threads
listlengths 1
2.99k
|
|---|
[
{
"msg_contents": "Hello, \n\nI have encountered a bug in 6.4 which reveals itself\nin the following query:\n\nQuery:\n\nSELECT supply, description, quantity, supplyunit, company||costcntr,\ntarget, saledate, \nprice, 'pad' FROM sales WHERE target IN\n('13132','13042','13070','12989','13120','12972') \nAND supply IN \n('1300600','1400034','0210300','1600966','1132050',\n'1305522','0135250',\n'1300326','1301695','1302009','1305169','0216190',\n'1600735') AND saletype \nIN ('Requisition','Replentish','Transfer') \nORDER BY supply, target, saledate;\n\nI recompiled the postmaster with -g to yeild the \nfollowing backtrace:\n\nGDB Trace:\n\n#0 varchareq (arg1=0x32f9 <Address 0x32f9 out of bounds>,\narg2=0x81dd558 \"\\013\") at varchar.c:636\n636 len1 = VARSIZE(arg1) - VARHDRSZ;\n\n(gdb) bt\n\n#0 varchareq (arg1=0x32f9 <Address 0x32f9 out of bounds>,\narg2=0x81dd558 \"\\013\") at varchar.c:636\n#1 0x80f7ad3 in fmgr_c (finfo=0x81e58a8, values=0xbfffcefc,\nisNull=0xbfffcf87 \"\") at fmgr.c:104\n#2 0x808c92d in ExecMakeFunctionResult (node=0x81dd4a8,\narguments=0x81e03a8, econtext=0x81e2fa0, \n isNull=0xbfffcf87 \"\", \n isDone=0xbfffcf4b\n\"\\001l���������������\\b\\b���\\003\\036\\b���/\\036\\b\\207���������\\020\\004\\036\\b���/\\036\\bH0\\036\\b\\210���������=���\\b\\b���\\0\n03\\036\\b���/\\036\\b\\207���������\\206���������!���\\001\") at execQual.c:826\n#3 0x808c970 in ExecEvalOper (opClause=0x81e03e8, econtext=0x81e2fa0,\nisNull=0xbfffcf87 \"\")\n at execQual.c:888\n#4 0x808cbb4 in ExecEvalExpr (expression=0x81e03e8,\necontext=0x81e2fa0, isNull=0xbfffcf87 \"\", \n isDone=0xbfffcf86 \"\\001\") at execQual.c:1219\n#5 0x808cc3d in ExecQualClause (clause=0x81e03e8, econtext=0x81e2fa0)\nat execQual.c:1281\n#6 0x808cc71 in ExecQual (qual=0x81e0410, econtext=0x81e2fa0) at\nexecQual.c:1347\n#7 0x80908d1 in IndexNext (node=0x81e0ed0) at nodeIndexscan.c:157\n#8 0x808cf64 in ExecScan (node=0x81e0ed0, accessMtd=0x80907f8\n<IndexNext>) at execScan.c:102\n#9 0x809098c in ExecIndexScan (node=0x81e0ed0) at nodeIndexscan.c:211\n#10 0x808bb16 in ExecProcNode (node=0x81e0ed0, parent=0x81e1aa0) at\nexecProcnode.c:271\n#11 0x80fb22e in createfirstrun (node=0x81e1aa0) at psort.c:402\n#12 0x80fb068 in initialrun (node=0x81e1aa0) at psort.c:293\n#13 0x80faf47 in psort_begin (node=0x81e1aa0, nkeys=3, key=0x81e47a8)\nat psort.c:155\n#14 0x809213f in ExecSort (node=0x81e1aa0) at nodeSort.c:156\n#15 0x808bb3a in ExecProcNode (node=0x81e1aa0, parent=0x81e1aa0) at\nexecProcnode.c:295\n#16 0x808b34f in ExecutePlan (estate=0x81e1b50, plan=0x81e1aa0,\nparseTree=0x81a32b0, operation=CMD_SELECT, \n numberTuples=0, direction=ForwardScanDirection,\nprintfunc=0x8061bcc <printtup>) at execMain.c:733\n#17 0x808af00 in ExecutorRun (queryDesc=0x81e1b28, estate=0x81e1b50,\nfeature=3, count=0) at execMain.c:236\n#18 0x80d4f83 in ProcessQueryDesc (queryDesc=0x81e1b28) at pquery.c:333\n#19 0x80d4fe2 in ProcessQuery (parsetree=0x81a32b0, plan=0x81e1aa0,\ndest=Remote) at pquery.c:376\n#20 0x80d3ad4 in pg_exec_query_dest (\n query_string=0xbfffd20c \"SELECT supply, description, quantity,\nsupplyunit, company||costcntr, target, sal\nedate,\\nprice, 'pad' FROM sales WHERE target IN\n('13132','13042','13070','12989','13120','12972')\\nAND supply\n IN ('1300600\"..., dest=Remote, aclOverride=0) at postgres.c:800\n#21 0x80d39cb in pg_exec_query (\n query_string=0xbfffd20c \"SELECT supply, description, quantity,\nsupplyunit, company||costcntr, target, sal\nedate,\\nprice, 'pad' FROM sales WHERE target IN\n('13132','13042','13070','12989','13120','12972')\\nAND supply\n IN ('1300600\"...) at postgres.c:699\n#22 0x80d49e8 in PostgresMain (argc=7, argv=0xbffff284, real_argc=5,\nreal_argv=0xbffffc28)\n---Type <return> to continue, or q <return> to quit---\n at postgres.c:1645\n#23 0x80c0652 in DoBackend (port=0x8167900) at postmaster.c:1528\n#24 0x80c01a9 in BackendStartup (port=0x8167900) at postmaster.c:1299\n#25 0x80bf921 in ServerLoop () at postmaster.c:757\n#26 0x80bf500 in PostmasterMain (argc=5, argv=0xbffffc28) at\npostmaster.c:563\n#27 0x8099b54 in main (argc=5, argv=0xbffffc28) at main.c:93\n\nThe sales table is a table with ~250K rows and is \ndefined as follows:\n\nSales Table:\n\nTable = sales\nsupplysource int4 not null\nsupply varchar(16) not null\nsupplyunit varchar(2) not null\nquantity float8 not null\ntarget int4 not null\ncompany varchar(8) not null\ncostcntr varchar(8) not null\nsaletype varchar(16) not null\nsaledate datetime not null\nprice float8 not null\nvendorname varchar(32) not null\nvendor varchar(16) not null\nvendorgroup varchar(16)\ndescription varchar(64) not null\n\nIndices: k_sales1\n k_sales2\n k_sales3\n k_sales4\n\nThis bug is reproducable and occurred after I added\nthe description column to the sales table.\n\nThanks for any help,\n\nMarcus Mascari ([email protected])\n\n\n\n\n\n\n\n\n\n_________________________________________________________\nDO YOU YAHOO!?\nGet your free @yahoo.com address at http://mail.yahoo.com\n\n",
"msg_date": "Tue, 1 Dec 1998 16:15:38 -0800 (PST)",
"msg_from": "Marcus Mascari <[email protected]>",
"msg_from_op": true,
"msg_subject": "Bug in 6.4 query of table in varchar.c"
}
] |
[
{
"msg_contents": "[Charset iso-8859-1 unsupported, filtering to ASCII...]\n> Hello,\n> \n> You have told me about notifying postgres developers with PL/pgSQL troubles\n> in BSDI-3.1.\n> How can I know, if any patches availibe, will it appear in GENERAL mailing\n> list ?\n\nI am CC'ing this to the hackers list. PL/pgsql is failing on BSDI 3.1,\nbut I don't run that anymore. Please report the problem to the hackers\nlist with the regression test output for the PL/pgsql part of the test. \nThanks.\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": "Tue, 1 Dec 1998 20:28:48 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: PL/pgSQL on BSDI-3.1"
}
] |
[
{
"msg_contents": "I've posted a new statically-linked CVSup client for linux-libc5 boxes\nin:\n\n ftp://postgresql.org/pub/CVSup/cvsup-15.4.2-client-linux-libc5.tar.gz\n\nI've noticed many checksum errors when using my older cvsup-15.2\ninstallation, but have not yet had a real update to test this newer\npackage to see if it helps.\n\nI used my existing from-sources Modula-3 installation to build it. That\nform of Modula-3 installation may not be for everyone, since it took\nover 200MB of disk space during the installation and ~50MB after. But...\n\nI found an rpm-packaged version of Modula-3 which was trivial to install\non my glibc2 box, but have not yet been successful at building cvsup for\nthat machine. It looks like a configuration problem of some sort, since\nit fails on the very first m3makefile statement. If anyone has an\ninterest in it I can give them the urls and point them at the CVSup\nauthor; not sure if I'll have time to pursue it for a while...\n\n - Tom\n",
"msg_date": "Wed, 02 Dec 1998 05:49:48 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "New Linux/libc5 CVSup client"
},
{
"msg_contents": "Hi,\n\nI'm the author of CVSup. I'm really glad you cc'd me on your mail.\n\nOn 02-Dec-98 Thomas G. Lockhart wrote:\n> I've posted a new statically-linked CVSup client for linux-libc5 boxes\n> in:\n> \n> ftp://postgresql.org/pub/CVSup/cvsup-15.4.2-client-linux-libc5.tar.gz\n\nGreat! I've been meaning to try to get some newer binaries for\nplatforms I don't have.\n\n> I've noticed many checksum errors when using my older cvsup-15.2\n> installation, but have not yet had a real update to test this newer\n> package to see if it helps.\n\nIt should fix that problem. The checksum errors were caused by\nsome whitespace changes in RCS files that were introduced in recent\nversions of CVS. I made the checksumming algorithm for RCS files\nignore inconsequential whitespace differences, so this problem\nshouldn't recur no matter what they do to CVS in the future.\n\nThere's still one potential source of checksum errors. The very\nnewest versions of CVS support some new features such as preserving\nfile modes and handling links. These are implemented using RCS file\nextensions (see rcsfile(5) for details) which are not yet supported by\nCVSup. Luckily, it seems that few people use those extensions. I am\ngoing to implement a general scheme for handling them, hopefully for\nthe next release of CVSup. Even if some files use the new extensions,\nthe current version will update them OK. You'll see checksum errors,\nbut the \"fixups\" at the end will correct them (slower, of course).\n\n> I used my existing from-sources Modula-3 installation to build it. That\n> form of Modula-3 installation may not be for everyone, since it took\n> over 200MB of disk space during the installation and ~50MB after. But...\n> \n> I found an rpm-packaged version of Modula-3 which was trivial to install\n> on my glibc2 box, but have not yet been successful at building cvsup for\n> that machine. It looks like a configuration problem of some sort, since\n> it fails on the very first m3makefile statement.\n\nThis is the so-called \"PM3\" release of Modula-3, right? I've made\nthe necessary changes to build CVSup under that version of Modula-3.\nI've been holding off on releasing it because I'd like to get some\nother new features into the program first. But it's well tested on\na couple of FreeBSD mirror sites. I put the sources up for FTP at:\n\n ftp://ftp.polstra.com/private/lockhart/cvsup-15.5.tar.gz\n\nThe MD5 checksum is 1f2466e83d06beb1ab2a2400515f9817.\n\nYou should be able to build it with \"make M3TARGET=LINUXLIBC6\". I'd\nlove to find out whether it works OK under Linux, and fix it if it\ndoesn't.\n\nI'm going to be on vacation next week, so I may be a bit unresponsive\nto mail during that time.\n\nJohn\n---\n John Polstra [email protected]\n John D. Polstra & Co., Inc. Seattle, Washington USA\n \"Nobody ever went broke underestimating the taste of the American public.\"\n -- H. L. Mencken\n",
"msg_date": "Wed, 02 Dec 1998 09:03:10 -0800 (PST)",
"msg_from": "John Polstra <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: New Linux/libc5 CVSup client"
},
{
"msg_contents": "> > I've noticed many checksum errors when using my older cvsup-15.2\n> > installation, but have not yet had a real update to test this newer\n> > package to see if it helps.\n> It should fix that problem. The checksum errors were caused by\n> some whitespace changes in RCS files that were introduced in recent\n> versions of CVS. I made the checksumming algorithm for RCS files\n> ignore inconsequential whitespace differences, so this problem\n> shouldn't recur no matter what they do to CVS in the future.\n\nI wasn't complaining; it's nice that your fallback technique of full\ntransfer made it a \"don't care\".\n\n> This is the so-called \"PM3\" release of Modula-3, right?\n\nYes. Any opinions on it? Hopefully it isn't too incompatible with other\nreleases. They seem to be enthusiastic, and RPM support really helps\nwith the initial installation...\n\n> I'm going to be on vacation next week, so I may be a bit unresponsive\n> to mail during that time.\n\nNo problem. My glibc2 machine is at work, so I'll test there tomorrow if\nI have time. Have a good vacation if I don't talk to you before.\n\n - Tom\n",
"msg_date": "Thu, 03 Dec 1998 01:00:45 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: New Linux/libc5 CVSup client"
},
{
"msg_contents": "On 03-Dec-98 Thomas G. Lockhart wrote:\n\n> I wasn't complaining; it's nice that your fallback technique of full\n> transfer made it a \"don't care\".\n\nOh, I didn't think you were complaining; I was just explaining. I\nalways worry that when people see those \"checksum mismatch\" messages,\nthey'll think it's because the program is buggy. Then they might\ndecide not to trust it at all. But because of the safeguards that it\ncontains (\"fixups\"), it's impossible (to the best of my knowledge) for\nthe package to deliver a corrupted file. Even if you kill it manually\nat any point, you'll either get a correct file or you'll still have\nyour original version.\n\n>> This is the so-called \"PM3\" release of Modula-3, right?\n> \n> Yes. Any opinions on it?\n\nIt's _the_ free release to use these days. (There's also a commercial\nM3 put out by Critical Mass <http://www.cmass.com/>.) DEC SRC hasn't\ndone anything with their M3 release for ages now. Some of the\nstalwarts took over maintenance in their spare time, and the result\nis PM3. Most of the progress has been in the Linux arena, since most\nof the PM3 people run Linux. I've gotten it ported to FreeBSD, but\nhaven't submitted the changes back to the PM3 folks yet. There are\nstill some wrinkles to iron out, due to the fact that FreeBSD is in\na transition period from a.out to ELF.\n\nAnyway, for Linux PM3 seems to be a big step forward. At least they\nare keeping up with the libc changes that keep coming along lately.\n\n> Hopefully it isn't too incompatible with other releases.\n\nIt's mostly compatible except for some details of the build system\nthat most programs never run into.\n\n> They seem to be enthusiastic, and RPM support really helps with the\n> initial installation...\n\nYes indeed.\n\n> My glibc2 machine is at work, so I'll test there tomorrow if I have\n> time. Have a good vacation if I don't talk to you before.\n\nThanks!\n\nJohn\n---\n John Polstra [email protected]\n John D. Polstra & Co., Inc. Seattle, Washington USA\n \"Nobody ever went broke underestimating the taste of the American public.\"\n -- H. L. Mencken\n",
"msg_date": "Wed, 02 Dec 1998 18:24:33 -0800 (PST)",
"msg_from": "John Polstra <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: New Linux/libc5 CVSup client"
}
] |
[
{
