader/irix5.c: File not found\nThis seems a little fundamental!!!\n\n> \n> \n> 2) compiling the backend fails:\n> in src/backend the gcc call \n> gcc -mabi=64 -o postgres access/SUBSYS.o bootstrap/SUBSYS.o\n> catalog/SUBSYS.o commands/SUBSYS.o executor/SUBSYS.o lib/SUBSYS.o\n> libpq/SUBSYS.o main/SUBSYS.o nodes/SUBSYS.o optimizer/SUBSYS.o\n> parser/SUBSYS.o port/SUBSYS.o postmaster/SUBSYS.o regex/SUBSYS.o\n> rewrite/SUBSYS.o storage/SUBSYS.o tcop/SUBSYS.o utils/SUBSYS.o\n> ../utils/version.o -64 -lPW -lcrypt -lm -lbsd -lreadline -lhistory\n> -ltermcap -lcurses \n> \n> returns \n> [...]\n> ld64: ERROR 33: Unresolved text symbol \"S_INIT_LOCK\" -- 1st referenced by\n> storage/SUBSYS.o.\n> ld64: ERROR 33: Unresolved text symbol \"S_UNLOCK\" -- 1st referenced by\n> storage/SUBSYS.o.\n> ld64: ERROR 33: Unresolved text symbol \"S_LOCK\" -- 1st referenced by\n> storage/SUBSYS.o\n> \n> \n> these functions just seem to be implemented for linux/alpha ??\nI'm generally a little dubious when people report a problem in actually\ncompiling 'cos of the Irix 6.x bug in ld. However, this guy is using\ngcc, not native Irix cc/ld\n\n> Fuad Abdallah\n> Max-Planck-Institut fuer Zuechtungsforschung / ZWDV\n> Tel.: 0221/5062-739 / Priv: 0221/584563\n> \n\nWith any luck I should have a bit of time next week to do some testing under\nIrix - things have been rather hectic the last few weeks.\n\n\nAndrew\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Mon, 2 Feb 1998 15:13:48 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] snapshot won't compile on irix6.2"
},
{
"msg_contents": "On Mon, 2 Feb 1998, Andrew Martin wrote:\n\n> > [...]\n> > linking ./backend/port/tas/i386_solaris.s to backend/port/tas.s\n> > linking ./backend/port/dynloader/irix5.c to backend/port/dynloader.c\n> > configure: error: ./backend/port/dynloader/irix5.c: File not found\n> This seems a little fundamental!!!'\n\n\tFixed now...\n\n> > returns \n> > [...]\n> > ld64: ERROR 33: Unresolved text symbol \"S_INIT_LOCK\" -- 1st referenced by\n> > storage/SUBSYS.o.\n> > ld64: ERROR 33: Unresolved text symbol \"S_UNLOCK\" -- 1st referenced by\n> > storage/SUBSYS.o.\n> > ld64: ERROR 33: Unresolved text symbol \"S_LOCK\" -- 1st referenced by\n> > storage/SUBSYS.o\n> > \n> > \n> > these functions just seem to be implemented for linux/alpha ??\n> I'm generally a little dubious when people report a problem in actually\n> compiling 'cos of the Irix 6.x bug in ld. However, this guy is using\n> gcc, not native Irix cc/ld\n\n>From include/storage/s_lock.h:\n\n#if defined(irix5)\n\n\tUsing tools/ccsym, what shoudl this be set to. There is no longer\nany concept of a \"port\" at the Makefile level, and hasn't been for over a\nmonth now...From henceforth, its either determined based on what configure\nknows about, or the compiler itself...in this case, tools/ccsym will have\nto be used to determine what the compiler defines internally and the\nsettings changed...\n\n",
"msg_date": "Mon, 2 Feb 1998 10:28:12 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] snapshot won't compile on irix6.2"
}
] |
[
{
"msg_contents": "\nThe link from the main English FAQ:\nhttp://www.PostgreSQL.ORG/docs/faq-english.shtml\nto the Linux and Irix specific FAQs seem to be broken. I get \"Document contains\nno data\" error messages. The links are:\n http://postgreSQL.org/docs/FAQ-linux.shtml\n http://postgreSQL.org/docs/FAQ-irix.shtml\n\nIt appears that FAQ should be in lower case.\n\n\nBTW a quick reminder to make sure that these latest versions get into\nthe 6.3 release (unlike with 6.2 :-)\n\n\nAndrew\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Mon, 2 Feb 1998 15:22:11 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "FAQs"
},
{
"msg_contents": "On Mon, 2 Feb 1998, Andrew Martin wrote:\n\n> \n> The link from the main English FAQ:\n> http://www.PostgreSQL.ORG/docs/faq-english.shtml\n> to the Linux and Irix specific FAQs seem to be broken. I get \"Document contains\n> no data\" error messages. The links are:\n> http://postgreSQL.org/docs/FAQ-linux.shtml\n> http://postgreSQL.org/docs/FAQ-irix.shtml\n> \n> It appears that FAQ should be in lower case.\n\n\n\tFixed...\n\n> BTW a quick reminder to make sure that these latest versions get into\n> the 6.3 release (unlike with 6.2 :-)\n\n\tUpdated...\n\n\n",
"msg_date": "Mon, 2 Feb 1998 10:54:29 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] FAQs"
},
{
"msg_contents": "> \n> \n> The link from the main English FAQ:\n> http://www.PostgreSQL.ORG/docs/faq-english.shtml\n> to the Linux and Irix specific FAQs seem to be broken. I get \"Document contains\n> no data\" error messages. The links are:\n> http://postgreSQL.org/docs/FAQ-linux.shtml\n> http://postgreSQL.org/docs/FAQ-irix.shtml\n> \n> It appears that FAQ should be in lower case.\n> \n> \n> BTW a quick reminder to make sure that these latest versions get into\n> the 6.3 release (unlike with 6.2 :-)\n\nI forgot to update them in the 6.3 beta distribution. I always forget\nthem.\n\nMarc, would you do the honors?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 12:09:26 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] FAQs"
}
] |
[
{
"msg_contents": "> > > Looks like this variable size stuff will be the first new feature\n> > > for 6.4 then, eh? :)\n> >\n> > What is the chance of having this ready by Friday? Still would\n> > give us 3weeks of debugging...\n> \n> Well, it's OK to leave _one_ feature for v6.4. Bruce is probably tired of typing\n> the feature list :)\n> \n> Also, we still have a few loose ends for v6.3. But, I'll bet Darren's stuff can\n> go in and not affect anything if we run using the current block size.\n\nGreat chance of having it by Friday if not sooner. I'll get on it tonite.\n\nWasn't sure where the line in the sand was for \"adding features\" vs \"fixing bugs.\"\n\nThanks,\n\nDarren\n",
"msg_date": "Mon, 2 Feb 1998 10:56:55 -0500",
"msg_from": "[email protected] (Darren King)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Variable block size..."
},
{
"msg_contents": "On Mon, 2 Feb 1998, Darren King wrote:\n\n> > > > Looks like this variable size stuff will be the first new feature\n> > > > for 6.4 then, eh? :)\n> > >\n> > > What is the chance of having this ready by Friday? Still would\n> > > give us 3weeks of debugging...\n> > \n> > Well, it's OK to leave _one_ feature for v6.4. Bruce is probably tired of typing\n> > the feature list :)\n> > \n> > Also, we still have a few loose ends for v6.3. But, I'll bet Darren's stuff can\n> > go in and not affect anything if we run using the current block size.\n> \n> Great chance of having it by Friday if not sooner. I'll get on it tonite.\n> \n> Wasn't sure where the line in the sand was for \"adding features\" vs\n> \"fixing bugs.\" \n\n\tThe only *fixed* date is march 1st...certain developers have a\nlittle leeway as far as the beta release is concerned...leeway tending to\ndepend on how quickly bugs they generate get fixed...anyway, fixed sized\nblock sizes are a bug, aren't they? *grin*\n\n\n",
"msg_date": "Mon, 2 Feb 1998 11:32:44 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Variable block size..."
},
{
"msg_contents": "> \n> > > > Looks like this variable size stuff will be the first new feature\n> > > > for 6.4 then, eh? :)\n> > >\n> > > What is the chance of having this ready by Friday? Still would\n> > > give us 3weeks of debugging...\n> > \n> > Well, it's OK to leave _one_ feature for v6.4. Bruce is probably tired of typing\n> > the feature list :)\n> > \n> > Also, we still have a few loose ends for v6.3. But, I'll bet Darren's stuff can\n> > go in and not affect anything if we run using the current block size.\n> \n> Great chance of having it by Friday if not sooner. I'll get on it tonite.\n> \n> Wasn't sure where the line in the sand was for \"adding features\" vs \"fixing bugs.\"\n\nI think Marc is getting softer on this issue. [Ducks head]\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 12:13:04 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Variable block size..."
},
{
"msg_contents": "On Mon, 2 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > > > > Looks like this variable size stuff will be the first new feature\n> > > > > for 6.4 then, eh? :)\n> > > >\n> > > > What is the chance of having this ready by Friday? Still would\n> > > > give us 3weeks of debugging...\n> > > \n> > > Well, it's OK to leave _one_ feature for v6.4. Bruce is probably \n> > > tired of typing the feature list :)\n> > > \n> > > Also, we still have a few loose ends for v6.3. But, I'll bet Darren's stuff can\n> > > go in and not affect anything if we run using the current block size.\n\n> > \n> > Great chance of having it by Friday if not sooner. I'll get on it tonite.\n> > \n> > Wasn't sure where the line in the sand was for \"adding features\" vs \"fixing bugs.\"\n> \n> I think Marc is getting softer on this issue. [Ducks head]\n\n\tMoi? Never...some ppl have shown over time a...dedication towards\nfixing any bugs they introduce in a *very* timely manner *grin* A little\nleeway, as appropriate, tends to reap large rewards for all\nparties...specially if we can help improve disk space usage by providing\nthe ability to tailor disk block size used :)\n\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 2 Feb 1998 20:45:10 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Variable block size..."
}
] |
[
{
"msg_contents": "Hello everybody,\n\nGreat job for 6.3! On my Linux box a nightmare interogation got down from\n15 seconds to 3 seconds.\n\nBad surprize: The parser does not accept anymore field names like \"valid\"\nand \"no\". Will this be the normal behaviour from now on or just a\nmisunderstanding?\n\nCostin Oproiu\n\n\n\n",
"msg_date": "Tue, 3 Feb 1998 00:12:16 +0200 (EET)",
"msg_from": "PostgreSQL <[email protected]>",
"msg_from_op": true,
"msg_subject": "Speed boost + Others"
},
{
"msg_contents": "> \n> Hello everybody,\n> \n> Great job for 6.3! On my Linux box a nightmare interogation got down from\n> 15 seconds to 3 seconds.\n> \n> Bad surprize: The parser does not accept anymore field names like \"valid\"\n> and \"no\". Will this be the normal behaviour from now on or just a\n> misunderstanding?\n> \n> Costin Oproiu\n\nWhat was the old version, 6.2.1?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 17:33:02 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Speed boost + Others"
},
{
"msg_contents": "On Tue, 3 Feb 1998, PostgreSQL wrote:\n\n> Hello everybody,\n> \n> Great job for 6.3! On my Linux box a nightmare interogation got down from\n> 15 seconds to 3 seconds.\n> \n> Bad surprize: The parser does not accept anymore field names like \"valid\"\n> and \"no\". Will this be the normal behaviour from now on or just a\n> misunderstanding?\n\n\tWe've tighted down on use of 'reserved words' for field\nnames...this is now 'normal behavior'\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 2 Feb 1998 20:47:40 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Speed boost + Others"
},
{
"msg_contents": "\n\nOn Mon, 2 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > Hello everybody,\n> > \n> > Great job for 6.3! On my Linux box a nightmare interogation got down from\n> > 15 seconds to 3 seconds.\n> > \n> > Bad surprize: The parser does not accept anymore field names like \"valid\"\n> > and \"no\". Will this be the normal behaviour from now on or just a\n> > misunderstanding?\n> > \n> > Costin Oproiu\n> \n> What was the old version, 6.2.1?\n\n\tyep\n\n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n\n",
"msg_date": "Tue, 3 Feb 1998 10:53:28 +0200 (EET)",
"msg_from": "PostgreSQL <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Speed boost + Others"
}
] |
[
{
"msg_contents": "I'm getting the following this evening from CVS?\n\nAnything wrong?\n\nKeith.\n\nmtcc:[/postgres95/CVS](44)% cvsup CVSupfile\nConnected to postgresql.org\nUpdating collection pgsql/cvs\n Edit pgsql/src/test/regress/regression.FreeBSD\nUpdater failed: pgsql/src/test/regress/regression.FreeBSD: Cannot open: Invalid \nargument\nmtcc:[/postgres95/CVS](45)% ls -al pgsql/src/test/regress/regression.*\n-rw-r--r-- 1 emkxp01 isu 0 Feb 1 11:42 \npgsql/src/test/regress/regression.FreeBSD\nmtcc:[/postgres95/CVS](46)% \n\n",
"msg_date": "Mon, 2 Feb 1998 22:13:49 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "CVS funny with cvsup."
},
{
"msg_contents": "On Mon, 2 Feb 1998, Keith Parks wrote:\n\n> I'm getting the following this evening from CVS?\n> \n> Anything wrong?\n\n\tJust checked everything over here, and made sure that CVSup that\nis used for the snapshots isn't failing...is this still a problem?\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 2 Feb 1998 20:52:55 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] CVS funny with cvsup."
}
] |
[
{
"msg_contents": "vac=> \\h vacuum\nCommand: vacuum\nDescription: vacuum the database, i.e. cleans out deleted records, updates statistics\nSyntax:\nvacuum [verbose] [analyze]\n or\nvacuum [verbose] [analyze] table [analyze [(attr1, ... attrN)] ];\n ^^^^^^^\nImho, new ANALYZE syntax with attributes is quite unclear to me:\n\nvac=> vacuum x (y);\nPQexec() -- Request was sent to backend, but backend closed the channel before responding.\n This probably means the backend terminated abnormally before or while processing the request.\n\nHow about the next:\n\nSyntax:\nvacuum [verbose] [analyze] [table]\n or\nvacuum [verbose] analyze table (attr1, ... attrN);\n\nVadim\n",
"msg_date": "Tue, 03 Feb 1998 11:07:04 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "vacuum analyze syntax in psql' help"
},
{
"msg_contents": "> \n> vac=> \\h vacuum\n> Command: vacuum\n> Description: vacuum the database, i.e. cleans out deleted records, updates statistics\n> Syntax:\n> vacuum [verbose] [analyze]\n> or\n> vacuum [verbose] [analyze] table [analyze [(attr1, ... attrN)] ];\n> ^^^^^^^\n> Imho, new ANALYZE syntax with attributes is quite unclear to me:\n\nOops, forgot to remove that second analyze.\n\n> \n> vac=> vacuum x (y);\n> PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> This probably means the backend terminated abnormally before or while processing the request.\n> \n> How about the next:\n> \n> Syntax:\n> vacuum [verbose] [analyze] [table]\n> or\n> vacuum [verbose] analyze table (attr1, ... attrN);\n> \n> Vadim\n> \n\nMuch better.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 2 Feb 1998 23:33:59 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: vacuum analyze syntax in psql' help"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> > vac=> vacuum x (y);\n> > PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> > This probably means the backend terminated abnormally before or while processing the request.\n\nWill you fix this ?\n\n> >\n> > How about the next:\n> >\n> > Syntax:\n> > vacuum [verbose] [analyze] [table]\n> > or\n> > vacuum [verbose] analyze table (attr1, ... attrN);\n\nAnd this ?\n\nVadim\n",
"msg_date": "Tue, 03 Feb 1998 12:11:03 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: vacuum analyze syntax in psql' help"
},
{
"msg_contents": "Done. Manual page, psql help, and error message fixed.\n\n> \n> vac=> \\h vacuum\n> Command: vacuum\n> Description: vacuum the database, i.e. cleans out deleted records, updates statistics\n> Syntax:\n> vacuum [verbose] [analyze]\n> or\n> vacuum [verbose] [analyze] table [analyze [(attr1, ... attrN)] ];\n> ^^^^^^^\n> Imho, new ANALYZE syntax with attributes is quite unclear to me:\n> \n> vac=> vacuum x (y);\n> PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> This probably means the backend terminated abnormally before or while processing the request.\n> \n> How about the next:\n> \n> Syntax:\n> vacuum [verbose] [analyze] [table]\n> or\n> vacuum [verbose] analyze table (attr1, ... attrN);\n> \n> Vadim\n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 3 Feb 1998 14:25:04 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: vacuum analyze syntax in psql' help"
}
] |
[
{
"msg_contents": "Hullo,\n\nI seen the new lock table help in psql:\nI think we should stick to the (pseudo)standard syntax right from the\nbeginning.\nOr maybe allow both, but i think at least the table keyword should be\nmandatory \nto simplify implementation of other locks (like database).\n\nlock table <tablename> in {exclusive|share} mode;\n\nWhat we have now would conform to exclusive mode (works inside a\ntransaction only, ok but\nthat's the most important case :-).\n\nAndreas\n",
"msg_date": "Tue, 3 Feb 1998 11:37:33 +0100",
"msg_from": "Zeugswetter Andreas DBT <[email protected]>",
"msg_from_op": true,
"msg_subject": "lock table syntax"
},
{
"msg_contents": "> \n> Hullo,\n> \n> I seen the new lock table help in psql:\n> I think we should stick to the (pseudo)standard syntax right from the\n> beginning.\n> Or maybe allow both, but i think at least the table keyword should be\n> mandatory \n> to simplify implementation of other locks (like database).\n> \n> lock table <tablename> in {exclusive|share} mode;\n> \n> What we have now would conform to exclusive mode (works inside a\n> transaction only, ok but\n> that's the most important case :-).\n\nAdded to TODO list.\n\n-- \nBruce Momjian | 830 Blythe Avenue\[email protected] | Drexel Hill, Pennsylvania 19026\n + If your life is a hard drive, | (610) 353-9879(w)\n + Christ can be your backup. | (610) 853-3000(h)\n",
"msg_date": "Sun, 15 Mar 1998 23:40:06 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] lock table syntax"
}
] |
[
{
"msg_contents": "\n\nNobody likes to take the time to move discussions, do they? And ya, I'm\nabout as guilty *sigh*\n\nMove to [email protected]...\n\n\nOn Tue, 3 Feb 1998, Herouth Maoz wrote:\n\n> At 15:01 +0200 on 2/2/98, The Hermit Hacker wrote:\n> \n> I think the garbage collection should be separated from the statistics.\n> Garbage collection needs a write lock, statistics only a read lock. If they\n> are not done at the same time, the various tables would be locked for\n> shorter periods.\n\n\tHrmmmm...good point, I think. Bruce? Vadim? When vacuum'ng a\nlarge table, how much time is spend 'garbage collecting' vs 'statistics'?\nI thought that 'vacuum analyze' *was* the statistics aspect of it? Where\njust 'vacuum' was only garbage collection...?\n\n\n",
"msg_date": "Tue, 3 Feb 1998 07:10:16 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "> \n> \n> \n> Nobody likes to take the time to move discussions, do they? And ya, I'm\n> about as guilty *sigh*\n> \n> Move to [email protected]...\n> \n> \n> On Tue, 3 Feb 1998, Herouth Maoz wrote:\n> \n> > At 15:01 +0200 on 2/2/98, The Hermit Hacker wrote:\n> > \n> > I think the garbage collection should be separated from the statistics.\n> > Garbage collection needs a write lock, statistics only a read lock. If they\n> > are not done at the same time, the various tables would be locked for\n> > shorter periods.\n> \n> \tHrmmmm...good point, I think. Bruce? Vadim? When vacuum'ng a\n> large table, how much time is spend 'garbage collecting' vs 'statistics'?\n> I thought that 'vacuum analyze' *was* the statistics aspect of it? Where\n> just 'vacuum' was only garbage collection...?\n\nThis is correct. Vacuum is fast, vacuum analyze is pretty slow. We\ncould separate them, I guess, and that would eliminate the write-lock\nand be only a readlock.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 3 Feb 1998 13:22:22 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "On Tue, 3 Feb 1998, Bruce Momjian wrote:\n\n> > On Tue, 3 Feb 1998, Herouth Maoz wrote:\n> > \n> > > At 15:01 +0200 on 2/2/98, The Hermit Hacker wrote:\n> > > \n> > > I think the garbage collection should be separated from the statistics.\n> > > Garbage collection needs a write lock, statistics only a read lock. If they\n> > > are not done at the same time, the various tables would be locked for\n> > > shorter periods.\n> > \n> > \tHrmmmm...good point, I think. Bruce? Vadim? When vacuum'ng a\n> > large table, how much time is spend 'garbage collecting' vs 'statistics'?\n> > I thought that 'vacuum analyze' *was* the statistics aspect of it? Where\n> > just 'vacuum' was only garbage collection...?\n> \n> This is correct. Vacuum is fast, vacuum analyze is pretty slow. We\n> could separate them, I guess, and that would eliminate the write-lock\n> and be only a readlock.\n\n\tPossible to slip it in for v6.3? Would make it so that an analyze\ncould be done nightly, to keep statistics up, and then a vacuum once a\nweek or so just for garbage collection...?\n\n\n",
"msg_date": "Tue, 3 Feb 1998 13:46:22 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "> > This is correct. Vacuum is fast, vacuum analyze is pretty slow. We\n> > could separate them, I guess, and that would eliminate the write-lock\n> > and be only a readlock.\n> \n> \tPossible to slip it in for v6.3? Would make it so that an analyze\n> could be done nightly, to keep statistics up, and then a vacuum once a\n> week or so just for garbage collection...?\n\nWhen I added analyze, I did not understand the issues, so I was able to\nwork from Vadim's code in vacuum. I put it on the TODO list. Don't\nknow if it can make 6.3. I am working on cleaning up the cacheoffset\ncode right now.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 3 Feb 1998 14:30:09 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "On Tue, 3 Feb 1998, Bruce Momjian wrote:\n\n> > > This is correct. Vacuum is fast, vacuum analyze is pretty slow. We\n> > > could separate them, I guess, and that would eliminate the write-lock\n> > > and be only a readlock.\n> > \n> > \tPossible to slip it in for v6.3? Would make it so that an analyze\n> > could be done nightly, to keep statistics up, and then a vacuum once a\n> > week or so just for garbage collection...?\n> \n> When I added analyze, I did not understand the issues, so I was able to\n> work from Vadim's code in vacuum. I put it on the TODO list. Don't\n> know if it can make 6.3. I am working on cleaning up the cacheoffset\n> code right now.\n\n\tOkay...personally, I'm finding 'vacuum <table>' an acceptable work\naround, so it isn't too big of a priority :)\n\n\n",
"msg_date": "Tue, 3 Feb 1998 14:30:57 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "> \n> On Tue, 3 Feb 1998, Bruce Momjian wrote:\n> \n> > > > This is correct. Vacuum is fast, vacuum analyze is pretty slow. We\n> > > > could separate them, I guess, and that would eliminate the write-lock\n> > > > and be only a readlock.\n> > > \n> > > \tPossible to slip it in for v6.3? Would make it so that an analyze\n> > > could be done nightly, to keep statistics up, and then a vacuum once a\n> > > week or so just for garbage collection...?\n> > \n> > When I added analyze, I did not understand the issues, so I was able to\n> > work from Vadim's code in vacuum. I put it on the TODO list. Don't\n> > know if it can make 6.3. I am working on cleaning up the cacheoffset\n> > code right now.\n> \n> \tOkay...personally, I'm finding 'vacuum <table>' an acceptable work\n> around, so it isn't too big of a priority :)\n> \n\nVacuum probably write-locks the pg_class table because it updates the\ntable statistics. By vacuuming one table at a time, your lock is\nremoved and re-asserted, allowing other people to get into pg_class, and\na scan of pg_class is not necessary becuase you supply the table names. \n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 3 Feb 1998 14:40:03 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "On Tue, 3 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > On Tue, 3 Feb 1998, Bruce Momjian wrote:\n> > \n> > > > > This is correct. Vacuum is fast, vacuum analyze is pretty slow. We\n> > > > > could separate them, I guess, and that would eliminate the write-lock\n> > > > > and be only a readlock.\n> > > > \n> > > > \tPossible to slip it in for v6.3? Would make it so that an analyze\n> > > > could be done nightly, to keep statistics up, and then a vacuum once a\n> > > > week or so just for garbage collection...?\n> > > \n> > > When I added analyze, I did not understand the issues, so I was able to\n> > > work from Vadim's code in vacuum. I put it on the TODO list. Don't\n> > > know if it can make 6.3. I am working on cleaning up the cacheoffset\n> > > code right now.\n> > \n> > \tOkay...personally, I'm finding 'vacuum <table>' an acceptable work\n> > around, so it isn't too big of a priority :)\n> > \n> \n> Vacuum probably write-locks the pg_class table because it updates the\n> table statistics. By vacuuming one table at a time, your lock is\n> removed and re-asserted, allowing other people to get into pg_class, and\n> a scan of pg_class is not necessary becuase you supply the table names. \n\n\tWait, then I think I got this backwards.\n\n\tVacuum right now locks pg_class because of the statistics? If\nthat is the case, if we made vacuum *just* garbage collecting,it wouldn't\nhave to lock pg_class, only \"vacuum analyze\" wouldhave to do that?\n\n\tSo, I was misunderstanding in that I was thinking that 'vacuum\nanalyze' only needed the read-lock :(\n\n\n",
"msg_date": "Tue, 3 Feb 1998 14:42:18 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "> \n> On Tue, 3 Feb 1998, Bruce Momjian wrote:\n> \n> > > \n> > > On Tue, 3 Feb 1998, Bruce Momjian wrote:\n> > > \n> > > > > > This is correct. Vacuum is fast, vacuum analyze is pretty slow. We\n> > > > > > could separate them, I guess, and that would eliminate the write-lock\n> > > > > > and be only a readlock.\n> > > > > \n> > > > > \tPossible to slip it in for v6.3? Would make it so that an analyze\n> > > > > could be done nightly, to keep statistics up, and then a vacuum once a\n> > > > > week or so just for garbage collection...?\n> > > > \n> > > > When I added analyze, I did not understand the issues, so I was able to\n> > > > work from Vadim's code in vacuum. I put it on the TODO list. Don't\n> > > > know if it can make 6.3. I am working on cleaning up the cacheoffset\n> > > > code right now.\n> > > \n> > > \tOkay...personally, I'm finding 'vacuum <table>' an acceptable work\n> > > around, so it isn't too big of a priority :)\n> > > \n> > \n> > Vacuum probably write-locks the pg_class table because it updates the\n> > table statistics. By vacuuming one table at a time, your lock is\n> > removed and re-asserted, allowing other people to get into pg_class, and\n> > a scan of pg_class is not necessary becuase you supply the table names. \n> \n> \tWait, then I think I got this backwards.\n> \n> \tVacuum right now locks pg_class because of the statistics? If\n> that is the case, if we made vacuum *just* garbage collecting,it wouldn't\n> have to lock pg_class, only \"vacuum analyze\" wouldhave to do that?\n> \n> \tSo, I was misunderstanding in that I was thinking that 'vacuum\n> analyze' only needed the read-lock :(\n\nMaybe I am wrong. I have not looked at it.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 3 Feb 1998 14:43:31 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "On Tue, 3 Feb 1998, Bruce Momjian wrote:\n\n> > \tVacuum right now locks pg_class because of the statistics? If\n> > that is the case, if we made vacuum *just* garbage collecting,it wouldn't\n> > have to lock pg_class, only \"vacuum analyze\" wouldhave to do that?\n> > \n> > \tSo, I was misunderstanding in that I was thinking that 'vacuum\n> > analyze' only needed the read-lock :(\n> \n> Maybe I am wrong. I have not looked at it.\n\n\tOkay, just sitting here thinking about it, and that doesn't really\nmake sense (if its true)...\n\n\tVacuum should be locking the table itself for a garbage cleanup,\nsince it has to move around records, and I wouldn't imagine you'd want to\nhave someone doing a SELECT at the same time. So, that locks the *table*\nitself, but shouldn't affect pg_class (statistically)\n\n\tOnce the vacuum is finished its garbage cleanup phase (which,\ngranted, could take several minutes), then the statistics phase would come\ninto play...but again, a lock on pg_class shouldn't have to be imposed\nuntil the 'update' of the table takes place, should it? \n\n\tSo, why is pg_class locked for the duration of a vacuum when the\nvacuum is being performed for the whole database when it should (I think)\nonly need to be locked when updates are happening to it?\n\n\n",
"msg_date": "Tue, 3 Feb 1998 15:02:00 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "The Hermit Hacker wrote:\n> \n> So, why is pg_class locked for the duration of a vacuum when the\n> vacuum is being performed for the whole database when it should (I think)\n> only need to be locked when updates are happening to it?\n\nYes, this is how it should be. This long lock comes from very old time.\nWhen I visited vacuum last time (year ago) I hadn't time to decide\nwhy this lock is here and could it be just removed or not. \nStill no time :)\n\nVadim\n",
"msg_date": "Wed, 04 Feb 1998 09:14:09 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "> \n> The Hermit Hacker wrote:\n> > \n> > So, why is pg_class locked for the duration of a vacuum when the\n> > vacuum is being performed for the whole database when it should (I think)\n> > only need to be locked when updates are happening to it?\n> \n> Yes, this is how it should be. This long lock comes from very old time.\n> When I visited vacuum last time (year ago) I hadn't time to decide\n> why this lock is here and could it be just removed or not. \n> Still no time :)\n> \n> Vadim\n> \n> \n\nAdded to TODO:\n\n\t* Make VACUUM on database not lock pg_class\n\t* Make VACUUM ANALYZE only use a readlock\n\n\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 3 Feb 1998 21:37:51 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page\\"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> Added to TODO:\n> \n> * Make VACUUM on database not lock pg_class\n\nOk.\n\n> * Make VACUUM ANALYZE only use a readlock\n\n??? Vacuum analyze means: do vacuuming and analyzing, yes ?\nVacuuming need in write-lock... \nThere should be seperate command - ANALYZE, - when only\nread-lock would be used.\n\nVadim\n",
"msg_date": "Wed, 04 Feb 1998 09:59:35 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page\\"
},
{
"msg_contents": "> \n> Bruce Momjian wrote:\n> > \n> > Added to TODO:\n> > \n> > * Make VACUUM on database not lock pg_class\n> \n> Ok.\n> \n> > * Make VACUUM ANALYZE only use a readlock\n> \n> ??? Vacuum analyze means: do vacuuming and analyzing, yes ?\n> Vacuuming need in write-lock... \n> There should be seperate command - ANALYZE, - when only\n> read-lock would be used.\n\nOK.\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 3 Feb 1998 23:45:53 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page\\"
}
] |
[
{
"msg_contents": "\n> On Tue, 3 Feb 1998, Andrew Martin wrote:\n> \n> > > > > > I'll say it again and again - glibc-2.0 is the _STANDARD_ (actually\n> > > > > > reference) platform for Unix. All Unix. Not just Linux.\n> > > > > > Adopted last year.\n> > > > > \n> > > > > \tAnd...how many Unix (other then Linux) are *actually* using it?\n> > > > > Any idea on how we can test whether it is being used or not?\n> > > > \n> > > > AFAIK some BSD's are now using it - but I am probably wrong. I gave up on\n> > > > BSD in '92.\n> > > \n> > > \tGeez, about the time I gave up on Linux and converted to *BSD\n> > > *grin*\n> > > \n> > No wonder you're so anti-Linux if you stopped using it in '92. I started with\n> > 0.99pl13 in November '93. It must have been around v0.2 in '92 and of course\n> > it wasn't stable then....\n\nLinux is a lot more stable. I've had an old 386 notebook running \nas a DNS server up for 519 days!\n\n> \n> \tChecking now...oops, you are right:\n> \n> Record created on 03-Jun-93.\n> \n> \tThat was when we created the DNS record for my first company,\n> which ran 2/3rds on Linux...so whatever version existed back there.\n> \n> \tIn Linux's favor here, though...even at that point, if you ran\n> Linux on a *non-networked* system (ie. we had some dialup lines into it,\n> but not networking)...that thing ran pretty much rock-solid. It wasn't\n> until we actually put some load on her that she wouldn't run for more then\n> a day or so at a time...\n> \n> \n> \n> \n\n* Michael J. Rogan, Network Administrator, 905-624-3020 *\n* Mark IV Industries, F-P Electronics & I.V.H.S. Divisions *\n* [email protected] [email protected] *\n",
"msg_date": "Tue, 3 Feb 1998 12:27:15 +0000",
"msg_from": "\"Michael J. Rogan\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question abou"
},
{
"msg_contents": "On Tue, 3 Feb 1998, Andrew Martin wrote:\n\n> > > > > I'll say it again and again - glibc-2.0 is the _STANDARD_ (actually\n> > > > > reference) platform for Unix. All Unix. Not just Linux.\n> > > > > Adopted last year.\n> > > > \n> > > > \tAnd...how many Unix (other then Linux) are *actually* using it?\n> > > > Any idea on how we can test whether it is being used or not?\n> > > \n> > > AFAIK some BSD's are now using it - but I am probably wrong. I gave up on\n> > > BSD in '92.\n> > \n> > \tGeez, about the time I gave up on Linux and converted to *BSD\n> > *grin*\n> > \n> No wonder you're so anti-Linux if you stopped using it in '92. I started with\n> 0.99pl13 in November '93. It must have been around v0.2 in '92 and of course\n> it wasn't stable then....\n\n\tChecking now...oops, you are right:\n\n Record created on 03-Jun-93.\n\n\tThat was when we created the DNS record for my first company,\nwhich ran 2/3rds on Linux...so whatever version existed back there.\n\n\tIn Linux's favor here, though...even at that point, if you ran\nLinux on a *non-networked* system (ie. we had some dialup lines into it,\nbut not networking)...that thing ran pretty much rock-solid. It wasn't\nuntil we actually put some load on her that she wouldn't run for more then\na day or so at a time...\n\n\n",
"msg_date": "Tue, 3 Feb 1998 10:07:47 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Tue, 3 Feb 1998, Michael J. Rogan wrote:\n\n> > > No wonder you're so anti-Linux if you stopped using it in '92. I started with\n> > > 0.99pl13 in November '93. It must have been around v0.2 in '92 and of course\n> > > it wasn't stable then....\n> \n> Linux is a lot more stable. I've had an old 386 notebook running \n> as a DNS server up for 519 days!\n\t\n\tGranted, and I had a FreeBSD system running PostgreSQL up for\nalmost 365 days before an employee accidentally hit the power switch on it\ninstead of the one beside it *shrug* I'm not going to switch from what\nworks for me just because Linux is better then it was 4 years\nago...someday I'd like to add a second system to my home \"network\" running\nLinux, but it wouldn't be my primary system *shrug*\n\n\n",
"msg_date": "Tue, 3 Feb 1998 12:45:05 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question abou"
}
] |
[
{
"msg_contents": "> >From include/storage/s_lock.h:\n> \n> #if defined(irix5)\n> \n> \tUsing tools/ccsym, what shoudl this be set to. There is no longer\n> any concept of a \"port\" at the Makefile level, and hasn't been for over a\n> month now...From henceforth, its either determined based on what configure\n> knows about, or the compiler itself...in this case, tools/ccsym will have\n> to be used to determine what the compiler defines internally and the\n> settings changed...\n> \n\nOK, I just ran ccsym (from PGSQL V6.1 - I know it's old...) using the native\nIrix cc and with gcc. (See attached results)\n\nI guess __sgi or _SGI_SOURCE look like the best options.\n\n\nAndrew\n\n\nNative C compiler\n=================\n\nIrix5 Irix6\n----- -----\nLANGUAGE_C -Ac=xansi \nMIPSEB LANGUAGE_C \nSYSTYPE_SVR4 MIPSEB \n_ABIO32=1 _ABIN32=2 \n_CFE _COMPILER_VERSION=710 \n_LANGUAGE_C _LANGUAGE_C \n_LONGLONG _LONGLONG \n_MIPSEB _MIPSEB \n_MIPS_FPSET=16 _MIPS_FPSET=32 \n_MIPS_ISA=1 _MIPS_ISA=4 \n_MIPS_SIM=_ABIO32 _MIPS_SIM=_ABIN32 \n_MIPS_SZINT=32 _MIPS_SZINT=32 \n_MIPS_SZLONG=32 _MIPS_SZLONG=32 \n_MIPS_SZPTR=32 _MIPS_SZPTR=32 \n_MODERN_C _MODERN_C \n_PIC _PIC \n_SGI_SOURCE _SGI_SOURCE \n_SVR4_SOURCE _SVR4_SOURCE \n_SYSTYPE_SVR4 _SYSTYPE_SVR4 \n__DSO__ __DSO__ \n__EXTENSIONS__ __EXTENSIONS__ \n__INLINE_INTRINSICS __INLINE_INTRINSICS \n__STDC__=1 __MATH_HAS_NO_SIDE_EFFECTS \n__host_mips __host_mips \n__mips=1 __mips=4 \n__sgi __sgi \n__unix __unix \nhost_mips host_mips \nmachine=mips machine=mips \nmips mips \nsgi sgi \nsystem=unix system=unix \nunix unix \n\n\nGCC\n===\n\nIrix5 Irix6\n----- -----\nLANGUAGE_C LANGUAGE_C \nMIPSEB MIPSEB \nSYSTYPE_SVR4 SYSTYPE_SVR4 \n_LANGUAGE_C _LANGUAGE_C \n_LONGLONG _LONGLONG \n_MIPSEB _MIPSEB \n_MIPSEB _MIPSEB \n_MIPS_FPSET=16 _MIPS_FPSET=16 \n_MIPS_ISA=_MIPS_ISA_MIPS1 _MIPS_ISA=_MIPS_ISA_MIPS1 \n_MIPS_SIM=_MIPS_SIM_ABI32 _MIPS_SIM=_MIPS_SIM_ABI32 \n_MIPS_SIM=_MIPS_SIM_ABI32 _MIPS_SIM=_MIPS_SIM_ABI32 \n_MIPS_SZINT=32 _MIPS_SZINT=32 \n_MIPS_SZLONG=32 _MIPS_SZLONG=32 \n_MIPS_SZPTR=32 _MIPS_SZPTR=32 \n_MIPS_SZPTR=32 _MIPS_SZPTR=32 \n_MODERN_C _MODERN_C \n_MODERN_C _MODERN_C \n_SGI_SOURCE _SGI_SOURCE \n_SVR4_SOURCE _SVR4_SOURCE \n_SVR4_SOURCE _SVR4_SOURCE \n__CHAR_UNSIGNED__ __CHAR_UNSIGNED__ \n__DSO__ __DSO__ \n__DSO__ __DSO__ \n__EXTENSIONS__ __EXTENSIONS__ \n__GNUC_MINOR__=7 __GNUC_MINOR__=7 \n__GNUC__=2 __GNUC__=2 \n__MIPSEB __MIPSEB \n__MIPSEB__ __MIPSEB__ \n__PTRDIFF_TYPE__=int __PTRDIFF_TYPE__=int \n__SIZE_TYPE__=unsigned __SIZE_TYPE__=unsigned \n__SYSTYPE_SVR4 __SYSTYPE_SVR4 \n__SYSTYPE_SVR4__ __SYSTYPE_SVR4__ \n__host_mips __host_mips \n__host_mips__ __host_mips__ \n__mips __mips \n__mips__ __mips__ \n__sgi __sgi \n__sgi__ __sgi__ \n__unix __unix \n__unix__ __unix__ \ncpu=mips cpu=mips \nhost_mips host_mips \nmachine=sgi machine=sgi \nmips mips \nsgi sgi \nsystem=svr4 system=svr4 \nsystem=unix system=unix \nunix unix \n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Tue, 3 Feb 1998 14:36:53 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] snapshot won't compile on irix6.2"
},
{
"msg_contents": "On Tue, 3 Feb 1998, Andrew Martin wrote:\n\n> > >From include/storage/s_lock.h:\n> > \n> > #if defined(irix5)\n> > \n> > \tUsing tools/ccsym, what shoudl this be set to. There is no longer\n> > any concept of a \"port\" at the Makefile level, and hasn't been for over a\n> > month now...From henceforth, its either determined based on what configure\n> > knows about, or the compiler itself...in this case, tools/ccsym will have\n> > to be used to determine what the compiler defines internally and the\n> > settings changed...\n> > \n> \n> OK, I just ran ccsym (from PGSQL V6.1 - I know it's old...) using the native\n> Irix cc and with gcc. (See attached results)\n> \n> I guess __sgi or _SGI_SOURCE look like the best options.\n\n\tDone\n\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Wed, 4 Feb 1998 23:31:02 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] snapshot won't compile on irix6.2"
}
] |
[
{
"msg_contents": "\n> > > > I'll say it again and again - glibc-2.0 is the _STANDARD_ (actually\n> > > > reference) platform for Unix. All Unix. Not just Linux.\n> > > > Adopted last year.\n> > > \n> > > And...how many Unix (other then Linux) are *actually* using it?\n> > > Any idea on how we can test whether it is being used or not?\n> > \n> > AFAIK some BSD's are now using it - but I am probably wrong. I gave up on\n> > BSD in '92.\n> \n> Geez, about the time I gave up on Linux and converted to *BSD\n> *grin*\n> \nNo wonder you're so anti-Linux if you stopped using it in '92. I started with\n0.99pl13 in November '93. It must have been around v0.2 in '92 and of course\nit wasn't stable then....\n\nAndrew\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Tue, 3 Feb 1998 16:11:56 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
}
] |
[
{
"msg_contents": "> > > > > > I'll say it again and again - glibc-2.0 is the _STANDARD_ (actually\n> > > > > > reference) platform for Unix. All Unix. Not just Linux.\n> > > > > > Adopted last year.\n> > > > > \n> > > > > \tAnd...how many Unix (other then Linux) are *actually* using it?\n> > > > > Any idea on how we can test whether it is being used or not?\n> > > > \n> > > > AFAIK some BSD's are now using it - but I am probably wrong. I gave up on\n> > > > BSD in '92.\n> > > \n> > > \tGeez, about the time I gave up on Linux and converted to *BSD\n> > > *grin*\n> > > \n> > No wonder you're so anti-Linux if you stopped using it in '92. I started with\n> > 0.99pl13 in November '93. It must have been around v0.2 in '92 and of course\n> > it wasn't stable then....\n> \n> \tChecking now...oops, you are right:\n> \n> Record created on 03-Jun-93.\n\nStill can't have been much beyond 0.8\n\nAt last I understand why Marc is anti-Linux :-)\n\n\n> \n> \tThat was when we created the DNS record for my first company,\n> which ran 2/3rds on Linux...so whatever version existed back there.\n> \n> \tIn Linux's favor here, though...even at that point, if you ran\n> Linux on a *non-networked* system (ie. we had some dialup lines into it,\n> but not networking)...that thing ran pretty much rock-solid. It wasn't\n> until we actually put some load on her that she wouldn't run for more then\n> a day or so at a time...\n\nFAINTS....... :-)\n\n\nAndrew\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Tue, 3 Feb 1998 16:14:23 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