"msg_contents": "> Just a quick question. What is the current state of \n> PostgreSQL on Windows\n> NT (yeuch).\n> \n> I'm asking, because I'm pushing PostgreSQL here, and their \n> eye's lit up as\n> soon as they heared that someone has got it running (even partially\n> helps).\n\nPlease look at , there are some details what to do. My last patch for final\n6.4 and Cygwin B19 is included (after applying pgsql64.diff apply two other\npatches too). Cygwin B20 needs some minor changes (cygwin32 -> cygwin, ...).\nThis version needs binary mounts in Cygwin.\n\nIt should run also the regression tests with some tests failed - some\ndate/time and every test where dynamic loading of some modules is needed.\nThe dynamic loaded modules are created by the makefiles, but there are\nproblems with accessing exported variables - we are working on a solution\n(dllhelpers from Mumit Khan should help).\n\n\t\t\t\tDan Horak",
"msg_date": "Wed, 2 Dec 1998 08:52:00 +0100 ",
"msg_from": "Horak Daniel <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] NT port"
}
] |
[
{
"msg_contents": ">> You can specify a nonzero value for a serial column \n>> (as long as it does not duplicate any existing value in that column), ...\n>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n>> \n>> ?!!!\n\n>You can assign a value to a serial column, as long as it is unique.\n\nOnly an explicitly created unique index will enforce behavior this in\nInformix.\n(I tested this just to be sure)\n\nAndreas\n",
"msg_date": "Wed, 2 Dec 1998 10:43:04 +0100 ",
"msg_from": "Zeugswetter Andreas IZ5 <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] DROPping tables with SERIALs"
}
] |
[
{
"msg_contents": "> Please look at , there are some details what to do. My last \n> patch for final\nI forgot the URL, it is http://www.askesis.nl/AskesisPostgresIndex.html\n\n\t\t\t\tDan Horak\n",
"msg_date": "Wed, 2 Dec 1998 11:38:24 +0100 ",
"msg_from": "Horak Daniel <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] NT port"
},
{
"msg_contents": "> > Please look at , there are some details what to do. My last \n> > patch for final\n> I forgot the URL, it is http://www.askesis.nl/AskesisPostgresIndex.html\n\nAlready in the FAQ. Glad you updated it for 6.4 final. Looks very\nlean. We should probably merge this into the main tree when you feel\ncomfortable, but having a URL for the patch is a nice idea, and allows\nyou to independently improve it.\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": "Wed, 2 Dec 1998 10:21:03 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] NT port"
},
{
"msg_contents": "Hi,\n\nthere is a port of Postgresql to NT which works under Uwin\nI think it could be also added to the FAQ\n\n\tRegards,\n\t\tOleg\n\nOn Wed, 2 Dec 1998, Bruce Momjian wrote:\n\n> Date: Wed, 2 Dec 1998 10:21:03 -0500 (EST)\n> From: Bruce Momjian <[email protected]>\n> To: Horak Daniel <[email protected]>\n> Cc: [email protected], [email protected]\n> Subject: Re: [HACKERS] NT port\n> \n> > > Please look at , there are some details what to do. My last \n> > > patch for final\n> > I forgot the URL, it is http://www.askesis.nl/AskesisPostgresIndex.html\n> \n> Already in the FAQ. Glad you updated it for 6.4 final. Looks very\n> lean. We should probably merge this into the main tree when you feel\n> comfortable, but having a URL for the patch is a nice idea, and allows\n> you to independently improve it.\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> \n\n_____________________________________________________________\nOleg Bartunov, sci.researcher, hostmaster of AstroNet,\nSternberg Astronomical Institute, Moscow University (Russia)\nInternet: [email protected], http://www.sai.msu.su/~megera/\nphone: +007(095)939-16-83, +007(095)939-23-83\n\n",
"msg_date": "Wed, 2 Dec 1998 19:08:51 +0300 (MSK)",
"msg_from": "Oleg Bartunov <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] NT port"
}
] |
[
{
"msg_contents": "My database is in weird state after some operation. Here is an example:\n\n[david@kerberos david]$ psql\n.....\ndavid=> begin; \nBEGIN\ndavid=> create table tbl ( i int );\nCREATE\ndavid=> insert into tbl values(1);\nINSERT 104137 1\nEOFid=> [<ctrl-d> here]\n[david@kerberos david]$ psql\n....\ndavid=> create table tbl ( a text );\nERROR: cannot create tbl\ndavid=> \\d\nCouldn't find any tables, sequences or indices!\n ====================== cut here ======================\nSimple question: why 'ERROR: cannot create tbl' ?\nPlatform: linux i586, glibc, regression tests seems ok (except minor\nproblem with floating point. The some problem have my colleague.\n\n\t\t\t\tthanks,\n-- \n* David Sauer, student of Czech Technical University\n* electronic mail: [email protected] (mime compatible)\n* boycott czech telecom >>>> http://www.bojkot.cz/ (czech)\n* >>>> http://www.bojkot.cz/english/ (english)\n",
"msg_date": "02 Dec 1998 12:17:26 +0100",
"msg_from": "David Sauer <[email protected]>",
"msg_from_op": true,
"msg_subject": "weird state after aborted transaction in pgsql 6.4"
},
{
"msg_contents": ">\n> My database is in weird state after some operation. Here is an example:\n>\n> [david@kerberos david]$ psql\n> .....\n> david=> begin;\n> BEGIN\n> david=> create table tbl ( i int );\n> CREATE\n> david=> insert into tbl values(1);\n> INSERT 104137 1\n> EOFid=> [<ctrl-d> here]\n> [david@kerberos david]$ psql\n> ....\n> david=> create table tbl ( a text );\n> ERROR: cannot create tbl\n> david=> \\d\n> Couldn't find any tables, sequences or indices!\n> ====================== cut here ======================\n> Simple question: why 'ERROR: cannot create tbl' ?\n> Platform: linux i586, glibc, regression tests seems ok (except minor\n> problem with floating point. The some problem have my colleague.\n\n The problem is, that the CREATE TABLE is inside the\n transaction. So the catalog updates for the new table get\n rolled back, but the file is still there. On the next CREATE\n TABLE the file creation (made with O_EXCL) fails.\n\n You can safely remove the file .../data/base/david/tbl and\n you should execute CREATE/DROP statements outside of\n transactions.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Wed, 2 Dec 1998 13:56:54 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] weird state after aborted transaction in pgsql 6.4"
},
{
"msg_contents": "Jan Wieck wrote:\n> \n> The problem is, that the CREATE TABLE is inside the\n> transaction. So the catalog updates for the new table get\n> rolled back, but the file is still there. On the next CREATE\n> TABLE the file creation (made with O_EXCL) fails.\n> \n> You can safely remove the file .../data/base/david/tbl and\n> you should execute CREATE/DROP statements outside of\n> transactions.\n\nIs there any plans (in TODO) to fix this ?\n\nEither by making the CREATE/DROP totally ignore transactions (probably\nhard),\not then update the transaction machanism to be able to delete files\ncreated \ninside the transaction.\n\n--------------\nHannu\n",
"msg_date": "Wed, 02 Dec 1998 17:00:53 +0200",
"msg_from": "Hannu Krosing <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] weird state after aborted transaction in pgsql 6.4"
},
{
"msg_contents": "> >\n> > My database is in weird state after some operation. Here is an example:\n> >\n> > [david@kerberos david]$ psql\n> > .....\n> > david=> begin;\n> > BEGIN\n> > david=> create table tbl ( i int );\n> > CREATE\n> > david=> insert into tbl values(1);\n> > INSERT 104137 1\n> > EOFid=> [<ctrl-d> here]\n> > [david@kerberos david]$ psql\n> > ....\n> > david=> create table tbl ( a text );\n> > ERROR: cannot create tbl\n> > david=> \\d\n> > Couldn't find any tables, sequences or indices!\n> > ====================== cut here ======================\n> > Simple question: why 'ERROR: cannot create tbl' ?\n> > Platform: linux i586, glibc, regression tests seems ok (except minor\n> > problem with floating point. The some problem have my colleague.\n> \n> The problem is, that the CREATE TABLE is inside the\n> transaction. So the catalog updates for the new table get\n> rolled back, but the file is still there. On the next CREATE\n> TABLE the file creation (made with O_EXCL) fails.\n> \n> You can safely remove the file .../data/base/david/tbl and\n> you should execute CREATE/DROP statements outside of\n> transactions.\n> \n\nAdded to TODO:\n\t\n\t* CREATE TABLE inside aborted transaction causes stray table file\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, 12 Dec 1998 23:07:17 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] weird state after aborted transaction in pgsql 6.4"
}
] |
[
{
"msg_contents": "Hi,\n\nI had little time the last few months but I will keep up the web page with\ninfo about the Windows NT port.\n\nThere are several people working (mostly Dan) on the port. If anyone has\nadditional patches (in whatever form) I am happy to provide them at my\nwebpage. Any info that will get to me about running Postgres on Windows NT\nwill appear on the same webpage as soon as I have it.\n\nAs far as I know the port works more or less (even with dll's, sometimes\nmore, sometimes less ;-)) if all mounting is done binary and one uses Cygnus\nB19 (Dan?).\n\nMy main target now is to run it with default mounting (text) because that is\nthe default way in Windows. At least one person fixed all the fopen's but\nalas I do not have his/hers patches and/or name.\n\nFor any info one can contact me.\n\nJoost Kraaijeveld\n\nAskesis B.V.\nGroenewoudeseweg 46\n6524VB Nijmegen\ntel: 024-3888063\nfax: 024-3608416\nweb: www.askesis.nl\n\n",
"msg_date": "Wed, 2 Dec 1998 20:59:30 +0100",
"msg_from": "\"Joost Kraaijeveld\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "NT port"
}
] |
[
{
"msg_contents": "============================================================================\n POSTGRESQL BUG REPORT TEMPLATE\n============================================================================\n\n\nYour name\t\t: Billy G. Allie\t\nYour email address\t: [email protected]\n\n\nSystem Configuration\n---------------------\n Architecture (example: Intel Pentium) \t: Intel i486DX2\n\n Operating System (example: Linux 2.0.26 ELF) \t: UnixWare 7.0.1\n\n PostgreSQL version (example: PostgreSQL-6.4) : PostgreSQL-6.4\n\n Compiler used (example: gcc 2.8.0)\t\t: Optimizing C Compilation Sys.\n (CCS) 3.2 08/18/98 (u701)\n\n\nPlease enter a FULL description of your problem:\n------------------------------------------------\nPL/pgSQL functions will fail to run if a table with a mixed case name is \nused/referenced.\n\n\nPlease describe a way to repeat the problem. Please try to provide a\nconcise reproducible example, if at all possible: \n----------------------------------------------------------------------\n$ psql mug <test.func\nCREATE FUNCTION \"In_Committees\" (\"int4\" ) RETURNS \"text\" AS '\nDECLARE\n committee_rec \"Committees\"%ROWTYPE;\nBEGIN\n SELECT * INTO committee_rec FROM Committees WHERE CommitteeID = $1;\n IF NOT FOUND THEN\n RETURN ''FALSE'';\n END IF;\n RETURN ''TRUE'';\nEND;' LANGUAGE 'plpgsql';\nCREATE\n\nselect \"In_Committees\" (2);\nNOTICE: plpgsql: ERROR during compile of In_Committees near line 2\nERROR: parse error at or near \"\"\"\nEOF\n$\n\n\nIf you know how this problem might be fixed, list the solution below:\n---------------------------------------------------------------------\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\n-- \n____ | Billy G. Allie | Domain....: [email protected]\n| /| | 7436 Hartwell | Compuserve: 76337,2061\n|-/-|----- | Dearborn, MI 48126| MSN.......: [email protected]\n|/ |LLIE | (313) 582-1540 |",
"msg_date": "Wed, 02 Dec 1998 20:56:25 -0500",
"msg_from": "\"Billy G. Allie\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Bug report: PL/pgSQL and mixed case table/field names."
},
{
"msg_contents": "Can someone comment on this?\n\n> ============================================================================\n> POSTGRESQL BUG REPORT TEMPLATE\n> ============================================================================\n> \n> \n> Your name\t\t: Billy G. Allie\t\n> Your email address\t: [email protected]\n> \n> \n> System Configuration\n> ---------------------\n> Architecture (example: Intel Pentium) \t: Intel i486DX2\n> \n> Operating System (example: Linux 2.0.26 ELF) \t: UnixWare 7.0.1\n> \n> PostgreSQL version (example: PostgreSQL-6.4) : PostgreSQL-6.4\n> \n> Compiler used (example: gcc 2.8.0)\t\t: Optimizing C Compilation Sys.\n> (CCS) 3.2 08/18/98 (u701)\n> \n> \n> Please enter a FULL description of your problem:\n> ------------------------------------------------\n> PL/pgSQL functions will fail to run if a table with a mixed case name is \n> used/referenced.\n> \n> \n> Please describe a way to repeat the problem. Please try to provide a\n> concise reproducible example, if at all possible: \n> ----------------------------------------------------------------------\n> $ psql mug <test.func\n> CREATE FUNCTION \"In_Committees\" (\"int4\" ) RETURNS \"text\" AS '\n> DECLARE\n> committee_rec \"Committees\"%ROWTYPE;\n> BEGIN\n> SELECT * INTO committee_rec FROM Committees WHERE CommitteeID = $1;\n> IF NOT FOUND THEN\n> RETURN ''FALSE'';\n> END IF;\n> RETURN ''TRUE'';\n> END;' LANGUAGE 'plpgsql';\n> CREATE\n> \n> select \"In_Committees\" (2);\n> NOTICE: plpgsql: ERROR during compile of In_Committees near line 2\n> ERROR: parse error at or near \"\"\"\n> EOF\n> $\n> \n> \n> If you know how this problem might be fixed, list the solution below:\n> ---------------------------------------------------------------------\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> \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[application/pgp-signature is not supported, skipping...]\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, 12 Dec 1998 23:09:43 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug report: PL/pgSQL and mixed case table/field names."
},
{
"msg_contents": "Bruce Momjian wrote:\n\n> Can someone comment on this?\n>\n> > ============================================================================\n> > POSTGRESQL BUG REPORT TEMPLATE\n> > ============================================================================\n> >\n> >\n> > Your name : Billy G. Allie\n> > Your email address : [email protected]\n> > [...]\n> >\n> > Please enter a FULL description of your problem:\n> > ------------------------------------------------\n> > PL/pgSQL functions will fail to run if a table with a mixed case name is\n> > used/referenced.\n> >\n> >\n> > Please describe a way to repeat the problem. Please try to provide a\n> > concise reproducible example, if at all possible:\n> > ----------------------------------------------------------------------\n> > $ psql mug <test.func\n> > CREATE FUNCTION \"In_Committees\" (\"int4\" ) RETURNS \"text\" AS '\n> > DECLARE\n> > committee_rec \"Committees\"%ROWTYPE;\n> > BEGIN\n> > SELECT * INTO committee_rec FROM Committees WHERE CommitteeID = $1;\n> > IF NOT FOUND THEN\n> > RETURN ''FALSE'';\n> > END IF;\n> > RETURN ''TRUE'';\n> > END;' LANGUAGE 'plpgsql';\n> > CREATE\n> >\n> > select \"In_Committees\" (2);\n> > NOTICE: plpgsql: ERROR during compile of In_Committees near line 2\n> > ERROR: parse error at or near \"\"\"\n> > EOF\n> > $\n\n Sure :-(\n\n PL/pgSQL's scanner/parser does not handle double quoted\n syntax for case mixed object names. Not a real bug, just a\n lack. Please add to TODO.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 14 Dec 1998 10:33:16 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug report: PL/pgSQL and mixed case table/field names."
},
{
"msg_contents": "> PL/pgSQL's scanner/parser does not handle double quoted\n> syntax for case mixed object names. Not a real bug, just a\n> lack. Please add to TODO.\n\nAdded to TODO:\n\t\n\t* plpgsql does not handle quoted mixed-case identifiers\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": "Mon, 14 Dec 1998 10:01:37 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug report: PL/pgSQL and mixed case table/field names."
}
] |
[
{
"msg_contents": "Hi, \n\tThis is a minor problem. I thought it would be better to post \nit and let someone else decide if it mattered or not. The problem is \nthat psql crashes when it encounters a relation name greater than 32 \ncharacters. However, a 32 character name is handled correctly with a \nlovely error. You can clearly see the problem below. \n\tSorry I can't be more helpful. If there is something I can do, \npoint me in the right direction. (email me, since I'm not subscribed)\n\nAaron.\n\ngeopow=> create table abcdefghijklmnopqrstuvwxy ( id serial );\nNOTICE: CREATE TABLE will create implicit sequence \nabcdefghijklmnopqrstuvwxy_id_seq for SERIAL column \nabcdefghijklmnopqrstuvwxy.id\nNOTICE: CREATE TABLE/UNIQUE will create implicit index \nabcdefghijklmnopqrstuvwxy_id_key for table abcdefghijklmnopqrstuvwxy\nERROR: the relation name abcdefghijklmnopqrstuvwxy_id_seq is >= 32 \ncharacters long\n\ngeopow=> create table abcdefghijklmnopqrstuvwxyz ( id serial );\npqReadData() -- backend closed the channel unexpectedly.\n This probably means the backend terminated abnormally before or \nwhile processing the request.\nWe have lost the connection to the backend, so further processing is \nimpossible. Terminating.\n",
"msg_date": "Wed, 2 Dec 1998 23:56:32 -0500 (EST)",
"msg_from": "aaron ross <[email protected]>",
"msg_from_op": true,
"msg_subject": "Problem w/ psql crashing"
}
] |
[
{
"msg_contents": "Hi all\n\nI have a simple patch about the treatment of functions.\nBut it may be self-satisfied. \nPlease check my patch at the end of this posting.\n\nCase 1. executor evaluates functions twice\n\n create table a (elem int4);\n create function ax2(int4) returns a as\n 'select $1*2;' language 'sql';\n select elem(ax2(1));\n\n the result is\n\n elem\n ----\n 4\n (1 row)\n\n it's wrong ?\n it must be the following ?\n\n elem\n ----\n 2\n (1 row)\n\nCase 2.parser rejects functions\n\n create function elemout(a) returns int4 as\n 'select $1.elem;' language 'sql';\n select elemout(ax2(1));\n\n ERROR: Function 'elemout' has bad return type 315169\n\n this elog() is necessary ?\n In my patch I removed it and the result is\n\n elemout\n -------\n 2\n (1 row)\n\nThanks\n\nHiroshi Inoue\[email protected]\n\n*** backend/parser/parse_func.c.orig\tThu Dec 3 18:25:31 1998\n--- backend/parser/parse_func.c\tThu Dec 3 18:39:18 1998\n***************\n*** 1352,1358 ****\n \t\t\t\t\t\tnewexpr->typeOid = funcnode->functype;\n \t\t\t\t\t\tnewexpr->opType = FUNC_EXPR;\n \t\t\t\t\t\tnewexpr->oper = (Node *) funcnode;\n! \t\t\t\t\t\tnewexpr->args = lcons(first_arg, NIL);\n \n \t\t\t\t\t\treturn (Node *) newexpr;\n \t\t\t\t\t}\n--- 1352,1358 ----\n \t\t\t\t\t\tnewexpr->typeOid = funcnode->functype;\n \t\t\t\t\t\tnewexpr->opType = FUNC_EXPR;\n \t\t\t\t\t\tnewexpr->oper = (Node *) funcnode;\n! \t\t\t\t\t\tnewexpr->args = expr->args;\n \n \t\t\t\t\t\treturn (Node *) newexpr;\n \t\t\t\t\t}\n***************\n*** 1359,1366 ****\n \n \t\t\t\t}\n \n- \t\t\t\telog(ERROR, \"Function '%s' has bad return type %d\",\n- \t\t\t\t\t funcname, argtype);\n \t\t\t\tbreak;\n \t\t\t}\n \t\tcase T_Param:\n--- 1359,1364 ----\n",
"msg_date": "Thu, 3 Dec 1998 19:06:07 +0900",
"msg_from": "\"Hiroshi Inoue\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Projecting attributes of function return values "
},
{
"msg_contents": "I have applied this to the CURRENT tree, so the fix will appear in 6.5. \nI have not applied it to the RELEASE tree, so it will not be in 6.4.1. \nNot sure there is going to be enough testing for that.\n\nThanks for the patch.\n\n> \n> Hi all\n> \n> I have a simple patch about the treatment of functions.\n> But it may be self-satisfied. \n> Please check my patch at the end of this posting.\n> \n> Case 1. executor evaluates functions twice\n> \n> create table a (elem int4);\n> create function ax2(int4) returns a as\n> 'select $1*2;' language 'sql';\n> select elem(ax2(1));\n> \n> the result is\n> \n> elem\n> ----\n> 4\n> (1 row)\n> \n> it's wrong ?\n> it must be the following ?\n> \n> elem\n> ----\n> 2\n> (1 row)\n> \n> Case 2.parser rejects functions\n> \n> create function elemout(a) returns int4 as\n> 'select $1.elem;' language 'sql';\n> select elemout(ax2(1));\n> \n> ERROR: Function 'elemout' has bad return type 315169\n> \n> this elog() is necessary ?\n> In my patch I removed it and the result is\n> \n> elemout\n> -------\n> 2\n> (1 row)\n> \n> Thanks\n> \n> Hiroshi Inoue\n> [email protected]\n> \n> *** backend/parser/parse_func.c.orig\tThu Dec 3 18:25:31 1998\n> --- backend/parser/parse_func.c\tThu Dec 3 18:39:18 1998\n> ***************\n> *** 1352,1358 ****\n> \t\t\t\t\t\tnewexpr->typeOid = funcnode->functype;\n> \t\t\t\t\t\tnewexpr->opType = FUNC_EXPR;\n> \t\t\t\t\t\tnewexpr->oper = (Node *) funcnode;\n> ! \t\t\t\t\t\tnewexpr->args = lcons(first_arg, NIL);\n> \n> \t\t\t\t\t\treturn (Node *) newexpr;\n> \t\t\t\t\t}\n> --- 1352,1358 ----\n> \t\t\t\t\t\tnewexpr->typeOid = funcnode->functype;\n> \t\t\t\t\t\tnewexpr->opType = FUNC_EXPR;\n> \t\t\t\t\t\tnewexpr->oper = (Node *) funcnode;\n> ! \t\t\t\t\t\tnewexpr->args = expr->args;\n> \n> \t\t\t\t\t\treturn (Node *) newexpr;\n> \t\t\t\t\t}\n> ***************\n> *** 1359,1366 ****\n> \n> \t\t\t\t}\n> \n> - \t\t\t\telog(ERROR, \"Function '%s' has bad return type %d\",\n> - \t\t\t\t\t funcname, argtype);\n> \t\t\t\tbreak;\n> \t\t\t}\n> \t\tcase T_Param:\n> --- 1359,1364 ----\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, 12 Dec 1998 23:18:55 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Projecting attributes of function return values"
},
{
"msg_contents": ">\n> I have applied this to the CURRENT tree, so the fix will appear in 6.5.\n> I have not applied it to the RELEASE tree, so it will not be in 6.4.1.\n> Not sure there is going to be enough testing for that.\n>\n> Thanks for the patch.\n>\n> >\n> > Hi all\n> >\n> > I have a simple patch about the treatment of functions.\n> > But it may be self-satisfied.\n> > Please check my patch at the end of this posting.\n> >\n> > Case 1. executor evaluates functions twice\n> >\n> > create table a (elem int4);\n> > create function ax2(int4) returns a as\n> > 'select $1*2;' language 'sql';\n\n It addresses a problem I've mentioned some time ago.\n Functions returning sets or complex types have targetlists,\n but these aren't handled correctly in the executor.\n\n Me and Thomas G. Lockhard where discussing another rangetable\n entry type lately, which is in fact a subselect. That might\n cover this problem more completely in the future.\n\n So pls let it out of the REL6_4 tree for now.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 14 Dec 1998 10:38:02 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Projecting attributes of function return values"
},
{
"msg_contents": "> -----Original Message-----\n> From: Jan Wieck [mailto:[email protected]]\n> Sent: Monday, December 14, 1998 6:38 PM\n> To: Bruce Momjian\n> Cc: [email protected]; [email protected]\n> Subject: Re: [HACKERS] Projecting attributes of function return values\n> \n> \n> >\n> > I have applied this to the CURRENT tree, so the fix will appear in 6.5.\n> > I have not applied it to the RELEASE tree, so it will not be in 6.4.1.\n> > Not sure there is going to be enough testing for that.\n> >\n> > Thanks for the patch.\n> >\n> > >\n> > > Hi all\n> > >\n> > > I have a simple patch about the treatment of functions.\n> > > But it may be self-satisfied.\n> > > Please check my patch at the end of this posting.\n> > >\n> > > Case 1. executor evaluates functions twice\n> > >\n> > > create table a (elem int4);\n> > > create function ax2(int4) returns a as\n> > > 'select $1*2;' language 'sql';\n> \n> It addresses a problem I've mentioned some time ago.\n> Functions returning sets or complex types have targetlists,\n> but these aren't handled correctly in the executor.\n>\n\nIt includes the fact that func_tlist of Func node is not used to project \nattributes of function return values except when functions are written \nin 'sql' language ?\n\nand \nthe fact that PL/pgSQL functions return different type(HeapTuple) \nfrom 'sql' functions which return the pointer to TupleTableSlot node ?\n\nMy samples are all written in 'sql' language.\nSeems the cases I mentioned are simple bugs.\n\nThanks.\n\nHiroshi Inoue\[email protected] \n",
"msg_date": "Tue, 15 Dec 1998 08:48:58 +0900",
"msg_from": "\"Hiroshi Inoue\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Projecting attributes of function return values"
}
] |
[
{
"msg_contents": "Hi all,\n\nI'm trying to make parser recognize SELECTs between parenthesis.\nI need to do this work to have m$access working with PostgreSQL.\nThe micro$oft access jet (great wisdom) translates the UNIONs as:\n(SELECT ...) UNION (SELECT ...)\nTo have PostgreSQL understand this syntax I edited gram.y\nand I modified the \"SelectStmt:\" and the \"SubUnion:\" as:\n\nSelectStmt: opt_left_paren SELECT\n opt_unique\n res_target_list2\n result\n from_clause\n where_clause\n group_clause\n having_clause\n opt_right_paren\n union_clause\n sort_clause\n {\n SelectStmt *n = makeNode(SelectStmt);\n n->unique = $3;\n n->targetList = $4;\n n->into = $5;\n n->fromClause = $6;\n n->whereClause = $7;\n n->groupClause = $8;\n n->havingClause = $9;\n n->unionClause = $11;\n n->sortClause = $12;\n $$ = (Node *)n;\n }\n ;\n\nSubUnion: opt_left_paren SELECT opt_unique res_target_list2\n from_clause where_clause\n group_clause having_clause opt_right_paren\n {\n SelectStmt *n = makeNode(SelectStmt);\n n->unique = $3;\n n->unionall = FALSE;\n n->targetList = $4;\n n->fromClause = $5;\n n->whereClause = $6;\n n->groupClause = $7;\n n->havingClause = $8;\n $$ = (Node *)n;\n }\n ;\n\n\nand now with these changes I can specify SELECTs inside () and\nwithout parenthesis as before, I tried also subselects and seems that\nall works well...but when I tried a select with a function which contains\nparenthesis, like SUM(), VERSION() for example, then it doesn't work anymore.\nI'm not very good programing in C.\nIs there anybody that can help me with this question ?\nAny help would be very appreciated.\n\nEXAMPLES:\n\n(select (2-3)*3);\n?column?\n--------\n -3\n(1 row)\n\n(select (132-3)*3) union all ( select ((983+1)/2) );\n?column?\n--------\n 387\n 492\n(2 rows)\n\nselect (132-3)*3 union all select ((983+1)/2);\n?column?\n--------\n 387\n 492\n(2 rows)\n\nselect ename from emp where ename in (\n select ename from emp where ename like 'K%'\n );\nename\n-----\nKING\n(1 row)\n\nselect version();\nERROR: parser: parse error at or near \"(\"\n\n-Jose'-\n\n\n",
"msg_date": "Thu, 3 Dec 1998 13:22:45 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": true,
"msg_subject": "gram.y"
},
{
"msg_contents": "> I'm trying to make parser recognize SELECTs between parenthesis.\n> I need to do this work to have m$access working with PostgreSQL.\n> The micro$oft access jet (great wisdom) translates the UNIONs as:\n> (SELECT ...) UNION (SELECT ...)\n\nOK, just assume I have mentioned that you should ask M$ to support\nstandard syntax. It isn't entirely clear from the BNF definition of\nSQL92 (parens near SELECT are allowed in some places), but I was hard\npressed to see where this was legal syntax.\n\n> To have PostgreSQL understand this syntax I edited gram.y\n> and I modified the \"SelectStmt:\" and the \"SubUnion:\" as:\n> SelectStmt: opt_left_paren SELECT\n<snip>\n\nOne problem with this is the parens must of course be balanced, so\nmaking them individually optional is probably not the right thing to do.\nJust to experiment I tried adding parens on some of the several places\nwhere selects with unions are allowed, and started running into trouble\non the second one I tried. I'm guessing that a yacc-based parser will\nrun into serious trouble, but I haven't thought of a completely fatal\nexample yet.\n\nAlso, the way gram.y is currently laid out it would be difficult to\nensure that all clauses of the UNION are surrounded by parens if one\nclause has those parens. But I think this would be desirable if the\nparens are going to be allowed at all.\n\n - Tom\n",