},
{
"msg_contents": "On Tue, 3 Feb 1998, Andrew Martin wrote:\n\n> \n> Still can't have been much beyond 0.8\n\n\tActually, I believe we were running v0.99pl++\n\n\n",
"msg_date": "Tue, 3 Feb 1998 11:50:53 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] (: JDBC+(Sun ~3:pm MST) CVS :) -also question about\n\tregression tests"
}
] |
[
{
"msg_contents": "In doing some work with the PostgreSQL ODBC driver today I noticed that I\nwas unable to connect to any of my database (how long has this been a\nproblem, I'm not sure). I then tried to use psql to connect from outside\nthe local machine. and I get the following errors on the server...\n\tConnStartup: No Postgres username specified in startup packet.\n\tand the psql errors out with \n\t-- couldn't recieve un/salt packet: errno=61\n\n\tWith the odbc driver after I send the login info I wait for a\nresponse and finally get a character value 205 back... \n\n\tI am afraid I've done something really silly but can't find out\nwhere I went wrong.\n\nJulie\n\n-- \n[ Julia Anne Case ] [ Ships are safe inside the harbor, ]\n[Programmer at large] [ but is that what ships are really for. ] \n[ Admining Linux ] [ To thine own self be true. ]\n[ Windows/WindowsNT ] [ Fair is where you take your cows to be judged. ]\n",
"msg_date": "Tue, 3 Feb 1998 14:25:04 -0500",
"msg_from": "Julia Anne Case <[email protected]>",
"msg_from_op": true,
"msg_subject": "connection troubles "
},
{
"msg_contents": "Quoting Julia Anne Case ([email protected]):\n> \tWith the odbc driver after I send the login info I wait for a\n> response and finally get a character value 205 back... \n> \n\tI've also found if I kill my debuging process during the wait for\nresponse from the server it often crashes the server... \n\nJulie\n\nBTW, this is using 6.2.1\n\n-- \n[ Julia Anne Case ] [ Ships are safe inside the harbor, ]\n[Programmer at large] [ but is that what ships are really for. ] \n[ Admining Linux ] [ To thine own self be true. ]\n[ Windows/WindowsNT ] [ Fair is where you take your cows to be judged. ]\n",
"msg_date": "Wed, 4 Feb 1998 13:19:29 -0500",
"msg_from": "Julia Anne Case <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] connection troubles"
},
{
"msg_contents": "On Wed, 4 Feb 1998, Julia Anne Case wrote:\n\n> Quoting Julia Anne Case ([email protected]):\n> > \tWith the odbc driver after I send the login info I wait for a\n> > response and finally get a character value 205 back... \n> > \n> \tI've also found if I kill my debuging process during the wait for\n> response from the server it often crashes the server... \n> \n> Julie\n> \n> BTW, this is using 6.2.1\n\nJulie...\n\n\tWill PostODBC be v6.3 tested in the next little while? I know\nthere have been *alot* of changes to v6.3 as far as authentication is\nconcerned, which might prove a problem once we release...\n\n\n",
"msg_date": "Wed, 4 Feb 1998 13:26:58 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] connection troubles"
},
{
"msg_contents": "Quoting The Hermit Hacker ([email protected]):\n> Julie...\n> \n> \tWill PostODBC be v6.3 tested in the next little while? I know\n> there have been *alot* of changes to v6.3 as far as authentication is\n> concerned, which might prove a problem once we release...\n> \n\tI just brought down the latest snapshot to my Sparc/Linux\nmachine... I'll let you know how it works out... Could take a little\nwhile to compile it's just a IPX laptop...\n\nJulie\n\n\n-- \n[ Julia Anne Case ] [ Ships are safe inside the harbor, ]\n[Programmer at large] [ but is that what ships are really for. ] \n[ Admining Linux ] [ To thine own self be true. ]\n[ Windows/WindowsNT ] [ Fair is where you take your cows to be judged. ]\n",
"msg_date": "Wed, 4 Feb 1998 13:53:21 -0500",
"msg_from": "Julia Anne Case <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] connection troubles"
},
{
"msg_contents": "On Wed, 4 Feb 1998, Julia Anne Case wrote:\n\n> Quoting The Hermit Hacker ([email protected]):\n> > Julie...\n> > \n> > \tWill PostODBC be v6.3 tested in the next little while? I know\n> > there have been *alot* of changes to v6.3 as far as authentication is\n> > concerned, which might prove a problem once we release...\n> > \n> \tI just brought down the latest snapshot to my Sparc/Linux\n> machine... I'll let you know how it works out... Could take a little\n> while to compile it's just a IPX laptop...\n\t\n\n\tI don't have any knowledge of ODBC, but would it help if I setup\nand maintained a (reasonably) current server that you can connect remotely\ntoo for testing? I'll do periodic recompile/installs of it so that it\nstays up to date? I'd like to see the ODBC as rock solid as possible\nbefore the 1st, in accordance with v6.3 specs, and if it means doing this,\nits a cheap price to pay :)\n\n\tPeter...would this help any with the JDBC stuff too, or are you\nokay with that one?\n\n\n\n",
"msg_date": "Wed, 4 Feb 1998 13:58:50 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] connection troubles"
},
{
"msg_contents": "On Wed, 4 Feb 1998, The Hermit Hacker wrote:\n\n> On Wed, 4 Feb 1998, Julia Anne Case wrote:\n> \tI don't have any knowledge of ODBC, but would it help if I setup\n> and maintained a (reasonably) current server that you can connect remotely\n> too for testing? I'll do periodic recompile/installs of it so that it\n> stays up to date? I'd like to see the ODBC as rock solid as possible\n> before the 1st, in accordance with v6.3 specs, and if it means doing this,\n> its a cheap price to pay :)\n> \n> \tPeter...would this help any with the JDBC stuff too, or are you\n> okay with that one?\n\nI'm ok here, as I keep the backend relatively up to date. It may be\nuseful to have one though, as it would allow us to test over a long\ndistance, and could highlight any problems that don't show up locally.\n\n-- \nPeter T Mount [email protected] or [email protected]\nMain Homepage: http://www.demon.co.uk/finder\nWork Homepage: http://www.maidstone.gov.uk Work EMail: [email protected]\n\n",
"msg_date": "Thu, 5 Feb 1998 15:15:38 +0000 (GMT)",
"msg_from": "Peter T Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] connection troubles"
}
] |
[
{
"msg_contents": "Another problem I found with the PL code was that it took the fmgr()\nmacro, and made it call a function call to fmgr_addr, which just killed\nperformance.\n\nI made fmgr_addr() a macro too.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 3 Feb 1998 14:57:46 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "PL code and fmgr_addr"
},
{
"msg_contents": "> \n> Another problem I found with the PL code was that it took the fmgr()\n> macro, and made it call a function call to fmgr_addr, which just killed\n> performance.\n> \n> I made fmgr_addr() a macro too.\n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n> \n\n Just to let you know - after fixing some other problems\n caused by the bpchar trouble my PLtcl tests went through\n again. Looks the macroization is O.K.\n\n\nUntil later, Jan\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] (Jan Wieck) #\n\n",
"msg_date": "Thu, 5 Feb 1998 16:55:16 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] PL code and fmgr_addr"
},
{
"msg_contents": "On Thu, 5 Feb 1998, Jan Wieck wrote:\n\n> > \n> > Another problem I found with the PL code was that it took the fmgr()\n> > macro, and made it call a function call to fmgr_addr, which just killed\n> > performance.\n> > \n> > I made fmgr_addr() a macro too.\n> > \n> > -- \n> > Bruce Momjian\n> > [email protected]\n> > \n> > \n> \n> Just to let you know - after fixing some other problems\n> caused by the bpchar trouble my PLtcl tests went through\n> again. Looks the macroization is O.K.\n\n\tDo we have regression tests for this?\n\n\n",
"msg_date": "Thu, 5 Feb 1998 10:57:45 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] PL code and fmgr_addr"
},
{
"msg_contents": ">\n> On Thu, 5 Feb 1998, Jan Wieck wrote:\n>\n> > >\n> > > Another problem I found with the PL code was that it took the fmgr()\n> > > macro, and made it call a function call to fmgr_addr, which just killed\n> > > performance.\n> > >\n> > > I made fmgr_addr() a macro too.\n> > >\n> > > --\n> > > Bruce Momjian\n> > > [email protected]\n> > >\n> > >\n> >\n> > Just to let you know - after fixing some other problems\n> > caused by the bpchar trouble my PLtcl tests went through\n> > again. Looks the macroization is O.K.\n>\n> Do we have regression tests for this?\n>\n>\n>\n\n No - since PLtcl isn't part of the distribution up to now.\n And I think that even if we include PLtcl into the dist we\n shouldn't include it into the regression tests because\n building PLtcl requires a Tcl installation (at least the\n libtcl??.so and tclConfig.sh).\n\n But it would be O.K. for me to include the PLtcl to the\n contrib directory and setup a separate test suite there.\n\n\nUntil later, Jan\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] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 5 Feb 1998 17:13:36 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] PL code and fmgr_addr"
},
{
"msg_contents": "On Thu, 5 Feb 1998, Jan Wieck wrote:\n\n> >\n> > On Thu, 5 Feb 1998, Jan Wieck wrote:\n> >\n> > > >\n> > > > Another problem I found with the PL code was that it took the fmgr()\n> > > > macro, and made it call a function call to fmgr_addr, which just killed\n> > > > performance.\n> > > >\n> > > > I made fmgr_addr() a macro too.\n> > > >\n> > > > --\n> > > > Bruce Momjian\n> > > > [email protected]\n> > > >\n> > > >\n> > >\n> > > Just to let you know - after fixing some other problems\n> > > caused by the bpchar trouble my PLtcl tests went through\n> > > again. Looks the macroization is O.K.\n> >\n> > Do we have regression tests for this?\n> >\n> >\n> >\n> \n> No - since PLtcl isn't part of the distribution up to now.\n> And I think that even if we include PLtcl into the dist we\n> shouldn't include it into the regression tests because\n> building PLtcl requires a Tcl installation (at least the\n> libtcl??.so and tclConfig.sh).\n> \n> But it would be O.K. for me to include the PLtcl to the\n> contrib directory and setup a separate test suite there.\n\n\tWait...didn't we just do a patch so that PLs could be used? Do we\nhave regression tests for that? LIke we have to triggers?\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Thu, 5 Feb 1998 19:03:09 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] PL code and fmgr_addr"
},
{
"msg_contents": ">\n> On Thu, 5 Feb 1998, Jan Wieck wrote:\n>\n> > >\n> > > On Thu, 5 Feb 1998, Jan Wieck wrote:\n> > >\n> > > > >\n> > > > > Another problem I found with the PL code was that it took the fmgr()\n> > > > > macro, and made it call a function call to fmgr_addr, which just killed\n> > > > > performance.\n> > > > >\n> > > > > I made fmgr_addr() a macro too.\n> > > > >\n> > > > > --\n> > > > > Bruce Momjian\n> > > > > [email protected]\n> > > > >\n> > > > >\n> > > >\n> > > > Just to let you know - after fixing some other problems\n> > > > caused by the bpchar trouble my PLtcl tests went through\n> > > > again. Looks the macroization is O.K.\n> > >\n> > > Do we have regression tests for this?\n> > >\n> > >\n> > >\n> >\n> > No - since PLtcl isn't part of the distribution up to now.\n> > And I think that even if we include PLtcl into the dist we\n> > shouldn't include it into the regression tests because\n> > building PLtcl requires a Tcl installation (at least the\n> > libtcl??.so and tclConfig.sh).\n> >\n> > But it would be O.K. for me to include the PLtcl to the\n> > contrib directory and setup a separate test suite there.\n>\n> Wait...didn't we just do a patch so that PLs could be used? Do we\n> have regression tests for that? LIke we have to triggers?\n\n Up to now we don't have a PL that is independent. The patch\n only prepared the backend for a generic PL interface (CREATE\n PROCEDURAL LANGUAGE and the fmgr changes that a handler is\n called for functions of that language) and I tested all that\n with PLtcl, which for sake isn't in the dist right now.\n\n The current regression tests only enshure that the slightly\n modified fmgr doesn't break anything that worked before. And\n I think this should be the state until we have a real\n PL/pgSQL.\n\n I have some ideas for a PL/pgSQL. And I will create a\n regression test for it while implementing.\n\n>\n> Marc G. Fournier\n> Systems Administrator @ hub.org\n> primary: [email protected] secondary: scrappy@{freebsd|postgresql}.org\n>\n>\n>\n\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] (Jan Wieck) #\n\n\n",
"msg_date": "Fri, 6 Feb 1998 11:15:58 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] PL code and fmgr_addr"
}
] |
[
{
"msg_contents": "Hi All,\n\nJust attempted to configure the latest CVS on S/Linux and got the\nfollowing error:-\n\ncreating include/config.h\nlinking ./backend/port/tas/linux.s to backend/port/tas.s\nconfigure: error: ./backend/port/tas/linux.s: File not found\n\nJust touching the file backend/port/tas/linux.s seems to have gotten\nme through the configure for now.\n\nI've just started a compile.....\n\n.\n.\n\nCompleted OK, on to the regression tests...\n\n\nKeith.\n\n",
"msg_date": "Tue, 3 Feb 1998 23:36:00 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "Snapshot 030298 on SPARC Linux."
}
] |
[
{
"msg_contents": "Thomas G. Lockhart <[email protected]>\n> Keith Parks <[email protected]>\n> \n> > I'm getting the following failure in the \"strings\" regression test:-\n> \n> What is the date/time on your snapshot? I ran a fresh build yesterday\n> from the development tree as of about 980202 15:20UTC. Didn't see the\n> problem...\n\nFalse alarm, I've just found this message on [COMMITTERS].\n\nKeith.\n\n------------------------\nDate: Sun, 1 Feb 1998 21:12:38 -0500 (EST)\nFrom: \"Marc G. Fournier\" <[email protected]>\nTo: [email protected]\nSubject: [COMMITTERS] 'pgsql/src/backend/parser parse_func.c'\n\nUpdate of /usr/local/cvsroot/pgsql/src/backend/parser\nIn directory hub.org:/tmp/cvs-serv20139/parser\n\nModified Files:\n\tparse_func.c \nLog Message:\n\nset \"caller\" value in parse_func.h to func_get_detail instead of just \"\"\n\n---------------------------\n\n\n> \n> - Tom\n> \n> > ====== strings ======\n> > 15c15\n> > < ERROR: func_get_detail: function text(bpchar) does not exist\n> > ---\n> > > ERROR: : function text(bpchar) does not exist\n> > 17c17\n> > < ERROR: func_get_detail: function text(varchar) does not exist\n> > ---\n> > > ERROR: : function text(varchar) does not exist\n> >\n> > This message comes from a call to func_error() but the 1st arg to the\n> > call has changed from \"func_get_detail\" to \"\".\n> >\n> > I can only think this change is deliberate, but why.\n> >\n> > All the code is in backend/parser/parse_func.c\n> >\n> > Thanks,\n> > Keith.\n> \n> \n> \n\n",
"msg_date": "Wed, 4 Feb 1998 02:15:19 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] regression test \"strings\" failure."
}
] |
[
{
"msg_contents": "> \n> > There are several things I have attempted to do, that my reference\n> > book says are possible, that do not seem to function. To wit:\n> >\n> > wwwlogs=> select * from dates where '1998-01-06 15:59:59' between dt1\n> > and dt2;\n> > WARN:parser: parse error at or near \"dt1\"\n> \n> Oh! None of us even thought of writing an expression like that.\n> Currently, the parser requires that only constants follow BETWEEN.\n> Bruce, I have a patch which will allow the above expression (column\n> names in the expression) but I think we will have trouble allowing full\n> expressions there since the AND looks like a boolean expression. I'll go\n> ahead and apply my patches soon (tomorrow?) unless you want something\n> different...\n\nGo head.\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 3 Feb 1998 21:36:11 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [QUESTIONS] timestamp questions"
}
] |
[
{
"msg_contents": "Hi,\n\ndespite bein chronically short of time I'd like to be involved with some\ndatabase development in free software. It seems to me that postgresql is\nthe most complete package out there and it also has some nice features\nlike object identity. I consider myself fairly knowledgable when it\ncomes to databases (at least it brought me my Ph.D.) so I think I could\ndo some. But before I do so I'd like to ask some questions about the\nstatus of development:\n\n1) Is there an embedded SQL preprocessor available? I read about an\nalpha version, but is it still maintained?\n2) How about ODBC? I know there is a driver but apparently I cannot use\nall ODBC features. I have an application on Windows that I'd like to use\nto stress test postgresql a little bit, but I get an SQLError -1\neverytime I try to fetch a record. Is this a knwon bug?\n3) How would you describe the general goal of this project: Offer as\ngood a database system as you can create as free software? That is\nshould it be able to handle huge amounts of data, or is it seen more as\na tool for a private person/small business?\n4) Are there areas/Which ares need help?\n\nPlease tell me what you think\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n\n\n",
"msg_date": "Wed, 04 Feb 1998 12:49:27 +0100",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": true,
"msg_subject": "Hi"
},
{
"msg_contents": "On Wed, 4 Feb 1998, Michael Meskes wrote:\n\n> Hi,\n> \n> despite bein chronically short of time I'd like to be involved with some\n> database development in free software. It seems to me that postgresql is\n> the most complete package out there and it also has some nice features\n> like object identity. I consider myself fairly knowledgable when it\n> comes to databases (at least it brought me my Ph.D.) so I think I could\n> do some. But before I do so I'd like to ask some questions about the\n> status of development:\n\n\tWelcome :)\n\n> 2) How about ODBC? I know there is a driver but apparently I cannot use\n> all ODBC features. I have an application on Windows that I'd like to use\n> to stress test postgresql a little bit, but I get an SQLError -1\n> everytime I try to fetch a record. Is this a knwon bug?\n\n\tJulie (maintainer of PostODBC(src/interfaces/odbc)) has just\njoined this list, and will hopefully be able to answer more completely,\nbut the PostODBC driver is currently listed as v0.21, so is still young... \n\n\tAny aid you can provide in helping to mature it are most welcome,\nboth on our part, and, I imagine, Julie's...\n\n> 3) How would you describe the general goal of this project: Offer as\n> good a database system as you can create as free software? That is\n> should it be able to handle huge amounts of data, or is it seen more as\n> a tool for a private person/small business? \n\n\tBoth...We have at least two ppl so far, that I'm aware of, that\nare dealing with systems of several GB of data through it, including one\nof the core developers...like *BSD/Linux \"competing\" with the more\nentrenched OSs like Solaris (and Microsloth), we're moving towards the\nconcept of being a viable alternative to such monolithic RDBMS systems\nsuch as Oracle/Informix...but we are also working at making sure that we\ndon't loss site of the majority of our users, which tend to be ppl using\nit for WWW based projects.\n\n> 4) Are there areas/Which ares need help?\n\n\thttp://www.postgresql.org/docs/todo.shtml :)\n\n\tThere is a long list of things that are \"Wish list\" items...if you\ndecide to dive into something, all we ask is that you post something here\nso that others know and to prevent \"recreating the wheel\"...just in case\nsomeone else is already working on it.\n\n",
"msg_date": "Wed, 4 Feb 1998 08:05:50 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Hi"
},
{
"msg_contents": "> 1) Is there an embedded SQL preprocessor available? I read about an\n> alpha version, but is it still maintained?\n\nI remember mention of it on the list a while ago, but have not heard\nanything since. Also, no code for this was contributed for the\ndistribution, and it should be included. It would be _great_ to have a real\npreprocessor.\n\n> 3) How would you describe the general goal of this project: Offer as\n> good a database system as you can create as free software? That is\n> should it be able to handle huge amounts of data, or is it seen more as\n> a tool for a private person/small business?\n\nAs scrappy pointed out, we would like to not forget either kind of user.\nHowever, afaik PostgreSQL is the closest to a \"heavyweight\" RDBMS available\nin the freeware world, so if we had to choose we would probably put\nourselves on the \"huge amounts of data\" end of the spectrum.\n\n> 4) Are there areas/Which areas need help?\n\nThe ToDo has lots of projects. Also, as you get started and learn the\ndetails of Postgres internals the docs project could use some contributions\n:).\n\nThere are other areas which are not yet called out in the ToDo list,\nincluding\n\n - recovery of munged databases (I think disk full conditions may cause\ndamage)\n - transaction-only and session-only tables, variables, etc.\n - re-think the table/type interface to allow more extensibility of types\n\nBut, picking simpler items from the ToDo is probably the best way to get\nfamiliar with pieces of the code.\n\nWelcome.\n\n - Tom\n\n",
"msg_date": "Wed, 04 Feb 1998 14:29:07 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Hi"
},
{
"msg_contents": "> \n> Hi,\n> \n> despite bein chronically short of time I'd like to be involved with some\n> database development in free software. It seems to me that postgresql is\n> the most complete package out there and it also has some nice features\n> like object identity. I consider myself fairly knowledgable when it\n> comes to databases (at least it brought me my Ph.D.) so I think I could\n> do some. But before I do so I'd like to ask some questions about the\n> status of development:\n> \n\n> 3) How would you describe the general goal of this project: Offer as\n> good a database system as you can create as free software? That is\n> should it be able to handle huge amounts of data, or is it seen more as\n> a tool for a private person/small business?\n\nAbout a year ago, our goal was to make it more reliable. At this point,\nour goal is SQL compliance, speed, and features. See the TODO list in\nthe distribution or on our web page for an idea of our open items.\n\n> 4) Are there areas/Which ares need help?\n\nWhat we really need are more people who can get a handle on the entire\nsource tree, so we can implement some of the more complex SQL features,\nor improve some of the older code.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 4 Feb 1998 12:09:46 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Hi"
},
{
"msg_contents": "> About a year ago, our goal was to make it more reliable. At this point,\n> our goal is SQL compliance, speed, and features. See the TODO list in\n> the distribution or on our web page for an idea of our open items.\n\nGood summary Bruce. We haven't emphasized long term goals, since these\nshorter-term ones don't happen overnight. In the long run, though, other\ntopics of advanced interest, such as storage managers, distributed databases,\nlogging and error recovery, more OO features, adaptation to emerging\nstandards (SQL3 and beyond), certainly will be pertinent.\n\n> > 4) Are there areas/Which ares need help?\n>\n> What we really need are more people who can get a handle on the entire\n> source tree, so we can implement some of the more complex SQL features,\n> or improve some of the older code.\n\nWe have just recently gotten to the point where a very few of the current\ndevelopers have an understanding of _most_ of the backend code. Achieving\nlonger-term advances require that the development be planned by folks with\nthat level of understanding, though others can and do contribute\nsubstantially.\n\nEveryone starts on Postgres without understanding the code. As we do small\nprojects, we learn as we go and become more capable of dealing with larger\nproblems. Enjoy.\n\n - Tom\n\n",
"msg_date": "Wed, 04 Feb 1998 17:39:15 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Hi"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n>\n> - recovery of munged databases (I think disk full conditions may cause\n> damage)\n\nI have filled up the disk lots of times, but never had any problems\nwith corrupted data. When inserting, the backend simply says:\n\"mytable: cannot extend\" and the insert fails.\n\nThe only problem I've had is with SELECT and ORDER BY, where the\nsort files fill up the disk, and the backend either hangs or returns\nzero rows. This can of course cause serious problems if an application\ndoes something bad when a select returns zero rows (like re-initializing\nthe whole database or something... ooooh noooo...).\n\n/* m */\n",
"msg_date": "Thu, 05 Feb 1998 15:56:51 +0100",
"msg_from": "Mattias Kregert <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Hi"
}
] |
[
{
"msg_contents": "Okay. Sounds good. :-)\n\nHow do I get the latest source tree and how is it synchronised?\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tThe Hermit Hacker [SMTP:[email protected]]\n> Sent:\tWednesday, February 04, 1998 2:06 PM\n> To:\tMichael Meskes\n> Cc:\[email protected]\n> Subject:\tRe: [HACKERS] Hi\n> \n> On Wed, 4 Feb 1998, Michael Meskes wrote:\n> \n> > Hi,\n> > \n> > despite bein chronically short of time I'd like to be involved with\n> some\n> > database development in free software. It seems to me that\n> postgresql is\n> > the most complete package out there and it also has some nice\n> features\n> > like object identity. I consider myself fairly knowledgable when it\n> > comes to databases (at least it brought me my Ph.D.) so I think I\n> could\n> > do some. But before I do so I'd like to ask some questions about the\n> > status of development:\n> \n> \tWelcome :)\n> \n> > 2) How about ODBC? I know there is a driver but apparently I cannot\n> use\n> > all ODBC features. I have an application on Windows that I'd like to\n> use\n> > to stress test postgresql a little bit, but I get an SQLError -1\n> > everytime I try to fetch a record. Is this a knwon bug?\n> \n> \tJulie (maintainer of PostODBC(src/interfaces/odbc)) has just\n> joined this list, and will hopefully be able to answer more\n> completely,\n> but the PostODBC driver is currently listed as v0.21, so is still\n> young... \n> \n> \tAny aid you can provide in helping to mature it are most\n> welcome,\n> both on our part, and, I imagine, Julie's...\n> \n> > 3) How would you describe the general goal of this project: Offer as\n> > good a database system as you can create as free software? That is\n> > should it be able to handle huge amounts of data, or is it seen more\n> as\n> > a tool for a private person/small business? \n> \n> \tBoth...We have at least two ppl so far, that I'm aware of, that\n> are dealing with systems of several GB of data through it, including\n> one\n> of the core developers...like *BSD/Linux \"competing\" with the more\n> entrenched OSs like Solaris (and Microsloth), we're moving towards the\n> concept of being a viable alternative to such monolithic RDBMS systems\n> such as Oracle/Informix...but we are also working at making sure that\n> we\n> don't loss site of the majority of our users, which tend to be ppl\n> using\n> it for WWW based projects.\n> \n> > 4) Are there areas/Which ares need help?\n> \n> \thttp://www.postgresql.org/docs/todo.shtml :)\n> \n> \tThere is a long list of things that are \"Wish list\" items...if\n> you\n> decide to dive into something, all we ask is that you post something\n> here\n> so that others know and to prevent \"recreating the wheel\"...just in\n> case\n> someone else is already working on it.\n> \n",
"msg_date": "Wed, 4 Feb 1998 14:31:15 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Hi"
},
{
"msg_contents": "On Wed, 4 Feb 1998, Meskes, Michael wrote:\n\n> Okay. Sounds good. :-)\n> \n> How do I get the latest source tree and how is it synchronised?\n\n\tWe recommend that everyone doing development work use or grab\nCVSup (see ftp://ftp.postgresql.org/pub/CVSup for details and some\nbinaries), which is the fastest/most uptodate way of grabbing the sources. \n\n\tRight now, being in beta mode, there is a daily snapshot being\nmade and placed in ftp://ftp.postgresql.org/pub/postgresql.snapshot.tar.gz\n\n\tAfter v6.4 is released, that will go back down to weekly snapshots\nvs daily\n\n\n\n > \n> Michael\n> \n> --\n> Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> [email protected] | Europark A2, Adenauerstr. 20\n> [email protected] | 52146 Wuerselen\n> Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n> \n> > -----Original Message-----\n> > From:\tThe Hermit Hacker [SMTP:[email protected]]\n> > Sent:\tWednesday, February 04, 1998 2:06 PM\n> > To:\tMichael Meskes\n> > Cc:\[email protected]\n> > Subject:\tRe: [HACKERS] Hi\n> > \n> > On Wed, 4 Feb 1998, Michael Meskes wrote:\n> > \n> > > Hi,\n> > > \n> > > despite bein chronically short of time I'd like to be involved with\n> > some\n> > > database development in free software. It seems to me that\n> > postgresql is\n> > > the most complete package out there and it also has some nice\n> > features\n> > > like object identity. I consider myself fairly knowledgable when it\n> > > comes to databases (at least it brought me my Ph.D.) so I think I\n> > could\n> > > do some. But before I do so I'd like to ask some questions about the\n> > > status of development:\n> > \n> > \tWelcome :)\n> > \n> > > 2) How about ODBC? I know there is a driver but apparently I cannot\n> > use\n> > > all ODBC features. I have an application on Windows that I'd like to\n> > use\n> > > to stress test postgresql a little bit, but I get an SQLError -1\n> > > everytime I try to fetch a record. Is this a knwon bug?\n> > \n> > \tJulie (maintainer of PostODBC(src/interfaces/odbc)) has just\n> > joined this list, and will hopefully be able to answer more\n> > completely,\n> > but the PostODBC driver is currently listed as v0.21, so is still\n> > young... \n> > \n> > \tAny aid you can provide in helping to mature it are most\n> > welcome,\n> > both on our part, and, I imagine, Julie's...\n> > \n> > > 3) How would you describe the general goal of this project: Offer as\n> > > good a database system as you can create as free software? That is\n> > > should it be able to handle huge amounts of data, or is it seen more\n> > as\n> > > a tool for a private person/small business? \n> > \n> > \tBoth...We have at least two ppl so far, that I'm aware of, that\n> > are dealing with systems of several GB of data through it, including\n> > one\n> > of the core developers...like *BSD/Linux \"competing\" with the more\n> > entrenched OSs like Solaris (and Microsloth), we're moving towards the\n> > concept of being a viable alternative to such monolithic RDBMS systems\n> > such as Oracle/Informix...but we are also working at making sure that\n> > we\n> > don't loss site of the majority of our users, which tend to be ppl\n> > using\n> > it for WWW based projects.\n> > \n> > > 4) Are there areas/Which ares need help?\n> > \n> > \thttp://www.postgresql.org/docs/todo.shtml :)\n> > \n> > \tThere is a long list of things that are \"Wish list\" items...if\n> > you\n> > decide to dive into something, all we ask is that you post something\n> > here\n> > so that others know and to prevent \"recreating the wheel\"...just in\n> > case\n> > someone else is already working on it.\n> > \n> \n\n",
"msg_date": "Wed, 4 Feb 1998 08:42:11 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] Hi"
},
{
"msg_contents": "> \n> Okay. Sounds good. :-)\n> \n> How do I get the latest source tree and how is it synchronised?\n> \n\nHere is the developers FAQ. It will be on our web page soon, too.\n\n---------------------------------------------------------------------------\nDevelopers Frequently Asked Questions (FAQ) for PostgreSQL\n\nLast updated: Wed Feb 4 12:43:43 EST 1998\n\nCurrent maintainer: Bruce Momjian ([email protected])\n\nThe most recent version of this document can be viewed at the postgreSQL Web\nsite, http://postgreSQL.org.\n\n ------------------------------------------------------------------------\n\nQuestions answered:\n\n1) What tools are available for developers?\n2) What books are good for developers?\n3) Why do we use palloc() and pfree() to allocate memory?\n4) Why do we use Node and List to make data structures?\n5) How do I add a feature or fix a bug?\n6) How do I download/update the current source tree?\n ------------------------------------------------------------------------\n\n1) What tools are available for developers?\n\nAside from the User documentation mentioned in the regular FAQ, there are\nseveral development tools available. First, all the files in the /tools\ndirectory are designed for developers.\n\n RELEASE_CHANGES changes we have to make for each release\n SQL_keywords standard SQL'92 keywords\n backend web flowchart of the backend directories\n ccsym find standard defines made by your compiler\n entab converts tabs to spaces, used by pgindent\n find_static finds functions that could be made static\n find_typedef get a list of typedefs in the source code\n make_ctags make vi 'tags' file in each directory\n make_diff make *.orig and diffs of source\n make_etags make emacs 'etags' files\n make_keywords.README make comparison of our keywords and SQL'92\n make_mkid make mkid ID files\n mkldexport create AIX exports file\n pgindent indents C source files\n\nLet me note some of these. If you point your browser at the tools/backend\ndirectory, you will see all the backend components in a flow chart. You can\nclick on any one to see a description. If you then click on the directory\nname, you will be taken to the source directory, to browse the actual source\ncode behind it. We also have several README files in some source directories\nto describe the function of the module. The browser will display these when\nyou enter the directory also. The tools/backend directory is also contained\non our web page under the title Backend Flowchart.\n\nSecond, you really should have an editor that can handle tags, so you can\ntag a function call to see the function definition, and then tag inside that\nfunction to see an even lower-level function, and then back out twice to\nreturn to the original function. Most editors support this via tags or etags\nfiles.\n\nThird, you need to get mkid from ftp.postgresql.org. By running\ntools/make_mkid, an archive of source symbols can be created that can be\nrapidly queried like grep or edited.\n\nmake_diff has tools to create patch diff files that can be applied to the\ndistribution.\n\npgindent will format source files to match our standard format, which has\nfour-space tabs, and an indenting format specified by flags to the your\noperating system's utility indent.\n\n2) What books are good for developers?\n\nI have two good books, An Introduction to Database Systems, by C.J. Date,\nAddison, Wesley and A Guide to the SQL Standard, by C.J. Date, et. al,\nAddison, Wesley.\n\n3) Why do we use palloc() and pfree() to allocate memory?\n\npalloc() and pfree() are used in place of malloc() and free() because we\nautomatically free all memory allocated when a transaction completes. This\nmakes it easier to make sure we free memory that gets allocated in one\nplace, but only freed much later. There are several contexts that memory can\nbe allocated in, and this controls when the allocated memory is\nautomatically freed by the backend.\n\n4) Why do we use Node and List to make data structures?\n\nWe do this because this allows a consistent way to pass data inside the\nbackend in a flexible way. Every node has a NodeTag which specifies what\ntype of data is inside the Node. Lists are lists of Nodes. lfirst(),\nlnext(), and foreach() are used to get, skip, and traverse through Lists.\n\n5) How do I add a feature or fix a bug?\n\nThe source code is over 250,000 lines. Many problems/features are isolated\nto one specific area of the code. Others require knowledge of much of the\nsource. If you are confused about where to start, ask the hackers list, and\nthey will be glad to assess the complexity and give pointers on where to\nstart.\n\nAnother thing to keep in mind is that many fixes and features can be added\nwith surprisingly little code. I often start by adding code, then looking at\nother areas in the code where similar things are done, and by the time I am\nfinished, the patch is quite small and compact.\n\nWhen adding code, keep in mind that it should use the existing facilities in\nthe source, for performance reasons and for simplicity. Often a review of\nexisting code doing similar things is helpful.\n\n6) How do I download/update the current source tree?\n\nThere are several ways to obtain the source tree. Occasional developers can\njust get the most recent source tree snapshot from ftp.postgresql.org. For\nregular developers, you can get cvsup, which is available from\nftp.postgresql.org too. cvsup allows you to download the source tree, then\noccasionally update your copy of the source tree with any new changes. Using\ncvsup, you don't have to download the entire source each time, only the\nchanged files. cvsup does not allow developers to update the source tree.\n\nTo update the source tree, there are two ways. You can generate a patch\nagainst your current source tree, perhaps using the make_diff tools\nmentioned above, and send them to the patches list. They will be reviewed,\nand applied in a timely manner. If the patch is major, and we are in beta\ntesting, the developers may wait for the final release before applying your\npatches.\n\nFor hard-core developers, Marc([email protected]) will give you a Unix\nshell account on postgresql.org, and you can ftp your files into your\naccount, patch, and cvs install the changes directly in the source tree.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 4 Feb 1998 12:47:32 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Hi"
},
{
"msg_contents": "Thanks to Neil, our webman, this is now on the web page, and will be\nupdated regularly.\n\n> \n> > \n> > Okay. Sounds good. :-)\n> > \n> > How do I get the latest source tree and how is it synchronised?\n> > \n> \n> Here is the developers FAQ. It will be on our web page soon, too.\n> \n> ---------------------------------------------------------------------------\n> Developers Frequently Asked Questions (FAQ) for PostgreSQL\n> \n> Last updated: Wed Feb 4 12:43:43 EST 1998\n> \n> Current maintainer: Bruce Momjian ([email protected])\n> \n> The most recent version of this document can be viewed at the postgreSQL Web\n> site, http://postgreSQL.org.\n> \n> ------------------------------------------------------------------------\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 4 Feb 1998 12:58:37 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Hi"
}
] |
[
{
"msg_contents": "\nAnyone out there know more about autoconf then I've been able to learn so\nfar? I've found *where* the problem is with the tas.s stuff, just haven't\nbeen able to find a clean way of fixing it :(\n\nif test \"X$need_tas\" = \"Xyes\"\nthen\n echo then do this?\n AC_LINK_FILES(backend/port/tas/${os}.s, backend/port/tas.s)\n TAS=tas.o\n AC_SUBST(TAS)\nfi\n\n\nIf you look at the above in configure (the above is from configure.in)\nafter running autoconf on it, it looks like:\n\nif test \"X$need_tas\" = \"Xyes\"\nthen\n echo then do this?\n\n TAS=tas.o\n\nfi\n\n\tThey tack the AC_LINK_FILES 'macro' and move it to the end of the\nfile, so, effectively, you can't conditionally add it...or so I've been\nable to determine. So, of course, the configure fails for those platforms\nthat don't have a tas.s file *sigh*\n\n\tNow, what I *could* do is put in a 'generic.s' file in that\nsubdirectory so that it has something to link, and then, because TAS isn't\nset, nothing would be attempted for the compile, but that feels sooooo\nmuch like a kludge to me :(\n\n\tComments?\n\n\n",
"msg_date": "Wed, 4 Feb 1998 08:34:08 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Configure question..."
}
] |
[
{
"msg_contents": "Okay, I'm just checking out sources.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tThe Hermit Hacker [SMTP:[email protected]]\n> Sent:\tWednesday, February 04, 1998 2:42 PM\n> To:\tMeskes, Michael\n> Cc:\[email protected]\n> Subject:\tRE: [HACKERS] Hi\n> \n> On Wed, 4 Feb 1998, Meskes, Michael wrote:\n> \n> > Okay. Sounds good. :-)\n> > \n> > How do I get the latest source tree and how is it synchronised?\n> \n> \tWe recommend that everyone doing development work use or grab\n> CVSup (see ftp://ftp.postgresql.org/pub/CVSup for details and some\n> binaries), which is the fastest/most uptodate way of grabbing the\n> sources. \n> \n> \tRight now, being in beta mode, there is a daily snapshot being\n> made and placed in\n> ftp://ftp.postgresql.org/pub/postgresql.snapshot.tar.gz\n> \n> \tAfter v6.4 is released, that will go back down to weekly\n> snapshots\n> vs daily\n> \n> \n> \n> > \n> > Michael\n> > \n> > --\n> > Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> > [email protected] | Europark A2, Adenauerstr.\n> 20\n> > [email protected] | 52146 Wuerselen\n> > Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> > Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n> > \n> > > -----Original Message-----\n> > > From:\tThe Hermit Hacker [SMTP:[email protected]]\n> > > Sent:\tWednesday, February 04, 1998 2:06 PM\n> > > To:\tMichael Meskes\n> > > Cc:\[email protected]\n> > > Subject:\tRe: [HACKERS] Hi\n> > > \n> > > On Wed, 4 Feb 1998, Michael Meskes wrote:\n> > > \n> > > > Hi,\n> > > > \n> > > > despite bein chronically short of time I'd like to be involved\n> with\n> > > some\n> > > > database development in free software. It seems to me that\n> > > postgresql is\n> > > > the most complete package out there and it also has some nice\n> > > features\n> > > > like object identity. I consider myself fairly knowledgable when\n> it\n> > > > comes to databases (at least it brought me my Ph.D.) so I think\n> I\n> > > could\n> > > > do some. But before I do so I'd like to ask some questions about\n> the\n> > > > status of development:\n> > > \n> > > \tWelcome :)\n> > > \n> > > > 2) How about ODBC? I know there is a driver but apparently I\n> cannot\n> > > use\n> > > > all ODBC features. I have an application on Windows that I'd\n> like to\n> > > use\n> > > > to stress test postgresql a little bit, but I get an SQLError -1\n> > > > everytime I try to fetch a record. Is this a knwon bug?\n> > > \n> > > \tJulie (maintainer of PostODBC(src/interfaces/odbc)) has just\n> > > joined this list, and will hopefully be able to answer more\n> > > completely,\n> > > but the PostODBC driver is currently listed as v0.21, so is still\n> > > young... \n> > > \n> > > \tAny aid you can provide in helping to mature it are most\n> > > welcome,\n> > > both on our part, and, I imagine, Julie's...\n> > > \n> > > > 3) How would you describe the general goal of this project:\n> Offer as\n> > > > good a database system as you can create as free software? That\n> is\n> > > > should it be able to handle huge amounts of data, or is it seen\n> more\n> > > as\n> > > > a tool for a private person/small business? \n> > > \n> > > \tBoth...We have at least two ppl so far, that I'm aware of, that\n> > > are dealing with systems of several GB of data through it,\n> including\n> > > one\n> > > of the core developers...like *BSD/Linux \"competing\" with the more\n> > > entrenched OSs like Solaris (and Microsloth), we're moving towards\n> the\n> > > concept of being a viable alternative to such monolithic RDBMS\n> systems\n> > > such as Oracle/Informix...but we are also working at making sure\n> that\n> > > we\n> > > don't loss site of the majority of our users, which tend to be ppl\n> > > using\n> > > it for WWW based projects.\n> > > \n> > > > 4) Are there areas/Which ares need help?\n> > > \n> > > \thttp://www.postgresql.org/docs/todo.shtml :)\n> > > \n> > > \tThere is a long list of things that are \"Wish list\" items...if\n> > > you\n> > > decide to dive into something, all we ask is that you post\n> something\n> > > here\n> > > so that others know and to prevent \"recreating the wheel\"...just\n> in\n> > > case\n> > > someone else is already working on it.\n> > > \n> > \n",
"msg_date": "Wed, 4 Feb 1998 15:30:11 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Hi"
},
{
"msg_contents": "On Wed, 4 Feb 1998, Meskes, Michael wrote:\n\n> Okay, I'm just checking out sources.\n\n\tDiscussions on development issues should be discussed here, in\npgsql-hackers, but any patches should be submitted to\npgsql-patches...other then...enjoy :)\n\n\n\n",
"msg_date": "Wed, 4 Feb 1998 09:39:56 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] Hi"
}
] |
[
{
"msg_contents": "I think I will start to get into the code slowly. But I agree that a\npreprocessor would be fine. I found some old 0.1 code on the internet.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tThomas G. Lockhart [SMTP:[email protected]]\n> Sent:\tWednesday, February 04, 1998 3:29 PM\n> To:\tMichael Meskes\n> Cc:\[email protected]\n> Subject:\tRe: [HACKERS] Hi\n> \n> > 1) Is there an embedded SQL preprocessor available? I read about an\n> > alpha version, but is it still maintained?\n> \n> I remember mention of it on the list a while ago, but have not heard\n> anything since. Also, no code for this was contributed for the\n> distribution, and it should be included. It would be _great_ to have a\n> real\n> preprocessor.\n> \n> > 3) How would you describe the general goal of this project: Offer as\n> > good a database system as you can create as free software? That is\n> > should it be able to handle huge amounts of data, or is it seen more\n> as\n> > a tool for a private person/small business?\n> \n> As scrappy pointed out, we would like to not forget either kind of\n> user.\n> However, afaik PostgreSQL is the closest to a \"heavyweight\" RDBMS\n> available\n> in the freeware world, so if we had to choose we would probably put\n> ourselves on the \"huge amounts of data\" end of the spectrum.\n> \n> > 4) Are there areas/Which areas need help?\n> \n> The ToDo has lots of projects. Also, as you get started and learn the\n> details of Postgres internals the docs project could use some\n> contributions\n> :).\n> \n> There are other areas which are not yet called out in the ToDo list,\n> including\n> \n> - recovery of munged databases (I think disk full conditions may\n> cause\n> damage)\n> - transaction-only and session-only tables, variables, etc.\n> - re-think the table/type interface to allow more extensibility of\n> types\n> \n> But, picking simpler items from the ToDo is probably the best way to\n> get\n> familiar with pieces of the code.\n> \n> Welcome.\n> \n> - Tom\n",
"msg_date": "Wed, 4 Feb 1998 15:31:12 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Hi"
}
] |
[
{
"msg_contents": "Are changes incorporated via patches? That means there's one person\nmerging all the patches?\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tThe Hermit Hacker [SMTP:[email protected]]\n> Sent:\tWednesday, February 04, 1998 3:40 PM\n> To:\tMeskes, Michael\n> Cc:\[email protected]\n> Subject:\tRE: [HACKERS] Hi\n> \n> On Wed, 4 Feb 1998, Meskes, Michael wrote:\n> \n> > Okay, I'm just checking out sources.\n> \n> \tDiscussions on development issues should be discussed here, in\n> pgsql-hackers, but any patches should be submitted to\n> pgsql-patches...other then...enjoy :)\n> \n> \n",
"msg_date": "Wed, 4 Feb 1998 15:41:51 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Hi"
},
{
"msg_contents": "On Wed, 4 Feb 1998, Meskes, Michael wrote:\n\n> Are changes incorporated via patches? That means there's one person\n> merging all the patches?\n\n\tGenerally, I keep on top of the patches (or try to), but Bruce,\nVadim and Thomas all have direct access to the CVS repository for applying\npatches and changes also..\n\n\n> \n> Michael\n> \n> --\n> Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> [email protected] | Europark A2, Adenauerstr. 20\n> [email protected] | 52146 Wuerselen\n> Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n> \n> > -----Original Message-----\n> > From:\tThe Hermit Hacker [SMTP:[email protected]]\n> > Sent:\tWednesday, February 04, 1998 3:40 PM\n> > To:\tMeskes, Michael\n> > Cc:\[email protected]\n> > Subject:\tRE: [HACKERS] Hi\n> > \n> > On Wed, 4 Feb 1998, Meskes, Michael wrote:\n> > \n> > > Okay, I'm just checking out sources.\n> > \n> > \tDiscussions on development issues should be discussed here, in\n> > pgsql-hackers, but any patches should be submitted to\n> > pgsql-patches...other then...enjoy :)\n> > \n> > \n> \n\n",
"msg_date": "Wed, 4 Feb 1998 09:52:12 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] Hi"
}
] |
[
{
"msg_contents": "> Is Postgres an SQL-based database or SQL is only an option?\n\nYes.\n\n> I'd uploaded 6.3 and I'm palying with it. Congratulations to Developers\n> PostgreSQL is now more SQL92-compliant.\n> But I can't understand one thing; seems that Postgres is SQL-based. (PostgreSQL)\n> ^^^\n> Is there a reason why developers implemnts new functions not SQL-compatible ?\n\nYes. There is more to SQL and ORDBMS than SQL87/89/92.\n\n> example:\n> In this release there's a very useful function LENGTH(), thanks to Thomas.\n> but it's not SQL. CHARACTER_LENGTH() or CHAR_LENGTH() this is SQL.\n\nPostgreSQL is an ORDBMS engine with an SQL front end. SQL92, which does not address\nthe possibility of type extensibility, tends to define type-specific functions,\nsuch as you mention, in a heavy and crude manner. We have other types for which\n\"length\" is an obvious useful quality; why put the type name into the function\nname? And why require two forms of the same function for every type??\n\nThe real implementation issue is this: for built-in functions, every function call\nmust currently have a unique name. For generic functions such as length, we define\na second \"sql function\" with the generic name which then refers to the built-in\nunique function name. For character types, we would need two more of these \"sql\nfunctions\" for each character type. That's four function definitions in pg_proc for\neach character type as opposed to the two definitions we currently have.\n\nYou raise a good point, however, in that we should provide the SQL92-compatible\nfunction name where possible (I think you have found one of the few cases where we\ndo not). Perhaps we can translate function names in the parser as we do for type\nnames? I'll look into it...\n\n> Maybe there's a reason why somebody needs to invent again the wheel but I cannot\n> understand it.\n\nWe are already working with a round wheel (well, at least ovoid), and are trying to\nprevent it from becoming square :)\n\n - Tom\n\n",
"msg_date": "Wed, 04 Feb 1998 14:54:36 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [QUESTIONS] is Postgres an SQL-based database?"
},
{
"msg_contents": "> > In this release there's a very useful function LENGTH(), thanks to Thomas.\n> > but it's not SQL. CHARACTER_LENGTH() or CHAR_LENGTH() this is SQL.\n> Perhaps we can translate function names in the parser as we do for type\n> names? I'll look into it...\n\nI have gram.y patches to do this. No visible effect on the regression execution speed,\nwhich thanks to Bruce is on my machine down to a record 2:23 elapsed execution time;\nthe last speeds were in the 2:31 range.\n\nI'm planning on bumping the gram.y cvs version to 2.0 for this change, due to the\n_large_ number of accumulated improvements. Any objections?\n\n - Tom\n\n",
"msg_date": "Wed, 04 Feb 1998 17:45:01 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] is Postgres an SQL-based database?"
},
{
"msg_contents": "> \n> > > In this release there's a very useful function LENGTH(), thanks to Thomas.\n> > > but it's not SQL. CHARACTER_LENGTH() or CHAR_LENGTH() this is SQL.\n> > Perhaps we can translate function names in the parser as we do for type\n> > names? I'll look into it...\n> \n> I have gram.y patches to do this. No visible effect on the regression execution speed,\n> which thanks to Bruce is on my machine down to a record 2:23 elapsed execution time;\n> the last speeds were in the 2:31 range.\n> \n> I'm planning on bumping the gram.y cvs version to 2.0 for this change, due to the\n> _large_ number of accumulated improvements. Any objections?\n\nGo ahead. Are there security problems?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 4 Feb 1998 12:57:48 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] is Postgres an SQL-based database?"
},
{
"msg_contents": "> > I have gram.y patches to do this. No visible effect on the regression execution speed,\n> > which thanks to Bruce is on my machine down to a record 2:23 elapsed execution time;\n> > the last speeds were in the 2:31 range.\n> >\n> > I'm planning on bumping the gram.y cvs version to 2.0 for this change, due to the\n> > _large_ number of accumulated improvements. Any objections?\n>\n> Go ahead. Are there security problems?\n\nSecurity problems with the patch? No, I don't think so. The only thing I'm doing is\nimplementing xlateSqlFunc() similarly to the existing xlateSqlType(), and defining a\n\"func_name\" state in place of just \"name\" which was used before.\n\nThe only downside is that backend functions implemented with names \"char_length\" and\n\"character_length\" will be masked by the parser translation so cannot be executed. In this\ncase I think that is OK, but I should put something in the new docs about that :)\n\n - Tom\n\n",
"msg_date": "Thu, 05 Feb 1998 01:52:08 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] is Postgres an SQL-based database?"
},
{
"msg_contents": "On Wed, 4 Feb 1998, Thomas G. Lockhart wrote:\n\n> > Is Postgres an SQL-based database or SQL is only an option?\n> \n> Yes.\n> \n> > I'd uploaded 6.3 and I'm palying with it. Congratulations to Developers\n> > PostgreSQL is now more SQL92-compliant.\n> > But I can't understand one thing; seems that Postgres is SQL-based. (PostgreSQL)\n> > ^^^\n> > Is there a reason why developers implemnts new functions not SQL-compatible ?\n> \n> Yes. There is more to SQL and ORDBMS than SQL87/89/92.\n\nI agree with you Tom, PostgreSQL should be more than standard,\nbut users expect that PostgreSQL at least supports SQL syntax.\nI agree with you if you talk about implement functions not supported by SQL\nbut I can't understand if you write the same function with another syntax.\nPostgreSQL has many functions that doesn't follow SQL-standard syntax.\nI can understand this:\n\n ROLLBACK [ WORK ] -- SQL-syntax\n rollback [transaction|work] -- PostgreSQL-syntax (this is more than standard)\n \nand\n\n ALTER TABLE <class_name> ADD [COLUMN] <attr> <type>; --SQL\n alter table <class_name> [*] add column <attr> <type>; --PostgreSQL\n\n In this case [*] show that PostgreSQL is an ORDBMS,\n but what about keyword COLUMN? It sould be optional.\n\nbut I can't understand things like:\n\n \\connect <dbname|-> <user> --PostgreSQL\ninsted of:\n CONNECT TO { DEFAULT | <SQL-server name> --SQL\n [ AS <connection name> ]\n [ USER <user name> ] }\nor\n\n CAST expression AS data-type --PostgreSQL\ninsted of:\n CAST ( expression AS { data-type | domain } ) --SQL\n\n\n> \n> > example:\n> > In this release there's a very useful function LENGTH(), thanks to Thomas.\n> > but it's not SQL. CHARACTER_LENGTH() or CHAR_LENGTH() this is SQL.\n> \n> PostgreSQL is an ORDBMS engine with an SQL front end. SQL92, which does not address\n> the possibility of type extensibility, tends to define type-specific functions,\n> such as you mention, in a heavy and crude manner. We have other types for which\n> \"length\" is an obvious useful quality; why put the type name into the function\n> name? And why require two forms of the same function for every type??\n> \n> The real implementation issue is this: for built-in functions, every function call\n> must currently have a unique name. For generic functions such as length, we define\n> a second \"sql function\" with the generic name which then refers to the built-in\n> unique function name. For character types, we would need two more of these \"sql\n> functions\" for each character type. That's four function definitions in pg_proc for\n> each character type as opposed to the two definitions we currently have.\n\nOk. That's a good reason. Thanks for your explanation. I understand now.\n\n> \n> You raise a good point, however, in that we should provide the SQL92-compatible\n> function name where possible (I think you have found one of the few cases where we\n> do not). Perhaps we can translate function names in the parser as we do for type\n> names? I'll look into it...\nGreat!\n> \n> > Maybe there's a reason why somebody needs to invent again the wheel but I cannot\n> > understand it.\n> \n> We are already working with a round wheel (well, at least ovoid), and are trying to\n> prevent it from becoming square :)\n> \nOk, you are doing an excellent work. Thanks for this great Database.\n\n Ciao, Jose'\n\n",
"msg_date": "Thu, 5 Feb 1998 14:34:52 +0000 (GMT)",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [QUESTIONS] is Postgres an SQL-based database?"
},
{
"msg_contents": "> I agree with you Tom, PostgreSQL should be more than standard,\n> but users expect that PostgreSQL at least supports SQL syntax.\n> I agree with you if you talk about implement functions not supported by SQL\n> but I can't understand if you write the same function with another syntax.\n\nWell, one explanation is that sometimes I get it wrong :)\n\n> PostgreSQL has many functions that doesn't follow SQL-standard syntax.\n> I can understand this:\n>\n> ROLLBACK [ WORK ] -- SQL-syntax\n> rollback [transaction|work] -- PostgreSQL-syntax (this is more than standard)\n>\n> and\n>\n> ALTER TABLE <class_name> ADD [COLUMN] <attr> <type>; --SQL\n> alter table <class_name> [*] add column <attr> <type>; --PostgreSQL\n>\n> In this case [*] show that PostgreSQL is an ORDBMS,\n> but what about keyword COLUMN? It sould be optional.\n>\n> but I can't understand things like:\n>\n> \\connect <dbname|-> <user> --PostgreSQL\n> insted of:\n> CONNECT TO { DEFAULT | <SQL-server name> --SQL\n> [ AS <connection name> ]\n> [ USER <user name> ] }\n\nConnections are done by the frontend, which does not parse the SQL. We could think about\nimplementing this if we allowed the backend to force a database change. Don't know if\nthis is easy or reasonably so.\n\n> CAST expression AS data-type --PostgreSQL\n> insted of:\n> CAST ( expression AS { data-type | domain } ) --SQL\n\nWow, I never noticed that parens are in the syntax! Will look at changing it...\n\n - Tom\n\n",
"msg_date": "Fri, 06 Feb 1998 15:30:41 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] is Postgres an SQL-based database?"
}
] |
[
{
"msg_contents": "\nIn article <[email protected]>,\nFrans van Dorsselaer <[email protected]> wrote:\n>Hello,\n>\n\n>I've always hesitated to send in bug reports, but after examining this\n> group for a while I think I'm now ready. \n\n> I'm running a developer release Linux 2.1.99, compiled with a developer\n> release gcc 2.8.13, which was compiled using itself. Both contain >\nnumerous >source bugs, as well as binary bugs due to the source bugs of >\nthe latter. I >use an alpha release Wine compiled with the very same >\ndeveloper release >compiler to run Windows binaries, each of which are per\ndefinition a >collection of bugs. Of course I run everything as root. \n\n>\n>Now I found a bug. Can somebody fix this ?\n\n I love this...have to add this to the PostgreSQL FAQ on how not to\npost for help :) Best example I've seen yet...\n\n",
"msg_date": "Wed, 4 Feb 1998 11:45:29 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Best problem report *I've* ever seen ... NAWT!!"
}
] |
[
{
"msg_contents": "Here is what I get from the current cvsup:\n\n---------------------------------------------------------------------------\n\n/usr/bin/yacc -d gram.y\n/usr/bin/yacc: f - maximum table size exceeded\ngmake[2]: *** [parse.h] Error 2\ngmake[2]: Leaving directory `/usr/local/src/pgsql/pgsql/src/backend/parser'\ngmake[1]: *** [parser.dir] Error 2\ngmake[1]: Leaving directory `/usr/local/src/pgsql/pgsql/src/backend'\ngmake: *** [all] Error 2\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 4 Feb 1998 13:19:57 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "new yacc problem"
},
{
"msg_contents": "> \n> Hi, \n> \n> GNU bison processes the current gram.y without objections...\n> \n\nGot bison. Fixed.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 4 Feb 1998 14:11:44 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] new yacc problem"
},
{
"msg_contents": "Hi, \n\nGNU bison processes the current gram.y without objections...\n\nbest regards,\n\nErnst\n",
"msg_date": "Wed, 4 Feb 1998 20:59:38 GMT",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] new yacc problem"
},
{
"msg_contents": "> Here is what I get from the current cvsup:\n>\n> ---------------------------------------------------------------------------\n>\n> /usr/bin/yacc -d gram.y\n> /usr/bin/yacc: f - maximum table size exceeded\n> gmake[2]: *** [parse.h] Error 2\n> gmake[2]: Leaving directory `/usr/local/src/pgsql/pgsql/src/backend/parser'\n> gmake[1]: *** [parser.dir] Error 2\n> gmake[1]: Leaving directory `/usr/local/src/pgsql/pgsql/src/backend'\n> gmake: *** [all] Error 2\n\nUh oh. Are there any switches on yacc which would let you increase some\ninternal parameters?\n\nIf not, we could try delivering gram.c as we do with scan.c. With the current\nparser scheme, I don't see how we could shrink things very much :<(\n\n - Tom\n\nI haven't yet tested with both our our recent changes in; will let you know if\nbison has trouble...\n\n",
"msg_date": "Thu, 05 Feb 1998 02:01:59 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: new yacc problem"
},
{
"msg_contents": "> \n> > Here is what I get from the current cvsup:\n> >\n> > ---------------------------------------------------------------------------\n> >\n> > /usr/bin/yacc -d gram.y\n> > /usr/bin/yacc: f - maximum table size exceeded\n> > gmake[2]: *** [parse.h] Error 2\n> > gmake[2]: Leaving directory `/usr/local/src/pgsql/pgsql/src/backend/parser'\n> > gmake[1]: *** [parser.dir] Error 2\n> > gmake[1]: Leaving directory `/usr/local/src/pgsql/pgsql/src/backend'\n> > gmake: *** [all] Error 2\n> \n> Uh oh. Are there any switches on yacc which would let you increase some\n> internal parameters?\n\nNope. I don't see any.\n\n> If not, we could try delivering gram.c as we do with scan.c. With the current\n> parser scheme, I don't see how we could shrink things very much :<(\n\nWould be a good idea.\n\n> I haven't yet tested with both our our recent changes in; will let you know if\n> bison has trouble...\n\nGot bison. Works fine.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 4 Feb 1998 22:27:18 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: new yacc problem"
}
] |
[
{
"msg_contents": "Compiling the latests snapshot on Sparc/Linux (RedHat 4.2/Kernel 2.0.30)\n\n\tI recall geting this with the 6.2.1 release too\n\tin backend/storaage/file/fd.c\n\t_SC_OPEN_MAX undefined\n\tI had just defaulted no_files to 255 and moved on last time...\nThought I might mention this now.\n\nJulie\n\n-- \n[ Julia Anne Case ] [ Ships are safe inside the harbor, ]\n[Programmer at large] [ but is that what ships are really for. ] \n[ Admining Linux ] [ To thine own self be true. ]\n[ Windows/WindowsNT ] [ Fair is where you take your cows to be judged. ]\n",
"msg_date": "Wed, 4 Feb 1998 15:39:54 -0500",
"msg_from": "Julia Anne Case <[email protected]>",
"msg_from_op": true,
"msg_subject": "Compile error ..."
}
] |
[
{
"msg_contents": "Hi,\n\nI have some questions about implementing additional features to the fmgr\ncode:\n\n1. How do I return a NULL value from a function. In fmgr.c I found out\nthat only the C-function with 1 parameter has an additional parameter\n&isNull which might be used to set the return value to a SQL NULL value.\nWhy not for functions with more paramaters?\n\n2. How do I use sets as a parameter to my code. E.g. in Illustra's API,\nwith which I'm fairly familiar, I wrote a function to send an email to\ncertain people in the following fashion:\n\nreturn sendmail((select email from table where condition), 'from email',\n'subject', 'body of email');\n\nOf course it would be possible to do something like this:\n\nselect\n sendmail(email, 'from email', 'subject', 'body of email')\nfrom\n table\nwhere\n condition;\n\nBut this would have a detrimental impact on the server, if it has to\nopen say 400 pipes to /usr/lib/sendmail before this query returns\n(probably my machine dies on this).\n\nThe function definition in Illustra would be something like:\n\ncreate function sendmail(setof(text), text, text, text) returns integer\nas external name '/path/to/sendmail.so';\n\n3. Can I return sets from a function, e.g. select * from range(1,6)\nwould give me six rows?\n\n(Again, in Illustra: create function range(integer, integer) returns\nsetof(integer) as external '/path/to/range.so').\n\nCan this be done using PostgreSQL's API and if not where do I have to\nstart to try implementing this?\n\n4. And finally: I hardly dare to ask this one: setof(user defined type)?\n\nPlease let me know if you have any information that I can use.\nIf you want me to elaborate some more on Illustra's API (which is quite\ngood on implementing sets except for naming conventions) please let me\nknow.\n\nCheers,\n\nJeroen van Vianen\n",
"msg_date": "Wed, 04 Feb 1998 22:53:11 +0100",
"msg_from": "Jeroen van Vianen <[email protected]>",
"msg_from_op": true,
"msg_subject": "User-defined functions with NULL values and sets as parameter and/or\n\treturn type"
},
{
"msg_contents": "> 1. How do I return a NULL value from a function. In fmgr.c I found out\n> that only the C-function with 1 parameter has an additional parameter\n> &isNull which might be used to set the return value to a SQL NULL value.\n> Why not for functions with more paramaters?\n\nFor functions which return a pointer, why not just check for a null\npointer? At the moment, returning a null pointer crashes the backend, but\nit would seem to be fairly easy to fix since it sort-of handles null\nreturns as you found.\n\nI was planning on looking at this someday, but am far away from getting to\nit. It's all yours :)\n\n> Please let me know if you have any information that I can use.\n> If you want me to elaborate some more on Illustra's API (which is quite\n> good on implementing sets except for naming conventions) please let me\n> know.\n\nI'd be interested in knowing more about Illustra's features. Since we are\nstarting with v6.3 testing and bug fixing, you might want to bring this up\nagain in ~3 weeks to get more of a response...\n\n - Tom\n\n",
"msg_date": "Thu, 05 Feb 1998 02:38:16 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] User-defined functions with NULL values and sets as\n\tparameter and/or return type"
},
{
"msg_contents": "Hi,\n\n I got somewhat familiar with the fmgr code when implementing\n the new procedural language interface. All of your points are\n on my personal TODO too :-). But I'm currently very short on\n time :-(.\n\n>\n> Hi,\n>\n> I have some questions about implementing additional features to the fmgr\n> code:\n>\n> 1. How do I return a NULL value from a function. In fmgr.c I found out\n> that only the C-function with 1 parameter has an additional parameter\n> &isNull which might be used to set the return value to a SQL NULL value.\n> Why not for functions with more paramaters?\n\n The bool isNull is only passed to fmgr_c(), not to fmgr()\n which is used in other places of the executor. In addition I\n would like to be able to identify NULL values passed to the\n function. Currently the isNull is set to true if one of all\n arguments is NULL (yes it's an IN/OUT parameter). But read\n on - below is more on that.\n\n>\n> 2. How do I use sets as a parameter to my code. E.g. in Illustra's API,\n> with which I'm fairly familiar, I wrote a function to send an email to\n> certain people in the following fashion:\n>\n> return sendmail((select email from table where condition), 'from email',\n> 'subject', 'body of email');\n>\n> Of course it would be possible to do something like this:\n>\n> select\n> sendmail(email, 'from email', 'subject', 'body of email')\n> from\n> table\n> where\n> condition;\n>\n> But this would have a detrimental impact on the server, if it has to\n> open say 400 pipes to /usr/lib/sendmail before this query returns\n> (probably my machine dies on this).\n>\n> The function definition in Illustra would be something like:\n>\n> create function sendmail(setof(text), text, text, text) returns integer\n> as external name '/path/to/sendmail.so';\n\n This would be absolutely nice. But currently there is no\n interface to C functions to pass a set into. Well, this would\n be kind of a subselect and currently it could be simulated by\n giving the subselect as a string into the function and having\n the function executing the select via SPI. Using prepared\n plans variable arguments for the qualification might be given\n as separate arguments.\n\n>\n> 3. Can I return sets from a function, e.g. select * from range(1,6)\n> would give me six rows?\n>\n> (Again, in Illustra: create function range(integer, integer) returns\n> setof(integer) as external '/path/to/range.so').\n>\n> Can this be done using PostgreSQL's API and if not where do I have to\n> start to try implementing this?\n\n The other way round :-) yes - this feature is really missing.\n But the executor only supports it for SQL language functions\n right now. The reason for this is that the executor really\n hacks on the SQL function in this case.\n\n The last statement in an SQL function returning a set must be\n a select. When the executor comes to evaluate such a\n function it calls the function the first time through and\n then replaces some internal states of the function cache to\n return subsequent tuples directly from the resulting\n scan/join/merge node of the last select in the function.\n\n For C or PL functions, this cannot work because they could\n return data not coming from a relation. The range() function\n is a really nice little example for this. There isn't any\n node generated to return the results. It cannot get\n implemented as a for(...) loop as it must return it's first\n result on the first call, and the others on subsequent calls.\n So it will look more like a state machine.\n\n Solutions:\n\n For the NULL values in and out I think it wouldn't be too\n hard to implement. A mechanism like CurrentTriggerData used\n in SPI with a global pointer used by the functions that deal\n with NULL's would do.\n\n A function receiving a set as an argument might get the\n executor node returning the set passed in and then evaluate\n that node to get the tuples in. This way, the set passed in\n could be anything that retrieves tuples (seqscan, indexscan,\n merge etc.) and the data is presented on the fly, not\n buffered somewhere like in a temp relation.\n\n Returning a set - hmmm - tricky. On the first evaluation of\n the functions node the executor calls the function than\n returns another executor node (seqscan on a temp relation or\n something else). This returned node is then remembered in\n the functions node and the real return tuples will be\n received by evaluating the return node.\n\n I would really like to start on all of these issues. But\n before hacking around we should discuss that all in detail.\n And I think these things should be available in 6.4 someday.\n\n>\n> 4. And finally: I hardly dare to ask this one: setof(user defined type)?\n\n Since any relation is a new user defined type, this would\n work as soon as sets at all get implemented.\n\n>\n> Please let me know if you have any information that I can use.\n> If you want me to elaborate some more on Illustra's API (which is quite\n> good on implementing sets except for naming conventions) please let me\n> know.\n>\n> Cheers,\n>\n> Jeroen van Vianen\n>\n>\n\n\nUntil later, Jan.\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] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 5 Feb 1998 13:32:03 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] User-defined functions with NULL values and sets as\n\tparameter and/or return type"
}
] |
[
{
"msg_contents": "Dear PostgreSQL gurus, \n\nwith the current CVSup sources, I came across an error\nmessage during the configure step. Using my personal\nconfigure template, linux-elf-libc6, I stumbled into an\nerror message saying that .../tas.s could not be linked.\n\nTo get rid of the message, I had to change the configure\nfile:\n--- snip, snip ---\n*** 5258,5265 ****\n EOF\n \n cat >> $CONFIG_STATUS <<EOF\n! ac_sources=\"backend/port/tas/${os}.s backend/port/dynloader/${os}.c backend/port/dynloader/${os}.h include/port/${os}.h makefiles/Makefile.${os}\"\n! ac_dests=\"backend/port/tas.s backend/port/dynloader.c include/dynloader.h include/os.h Makefile.port\"\n EOF\n \n cat >> $CONFIG_STATUS <<\\EOF\n--- 5261,5269 ----\n EOF\n \n cat >> $CONFIG_STATUS <<EOF\n! ac_sources=\"backend/port/dynloader/${os}.c backend/port/dynloader/${os}.h include/port/${os}.h makefiles/Makefile.${os}\"\n! ac_dests=\"backend/port/dynloader.c include/dynloader.h include/os.h Makefile.port\"\n EOF\n \n cat >> $CONFIG_STATUS <<\\EOF\n--- snip, snip ---\n\nBTW: my template for Linux-GNU-GLIBC-2.x (aka libc6) looks like this:\n\n--- snip, snip ---\nAROPT:crs\nCFLAGS:-O2 -m486 -DHAVE_CRYPT_H=1 -DHAVE_UNION_SEMUN=1\nSHARED_LIB:-fpic\nALL:\nSRCH_INC:/usr/local/include /usr/local/include/ncurses /usr/local/include/readline\nSRCH_LIB:/usr/local/lib\nUSE_LOCALE:yes\nDLSUFFIX:.so\nYFLAGS:-d\nYACC:bison -y\n--- snip, snip ---\n\nBest regards,\n\nErnst\n",
"msg_date": "Wed, 4 Feb 1998 23:13:48 GMT",
"msg_from": "[email protected]",
"msg_from_op": true,
"msg_subject": "Linux: linux.s / tas.s not found..."