"msg_date": "Sun, 06 Dec 1998 07:06:04 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] gram.y"
},
{
"msg_contents": "Hello Thomas,\n\ndomenica, 6 dicembre 98, you wrote:\n\n>> I'm trying to make parser recognize SELECTs between parenthesis.\n>> I need to do this work to have m$access working with PostgreSQL.\n>> The micro$oft access jet (great wisdom) translates the UNIONs as:\n>> (SELECT ...) UNION (SELECT ...)\n\nTGL> OK, just assume I have mentioned that you should ask M$ to support\nTGL> standard syntax. It isn't entirely clear from the BNF definition of\nTGL> SQL92 (parens near SELECT are allowed in some places), but I was hard\nTGL> pressed to see where this was legal syntax.\n\nThe problem is: I can't say this syntax is illegal, I can't find\nnothing about it on SQL documentation.\n\nAny way. I think it would be interesting to have parser recognize every\nstatement enclosed into parens.\nI have also Informix-SE on my Linux box, I tried statements enclosed\ninto parens and Informix understand it.\n\n-Jose'-\n\n\n",
"msg_date": "Thu, 10 Dec 1998 14:31:59 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re[2]: [HACKERS] gram.y"
}
] |
[
{
"msg_contents": "I am installing postgres 6.4 on my Solaris 2.5.1 machine. I ran the configure script with\nthe --prefix option set. It found my machine type ok and used the Solaris_sparc_gcc template. I\nmade the binaries with gmake and that completed fine and the install went fine also. I run\nthe initdb script and it seems to complete without any problems. Here is the output from initdb\n\nWe are initializing the database system with username sdbm (uid=1030).\nThis user will own all the files and must also own the server process.\n\nCreating Postgres database system directory /vol3/postgresql-6.4-solaris/data\nCreating Postgres database system directory /vol3/postgresql-6.4-solaris/data/base\nCreating template database in /vol3/postgresql-6.4-solaris/data/base/template1\nCreating global classes in /vol3/postgresql-6.4-solaris/data/base\nAdding template1 database to pg_database...\nVacuuming template1\nCreating public pg_user view\nCreating view pg_rules\nCreating view pg_views\nCreating view pg_tables\nCreating view pg_indexes\nLoading pg_description\n\nThis all looks fine. I start the postmaster process without any problems. Next when I try to do a\ncreatedb as the postgres superuser I get the following:\n> createdb test\nConnection to database 'template1' failed.\nUser authentication failed\ncreatedb: database creation failed on test.\n\nAny idea why this is happening? I am lost. I had version 6.3.2 installed previously and started to\nget the same type of errors. I thought upgrading would help. I have made sure all my paths were\ncorrect and that no other postmasters were running. I also did the initdb in debug mode but I have\nno clue what to look for. Maybe if someone give me ideas what might be wrong there I could look\ninto it more. I am REALLY screwed and need some help.\nThanks\n---------\nChris Williams\nSterling Software\nRome, New York\nPhone: (315) 336-0500\nEmail: [email protected]\n\n",
"msg_date": "Thu, 3 Dec 1998 08:34:13 -0500",
"msg_from": "\"Chris Williams\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Need immediate help with install"
}
] |
[
{
"msg_contents": "Hi all,\n\nSeems that I found a bug on HAVING clause, see attached file.\n\n-Jose'-",
"msg_date": "Thu, 3 Dec 1998 15:01:51 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": true,
"msg_subject": "BUG ON HAVING CLAUSE"
},
{
"msg_contents": "Sferacarta Software wrote:\n> \n> Hi all,\n> \n> Seems that I found a bug on HAVING clause, see attached file.\n> \n\nCould you post me data for 8342 rows ?\n\nVadim\n",
"msg_date": "Fri, 04 Dec 1998 09:53:39 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] BUG ON HAVING CLAUSE"
},
{
"msg_contents": "Hello Vadim,\n\nvenerd�, 4 dicembre 98, you wrote:\n\nVM> Sferacarta Software wrote:\n>> \n>> Hi all,\n>> \n>> Seems that I found a bug on HAVING clause, see attached file.\n>> \n\nVM> Could you post me data for 8342 rows ?\n\nI think this bug is not on HAVING but on IN/ANY/ALL, I tried all these\nstatements and no one returns right values.\nI compared it with Informix, take a look at attachment.\n\nBTW. Thomas said me that you know something about the following\nproblem:\n\nSELECT esito1,esito2\n FROM brogliacci\n WHERE NOT esito1 IS NULL AND NOT esito2 IS NULL;\nesito1|esito2\n------+------\nN |N \n(1 row)\n\n\nSELECT esito1,esito2\n FROM brogliacci\n WHERE NOT esito1 IS NULL OR NOT esito2 IS NULL;\n\npqReadData() -- backend closed the channel unexpectedly.\n This probably means the backend terminated abnormally before or while pr\nocessing the request.\nWe have lost the connection to the backend, so further processing is impossible.\n Terminating.\n\n May you help me?\n\n-Jose'-",
"msg_date": "Fri, 4 Dec 1998 14:36:19 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re[2]: [HACKERS] BUG ON HAVING CLAUSE"
},
{
"msg_contents": "Sferacarta Software wrote:\n> \n> >> Seems that I found a bug on HAVING clause, see attached file.\n> >>\n> \n> VM> Could you post me data for 8342 rows ?\n> \n> I think this bug is not on HAVING but on IN/ANY/ALL, I tried all these\n\nThis is bug on handling HAVING in subqueries.\nThere was no HAVING when I was implementing subqueries and\nso I didn't care... The bug is in optimizer:\n\nvac=> explain select * from test where x in (select * from test group by x having 1 < count(x));\nNOTICE: QUERY PLAN:\n\nSeq Scan on test (cost=0.00 size=0 width=4)\n SubPlan\n -> Aggregate (cost=0.00 size=0 width=0)\n! InitPlan\n! -> Aggregate (cost=0.00 size=0 width=0)\n! -> Seq Scan on test (cost=0.00 size=0 width=4)\n\nThere must be no InitPlan here...\n\n -> Group (cost=0.00 size=0 width=0)\n -> Sort (cost=0.00 size=0 width=0)\n -> Seq Scan on test (cost=0.00 size=0 width=4)\n\nCurrently, my local copy of dev-tree is broken and so I can't\nfix this. I'll return to this bug latter if no one else...\n\nVadim\n",
"msg_date": "Sat, 05 Dec 1998 18:32:20 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] BUG ON HAVING CLAUSE"
},
{
"msg_contents": "Hello Vadim,\n\nsabato, 5 dicembre 98, you wrote:\n\nVM> Sferacarta Software wrote:\n>> \n>> >> Seems that I found a bug on HAVING clause, see attached file.\n>> >>\n>> \n>> VM> Could you post me data for 8342 rows ?\n>> \n>> I think this bug is not on HAVING but on IN/ANY/ALL, I tried all these\n\nVM> This is bug on handling HAVING in subqueries.\nVM> There was no HAVING when I was implementing subqueries and\nVM> so I didn't care... The bug is in optimizer:\n\nWell in such case I can send you my 8342 rows if you want it still.\n\n-Jose'-\n\n\n",
"msg_date": "Wed, 9 Dec 1998 15:05:49 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re[2]: [HACKERS] BUG ON HAVING CLAUSE"
},
{
"msg_contents": "> Hi all,\n> \n> Seems that I found a bug on HAVING clause, see attached file.\n> \n> -Jose'-\n\nCan you summarize what was the result of your several bug reports on IN\nand HAVING. Is there a problem that needs to be added to the TODO list?\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, 12 Dec 1998 23:20:28 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] BUG ON HAVING CLAUSE"
},
{
"msg_contents": "Added to TODO:\n\n\t* subqueries containing HAVING return incorrect results\n\n\n> Sferacarta Software wrote:\n> > \n> > >> Seems that I found a bug on HAVING clause, see attached file.\n> > >>\n> > \n> > VM> Could you post me data for 8342 rows ?\n> > \n> > I think this bug is not on HAVING but on IN/ANY/ALL, I tried all these\n> \n> This is bug on handling HAVING in subqueries.\n> There was no HAVING when I was implementing subqueries and\n> so I didn't care... The bug is in optimizer:\n> \n> vac=> explain select * from test where x in (select * from test group by x having 1 < count(x));\n> NOTICE: QUERY PLAN:\n> \n> Seq Scan on test (cost=0.00 size=0 width=4)\n> SubPlan\n> -> Aggregate (cost=0.00 size=0 width=0)\n> ! InitPlan\n> ! -> Aggregate (cost=0.00 size=0 width=0)\n> ! -> Seq Scan on test (cost=0.00 size=0 width=4)\n> \n> There must be no InitPlan here...\n> \n> -> Group (cost=0.00 size=0 width=0)\n> -> Sort (cost=0.00 size=0 width=0)\n> -> Seq Scan on test (cost=0.00 size=0 width=4)\n> \n> Currently, my local copy of dev-tree is broken and so I can't\n> fix this. I'll return to this bug latter if no one else...\n> \n> Vadim\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, 12 Dec 1998 23:27:11 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] BUG ON HAVING CLAUSE"
},
{
"msg_contents": "Hello Bruce,\n\ndomenica, 13 dicembre 98, you wrote:\n\n>> Hi all,\n>> \n>> Seems that I found a bug on HAVING clause, see attached file.\n>> \n>> -Jose'-\n\nBM> Can you summarize what was the result of your several bug reports on IN\nBM> and HAVING. Is there a problem that needs to be added to the TODO list?\n\nWell, I tried IN-ANY-ALL on a subselect containing an HAVING clause\nlike:\n\nselect * from emp where ename in (\n select ename from emp group by ename having 1 < count(ename)\n );\n\nand I was thought that IN-ANY-ALL were buggy, then I tried this:\n\nselect * from emp where ename IN (\n select ename from emp where deptno=20);\n\nand now I know the bug is only in the HAVING clause.\n\n-Jose'-\n\n\n",
"msg_date": "Mon, 14 Dec 1998 15:21:41 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re[2]: [HACKERS] BUG ON HAVING CLAUSE"
},
{
"msg_contents": "> Well, I tried IN-ANY-ALL on a subselect containing an HAVING clause\n> like:\n> \n> select * from emp where ename in (\n> select ename from emp group by ename having 1 < count(ename)\n> );\n> \n> and I was thought that IN-ANY-ALL were buggy, then I tried this:\n> \n> select * from emp where ename IN (\n> select ename from emp where deptno=20);\n> \n> and now I know the bug is only in the HAVING clause.\n\nOK, that's good. We now have TODO item:\n\n\t* subqueries containing HAVING return incorrect results\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": "Mon, 14 Dec 1998 10:09:40 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Re[2]: [HACKERS] BUG ON HAVING CLAUSE"
}
] |
[
{
"msg_contents": "Hi all,\nSeems there is a bug on IN (subselect)\n\nI'm looking for duplicate rows and I found 2:\n\nselect ename from emp group by ename having 1 < count(ename);\nename\n-----\nALLEN\nBLAKE\n(2 rows)\n\nUsing IN to display all columns of duplicate rows I have the\nfollowing:\n\n\nselect * from emp where ename in (\n select ename from emp group by ename having 1 < count(ename)\n );\nename |empno|job | hiredate|sal |comm|deptno|level| mgr\n------+-----+----------+----------+---------+----+------+-----+----\nALLEN | 7499|SALESMAN |1981-02-20|$1,600.00| 300| 20| 4|7782\nBLAKE | 7698|MANAGER |1981-05-01|$2,850.00| | 30| 3|7782\nJONES | 7900|CLERK |1981-12-03|$950.00 | | 30| 2|7782\nMILLER| 7654|SALESMAN |1981-09-28|$1,250.00| 400| 30| 3|7839\nCLARK | 7844|SALESMAN |1981-09-08|$1,500.00| | 10| 2|7839\nKING | 7521|SALESMAN |1981-02-22|$1,250.00| 500| 10| 1|7782\nALLEN | 7499|SALESMAN |1981-02-20|$1,600.00| 300| 20| 4|7782\nBLAKE | 7698|MANAGER |1981-05-01|$2,850.00| | 30| 3|7782\n(8 rows)\n\n\nbut I expect the following instead:\n\nename |empno|job | hiredate|sal |comm|deptno|level| mgr\n------+-----+----------+----------+---------+----+------+-----+----\nALLEN | 7499|SALESMAN |1981-02-20|$1,600.00| 300| 20| 4|7782\nBLAKE | 7698|MANAGER |1981-05-01|$2,850.00| | 30| 3|7782\nALLEN | 7499|SALESMAN |1981-02-20|$1,600.00| 300| 20| 4|7782\nBLAKE | 7698|MANAGER |1981-05-01|$2,850.00| | 30| 3|7782\n(4 rows)\n\n-Jose'-\n\n\n",
"msg_date": "Thu, 3 Dec 1998 15:35:34 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": true,
"msg_subject": "IN bug"
}
] |
[
{
"msg_contents": "Hi all,\n\nBecause IN is semantically identical to =ANY\nhere the same error using ANY:\n\nselect * from emp where ename =ANY (\n select ename from emp group by ename having 1 < count(ename)\n );\nename |empno|job | hiredate|sal |comm|deptno|level| mgr\n------+-----+----------+----------+---------+----+------+-----+----\nALLEN | 7499|SALESMAN |1981-02-20|$1,600.00| 300| 20| 4|7782\nBLAKE | 7698|MANAGER |1981-05-01|$2,850.00| | 30| 3|7782\nJONES | 7900|CLERK |1981-12-03|$950.00 | | 30| 2|7782\nMILLER| 7654|SALESMAN |1981-09-28|$1,250.00| 400| 30| 3|7839\nCLARK | 7844|SALESMAN |1981-09-08|$1,500.00| | 10| 2|7839\nKING | 7521|SALESMAN |1981-02-22|$1,250.00| 500| 10| 1|7782\nALLEN | 7499|SALESMAN |1981-02-20|$1,600.00| 300| 20| 4|7782\nBLAKE | 7698|MANAGER |1981-05-01|$2,850.00| | 30| 3|7782\n(8 rows)\n\n\nbut I expect the following instead:\n\nename |empno|job | hiredate|sal |comm|deptno|level| mgr\n------+-----+----------+----------+---------+----+------+-----+----\nALLEN | 7499|SALESMAN |1981-02-20|$1,600.00| 300| 20| 4|7782\nBLAKE | 7698|MANAGER |1981-05-01|$2,850.00| | 30| 3|7782\nALLEN | 7499|SALESMAN |1981-02-20|$1,600.00| 300| 20| 4|7782\nBLAKE | 7698|MANAGER |1981-05-01|$2,850.00| | 30| 3|7782\n(4 rows)\n\n-Jose'-\n\n\n",
"msg_date": "Thu, 3 Dec 1998 16:06:36 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": true,
"msg_subject": "=ANY bug"
},
{
"msg_contents": "> Hi all,\n> \n> Because IN is semantically identical to =ANY\n> here the same error using ANY:\n> \n> select * from emp where ename =ANY (\n> select ename from emp group by ename having 1 < count(ename)\n> );\n\nCan I just say that using HAVING inside a subquery is buggy? Vadim, was\nthis already fixed?\n\n\n-- \n Bruce Momjian | http://www.op.net/~candle\n [email protected] | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n",
"msg_date": "Sat, 12 Dec 1998 23:21:43 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] =ANY bug"
}
] |
[
{
"msg_contents": "Had some problems with the mail so I send this message again, sorry if\nit comes twice !\n\n--------------------------------------\n\nFound today two ugly bugs in pg_dump :\n\nEnvironment : Linux i386 RedHat 5.2, 2.0.36 kernel, Pentium 233 MMX\nPg version : PostgreSQL 6.4\n\n1. It's easy reproductible :\n\n(login as user teo, createdb rights granted)\n$ createdb test\n$ psql test\npsql=> create table people (id int4, name text);\npsql=> grant select on people to teo;\npsql=> \\q\n$ pg_dump -z test\n\\connect - teo\nCREATE TABLE \"people\" (\n \"id\" \"int4\",\n \"name\" \"text\");\nREVOKE ALL on \"people\" from PUBLIC;\nGRANT SELECT on \"people\" to \"people\";\n...\n\nThe error is on the last line : grant select on people to PEOPLE ? Not\nto teo ?\n\nMy pg_hba.conf is :\nlocal all trust\nhost all 127.0.0.1 255.255.255.255 trust\nhost all 133.210.115.4 255.255.255.255 password\nhost all 133.210.115.9 255.255.255.255 password \n\n---------------------------------------------------\n\n2. Got a medium size database (640 Kb dumped) that I have recently moved\nto Pg 6.4. After grant-ing and revoke-ing rights to various people,\ndumping with :\n\npg_dump -z showroom >showroom.dmp\n\nis dumping also core :-)\n\nLooking into the showroom.dmp I noticed that it is dumping core when it\nshould start dumping data from tables.\nTable structures (also table rights) , functions and sequences are\ndumped corectly, when it should begin with table data, it dumpes core.\npg_dump without -z is working fine, database showroom is not broken,\neverything is ok.\n\n\nAll the best and happy bug hunting !\n\nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Thu, 03 Dec 1998 20:32:55 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "Two ugly bugs in pg_dump :-("
}
] |
[
{
"msg_contents": "Hello!\n\n The following gave me wrong result:\nxxx=> select datetime(current_date, '11:00');\ndatetime \n----------------------------\nThu 03 Dec 17:00:00 1998 MSK\n(1 row)\n\n and I started investigation. Soon I found I only have the problem on\nglibc2-based linuxes (RedHat 5.1 and Debian 2.0), RedHat 4.2 and Solaris\nare Ok.\n Well, it looked like an error in glibc2, and I continued. I wrote the\nprogram:\n-----\n#include <stdio.h>\n#include <stdlib.h>\n#include <time.h>\n\n\nint main(int argc, char **argv) {\n char strf_buf[100];\n time_t today_t;\n struct tm * today;\n\n memset(strf_buf, 0, 99);\n today_t = time(NULL);\n today = localtime(&today_t);\n strftime(strf_buf, 100, \"%Z\", today);\n printf(\"%s\\n\", strf_buf);\n\n /* In Moscow I expect MSK/MSD/1/10800 */\n printf(\"%s\\n%s\\n%d\\n%ld\\n\", tzname[0], tzname[1], daylight, timezone);\n\n return 0;\n}\n-----\n\n I expected to have results:\n-----\nMSK\nMSK\nMSD\n1\n-10800\n-----\n\n Ok, I've got it as expected on RH4.2 and Solaris. RH5.1 and Debian2.0\ngave wrong results (although 1st line was Ok: MSK - produced by strftime).\n\n Continued, I found to my big surprize that Python programs do date/time\narithmetic right an all platforms. I looked into python's timemodule.c and\nfound some code with ifdef's. When I looked into config.h, I found that on\n\"good\" platforms there is HAVE_TZNAME defined, but on \"bad\" platforms it is\nundefined. Because of this definition \"bad\" platforms execute the following\ncode:\n#else /* !HAVE_TZNAME */\n#if HAVE_TM_ZONE\n {\n#define YEAR ((time_t)((365 * 24 + 6) * 3600))\n time_t t;\n struct tm *p;\n long winterzone, summerzone;\n char wintername[10], summername[10];\n /* XXX This won't work on the southern hemisphere.\n XXX Anybody got a better idea? */\n t = (time((time_t *)0) / YEAR) * YEAR;\n p = localtime(&t);\n winterzone = -p->tm_gmtoff;\n strncpy(wintername, p->tm_zone ? p->tm_zone : \" \", 9);\n wintername[9] = '\\0';\n t += YEAR/2;\n p = localtime(&t);\n summerzone = -p->tm_gmtoff;\n strncpy(summername, p->tm_zone ? p->tm_zone : \" \", 9);\n summername[9] = '\\0';\n ins(d, \"timezone\", PyInt_FromLong(winterzone));\n ins(d, \"altzone\", PyInt_FromLong(summerzone));\n ins(d, \"daylight\",\n PyInt_FromLong((long)(winterzone != summerzone)));\n ins(d, \"tzname\",\n Py_BuildValue(\"(zz)\", wintername, summername));\n }\n#else\n\n WOW!!! Look, look here - how tzname and timezone calculated! (Actually,\nrecalculated - glibc2 already has tzname/timezone, but initialized to wrong\nvalues.)\n\n I am pretty sure date/time arithmetic on Postgres should be changed\naccordingly. Where can I start in postgres sources? Anyone to help?\n\nOleg.\n---- \n Oleg Broytmann http://members.tripod.com/~phd2/ [email protected]\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Thu, 3 Dec 1998 21:59:42 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Date/time on glibc2 linux"
},
{
"msg_contents": "> The following gave me wrong result:\n> I am pretty sure date/time arithmetic on Postgres should be changed\n> accordingly. Where can I start in postgres sources? Anyone to help?\n\nI see the same symptom on my RH5.1 (glibc2) box here:\n\nlockhart=> select datetime('today'::date);\ndatetime\n----------------------------\nFri Dec 04 06:00:00 1998 MSK\n(1 row)\n\nWhereas my RH4.2 (libc5) box gets it right:\n\npostgres=> select datetime('today'::date);\ndatetime\n----------------------------\nFri Dec 04 00:00:00 1998 MSK\n(1 row)\n\nThis isn't specific to your timezone; I see the same problem with\nPST8PST. The routine actually getting invoked is date_datetime() in\nbackend/utils/adt/datetime.c. I would be reluctant to scatter\nglibc2-specific bug-fix code throughout the date/time files (the issue\nhas come up before), but once we've identified the actual cause of the\nproblem we can consider a workaround.\n\nTo help look at the problem do a \"make clean\" in that directory, and\nmodify the Makefile to add to CFLAGS the argument \"-DDATEDEBUG\" which\nwill enable some print statements.\n\nIf it is a glibc2 bug then presumably someone will be applying a fix to\nthat fairly soon. If that is the case perhaps we can either\nautomatically test for the deficiency in configure or we can post the\nfix as a patch rather than putting it into the main source tree.\n\nbtw, can you see a problem in the regression tests? I wonder if there\nare other date/time routines which show problems. I don't usually run\nPostgres regression tests on my RH5.1 box since that is at work...\n\n - Tom\n\nRH4.2 (libc-5.3.12):\n\npostgres=> select datetime(current_date, '11:00');\ndatetime\n----------------------------\nFri Dec 04 11:00:00 1998 MSK\n(1 row)\n\npostgres=> select 'now'::datetime;\n?column?\n----------------------------\nFri Dec 04 09:18:43 1998 MSK\n(1 row)\n\nRH5.1 (glibc-2.0.7):\n\nlockhart=> select datetime(current_date, '11:00');\ndatetime\n----------------------------\nFri Dec 04 17:00:00 1998 MSK\n(1 row)\n\nlockhart=> select 'now'::datetime;\n?column?\n----------------------------\nFri Dec 04 09:16:21 1998 MSK\n(1 row)\n",
"msg_date": "Fri, 04 Dec 1998 06:45:41 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "On Thu, Dec 03, 1998 at 09:59:42PM +0300, Oleg Broytmann wrote:\n> Well, it looked like an error in glibc2, and I continued. I wrote the\n> program:\n> ...\n> \n> I expected to have results:\n> -----\n> MSK\n> MSK\n> MSD\n> 1\n> -10800\n\nAnd what did you get? I get this on Debian 2.1:\n\n����\u0001\b\nMEZ\nCEST\n1\n18000\n\nMichael\n-- \nDr. Michael Meskes, Manager of the Western Branch Office, Datenrevision GmbH\nwork: Cuxhavener Str. 36, D-21149 Hamburg, [email protected]\nhome: Th.-Heuss-Str. 61, D-41812 Erkelenz, [email protected]\nGo SF49ers! Go Rhein Fire! Use Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Fri, 4 Dec 1998 16:04:15 +0100",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hello!\n\nOn Fri, 4 Dec 1998, Thomas G. Lockhart wrote:\n> This isn't specific to your timezone; I see the same problem with\n> PST8PST. The routine actually getting invoked is date_datetime() in\n> backend/utils/adt/datetime.c. I would be reluctant to scatter\n> glibc2-specific bug-fix code throughout the date/time files (the issue\n> has come up before), but once we've identified the actual cause of the\n> problem we can consider a workaround.\n\n I was pretty sure the problem is not with my timezone. The actual cause\nof the problem is very simple - glibc2 has invalid timezone/tzname even\nafter tzset(). To my surprise, strftime() works right! It looks like\nglibc2 has two different tzname's - public and internal. Because of this\nALL date/time arithmetic will fail sooner or later.\n The code I stole from python and displayed yesterday works around the\nproblem. It seems python's configure can safely detect this and avoid. I\nasked python author about this. Waiting for replay...\n\n> btw, can you see a problem in the regression tests? I wonder if there\n> are other date/time routines which show problems. I don't usually run\n> Postgres regression tests on my RH5.1 box since that is at work...\n\n Regression test passed Ok on datetime, abstime, reltime, tinterval tests\non Debian 2.0. That is - the error is not triggered. Should we change the\nregression test?\n\nOleg.\n---- \n Oleg Broytmann National Research Surgery Centre http://sun.med.ru/~phd/\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Fri, 4 Dec 1998 19:06:13 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "That Python code is a huge hunk of workaround kludge. I haven't looked\nto see how Postgres configures itself on the glibc2 box, but we should\nfirst check (if you haven't done so already) that it is getting the\nright choices for tz configuration. \n\nMy recollection is that the two possible Unix-style setups are:\n\n1) use global variables for time zone offsets and names. This can't work\nwith re-entrant or thread-safe libraries such as glibc2.\n\n2) use variables embedded in the tz structure. I had thought that this\nwas an older style, but is clearly required for thread-safe code.\n\nWe should confirm that our glibc2 machines are being built with option\n(2), that they don't implement a third option, and that it is indeed\nmassively broken. Also, we should characterize under what conditions it\nis broken; so far the only case we have is when converting date to\ndatetime, so the problem may be within that Postgres code rather than a\ngeneral problem.\n\n> Regression test passed Ok on datetime, abstime, reltime, tinterval \n> tests on Debian 2.0. That is - the error is not triggered. Should we \n> change the regression test?\n\nSure we should. Send patches...\n\n - Tom\n",
"msg_date": "Fri, 04 Dec 1998 16:34:30 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "On Fri, 4 Dec 1998, Michael Meskes wrote:\n\n> On Thu, Dec 03, 1998 at 09:59:42PM +0300, Oleg Broytmann wrote:\n> > Well, it looked like an error in glibc2, and I continued. I wrote the\n> > program:\n> > ...\n> > \n> > I expected to have results:\n> > -----\n> > MSK\n> > MSK\n> > MSD\n> > 1\n> > -10800\n> \n> And what did you get? I get this on Debian 2.1:\n> \n> О©╫О©╫О©╫О©╫\u0001\b\n> MEZ\n> CEST\n> 1\n> 18000\n\n Definetely wrong! The first 2 lines must be identical - they are\nexpexted to be your timezone (MSK, in my case).\n I got different results on different glibc2-based linux systems, but all\nwas wrong. On Debian 2.0, e.g.:\n-----\nMSK\nEET\nEST\n-9200\n-----\n\n MSK is Ok here (it is output from strftime()), but other lines (global\nvariables) are wrong.\n Thomas Lockhart gave an idea that global variables cannot be used at all\nin thread-safe library, so the problem is - how to detect glibc2 at\nconfigure stage, and how to patch date/time arithmetic in postgres. I am\nworking on it...\n\nOleg.\n---- \n Oleg Broytmann National Research Surgery Centre http://sun.med.ru/~phd/\n Programmers don't die, they just GOSUB without RETURN.\n\n\n",
"msg_date": "Sat, 5 Dec 1998 16:40:02 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hello!\n\nOn Fri, 4 Dec 1998, Thomas G. Lockhart wrote:\n> > Regression test passed Ok on datetime, abstime, reltime, tinterval \n> > tests on Debian 2.0. That is - the error is not triggered. Should we \n> > change the regression test?\n> \n> Sure we should. Send patches...\n\n I have problems creating the patch. One of my systems (Linux) returns\nThu 01 Oct 11:00:00 1998 PST\n where another (Solaris) reports\nThu Oct 01 11:00:00 1998 PST\n\n With this, datetime regression test always failed. It's minor\ndifference, but difference anyway.\n\n I am always run postgres with -o -e\n\nOleg.\n---- \n Oleg Broytmann http://members.tripod.com/~phd2/ [email protected]\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Mon, 7 Dec 1998 20:40:31 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Oleg Broytmann wrote:\n> \n> Hello!\n> \n> On Fri, 4 Dec 1998, Thomas G. Lockhart wrote:\n> > > Regression test passed Ok on datetime, abstime, reltime, tinterval\n> > > tests on Debian 2.0. That is - the error is not triggered. Should \n> > > we change the regression test?\n> > Sure we should. Send patches...\n> I have problems creating the patch. One of my systems (Linux) returns\n> Thu 01 Oct 11:00:00 1998 PST\n> where another (Solaris) reports\n> Thu Oct 01 11:00:00 1998 PST\n> I am always run postgres with -o -e\n\nUh, these differences are coming from one of your systems running on\nUS/NonEuropean date style and the other running with the Europena style\nper your \"-o -e\" flags above. The Postgres date/time output routines do\nnot have system-specific formatting differences, other than the\ndifferences lower down between the two main flavors of Unix libc\nsupport.\n\nExample follows. Talk to you soon.\n\n - Tom\n\npostgres=> set DateStyle = 'Postgres,NonEuropean';\nSET VARIABLE\npostgres=> select 'now'::datetime;\n?column?\n----------------------------\nTue Dec 08 04:16:28 1998 GMT\n(1 row)\n\npostgres=> set DateStyle = 'Postgres,European';\nSET VARIABLE\npostgres=> select 'now'::datetime;\n?column?\n----------------------------\nTue 08 Dec 04:16:46 1998 GMT\n(1 row)\n",