}
] |
[
{
"msg_contents": "Okay, here's my idea of how I will get into it:\n\n1) Try to get my apps going. That is I'll try to port my ORACLE\napplications to postgresql to have some programs to do testing.\nTherefore I need:\n2) Add Linus' ecpg precompiler for embedded SQL to the dist.\n3) Try to implement some speed-ups I encounter while playing with my\napps.\n\nI take it some of you have already run the backend against a profiler.\nCould anyone please tell me which functions need works most?\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tBruce Momjian [SMTP:[email protected]]\n> Sent:\tWednesday, February 04, 1998 6:10 PM\n> To:\[email protected]\n> Cc:\[email protected]\n> Subject:\tRe: [HACKERS] Hi\n> \n> > \n> > Hi,\n> > \n> > despite bein chronically short of time I'd like to be involved with\n> some\n> > database development in free software. It seems to me that\n> postgresql is\n> > the most complete package out there and it also has some nice\n> features\n> > like object identity. I consider myself fairly knowledgable when it\n> > comes to databases (at least it brought me my Ph.D.) so I think I\n> could\n> > do some. But before I do so I'd like to ask some questions about the\n> > status of development:\n> > \n> \n> > 3) How would you describe the general goal of this project: Offer as\n> > good a database system as you can create as free software? That is\n> > should it be able to handle huge amounts of data, or is it seen more\n> as\n> > a tool for a private person/small business?\n> \n> About a year ago, our goal was to make it more reliable. At this\n> point,\n> our goal is SQL compliance, speed, and features. See the TODO list in\n> the distribution or on our web page for an idea of our open items.\n> \n> > 4) Are there areas/Which ares need help?\n> \n> What we really need are more people who can get a handle on the entire\n> source tree, so we can implement some of the more complex SQL\n> features,\n> or improve some of the older code.\n> \n> -- \n> Bruce Momjian\n> [email protected]\n",
"msg_date": "Thu, 5 Feb 1998 09:30:27 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Hi"
},
{
"msg_contents": "> \n> Okay, here's my idea of how I will get into it:\n> \n> 1) Try to get my apps going. That is I'll try to port my ORACLE\n> applications to postgresql to have some programs to do testing.\n> Therefore I need:\n> 2) Add Linus' ecpg precompiler for embedded SQL to the dist.\n> 3) Try to implement some speed-ups I encounter while playing with my\n> apps.\n> \n> I take it some of you have already run the backend against a profiler.\n> Could anyone please tell me which functions need works most?\n> \n\nI have already hit many of the speedups people have reported in the\nprofiler by changing the logic, or inlining part/all of the function. I\nrecommend running your own profile and see what it shows. \n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 11:47:44 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Hi"
},
{
"msg_contents": "Meskes, Michael wrote:\n>\n> I take it some of you have already run the backend against a profiler.\n> Could anyone please tell me which functions need works most?\n\nSYSTEM: Linux 2.0.29, gcc 2.7.2\n-------------------------------\n[make distclean]\n[CVSup'ed current source 1998-02-05]\n[./configure]\n[make]\n [CUSTOM_LDFLAGS not settable in Makefile.custom ...]\n [... setting LDFLAGS=-pg ins Makefile.global]\n\n --> fmgr.c: In function `fmgr_pl':\n --> fmgr.c:46: `fmgr_pl_finfo' undeclared (first use this function)\n [fix] cp backend/utils/fmgr.h backend/utils/fmgr/fmgr.h\n\n[make install]\n[Run initdb]\n[Regression tests]\n strings .. failed [ok]\n lseg .. failed [new operator <= on lseg]\n horology .. failed [daylight savings error?]\n triggers .. failed\n select_views .. failed\n\n\nPROFILE OF REGRESSION TEST:\n---------------------------\n time seconds seconds calls ms/call ms/call name \n 35.16 7.97 7.97 mcount (profiler overhead)\n 5.91 9.31 1.34 9924 0.14 0.20 heapgettup\n 4.54 10.34 1.03 234597 0.00 0.01 hash_search\n 2.12 10.82 0.48 151781 0.00 0.00 SpinAcquire\n 2.03 11.28 0.46 46635 0.01 0.02 SearchSysCache\n 1.63 11.65 0.37 171345 0.00 0.00 tag_hash\n 1.46 11.98 0.33 20511 0.02 0.02 yylex\n 1.41 12.30 0.32 31311 0.01 0.03 LockAcquire\n 1.28 12.59 0.29 1522 0.19 0.61 yyparse\n 1.24 12.87 0.28 41176 0.01 0.01 fmgr_isbuiltin\n 1.01 13.10 0.23 201572 0.00 0.00 AllocSetAlloc\n 0.84 13.29 0.19 186214 0.00 0.00 OrderedElemPop\n 0.75 13.46 0.17 42896 0.00 0.00 nocachegetattr\n 0.71 13.62 0.16 29869 0.01 0.04 LockRelease\n 0.66 13.77 0.15 151781 0.00 0.00 SpinRelease\n 0.66 13.92 0.15 88766 0.00 0.01 newNode\n 0.66 14.07 0.15 52248 0.00 0.00 _bt_compare\n... around 1850 functions follows (0.66 - 0.0%)\n\nheapgettup (called by, calling):\n--------------------------------\nindex % time self children called name\n 0.01 0.01 96/9924 heap_markpos [448]\n 0.21 0.11 1566/9924 heap_restrpos [102]\n 1.12 0.58 8262/9924 heap_getnext [20]\n[14] 13.8 1.34 0.69 9924 heapgettup [14]\n 0.04 0.14 9924/9924 RelationGetBufferWithBuffer [148]\n 0.03 0.15 5642/5702 ReleaseAndReadBuffer [145]\n 0.10 0.00 26276/42896 nocachegetattr [158]\n 0.01 0.08 7111/9607 HeapTupleSatisfiesVisibility [185]\n 0.04 0.00 117785/126582 char16eq [339]\n 0.02 0.00 111941/111994 int4eq [427]\n 0.02 0.00 109647/112329 nameeq [426]\n 0.00 0.01 1770/31585 ReleaseBuffer [134]\n 0.01 0.00 14000/50173 PageGetMaxOffsetNumber [341]\n 0.01 0.00 5185/5433 chareq [537]\n 0.01 0.00 15566/90147 BufferGetBlock [311]\n 0.01 0.00 17336/240911 BufferIsValid [207]\n 0.00 0.00 4/1973 int2eq [575]\n 0.00 0.00 7412/7412 nextpage [1186]\n 0.00 0.00 951/1358 SetBufferCommitInfoNeedsSave [1253]\n--------------------------------\n\nThe *complete* gprof output is here:\n ftp://postgresql.org/pub/incoming/regression-profile-980205.gz 156k\n\n/* m */\n",
"msg_date": "Thu, 05 Feb 1998 20:57:21 +0100",
"msg_from": "Mattias Kregert <[email protected]>",
"msg_from_op": false,
"msg_subject": "Profile of current backend"
},
{
"msg_contents": "Interesting. Nothing is jumping out at me. Looks like we could try to\nclean up heapgettup() to see if there is anything in there that can be\nspeeded up.\n\nNone of the calls looks like it should be inlined. Do you see any that\nlook good for inlining?\n\n\n> \n> Meskes, Michael wrote:\n> >\n> > I take it some of you have already run the backend against a profiler.\n> > Could anyone please tell me which functions need works most?\n> \n> SYSTEM: Linux 2.0.29, gcc 2.7.2\n> -------------------------------\n> [make distclean]\n> [CVSup'ed current source 1998-02-05]\n> [./configure]\n> [make]\n> [CUSTOM_LDFLAGS not settable in Makefile.custom ...]\n> [... setting LDFLAGS=-pg ins Makefile.global]\n> \n> --> fmgr.c: In function `fmgr_pl':\n> --> fmgr.c:46: `fmgr_pl_finfo' undeclared (first use this function)\n> [fix] cp backend/utils/fmgr.h backend/utils/fmgr/fmgr.h\n> \n> [make install]\n> [Run initdb]\n> [Regression tests]\n> strings .. failed [ok]\n> lseg .. failed [new operator <= on lseg]\n> horology .. failed [daylight savings error?]\n> triggers .. failed\n> select_views .. failed\n> \n> \n> PROFILE OF REGRESSION TEST:\n> ---------------------------\n> time seconds seconds calls ms/call ms/call name \n> 35.16 7.97 7.97 mcount (profiler overhead)\n> 5.91 9.31 1.34 9924 0.14 0.20 heapgettup\n> 4.54 10.34 1.03 234597 0.00 0.01 hash_search\n> 2.12 10.82 0.48 151781 0.00 0.00 SpinAcquire\n> 2.03 11.28 0.46 46635 0.01 0.02 SearchSysCache\n> 1.63 11.65 0.37 171345 0.00 0.00 tag_hash\n> 1.46 11.98 0.33 20511 0.02 0.02 yylex\n> 1.41 12.30 0.32 31311 0.01 0.03 LockAcquire\n> 1.28 12.59 0.29 1522 0.19 0.61 yyparse\n> 1.24 12.87 0.28 41176 0.01 0.01 fmgr_isbuiltin\n> 1.01 13.10 0.23 201572 0.00 0.00 AllocSetAlloc\n> 0.84 13.29 0.19 186214 0.00 0.00 OrderedElemPop\n> 0.75 13.46 0.17 42896 0.00 0.00 nocachegetattr\n> 0.71 13.62 0.16 29869 0.01 0.04 LockRelease\n> 0.66 13.77 0.15 151781 0.00 0.00 SpinRelease\n> 0.66 13.92 0.15 88766 0.00 0.01 newNode\n> 0.66 14.07 0.15 52248 0.00 0.00 _bt_compare\n> ... around 1850 functions follows (0.66 - 0.0%)\n> \n> heapgettup (called by, calling):\n> --------------------------------\n> index % time self children called name\n> 0.01 0.01 96/9924 heap_markpos [448]\n> 0.21 0.11 1566/9924 heap_restrpos [102]\n> 1.12 0.58 8262/9924 heap_getnext [20]\n> [14] 13.8 1.34 0.69 9924 heapgettup [14]\n> 0.04 0.14 9924/9924 RelationGetBufferWithBuffer [148]\n> 0.03 0.15 5642/5702 ReleaseAndReadBuffer [145]\n> 0.10 0.00 26276/42896 nocachegetattr [158]\n> 0.01 0.08 7111/9607 HeapTupleSatisfiesVisibility [185]\n> 0.04 0.00 117785/126582 char16eq [339]\n> 0.02 0.00 111941/111994 int4eq [427]\n> 0.02 0.00 109647/112329 nameeq [426]\n> 0.00 0.01 1770/31585 ReleaseBuffer [134]\n> 0.01 0.00 14000/50173 PageGetMaxOffsetNumber [341]\n> 0.01 0.00 5185/5433 chareq [537]\n> 0.01 0.00 15566/90147 BufferGetBlock [311]\n> 0.01 0.00 17336/240911 BufferIsValid [207]\n> 0.00 0.00 4/1973 int2eq [575]\n> 0.00 0.00 7412/7412 nextpage [1186]\n> 0.00 0.00 951/1358 SetBufferCommitInfoNeedsSave [1253]\n> --------------------------------\n> \n> The *complete* gprof output is here:\n> ftp://postgresql.org/pub/incoming/regression-profile-980205.gz 156k\n> \n> /* m */\n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 15:11:14 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "Bruce Momjian writes:\n> Interesting. Nothing is jumping out at me. Looks like we could try to\n> clean up heapgettup() to see if there is anything in there that can be\n> speeded up.\n> \n> None of the calls looks like it should be inlined. Do you see any that\n> look good for inlining?\n\nNot exactly. But to get my application running we have to do something to\nspeed it up. This morning I started my program on a database with 15165\nrecords in one table and something like 100 in a second plus 2 in a third.\nThen my software tries to add records to the first table and for each record\nadded it updates a record in the other two. This doesn't count the select\netc.\n\nAnyway, the data isn't processed fast enough. I tried adding about 600\nrecords which should be done (the time in which the data was send) in 5\nminutes, but our system needed almost 8 minutes to insert the data. And this\ndoesn't cause Oracle too much trouble.\n\nSo I guess there's need for some speed-up. :-)\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Fri, 6 Feb 1998 09:54:03 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> Interesting. Nothing is jumping out at me. Looks like we could try to\n> clean up heapgettup() to see if there is anything in there that can be\n> speeded up.\n> \n> None of the calls looks like it should be inlined. Do you see any that\n> look good for inlining?\n\nExecScan() seems to be the only func which calls SeqNext(), which in\nturn accounts for 60% of the calls to heap_getnext(), which does 80% of\nthe calls to heapgettup().\n\n- Put SeqNext() into ExecScan() to lower function call overhead? [minimal optim.]\n\n- In heapgettup(), 50% is the func itself and 50% is called funcs.\n Top four CPU consumers:\n\t0.04 0.14 9924/9924 RelationGetBufferWithBuffer [148]\n\t0.03 0.15 5642/5702 ReleaseAndReadBuffer [145]\n\t0.10 0.00 26276/42896 nocachegetattr [158]\n\t0.01 0.08 7111/9607 HeapTupleSatisfiesVisibility [185]\n\n RelationGetBufferWithBuffer() seems to be called from here only. If so, inline.\n\n- Looking at RelationGetBufferWithBuffer():\n\t0.00 0.10 4603/32354 ReadBuffer [55]\n ReadBuffer() is the biggest cpu consumer called by RelationGetBufferWithBuffer(). (55%)\n\n -> *** 97% of ReadBuffer() CPU time is in calling ReadBufferWithBufferLock()\n\n -> 85% of ReadBufferWithBufferLock() CPU time is in calling BufferAlloc().\n -> ReadBufferWithBufferLock() is the only func calling BufferAlloc().\n -> Conclusion: INLINE BufferAlloc().\n\n- Looking at BufferAlloc():\n\t0.04 0.25 37974/37974 BufTableLookup [114]\n\t0.10 0.00 32340/151781 SpinAcquire [81]\n\t0.10 0.00 37470/40585 PinBuffer [209]\n\t0.08 0.00 38478/43799 RelationGetLRelId [234]\n\t0.04 0.00 37974/151781 SpinRelease [175]\n\n -> 40% of BufferAlloc() CPU time is in calling BufTableLookup().\n -> BufferAlloc() is the only func calling BufTableLookup().\n -> Conclusion: INLINE BufTableLookup().\n\n- Looking at BufTableLookup():\n 86% of CPU time is in calling hash_search(). The rest is own time.\n\n- Looking at hash_search():\n\t0.13 0.41 179189/179189 call_hash [69]\n\t0.00 0.00 6/6 bucket_alloc [1084]\n -> Conclusion: INLINE call_hash() [and bucket_alloc()] into hash_search(). \n\n- Looking at call_hash():\n\t0.37 0.00 171345/171345 tag_hash [94]\n\t0.04 0.00 7844/7844 string_hash [348]\n -> Conclusion: INLINE tag_hash() [and string_hash()] into call_hash().\n -> Perhaps disk_hash() could be used in some way? It is currently #ifdef'd away.\n -> Could we use a lookup table instead of doing hash calculations? Would not that\n -> be much faster?\n\n\nIt looks to me as if there are too many levels of function calls.\nPerhaps all functions which are called by only one other func should be inlined?\n\n\nGuesstimate:\n This would speed up heapgettup() by 10% ???\n Other functions would speed up too.\n\n\n/* m */\n",
"msg_date": "Fri, 06 Feb 1998 10:58:08 +0100",
"msg_from": "Mattias Kregert <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "> Not exactly. But to get my application running we have to do something to\n> speed it up. This morning I started my program on a database with 15165\n> records in one table and something like 100 in a second plus 2 in a third.\n> Then my software tries to add records to the first table and for each record\n> added it updates a record in the other two. This doesn't count the select\n> etc.\n>\n> Anyway, the data isn't processed fast enough. I tried adding about 600\n> records which should be done (the time in which the data was send) in 5\n> minutes, but our system needed almost 8 minutes to insert the data. And this\n> doesn't cause Oracle too much trouble.\n>\n> So I guess there's need for some speed-up. :-)\n\nI (and others) had done some benchmarking on simple inserts (6 months ago?) and\nhad concluded that the speed was similar to other commercial systems (I was\ncomparing against Ingres). I recall getting ~50TPS.\n\nThis was all before Bruce did his work on startup and runtime speeds. You\nreally think your performance is that far off? You are doing selects on the big\ntable before inserting? Do you have indices set up?? Our results were for\ninserts on a heap table, which has the least overhead...\n\n - Tom\n\n",
"msg_date": "Fri, 06 Feb 1998 15:09:53 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "On Fri, 6 Feb 1998, Thomas G. Lockhart wrote:\n\n> > Not exactly. But to get my application running we have to do something to\n> > speed it up. This morning I started my program on a database with 15165\n> > records in one table and something like 100 in a second plus 2 in a third.\n> > Then my software tries to add records to the first table and for each record\n> > added it updates a record in the other two. This doesn't count the select\n> > etc.\n> >\n> > Anyway, the data isn't processed fast enough. I tried adding about 600\n> > records which should be done (the time in which the data was send) in 5\n> > minutes, but our system needed almost 8 minutes to insert the data. And this\n> > doesn't cause Oracle too much trouble.\n> >\n> > So I guess there's need for some speed-up. :-)\n> \n> I (and others) had done some benchmarking on simple inserts (6 months ago?) and\n> had concluded that the speed was similar to other commercial systems (I was\n> comparing against Ingres). I recall getting ~50TPS.\n> \n> This was all before Bruce did his work on startup and runtime speeds. You\n> really think your performance is that far off? You are doing selects on the big\n> table before inserting? Do you have indices set up?? Our results were for\n> inserts on a heap table, which has the least overhead...\n\n\tJust curious, but do you have -F set to disable fsync()? We\nreally really should disable that by default :(\n\n\n",
"msg_date": "Fri, 6 Feb 1998 10:20:38 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "On Fri, 6 Feb 1998, Vadim B. Mikheev wrote:\n\n> > Just curious, but do you have -F set to disable fsync()? We\n> > really really should disable that by default :(\n> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n> Oh no!\n\n\tIsn't that our standard recommended operating parameter anyway?\nOops, that's just our Performance improving recommendation...sorry...:)\n\n\n",
"msg_date": "Fri, 6 Feb 1998 10:55:15 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "The Hermit Hacker wrote:\n> \n> On Fri, 6 Feb 1998, Thomas G. Lockhart wrote:\n> \n> > > Not exactly. But to get my application running we have to do something to\n> > > speed it up. This morning I started my program on a database with 15165\n> > > records in one table and something like 100 in a second plus 2 in a third.\n> > > Then my software tries to add records to the first table and for each record\n> > > added it updates a record in the other two. This doesn't count the select\n> > > etc.\n> > >\n> > > Anyway, the data isn't processed fast enough. I tried adding about 600\n> > > records which should be done (the time in which the data was send) in 5\n> > > minutes, but our system needed almost 8 minutes to insert the data. And this\n> > > doesn't cause Oracle too much trouble.\n> > >\n> > > So I guess there's need for some speed-up. :-)\n> >\n> > I (and others) had done some benchmarking on simple inserts (6 months ago?) and\n> > had concluded that the speed was similar to other commercial systems (I was\n> > comparing against Ingres). I recall getting ~50TPS.\n> >\n> > This was all before Bruce did his work on startup and runtime speeds. You\n> > really think your performance is that far off? You are doing selects on the big\n> > table before inserting? Do you have indices set up?? Our results were for\n> > inserts on a heap table, which has the least overhead...\n\nAnd did you use BEGIN/END ?\nAuto-commit is ON in postgres...\n\n> \n> Just curious, but do you have -F set to disable fsync()? We\n> really really should disable that by default :(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nOh no!\n\nVadim\n",
"msg_date": "Fri, 06 Feb 1998 23:00:54 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "> \n> Bruce Momjian wrote:\n> > \n> > Interesting. Nothing is jumping out at me. Looks like we could try to\n> > clean up heapgettup() to see if there is anything in there that can be\n> > speeded up.\n> > \n> > None of the calls looks like it should be inlined. Do you see any that\n> > look good for inlining?\n> \n> ExecScan() seems to be the only func which calls SeqNext(), which in\n> turn accounts for 60% of the calls to heap_getnext(), which does 80% of\n> the calls to heapgettup().\n\nThis is certainly the type of analysis that will help us. I disagree on\ninlining everything, because the code maintenance becomes a nightmare,\nbut we certainly could improve things.\n\nCurrently, if a function is declared static, is it only called from\nwithin that file. If there is only one call to that function in the\nfile, it could be inlined. I suggest we only attack things like the\nbelow where we have a large amount of function call overhead. Also,\nthere are cases where inlining the 'top' of the function can be really\ngood. For heap_getattr(), many cases were handled in the top that just\ndid a quick return. In those cases, I took the top and made a macro out\nof it, and only called the function if I needed real processing.\n\nThis is a great help, and any patches would be good. The other nice\nthing about this is that it is pretty compartementalized. You just need\nto check that the new code behaves like the old code, and you are safe.\n\nThis is great analysis of the problem, too.\n\n> \n> - Put SeqNext() into ExecScan() to lower function call overhead? [minimal optim.]\n\nSure. Just comment it so it is clear for readers, so it is not pulled\nout into a separate function some day.\n\n\n\n\n\n> \n> - In heapgettup(), 50% is the func itself and 50% is called funcs.\n> Top four CPU consumers:\n> \t0.04 0.14 9924/9924 RelationGetBufferWithBuffer [148]\n> \t0.03 0.15 5642/5702 ReleaseAndReadBuffer [145]\n> \t0.10 0.00 26276/42896 nocachegetattr [158]\n> \t0.01 0.08 7111/9607 HeapTupleSatisfiesVisibility [185]\n> \n> RelationGetBufferWithBuffer() seems to be called from here only. If so, inline.\n\nOK.\n\n> \n> - Looking at RelationGetBufferWithBuffer():\n> \t0.00 0.10 4603/32354 ReadBuffer [55]\n> ReadBuffer() is the biggest cpu consumer called by RelationGetBufferWithBuffer(). (55%)\n> \n> -> *** 97% of ReadBuffer() CPU time is in calling ReadBufferWithBufferLock()\n> \n> -> 85% of ReadBufferWithBufferLock() CPU time is in calling BufferAlloc().\n> -> ReadBufferWithBufferLock() is the only func calling BufferAlloc().\n> -> Conclusion: INLINE BufferAlloc().\n\nMy only warning is that I have never inlined a function, AND left\nanother copy of the function non-inlined. The reason is that you are\nleft with two copies of the function, and that can get really confusing.\nI don't know how others feel about that, but the code is already so\ncomplicated, I hesitate to start duplicating blocks of code. Of course,\nif you make it a macro, you have it in one place, and the preprocessor\nis duplicating it for you.\n\nI have converted a function to a macro, and I have inlined\nfrequently-called functions. \n\n\n> \n> - Looking at BufferAlloc():\n> \t0.04 0.25 37974/37974 BufTableLookup [114]\n> \t0.10 0.00 32340/151781 SpinAcquire [81]\n> \t0.10 0.00 37470/40585 PinBuffer [209]\n> \t0.08 0.00 38478/43799 RelationGetLRelId [234]\n> \t0.04 0.00 37974/151781 SpinRelease [175]\n> \n> -> 40% of BufferAlloc() CPU time is in calling BufTableLookup().\n> -> BufferAlloc() is the only func calling BufTableLookup().\n> -> Conclusion: INLINE BufTableLookup().\n> \n> - Looking at BufTableLookup():\n> 86% of CPU time is in calling hash_search(). The rest is own time.\n> \n> - Looking at hash_search():\n> \t0.13 0.41 179189/179189 call_hash [69]\n> \t0.00 0.00 6/6 bucket_alloc [1084]\n> -> Conclusion: INLINE call_hash() [and bucket_alloc()] into hash_search(). \n> \n> - Looking at call_hash():\n> \t0.37 0.00 171345/171345 tag_hash [94]\n> \t0.04 0.00 7844/7844 string_hash [348]\n> -> Conclusion: INLINE tag_hash() [and string_hash()] into call_hash().\n> -> Perhaps disk_hash() could be used in some way? It is currently #ifdef'd away.\n> -> Could we use a lookup table instead of doing hash calculations? Would not that\n> -> be much faster?\n> \n> \n> It looks to me as if there are too many levels of function calls.\n> Perhaps all functions which are called by only one other func should be inlined?\n> \n> \n> Guesstimate:\n> This would speed up heapgettup() by 10% ???\n> Other functions would speed up too.\n\nMakes sense. If the functions are in the same source file, and are\ndeclared static, doesn't the compiler inline some of them.\n\nSee my heap_getattr() macro for a way perhaps to macro-ize some of\nthese. The macro looks much clearer than the old style macros.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 6 Feb 1998 12:09:38 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "> > I (and others) had done some benchmarking on simple inserts (6 months ago?) and\n> > had concluded that the speed was similar to other commercial systems (I was\n> > comparing against Ingres). I recall getting ~50TPS.\n> > \n> > This was all before Bruce did his work on startup and runtime speeds. You\n> > really think your performance is that far off? You are doing selects on the big\n> > table before inserting? Do you have indices set up?? Our results were for\n> > inserts on a heap table, which has the least overhead...\n> \n> \tJust curious, but do you have -F set to disable fsync()? We\n> really really should disable that by default :(\n\nAgreed. But maybe we should wait until we get pg_log syncing so we have\n30-second reliability. Don't know where that fits on Vadim's list.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 6 Feb 1998 12:46:05 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "The Hermit Hacker writes:\n> \tJust curious, but do you have -F set to disable fsync()? We\n> really really should disable that by default :(\n\nI tried with -F and it runs nicely. No difference to see between PostgreSQL\nand Oracle. I just ran another test which includes table creation, inserts\nand drop (source follows). Here's the result:\n\nOracle 7.3.3.4.0:\nI needed 21 seconds and -345682 microseconds for this test\n\nPostgreSQL without -F:\nI needed 152 seconds and -623545 microseconds for this test\n\nPostgreSQL with -F:\nI needed 5 seconds and 84411 microseconds for this test\n\nWhow!\n\nHere's the source (yes, our precompiler can handle this kind of program\nalready :-)):\n\n#include <stdio.h>\n#include <sys/time.h>\n#include <unistd.h>\n\nexec sql include sqlca;\n\n#define SQLCODE sqlca.sqlcode\n\nvoid\ndb_error (char *msg)\n{\n\tsqlca.sqlerrm.sqlerrmc[sqlca.sqlerrm.sqlerrml] = '\\0';\n\tprintf (\"%s: db error %s\\n\", msg, sqlca.sqlerrm.sqlerrmc);\n\texit (1);\n}\n\nint\nmain ()\n{\nexec sql begin declare section;\n\tlong i;\nexec sql end declare section;\n\tstruct timeval tvs, tve;\n\n\tgettimeofday(&tvs, NULL);\n\n\texec sql connect 'mm';\n\tif (SQLCODE)\n\t\tdb_error (\"connect\");\n\n\texec sql create table perftest(number int4, ascii char16);\n\tif (SQLCODE)\n db_error (\"create t\");\n\n\texec sql create unique index number on perftest(number);\n\tif (SQLCODE)\n db_error (\"create i\");\n\n\tfor (i = 0;i < 1407; i++)\n\t{\n\t\texec sql begin declare section;\n\t\t\tchar text[16];\n\t\texec sql end declare section;\n\n\t\tsprintf(text, \"%ld\", i);\n\t\texec sql insert into perftest(number, ascii) values (:i, :text);\n\t\tif (SQLCODE)\n\t\t\tdb_error (\"insert\");\n\n\t\texec sql commit;\n\t\tif (SQLCODE)\t\n\t\t\tdb_error (\"commit\");\n\t}\n\n\texec sql drop index number;\n\tif (SQLCODE)\t\n\t\tdb_error (\"drop i\");\n\n\texec sql drop table perftest;\n\tif (SQLCODE)\t\n\t\tdb_error (\"drop t\");\n\n\texec sql commit;\n\tif (SQLCODE)\t\n\t\tdb_error (\"commit\");\n\n\tgettimeofday(&tve, NULL);\n\n\tprintf(\"I needed %ld seconds and %ld microseconds for this test\\n\", tve.tv_sec - tvs.tv_sec, tve.tv_usec - tvs.tv_usec);\n\n\treturn (0);\n}\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Tue, 10 Feb 1998 16:10:16 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "> Here's the source (yes, our precompiler can handle this kind of program\n> already :-)):\n\nLooks nice!\n\n - Tom\n\n",
"msg_date": "Tue, 10 Feb 1998 15:24:53 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "> \n> The Hermit Hacker writes:\n> > \tJust curious, but do you have -F set to disable fsync()? We\n> > really really should disable that by default :(\n> \n> I tried with -F and it runs nicely. No difference to see between PostgreSQL\n> and Oracle. I just ran another test which includes table creation, inserts\n> and drop (source follows). Here's the result:\n> \n> Oracle 7.3.3.4.0:\n> I needed 21 seconds and -345682 microseconds for this test\n> \n> PostgreSQL without -F:\n> I needed 152 seconds and -623545 microseconds for this test\n> \n> PostgreSQL with -F:\n> I needed 5 seconds and 84411 microseconds for this test\n> \n> Whow!\n> \n\n\nThis is good news. Is this with 6.3 or 6.2.1?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 10 Feb 1998 10:40:51 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "Can we go anywhere with this?\n\n> \n> Bruce Momjian wrote:\n> > \n> > Interesting. Nothing is jumping out at me. Looks like we could try to\n> > clean up heapgettup() to see if there is anything in there that can be\n> > speeded up.\n> > \n> > None of the calls looks like it should be inlined. Do you see any that\n> > look good for inlining?\n> \n> ExecScan() seems to be the only func which calls SeqNext(), which in\n> turn accounts for 60% of the calls to heap_getnext(), which does 80% of\n> the calls to heapgettup().\n> \n> - Put SeqNext() into ExecScan() to lower function call overhead? [minimal optim.]\n> \n> - In heapgettup(), 50% is the func itself and 50% is called funcs.\n> Top four CPU consumers:\n> \t0.04 0.14 9924/9924 RelationGetBufferWithBuffer [148]\n> \t0.03 0.15 5642/5702 ReleaseAndReadBuffer [145]\n> \t0.10 0.00 26276/42896 nocachegetattr [158]\n> \t0.01 0.08 7111/9607 HeapTupleSatisfiesVisibility [185]\n> \n> RelationGetBufferWithBuffer() seems to be called from here only. If so, inline.\n> \n> - Looking at RelationGetBufferWithBuffer():\n> \t0.00 0.10 4603/32354 ReadBuffer [55]\n> ReadBuffer() is the biggest cpu consumer called by RelationGetBufferWithBuffer(). (55%)\n> \n> -> *** 97% of ReadBuffer() CPU time is in calling ReadBufferWithBufferLock()\n> \n> -> 85% of ReadBufferWithBufferLock() CPU time is in calling BufferAlloc().\n> -> ReadBufferWithBufferLock() is the only func calling BufferAlloc().\n> -> Conclusion: INLINE BufferAlloc().\n> \n> - Looking at BufferAlloc():\n> \t0.04 0.25 37974/37974 BufTableLookup [114]\n> \t0.10 0.00 32340/151781 SpinAcquire [81]\n> \t0.10 0.00 37470/40585 PinBuffer [209]\n> \t0.08 0.00 38478/43799 RelationGetLRelId [234]\n> \t0.04 0.00 37974/151781 SpinRelease [175]\n> \n> -> 40% of BufferAlloc() CPU time is in calling BufTableLookup().\n> -> BufferAlloc() is the only func calling BufTableLookup().\n> -> Conclusion: INLINE BufTableLookup().\n> \n> - Looking at BufTableLookup():\n> 86% of CPU time is in calling hash_search(). The rest is own time.\n> \n> - Looking at hash_search():\n> \t0.13 0.41 179189/179189 call_hash [69]\n> \t0.00 0.00 6/6 bucket_alloc [1084]\n> -> Conclusion: INLINE call_hash() [and bucket_alloc()] into hash_search(). \n> \n> - Looking at call_hash():\n> \t0.37 0.00 171345/171345 tag_hash [94]\n> \t0.04 0.00 7844/7844 string_hash [348]\n> -> Conclusion: INLINE tag_hash() [and string_hash()] into call_hash().\n> -> Perhaps disk_hash() could be used in some way? It is currently #ifdef'd away.\n> -> Could we use a lookup table instead of doing hash calculations? Would not that\n> -> be much faster?\n> \n> \n> It looks to me as if there are too many levels of function calls.\n> Perhaps all functions which are called by only one other func should be inlined?\n> \n> \n> Guesstimate:\n> This would speed up heapgettup() by 10% ???\n> Other functions would speed up too.\n> \n> \n> /* m */\n> \n\n\n-- \nBruce Momjian | 830 Blythe Avenue\[email protected] | Drexel Hill, Pennsylvania 19026\n + If your life is a hard drive, | (610) 353-9879(w)\n + Christ can be your backup. | (610) 853-3000(h)\n",
"msg_date": "Sun, 15 Mar 1998 23:57:19 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "In src/test/regress:\n it is time to remove the \"PST8PDT\" stuff from the files\n README and regress.sh\n\n\"set timezone to 'PST8PDT'\" should be added to sql/datetime.sql\n\n/* m */\n",
"msg_date": "Tue, 17 Mar 1998 15:28:30 +0100",
"msg_from": "Mattias Kregert <[email protected]>",
"msg_from_op": false,
"msg_subject": "Regression test"
},
{
"msg_contents": "> it is time to remove the \"PST8PDT\" stuff from the files\n> README and regress.sh\n> \n> \"set timezone to 'PST8PDT'\" should be added to sql/datetime.sql\n\nHmm. It would have to be added in several regression tests, not just the\ndatetime one. Also, having it isolated to the client configuration in\nregress.sh gives one the opportunity to fix it or adjust it for your\nplatform.\n\nI'm not sure what the README says about it; that may be obsolete. Is the\nPGTZ setting in regress.sh causing you trouble??\n\n - Tom\n",
"msg_date": "Wed, 18 Mar 1998 06:42:11 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Regression test"
},
{
"msg_contents": "Bruce Momjian writes:\n> Can we go anywhere with this?\n\nDid anyone do anything with this already?\n\n> > ExecScan() seems to be the only func which calls SeqNext(), which in\n> > turn accounts for 60% of the calls to heap_getnext(), which does 80% of\n> > the calls to heapgettup().\n> > \n> > - Put SeqNext() into ExecScan() to lower function call overhead? [minimal optim.]\n> > \n> > - In heapgettup(), 50% is the func itself and 50% is called funcs.\n> > Top four CPU consumers:\n> > \t0.04 0.14 9924/9924 RelationGetBufferWithBuffer [148]\n> > \t0.03 0.15 5642/5702 ReleaseAndReadBuffer [145]\n> > \t0.10 0.00 26276/42896 nocachegetattr [158]\n> > \t0.01 0.08 7111/9607 HeapTupleSatisfiesVisibility [185]\n> > \n> > RelationGetBufferWithBuffer() seems to be called from here only. If so, inline.\n> > \n> > - Looking at RelationGetBufferWithBuffer():\n> > \t0.00 0.10 4603/32354 ReadBuffer [55]\n> > ReadBuffer() is the biggest cpu consumer called by RelationGetBufferWithBuffer(). (55%)\n> > \n> > -> *** 97% of ReadBuffer() CPU time is in calling ReadBufferWithBufferLock()\n> > \n> > -> 85% of ReadBufferWithBufferLock() CPU time is in calling BufferAlloc().\n> > -> ReadBufferWithBufferLock() is the only func calling BufferAlloc().\n> > -> Conclusion: INLINE BufferAlloc().\n> > \n> > - Looking at BufferAlloc():\n> > \t0.04 0.25 37974/37974 BufTableLookup [114]\n> > \t0.10 0.00 32340/151781 SpinAcquire [81]\n> > \t0.10 0.00 37470/40585 PinBuffer [209]\n> > \t0.08 0.00 38478/43799 RelationGetLRelId [234]\n> > \t0.04 0.00 37974/151781 SpinRelease [175]\n> > \n> > -> 40% of BufferAlloc() CPU time is in calling BufTableLookup().\n> > -> BufferAlloc() is the only func calling BufTableLookup().\n> > -> Conclusion: INLINE BufTableLookup().\n> > \n> > - Looking at BufTableLookup():\n> > 86% of CPU time is in calling hash_search(). The rest is own time.\n> > \n> > - Looking at hash_search():\n> > \t0.13 0.41 179189/179189 call_hash [69]\n> > \t0.00 0.00 6/6 bucket_alloc [1084]\n> > -> Conclusion: INLINE call_hash() [and bucket_alloc()] into hash_search(). \n> > \n> > - Looking at call_hash():\n> > \t0.37 0.00 171345/171345 tag_hash [94]\n> > \t0.04 0.00 7844/7844 string_hash [348]\n> > -> Conclusion: INLINE tag_hash() [and string_hash()] into call_hash().\n> > -> Perhaps disk_hash() could be used in some way? It is currently #ifdef'd away.\n> > -> Could we use a lookup table instead of doing hash calculations? Would not that\n> > -> be much faster?\n> > \n> > \n> > It looks to me as if there are too many levels of function calls.\n> > Perhaps all functions which are called by only one other func should be inlined?\n\nIsn't this a good solution? A function only called by one other function has\nits right to exist only for readability. And this optimization could be done\nautomatically.\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Mon, 23 Mar 1998 15:06:34 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "> \n> Bruce Momjian writes:\n> > Can we go anywhere with this?\n> \n> Did anyone do anything with this already?\n\nNo one yet.\n\n> \n> > > ExecScan() seems to be the only func which calls SeqNext(), which in\n> > > turn accounts for 60% of the calls to heap_getnext(), which does 80% of\n> > > the calls to heapgettup().\n> > > \n> > > - Put SeqNext() into ExecScan() to lower function call overhead? [minimal optim.]\n> > > \n> > > - In heapgettup(), 50% is the func itself and 50% is called funcs.\n> > > Top four CPU consumers:\n> > > \t0.04 0.14 9924/9924 RelationGetBufferWithBuffer [148]\n> > > \t0.03 0.15 5642/5702 ReleaseAndReadBuffer [145]\n> > > \t0.10 0.00 26276/42896 nocachegetattr [158]\n> > > \t0.01 0.08 7111/9607 HeapTupleSatisfiesVisibility [185]\n> > > \n> > > RelationGetBufferWithBuffer() seems to be called from here only. If so, inline.\n> > > \n> > > - Looking at RelationGetBufferWithBuffer():\n> > > \t0.00 0.10 4603/32354 ReadBuffer [55]\n> > > ReadBuffer() is the biggest cpu consumer called by RelationGetBufferWithBuffer(). (55%)\n> > > \n> > > -> *** 97% of ReadBuffer() CPU time is in calling ReadBufferWithBufferLock()\n> > > \n> > > -> 85% of ReadBufferWithBufferLock() CPU time is in calling BufferAlloc().\n> > > -> ReadBufferWithBufferLock() is the only func calling BufferAlloc().\n> > > -> Conclusion: INLINE BufferAlloc().\n> > > \n> > > - Looking at BufferAlloc():\n> > > \t0.04 0.25 37974/37974 BufTableLookup [114]\n> > > \t0.10 0.00 32340/151781 SpinAcquire [81]\n> > > \t0.10 0.00 37470/40585 PinBuffer [209]\n> > > \t0.08 0.00 38478/43799 RelationGetLRelId [234]\n> > > \t0.04 0.00 37974/151781 SpinRelease [175]\n> > > \n> > > -> 40% of BufferAlloc() CPU time is in calling BufTableLookup().\n> > > -> BufferAlloc() is the only func calling BufTableLookup().\n> > > -> Conclusion: INLINE BufTableLookup().\n> > > \n> > > - Looking at BufTableLookup():\n> > > 86% of CPU time is in calling hash_search(). The rest is own time.\n> > > \n> > > - Looking at hash_search():\n> > > \t0.13 0.41 179189/179189 call_hash [69]\n> > > \t0.00 0.00 6/6 bucket_alloc [1084]\n> > > -> Conclusion: INLINE call_hash() [and bucket_alloc()] into hash_search(). \n> > > \n> > > - Looking at call_hash():\n> > > \t0.37 0.00 171345/171345 tag_hash [94]\n> > > \t0.04 0.00 7844/7844 string_hash [348]\n> > > -> Conclusion: INLINE tag_hash() [and string_hash()] into call_hash().\n> > > -> Perhaps disk_hash() could be used in some way? It is currently #ifdef'd away.\n> > > -> Could we use a lookup table instead of doing hash calculations? Would not that\n> > > -> be much faster?\n> > > \n> > > \n> > > It looks to me as if there are too many levels of function calls.\n> > > Perhaps all functions which are called by only one other func should be inlined?\n> \n> Isn't this a good solution? A function only called by one other function has\n> its right to exist only for readability. And this optimization could be done\n> automatically.\n\nWouldn't be such a big deal if it was not call so many times.\n\n-- \nBruce Momjian | 830 Blythe Avenue\[email protected] | Drexel Hill, Pennsylvania 19026\n + If your life is a hard drive, | (610) 353-9879(w)\n + Christ can be your backup. | (610) 853-3000(h)\n",
"msg_date": "Mon, 23 Mar 1998 10:08:05 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
},
{
"msg_contents": "> \n> Bruce Momjian wrote:\n> > \n> > Interesting. Nothing is jumping out at me. Looks like we could try to\n> > clean up heapgettup() to see if there is anything in there that can be\n> > speeded up.\n> > \n> > None of the calls looks like it should be inlined. Do you see any that\n> > look good for inlining?\n> \n> ExecScan() seems to be the only func which calls SeqNext(), which in\n> turn accounts for 60% of the calls to heap_getnext(), which does 80% of\n> the calls to heapgettup().\n> \n> - Put SeqNext() into ExecScan() to lower function call overhead? [minimal optim.]\n> \n> - In heapgettup(), 50% is the func itself and 50% is called funcs.\n> Top four CPU consumers:\n> \t0.04 0.14 9924/9924 RelationGetBufferWithBuffer [148]\n> \t0.03 0.15 5642/5702 ReleaseAndReadBuffer [145]\n> \t0.10 0.00 26276/42896 nocachegetattr [158]\n> \t0.01 0.08 7111/9607 HeapTupleSatisfiesVisibility [185]\n> \n> RelationGetBufferWithBuffer() seems to be called from here only. If so, inline.\n> \n> - Looking at RelationGetBufferWithBuffer():\n> \t0.00 0.10 4603/32354 ReadBuffer [55]\n> ReadBuffer() is the biggest cpu consumer called by RelationGetBufferWithBuffer(). (55%)\n> \n> -> *** 97% of ReadBuffer() CPU time is in calling ReadBufferWithBufferLock()\n> \n> -> 85% of ReadBufferWithBufferLock() CPU time is in calling BufferAlloc().\n> -> ReadBufferWithBufferLock() is the only func calling BufferAlloc().\n> -> Conclusion: INLINE BufferAlloc().\n> \n> - Looking at BufferAlloc():\n> \t0.04 0.25 37974/37974 BufTableLookup [114]\n> \t0.10 0.00 32340/151781 SpinAcquire [81]\n> \t0.10 0.00 37470/40585 PinBuffer [209]\n> \t0.08 0.00 38478/43799 RelationGetLRelId [234]\n> \t0.04 0.00 37974/151781 SpinRelease [175]\n> \n> -> 40% of BufferAlloc() CPU time is in calling BufTableLookup().\n> -> BufferAlloc() is the only func calling BufTableLookup().\n> -> Conclusion: INLINE BufTableLookup().\n> \n> - Looking at BufTableLookup():\n> 86% of CPU time is in calling hash_search(). The rest is own time.\n> \n> - Looking at hash_search():\n> \t0.13 0.41 179189/179189 call_hash [69]\n> \t0.00 0.00 6/6 bucket_alloc [1084]\n> -> Conclusion: INLINE call_hash() [and bucket_alloc()] into hash_search(). \n> \n> - Looking at call_hash():\n> \t0.37 0.00 171345/171345 tag_hash [94]\n> \t0.04 0.00 7844/7844 string_hash [348]\n> -> Conclusion: INLINE tag_hash() [and string_hash()] into call_hash().\n> -> Perhaps disk_hash() could be used in some way? It is currently #ifdef'd away.\n> -> Could we use a lookup table instead of doing hash calculations? Would not that\n> -> be much faster?\n> \n> \n> It looks to me as if there are too many levels of function calls.\n> Perhaps all functions which are called by only one other func should be inlined?\n> \n> \n> Guesstimate:\n> This would speed up heapgettup() by 10% ???\n> Other functions would speed up too.\n\nI have reviewed your findings, and have inlined 9 small functions that\nare called for every row. Unfortunately, several of the ones you\nmention above are called via function pointers, so it was not possible\nto inline them.\n\nIn my testing, I created a table with 160k rows, and did a SELECT that\nwould do a sequential scan of the table, and looked at the functions\nthat were called the most.\n\n---------------------------------------------------------------------------\n\n % cumulative self self total \n time seconds seconds calls ms/call ms/call name \n 54.3 2.65 2.65 mcount (472)\n 16.2 3.44 0.79 2999 0.26 0.26 _read [18]\n 3.5 3.61 0.17 160148 0.00 0.01 _heapgettup [15]\n 2.5 3.73 0.12 174007 0.00 0.00 _SpinAcquire [36]\n 2.3 3.84 0.11 981732 0.00 0.00 _BufferIsValid [37]\n 2.3 3.95 0.11 160148 0.00 0.01 _heap_getnext [14]\n 2.0 4.05 0.10 160096 0.00 0.00 _ExecEvalVar [38]\n 2.0 4.15 0.10 160096 0.00 0.00 _ExecMakeFunctionResult <cycle 3> [26]\n 1.4 4.22 0.07 174007 0.00 0.00 _SpinRelease [41]\n 1.2 4.28 0.06 18176 0.00 0.00 _hash_search [40]\n 1.0 4.33 0.05 480288 0.00 0.00 _ExecEvalExpr <cycle 3> [34]\n 1.0 4.38 0.05 167313 0.00 0.00 _RelationGetLRelId [47]\n 1.0 4.43 0.05 160096 0.00 0.00 _ExecEvalFuncArgs <cycle 3> [48]\n 0.8 4.47 0.04 160097 0.00 0.01 _SeqNext [13]\n 0.8 4.51 0.04 160096 0.00 0.00 _ExecQual [22]\n 0.6 4.54 0.03 160096 0.00 0.00 _ExecEvalOper <cycle 3> [51]\n 0.4 4.56 0.02 163015 0.00 0.00 _IncrBufferRefCount [61]\n 0.4 4.58 0.02 161342 0.00 0.00 _fmgr_c <cycle 5> [29]\n 0.4 4.60 0.02 160451 0.00 0.00 _TransactionIdEquals [62]\n 0.4 4.62 0.02 160148 0.00 0.00 _RelationGetBufferWithBuffer [28]\n 0.4 4.64 0.02 160098 0.00 0.00 _TupIsNull [63]\n 0.4 4.66 0.02 160097 0.00 0.00 _ExecIncrSlotBufferRefcnt [45]\n 0.4 4.68 0.02 160096 0.00 0.00 _ExecQualClause [23]\n 0.4 4.70 0.02 6161 0.00 0.00 _strcpy [64]\n 0.4 4.72 0.02 31 0.65 0.65 _open [65]\n 0.2 4.73 0.01 164288 0.00 0.00 _ReleaseBuffer [81]\n 0.2 4.74 0.01 164048 0.00 0.00 _PageGetMaxOffsetNumber [97]\n 0.2 4.75 0.01 160450 0.00 0.00 _HeapTupleSatisfiesVisibility [50]\n 0.2 4.76 0.01 160449 0.00 0.00 _HeapTupleSatisfiesNow [98]\n 0.2 4.77 0.01 160100 0.00 0.00 _ExecClearTuple [49]\n 0.2 4.78 0.01 160098 0.00 0.00 _ExecStoreTuple [46]\n 0.2 4.79 0.01 160096 0.00 0.00 _bpchareq [102]\n 0.2 4.80 0.01 15696 0.00 0.00 _call_hash [99]\n 0.2 4.81 0.01 12187 0.00 0.00 _malloc [96]\n 0.2 4.82 0.01 9182 0.00 0.00 _PortalHeapMemoryAlloc [73]\n 0.2 4.83 0.01 7257 0.00 0.00 _newNode [54]\n 0.2 4.84 0.01 2973 0.00 0.20 _ReleaseAndReadBuffer [25]\n 0.2 4.85 0.01 1241 0.01 0.02 _LockAcquire [58]\n 0.2 4.86 0.01 189 0.05 0.05 _IsBootstrapProcessingMode [103]\n 0.2 4.87 0.01 3 3.33 3.33 _shmdt [101]\n 0.2 4.88 0.01 1 10.00 2172.84 _ExecScan [11]\n 0.0 4.88 0.00 166214 0.00 0.00 _BufferGetBlock [72]\n 0.0 4.88 0.00 160449 0.00 0.00 _heapisoverride [477]\n 0.0 4.88 0.00 15303 0.00 0.00 _tag_hash [478]\n 0.0 4.88 0.00 11718 0.00 0.00 _AllocSetContains [479]\n 0.0 4.88 0.00 11718 0.00 0.00 _OrderedSetContains [480]\n 0.0 4.88 0.00 11366 0.00 0.00 _OrderedElemPush [481]\n 0.0 4.88 0.00 11366 0.00 0.00 _OrderedElemPushHead [482]\n 0.0 4.88 0.00 11366 0.00 0.00 _OrderedElemPushInto [483]\n 0.0 4.88 0.00 11361 0.00 0.00 _AllocSetAlloc [104]\n 0.0 4.88 0.00 11361 0.00 0.00 _MemoryContextAlloc [67]\n 0.0 4.88 0.00 11339 0.00 0.00 _palloc [68]\n 0.0 4.88 0.00 10515 0.00 0.00 _free [484]\n 0.0 4.88 0.00 10488 0.00 0.00 _AllocSetFree [485]\n 0.0 4.88 0.00 10488 0.00 0.00 _OrderedElemPop [486]\n 0.0 4.88 0.00 8176 0.00 0.00 _get_expr [487]\n 0.0 4.88 0.00 7953 0.00 0.00 _AllocSetGetFirst [488]\n 0.0 4.88 0.00 7953 0.00 0.00 _OrderedSetGetHead [489]\n 0.0 4.88 0.00 7952 0.00 0.00 _OrderedElemGetBase [490]\n 0.0 4.88 0.00 7424 0.00 0.00 _memmove [491]\n\n\n---------------------------------------------------------------------------\n\nOf these, I inlined:\n\n\t#define TransactionIdEquals(id1, id2) \\\n\t#define TupIsNull(slot) \\\n\t#define ExecIncrSlotBufferRefcnt(slot) \\\n\t#define BufferIsValid(bufnum) \\\n\t#define IncrBufferRefCount(buffer) \\\n\t#define BufferGetBlock(buffer) \\\n\t#define PageGetMaxOffsetNumber(page) \\\n\t#define HeapTupleSatisfiesVisibility(tuple, seeself) \\\n\t#define\theapisoverride() \\\n\nWhen I say linined, I mean I made macros of them. I did not inline the\ncode into other functions, because you loose the modularity of the\nfunction name, and many are called multiple times.\n\nThese were all small functions, only a few lines of actual code. \nBufferIsValid was called several times per row.\n\nWith that said, the execution times did not change at all. What I did\nsee is that the mcount row in gprof output:\n\n 54.3 2.65 2.65 mcount (472)\n\ndecreased from 2.65 to 1.79, which means that less time was spent in\nfunction calls. Not sure why that time did not translate to better\noverall performance, but the inlining is a clear win for these small\nfunctions. mcount is artificially high anyway because it is doing the\nprofiling. The resulting gprof output shows a much smaller list of\nfunctions called 160k times.\n\nA patch will be posted to the patches list, for any of you who subscribe\nto that.\n\n-- \nBruce Momjian | 830 Blythe Avenue\[email protected] | Drexel Hill, Pennsylvania 19026\n + If your life is a hard drive, | (610) 353-9879(w)\n + Christ can be your backup. | (610) 853-3000(h)\n",
"msg_date": "Fri, 24 Apr 1998 11:03:25 -0400 (EDT)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Profile of current backend"
}
] |
[
{
"msg_contents": "Could we please add the ODBC driver to the source tree? I think it would\nlook much better if we distribute a 'complete' system.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tThe Hermit Hacker [SMTP:[email protected]]\n> Sent:\tWednesday, February 04, 1998 7:59 PM\n> To:\tJulia Anne Case\n> Cc:\[email protected]\n> Subject:\tRe: [HACKERS] connection troubles\n> \n> On Wed, 4 Feb 1998, Julia Anne Case wrote:\n> \n> > Quoting The Hermit Hacker ([email protected]):\n> > > Julie...\n> > > \n> > > \tWill PostODBC be v6.3 tested in the next little while? I know\n> > > there have been *alot* of changes to v6.3 as far as authentication\n> is\n> > > concerned, which might prove a problem once we release...\n> > > \n> > \tI just brought down the latest snapshot to my Sparc/Linux\n> > machine... I'll let you know how it works out... Could take a\n> little\n> > while to compile it's just a IPX laptop...\n> \t\n> \n> \tI don't have any knowledge of ODBC, but would it help if I setup\n> and maintained a (reasonably) current server that you can connect\n> remotely\n> too for testing? I'll do periodic recompile/installs of it so that it\n> stays up to date? I'd like to see the ODBC as rock solid as possible\n> before the 1st, in accordance with v6.3 specs, and if it means doing\n> this,\n> its a cheap price to pay :)\n> \n> \tPeter...would this help any with the JDBC stuff too, or are you\n> okay with that one?\n> \n> \n> \n",
"msg_date": "Thu, 5 Feb 1998 09:31:32 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] connection troubles"
},
{
"msg_contents": "On Thu, 5 Feb 1998, Meskes, Michael wrote:\n\n> Could we please add the ODBC driver to the source tree? I think it would\n> look much better if we distribute a 'complete' system.\n\n\tsrc/interfaces/odbc ... been there for a couple of weeks now :)\nJulie has direct access to the source repository, and, like Peter, sort of\nhas to review any patches applied to it...\n\n> Michael\n> \n> --\n> Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> [email protected] | Europark A2, Adenauerstr. 20\n> [email protected] | 52146 Wuerselen\n> Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n> \n> > -----Original Message-----\n> > From:\tThe Hermit Hacker [SMTP:[email protected]]\n> > Sent:\tWednesday, February 04, 1998 7:59 PM\n> > To:\tJulia Anne Case\n> > Cc:\[email protected]\n> > Subject:\tRe: [HACKERS] connection troubles\n> > \n> > On Wed, 4 Feb 1998, Julia Anne Case wrote:\n> > \n> > > Quoting The Hermit Hacker ([email protected]):\n> > > > Julie...\n> > > > \n> > > > \tWill PostODBC be v6.3 tested in the next little while? I know\n> > > > there have been *alot* of changes to v6.3 as far as authentication\n> > is\n> > > > concerned, which might prove a problem once we release...\n> > > > \n> > > \tI just brought down the latest snapshot to my Sparc/Linux\n> > > machine... I'll let you know how it works out... Could take a\n> > little\n> > > while to compile it's just a IPX laptop...\n> > \t\n> > \n> > \tI don't have any knowledge of ODBC, but would it help if I setup\n> > and maintained a (reasonably) current server that you can connect\n> > remotely\n> > too for testing? I'll do periodic recompile/installs of it so that it\n> > stays up to date? I'd like to see the ODBC as rock solid as possible\n> > before the 1st, in accordance with v6.3 specs, and if it means doing\n> > this,\n> > its a cheap price to pay :)\n> > \n> > \tPeter...would this help any with the JDBC stuff too, or are you\n> > okay with that one?\n> > \n> > \n> > \n> \n\n",
"msg_date": "Thu, 5 Feb 1998 06:54:32 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] connection troubles"
}
] |
[
{
"msg_contents": "I also use linux-ELF-libc6 but I don't have a templeate for it. It\nappears the standard linux-ELF template worked and created libc6\nversions of the binaries. Why do you need a special template? If it's\nreally needed we should include IMHO.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\[email protected] [SMTP:[email protected]]\n> Sent:\tThursday, February 05, 1998 12:14 AM\n> To:\[email protected]\n> Subject:\t[HACKERS] Linux: linux.s / tas.s not found...\n> \n> Dear PostgreSQL gurus, \n> \n> with the current CVSup sources, I came across an error\n> message during the configure step. Using my personal\n> configure template, linux-elf-libc6, I stumbled into an\n> error message saying that .../tas.s could not be linked.\n> \n> To get rid of the message, I had to change the configure\n> file:\n> --- snip, snip ---\n> *** 5258,5265 ****\n> EOF\n> \n> cat >> $CONFIG_STATUS <<EOF\n> ! ac_sources=\"backend/port/tas/${os}.s backend/port/dynloader/${os}.c\n> backend/port/dynloader/${os}.h include/port/${os}.h\n> makefiles/Makefile.${os}\"\n> ! ac_dests=\"backend/port/tas.s backend/port/dynloader.c\n> include/dynloader.h include/os.h Makefile.port\"\n> EOF\n> \n> cat >> $CONFIG_STATUS <<\\EOF\n> --- 5261,5269 ----\n> EOF\n> \n> cat >> $CONFIG_STATUS <<EOF\n> ! ac_sources=\"backend/port/dynloader/${os}.c\n> backend/port/dynloader/${os}.h include/port/${os}.h\n> makefiles/Makefile.${os}\"\n> ! ac_dests=\"backend/port/dynloader.c include/dynloader.h include/os.h\n> Makefile.port\"\n> EOF\n> \n> cat >> $CONFIG_STATUS <<\\EOF\n> --- snip, snip ---\n> \n> BTW: my template for Linux-GNU-GLIBC-2.x (aka libc6) looks like this:\n> \n> --- snip, snip ---\n> AROPT:crs\n> CFLAGS:-O2 -m486 -DHAVE_CRYPT_H=1 -DHAVE_UNION_SEMUN=1\n> SHARED_LIB:-fpic\n> ALL:\n> SRCH_INC:/usr/local/include /usr/local/include/ncurses\n> /usr/local/include/readline\n> SRCH_LIB:/usr/local/lib\n> USE_LOCALE:yes\n> DLSUFFIX:.so\n> YFLAGS:-d\n> YACC:bison -y\n> --- snip, snip ---\n> \n> Best regards,\n> \n> Ernst\n",
"msg_date": "Thu, 5 Feb 1998 09:33:42 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Linux: linux.s / tas.s not found..."
},
{
"msg_contents": "Dear fellow PostgreSQL fans,\n\n>I also use linux-ELF-libc6 but I don't have a templeate for it. It\n>appears the standard linux-ELF template worked and created libc6\n>versions of the binaries. Why do you need a special template? If it's\n>really needed we should include IMHO.\n\nwith the current source tree, a special template for linux-libc6 is\n_not_ required, I stand corrected. In a previous version, without it,\nlibcrypt had not been linked into the postmaster on my box, which\ncaused some trouble. In addition, I felt that with the locale support\nincluded in GLIBC-2.x, the USE_LOCALE should be set to yes, but\ncurrently, this does not seem to matter too much ...\n\nRegards,\n\nErnst\n",
"msg_date": "Thu, 5 Feb 1998 09:40:07 GMT",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Linux: linux.s / tas.s not found..."
},
{
"msg_contents": "> caused some trouble. In addition, I felt that with the locale support\n> included in GLIBC-2.x, the USE_LOCALE should be set to yes, but\n> currently, this does not seem to matter too much ...\n\nThe default for _all_ ports is to have USE_LOCALE turned off since there\nis a performance hit to use it. Any installation is welcome to turn it on\nbut it should not be turned on by default.\n\n - Tom\n\n",
"msg_date": "Thu, 05 Feb 1998 15:22:46 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Linux: linux.s / tas.s not found..."
}
] |
[
{
"msg_contents": "I'm currently back to work with version 6.2.1 since I cannot connect with\n6.3 via ODBC.\n\nAnyway, I got my application running after finding a problem (inside the\napp) with data conversion (it read double from a long data field). But it\nseems I also encountered what I believe to be a bug. Since I'm nor sure\nwhether it's known I bring it up here. If I try to insert 199802051215 to a\nfloat8 field it doesn't work because the parser believes this is a long and\ntruncates it to 2147...... Using 199802051215.0 to make sure it's a float\nworks fine. Shouldn't the parser be able to handle this?\n\nMichael\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Thu, 5 Feb 1998 12:12:31 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": true,
"msg_subject": "Bug?"
},
{
"msg_contents": "> I'm currently back to work with version 6.2.1 since I cannot connect with\n> 6.3 via ODBC.\n>\n> Anyway, I got my application running after finding a problem (inside the\n> app) with data conversion (it read double from a long data field). But it\n> seems I also encountered what I believe to be a bug. Since I'm nor sure\n> whether it's known I bring it up here. If I try to insert 199802051215 to a\n> float8 field it doesn't work because the parser believes this is a long and\n> truncates it to 2147...... Using 199802051215.0 to make sure it's a float\n> works fine. Shouldn't the parser be able to handle this?\n\nThe current v6.3beta behavior is this:\n\npostgres=> select 199802051215;\nERROR: Bad integer input '199802051215'\n\nWe had a bit of a discussion about the best way to handle this, and decided to\ntry Bruce's solution to reject the input as a first step. I have patches to do\nthe automatic conversion to a float, but have not applied them.\n\nThe other approach would lead to an error looking like:\n\npostgres=> insert into t values (199802051215.0);\nERROR: pg_atoi: error reading \"199802051215.000000\": Math result not\nrepresentable\n\nI can see arguments for both approaches; do you have a strong opinion either\nway?\n\n - Tom\n\n",
"msg_date": "Thu, 05 Feb 1998 15:39:54 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
},
{
"msg_contents": "> \n> I'm currently back to work with version 6.2.1 since I cannot connect with\n> 6.3 via ODBC.\n> \n> Anyway, I got my application running after finding a problem (inside the\n> app) with data conversion (it read double from a long data field). But it\n> seems I also encountered what I believe to be a bug. Since I'm nor sure\n> whether it's known I bring it up here. If I try to insert 199802051215 to a\n> float8 field it doesn't work because the parser believes this is a long and\n> truncates it to 2147...... Using 199802051215.0 to make sure it's a float\n> works fine. Shouldn't the parser be able to handle this?\n\nI think it converts it to an integer, and then by the time it tries to\nconvert it, it has already chopped off the top of the number. The only\nfix for this would be to read all integers in as 64-bit integers, then\ndo the conversion, but that could be a performance problem.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 11:50:31 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
},
{
"msg_contents": "Bruce Momjian writes:\n> I think it converts it to an integer, and then by the time it tries to\n> convert it, it has already chopped off the top of the number. The only\n> fix for this would be to read all integers in as 64-bit integers, then\n> do the conversion, but that could be a performance problem.\n\nI agree. And performance is important. I think explicit type conversion is\nwhat we should do. Or is it asked for too much if the user has to add a\n::float8 to the number?\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Fri, 6 Feb 1998 14:25:24 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Bug?"
},
{
"msg_contents": "> I think it converts it to an integer, and then by the time it tries to\n> convert it, it has already chopped off the top of the number. The only\n> fix for this would be to read all integers in as 64-bit integers, then\n> do the conversion, but that could be a performance problem.\n\nWell, the other possibility is to try converting to float8 only if the int4\nconversion fails. If both fail, then throw an elog(ERROR). I have patches for\nthis...\n\n",
"msg_date": "Fri, 06 Feb 1998 15:53:08 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
},
{
"msg_contents": "Bruce Momjian said:\n\n> The only fix for this would be to read all integers in as 64-bit\n> integers, then do the conversion, but that could be a performance\n> problem.\n\nMichael Meskes answered:\n\n> I agree. And performance is important. I think explicit type\n> conversion is what we should do. Or is it asked for too much if the\n> user has to add a ::float8 to the number?\n\nAm I being dense here? Can there really be a significant performance\nhit in the parsing of a query? Let's say that it takes a millisecond\nextra to do the right thing with a number. Does it matter? How many\nqueries per second can we expect to process anyway?\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n\n",
"msg_date": "Fri, 6 Feb 1998 17:55:57 +0100 (CET)",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
},
{
"msg_contents": "> \n> Bruce Momjian said:\n> \n> > The only fix for this would be to read all integers in as 64-bit\n> > integers, then do the conversion, but that could be a performance\n> > problem.\n> \n> Michael Meskes answered:\n> \n> > I agree. And performance is important. I think explicit type\n> > conversion is what we should do. Or is it asked for too much if the\n> > user has to add a ::float8 to the number?\n> \n> Am I being dense here? Can there really be a significant performance\n> hit in the parsing of a query? Let's say that it takes a millisecond\n> extra to do the right thing with a number. Does it matter? How many\n> queries per second can we expect to process anyway?\n\nI don't know of standard ways to read in 64-bit integers.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 6 Feb 1998 13:00:12 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
},
{
"msg_contents": "On Fri, 6 Feb 1998, Bruce Momjian wrote:\n\n> > Am I being dense here? Can there really be a significant performance\n> > hit in the parsing of a query? Let's say that it takes a millisecond\n> > extra to do the right thing with a number. Does it matter? How many\n> > queries per second can we expect to process anyway?\n> \n> I don't know of standard ways to read in 64-bit integers.\n\nOh, sorry -- I wasn't being clear. Of course you don't, since we\ndon't even have standard 64-bit integers. My point was that I\ncouldn't see was how special handling of constants during the parsing\nof the query string could have significant performance impact, even if\nyou did read them as 64-bit integers, which would mean adding bignum\ncode to PostgreSQL. In other words, performance isn't the argument to\nbe used against doing the right thing during parsing.\n\nAs for implementation, I was thinking more along the lines of:\n\n{integer}\t\t{\n\t\t\t\t\tchar* endptr;\n\n\t\t\t\t\terrno = 0;\n\t\t\t\t\tyylval.ival = strtol((char *)yytext,&endptr,10);\n\t\t\t\t\tif (*endptr != '\\0' || errno == ERANGE) {\n\t\t\t\t\t\terrno = 0;\n\t\t\t\t\t\tyylval.dval = strtod(((char *)yytext),&endptr);\n\t\t\t\t\t\tif (*endptr != '\\0' || errno == ERANGE) {\n\t\t\t\t\t\t\telog(ERROR,\"Bad integer input '%s'\",yytext);\n\t\t\t\t\t\t\treturn (ICONST);\n\t\t\t\t\t\t}\n\t\t\t\t\t\tCheckFloat8Val(yylval.dval);\n\t\t\t\t\t\treturn (FCONST);\n\t\t\t\t\t}\n\t\t\t\t\treturn (ICONST);\n\t\t\t\t}\n\nHowever: do we really want to do this anyway? If you demand that the\nuser indicate whether a given constant is integer or real, you lessen\nthe risk of doing the wrong thing with his or her data. Specifically,\ngoing to floating point means giving up accuracy in representation,\nand this may not be something we want to do without user permission.\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n\n",
"msg_date": "Sat, 7 Feb 1998 04:53:32 +0100 (CET)",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
},
{
"msg_contents": "> However: do we really want to do this anyway? If you demand that the\n> user indicate whether a given constant is integer or real, you lessen\n> the risk of doing the wrong thing with his or her data. Specifically,\n> going to floating point means giving up accuracy in representation,\n> and this may not be something we want to do without user permission.\n> \n\nI never auto-convert integer to float unless I have to.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sat, 7 Feb 1998 00:36:26 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
}
] |
[
{
"msg_contents": "Oops. .-)\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tThe Hermit Hacker [SMTP:[email protected]]\n> Sent:\tThursday, February 05, 1998 12:55 PM\n> To:\tMeskes, Michael\n> Cc:\tJulia Anne Case; [email protected]\n> Subject:\tRE: [HACKERS] connection troubles\n> \n> On Thu, 5 Feb 1998, Meskes, Michael wrote:\n> \n> > Could we please add the ODBC driver to the source tree? I think it\n> would\n> > look much better if we distribute a 'complete' system.\n> \n> \tsrc/interfaces/odbc ... been there for a couple of weeks now :)\n> Julie has direct access to the source repository, and, like Peter,\n> sort of\n> has to review any patches applied to it...\n> \n> > Michael\n> > \n> > --\n> > Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> > [email protected] | Europark A2, Adenauerstr.\n> 20\n> > [email protected] | 52146 Wuerselen\n> > Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> > Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n> > \n> > > -----Original Message-----\n> > > From:\tThe Hermit Hacker [SMTP:[email protected]]\n> > > Sent:\tWednesday, February 04, 1998 7:59 PM\n> > > To:\tJulia Anne Case\n> > > Cc:\[email protected]\n> > > Subject:\tRe: [HACKERS] connection troubles\n> > > \n> > > On Wed, 4 Feb 1998, Julia Anne Case wrote:\n> > > \n> > > > Quoting The Hermit Hacker ([email protected]):\n> > > > > Julie...\n> > > > > \n> > > > > \tWill PostODBC be v6.3 tested in the next little while?\n> I know\n> > > > > there have been *alot* of changes to v6.3 as far as\n> authentication\n> > > is\n> > > > > concerned, which might prove a problem once we release...\n> > > > > \n> > > > \tI just brought down the latest snapshot to my\n> Sparc/Linux\n> > > > machine... I'll let you know how it works out... Could take a\n> > > little\n> > > > while to compile it's just a IPX laptop...\n> > > \t\n> > > \n> > > \tI don't have any knowledge of ODBC, but would it help if I setup\n> > > and maintained a (reasonably) current server that you can connect\n> > > remotely\n> > > too for testing? I'll do periodic recompile/installs of it so\n> that it\n> > > stays up to date? I'd like to see the ODBC as rock solid as\n> possible\n> > > before the 1st, in accordance with v6.3 specs, and if it means\n> doing\n> > > this,\n> > > its a cheap price to pay :)\n> > > \n> > > \tPeter...would this help any with the JDBC stuff too, or are you\n> > > okay with that one?\n> > > \n> > > \n> > > \n> > \n> \n",
"msg_date": "Thu, 5 Feb 1998 12:52:41 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] connection troubles"
}
] |
[
{
"msg_contents": "subscribe\n",
"msg_date": "Thu, 5 Feb 1998 09:15:08 -0500 (EST)",
"msg_from": "[email protected]",
"msg_from_op": true,
"msg_subject": "None"
}
] |
[
{
"msg_contents": "Hi,\n\n after spending 3 hours in gdb I found it :-)\n\n Symptom:\n select from a table with attrs (a int, b char(20))\n crashed in bpcharout() (palloc of -1 bytes). But a table\n with attrs (a int, b varchar(20)) worked.\n\n Bug location:\n Function nocachegetattr() in access/common/heaptuple.c\n line 557. Really surprising - and really hard to find -\n phew.\n\n\nUntil later, Jan\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] (Jan Wieck) #\n\n\n*** heaptuple.c\tThu Feb 5 15:21:50 1998\n--- heaptuple.c.old\tThu Feb 5 15:21:39 1998\n***************\n*** 554,560 ****\n \t\twhile (att[j]->attcacheoff > 0)\n \t\t\tj++;\n \n! \t\tif (!VARLENA_FIXED_SIZE(att[j - 1]))\n \t\t\toff = att[j - 1]->attcacheoff + att[j - 1]->attlen;\n \t\telse\n \t\t\toff = att[j - 1]->attcacheoff + att[j - 1]->atttypmod;\n--- 554,560 ----\n \t\twhile (att[j]->attcacheoff > 0)\n \t\t\tj++;\n \n! \t\tif (!VARLENA_FIXED_SIZE(att[j]))\n \t\t\toff = att[j - 1]->attcacheoff + att[j - 1]->attlen;\n \t\telse\n \t\t\toff = att[j - 1]->attcacheoff + att[j - 1]->atttypmod;\n",
"msg_date": "Thu, 5 Feb 1998 15:32:19 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Bugfix for bpchar"
}
] |
[
{
"msg_contents": "I've just send the tar file to the patches list (hope that was correct). I'd\nlike to have it included in the archive before starting more work on it.\n\nWe have to decide which extension to use for embedded SQL code. Oracle\nalready uses *.pc so that doesn't look good. Or shall we make it as\ncompatible to Oracle as possible and use *.pc. Alternatively, how about\n*.pqc?\n\nMichael\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Thu, 5 Feb 1998 15:50:54 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": true,
"msg_subject": "Preprocessor"
},
{
"msg_contents": "> \n> I've just send the tar file to the patches list (hope that was correct). I'd\n> like to have it included in the archive before starting more work on it.\n> \n> We have to decide which extension to use for embedded SQL code. Oracle\n> already uses *.pc so that doesn't look good. Or shall we make it as\n> compatible to Oracle as possible and use *.pc. Alternatively, how about\n> *.pqc?\n\n *.pgc maybe\n\n> \n> Michael\n> -- \n> Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> [email protected] | Europark A2, Adenauerstr. 20\n> [email protected] | 52146 Wuerselen\n> Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n> \n> \n\n\nUntil later, Jan\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] (Jan Wieck) #\n\n",
"msg_date": "Thu, 5 Feb 1998 16:49:05 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Preprocessor"
},
{
"msg_contents": "Michael Meskes wrote:\n\n> I've just send the tar file to the patches list (hope that was correct). I'd\n> like to have it included in the archive before starting more work on it.\n>\n> We have to decide which extension to use for embedded SQL code. Oracle\n> already uses *.pc so that doesn't look good. Or shall we make it as\n> compatible to Oracle as possible and use *.pc. Alternatively, how about\n> *.pqc?\n\nIngres uses \".sc\", which I imagine represented \"sql c\". Applix ships files for\ntheir database interface labeled with \".ec\", which I assume represents\n\"embedded c\". I like the \".sc\" myself, but that may be because I've worked with\nit for years :) I would think any label you choose does not need to be unique.\nWhat does \".pc\" indicate?\n\n - Tom\n\n",
"msg_date": "Thu, 05 Feb 1998 16:13:52 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Preprocessor"
},
{
"msg_contents": "On Thu, 5 Feb 1998, Thomas G. Lockhart wrote:\n\n> Michael Meskes wrote:\n> \n> > I've just send the tar file to the patches list (hope that was correct). I'd\n> > like to have it included in the archive before starting more work on it.\n> >\n> > We have to decide which extension to use for embedded SQL code. Oracle\n> > already uses *.pc so that doesn't look good. Or shall we make it as\n> > compatible to Oracle as possible and use *.pc. Alternatively, how about\n> > *.pqc?\n> \n> Ingres uses \".sc\", which I imagine represented \"sql c\". Applix ships files for\n> their database interface labeled with \".ec\", which I assume represents\n> \"embedded c\". I like the \".sc\" myself, but that may be because I've worked with\n> it for years :) I would think any label you choose does not need to be unique.\n> What does \".pc\" indicate?\n\n\tpostgresql c?\n\n\n",
"msg_date": "Thu, 5 Feb 1998 12:15:24 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Preprocessor"
},
{
"msg_contents": "Thomas G. Lockhart writes:\n> Michael Meskes wrote:\n> \n> > I've just send the tar file to the patches list (hope that was correct). I'd\n> > like to have it included in the archive before starting more work on it.\n> >\n> > We have to decide which extension to use for embedded SQL code. Oracle\n> > already uses *.pc so that doesn't look good. Or shall we make it as\n> > compatible to Oracle as possible and use *.pc. Alternatively, how about\n> > *.pqc?\n> \n> Ingres uses \".sc\", which I imagine represented \"sql c\". Applix ships files for\n> their database interface labeled with \".ec\", which I assume represents\n> \"embedded c\". I like the \".sc\" myself, but that may be because I've worked with\n> it for years :) I would think any label you choose does not need to be unique.\n> What does \".pc\" indicate?\n\n.pc means Proc-C with proc being Oracle's preprocessor.\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Fri, 6 Feb 1998 09:42:57 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Preprocessor"
}
] |
[
{
"msg_contents": "> There's an error while inserting data into a column defined as VARCHAR without\n> length in PostgreSQL 6.3.\n> Take a look:\n>\n> postgres=> CREATE TABLE prova ( uno VARCHAR(10) );\n> CREATE\n> postgres=> INSERT INTO prova VALUES ( 'OK' );\n> INSERT 153042 1\n>\n> postgres=> CREATE TABLE prova2 ( uno VARCHAR );\n> CREATE\n> postgres=> INSERT INTO prova2 VALUES ( 'NOT OK' );\n> PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> This probably means the backend terminated abnormally before or while processing the request.\n\nJose, please post problem reports to hackers, not to questions, on not-yet-released versions. I've moved\nthis to hackers.\n\nbtw, I saw this too but in a different context and wasn't certain if it was a new problem:\n\npostgres=> select char_length('hi'::text);\nlength\n------\n 2\n(1 row)\n\npostgres=> select character_length('hi'::varchar);\nPQexec() -- Request was sent to backend, but backend closed the channel before responding.\n This probably means the backend terminated abnormally before or while processing the request.\n\n - Tom\n\n",
"msg_date": "Thu, 05 Feb 1998 15:18:27 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [QUESTIONS] impossible insert data into VARCHAR"
},
{
"msg_contents": "Hi,\n\n>\n> > There's an error while inserting data into a column defined as VARCHAR without\n> > length in PostgreSQL 6.3.\n> > Take a look:\n> >\n> > postgres=> CREATE TABLE prova ( uno VARCHAR(10) );\n> > CREATE\n> > postgres=> INSERT INTO prova VALUES ( 'OK' );\n> > INSERT 153042 1\n> >\n> > postgres=> CREATE TABLE prova2 ( uno VARCHAR );\n> > CREATE\n> > postgres=> INSERT INTO prova2 VALUES ( 'NOT OK' );\n> > PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> > This probably means the backend terminated abnormally before or while processing the request.\n>\n> Jose, please post problem reports to hackers, not to questions, on not-yet-released versions. I've moved\n> this to hackers.\n>\n> btw, I saw this too but in a different context and wasn't certain if it was a new problem:\n>\n> postgres=> select char_length('hi'::text);\n> length\n> ------\n> 2\n> (1 row)\n>\n> postgres=> select character_length('hi'::varchar);\n> PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> This probably means the backend terminated abnormally before or while processing the request.\n>\n> - Tom\n\n ARRRRG - then next problem with VARSIZE - damn thing.\n\n This time it's during the parser (gdb told me). varcharin()\n is called with a atttypmod of 0 causing a palloc() of 0\n bytes. How should a VARCHAR type whithout a specified length\n behave? Is this type 1 character or a variable size up to\n 4096?\n\n\nUntil later, Jan\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] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 5 Feb 1998 17:06:20 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] impossible insert data into VARCHAR"
},
{
"msg_contents": "Jan Wieck wrote:\n\n> Hi,\n>\n> >\n> > > There's an error while inserting data into a column defined as VARCHAR without\n> > > length in PostgreSQL 6.3.\n> > > Take a look:\n> > >\n> > > postgres=> CREATE TABLE prova ( uno VARCHAR(10) );\n> > > CREATE\n> > > postgres=> INSERT INTO prova VALUES ( 'OK' );\n> > > INSERT 153042 1\n> > >\n> > > postgres=> CREATE TABLE prova2 ( uno VARCHAR );\n> > > CREATE\n> > > postgres=> INSERT INTO prova2 VALUES ( 'NOT OK' );\n> > > PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> > > This probably means the backend terminated abnormally before or while processing the request.\n> >\n> > Jose, please post problem reports to hackers, not to questions, on not-yet-released versions. I've moved\n> > this to hackers.\n> >\n> > btw, I saw this too but in a different context and wasn't certain if it was a new problem:\n> >\n> > postgres=> select char_length('hi'::text);\n> > length\n> > ------\n> > 2\n> > (1 row)\n> >\n> > postgres=> select character_length('hi'::varchar);\n> > PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> > This probably means the backend terminated abnormally before or while processing the request.\n> >\n> > - Tom\n>\n> ARRRRG - then next problem with VARSIZE - damn thing.\n>\n> This time it's during the parser (gdb told me). varcharin()\n> is called with a atttypmod of 0 causing a palloc() of 0\n> bytes. How should a VARCHAR type whithout a specified length\n> behave? Is this type 1 character or a variable size up to\n> 4096?\n\nIt should allow any length (up to the implementation maximum of 4096). The usage is not defined for SQL92,\nunlike the case for \"char\" with no explicit size which defaults to one.\n\n - Tom\n\n",
"msg_date": "Thu, 05 Feb 1998 16:18:25 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] impossible insert data into VARCHAR"
},
{
"msg_contents": ">\n> Jan Wieck wrote:\n>\n> > > [... quoting level exceeded ...]\n> >\n> > ARRRRG - then next problem with VARSIZE - damn thing.\n> >\n> > This time it's during the parser (gdb told me). varcharin()\n> > is called with a atttypmod of 0 causing a palloc() of 0\n> > bytes. How should a VARCHAR type whithout a specified length\n> > behave? Is this type 1 character or a variable size up to\n> > 4096?\n>\n> It should allow any length (up to the implementation maximum of 4096). The usage is not defined for SQL92,\n> unlike the case for \"char\" with no explicit size which defaults to one.\n>\n\n Could be hacked into varcharin(). If the test on atttypmod is\n changed from \"!= -1\" into \"> 0\" it works that way. Don't\n have the time right now to run a regression test, but some\n typed queries with\n\n select into t values ...\n\n and\n\n select into t select ...\n\n worked O.K.\n\n\nUntil later, Jan\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] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 5 Feb 1998 17:54:14 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] impossible insert data into VARCHAR"
},
{
"msg_contents": "Do we need to disallow varchar() and char() without a length?\n\n> \n> > There's an error while inserting data into a column defined as VARCHAR without\n> > length in PostgreSQL 6.3.\n> > Take a look:\n> >\n> > postgres=> CREATE TABLE prova ( uno VARCHAR(10) );\n> > CREATE\n> > postgres=> INSERT INTO prova VALUES ( 'OK' );\n> > INSERT 153042 1\n> >\n> > postgres=> CREATE TABLE prova2 ( uno VARCHAR );\n> > CREATE\n> > postgres=> INSERT INTO prova2 VALUES ( 'NOT OK' );\n> > PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> > This probably means the backend terminated abnormally before or while processing the request.\n> \n> Jose, please post problem reports to hackers, not to questions, on not-yet-released versions. I've moved\n> this to hackers.\n> \n> btw, I saw this too but in a different context and wasn't certain if it was a new problem:\n> \n> postgres=> select char_length('hi'::text);\n> length\n> ------\n> 2\n> (1 row)\n> \n> postgres=> select character_length('hi'::varchar);\n> PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> This probably means the backend terminated abnormally before or while processing the request.\n> \n> - Tom\n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 12:00:48 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] impossible insert data into VARCHAR"
},
{
"msg_contents": "On Thu, 5 Feb 1998, Thomas G. Lockhart wrote:\n\n> > There's an error while inserting data into a column defined as VARCHAR without\n> > length in PostgreSQL 6.3.\n> > Take a look:\n> >\n> > postgres=> CREATE TABLE prova ( uno VARCHAR(10) );\n> > CREATE\n> > postgres=> INSERT INTO prova VALUES ( 'OK' );\n> > INSERT 153042 1\n> >\n> > postgres=> CREATE TABLE prova2 ( uno VARCHAR );\n> > CREATE\n> > postgres=> INSERT INTO prova2 VALUES ( 'NOT OK' );\n> > PQexec() -- Request was sent to backend, but backend closed the channel before responding.\n> > This probably means the backend terminated abnormally before or while processing the request.\n> \n> Jose, please post problem reports to hackers, not to questions, on not-yet-released versions. I've moved\n> this to hackers.\nOk, sorry.\n> \n> btw, I saw this too but in a different context and wasn't certain if it was a new problem:\n> \n> postgres=> select char_length('hi'::text);\n> length\n> ------\n> 2\n> (1 row)\n> \n\nI've another problem about connection\nwhen I try to connect to PostgreSQL using:\n psql postgres -h server\nI have this error:\n\nConnection to database 'postgres' failed.\nUser authentication failed\n\n\nI'd ran postmaster with -i as:\n\npostmaster -i -o -F -B 512 -S > server.log 2>&1\n\nAny idea?\n Thanks, Jose'\n\n",
"msg_date": "Thu, 5 Feb 1998 17:15:55 +0000 (GMT)",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "connection error"
},
{
"msg_contents": "I am applying such a fix right now.\n\n> \n> Could be hacked into varcharin(). If the test on atttypmod is\n> changed from \"!= -1\" into \"> 0\" it works that way. Don't\n> have the time right now to run a regression test, but some\n> typed queries with\n> \n> select into t values ...\n> \n> and\n> \n> select into t select ...\n> \n> worked O.K.\n> \n> \n> Until later, Jan\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] (Jan Wieck) #\n> \n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 12:18:28 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] impossible insert data into VARCHAR"
},
{
"msg_contents": "> Do we need to disallow varchar() and char() without a length?\n\nThe question might be moot since Jan's suggestion may fix the behavior. But, just for completeness...\n\nchar, char(), and varchar() are SQL92 character types. varchar - with no arguments - is not, and seems as\nthough it is functionally equivalent to \"text\". Could do without it if necessary, but it is a nice extension.\n\n - Tom\n\n",
"msg_date": "Fri, 06 Feb 1998 02:28:07 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] impossible insert data into VARCHAR"
}
] |
[
{
"msg_contents": "Oops, that's what I meant to say. :-)\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\[email protected] [SMTP:[email protected]]\n> Sent:\tThursday, February 05, 1998 4:49 PM\n> To:\[email protected]\n> Cc:\[email protected]\n> Subject:\tRe: [HACKERS] Preprocessor\n> \n> > \n> > I've just send the tar file to the patches list (hope that was\n> correct). I'd\n> > like to have it included in the archive before starting more work on\n> it.\n> > \n> > We have to decide which extension to use for embedded SQL code.\n> Oracle\n> > already uses *.pc so that doesn't look good. Or shall we make it as\n> > compatible to Oracle as possible and use *.pc. Alternatively, how\n> about\n> > *.pqc?\n> \n> *.pgc maybe\n> \n> > \n> > Michael\n> > -- \n> > Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> > [email protected] | Europark A2, Adenauerstr.\n> 20\n> > [email protected] | 52146 Wuerselen\n> > Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> > Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n> > \n> > \n> \n> \n> Until later, Jan\n> \n> -- \n> \n> #=====================================================================\n> =#\n> # It's easier to get forgiveness for being wrong than for being right.\n> #\n> # Let's break this rule - forgive me.\n> #\n> #======================================== [email protected] (Jan Wieck)\n> #\n> \n",
"msg_date": "Thu, 5 Feb 1998 16:43:18 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Preprocessor"
}
] |
[
{
"msg_contents": "Is there a way to explicitely tell the parser which kind of variable we\nwant? That is something like float(1) to get 1.0.\n\nIt's obviously easier in Oracle (where my source came from) since they\nonly have one number type. But expclicitely converting would be okay\nIMO.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tThomas G. Lockhart [SMTP:[email protected]]\n> Sent:\tThursday, February 05, 1998 4:40 PM\n> To:\tMichael Meskes\n> Cc:\tPostgreSQL Hacker\n> Subject:\tRe: [HACKERS] Bug?\n> \n> > I'm currently back to work with version 6.2.1 since I cannot connect\n> with\n> > 6.3 via ODBC.\n> >\n> > Anyway, I got my application running after finding a problem (inside\n> the\n> > app) with data conversion (it read double from a long data field).\n> But it\n> > seems I also encountered what I believe to be a bug. Since I'm nor\n> sure\n> > whether it's known I bring it up here. If I try to insert\n> 199802051215 to a\n> > float8 field it doesn't work because the parser believes this is a\n> long and\n> > truncates it to 2147...... Using 199802051215.0 to make sure it's a\n> float\n> > works fine. Shouldn't the parser be able to handle this?\n> \n> The current v6.3beta behavior is this:\n> \n> postgres=> select 199802051215;\n> ERROR: Bad integer input '199802051215'\n> \n> We had a bit of a discussion about the best way to handle this, and\n> decided to\n> try Bruce's solution to reject the input as a first step. I have\n> patches to do\n> the automatic conversion to a float, but have not applied them.\n> \n> The other approach would lead to an error looking like:\n> \n> postgres=> insert into t values (199802051215.0);\n> ERROR: pg_atoi: error reading \"199802051215.000000\": Math result not\n> representable\n> \n> I can see arguments for both approaches; do you have a strong opinion\n> either\n> way?\n> \n> - Tom\n",
"msg_date": "Thu, 5 Feb 1998 16:45:17 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Bug?"