"msg_date": "Tue, 08 Dec 1998 04:19:10 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hi!\n\nOn Fri, 4 Dec 1998, Thomas G. Lockhart wrote:\n> To help look at the problem do a \"make clean\" in that directory, and\n> modify the Makefile to add to CFLAGS the argument \"-DDATEDEBUG\" which\n> will enable some print statements.\n\n Where do the output goes? To postmaster's stdout, I guess?\n\nOleg.\n---- \n Oleg Broytmann National Research Surgery Centre http://sun.med.ru/~phd/\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Tue, 8 Dec 1998 15:59:33 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "> > To help look at the problem do a \"make clean\" in that directory, and\n> > modify the Makefile to add to CFLAGS the argument \"-DDATEDEBUG\" \n> > which will enable some print statements.\n> Where do the output goes? To postmaster's stdout, I guess?\n\nYes. I always have X up on my systems, and for debugging usually run\npostmaster from the command line on one terminal while running psql from\nanother.\n\nRecently, Bruce et al got me going with gdb directly on the Postgres\nbackend, and that works very well also. You can do the following:\n\n$ gdb postgres\n<snip>\n(gdb) b datetime_out\nBreakpoint 1 at 0x80ec6bc\n(gdb) run -d 99\n<snip>\n> select 'now'::datetime;\n<snip>\nBreakpoint 1, 0x80ec6bc in datetime_out ()\n(gdb) \n\nand you can then step through from there. Note that you should recompile\nwith \n CFLAGS+= -O0 -g\nto allow the debugger to grok the source code.\n\n - Tom\n",
"msg_date": "Tue, 08 Dec 1998 14:21:32 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hello!\n\n I think, finally, that the simplest way to solve this would be patch\nconfigure, not datetime.c, dt.c and all that.\n The struct tm on linux has tm_gmtoff and tm_zone, so it will be enough\nif postgres' configure undefine HAVE_TZSET and HAVE_INT_TIMEZONE on glibc2.\n\n Python's configure.in has the following:\n# checks for structures\nAC_HEADER_TIME\nAC_STRUCT_TM\nAC_STRUCT_TIMEZONE\n and this do all neccessary defines. I think postgres' configure will\nrequire some more tricks.\n\n Continue investigating...\n\nOleg.\n---- \n Oleg Broytmann http://members.tripod.com/~phd2/ [email protected]\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Wed, 9 Dec 1998 18:16:01 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hi!\n\n It looks I am doing something wrong. The file\n.../src/include/port/linux.h already has it:\n\n#if (__GLIBC__ >= 2)\n#ifdef HAVE_INT_TIMEZONE\n#undef HAVE_INT_TIMEZONE\n#endif\n\n but it isn't working. It is all that needs to solve the timezone\nproblem.\n\nOleg.\n---- \n Oleg Broytmann http://members.tripod.com/~phd2/ [email protected]\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Wed, 9 Dec 1998 18:48:47 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hi!\n\nOn Wed, 9 Dec 1998, Oleg Broytmann wrote:\n> #if (__GLIBC__ >= 2)\n> #ifdef HAVE_INT_TIMEZONE\n> #undef HAVE_INT_TIMEZONE\n> #endif\n> \n> but it isn't working.\n\n Sorry, my fault - it is working.\n\n Another problem I have - I cannot recompile postgres with DATEDEBUG\ndefined in config.h - I got the error:\n\ngcc -I../../include -I../../backend -O2 -Wall -Wmissing-prototypes -I..\n-c\nfmgrtab.c -o fmgrtab.o\nmake[2]: *** No rule to make target `adt/SUBSYS.o', needed by `SUBSYS.o'.\nStop.\nmake[2]: Leaving directory\n`/usr/local/src/PostgreSQL/postgresql-v6.4/src/backend\n/utils'\nmake[1]: *** [utils.dir] Error 2\nmake[1]: Leaving directory\n`/usr/local/src/PostgreSQL/postgresql-v6.4/src/backend\n'\nmake: *** [all] Error 2\n\nOleg.\n---- \n Oleg Broytmann http://members.tripod.com/~phd2/ [email protected]\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Wed, 9 Dec 1998 20:38:12 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "> Another problem I have - I cannot recompile postgres with DATEDEBUG\n> defined in config.h - I got the error:\n<snip>\n\nThese don't appear to be related. However, I usually just (slightly)\nedit the src/backend/utils/adt/Makefile to add in\n CFLAGS+= -DDATEDEBUG\n\nsince it is a symbol which afaik only affects files in that directory,\nand should not be kept in after you are finished.\n\nDo you have any test cases *other* than the date->datetime conversion\nexample? If you don't then we should probably be looking for a very\nlocal problem, not a system-wide problem. otoh I have a hard time\npicturing how this could not be a system-wide problem, given that I see\nno problems on the same test case on gcc/libc5 systems.\n\n - Tom\n",
"msg_date": "Wed, 09 Dec 1998 18:25:02 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hi!\n\nOn Wed, 9 Dec 1998, Thomas G. Lockhart wrote:\n\n> > Another problem I have - I cannot recompile postgres with DATEDEBUG\n> > defined in config.h - I got the error:\n> <snip>\n> \n> These don't appear to be related. However, I usually just (slightly)\n\n I am sure this is related. When I edited config.h and commented out\nDATEDEBUG the sources compiled just fine.\n\n> edit the src/backend/utils/adt/Makefile to add in\n> CFLAGS+= -DDATEDEBUG\n\n Well, I'll try this.\n\n> since it is a symbol which afaik only affects files in that directory,\n> and should not be kept in after you are finished.\n\n> Do you have any test cases *other* than the date->datetime conversion\n> example? If you don't then we should probably be looking for a very\n> local problem, not a system-wide problem. otoh I have a hard time\n> picturing how this could not be a system-wide problem, given that I see\n> no problems on the same test case on gcc/libc5 systems.\n\n Any use for daylight or tzname or timezone (global vars) produces\nincorrect results. Again, test program is very simple:\n\n#include <stdio.h>\n#include <stdlib.h>\n#include <time.h>\n\n\nint main(int argc, char **argv) {\n char strf_buf[100];\n time_t today_t;\n struct tm * today;\n\n memset(strf_buf, 0, 99);\n today_t = time(NULL);\n today = localtime(&today_t);\n strftime(strf_buf, 100, \"%Z\", today);\n printf(\"%s\\n\", strf_buf);\n\n /* In Moscow I expect MSK/MSD/1/10800 */\n printf(\"%s\\n%s\\n%d\\n%ld\\n\", tzname[0], tzname[1], daylight, timezone);\n\n return 0;\n}\n\n It looks like glibc2 defines these vars incorrectly. Correct values are\nin struct tm (including tmzone and gmtoff).\n Do you think it is local problem? I am pretty sure it is \"system-wide\".\n\nOleg.\n---- \n Oleg Broytmann National Research Surgery Centre http://sun.med.ru/~phd/\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Wed, 9 Dec 1998 21:33:28 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "> I am sure this is related. When I edited config.h and commented out\n> DATEDEBUG the sources compiled just fine.\n\nSo send my your config.h if you want someone to look at it.\n\n> Any use for daylight or tzname or timezone (global vars) produces\n> incorrect results.\n> It looks like glibc2 defines these vars incorrectly. Correct values \n> are in struct tm (including tmzone and gmtoff).\n> Do you think it is local problem? I am pretty sure it is \n> \"system-wide\".\n\nThe glibc2 is a thread-safe library, and I would expect that the *only*\nplace with reliable timezone info is in the tm structure. Global\nvariables are not to be trusted since they are not available in a\nreentrant way.\n\nIf the tm structure contains the timezone info (as it claims to on my\nRH5.1 glibc2 system) then for testing try to #undef HAVE_INT_TIMEZONE in\nconfig.h and see how it goes. If I have a chance tomorrow I'll try doing\nthe same at work. I'm guessing that our configure tests look for the\nglobal variable version first, and that the glibc2 passes that test even\nthough the other mechanism is the right one.\n\n - Tom\n",
"msg_date": "Thu, 10 Dec 1998 05:45:53 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hello!\n\nOn Thu, 10 Dec 1998, Thomas G. Lockhart wrote:\n\n> > I am sure this is related. When I edited config.h and commented out\n> > DATEDEBUG the sources compiled just fine.\n> \n> So send my your config.h if you want someone to look at it.\n\n I've solved this issue. I uncomment it to\n#define DATEDEBUG\n but forget to add definition. After writing\n#define DATEDEBUG 1\n postgres compiled.\n\n> > Any use for daylight or tzname or timezone (global vars) produces\n> > incorrect results.\n> > It looks like glibc2 defines these vars incorrectly. Correct values \n> > are in struct tm (including tmzone and gmtoff).\n> > Do you think it is local problem? I am pretty sure it is \n> > \"system-wide\".\n> \n> The glibc2 is a thread-safe library, and I would expect that the *only*\n> place with reliable timezone info is in the tm structure. Global\n> variables are not to be trusted since they are not available in a\n> reentrant way.\n\n Yes, I understood the idea after you explained this for a first time.\n\n> If the tm structure contains the timezone info (as it claims to on my\n> RH5.1 glibc2 system) then for testing try to #undef HAVE_INT_TIMEZONE in\n> config.h and see how it goes. If I have a chance tomorrow I'll try doing\n> the same at work. I'm guessing that our configure tests look for the\n> global variable version first, and that the glibc2 passes that test even\n> though the other mechanism is the right one.\n\n#undef HAVE_INT_TIMEZONE in config.h replaced by #define HAVE_INT_TIMEZONE 1\nafter configure.\n On the other hand, os.h (linked to port/linux.h) undefine it. I need to\ntest that it is still undefined when compiling datetime.c (and dt.c and all\nthat) - there are all provisions to use tm.tz_name/tm_gmtoff, compiler just\nneed right defines.\n\nOleg.\n---- \n Oleg Broytmann National Research Surgery Centre http://sun.med.ru/~phd/\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Thu, 10 Dec 1998 11:35:49 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hi!\n\n I run postgres with DATEDEBUG and got:\n\nGetCurrentAbsoluteTime- timezone is MSK -> -10800 seconds from UTC\ndate_in- input string is 1991-10-01\nParseDateTime- input string is 1991-10-01\nParseDateTime- set field[0] to 1991-10-01 type 2\nDecodeDateTime- field[0] is 1991-10-01 (type 2)\nDecodeNumber- 1991 is 1991 fmask=00000000 tmask=00000000\nDecodeNumber- match 1991 (1991) as year\nDecodeNumber- 10 is 10 fmask=00000004 tmask=00000000\nDecodeNumber- match 10 (10) as month\nDecodeNumber- 01 is 1 fmask=00000006 tmask=00000000\nDecodeNumber- (2) match 1 (01) as day\nDecodeDateTime- field[0] 1991 (00000000/0000000e) value is 1\nDecodeDateTime- mask 0000000e (0000000e) set y1991 m10 d01 00:00:00\nParseDateTime- input string is 11:00\nParseDateTime- set field[0] to 11:00 type 3\nDecodeTimeOnly- field[0] is 11:00 (type 3)\nDecodeTimeOnly- field[0] 11:00 value is -1073763148\nDecodeTimeOnly- mask 00001c0e (00001c00) 11:00:00 (0.000000)\ndate_datetime- date is 1991.10.01\ndate_datetime- time is 00:00:00 0.0000000\ntm2datetime- date is -260409600.000000 (-3014.000000 0.000000 0)\ntm2datetime- time is 0.000000 00:00:00 0.000000\ndatetime2tm- date is -260362800.000000 (2448531.000000 46800.000000)\ndatetime2tm- date is 1991.10.01\ndatetime2tm- time is 13:00:00\ndatetime2tm- time is 13:00:00 0.0000000\ndatetime2tm- (localtime) 91.09.01 15:00:00 EET dst=0\ndatetime2tm- date is 1991.10.01\ndatetime2tm- time is 15:00:00 0.0000000\nEncodeDateTime- timezone is EET (EET); offset is -7200 (-10800); daylight is 0 (0\nEncodeDateTime- day is 2448531\nEncodeDateTime- date result is Tue 01 Oct 15:00:00 1991 EET\n\n The last 3 lines shows an error. Continue searching...\n\nOleg.\n---- \n Oleg Broytmann National Research Surgery Centre http://sun.med.ru/~phd/\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Thu, 10 Dec 1998 11:57:21 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "\"Thomas G. Lockhart\" <[email protected]> writes:\n> If the tm structure contains the timezone info (as it claims to on my\n> RH5.1 glibc2 system) then for testing try to #undef HAVE_INT_TIMEZONE in\n> config.h and see how it goes. If I have a chance tomorrow I'll try doing\n> the same at work. I'm guessing that our configure tests look for the\n> global variable version first, and that the glibc2 passes that test even\n> though the other mechanism is the right one.\n\nYes, the autoconf test just checks for a global variable \"timezone\"\ndeclared in <time.h>. It'd probably be better to reverse the test to\ncheck for the timezone fields in a struct tm. I don't much care for\nthe name of the configuration symbol either ... maybe change it to\nHAVE_STRUCT_TM_TZ_FIELDS ?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 10 Dec 1998 10:48:41 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux "
},
{
"msg_contents": "Hi!\n\nOn Thu, 10 Dec 1998, Tom Lane wrote:\n> Yes, the autoconf test just checks for a global variable \"timezone\"\n> declared in <time.h>. It'd probably be better to reverse the test to\n> check for the timezone fields in a struct tm. I don't much care for\n> the name of the configuration symbol either ... maybe change it to\n> HAVE_STRUCT_TM_TZ_FIELDS ?\n\n It is enought to add AC_STRUCT_TIMEZONE to configure.in - and after it\nconfigure will #define HAVE_TM_ZONE 1 in config.h.\n I already did it - just haven't submitted the patch.\n\nOleg.\n---- \n Oleg Broytmann http://members.tripod.com/~phd2/ [email protected]\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Thu, 10 Dec 1998 19:34:12 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux "
},
{