},
{
"msg_contents": "> \n> Is there a way to explicitely tell the parser which kind of variable we\n> want? That is something like float(1) to get 1.0.\n\n '1'::float8\n\n> \n> It's obviously easier in Oracle (where my source came from) since they\n> only have one number type. But expclicitely converting would be okay\n> IMO.\n> \n> Michael\n> \n> --\n> Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> [email protected] | Europark A2, Adenauerstr. 20\n> [email protected] | 52146 Wuerselen\n> Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n\nUntil later, Jan\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] (Jan Wieck) #\n\n",
"msg_date": "Thu, 5 Feb 1998 17:16:01 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
},
{
"msg_contents": "Jan Wieck writes:\n> > \n> > Is there a way to explicitely tell the parser which kind of variable we\n> > want? That is something like float(1) to get 1.0.\n> \n> '1'::float8\n\nI'll try that for a complex statement. Seems I have to learn quite a lot\nabout postgresql. :-)\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Fri, 6 Feb 1998 10:09:21 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
}
] |
[
{
"msg_contents": "[Moved to hackers.]\n\nCan someone take this on?\n\n> \n> > \n> > Great, thanks for testing. I have a speedup patch I am working on now\n> > that will help with any table that has char(), and I think you probably\n> > have many of them. You will see my posting on the questions list.\n> > \n> \n> Will test it that's for sure but I would like to ask if it's possible \n> that you run also the crash-me test and the benchmark. When running \n> the crash-me test some strange things happened. I didn't had time to \n> take a look at it but it crashed (postgres) that's for sure .... \n> Further I got some errors I think is interesting for development \n> while running the benchmark.\n> Some other things I found:\n> postgreSQL isn't compiling with pgcc.\n> the memory usage of postgres is big. Very big. Trying to test the \n> maximal number of joins is taking 500 mb of memory testing a join of \n> 55 tables. When I run vacuum postgres takes approx 128 mb of memory. \n> That's a lot I think.\n> \n> When there is a new snapshot out just let me know and I will try to \n> test it ... \n> \n> Greetz...\n> \n> Luuk\n> |-----------------------------------------------------------------------|\n> | Luuk de Boer -> <[email protected]> |\n> | When you want to go fast ..... |\n> | Drive a cobra on the road and take a T3 line for internet ...... :-) |\n> | http://www.luuk.com/ -> The AC Cobra site |\n> |-----------------------------------------------------------------------| \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 11:45:17 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [QUESTIONS] MySQL benchmark page"
},
{
"msg_contents": "On Thu, 5 Feb 1998, Bruce Momjian wrote:\n\n> [Moved to hackers.]\n> \n> Can someone take this on?\n\n\tI have it all downloaded at home...I dive into it and get it\nincluded as part of our src/test directory...\n\n\n\n> \n> > \n> > > \n> > > Great, thanks for testing. I have a speedup patch I am working on now\n> > > that will help with any table that has char(), and I think you probably\n> > > have many of them. You will see my posting on the questions list.\n> > > \n> > \n> > Will test it that's for sure but I would like to ask if it's possible \n> > that you run also the crash-me test and the benchmark. When running \n> > the crash-me test some strange things happened. I didn't had time to \n> > take a look at it but it crashed (postgres) that's for sure .... \n> > Further I got some errors I think is interesting for development \n> > while running the benchmark.\n> > Some other things I found:\n> > postgreSQL isn't compiling with pgcc.\n> > the memory usage of postgres is big. Very big. Trying to test the \n> > maximal number of joins is taking 500 mb of memory testing a join of \n> > 55 tables. When I run vacuum postgres takes approx 128 mb of memory. \n> > That's a lot I think.\n> > \n> > When there is a new snapshot out just let me know and I will try to \n> > test it ... \n> > \n> > Greetz...\n> > \n> > Luuk\n> > |-----------------------------------------------------------------------|\n> > | Luuk de Boer -> <[email protected]> |\n> > | When you want to go fast ..... |\n> > | Drive a cobra on the road and take a T3 line for internet ...... :-) |\n> > | http://www.luuk.com/ -> The AC Cobra site |\n> > |-----------------------------------------------------------------------| \n> > \n> \n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n\n",
"msg_date": "Thu, 5 Feb 1998 12:46:57 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [QUESTIONS] MySQL benchmark page"
}
] |
[
{
"msg_contents": "Using a snapshot from two days ago on a i586-pc-linux-gnu using\nlinux-elf template:\n\ntest=> create table test (t1 integer);\nCREATE\ntest=> create function myfunc (integer) returns integer\ntest-> as 'select $1' language 'sql';\nERROR: ProcedureCreate: arg type 'integer' is not defined\ntest=> create function myfunc(int4) returns int4\ntest-> as 'select $1' language 'sql';\nCREATE\n\nI can use the integer type in tables, but not in functions. Is this a\nbug?\n\nEwan Mellor.\n",
"msg_date": "Thu, 05 Feb 1998 17:00:15 +0000",
"msg_from": "Ewan Mellor <[email protected]>",
"msg_from_op": true,
"msg_subject": "create function bug?"
},
{
"msg_contents": "\nI noticed that too, but since I don't use 'integer' in tables, I\nfigured postgres didn't have the \"integer\" type (which probably should\nhave struck me as odd). It might be one of these silly grammer\nrewrites, in which case (if there isn't central type/function parsing\ncode it seems like more trouble than it's worth, I mean why not just\nhave the extra type/functions in the catalogs? or have aliases somehow?)\n\nI'm sure you can work around it ;)\n\nOn Thu, 5 February 1998, at 17:00:15, Ewan Mellor wrote:\n\n> Using a snapshot from two days ago on a i586-pc-linux-gnu using\n> linux-elf template:\n> \n> test=> create table test (t1 integer);\n> CREATE\n> test=> create function myfunc (integer) returns integer\n> test-> as 'select $1' language 'sql';\n> ERROR: ProcedureCreate: arg type 'integer' is not defined\n> test=> create function myfunc(int4) returns int4\n> test-> as 'select $1' language 'sql';\n> CREATE\n> \n> I can use the integer type in tables, but not in functions. Is this a\n> bug?\n> \n> Ewan Mellor.\n",
"msg_date": "Thu, 5 Feb 1998 11:05:19 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] create function bug?"
},
{
"msg_contents": "\nAfter going over the grammer and function code, there's a little bit\nof weirdness I see. I don't know the correct terminology, but the\ngrammer for typname is a little odd. in the first case (i.e. txname)\nit does some type translating itself (flort => float8, decimal =>\ninteger, numeric => integer) before calling xlateSqlType, which does\nthe same thing, but with the following types: (int/integer => int4,\nsmalling => int2, real => float8, interval => timespan)\n\nin the second case (SETOF txname), none of this pre-xlateSqlType\ntranslating is done.\n\nis it safe to assume someone put this code in the wrong place (not in\nxlateSqlType so it would apply to all types)? is the grammar the best\nplace for this \"type aliasing\"? would it be safe to modify the create\nfunction grammer to call xlateSqlType (so long as opaque does not get\naliased to anything? ;)\n\n--brett\n\n> On Thu, 5 February 1998, at 17:00:15, Ewan Mellor wrote:\n> \n> > Using a snapshot from two days ago on a i586-pc-linux-gnu using\n> > linux-elf template:\n> > \n> > test=> create table test (t1 integer);\n> > CREATE\n> > test=> create function myfunc (integer) returns integer\n> > test-> as 'select $1' language 'sql';\n> > ERROR: ProcedureCreate: arg type 'integer' is not defined\n> > test=> create function myfunc(int4) returns int4\n> > test-> as 'select $1' language 'sql';\n> > CREATE\n> > \n> > I can use the integer type in tables, but not in functions. Is this a\n> > bug?\n> > \n> > Ewan Mellor.\n",
"msg_date": "Thu, 5 Feb 1998 13:14:30 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] create function bug?"
},
{
"msg_contents": "\nOn Thu, 5 Feb 1998, Ewan Mellor wrote:\n\n> test=> create function myfunc (integer) returns integer\n> test-> as 'select $1' language 'sql';\n\n On most PLs that I used, you need to specify a variable name for the\nparam, after all, how are you going to refer to it?\n\n Try:\n\ncreate function myfunc (myint integer) returns integer\nas 'select $1' language 'sql';\n\n\nTom\n\n",
"msg_date": "Thu, 5 Feb 1998 13:48:10 -0800 (PST)",
"msg_from": "Tom <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] create function bug?"
},
{
"msg_contents": "> is it safe to assume someone put this code in the wrong place (not in\n> xlateSqlType so it would apply to all types)? is the grammar the best\n> place for this \"type aliasing\"? would it be safe to modify the create\n> function grammer to call xlateSqlType (so long as opaque does not get\n> aliased to anything? ;)\n> > > Using a snapshot from two days ago on a i586-pc-linux-gnu using\n> > > linux-elf template: <snip>\n> > > I can use the integer type in tables, but not in functions. Is this a\n> > > bug?\n\nThis behavior must have been in the code for a long time. Anyway, I'll fix it\nfor v6.3 (I have some other similar fixes for function names ready to commit\nand I'll do both at the same time).\n\n\"integer\" is translated into \"int4\" in the parser, to avoid having to\nreplicate the declarations for the support code. Specify the actual type name\nand your declaration will work now.\n\n - Tom\n\n",
"msg_date": "Fri, 06 Feb 1998 02:53:37 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] create function bug?"
}
] |
[
{
"msg_contents": " This time it's during the parser (gdb told me). varcharin()\n is called with a atttypmod of 0 causing a palloc() of 0\n bytes. How should a VARCHAR type whithout a specified length\n behave? Is this type 1 character or a variable size up to\n 4096?\n\nI thought we fixed this on Feb 3. look at\nparse_expr.c line 104: it should read: \n if (con->typename != NULL)\n! result = parser_typecast(val,\ncon->typename, -1);\n else\n\nI think all funcs calling with atttypmod = 0 are allways wrong, should\nbe -1.\nor a number > 0 (or 4 if atttypmod includes the VARHDRSZ don't know) \n\nAndreas\n",
"msg_date": "Thu, 5 Feb 1998 18:23:59 +0100 ",
"msg_from": "Zeugswetter Andreas DBT <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: atttypmod of 0"
},
{
"msg_contents": "> \n> This time it's during the parser (gdb told me). varcharin()\n> is called with a atttypmod of 0 causing a palloc() of 0\n> bytes. How should a VARCHAR type whithout a specified length\n> behave? Is this type 1 character or a variable size up to\n> 4096?\n> \n> I thought we fixed this on Feb 3. look at\n> parse_expr.c line 104: it should read: \n> if (con->typename != NULL)\n> ! result = parser_typecast(val,\n> con->typename, -1);\n> else\n> \n> I think all funcs calling with atttypmod = 0 are allways wrong, should\n> be -1.\n> or a number > 0 (or 4 if atttypmod includes the VARHDRSZ don't know) \n> \n> Andreas\n> \n> \n\nYes, we did, but now I fixed varcharin, and bpcharin to test for\natttypmod of 0 and do the right thing, I think. If we need to make the\ndefault atttypmod value -1, then we can change it back. Let me know if\nthe current fix does not work.\n\nShould I make atttypmod default to -1?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 12:41:44 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> \n> This time it's during the parser (gdb told me). varcharin()\n> is called with a atttypmod of 0 causing a palloc() of 0\n> bytes. How should a VARCHAR type whithout a specified length\n> behave? Is this type 1 character or a variable size up to\n> 4096?\n> \n> I thought we fixed this on Feb 3. look at\n> parse_expr.c line 104: it should read: \n> if (con->typename != NULL)\n> ! result = parser_typecast(val,\n> con->typename, -1);\n> else\n> \n> I think all funcs calling with atttypmod = 0 are allways wrong, should\n> be -1.\n> or a number > 0 (or 4 if atttypmod includes the VARHDRSZ don't know) \n\nIt does include the VARHDRSZ.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 12:57:21 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> \n> > \n> > This time it's during the parser (gdb told me). varcharin()\n> > is called with a atttypmod of 0 causing a palloc() of 0\n> > bytes. How should a VARCHAR type whithout a specified length\n> > behave? Is this type 1 character or a variable size up to\n> > 4096?\n> > \n> > I thought we fixed this on Feb 3. look at\n> > parse_expr.c line 104: it should read: \n> > if (con->typename != NULL)\n> > ! result = parser_typecast(val,\n> > con->typename, -1);\n> > else\n> > \n> > I think all funcs calling with atttypmod = 0 are allways wrong, should\n> > be -1.\n> > or a number > 0 (or 4 if atttypmod includes the VARHDRSZ don't know) \n> \n> It does include the VARHDRSZ.\n\n So the right check would be \"atttypmod > 4\" instead of \"> 0\"\n But fixing all the other places to use one and the same\n value (0 or -1) would be a good idea.\n\n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n> \n\n\nUntil later, Jan\n\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] (Jan Wieck) #\n\n",
"msg_date": "Thu, 5 Feb 1998 19:19:37 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> > It does include the VARHDRSZ.\n> \n> So the right check would be \"atttypmod > 4\" instead of \"> 0\"\n> But fixing all the other places to use one and the same\n> value (0 or -1) would be a good idea.\n> \n\nI already have. 0 is the default, but I can change it.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 13:47:34 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: atttypmod of 0"
}
] |
[
{
"msg_contents": "As part of yesterday's patch, I fixed a problem with indexing of nulls. \nThe old code did:\n\n\n mask = (finalbit << 1) - 1;\n if ((~n) & mask)\n slow=1;\n\nwhen it meant to do:\n\n mask = (1 << finalbit) - 1;\n if ((~n) & mask)\n slow=1;\n\nThis may explain some problems people were reporing.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 12:24:16 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "index with nulls"
}
] |
[
{
"msg_contents": "Can someone figure out why atttypmod for system relations is not always\nzero? Sometimes it is -1 and other times it is 1, and I can't figure\nout why.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 12:50:03 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "atttypmod is wrong in system catalogs"
}
] |
[
{
"msg_contents": "Here is a query that will show the problem:\n\nselect relname, attname, atttypmod from pg_attribute, pg_class where \nattrelid = pg_class.oid;\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 12:51:06 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "atttypmod problem"
}
] |
[
{
"msg_contents": "Now for something completely different\n\n Somewhere in the Oracle doc's I read that a trigger when\n fired is executed under the permissions of the trigger owner.\n That makes really sense for me. This way a normal user can be\n revoked from modifying a table directly, but if he modifies\n another table where he has permissions for, a trigger fired\n for that update can access the protected table. Good feature\n on logging tables, so joe-user can delete a customer but a\n trigger will log that and joe-user cannot modify the log.\n\n In PostgreSQL, triggers are currently executed under the same\n username as the one caused it to fire. So the user must have\n the proper permissions on the tables, the trigger touches,\n too.\n\n I'm not sure if changing the current database user in the\n backend for the time a trigger function is active is easy.\n Looks like all the permissions are checked against what\n GetPgUserName() returns. But I'm not sure if this is true in\n all places that a trigger can invoke.\n\n I think it would be nice to to have something like setuid\n triggers. Some triggers cause a change of the user and some\n don't. A setuid_user attribute in pg_trigger could do that\n job. If that doesn't contain NULL, the trigger will be\n invoked under the permissions of the user in that field.\n Obviously to say that a normal user can only create setuid\n triggers with his own name.\n\n Should the syntax for CREATE TRIGGER be extended here or\n should we have a completely different command for that?\n\n Any suggestions/comments?\n\n\nUntil later, Jan\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] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 5 Feb 1998 19:10:08 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Trigger and permissions"
},
{
"msg_contents": "Jan Wieck wrote:\n> \n> I think it would be nice to to have something like setuid\n> triggers. Some triggers cause a change of the user and some\n> don't. A setuid_user attribute in pg_trigger could do that\n> job. If that doesn't contain NULL, the trigger will be\n> invoked under the permissions of the user in that field.\n> Obviously to say that a normal user can only create setuid\n> triggers with his own name.\n> \n> Should the syntax for CREATE TRIGGER be extended here or\n> should we have a completely different command for that?\n> \n> Any suggestions/comments?\n\nFirst, what standard says about execution permissions of triggers, \nfunctions and procedures ?\n\nSecond, if we would decide to have both setuid and non-setuid\ntriggers (etc) then I would suggest special command to set\nsetuid \"bit\" - just to don't change syntax of both CREATE TRIGGER\nand CREATE FUNCTION. Also, in \"normal\" dbsystem, user first\ndoes CREATE VIEW and only after that does GRANT ... TO ... (it likes\nsetting setuid bit on VIEW).\n\nVadim\n",
"msg_date": "Fri, 06 Feb 1998 08:54:19 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Trigger and permissions"
}
] |
[
{
"msg_contents": "Still have some problem:\n\t\n\ttest=> select attname,atttypmod from pg_attribute where atttypmod != 0;\n\tattname |atttypmod\n\t-------------+---------\n\tgrolist | -1\n\tattdisbursion| -1\n\trelacl | -1\n\t(3 rows)\n\nIt may have to do with columns that have special column types. Any\nideas? If these are wrong, other values on that row are wrong too.\n\nThis also shows disturbing results:\n\n\tselect attname,attcacheoff from pg_attribute where attcacheoff != -1;\n\nMany have a value of zero.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 14:08:12 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "atttypmod problem"
},
{
"msg_contents": "> \n> Still have some problem:\n> \t\n> \ttest=> select attname,atttypmod from pg_attribute where atttypmod != 0;\n> \tattname |atttypmod\n> \t-------------+---------\n> \tgrolist | -1\n> \tattdisbursion| -1\n> \trelacl | -1\n> \t(3 rows)\n> \n> It may have to do with columns that have special column types. Any\n> ideas? If these are wrong, other values on that row are wrong too.\n\nI found the problem. Was a stupid mistake I made when I was adding\natttypmod.\n\n\n> \n> This also shows disturbing results:\n> \n> \tselect attname,attcacheoff from pg_attribute where attcacheoff != -1;\n> \n> Many have a value of zero.\n\nNot sure about these yet.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 14:47:00 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] atttypmod problem"
}
] |
[
{
"msg_contents": "Here are my ideas.\n\nCan you look at fmgr_isbuiltin? It is doing a binary search looking to\nsee if it is builtin. Is it wrong? Is there a better way to do this? \nSeems like a lot of cpu for this.\n\nAlso, locking seems to be taking quite a bit of cpu, especially if you\nadd up all the locking calls together. Can you check and see if there\nis a cleaner way to do this?\n \nhash_search also looks like it could be improved. This is called a lot\nby the locking code. I know because when I added real deadlock\ndetection, I had to work with this stuff.\n\n---------------------------------------------------------------------------\n\nForwarded message:\n> From maillist Thu Feb 5 15:11:14 1998\n> Subject: Re: [HACKERS] Profile of current backend\n> To: [email protected] (Mattias Kregert)\n> Date: Thu, 5 Feb 1998 15:11:14 -0500 (EST)\n> Cc: [email protected], [email protected]\n> In-Reply-To: <[email protected]> from \"Mattias Kregert\" at Feb 5, 98 08:57:21 pm\n> X-Mailer: ELM [version 2.4 PL25]\n> MIME-Version: 1.0\n> Content-Type: text/plain; charset=US-ASCII\n> Content-Transfer-Encoding: 7bit\n> Content-Length: 4140 \n> \n> Interesting. Nothing is jumping out at me. Looks like we could try to\n> clean up heapgettup() to see if there is anything in there that can be\n> speeded up.\n> \n> None of the calls looks like it should be inlined. Do you see any that\n> look good for inlining?\n> \n> \n> > \n> > Meskes, Michael wrote:\n> > >\n> > > I take it some of you have already run the backend against a profiler.\n> > > Could anyone please tell me which functions need works most?\n> > \n> > SYSTEM: Linux 2.0.29, gcc 2.7.2\n> > -------------------------------\n> > [make distclean]\n> > [CVSup'ed current source 1998-02-05]\n> > [./configure]\n> > [make]\n> > [CUSTOM_LDFLAGS not settable in Makefile.custom ...]\n> > [... setting LDFLAGS=-pg ins Makefile.global]\n> > \n> > --> fmgr.c: In function `fmgr_pl':\n> > --> fmgr.c:46: `fmgr_pl_finfo' undeclared (first use this function)\n> > [fix] cp backend/utils/fmgr.h backend/utils/fmgr/fmgr.h\n> > \n> > [make install]\n> > [Run initdb]\n> > [Regression tests]\n> > strings .. failed [ok]\n> > lseg .. failed [new operator <= on lseg]\n> > horology .. failed [daylight savings error?]\n> > triggers .. failed\n> > select_views .. failed\n> > \n> > \n> > PROFILE OF REGRESSION TEST:\n> > ---------------------------\n> > time seconds seconds calls ms/call ms/call name \n> > 35.16 7.97 7.97 mcount (profiler overhead)\n> > 5.91 9.31 1.34 9924 0.14 0.20 heapgettup\n> > 4.54 10.34 1.03 234597 0.00 0.01 hash_search\n> > 2.12 10.82 0.48 151781 0.00 0.00 SpinAcquire\n> > 2.03 11.28 0.46 46635 0.01 0.02 SearchSysCache\n> > 1.63 11.65 0.37 171345 0.00 0.00 tag_hash\n> > 1.46 11.98 0.33 20511 0.02 0.02 yylex\n> > 1.41 12.30 0.32 31311 0.01 0.03 LockAcquire\n> > 1.28 12.59 0.29 1522 0.19 0.61 yyparse\n> > 1.24 12.87 0.28 41176 0.01 0.01 fmgr_isbuiltin\n> > 1.01 13.10 0.23 201572 0.00 0.00 AllocSetAlloc\n> > 0.84 13.29 0.19 186214 0.00 0.00 OrderedElemPop\n> > 0.75 13.46 0.17 42896 0.00 0.00 nocachegetattr\n> > 0.71 13.62 0.16 29869 0.01 0.04 LockRelease\n> > 0.66 13.77 0.15 151781 0.00 0.00 SpinRelease\n> > 0.66 13.92 0.15 88766 0.00 0.01 newNode\n> > 0.66 14.07 0.15 52248 0.00 0.00 _bt_compare\n> > ... around 1850 functions follows (0.66 - 0.0%)\n> > \n> > heapgettup (called by, calling):\n> > --------------------------------\n> > index % time self children called name\n> > 0.01 0.01 96/9924 heap_markpos [448]\n> > 0.21 0.11 1566/9924 heap_restrpos [102]\n> > 1.12 0.58 8262/9924 heap_getnext [20]\n> > [14] 13.8 1.34 0.69 9924 heapgettup [14]\n> > 0.04 0.14 9924/9924 RelationGetBufferWithBuffer [148]\n> > 0.03 0.15 5642/5702 ReleaseAndReadBuffer [145]\n> > 0.10 0.00 26276/42896 nocachegetattr [158]\n> > 0.01 0.08 7111/9607 HeapTupleSatisfiesVisibility [185]\n> > 0.04 0.00 117785/126582 char16eq [339]\n> > 0.02 0.00 111941/111994 int4eq [427]\n> > 0.02 0.00 109647/112329 nameeq [426]\n> > 0.00 0.01 1770/31585 ReleaseBuffer [134]\n> > 0.01 0.00 14000/50173 PageGetMaxOffsetNumber [341]\n> > 0.01 0.00 5185/5433 chareq [537]\n> > 0.01 0.00 15566/90147 BufferGetBlock [311]\n> > 0.01 0.00 17336/240911 BufferIsValid [207]\n> > 0.00 0.00 4/1973 int2eq [575]\n> > 0.00 0.00 7412/7412 nextpage [1186]\n> > 0.00 0.00 951/1358 SetBufferCommitInfoNeedsSave [1253]\n> > --------------------------------\n> > \n> > The *complete* gprof output is here:\n> > ftp://postgresql.org/pub/incoming/regression-profile-980205.gz 156k\n> > \n> > /* m */\n> > \n> > \n> \n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 5 Feb 1998 15:26:22 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Profile of current backend (fwd)"
}
] |
[
{
"msg_contents": "Snapshot of 2 feb:\n\npostgres=> create table one (str char(52));\nCREATE\npostgres=> insert into one values ('Salut!');\nINSERT 546025 1\npostgres=> create table two (str char(52));\nCREATE\npostgres=> insert into two select * from one;\nERROR: Length of str is longer than length of target column str\n\nGood luck!\n\nCostin Oproiu\n\n",
"msg_date": "Fri, 6 Feb 1998 00:01:34 +0200 (EET)",
"msg_from": "PostgreSQL <[email protected]>",
"msg_from_op": true,
"msg_subject": "Bug with char() type"
}
] |
[
{
"msg_contents": "\tIs there a way to have the server show me the queries and the\nresponses to those queries? -d 3 is way more info than I can shift\nthrough and while -d 2 shows me the query I don't see the response to\nit...\n\n\tI've got the odbc drive connecting to 6.3 but I'm not getting the\nresponses I expect for finding the tables in a database...\n\nJulie\n\n-- \n[ Julia Anne Case ] [ Ships are safe inside the harbor, ]\n[Programmer at large] [ but is that what ships are really for. ] \n[ Admining Linux ] [ To thine own self be true. ]\n[ Windows/WindowsNT ] [ Fair is where you take your cows to be judged. ]\n",
"msg_date": "Thu, 5 Feb 1998 19:30:33 -0500",
"msg_from": "Julia Anne Case <[email protected]>",
"msg_from_op": true,
"msg_subject": "Debug info"
}
] |
[
{
"msg_contents": "I thought I might provide a little more info...\n\nI'm sending \nQselect relname, usename from pg_class, pg_user where relkind = 'r' and\nrelname !~ '^Inv[0-9]+' and int4out(usesysid) = int4out(relowner) order by\nrelname\n\nand it looks like what I get back is \nPblank\n\nJulie\n\n\n-- \n[ Julia Anne Case ] [ Ships are safe inside the harbor, ]\n[Programmer at large] [ but is that what ships are really for. ] \n[ Admining Linux ] [ To thine own self be true. ]\n[ Windows/WindowsNT ] [ Fair is where you take your cows to be judged. ]\n",
"msg_date": "Thu, 5 Feb 1998 19:41:41 -0500",
"msg_from": "Julia Anne Case <[email protected]>",
"msg_from_op": true,
"msg_subject": "More info on Debug Info"
}
] |
[
{
"msg_contents": "vac=> \\d test\n\nTable = test\n+----------------------------------+----------------------------------+-------+\n| Field | Type | Length|\n+----------------------------------+----------------------------------+-------+\n| x | int4 | 4 |\n| y | int4 | 4 |\n+----------------------------------+----------------------------------+-------+\nvac=> select count(*) from test where exists (select t1.y from test t1 where t1.y = x);\n ^\nIs this correlated subquery or not ?\n(Note, that I don't use x with t1. prefix here)\nWith current parser this works as un-correlated subquery...\nIs this Ok and I have to re-write query as\n\nvac=> select count(*) from test t2 where exists \n ^^\n(select t1.y from test t1 where t1.y = t2.x);\n ^^^\nto get correlated one ?\n\nVadim\n",
"msg_date": "Fri, 06 Feb 1998 11:35:05 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "what standard say ..."
},
{
"msg_contents": "Vadim B. Mikheev wrote:\n\n> vac=> \\d test\n>\n> Table = test\n> +----------------------------------+----------------------------------+-------+\n> | Field | Type | Length|\n> +----------------------------------+----------------------------------+-------+\n> | x | int4 | 4 |\n> | y | int4 | 4 |\n> +----------------------------------+----------------------------------+-------+\n> vac=> select count(*) from test where exists (select t1.y from test t1 where t1.y = x);\n> ^\n> Is this correlated subquery or not ?\n> (Note, that I don't use x with t1. prefix here)\n> With current parser this works as un-correlated subquery...\n> Is this Ok and I have to re-write query as\n>\n> vac=> select count(*) from test t2 where exists\n> ^^\n> (select t1.y from test t1 where t1.y = t2.x);\n> ^^^\n> to get correlated one ?\n\n>From \"The SQL Standard\", 3rd ed., Date and Darwen:\n\n\"... each unqualified column name is _implicitly_ qualified by a range variable name\ndefined (explicitly or implicitly) in the nearest applicable FROM clause.\" (the emphasis\nis from the book, not me)\n\nIt goes on to recommend reading the standard for full understanding, but it is pretty\nclear that your interpretation is correct; in the example above x is implicitly equivalent\nto t1.x.\n\n - Tom\n\n",
"msg_date": "Fri, 06 Feb 1998 07:23:22 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] what standard say ..."
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> > vac=> select count(*) from test where exists (select t1.y from test t1 where t1.y = x);\n> > ^\n> > Is this correlated subquery or not ?\n> > (Note, that I don't use x with t1. prefix here)\n> > With current parser this works as un-correlated subquery...\n> \n> >From \"The SQL Standard\", 3rd ed., Date and Darwen:\n> \n> \"... each unqualified column name is _implicitly_ qualified by a range variable name\n> defined (explicitly or implicitly) in the nearest applicable FROM clause.\" (the emphasis\n> is from the book, not me)\n> \n> It goes on to recommend reading the standard for full understanding, but it is pretty\n> clear that your interpretation is correct; in the example above x is implicitly equivalent\n> to t1.x.\n\nOk. Nice to know that we are correct here :)\n\nVadim\n",
"msg_date": "Fri, 06 Feb 1998 14:48:58 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] what standard say ..."
},
{
"msg_contents": "> \n> vac=> \\d test\n> \n> Table = test\n> +----------------------------------+----------------------------------+-------+\n> | Field | Type | Length|\n> +----------------------------------+----------------------------------+-------+\n> | x | int4 | 4 |\n> | y | int4 | 4 |\n> +----------------------------------+----------------------------------+-------+\n> vac=> select count(*) from test where exists (select t1.y from test t1 where t1.y = x);\n> ^\n> Is this correlated subquery or not ?\n> (Note, that I don't use x with t1. prefix here)\n> With current parser this works as un-correlated subquery...\n> Is this Ok and I have to re-write query as\n> \n> vac=> select count(*) from test t2 where exists \n> ^^\n> (select t1.y from test t1 where t1.y = t2.x);\n> ^^^\n> to get correlated one ?\n> \n> Vadim\n> \n> \n\nI am almost sure this is uncorrelated. If an unqualified varaiable\nappears in a subquery, it matches the closest table it can find.\n\nI am not sure about the standard, but logic would suggest this is the\nway it should work.\n\nAnd, of course, that is what the parser does.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 6 Feb 1998 11:23:57 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] what standard say ..."
}
] |
[
{
"msg_contents": "Hi Marc,\n\nI was looking at the new 'isinf.c' function in the src/backend/port directory.\nI have a question about the inclusion of cases for FP_SNAN and FP_QNAN.\n\n\tDoes the 'isinf' function mean \"not finite\" or \"is infinity\"?\n\nIf it is to mean \"not finite\", then (IMHO) then checking for FP_SNAN and \nFP_QNAN is appropiate. If it is to mean \"is infinity\", then the function \nshould only be checking (again, IHMO) for FP_NINF and FP_PINF.\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",
"msg_date": "Thu, 05 Feb 1998 23:47:38 -0500",
"msg_from": "\"Billy G. Allie\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "The new 'isinf.c' function in port."
},
{
"msg_contents": "On Thu, 5 Feb 1998, Billy G. Allie wrote:\n\n> Hi Marc,\n> \n> I was looking at the new 'isinf.c' function in the src/backend/port directory.\n> I have a question about the inclusion of cases for FP_SNAN and FP_QNAN.\n> \n> \tDoes the 'isinf' function mean \"not finite\" or \"is infinity\"?\n> \n> If it is to mean \"not finite\", then (IMHO) then checking for FP_SNAN and \n> FP_QNAN is appropiate. If it is to mean \"is infinity\", then the function \n> should only be checking (again, IHMO) for FP_NINF and FP_PINF.\n\n\tGood question...\n\n\tFrom FreeBSD's man page:\n\nThe isninf() function returns 1 if the number is ``infinity'', otherwise\n0.\n\n\tIn my defence, those functions were just pulled in from the\nvarious ports themselves...I believe you are correct about removing\nFP_SNAN/FP_QNAN...\n\n\n",
"msg_date": "Fri, 6 Feb 1998 07:59:56 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: The new 'isinf.c' function in port."
}
] |
[
{
"msg_contents": "Bruce, Tom - could you fix this ?\n\nVadim\n",
"msg_date": "Fri, 06 Feb 1998 12:02:33 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Query->hasSubLinks is always FALSE..."
}
] |
[
{
"msg_contents": "\tI spent some more time with this...\n\n\tIt appears as if after I get the Pblank I get the T for a tuple\nand then the next 2 bytes should indicate the number of fields, but they\nare both 0... If I set the number of fields to 2 and then continue, the\nrest seems to happen fine.\n\nJulie\n\n-- \n[ Julia Anne Case ] [ Ships are safe inside the harbor, ]\n[Programmer at large] [ but is that what ships are really for. ] \n[ Admining Linux ] [ To thine own self be true. ]\n[ Windows/WindowsNT ] [ Fair is where you take your cows to be judged. ]\n",
"msg_date": "Fri, 6 Feb 1998 02:27:46 -0500",
"msg_from": "Julia Anne Case <[email protected]>",
"msg_from_op": true,
"msg_subject": "[BUG?] Frontend Protocol..."
}
] |
[
{
"msg_contents": "Yes, I think to change atttypmod to default to -1 would be the right\nthing,\nsince the empty string '' really has a length of 0,\nWe could think of -1 as: We don't know how long this field will be.\n0 would mean the field has 0 bytes.\n\nAndreas\n\n> > \n> > This time it's during the parser (gdb told me). varcharin()\n> > is called with a atttypmod of 0 causing a palloc() of 0\n> > bytes. How should a VARCHAR type whithout a specified length\n> > behave? Is this type 1 character or a variable size up to\n> > 4096?\n> > \n> > I thought we fixed this on Feb 3. look at\n> > parse_expr.c line 104: it should read: \n> > if (con->typename != NULL)\n> > ! result =\n> parser_typecast(val,\n> > con->typename, -1);\n> > else\n> > \n> > I think all funcs calling with atttypmod = 0 are allways wrong,\n> should\n> > be -1.\n> > or a number > 0 (or 4 if atttypmod includes the VARHDRSZ don't know)\n> \n> > \n> > Andreas\n> > \n> > \n> \n> Yes, we did, but now I fixed varcharin, and bpcharin to test for\n> atttypmod of 0 and do the right thing, I think. If we need to make\n> the\n> default atttypmod value -1, then we can change it back. Let me know\n> if\n> the current fix does not work.\n> \n> Should I make atttypmod default to -1?\n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n> \n",
"msg_date": "Fri, 6 Feb 1998 08:52:17 +0100",
"msg_from": "Zeugswetter Andreas DBT <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> \n> Yes, I think to change atttypmod to default to -1 would be the right\n> thing,\n> since the empty string '' really has a length of 0,\n> We could think of -1 as: We don't know how long this field will be.\n> 0 would mean the field has 0 bytes.\n\nField has varlena == 4, so zero is not possible, but I could still\nchange to -1 for clarity.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 6 Feb 1998 11:47:42 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> \n> Yes, I think to change atttypmod to default to -1 would be the right\n> thing,\n> since the empty string '' really has a length of 0,\n> We could think of -1 as: We don't know how long this field will be.\n> 0 would mean the field has 0 bytes.\n\nDo people want atttypmod at -1 at this time? It will require a\ndump/reload.\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 6 Feb 1998 14:12:47 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> > Yes, I think to change atttypmod to default to -1 would be the right\n> > thing,\n> > since the empty string '' really has a length of 0,\n> > We could think of -1 as: We don't know how long this field will be.\n> > 0 would mean the field has 0 bytes.\n>\n> Do people want atttypmod at -1 at this time? It will require a\n> dump/reload.\n\nIf it fixes problems, now is the time :)\n\n",
"msg_date": "Sat, 07 Feb 1998 01:54:51 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> \n> > > Yes, I think to change atttypmod to default to -1 would be the right\n> > > thing,\n> > > since the empty string '' really has a length of 0,\n> > > We could think of -1 as: We don't know how long this field will be.\n> > > 0 would mean the field has 0 bytes.\n> >\n> > Do people want atttypmod at -1 at this time? It will require a\n> > dump/reload.\n> \n> If it fixes problems, now is the time :)\n\nWell, right now, since it is not being used from anything by char() and\nvarchar(), it is not a problem. But it will have to be changed when we\nuse it for other types.\n\nSo do we change it now?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 6 Feb 1998 21:11:47 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> > > > Yes, I think to change atttypmod to default to -1 would be the right\n> > > > thing,\n> > > > since the empty string '' really has a length of 0,\n> > > > We could think of -1 as: We don't know how long this field will be.\n> > > > 0 would mean the field has 0 bytes.\n> > >\n> > > Do people want atttypmod at -1 at this time? It will require a\n> > > dump/reload.\n> >\n> > If it fixes problems, now is the time :)\n>\n> Well, right now, since it is not being used from anything by char() and\n> varchar(), it is not a problem. But it will have to be changed when we\n> use it for other types.\n>\n> So do we change it now?\n\nIf it make things more self-consistant, now is the time :)\n\nI would consider all of this related to or a followup on the char/varchar\nwork done earlier.\n\n - Tom\n\n",
"msg_date": "Sat, 07 Feb 1998 02:18:27 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> > So do we change it now?\n> \n> If it make things more self-consistant, now is the time :)\n> \n> I would consider all of this related to or a followup on the char/varchar\n> work done earlier.\n\nOK\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 6 Feb 1998 21:31:20 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> \n> > \n> > Yes, I think to change atttypmod to default to -1 would be the right\n> > thing,\n> > since the empty string '' really has a length of 0,\n> > We could think of -1 as: We don't know how long this field will be.\n> > 0 would mean the field has 0 bytes.\n> \n> Do people want atttypmod at -1 at this time? It will require a\n> dump/reload.\n\nOK, atttypmod now defaults to -1. Initdb everyone.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sat, 7 Feb 1998 01:09:22 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Re: atttypmod of 0"
}
] |
[
{
"msg_contents": "Hi there,\n\nI am starting with the conclusions to spare time:\n - persistent connections invention stands somewhere between the fire and the\n wheel \n - unix sockets is not much compared to tcpip (at least on my Linux 2.0.30\n box).\n - parsing and executing a trivial \"select 17;\" accounts for a lot of time,\n someone wanted hints on what is critical, profiler statistics coroborated.\n\n\n Stressing postmaster with 100 connections\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n Connection type Time (seconds)\n ----------------------------------------------------\n tcpip local 15\n tcpip local with select 26\n unix sockets 15\n unix sockets with select 26\n tcpip local persistent 0\n tcpip local persistent with select 2\n unix sockets persistent 0\n unix sockets persistent with select 2\n\n\n Stressing postmaster with 2000 persistent connections\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n Connection type Time (seconds)\n -----------------------------------------------------\n tcpip local 3\n tcpip local with select 45\n unix sockets 3\n unix sockets with select 45\n\n\n\nNOTES:\n - tests were (repeatedly!) run via php3 as apache module\n - scrips attached, comments wellcome\n - \"with select\" below means a pg_Exec(\"select 17;\") \n inserted between Connect and Close\n\n\n.co.\n\nCostin Oproiu ---------------- http://www2.deuroconsult.ro/~co\nCiprian Porumbescu 10, ap.13 tel/fax: +(40)-68-183528\nBrasov 2200, ROMANIA email: [email protected]\n--------------------------------------------------------------",
"msg_date": "Fri, 6 Feb 1998 12:00:10 +0200 (EET)",
"msg_from": "Costin Oproiu <[email protected]>",
"msg_from_op": true,
"msg_subject": "Some performance issues (since everybody is testing ... :)"
},
{
"msg_contents": "> I am starting with the conclusions to spare time:\n> - persistent connections invention stands somewhere between the fire and the\n> wheel\n> - unix sockets is not much compared to tcpip (at least on my Linux 2.0.30\n> box).\n> - parsing and executing a trivial \"select 17;\" accounts for a lot of time,\n> someone wanted hints on what is critical, profiler statistics coroborated.\n>\n> Stressing postmaster with 100 connections\n\nHi. Since you are testing with multiple connections, would you like to try additional testing to make sure the new deadlock\ndetection code behaves properly, especially under heavy load/multiple connections?\n\nThe large loading cases tend not to get exercised in beta testing because the systems are not in production yet...\n\n - Tom\n\n",
"msg_date": "Fri, 06 Feb 1998 15:39:14 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Some performance issues (since everybody is testing ...\n\t:)"
},
{
"msg_contents": "On Fri, 6 Feb 1998, Thomas G. Lockhart wrote:\n\n> >\n> > Stressing postmaster with 100 connections\n> \n> Hi. Since you are testing with multiple connections, would you like to try additional testing to make sure the new deadlock\n> detection code behaves properly, especially under heavy load/multiple connections?\n> \n> The large loading cases tend not to get exercised in beta testing because the systems are not in production yet...\n> \n> - Tom\n> \nOK, the title is a little bit misleading, it is about 100 successive\nconnections and wants to show how much time a trivial \n\n\t\"select 17\"\n\ncoul take compare to a connection.\n\nHowever your idea is very good, since I am dealing with benchmarking now.\n\n.co.\n\nCostin Oproiu ---------------- http://www2.deuroconsult.ro/~co\nCiprian Porumbescu 10, ap.13 tel/fax: +(40)-68-183528\nBrasov 2200, ROMANIA email: [email protected]\n--------------------------------------------------------------\n\n",
"msg_date": "Fri, 6 Feb 1998 20:57:59 +0200 (EET)",
"msg_from": "Costin Oproiu <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Some performance issues (since everybody is testing ...\n\t:)"
}
] |
[
{
"msg_contents": "> You misunderstand me. I didn't suggest removing the S_LOCK() et al\n> macros. What I meant was that the actual assembly implementation of\n> tas() itself might be better off in a separate source file. As an\n> example, here is my current version of the locking code for the VAX,\n> in s_lock.h (bbssi is \"branch on bit set and set, interlocked\"):\n> \n> static int tas(slock_t *lock) {\n\nWith the difficult bit done, maybe I should take a look at getting it\nrunning on the old VMS uVAX-II I have sitting at home :-)\n\nAndrew\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Fri, 6 Feb 1998 11:22:57 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "VAX portt was Re: [HACKERS] configure on linux"
},
{
"msg_contents": "On Fri, 6 Feb 1998, Andrew Martin wrote:\n\n> With the difficult bit done, maybe I should take a look at getting\n> it running on the old VMS uVAX-II I have sitting at home :-)\n\nThat would be cool -- but no, the difficult bit is not done. If you\nwant to get PostgreSQL to work well under VMS, especially on a II,\nwhere you can't run anything newer than 5.5, and thus can't use the\nPOSIX interface stuff from VMS 6, you'll need to rewrite the access\nmethods at the lowest level to utilize RMS properly, or performance\nwill absolutely suck. As an example, I got well over an order of\nmagnitude of speed improvement for the IOZONE benchmark under VMS by\nwriting a couple of tens of lines of RMS-savvy interface code...\n\nOh. You weren't serious. OK.\n\nHowever, the VAX port is running nicely under NetBSD. It seems to\nbe sane -- reading through 100 Kbytes of regression.diffs for the\nSparc port and 60 Kbytes for the VAX port is hard work, especially\nsince the diffs are all either slightly differently worded error\nmessages, tiny precision differences in floating point math hardware,\nzero instead of minus zero in math results or different limits on\nmax/min representable numbers. In addition, there are a number of\ndate and time differences that I don't understand -- like why a minute\nis sometimes \"1 min\" and sometimes \"1 min 60.00 sec\"...\n\nAnyway, the patches for NetBSD/vax are sent separately, with comments.\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n\n",
"msg_date": "Fri, 6 Feb 1998 20:58:27 +0100 (CET)",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: VAX portt was Re: [HACKERS] configure on linux"
}
] |
[
{
"msg_contents": "just installed 06.02.98 snapshot on linux 2.0.32, regression runs\n(almost) ok.\n\n\\d pg_description_index\n\ncoredumps immediately\n\n------------------------------------\nHannu Krosing\n\n\n",
"msg_date": "Fri, 06 Feb 1998 17:09:15 +0200",
"msg_from": "Hannu Krosing <[email protected]>",
"msg_from_op": true,
"msg_subject": "06.02.98 snapshot core dump"
}
] |
[
{
"msg_contents": "> \n> vac=> select count(*) from test where exists (select t1.y from test t1\nwhere t1.y = x);\n>\n^\n> Is this correlated subquery or not ?\n\nNo! Informix behaves in the same manner, I think it is ok.\n\nAndreas\n",
"msg_date": "Fri, 6 Feb 1998 17:29:10 +0100",
"msg_from": "Zeugswetter Andreas DBT <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: what standard say ..."
}
] |
[
{
"msg_contents": "> Is this Ok and I have to re-write query as\n> vac=> select count(*) from test t2 where exists \n> ^^\n> (select t1.y from test t1 where t1.y = t2.x);\n> ^^^\n> to get correlated one ?\n\nYes\n",
"msg_date": "Fri, 6 Feb 1998 17:31:26 +0100",
"msg_from": "Zeugswetter Andreas DBT <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: what standard say ..."
}
] |
[
{
"msg_contents": "I guess I don't have a strong opinion on that. Since atttypmod is also\nintended for\ndecimal and others, I think the default has to be type specific anyway, \nlike 0 digits after the comma for decimal.\nI would prefer an atttypmod that does not include the VARHDRSZ for\nvarchar though.\nThen it would simply be the max length (better for humans reading the\nsystem catalog).\nThe default of zero would still be no problem I guess, since max zero\nlength is still nonsense.\n\nAndreas \n\n> > Yes, I think to change atttypmod to default to -1 would be the right\n> > thing,\n> > since the empty string '' really has a length of 0,\n> > We could think of -1 as: We don't know how long this field will be.\n> > 0 would mean the field has 0 bytes.\n> \n> Field has varlena == 4, so zero is not possible, but I could still\n> change to -1 for clarity.\n> \n",
"msg_date": "Fri, 6 Feb 1998 18:14:49 +0100",
"msg_from": "Zeugswetter Andreas DBT <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: AW: [HACKERS] Re: atttypmod of 0"
},
{
"msg_contents": "> \n> I guess I don't have a strong opinion on that. Since atttypmod is also\n> intended for\n> decimal and others, I think the default has to be type specific anyway, \n> like 0 digits after the comma for decimal.\n> I would prefer an atttypmod that does not include the VARHDRSZ for\n> varchar though.\n> Then it would simply be the max length (better for humans reading the\n> system catalog).\n> The default of zero would still be no problem I guess, since max zero\n> length is still nonsense.\n\nI indented to change it to -1 when we start using atttypmod for more\ntypes that just char() and varchar().\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 6 Feb 1998 13:01:31 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: AW: [HACKERS] Re: atttypmod of 0"
}
] |
[
{
"msg_contents": "In the old system, many tables did not have an initial attcacheoff value\nof -1, especially system tables. It would have a value of zero, and\nthis is a valid offset for the first field.\n\nI have fixed it so they are all -1. With the new code that uses\nattcacheoff more, you may need to do a dump/reload. I am not sure.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 6 Feb 1998 14:15:39 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "attcacheoff fixed, possible dump/reload"
}
] |
[
{
"msg_contents": "> > > > > > Looks like this variable size stuff will be the first new feature\n> > > > > > for 6.4 then, eh? :)\n> > > > >\n> > > > > What is the chance of having this ready by Friday? Still would\n> > > > > give us 3weeks of debugging...\n> > > > \n> > > > Well, it's OK to leave _one_ feature for v6.4. Bruce is probably \n> > > > tired of typing the feature list :)\n> > > > \n> > > > Also, we still have a few loose ends for v6.3. But, I'll bet Darren's stuff can\n> > > > go in and not affect anything if we run using the current block size.\n> > > \n> > > Great chance of having it by Friday if not sooner. I'll get on it tonite.\n> > > \n> > > Wasn't sure where the line in the sand was for \"adding features\" vs \"fixing bugs.\"\n> > \n> > I think Marc is getting softer on this issue. [Ducks head]\n> \n> \tMoi? Never...some ppl have shown over time a...dedication towards\n> fixing any bugs they introduce in a *very* timely manner *grin* A little\n> leeway, as appropriate, tends to reap large rewards for all\n> parties...specially if we can help improve disk space usage by providing\n> the ability to tailor disk block size used :)\n\nThe variable block size stuff won't be done by tonite. Something I've done in\nthe tape sorting routines is causing a core dump during the constraints regression\ntest and I haven't been able to track it down the last two evenings.\n\nSorry about that folks. I'm going to keep on it, but I wouldn't feel too\ncomfortable putting in the beta any later than now.\n\ndarrenk\n",
"msg_date": "Fri, 6 Feb 1998 15:18:13 -0500",
"msg_from": "[email protected] (Darren King)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Variable block size..."
}
] |
[
{
"msg_contents": "\nOn 6 Feb 1998, Fedor Bezrukov wrote:\n> Probably that's a silly question, but...\n> \n> There are functions 'lo_export'/'lo_import' embedded in the backend.\n> They can be called from an SQL request like it is described in the\n> User Manual. But as they are executed from the server, not from the\n> client, I get the resulting file (from lo_export) owned by the\n> 'postgres' user and located on the server machine! This is not at all\n> what you need, and more, it is a security hole, using which you can\n> peek at any data in the database and even destroy it. Probably this\n> is not the correct place for these functions (and it is even mentioned\n> in the source :) ). Probably these functions should be removed from\n> the backend or at least restricted to use by the 'postgres' user only?\n\nYou do have a point here.\n\nI think these functions are obsolete. Do we still need them? We have\nexamples on how to implement these properly from the client to server in\nthe source.\n\nWhat does everyone else think?\n\n-- \nPeter T Mount [email protected] or [email protected]\nMain Homepage: http://www.demon.co.uk/finder\nWork Homepage: http://www.maidstone.gov.uk Work EMail: [email protected]\n\n",
"msg_date": "Fri, 6 Feb 1998 21:02:41 +0000 (GMT)",
"msg_from": "Peter T Mount <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [QUESTIONS] LArge object functions in the backend"
}
] |
[
{
"msg_contents": "Marc, I am getting messages like this in my mail.\n\n---------------------------------------------------------------------------\n\nForwarded message:\n> From [email protected] Sat Feb 7 00:43:36 1998\n> Date: Sat, 7 Feb 1998 00:43:15 -0500 (EST)\n> Message-Id: <[email protected]>\n> To: [email protected]\n> From: [email protected]\n> Subject: Majordomo results: Re: [HACKERS] Bug?\n> Reply-To: [email protected]\n> \n> --\n> \n> >>>> > However: do we really want to do this anyway? If you demand that the\n> **** Command '>' not recognized.\n> >>>> > user indicate whether a given constant is integer or real, you lessen\n> **** Command '>' not recognized.\n> >>>> > the risk of doing the wrong thing with his or her data. Specifically,\n> **** Command '>' not recognized.\n> >>>> > going to floating point means giving up accuracy in representation,\n> **** Command '>' not recognized.\n> >>>> > and this may not be something we want to do without user permission.\n> **** Command '>' not recognized.\n> >>>> > \n> **** Command '>' not recognized.\n> >>>> \n> >>>> I never auto-convert integer to float unless I have to.\n> **** Command 'i' not recognized.\n> >>>> \n> >>>> -- \n> END OF COMMANDS\n> **** Help for [email protected]:\n> \n> \n> This help message is being sent to you from the Majordomo mailing list\n> management system at [email protected].\n> \n> This is version 1.94.4 of Majordomo.\n> \n> If you're familiar with mail servers, an advanced user's summary of\n> Majordomo's commands appears at the end of this message.\n> \n> Majordomo is an automated system which allows users to subscribe\n> and unsubscribe to mailing lists, and to retrieve files from list\n> archives.\n> \n> You can interact with the Majordomo software by sending it commands\n> in the body of mail messages addressed to \"[email protected]\".\n> Please do not put your commands on the subject line; Majordomo does\n> not process commands in the subject line.\n> \n> You may put multiple Majordomo commands in the same mail message.\n> Put each command on a line by itself.\n> \n> If you use a \"signature block\" at the end of your mail, Majordomo may\n> mistakenly believe each line of your message is a command; you will\n> then receive spurious error messages. To keep this from happening,\n> either put a line starting with a hyphen (\"-\") before your signature,\n> or put a line with just the word\n> \n> \tend\n> \n> on it in the same place. This will stop the Majordomo software from\n> processing your signature as bad commands.\n> \n> Here are some of the things you can do using Majordomo:\n> \n> I.\tFINDING OUT WHICH LISTS ARE ON THIS SYSTEM\n> \n> To get a list of publicly-available mailing lists on this system, put the\n> following line in the body of your mail message to [email protected]:\n> \n> \tlists\n> \n> Each line will contain the name of a mailing list and a brief description\n> of the list.\n> \n> To get more information about a particular list, use the \"info\" command,\n> supplying the name of the list. For example, if the name of the list \n> about which you wish information is \"demo-list\", you would put the line\n> \n> \tinfo demo-list\n> \n> in the body of the mail message.\n> \n> II.\tSUBSCRIBING TO A LIST\n> \n> Once you've determined that you wish to subscribe to one or more lists on\n> this system, you can send commands to Majordomo to have it add you to the\n> list, so you can begin receiving mailings.\n> \n> To receive list mail at the address from which you're sending your mail,\n> simply say \"subscribe\" followed by the list's name:\n> \n> \tsubscribe demo-list\n> \n> If for some reason you wish to have the mailings go to a different address\n> (a friend's address, a specific other system on which you have an account,\n> or an address which is more correct than the one that automatically appears \n> in the \"From:\" header on the mail you send), you would add that address to\n> the command. For instance, if you're sending a request from your work\n> account, but wish to receive \"demo-list\" mail at your personal account\n> (for which we will use \"[email protected]\" as an example), you'd put\n> the line\n> \n> \tsubscribe demo-list [email protected]\n> \n> in the mail message body.\n> \n> Based on configuration decisions made by the list owners, you may be added \n> to the mailing list automatically. You may also receive notification\n> that an authorization key is required for subscription. Another message\n> will be sent to the address to be subscribed (which may or may not be the\n> same as yours) containing the key, and directing the user to send a\n> command found in that message back to [email protected]. (This can be\n> a bit of extra hassle, but it helps keep you from being swamped in extra\n> email by someone who forged requests from your address.) You may also\n> get a message that your subscription is being forwarded to the list owner\n> for approval; some lists have waiting lists, or policies about who may\n> subscribe. If your request is forwarded for approval, the list owner\n> should contact you soon after your request.\n> \n> Upon subscribing, you should receive an introductory message, containing\n> list policies and features. Save this message for future reference; it\n> will also contain exact directions for unsubscribing. If you lose the\n> intro mail and would like another copy of the policies, send this message\n> to [email protected]:\n> \n> \tintro demo-list\n> \n> (substituting, of course, the real name of your list for \"demo-list\").\n> \n> III.\tUNSUBSCRIBING FROM MAILING LISTS\n> \n> Your original intro message contains the exact command which should be\n> used to remove your address from the list. However, in most cases, you\n> may simply send the command \"unsubscribe\" followed by the list name:\n> \n> \tunsubscribe demo-list\n> \n> (This command may fail if your provider has changed the way your\n> address is shown in your mail.)\n> \n> To remove an address other than the one from which you're sending\n> the request, give that address in the command:\n> \n> \tunsubscribe demo-list [email protected]\n> \n> In either of these cases, you can tell [email protected] to remove\n> the address in question from all lists on this server by using \"*\"\n> in place of the list name:\n> \n> \tunsubscribe *\n> \tunsubscribe * [email protected]\n> \n> IV.\tFINDING THE LISTS TO WHICH AN ADDRESS IS SUBSCRIBED\n> \n> To find the lists to which your address is subscribed, send this command\n> in the body of a mail message to [email protected]:\n> \n> \twhich\n> \n> You can look for other addresses, or parts of an address, by specifying\n> the text for which Majordomo should search. For instance, to find which\n> users at my-isp.com are subscribed to which lists, you might send the\n> command\n> \n> \twhich my-isp.com\n> \n> Note that many list owners completely or fully disable the \"which\"\n> command, considering it a privacy violation.\n> \n> V.\tFINDING OUT WHO'S SUBSCRIBED TO A LIST\n> \n> To get a list of the addresses on a particular list, you may use the\n> \"who\" command, followed by the name of the list:\n> \n> \twho demo-list\n> \n> Note that many list owners allow only a list's subscribers to use the\n> \"who\" command, or disable it completely, believing it to be a privacy\n> violation.\n> \n> VI.\tRETRIEVING FILES FROM A LIST'S ARCHIVES\n> \n> Many list owners keep archives of files associated with a list. These\n> may include:\n> - back issues of the list\n> - help files, user profiles, and other documents associated with the list\n> - daily, monthly, or yearly archives for the list\n> \n> To find out if a list has any files associated with it, use the \"index\"\n> command:\n> \n> \tindex demo-list\n> \n> If you see files in which you're interested, you may retrieve them by\n> using the \"get\" command and specifying the list name and archive filename.\n> For instance, to retrieve the files called \"profile.form\" (presumably a\n> form to fill out with your profile) and \"demo-list.9611\" (presumably the\n> messages posted to the list in November 1996), you would put the lines\n> \n> \tget demo-list profile.form\n> \tget demo-list demo-list.9611\n> \n> in your mail to [email protected].\n> \n> VII.\tGETTING MORE HELP\n> \n> To contact a human site manager, send mail to [email protected].\n> To contact the owner of a specific list, send mail to that list's\n> approval address, which is formed by adding \"-approval\" to the user-name\n> portion of the list's address. For instance, to contact the list owner\n> for [email protected], you would send mail to [email protected].\n> \n> To get another copy of this help message, send mail to [email protected]\n> with a line saying\n> \n> \thelp\n> \n> in the message body.\n> \n> VIII.\tCOMMAND SUMMARY FOR ADVANCED USERS\n> \n> In the description below items contained in []'s are optional. When\n> providing the item, do not include the []'s around it. Items in angle\n> brackets, such as <address>, are meta-symbols that should be replaced\n> by appropriate text without the angle brackets.\n> \n> It understands the following commands:\n> \n> subscribe <list> [<address>]\n> \tSubscribe yourself (or <address> if specified) to the named <list>.\n> \t\n> unsubscribe <list> [<address>]\n> \tUnsubscribe yourself (or <address> if specified) from the named <list>.\n> \t\"unsubscribe *\" will remove you (or <address>) from all lists. This\n> \t_may not_ work if you have subscribed using multiple addresses.\n> \n> get <list> <filename>\n> Get a file related to <list>.\n> \n> index <list>\n> Return an index of files you can \"get\" for <list>.\n> \n> which [<address>]\n> \tFind out which lists you (or <address> if specified) are on.\n> \n> who <list>\n> \tFind out who is on the named <list>.\n> \n> info <list>\n> \tRetrieve the general introductory information for the named <list>.\n> \n> intro <list>\n> \tRetrieve the introductory message sent to new users. Non-subscribers\n> \tmay not be able to retrieve this.\n> \n> lists\n> \tShow the lists served by this Majordomo server.\n> \n> help\n> \tRetrieve this message.\n> \n> end\n> \tStop processing commands (useful if your mailer adds a signature).\n> \n> Commands should be sent in the body of an email message to\n> \"[email protected]\". Multiple commands can be processed provided\n> each occurs on a separate line.\n> \n> Commands in the \"Subject:\" line are NOT processed.\n> \n> If you have any questions or problems, please contact\n> \"[email protected]\".\n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sat, 7 Feb 1998 00:45:46 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Majordomo results: Re: [HACKERS] Bug? (fwd)"
},
{
"msg_contents": "On Sat, 7 Feb 1998, Bruce Momjian wrote:\n\n> Marc, I am getting messages like this in my mail.\n\n\tVery strange...has anyone else seen something similar? Just want\nto narrow this down as being a 'Bruce-only' thing, or an 'Everyone'...\n\n\tBruce...does your emamil get posted to the list regardless of\ngetting this, or you just get this and nothing gets posted?\n\n> ---------------------------------------------------------------------------\n> \n> Forwarded message:\n> > From [email protected] Sat Feb 7 00:43:36 1998\n> > Date: Sat, 7 Feb 1998 00:43:15 -0500 (EST)\n> > Message-Id: <[email protected]>\n> > To: [email protected]\n> > From: [email protected]\n> > Subject: Majordomo results: Re: [HACKERS] Bug?\n> > Reply-To: [email protected]\n> > \n> > --\n> > \n> > >>>> > However: do we really want to do this anyway? If you demand that the\n> > **** Command '>' not recognized.\n> > >>>> > user indicate whether a given constant is integer or real, you lessen\n> > **** Command '>' not recognized.\n> > >>>> > the risk of doing the wrong thing with his or her data. Specifically,\n> > **** Command '>' not recognized.\n> > >>>> > going to floating point means giving up accuracy in representation,\n> > **** Command '>' not recognized.\n> > >>>> > and this may not be something we want to do without user permission.\n> > **** Command '>' not recognized.\n> > >>>> > \n> > **** Command '>' not recognized.\n> > >>>> \n> > >>>> I never auto-convert integer to float unless I have to.\n> > **** Command 'i' not recognized.\n> > >>>> \n> > >>>> -- \n> > END OF COMMANDS\n> > **** Help for [email protected]:\n> > \n> > \n> > This help message is being sent to you from the Majordomo mailing list\n> > management system at [email protected].\n> > \n> > This is version 1.94.4 of Majordomo.\n> > \n> \n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sat, 7 Feb 1998 11:28:58 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Majordomo results: Re: [HACKERS] Bug? (fwd)"
},
{
"msg_contents": "> \n> On Sat, 7 Feb 1998, Bruce Momjian wrote:\n> \n> > Marc, I am getting messages like this in my mail.\n> \n> \tVery strange...has anyone else seen something similar? Just want\n> to narrow this down as being a 'Bruce-only' thing, or an 'Everyone'...\n> \n> \tBruce...does your emamil get posted to the list regardless of\n> getting this, or you just get this and nothing gets posted?\n> \n\n\nGet's posted regardless. Maybe there is a majordomo cc in there. I\nwill check next time.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sat, 7 Feb 1998 10:35:45 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Majordomo results: Re: [HACKERS] Bug? (fwd)"
},
{
"msg_contents": "On Sat, 7 Feb 1998, The Hermit Hacker wrote:\n\n> On Sat, 7 Feb 1998, Bruce Momjian wrote:\n> \n> > Marc, I am getting messages like this in my mail.\n> \n> \tVery strange...has anyone else seen something similar? Just want\n> to narrow this down as being a 'Bruce-only' thing, or an 'Everyone'...\n> \n> \tBruce...does your emamil get posted to the list regardless of\n> getting this, or you just get this and nothing gets posted?\n\nI've been getting these Majordomo error reports back for the last\nthree postings of mine to HACKERS, and, as in Bruce's case, they also\nget posted to the list. (At least I get a copy back that way -- if\nyou saw the NetBSD/vax patches I posted about 12 hours ago or so, then\nit works: I got the Majordomo error reply for that one.)\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n\n",
"msg_date": "Sat, 7 Feb 1998 16:53:52 +0100 (CET)",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Majordomo results: Re: [HACKERS] Bug? (fwd)"
}
] |
[
{
"msg_contents": "> I have just rejiggered the int8 support so that it only depends on\n> snprintf instead of sprintf and sscanf. The interesting thing about\n> this is that we have our own version of snprintf that we use if the\n> platform's C library hasn't got snprintf/vsnprintf --- and our version\n> knows about %lld. So if you have a compiler that offers \n> working 64-bit\n> arithmetic, you don't need any help from the C library to \n> make int8 go.\n\nThis solution will work also on the Win32 port, beacuse the Cygwin libs\ndon't have vsscanf that supports 64-bit ints. Great.\n\n\t\t\tDan\n",
"msg_date": "Sat, 7 Feb 1998 12:23:07 +0100 ",
"msg_from": "Horak Daniel <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Re: Postgres for Sunos 4.1.4 "
},
{
"msg_contents": "It seems to be snagging on 'char16'. It says that\npostgres.h is included for the declaration, yet,\nwhen I scan the current postgres.h, char16 is \nno where to be found.\n\nMy CVS snapshot is from yesterday afternoon.\n\n:) Clark\n",
"msg_date": "Fri, 12 Mar 1999 00:05:23 +0000",
"msg_from": "Clark Evans <[email protected]>",
"msg_from_op": false,
"msg_subject": "tutorial won't compile in current tree."
},
{
"msg_contents": "Clark Evans <[email protected]> writes:\n> It seems to be snagging on 'char16'.\n\nIIRC, char16 was deleted in 6.4 or maybe earlier (along with the other\nspecial \"charN\" type names). \"char(16)\" is the accepted spelling now.\n\nA quick glimpse shows an awful lot of appearances of \"char16\"\nin the contrib area, as well as the tutorial and some residual\nuses in the documentation. Sigh. I guess that stuff is not\nexercised as regularly as it oughta be.\n\nDo you want to contribute a patch to make the tutorial work again?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 12 Mar 1999 10:46:22 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] tutorial won't compile in current tree. "
},
{
"msg_contents": "> Clark Evans <[email protected]> writes:\n> > It seems to be snagging on 'char16'.\n> \n> IIRC, char16 was deleted in 6.4 or maybe earlier (along with the other\n> special \"charN\" type names). \"char(16)\" is the accepted spelling now.\n> \n> A quick glimpse shows an awful lot of appearances of \"char16\"\n> in the contrib area, as well as the tutorial and some residual\n> uses in the documentation. Sigh. I guess that stuff is not\n> exercised as regularly as it oughta be.\n\nIt was...I removed all references to the char16 type in the tutorial\nand documentation when I removed the type itself.\n\nDid someone accidentally commit it back with some other patch?\n\nDarren\n",
"msg_date": "Fri, 12 Mar 1999 18:36:53 -0500",
"msg_from": "\"Stupor Genius\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] tutorial won't compile in current tree. "
},
{
"msg_contents": "I was wondering if this function should be\n'wrapped up' and put into the SPI interface.\nI did some research and think I could pull \nit off.\n\nCurrently the tutorial uses this:\n\n> bool\n> c_overpaid( TUPLE t, /* the current instance of EMP */\n> int4 limit)\n> {\n> bool isnull = false;\n> int4 salary;\n> \n> salary = (int4) GetAttributeByName(t, \"salary\", &isnull);\n> \n> if (isnull)\n> return false;\n> return salary > limit;\n> }\n\nAnd the regression test uses this:\n\n> \n> char\n> overpaid(tuple)\n> TUPLE tuple;\n> {\n> bool isnull;\n> long salary;\n> \n> salary = (long) GetAttributeByName(tuple, \"salary\", &isnull);\n> return salary > 699;\n> }\n\n\nHere is a proposal:\n\na) An explanation between a TupleTableSlot\n and a HeapTuple / TupleDesc pair. (help?)\n\n Perhaps we call the (TupleTableSlot *) a\n QueryTuple ? It's called a TUPLE in the\n two above usages, where a TUPLE == (void *)\n\nb) Six more functions (overloaded?)\n\n Existing:\n\n Datum SPI_getbinval(HeapTuple, TupleDesc, int, bool *)\n char *SPI_getvalue (HeapTuple, TupleDesc, int)\n\n Two more by index:\n\n Datum SPI_getbinval(TupleTableSlot, int, bool *)\n char *SPI_getvalue (TupleTableSlot, int)\n\n Four more by name:\n\n Datum SPI_getbinval(HeapTuple, TupleDesc, char *, bool *)\n char *SPI_getvalue (HeapTuple, TupleDesc, char *)\n Datum SPI_getbinval(TupleTableSlot, char *, bool *)\n char *SPI_getvalue (TupleTableSlot, char *)\n\n\nHmmm. Better yet, is there a way to hide \nthe difference in the SPI by using a 'smart' \nstructure?\n\nClark\n",
"msg_date": "Sat, 13 Mar 1999 01:32:31 +0000",
"msg_from": "Clark Evans <[email protected]>",
"msg_from_op": false,
"msg_subject": "GetAttributeByName"
},
{
"msg_contents": "I was looking at the tutorial code (funcs.c) a bit more\nand it seems that a bulk of the work going on for the\n\"copytext\" and \"concat16\" (soon to be defunct *smile* )\nis converting from string to datum and then back again.\n\nPerhaps if two functions were added to the SPI that \ndid this operation...\n\nDatum SPI_stringtodatum(const char *type, const char *data);\nconst char *SPI_datumtostring(Datum data);\n\nThese two functions would lookup the type in the \ntype table, find the input/output functions for\nthe datatype and make the call.\n\nThoughts?\n\nClark\n",
"msg_date": "Sat, 13 Mar 1999 02:20:01 +0000",
"msg_from": "Clark Evans <[email protected]>",
"msg_from_op": false,
"msg_subject": "SPI_stringtodatum SPI_datumtostring ?"
},
{
"msg_contents": "[Charset iso-8859-1 unsupported, filtering to ASCII...]\n> > Clark Evans <[email protected]> writes:\n> > > It seems to be snagging on 'char16'.\n> > \n> > IIRC, char16 was deleted in 6.4 or maybe earlier (along with the other\n> > special \"charN\" type names). \"char(16)\" is the accepted spelling now.\n> > \n> > A quick glimpse shows an awful lot of appearances of \"char16\"\n> > in the contrib area, as well as the tutorial and some residual\n> > uses in the documentation. Sigh. I guess that stuff is not\n> > exercised as regularly as it oughta be.\n> \n> It was...I removed all references to the char16 type in the tutorial\n> and documentation when I removed the type itself.\n> \n> Did someone accidentally commit it back with some other patch?\n\nIt is the tutorual/func*.* files.\n\n-- \n Bruce Momjian | http://www.op.net/~candle\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, 12 Mar 1999 21:25:42 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] tutorial won't compile in current tree."
},
{
"msg_contents": "Bruce Momjian wrote:\n> It is the tutorual/func*.* files.\n\nYes.\n",
"msg_date": "Sat, 13 Mar 1999 03:03:48 +0000",
"msg_from": "Clark Evans <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] tutorial won't compile in current tree."
},
{
"msg_contents": "Thoughts:\n\na) In funcs.c, to get it to compile, simply remove\nthe concat16 function and replace TUPLE with TupleTableSlot .\n\nb) In funcs.c it does not make sense to fix concat16, it\nshould just be removed. \n\nFirst, the fix would make it almost identical to 'text', \nonly you truncate and pad extra spaces at the end, thus \nthis extra duplication does little for the tutorial. \nI think the tutorial should be modified to use another \nfixed length data type, mabye a square? This leads to:\n\nSecond, from a procedure/trigger builder's perspective,\na SPI_stringtodatum and SPI_datumtostring make more \nsence than having the programmer worry about the\ninternal representation of the data type, this is what\nthe conversion functions are for anyway.... thus, I \neven see the 'copytext' code being replaced with\nsomething seperated from the internal structure of the\ndatabase executor with these conversion functions.\n\nc) In funcs.source I could not get this function to compile..\n\n| clark=> \n| clark=> CREATE FUNCTION clean_EMP () RETURNS int4 \n| clark-> AS 'DELETE FROM EMP WHERE EMP.salary <= 0\\; \n| clark'> SELECT 1 AS ignore_this' \n| clark-> LANGUAGE 'sql'; \n| ERROR: parser: parse error at or near \"\"\n\nd) I had a weird problem with the columns not showing up..\n\n> [clark@monster clark]$ psql\n> Welcome to the POSTGRESQL interactive sql monitor:\n> Please read the file COPYRIGHT for copyright terms of POSTGRESQL\n> \n> type \\? for help on slash commands\n> type \\q to quit\n> type \\g or terminate with semicolon to execute query\n> You are currently connected to the database: clark\n> \n> clark=> CREATE TABLE EMP ( \n> clark-> name text, \n> clark-> salary int4, \n> clark-> age int4, \n> clark-> dept char(16) \n> clark-> ); \n> CREATE\n> clark=> \n> clark=> INSERT INTO EMP VALUES ('Sam', 1200, 16, 'toy'); \n> INSERT 182188 1\n> clark=> INSERT INTO EMP VALUES ('Claire', 5000, 32, 'shoe'); \n> INSERT 182189 1\n> clark=> INSERT INTO EMP VALUES ('Andy', -1000, 2, 'candy'); \n> INSERT 182190 1\n> clark=> INSERT INTO EMP VALUES ('Bill', 4200, 36, 'shoe'); \n> INSERT 182191 1\n> clark=> INSERT INTO EMP VALUES ('Ginger', 4800, 30, 'candy'); \n> INSERT 182192 1\n> clark=> \\d emp\n> \n> Table = emp\n> +----------------------------------+----------------------------------+-------+\n> | Field | Type | Length|\n> +----------------------------------+----------------------------------+-------+\n> | name | text | var |\n> | dept | char() | 16 |\n> +----------------------------------+----------------------------------+-------+\n> clark=> select * from emp;\n> name |salary|age|dept \n> ------+------+---+----------------\n> Sam | 1200| 16|toy \n> Claire| 5000| 32|shoe \n> Andy | -1000| 2|candy \n> Bill | 4200| 36|shoe \n> Ginger| 4800| 30|candy \n> (5 rows)\n> \n\n\nThomas,\n\nIf you still want a patch file, I can work on it Sunday.\n\n:) Clark\n",
"msg_date": "Sat, 13 Mar 1999 06:42:34 +0000",
"msg_from": "Clark Evans <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] tutorial won't compile in current tree."
},
{
"msg_contents": "Can you send me a patch? It would be easier, and less error-prone. Thanks.\n\n\n> Thoughts:\n> \n> a) In funcs.c, to get it to compile, simply remove\n> the concat16 function and replace TUPLE with TupleTableSlot .\n> \n> b) In funcs.c it does not make sense to fix concat16, it\n> should just be removed. \n> \n> First, the fix would make it almost identical to 'text', \n> only you truncate and pad extra spaces at the end, thus \n> this extra duplication does little for the tutorial. \n> I think the tutorial should be modified to use another \n> fixed length data type, mabye a square? This leads to:\n> \n> Second, from a procedure/trigger builder's perspective,\n> a SPI_stringtodatum and SPI_datumtostring make more \n> sence than having the programmer worry about the\n> internal representation of the data type, this is what\n> the conversion functions are for anyway.... thus, I \n> even see the 'copytext' code being replaced with\n> something seperated from the internal structure of the\n> database executor with these conversion functions.\n> \n> c) In funcs.source I could not get this function to compile..\n> \n> | clark=> \n> | clark=> CREATE FUNCTION clean_EMP () RETURNS int4 \n> | clark-> AS 'DELETE FROM EMP WHERE EMP.salary <= 0\\; \n> | clark'> SELECT 1 AS ignore_this' \n> | clark-> LANGUAGE 'sql'; \n> | ERROR: parser: parse error at or near \"\"\n> \n> d) I had a weird problem with the columns not showing up..\n> \n> > [clark@monster clark]$ psql\n> > Welcome to the POSTGRESQL interactive sql monitor:\n> > Please read the file COPYRIGHT for copyright terms of POSTGRESQL\n> > \n> > type \\? for help on slash commands\n> > type \\q to quit\n> > type \\g or terminate with semicolon to execute query\n> > You are currently connected to the database: clark\n> > \n> > clark=> CREATE TABLE EMP ( \n> > clark-> name text, \n> > clark-> salary int4, \n> > clark-> age int4, \n> > clark-> dept char(16) \n> > clark-> ); \n> > CREATE\n> > clark=> \n> > clark=> INSERT INTO EMP VALUES ('Sam', 1200, 16, 'toy'); \n> > INSERT 182188 1\n> > clark=> INSERT INTO EMP VALUES ('Claire', 5000, 32, 'shoe'); \n> > INSERT 182189 1\n> > clark=> INSERT INTO EMP VALUES ('Andy', -1000, 2, 'candy'); \n> > INSERT 182190 1\n> > clark=> INSERT INTO EMP VALUES ('Bill', 4200, 36, 'shoe'); \n> > INSERT 182191 1\n> > clark=> INSERT INTO EMP VALUES ('Ginger', 4800, 30, 'candy'); \n> > INSERT 182192 1\n> > clark=> \\d emp\n> > \n> > Table = emp\n> > +----------------------------------+----------------------------------+-------+\n> > | Field | Type | Length|\n> > +----------------------------------+----------------------------------+-------+\n> > | name | text | var |\n> > | dept | char() | 16 |\n> > +----------------------------------+----------------------------------+-------+\n> > clark=> select * from emp;\n> > name |salary|age|dept \n> > ------+------+---+----------------\n> > Sam | 1200| 16|toy \n> > Claire| 5000| 32|shoe \n> > Andy | -1000| 2|candy \n> > Bill | 4200| 36|shoe \n> > Ginger| 4800| 30|candy \n> > (5 rows)\n> > \n> \n> \n> Thomas,\n> \n> If you still want a patch file, I can work on it Sunday.\n> \n> :) Clark\n> \n> \n\n\n-- \n Bruce Momjian | http://www.op.net/~candle\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, 13 Mar 1999 21:59:12 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] tutorial won't compile in current tree."