"msg_contents": "> Yes, the autoconf test just checks for a global variable \"timezone\"\n> declared in <time.h>. It'd probably be better to reverse the test to\n> check for the timezone fields in a struct tm. I don't much care for\n> the name of the configuration symbol either ... maybe change it to\n> HAVE_STRUCT_TM_TZ_FIELDS ?\n\nCould we have a longer name please? :)\n\nThat would be fine. How about if we continue to test for the other too\nand then the code can choose which one to enable. I can flip around the\nsource blocks to get the right sense. I think this may help the Alpha\nport also, since I vaguely recall that Dec/Compaq (*sigh*) ran afoul of\nour tests too. And although the BSD-style TZ-in-tm seemed to be old\nfashioned, with reentrant/threaded code it is the wave of the future so\ncould/should be the first choice for the code to consider.\n\n - Tom\n",
"msg_date": "Thu, 10 Dec 1998 16:41:43 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "\"Thomas G. Lockhart\" <[email protected]> writes:\n> ... How about if we continue to test for the other too\n> and then the code can choose which one to enable. I can flip around the\n> source blocks to get the right sense. I think this may help the Alpha\n> port also, since I vaguely recall that Dec/Compaq (*sigh*) ran afoul of\n> our tests too. And although the BSD-style TZ-in-tm seemed to be old\n> fashioned, with reentrant/threaded code it is the wave of the future so\n> could/should be the first choice for the code to consider.\n\nI had some other fixes to make in configure.in today, so I went ahead\nand added a call to AC_STRUCT_TIMEZONE. So the configuration symbol\nHAVE_TM_ZONE is now available for use. You probably know better than\nI what parts of the code have to be touched to prefer using the tm_zone\nfield of struct tm over the global timezone variable, so I'll leave that\npart of the repair to you...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 13 Dec 1998 15:15:34 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux "
},
{
"msg_contents": "> I had some other fixes to make in configure.in today, so I went ahead\n> and added a call to AC_STRUCT_TIMEZONE. So the configuration symbol\n> HAVE_TM_ZONE is now available for use. You probably know better than\n> I what parts of the code have to be touched to prefer using the \n> tm_zone field of struct tm over the global timezone variable, so I'll \n> leave that part of the repair to you...\n\nOK, I'm pretty sure it's only one or two places (yeah for modular code\n:). So, should I make that change in both trees, or just in the\ndevelopment tree? My recommendation is to be *really* conservative and\nnot change it around for v6.4.1, since there may be a supported platform\nwhich doesn't conform to our assumptions. But I don't have a really\nstrong feeling about it, so if people other than Alpha and glibc2 folks\nthink it might be OK then...\n\n - Tom\n",
"msg_date": "Sun, 13 Dec 1998 21:06:58 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "\"Thomas G. Lockhart\" <[email protected]> writes:\n> ... So, should I make that change in both trees, or just in the\n> development tree? My recommendation is to be *really* conservative and\n> not change it around for v6.4.1, since there may be a supported platform\n> which doesn't conform to our assumptions.\n\nWell, we have a bug report against 6.4, so my inclination is to change\nit in both trees. We know that the current code is broken on at least\none major platform. I think that outweighs the unproven possibility\nthat the other way might be broken somewhere...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 13 Dec 1998 16:45:54 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux "
},
{
"msg_contents": "> Well, we have a bug report against 6.4, so my inclination is to change\n> it in both trees. We know that the current code is broken on at least\n> one major platform. I think that outweighs the unproven possibility\n> that the other way might be broken somewhere...\n\nOK, I'm not arguing strongly against it, but imho the current code is\nnot broken on any major *mature* platform :) The glibc2 stuff is fairly\nnew, and everyone is still working out the details. We apparently\nunderstand the problem, but we haven't seen Oleg's patches to know what\nhe actually did, whether the patches we are planning fix his problem, or\nwhat the effects are on machines (even of the same type) outside of his\nown.\n\nAnd remember, I'm actually running a linux RH/glibc2 machine at work, so\nI'm not just giving it a hard time for sport (hi scrappy!) ;) How about\nif I test some patches in the main tree on my work machine before we\ncommit to the v6.4.x tree?\n\n - Tom\n",
"msg_date": "Sun, 13 Dec 1998 22:03:18 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "> Well, we have a bug report against 6.4, so my inclination is to change\n> it in both trees. We know that the current code is broken on at least\n> one major platform. I think that outweighs the unproven possibility\n> that the other way might be broken somewhere...\n\nOK, I applied the fix (confined to backend/utils/adt/nabstime.c) to both\nCVS trees. This swaps the checking of HAVE_INT_TIMEZONE and HAVE_TM_ZONE\nso that the TM_ZONE takes precedence.\n\nThis single-file patch should be a candidate for removal if others run\ninto trouble when testing.\n\nOne thing I wasn't certain about was the use of #error in non-gcc\ncompilers. Is that a standard feature?\n\n - Tom\n",
"msg_date": "Mon, 14 Dec 1998 00:32:45 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "\"Thomas G. Lockhart\" <[email protected]> writes:\n> One thing I wasn't certain about was the use of #error in non-gcc\n> compilers. Is that a standard feature?\n\n#error was added by ANSI C. We might see some problems with it in\nold barely-ANSI-compatible compilers. I'd leave it there for now,\nbut keep an eye out for problems...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 13 Dec 1998 19:56:39 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux "
},
{
"msg_contents": "Hello!\n\nOn Sun, 13 Dec 1998, Thomas G. Lockhart wrote:\n> OK, I'm not arguing strongly against it, but imho the current code is\n> not broken on any major *mature* platform :) The glibc2 stuff is fairly\n> new, and everyone is still working out the details. We apparently\n> understand the problem, but we haven't seen Oleg's patches to know what\n> he actually did, whether the patches we are planning fix his problem, or\n> what the effects are on machines (even of the same type) outside of his\n> own.\n\n I haven't worked out a real patch yet. I did a little patch for\nconfigure.in/config.h.in, but you already committed the same patch. :)\n I did a patch for datetime regression test (attached). The patch\ntriggers an error on glibc2. Works right on Solaris. I am not sure whether\nwe should include the patch into regression test...\n\nOleg.\n---- \n Oleg Broytmann http://members.tripod.com/~phd2/ [email protected]\n Programmers don't die, they just GOSUB without RETURN.",
"msg_date": "Mon, 14 Dec 1998 16:44:39 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hi!\n\nOn Mon, 14 Dec 1998, Thomas G. Lockhart wrote:\n> OK, I applied the fix (confined to backend/utils/adt/nabstime.c) to both\n> CVS trees. This swaps the checking of HAVE_INT_TIMEZONE and HAVE_TM_ZONE\n> so that the TM_ZONE takes precedence.\n> \n> This single-file patch should be a candidate for removal if others run\n> into trouble when testing.\n\n I want to test this. (Recently I did a little patch, but I got even\nuglier results than I expected).\n I have no CVS, so I want a tarred snapshot. Where?\n\nOleg.\n---- \n Oleg Broytmann http://members.tripod.com/~phd2/ [email protected]\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Mon, 14 Dec 1998 16:47:17 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hello!\n\nOn Mon, 14 Dec 1998, Thomas G. Lockhart wrote:\n> OK, I applied the fix (confined to backend/utils/adt/nabstime.c) to both\n> CVS trees. This swaps the checking of HAVE_INT_TIMEZONE and HAVE_TM_ZONE\n> so that the TM_ZONE takes precedence.\n> \n> This single-file patch should be a candidate for removal if others run\n> into trouble when testing.\n> \n> One thing I wasn't certain about was the use of #error in non-gcc\n> compilers. Is that a standard feature?\n\n I just tested latest snapshot (Dec 16). The bug is still there.\nselect datetime('1998-10-01', '11:00') => 1998 Oct 01 18:00\n\nOleg.\n---- \n Oleg Broytmann http://members.tripod.com/~phd2/ [email protected]\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Thu, 17 Dec 1998 19:45:45 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "Hello!\n\nOn Mon, 14 Dec 1998, Thomas G. Lockhart wrote:\n> OK, I applied the fix (confined to backend/utils/adt/nabstime.c) to both\n> CVS trees. This swaps the checking of HAVE_INT_TIMEZONE and HAVE_TM_ZONE\n> so that the TM_ZONE takes precedence.\n\n There is a bug in nabstime.c, line 337 - pgsql 6.4.1. Well, I removed 2\nlines, recompile postgres with DATEDEBUG and got _surprising_ results (below).\n\n I ran the query SELECT datetime('1998-10-01', '11:00')\n Look, it started Ok: timezone = MSK = GMT + 3h. Ok.\n Then there is a line:\ndatetime2tm- time is 14:00:00\n Oops, it became 14:00 somewhere. Seems postgress added GMT offset. Well,\nlater:\ndatetime2tm- (localtime) 98.09.01 18:00:00 MSD dst=1\n How nice - postgres added GMT offset once more!\n\n I just want to know - where? It seems there is a line of code (I am pretty\nsure it is exactly one-line bug) - postgres add gmt_off where it should\nnot. Where?\n\n Detailed log:\n\nGetCurrentAbsoluteTime- timezone is MSK -> -10800 seconds from UTC\ndate_in- input string is 1998-10-01\nParseDateTime- input string is 1998-10-01\nParseDateTime- set field[0] to 1998-10-01 type 2\nDecodeDateTime- field[0] is 1998-10-01 (type 2)\nDecodeNumber- 1998 is 1998 fmask=00000000 tmask=00000000\nDecodeNumber- match 1998 (1998) as year\nDecodeNumber- 10 is 10 fmask=00000004 tmask=00000000\nDecodeNumber- match 10 (10) as month\nDecodeNumber- 01 is 1 fmask=00000006 tmask=00000000\nDecodeNumber- (2) match 1 (01) as day\nDecodeDateTime- field[0] 1998 (00000000/0000000e) value is 1\nDecodeDateTime- mask 0000000e (0000000e) set y1998 m10 d01 00:00:00\nParseDateTime- input string is 11:00\nParseDateTime- set field[0] to 11:00 type 3\nDecodeTimeOnly- field[0] is 11:00 (type 3)\nDecodeTimeOnly- field[0] 11:00 value is -1073763168\nDecodeTimeOnly- mask 00001c0e (00001c00) 11:00:00 (0.000000)\ndate_datetime- date is 1998.10.01\ndate_datetime- time is 00:00:00 0.0000000\ntm2datetime- date is -39484800.000000 (-457.000000 0.000000 0)\ntm2datetime- time is 0.000000 00:00:00 0.000000\ndatetime2tm- date is -39434400.000000 (2451088.000000 50400.000000)\ndatetime2tm- date is 1998.10.01\ndatetime2tm- time is 14:00:00\ndatetime2tm- time is 14:00:00 0.0000000\ndatetime2tm- (localtime) 98.09.01 18:00:00 MSD dst=1\ndatetime2tm- date is 1998.10.01\ndatetime2tm- time is 18:00:00 0.0000000\nEncodeDateTime- timezone is MSD (MSD); offset is -14400 (-10800); daylight is 1 (0)\nEncodeDateTime- day is 2451088\nEncodeDateTime- date result is Thu 01 Oct 18:00:00 1998 MSD\n\nOleg.\n---- \n Oleg Broytmann National Research Surgery Centre http://sun.med.ru/~phd/\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Wed, 23 Dec 1998 18:33:14 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
},
{
"msg_contents": "> I ran the query SELECT datetime('1998-10-01', '11:00')\n> How nice - postgres added GMT offset once more!\n> I just want to know - where? It seems there is a line of code (I am \n> pretty sure it is exactly one-line bug) - postgres add gmt_off where \n> it should not. Where?\n\npostgres=> SELECT datetime('1998-10-01', '11:00');\ndatetime\n----------------------------\nThu Oct 01 11:00:00 1998 MSK\n(1 row)\n\nLook for the #ifdef BREAK_OLEGS_LINUX_GLIBC2 :/\n\nI might have a chance today or next week (nothing in between, sorry) to\nfire up a debugging version of postgres on my linux/glibc2 machine at\nwork.\n\n - Tom\n",
"msg_date": "Wed, 23 Dec 1998 16:33:05 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Date/time on glibc2 linux"
}
] |
[
{
"msg_contents": "> Vadim Mikheev <[email protected]> writes:\n> > Tom Lane wrote:\n> >> No, I think it should *only* substitute for NULL. Why assume\n> >> zero is special?\n> \n> > As I remember this is how SERIAL works in Informix. \n> \n> Ah. OK, if that's what they do then I agree we ought to act the same.\nI hope that this wasn't to say that the SERIAL should substitute for 0\nas well as NULL. That would be quite annoying. I say stick to NULL.\nIf all the other DB's in the world want to shoot themselves in the foot\nI don't think we should follow their example.\n",
"msg_date": "Thu, 3 Dec 1998 13:49:40 -0600",
"msg_from": "\"Jackson, DeJuan\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] DROPping tables with SERIALs "
}
] |
[
{
"msg_contents": "\t>> >> No, I think it should *only* substitute for NULL. Why assume\n\t>> >> zero is special?\n\t>> \n\t>> > As I remember this is how SERIAL works in Informix. \n\t>> \n\t>> Ah. OK, if that's what they do then I agree we ought to act the\nsame.\n\t>I hope that this wasn't to say that the SERIAL should substitute\nfor 0\n\t>as well as NULL. That would be quite annoying. I say stick to\nNULL.\n\n\tYes, 0 is an allowed value for a serial, that should not be\nsubstituted.\n\tI like the behavior that a NULL is substituted.\n\tInformix is a little dull here. It forces a not null constraint, and\nwill therefore\n\traise an error if a null is inserted. I don't see any advantage in\nthis behavior.\n\t(You are forced to skip the field in the insert statement to get a\ngenerated serial)\n\n\tAndreas\n\n",
"msg_date": "Thu, 3 Dec 1998 21:20:38 +0100 ",
"msg_from": "Zeugswetter Andreas IZ5 <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: [HACKERS] DROPping tables with SERIALs "
}
] |
[
{
"msg_contents": "Hi..\n\nI write trigger function with PL/pgSQL Language.\n\nwhen Trigger was fired, i face with error... \"fmgr_info: function 90240:\ncache lookup failed\"\n\nwhat's mean..??\n\nDo you know??\n\n\n",
"msg_date": "Fri, 4 Dec 1998 10:48:40 +0900",
"msg_from": "\"Ki won, Song\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "[Question!] PL/pgSQL Error..???"