}
] |
[
{
"msg_contents": "On Sat, 7 Feb 1998, Magnus Hagander wrote:\n\n> Possible idea: SOmebody subscribed majordomo to the lists. That has\n> happened on one of the lists I run..\n\n\tFound it, thanks..\n\n\n> \n> //Magnus\n> \n> > -----Original Message-----\n> > From: Tom I Helbekkmo [mailto:[email protected]] \n> > Sent: den 7 februari 1998 16:54\n> > To: The Hermit Hacker\n> > Cc: Bruce Momjian; PostgreSQL-development\n> > Subject: Re: Majordomo results: Re: [HACKERS] Bug? (fwd)\n> > \n> > \n> > On Sat, 7 Feb 1998, The Hermit Hacker wrote:\n> > \n> > > On Sat, 7 Feb 1998, Bruce Momjian wrote:\n> > > \n> > > > Marc, I am getting messages like this in my mail.\n> > > \n> > > \tVery strange...has anyone else seen something similar? Just\n> want\n> > > to narrow this down as being a 'Bruce-only' thing, or an\n> 'Everyone'...\n> > > \n> > > \tBruce...does your emamil get posted to the list regardless of\n> > > getting this, or you just get this and nothing gets posted?\n> > \n> > I've been getting these Majordomo error reports back for the last\n> > three postings of mine to HACKERS, and, as in Bruce's case, they also\n> > get posted to the list. (At least I get a copy back that way -- if\n> > you saw the NetBSD/vax patches I posted about 12 hours ago or so, then\n> > it works: I got the Majordomo error reply for that one.)\n> > \n> > -tih\n> > -- \n> > Popularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n> > \n> > \n> \n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sat, 7 Feb 1998 14:21:46 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: FW: Majordomo results: "
}
] |
[
{
"msg_contents": "On Sat, 7 Feb 1998, Stan Brown wrote:\n\n> >\n> >\n> >\tNo probs...I've had worse hit me. My filters filtering something\n> >to a rarely viewed folder, then going back weeks later when it was a\n> >timely response needed :)\n> >\n> \n> \tWouldn't it e nice if we could go back to the days where spam filters\n> \twer not needed.\n> \n> \tBTW how do I create a user databse in 6.3? createdb stan while loged in\n> \tas stan doesn't seem to work.\n\n\tIs 'stan' the pg_superuser? If not, you have to be the\npg_superuser first...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sat, 7 Feb 1998 17:32:10 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [QUESTIONS] Please can someone help. pg_dump won't work!"
}
] |
[
{
"msg_contents": "We need the pg_user/password question in initdb. This is a know bug. \nOnly the postgres user can read pg_user currently.\n\nForwarded message:\n> From [email protected] Sat Feb 7 16:47:03 1998\n> Message-Id: <[email protected]>\n> Subject: [QUESTIONS] 6.3 question\n> To: [email protected] (Postgress Questions Mailing List)\n> Date: Sat, 7 Feb 1998 16:32:33 -0500 (EST)\n> From: \"Stan Brown\" <[email protected]>\n> X-Mailer: ELM [version 2.4 PL24]\n> Mime-Version: 1.0\n> Content-Type: text/plain; charset=US-ASCII\n> Content-Transfer-Encoding: 7bit\n> Sender: [email protected]\n> Precedence: bulk\n> \n> \tI have got 6.3 compiled on a deifferent machine than my main postgres\n> \tserver. Now I am trying to set ub the database. When I do a \\dt. I get\n> \tthe folwoing error message:\n> \n> \tstan=> \\dt\n> \tERROR: pg_user: Permission denied.\n> \n> \tWhat don't I have set up corectly yet? \n> \n> \tThis was as a normal user after having done a:\n> \n> \tCREATE DATABASE sta;\n> \tCREATE USER stan;\n> \n> \tas the postgres super user.\n> \n> -- \n> Stan Brown [email protected] 770-996-6955\n> Factory Automation Systems\n> Atlanta Ga.\n> -- \n> Look, look, see Windows 95. Buy, lemmings, buy! \n> Pay no attention to that cliff ahead... Henry Spencer\n> (c) 1998 Stan Brown. Redistribution via the Microsoft Network is prohibited.\n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sat, 7 Feb 1998 16:51:41 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "[QUESTIONS] 6.3 question (fwd)"
}
] |
[
{
"msg_contents": "50 sounds okay and sufficient.\n\nBut I do not get that much. I get about 50 of my transactions (2updates\n+ 1insert + some selects) per minute!\n\nI do not select the big table before inserting And yes, I do use\nindices. But for one key there are only two different values in the big\ntable.\n\nAlso, please take note that I'm still using 6.2.1 for this test. I will\nswitch to 6.3. as soon as the ODBC driver is working again.\n\nMichael\n--\nDr. Michael Meskes, Projekt-Manager | topystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Use Debian GNU/Linux! | Tel: (+49) 2405/4670-44\n\n> ----------\n> From: \tThomas G. Lockhart[SMTP:[email protected]]\n> Sent: \tFreitag, 6. Februar 1998 16:09\n> To: \tMichael Meskes\n> Cc: \tBruce Momjian; PostgreSQL Hacker\n> Subject: \tRe: [HACKERS] Profile of current backend\n> \n> I (and others) had done some benchmarking on simple inserts (6 months\n> ago?) and\n> had concluded that the speed was similar to other commercial systems\n> (I was\n> comparing against Ingres). I recall getting ~50TPS.\n> \n> This was all before Bruce did his work on startup and runtime speeds.\n> You\n> really think your performance is that far off? You are doing selects\n> on the big\n> table before inserting? Do you have indices set up?? Our results were\n> for\n> inserts on a heap table, which has the least overhead...\n> \n> - Tom\n> \n",
"msg_date": "Sun, 8 Feb 1998 11:16:17 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Profile of current backend"
}
] |
[
{
"msg_contents": "No, and that explains it all. Silly me. Argh!\n\nI used the standard start script that comes with Debian (in fact the\n6.2.1 version I use is the Debian package) and it doesn't use -F. And\nwith more disk access we have no chance.\n\nI'll check that as soon as I find the time.\n\nMichael\n--\nDr. Michael Meskes, Projekt-Manager | topystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Use Debian GNU/Linux! | Tel: (+49) 2405/4670-44\n\n> ----------\n> From: \tThe Hermit Hacker[SMTP:[email protected]]\n> Sent: \tFreitag, 6. Februar 1998 16:20\n> To: \tThomas G. Lockhart\n> Cc: \tMichael Meskes; Bruce Momjian; PostgreSQL Hacker\n> Subject: \tRe: [HACKERS] Profile of current backend\n> \n> \tJust curious, but do you have -F set to disable fsync()? We\n> really really should disable that by default :(\n> \n> \n> \n",
"msg_date": "Sun, 8 Feb 1998 11:16:18 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Profile of current backend"
}
] |
[
{
"msg_contents": "BEGIN/END of what? Transactions? Since I used ODBC auto commit is\nusually on. But all writing is done in one tranaction, i.e. we switch\nauto-commit off and handle it ourselves.\n\nMichael\n--\nDr. Michael Meskes, Projekt-Manager | topystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Use Debian GNU/Linux! | Tel: (+49) 2405/4670-44\n\n> ----------\n> From: \tVadim B. Mikheev[SMTP:[email protected]]\n> Sent: \tFreitag, 6. Februar 1998 17:00\n> To: \tThe Hermit Hacker\n> Cc: \tThomas G. Lockhart; Michael Meskes; Bruce Momjian; PostgreSQL\n> Hacker\n> Subject: \tRe: [HACKERS] Profile of current backend\n> \n> And did you use BEGIN/END ?\n> Auto-commit is ON in postgres...\n> \n> > \n> > Just curious, but do you have -F set to disable fsync()? We\n> > really really should disable that by default :(\n> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n> Oh no!\n> \n> Vadim\n> \n",
"msg_date": "Sun, 8 Feb 1998 11:16:20 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Profile of current backend"
}
] |
[
{
"msg_contents": "That's exactly what I meant with my first mail. IMO this is the correct\nway to handle it.\n\nMichael\n--\nDr. Michael Meskes, Projekt-Manager | topystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Use Debian GNU/Linux! | Tel: (+49) 2405/4670-44\n\n> ----------\n> From: \tThomas G. Lockhart[SMTP:[email protected]]\n> Sent: \tFreitag, 6. Februar 1998 16:53\n> To: \tBruce Momjian\n> Cc: \tMichael Meskes; [email protected]\n> Subject: \tRe: [HACKERS] Bug?\n> \n> > I think it converts it to an integer, and then by the time it tries\n> to\n> > convert it, it has already chopped off the top of the number. The\n> only\n> > fix for this would be to read all integers in as 64-bit integers,\n> then\n> > do the conversion, but that could be a performance problem.\n> \n> Well, the other possibility is to try converting to float8 only if the\n> int4\n> conversion fails. If both fail, then throw an elog(ERROR). I have\n> patches for\n> this...\n> \n",
"msg_date": "Sun, 8 Feb 1998 11:16:22 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Bug?"
}
] |
[
{
"msg_contents": "We're talking about quite selects/updates/inserts etc. If we say we have\n50TPS that makes one transaction every 20 milliseconds. So one more for\nparsing makes up for 5% more computation power. I doubt inlining offers\nmuch more speedup.\n\nMichael\n--\nDr. Michael Meskes, Projekt-Manager | topystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Use Debian GNU/Linux! | Tel: (+49) 2405/4670-44\n\n> ----------\n> From: \tTom I Helbekkmo[SMTP:[email protected]]\n> Sent: \tFreitag, 6. Februar 1998 17:55\n> To: \tMichael Meskes\n> Cc: \tBruce Momjian; PostgreSQL Hacker\n> Subject: \tRe: [HACKERS] Bug?\n> \n> Bruce Momjian said:\n> \n> > The only fix for this would be to read all integers in as 64-bit\n> > integers, then do the conversion, but that could be a performance\n> > problem.\n> \n> Michael Meskes answered:\n> \n> > I agree. And performance is important. I think explicit type\n> > conversion is what we should do. Or is it asked for too much if the\n> > user has to add a ::float8 to the number?\n> \n> Am I being dense here? Can there really be a significant performance\n> hit in the parsing of a query? Let's say that it takes a millisecond\n> extra to do the right thing with a number. Does it matter? How many\n> queries per second can we expect to process anyway?\n> \n> -tih\n> -- \n> Popularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n> \n",
"msg_date": "Sun, 8 Feb 1998 11:16:27 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Bug?"
}
] |
[
{
"msg_contents": "I like this code. I really do. I don't think accuracy is a problem. It\nwill work as it does right now if the number is a long. Only if it is\nout of range it will go to float instead of giving back an error\nmessage. Where could that be a problem?\n\nMichael\n--\nDr. Michael Meskes, Projekt-Manager | topystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Use Debian GNU/Linux! | Tel: (+49) 2405/4670-44\n\n> ----------\n> From: \tTom I Helbekkmo[SMTP:[email protected]]\n> Sent: \tSamstag, 7. Februar 1998 04:53\n> To: \tBruce Momjian\n> Cc: \[email protected]; [email protected]\n> Subject: \tRe: [HACKERS] Bug?\n> \n> Oh, sorry -- I wasn't being clear. Of course you don't, since we\n> don't even have standard 64-bit integers. My point was that I\n> couldn't see was how special handling of constants during the parsing\n> of the query string could have significant performance impact, even if\n> you did read them as 64-bit integers, which would mean adding bignum\n> code to PostgreSQL. In other words, performance isn't the argument to\n> be used against doing the right thing during parsing.\n> \n> As for implementation, I was thinking more along the lines of:\n> \n> {integer}\t\t{\n> \t\t\t\t\tchar* endptr;\n> \n> \t\t\t\t\terrno = 0;\n> \t\t\t\t\tyylval.ival = strtol((char\n> *)yytext,&endptr,10);\n> \t\t\t\t\tif (*endptr != '\\0' || errno ==\n> ERANGE) {\n> \t\t\t\t\t\terrno = 0;\n> \t\t\t\t\t\tyylval.dval =\n> strtod(((char *)yytext),&endptr);\n> \t\t\t\t\t\tif (*endptr != '\\0' ||\n> errno == ERANGE) {\n> \t\t\t\t\t\t\telog(ERROR,\"Bad\n> integer input '%s'\",yytext);\n> \t\t\t\t\t\t\treturn (ICONST);\n> \t\t\t\t\t\t}\n> \n> CheckFloat8Val(yylval.dval);\n> \t\t\t\t\t\treturn (FCONST);\n> \t\t\t\t\t}\n> \t\t\t\t\treturn (ICONST);\n> \t\t\t\t}\n> \n> However: do we really want to do this anyway? If you demand that the\n> user indicate whether a given constant is integer or real, you lessen\n> the risk of doing the wrong thing with his or her data. Specifically,\n> going to floating point means giving up accuracy in representation,\n> and this may not be something we want to do without user permission.\n> \n> -tih\n> -- \n> Popularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n> \n",
"msg_date": "Sun, 8 Feb 1998 11:16:33 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Bug?"
},
{
"msg_contents": "On Sun, 8 Feb 1998, Michael Meskes wrote:\n\n> Only if it is out of range it will go to float instead of giving\n> back an error message. Where could that be a problem?\n\nMy worry about that is for the (unlikely, but possible) case where a\nuser gives a large number as a constant in a query, believing it to be\nan integer. If we signal an error, the user will know that the number\nwas out of range. If not, we may end up doing calculations on floats\nthat the user wanted done on integers. This may mean losing accuracy,\nwhich is something you don't want to do behind the user's back.\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n\n",
"msg_date": "Sun, 8 Feb 1998 13:01:18 +0100 (CET)",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] Bug?"
},
{
"msg_contents": "I wrote:\n\n> My worry about that is for the (unlikely, but possible) case where a\n> user gives a large number as a constant in a query, believing it to be\n> an integer. If we signal an error, the user will know that the number\n> was out of range. If not, we may end up doing calculations on floats\n> that the user wanted done on integers. This may mean losing accuracy,\n> which is something you don't want to do behind the user's back.\n\nSomething just struck me... How's this for a workaround? It lets you\nenter large floats that happen to be integral without any explicit\nindication of the fact that they are floats, which was the original\nintent of this whole thread, while letting the user know that we're\ndoing it. In effect, we're saying \"sure, you can do this on the fly,\nand we'll do the right thing, but please be explicit in stored code\".\n\nHere's the modified code -- reindented for email purposes:\n\n{integer} {\n char* endptr;\n\n errno = 0;\n yylval.ival = strtol((char *)yytext,&endptr,10);\n if (*endptr != '\\0' || errno == ERANGE) {\n errno = 0;\n yylval.dval = strtod(((char *)yytext),&endptr);\n if (*endptr != '\\0' || errno == ERANGE) {\n elog(ERROR,\"Bad integer input '%s'\",yytext);\n return (ICONST);\n }\n elog(NOTICE,\"Out of range integer '%s' promoted to float\",yytext);\n CheckFloat8Val(yylval.dval);\n return (FCONST);\n }\n return (ICONST);\n}\n\nI don't know how the NOTICE is treated in all situations, though. If\nthe user doesn't see it, there isn't much point in having it. What\nhappens for the various interface methods? Will it be shown to a user\naccessing a database from MS Excel using ODBC, for instance?\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n\n",
"msg_date": "Sun, 8 Feb 1998 15:27:36 +0100 (CET)",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] Bug?"
},
{
"msg_contents": "> > My worry about that is for the (unlikely, but possible) case where a\n> > user gives a large number as a constant in a query, believing it to be\n> > an integer. If we signal an error, the user will know that the number\n> > was out of range. If not, we may end up doing calculations on floats\n> > that the user wanted done on integers. This may mean losing accuracy,\n> > which is something you don't want to do behind the user's back.\n>\n> Something just struck me... How's this for a workaround? It lets you\n> enter large floats that happen to be integral without any explicit\n> indication of the fact that they are floats, which was the original\n> intent of this whole thread, while letting the user know that we're\n> doing it. In effect, we're saying \"sure, you can do this on the fly,\n> and we'll do the right thing, but please be explicit in stored code\".\n> I don't know how the NOTICE is treated in all situations, though. If\n> the user doesn't see it, there isn't much point in having it. What\n> happens for the various interface methods? Will it be shown to a user\n> accessing a database from MS Excel using ODBC, for instance?\n\nPer Tom H's suggestion; what do you think Bruce?\n\npostgres=> select 100000000000;\nNOTICE: Integer input '100000000000' is out of range; promoted to float\n ?column?\n------------\n100000000000\n(1 row)\n\nThat would alleviate the \"hidden\" side effects, but still come closer to\ndoing something helpful...\n\n - Tom\n\n",
"msg_date": "Mon, 09 Feb 1998 05:45:05 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
},
{
"msg_contents": "> Per Tom H's suggestion; what do you think Bruce?\n> \n> postgres=> select 100000000000;\n> NOTICE: Integer input '100000000000' is out of range; promoted to float\n> ?column?\n> ------------\n> 100000000000\n> (1 row)\n> \n> That would alleviate the \"hidden\" side effects, but still come closer to\n> doing something helpful...\n\nWith the NOTICE, I like it.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 9 Feb 1998 10:49:27 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
},
{
"msg_contents": "Bruce Momjian wrote:\n\n> > Per Tom H's suggestion; what do you think Bruce?\n> >\n> > postgres=> select 100000000000;\n> > NOTICE: Integer input '100000000000' is out of range; promoted to float\n> > ?column?\n> > ------------\n> > 100000000000\n> > (1 row)\n> >\n> > That would alleviate the \"hidden\" side effects, but still come closer to\n> > doing something helpful...\n>\n> With the NOTICE, I like it.\n\nWill be done for v6.3 then. I'm finishing up on patches to gram.y to help\nwith CREATE FUNCTION, etc., wrt type names. Will commit the scan.l changes at\nthat time...\n\n - Tom\n\n",
"msg_date": "Mon, 09 Feb 1998 16:07:28 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug?"
}
] |
[
{
"msg_contents": "subscribe\n",
"msg_date": "Sun, 8 Feb 1998 21:58:46 EST",
"msg_from": "[email protected]",
"msg_from_op": true,
"msg_subject": "subscribe pgsql-hackers [email protected]"
}
] |
[
{
"msg_contents": "The compiler I'm using breaks while compiling aclchk.c, in particular, when \nprocessing the heap_getattr macro. Yes, I know that it may compile with gcc \n(may, because I have not tried it yet), but I would rather not have to rely on \ngcc. Would it be possible to include the non-macro version as a static \nfunction in the appropiate header file? This would allow the compiler to \nin-line the function if it would provide a speed increase.\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",
"msg_date": "Mon, 09 Feb 1998 01:57:54 -0500",
"msg_from": "\"Billy G. Allie\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "The use of macros for heap_getattr."
},
{
"msg_contents": "> \n> The compiler I'm using breaks while compiling aclchk.c, in particular, when \n> processing the heap_getattr macro. Yes, I know that it may compile with gcc \n> (may, because I have not tried it yet), but I would rather not have to rely on \n> gcc. Would it be possible to include the non-macro version as a static \n> function in the appropiate header file? This would allow the compiler to \n> in-line the function if it would provide a speed increase.\n\nSend me the error, and try and determine what makes it break. I will\ntry to change it so your compiler will work. I used no gcc-specific\nstuff in there that I know of. Also, try removing stuff until it\ncompiles so I can know what it exactly does not like.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 9 Feb 1998 10:50:56 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] The use of macros for heap_getattr."
}
] |
[
{
"msg_contents": "In my previous post, I meant to say:\n\nWould it be possible to include the non-macro version as a static function in \nthe appropiate header file, selectable by an appropiate defined tag?\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 | \n\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 | \n\n\n",
"msg_date": "Mon, 09 Feb 1998 02:05:30 -0500",
"msg_from": "\"Billy G. Allie\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "The use of macros for heap_getattr. (clarification). "
}
] |
[
{
"msg_contents": "I think the real problem right now is the lack of a decimal type.\nAn integer out of range could be converted to a decimal without \nloosing precision. I think this would be the right thing to do.\nThis is how Informix and DB/2 do it (and Oracle I guess).\nI personally think postgresql should not force the user to use \ntypecasts where it is not absolutely necessary.\n\nAndreas\n",
"msg_date": "Mon, 9 Feb 1998 09:59:46 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: integer to float8 propagation (missing decimal)"
}
] |
[
{
"msg_contents": "I have noticed with delight, that the defined(aix) directives have gone\naway (since it is not allways defined),\nnow there is another one in s_lock.h __AIX with two underscores,\nwich also is not defined by gcc.\nPlease use the singel underscore _AIX since it seems to be defined by\nall compilers.\n\nAndreas\n",
"msg_date": "Mon, 9 Feb 1998 14:14:20 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "AIX port s_lock.h __AIX --> _AIX"
},
{
"msg_contents": "On Mon, 9 Feb 1998, Zeugswetter Andreas SARZ wrote:\n\n> I have noticed with delight, that the defined(aix) directives have gone\n> away (since it is not allways defined),\n> now there is another one in s_lock.h __AIX with two underscores,\n> wich also is not defined by gcc.\n> Please use the singel underscore _AIX since it seems to be defined by\n> all compilers.\n\n\tDarren? can you confirm before we make any changes?\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 9 Feb 1998 23:00:24 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] AIX port s_lock.h __AIX --> _AIX"
}
] |
[
{
"msg_contents": "subscribe\n\n",
"msg_date": "Mon, 09 Feb 1998 15:09:40 +0100",
"msg_from": "Brano Vislocky <[email protected]>",
"msg_from_op": true,
"msg_subject": "(no subject)"
}
] |
[
{
"msg_contents": "Bruce Momjian wrote:\n> \n> You know, after all our lengthy discussions, this SubLink thing looks\n> pretty clean to me.\n> \n> How are subselects going? Are you having to write a lot of code to get\n> it working? I would say after all our discussions and thinking, it took\n\nMmm ... =~ 1000 lines of code :)\n\n> me about two or three days to do the parser and rewrite for sublinks.\n> How many coding days is the rest taking? I am curious.\n\nAll was near ready Feb 6, but ... I got problems from canonification\nroutines in optimizer: 'where A AND B OR C ' becomes 'where (A OR C) AND (B OR C)'\nand if C is subselect then ... C has to be processed twice (sometime) ...\nI remember your question about handling 'A AND (B OR SubSelect)' - unfortunately,\nI didn't see problems here (actually, there is no problem with this\nconjunctive normal form :), but there are troubles in common case.\n\nIdeas ?\n\nFor the moment I could leave this un-optimized, with possibility\nof double subselect processing...\n\nVadim\n",
"msg_date": "Mon, 09 Feb 1998 23:28:13 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [PATCHES] Re: [HACKERS] Query->hasSubLinks is always FALSE..."
},
{
"msg_contents": "> \n> Bruce Momjian wrote:\n> > \n> > You know, after all our lengthy discussions, this SubLink thing looks\n> > pretty clean to me.\n> > \n> > How are subselects going? Are you having to write a lot of code to get\n> > it working? I would say after all our discussions and thinking, it took\n> \n> Mmm ... =~ 1000 lines of code :)\n> \n> > me about two or three days to do the parser and rewrite for sublinks.\n> > How many coding days is the rest taking? I am curious.\n> \n> All was near ready Feb 6, but ... I got problems from canonification\n> routines in optimizer: 'where A AND B OR C ' becomes 'where (A OR C) AND (B OR C)'\n> and if C is subselect then ... C has to be processed twice (sometime) ...\n> I remember your question about handling 'A AND (B OR SubSelect)' - unfortunately,\n> I didn't see problems here (actually, there is no problem with this\n> conjunctive normal form :), but there are troubles in common case.\n> \n> Ideas ?\n> \n> For the moment I could leave this un-optimized, with possibility\n> of double subselect processing...\n\nI would leave it for later. We already have an optimizer explosion\nproblem when we have lots of OR's so at some point we may have to\nrevisit the cnf-ify process anyway. Commerical Ingres has or had the\nsame problem.\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 9 Feb 1998 12:33:59 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Re: [HACKERS] Query->hasSubLinks is always FALSE..."
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> >\n> > > me about two or three days to do the parser and rewrite for sublinks.\n> > > How many coding days is the rest taking? I am curious.\n> >\n> > All was near ready Feb 6, but ... I got problems from canonification\n> > routines in optimizer: 'where A AND B OR C ' becomes 'where (A OR C) AND (B OR C)'\n> > and if C is subselect then ... C has to be processed twice (sometime) ...\n> > I remember your question about handling 'A AND (B OR SubSelect)' - unfortunately,\n> > I didn't see problems here (actually, there is no problem with this\n> > conjunctive normal form :), but there are troubles in common case.\n> >\n> > Ideas ?\n> >\n> > For the moment I could leave this un-optimized, with possibility\n> > of double subselect processing...\n> \n> I would leave it for later. We already have an optimizer explosion\n> problem when we have lots of OR's so at some point we may have to\n> revisit the cnf-ify process anyway. Commerical Ingres has or had the\n> same problem.\n\nOk, follow this way...\n\nVadim\n",
"msg_date": "Tue, 10 Feb 1998 00:56:06 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [PATCHES] Re: [HACKERS] Query->hasSubLinks is always FALSE..."
}
] |
[
{
"msg_contents": "Here is a bug report that outlines out cnf-ify problem. Looks like we\nneed a general solution, not just for subselects.\n\n---------------------------------------------------------------------------\n\nForwarded message:\n> From [email protected] Wed Jan 14 10:20:16 1998\n> X-Authentication-Warning: u1.abs.net: nobody set sender to insightdist.com!daveh using -f\n> Message-Id: <[email protected]>\n> Date: Mon, 12 Jan 1998 15:53:18 -0500\n> From: David Hartwig <[email protected]>\n> Organization: Insight Distribution Systems\n> X-Mailer: Mozilla 4.04 [en] (Win95; I)\n> Mime-Version: 1.0\n> To: Bruce Momjian <[email protected]>\n> Subject: Re: [BUGS] General Bug Report: palloc fails with lots of ANDs and ORs\n> References: <[email protected]>\n> Content-Type: multipart/mixed; boundary=\"------------20C7AC27E8BCA117B23354BE\"\n> \n> This is a multi-part message in MIME format.\n> --------------20C7AC27E8BCA117B23354BE\n> Content-Type: text/plain; charset=us-ascii\n> Content-Transfer-Encoding: 7bit\n> \n> Bruce,\n> \n> I did some homework. Here is what I have. The default max data segment size on our (AIX 4.1.4) box is around 130000 kbytes.\n> \n> I put together a query which put me just past the threshold of the palloc \"out of memory error\". It is as follows:\n> \n> create table outlet (\n> number int,\n> name varchar(30),\n> ...\n> }\n> \n> create unique index outlet_key on outlet using btree (number);\n> \n> select count(*) from outlet\n> where\n> (number = 1 and number = 1 and number = 1) or\n> (number = 1 and number = 1 and number = 1) or\n> (number = 1 and number = 1 and number = 1) or\n> (number = 1 and number = 1 and number = 1) or\n> (number = 1 and number = 1 and number = 1) or\n> (number = 1 and number = 1 and number = 1) or\n> (number = 1 and number = 1 and number = 1) or\n> (number = 1 and number = 1 and number = 1) or\n> (number = 1 and number = 1 and number = 1);\n> \n> Not pretty but it makes the point. Take out two OR clauses and the query works fine (but a bit slow).\n> \n> The above query is all it takes to use up all 130000 Kbytes of memory. And, since the query takes a long time to finally fail, I was able to\n> observe the memory consumption.\n> \n> I extended the max data segment to 300000. And tried again. I could observer the memory consumption up to about 280000 when the system\n> suddenly got sick. I was getting all kinds of messages like \"cant fork\"; bad stuff. The system did finally recover on its own. I am not\n> sure happened there. I know that ulimit puts us right around the physical memory limits of out system.\n> \n> Using 300 meg for the above query seems like a bit of a problem. It is difficult to imagine where all that memory is being used. I will\n> research the problem further if you need more information.\n> \n> Bruce Momjian wrote:\n> \n> > Try changing your OS default memory size. Unsure how to do this under\n> > AIX.\n> >\n> > >\n> > >\n> > > ============================================================================\n> > > POSTGRESQL BUG REPORT TEMPLATE\n> > > ============================================================================\n> > >\n> > >\n> > > Your name : David Hartwig\n> > > Your email address : [email protected]\n> > >\n> > > Category : runtime: back-end: SQL\n> > > Severity : serious\n> > >\n> > > Summary: palloc fails with lots of ANDs and ORs\n> > >\n> > > System Configuration\n> > > --------------------\n> > > Operating System : AIX 4.1\n> > >\n> > > PostgreSQL version : 6.2\n> > >\n> > > Compiler used : native CC\n> > >\n> > > Hardware:\n> > > ---------\n> > > RS 6000\n> > >\n> > > Versions of other tools:\n> > > ------------------------\n> > > NA\n> > >\n> > > --------------------------------------------------------------------------\n> > >\n> > > Problem Description:\n> > > --------------------\n> > > The follow is a mail message describing the problem on the PostODBC mailing list:\n> > >\n> > >\n> > > I have run across this also. We traced it down to a failure in the PostgreSQL server. This occurs under the following conditions.\n> > >\n> > > 1. MS Access\n> > > 2. Specify a multi-part key in the link time setup with postgresql\n> > > 3. Click on table view.\n> > >\n> > > What happens is MS Access takes the following steps. First it selects all possible key values for the table being viewed. I\n> > > suspect it maps the key values to the relative row position in the display. Then it uses the mapping to generate future queries based\n> > > on the mapping and the rows showing on the screen. The queries take the following form:\n> > >\n> > > SELECT keypart1, keypart2, keypart3, col4, col5, col6 ... FROM example_table\n> > > WHERE\n> > > (keypart1 = row1keypartval1 AND keypart2 = row1keypartval2 AND keypart3 = row1keypartval3) OR\n> > > (keypart1 = row2keypartval1 AND keypart2 = row2keypartval2 AND keypart3 = row2keypartval3) OR\n> > > .\n> > > . -- 28 lines of this stuff. Why 28... Why not 28\n> > > .\n> > > (keypart1 = row27keypartval1 AND keypart2 = row27keypartval2 AND keypart3 = row27keypartval3) OR\n> > > (keypart1 = row28keypartval1 AND keypart2 = row28keypartval2 AND keypart3 = row28keypartval3);\n> > >\n> > >\n> > > The PostgreSQL sever chokes on this statement claiming it is out of memory. (palloc) In this example I used a three part key. I\n> > > do not recall if a three part key is enough to trash the backend. It has been a while. I have tried sending these kinds of statements\n> > > directly through the psql monitor and get the same result.\n> > >\n> > >\n> > > --------------------------------------------------------------------------\n> > >\n> > > Test Case:\n> > > ----------\n> > > select c1, c1 c3, c4, c5 ... from example_table\n> > > where\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something) or\n> > > (c1 = something and c2 = something and c3 = something and c4 = something);\n> > >\n> > >\n> > > --------------------------------------------------------------------------\n> > >\n> > > Solution:\n> > > ---------\n> > >\n> > >\n> > > --------------------------------------------------------------------------\n> > >\n> > >\n> > >\n> >\n> > --\n> > Bruce Momjian\n> > [email protected]\n> \n> \n> \n> --------------20C7AC27E8BCA117B23354BE\n> Content-Type: text/x-vcard; charset=us-ascii; name=\"vcard.vcf\"\n> Content-Transfer-Encoding: 7bit\n> Content-Description: Card for David Hartwig\n> Content-Disposition: attachment; filename=\"vcard.vcf\"\n> \n> begin: vcard\n> fn: David Hartwig\n> n: Hartwig;David\n> org: Insight Distribution Systems\n> adr: 222 Shilling Circle;;;Hunt Valley ;MD;21030;USA\n> email;internet: [email protected]\n> title: Manager Research & Development\n> tel;work: (410)403-2308\n> x-mozilla-cpt: ;0\n> x-mozilla-html: TRUE\n> version: 2.1\n> end: vcard\n> \n> \n> --------------20C7AC27E8BCA117B23354BE--\n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 9 Feb 1998 12:49:58 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "cnf-ify problem"
}
] |
[
{
"msg_contents": "I'm seeing what looks like an ugly bug in the February 8th snapshot,\nrunning under NetBSD/sparc 1.3. Here's the sequence of events (note\nthat it's doesn't have to be this particular table; I've reproduced\nthis with different ones):\n\n* I run initdb as 'postgres', and create a user 'tih'.\n* As 'tih', I then create a data base 'sr11', and connect to it with psql.\n* I then create a table and an index from psql, thus:\n\ncreate table food_des (\n\tndb_no\t\tchar(5)\t\tnot null,\n\tfdgp_cd\t\tchar4\t\tnot null,\n\tdescrip\t\ttext\t\tnot null,\n\tshrt_desc\tvarchar(60)\tnot null,\n\tref_desc\tvarchar(45)\tnot null,\n\trefuse\t\tint2,\n\tsciname\t\tvarchar(60),\n\tn_factor\tfloat4,\n\tpro_factor\tfloat4,\n\tfat_factor\tfloat4,\n\tcho_factor\tfloat4\n);\n\ncreate unique index food_des_ndb_no on food_des (ndb_no);\n\n* Then, in the psql session, I load data from an external file, thus:\n\ncopy food_des from '/u/tih/databases/SR11/food_des.load' using delimiters '^';\n\n* While this is running, in a separate shell I try to create the default\n* data base for user 'tih' (upon this error, disk activity stops):\n\nbarsoom:tih> createdb\nConnection to database 'template1' failed.\nPQexec() -- There is no connection to the backend.\ncreatedb: database creation failed on tih.\nbarsoom:tih>\n\n* When this happens, the postmaster process reports the following:\n\nERROR: cannot write block 6 of food_des_ndb_no [sr11] blind\n\n* The 'postmaster' and one 'postgres' process are still active.\n* Further attempts at starting psql sessions will just hang.\n\nAny ideas? I can reproduce this easily, so I can test any time.\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n\n",
"msg_date": "Mon, 9 Feb 1998 21:54:58 +0100 (CET)",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": true,
"msg_subject": "Bug: interference between two sessions"
}
] |
[
{
"msg_contents": "I have applied a patch to pass atttypmod all the way through to the\nexecutor. This removes some of the hacks for select into table for\nchar(), and cleans up the code quite a bit.\n\nI have added typmod to Resdom and Var, and this allows its use in\nTupleDescInitEntry.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 9 Feb 1998 23:15:00 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "atttypmod"
}
] |
[
{
"msg_contents": "Another thing the change does is to pass atttypmod to output functions\nas the third parameter.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 9 Feb 1998 23:21:56 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "atttypmod"
}
] |
[
{
"msg_contents": "Ok. I'll be setting up postgresql on 4 platforms (Linux, Solaris, AIX, IRIX)\nand I'm wondering where to get the latest and greatest versions.\n\nThanks,\nMaurice\n\n\n",
"msg_date": "Tue, 10 Feb 1998 10:07:31 +0100",
"msg_from": "\"Maurice Gittens\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Where to get snapshots"
},
{
"msg_contents": "On Tue, 10 Feb 1998, Maurice Gittens wrote:\n\n> Ok. I'll be setting up postgresql on 4 platforms (Linux, Solaris, AIX, IRIX)\n> and I'm wondering where to get the latest and greatest versions.\n\n\tftp.postgresql.org:/pub/postgresql-snapshot.tar.gz\n\n\n",
"msg_date": "Tue, 10 Feb 1998 06:54:51 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Where to get snapshots"
}
] |
[
{
"msg_contents": "tempx and tempx2 are tables with single int4 field. Having single record\nwith value of 1 in both tables:\n\nvac=> explain select * from tempx, tempx2 where tx = tx2;\nNOTICE: QUERY PLAN:\n\nHash Join (cost=17.20 size=100 width=8)\n -> Seq Scan on tempx2 (cost=4.30 size=100 width=4)\n -> Hash (cost=0.00 size=0 width=0)\n -> Seq Scan on tempx (cost=4.30 size=100 width=4)\n\nEXPLAIN\nvac=> begin;\nBEGIN\nvac=> declare c cursor for select * from tempx, tempx2 where tx = tx2;\nSELECT\nvac=> fetch 1 in c;\ntx|tx2\n--+---\n 0| 0\n(1 row)\n\nvac=> fetch 1 in c;\ntx|tx2\n--+---\n 1| 1\n(1 row)\nvac=> fetch backward 1 in c;\ntx|tx2\n--+---\n 0| 0\n(1 row)\n^^^^^^^^^^^^^^^^^^^^^^ - very nice!\nvac=> end;\nEND\nvac=> insert into tempx values (1);\nINSERT 320416 1\nvac=> begin;\nBEGIN\nvac=> declare c cursor for select * from tempx, tempx2 where tx = tx2;\nSELECT\nvac=> fetch 1 in c;\ntx|tx2\n--+---\n 0| 0\n(1 row)\n\nvac=> fetch 1 in c;\ntx|tx2\n--+---\n 1| 1\n(1 row)\n\nvac=> fetch backward 1 in c;\ntx|tx2\n--+---\n 1| 1\n(1 row)\n^^^^^^^^^^^^^^^^^^^^^^^^^^ \n\nHashJoin doesn't care about scan dirrection...\n\nThis is another story:\n\nvac=> fetch backward 1 in c;\ntx|tx2\n--+---\n 0| 0\n(1 row)\n\nvac=> fetch backward 1 in c;\ntx|tx2\n--+---\n(0 rows)\n\nvac=> fetch 1 in c;\nPQexec() -- Request was sent to backend, but backend closed the channel before responding.\n This probably means the backend terminated abnormally before or while processing the request.\n\nVadim\n",
"msg_date": "Tue, 10 Feb 1998 16:33:01 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "bug: HashJoin and backward cursors..."
}
] |
[
{
"msg_contents": "Great. Thanks.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tThomas G. Lockhart [SMTP:[email protected]]\n> Sent:\tMonday, February 09, 1998 5:07 PM\n> To:\tBruce Momjian\n> Cc:\[email protected]; [email protected];\n> [email protected]\n> Subject:\tRe: [HACKERS] Bug?\n> \n> Bruce Momjian wrote:\n> \n> > > Per Tom H's suggestion; what do you think Bruce?\n> > >\n> > > postgres=> select 100000000000;\n> > > NOTICE: Integer input '100000000000' is out of range; promoted to\n> float\n> > > ?column?\n> > > ------------\n> > > 100000000000\n> > > (1 row)\n> > >\n> > > That would alleviate the \"hidden\" side effects, but still come\n> closer to\n> > > doing something helpful...\n> >\n> > With the NOTICE, I like it.\n> \n> Will be done for v6.3 then. I'm finishing up on patches to gram.y to\n> help\n> with CREATE FUNCTION, etc., wrt type names. Will commit the scan.l\n> changes at\n> that time...\n> \n> - Tom\n",
"msg_date": "Tue, 10 Feb 1998 10:38:48 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Bug?"
}
] |
[
{
"msg_contents": "Hi,\n\nEvery now and then the postmaster dies without reason.\nI'm using vanilla 6.2.1 under Linux 2.0.33 (libc 5).\nI also make use of large objects.\ngdb and the core file says that the crash is happening in free().\n\nDoes anyone have a clue? \n\n#0 0x810ae53 in free ()\n#1 0x80e609e in AllocSetFree ()\n#2 0x80e6583 in pstrdup ()\n#3 0x80e62a2 in MemoryContextFree ()\n#4 0x80e647d in pfree ()\n#5 0x80c618d in HandleFunctionRequest ()\n#6 0x80c7768 in PostgresMain ()\n#7 0x8095d03 in main ()\n#8 0x80648cb in _start ()\n\nThanks in advance.\nMaurice\n",
"msg_date": "Tue, 10 Feb 1998 11:00:58 +0100",
"msg_from": "Maurice Gittens <[email protected]>",
"msg_from_op": true,
"msg_subject": "postmaster dies"
}
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.