}
] |
[
{
"msg_contents": "I seem to have a CASE statement implemented (examples at end). Is anyone\nin the middle of preparing a big patch, or can I go ahead and commit my\nchanges? They touch roughly a dozen files.\n\nAlso, I added arbitrary code to keep the rewriter from complaining, and\nhave not made changes to calculate costs for the optimizer. Would the\nresident experts (Jan and Bruce?) be willing to look at that once the\npatches are in?\n\n - Tom\n\nHere is the table:\n\npostgres=> select * from t2;\ni| f\n-+---\n1|2.2\n3|2.2\n4| 4\n(3 rows)\n\nAnd here is a CASE, including promoting ints to floats for the mixed\ntypes in the WHERE clause:\n\npostgres=> select case when i > f then f else i end from t2;\n?column?\n--------\n 1\n 2.2\n 4\n(3 rows)\n\nShows filling with NULLs if no default clause is specified:\n\npostgres=> select case when i > f then f end from t2;\n?column?\n--------\n\n 2.2\n\n(3 rows)\n\nAnd this last one shows that CASE is allowed inside a qualification\nclause, though I'm having a hard time picturing how that might be\nuseful:\n\npostgres=> select * from t2 where case when i > f then f end is not\nnull;\ni| f\n-+---\n3|2.2\n(1 row)\n",
"msg_date": "Fri, 04 Dec 1998 06:59:45 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Case statement ready?"
},
{
"msg_contents": "\"Thomas G. Lockhart\" wrote:\n> \n> I seem to have a CASE statement implemented (examples at end). Is anyone\n> in the middle of preparing a big patch, or can I go ahead and commit my\n> changes? They touch roughly a dozen files.\n\nThese are files I'm changing for MVCC now:\n\ngist.c indexam.c pg_am.h tqual.c\nMakefile.lmgr gistscan.c ipc.h pg_class.h\ntqual.h hashpage.c ipci.c proc.h\ntranssup.c buf_init.c heapam.c lmgr.c\nrtree.c xact.c buf_internals.h heapam.h\nlmgr.h rtscan.c bufmgr.c hio.c\nlock.c s_lock.c bufmgr.h hio.h\nlock.h s_lock.h execUtils.c htup.h\nnbtpage.c shmem.c\n\nVadim\n",
"msg_date": "Fri, 04 Dec 1998 14:10:52 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Case statement ready?"
},
{
"msg_contents": "> > I seem to have a CASE statement implemented...\n\nOK, it looked to me that I'm not touching any files Vadim is working on,\nso I've gone ahead and committed changes for the CASE construct to the\nCURRENT (main) tree.\n\nThe changes don't actually affect the system catalogs, but I would\nassume that we would not want this to be a feature added to the RELEASE\ntree. Speak up if you think it should be added there too.\n\nI've put in a rudimentary regression test for CASE, but I'll update that\nsoon with some better tests.\n\nIf anyone has a chance to look at the new code wrt the rewriter and\noptimizer costing that would be great. For the costing I would assume\nthat the cost should be related to the sum of the costs of the\nindividual clauses within the CASE statement (or maybe half of the sum).\n\n - Tom\n",
"msg_date": "Fri, 04 Dec 1998 15:44:11 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Case statement ready?"
},
{
"msg_contents": "> If anyone has a chance to look at the new code wrt the rewriter and\n> optimizer costing that would be great. For the costing I would assume\n> that the cost should be related to the sum of the costs of the\n> individual clauses within the CASE statement (or maybe half of the sum).\n\nIsn't only one of the case values going to be used? Wouldn't you want\nthe average of the case entry costs?\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, 4 Dec 1998 11:36:58 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Case statement ready?"
},
{
"msg_contents": ">\n> > > I seem to have a CASE statement implemented...\n>\n> OK, it looked to me that I'm not touching any files Vadim is working on,\n> so I've gone ahead and committed changes for the CASE construct to the\n> CURRENT (main) tree.\n>\n> The changes don't actually affect the system catalogs, but I would\n> assume that we would not want this to be a feature added to the RELEASE\n> tree. Speak up if you think it should be added there too.\n\n It's feature and thus must NOT be committed to the RELEASE\n tree. But if you send me the patches, I could add it to the\n v6.4.1 feature patch which I'll put into the patches\n directory on the ftp server after v6.4.1 is out.\n\n> If anyone has a chance to look at the new code wrt the rewriter and\n> optimizer costing that would be great. For the costing I would assume\n> that the cost should be related to the sum of the costs of the\n> individual clauses within the CASE statement (or maybe half of the sum).\n\n Did it add new node types? If not, the rewrite system might\n already be happy with it. Anyway - I'll take a look at it and\n check/extend/fix for the rewrite corner.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Fri, 4 Dec 1998 20:45:43 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Case statement ready?"
},
{
"msg_contents": "> > For the costing I would assume\n> > that the cost should be related to the sum of the costs of the\n> > individual clauses within the CASE statement (or maybe half of the \n> > sum).\n> Isn't only one of the case values going to be used? Wouldn't you want\n> the average of the case entry costs?\n\nHmm. Actually, on average half of the conditional clauses will be\nevaluated, and then one of the result clauses will be evaluated. So\nperhaps the cost should be:\n\n (0.5*sum(conditional clauses))+avg(result clauses)\n\nOr perhaps we should be pessimistic and leave off the 0.5.\n\n - Tom\n",
"msg_date": "Sat, 05 Dec 1998 06:01:13 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Case statement ready?"
},
{
"msg_contents": "> It's feature and thus must NOT be committed to the RELEASE\n> tree. But if you send me the patches, I could add it to the\n> v6.4.1 feature patch which I'll put into the patches\n> directory on the ftp server after v6.4.1 is out.\n\nI'm thinking that it will lead to nothing but trouble. How about waiting\nfor v6.5?\n\n> > If anyone has a chance to look at the new code wrt the rewriter\n> Did it add new node types? If not, the rewrite system might\n> already be happy with it. Anyway - I'll take a look at it and\n> check/extend/fix for the rewrite corner.\n\nYes, it added new node types (CaseExpr and CaseWhen), and I added a few\nlines to rewriteHandler.c to pretty much ignore them. Not sure if that\nwas the right thing to do.\n\n - Tom\n",
"msg_date": "Sat, 05 Dec 1998 06:03:22 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Case statement ready?"
},
{
"msg_contents": ">\n> > > If anyone has a chance to look at the new code wrt the rewriter\n> > Did it add new node types? If not, the rewrite system might\n> > already be happy with it. Anyway - I'll take a look at it and\n> > check/extend/fix for the rewrite corner.\n>\n> Yes, it added new node types (CaseExpr and CaseWhen), and I added a few\n> lines to rewriteHandler.c to pretty much ignore them. Not sure if that\n> was the right thing to do.\n\n I expect that the structures below the two new nodes can also\n contain Var nodes. And then it's definitely wrong to ignore\n them. They must be handled properly in rewriteHandler and\n rewriteManip.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 7 Dec 1998 10:57:17 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Case statement ready?"
},
{
"msg_contents": "> > > > If anyone has a chance to look at the new code wrt the rewriter\n> I expect that the structures below the two new nodes can also\n> contain Var nodes. And then it's definitely wrong to ignore\n> them. They must be handled properly in rewriteHandler and\n> rewriteManip.\n\nI've just committed changes to the main development tree which allow\nCASE statement handling for at least some cases of rules and views.\n\nStill have trouble (crashes) with CASE statements involving mixing\ncolumns from multiple tables during joins. But the single-table stuff\n(and multi-table when the case statement is confined to a single table)\nseems to work for my testing.\n\nI have an update to the CASE regression test which has many more tests,\nand will put it in the tree and on this list fairly soon (to get some\nOracle/etc testing of results).\n\n - Tom\n",
"msg_date": "Mon, 14 Dec 1998 00:29:15 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Case statement ready!"
}
] |
[
{
"msg_contents": "Such a module can be gotten from:\n\nftp.duke.edu/pub/perl/modules/by-module/Pg\n\ntom\n\nOn Fri, 04 Dec 1998, you wrote:\n>I've recently compiled and put into service Posgresql v6.4. All is working\n>well.\n>\n>I am looking for information on locating, compiling, and implementing a\n>Perl DBD or DBI for Postgresql. Can anyone assist?\n>\n>Bob\n",
"msg_date": "Fri, 4 Dec 1998 14:35:51 +0000",
"msg_from": "Tom Werges <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [GENERAL] Perl DBD / DBI modules"
},
{
"msg_contents": "\nI've recently compiled and put into service Posgresql v6.4. All is working\nwell.\n\nI am looking for information on locating, compiling, and implementing a\nPerl DBD or DBI for Postgresql. Can anyone assist?\n\nBob\n\n",
"msg_date": "Fri, 04 Dec 1998 15:20:43 -0600",
"msg_from": "Bob Kruger <[email protected]>",
"msg_from_op": false,
"msg_subject": "Perl DBD / DBI modules"
},
{
"msg_contents": "I've used PostgreSQL before, but has just joined this mailing list\nagain after having searched www.postgresql.org for any information\nabout 6.4. There's no mention of what is news in 6.4 anywhere to be\nfound. Why not? What _is_ new?\n\n",
"msg_date": "Fri, 4 Dec 1998 23:48:43 +0100 (CET)",
"msg_from": "Kaare Rasmussen <[email protected]>",
"msg_from_op": false,
"msg_subject": "6.4 ?"
},
{
"msg_contents": "On Fri, 4 Dec 1998, Bob Kruger wrote:\n\nJust as with any other perl module:\n\n/usr/local/bin/perl -MCPAN -e shell\ninstall DBI\ninstall DBD::Pg\n\n\n# \n# I've recently compiled and put into service Posgresql v6.4. All is working\n# well.\n# \n# I am looking for information on locating, compiling, and implementing a\n# Perl DBD or DBI for Postgresql. Can anyone assist?\n# \n# Bob\n# \n# \n# \n\n--\nSA, beyond.com My girlfriend asked me which one I like better.\npub 1024/3CAE01D5 1994/11/03 Dustin Sallings <[email protected]>\n| Key fingerprint = 87 02 57 08 02 D0 DA D6 C8 0F 3E 65 51 98 D8 BE \nL_______________________ I hope the answer won't upset her. ____________\n\n",
"msg_date": "Fri, 4 Dec 1998 14:57:05 -0800 (PST)",
"msg_from": "Dustin Sallings <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Perl DBD / DBI modules"
},
{
"msg_contents": "> I've used PostgreSQL before, but has just joined this mailing list\n> again after having searched www.postgresql.org for any information\n> about 6.4. There's no mention of what is news in 6.4 anywhere to be\n> found. Why not? What _is_ new?\n\nSee the TODO list on the support page, or see the HISTORY file in the\n6.4 distribution.\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, 4 Dec 1998 18:13:38 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] 6.4 ?"
},
{
"msg_contents": "I studied the TODO list without finding these elements. How about\nincluding then on the TODO list ?\n\n1. Better money types. As it is now, it is unusable for large amounts,\nbecause it really maps to an 4 byte integer. The best solution would be\nto make the size configurable, but maybe it could do to make it 8 byte\ninteger?\n\n2. An ALTER TABLE ALTER COLUMN new-type. Of course the will be a risk\nof loosing information here, but that's the users problem if he eg.\nwant to convert a 50 character field to 30 character.\n\nOn a related topic, how far down the list is the row level locking? Will\nit come in a not too distant release?\n\n",
"msg_date": "Thu, 10 Dec 1998 07:38:27 +0100 (CET)",
"msg_from": "Kaare Rasmussen <[email protected]>",
"msg_from_op": false,
"msg_subject": "TODO list elements"
},
{
"msg_contents": "On Thu, 10 Dec 1998, Kaare Rasmussen wrote:\n\n> On a related topic, how far down the list is the row level locking? Will\n> it come in a not too distant release?\n\n\tVadim has already starting throwing in code for this...his plan is\nfor v6.5 release (~3months) for this to be in place...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Thu, 10 Dec 1998 03:10:04 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] TODO list elements"
},
{
"msg_contents": "On Thu, 10 Dec 1998, The Hermit Hacker wrote:\n\n// \tVadim has already starting throwing in code for this...his plan\n// is for v6.5 release (~3months) for this to be in place...\n\n\tInteresting...will this help out vacuuming any?\n\n--\nSA, beyond.com The world is watching America,\npub 1024/3CAE01D5 1994/11/03 Dustin Sallings <[email protected]>\n| Key fingerprint = 87 02 57 08 02 D0 DA D6 C8 0F 3E 65 51 98 D8 BE \nL________________________________________ and America is watching TV. __\n\n",
"msg_date": "Thu, 10 Dec 1998 00:21:57 -0800 (PST)",
"msg_from": "dustin sallings <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] TODO list elements"
},
{
"msg_contents": "dustin sallings wrote:\n> \n> On Thu, 10 Dec 1998, The Hermit Hacker wrote:\n> \n> // Vadim has already starting throwing in code for this...his plan\n> // is for v6.5 release (~3months) for this to be in place...\n> \n> Interesting...will this help out vacuuming any?\n\nNo. Vacuum will still exclusively lock relation. But\nI'll fix vacuum to unlock pg_class (when vacuuming all tables\nin database).\n\nAnd note - this will be not row level locking, but\nmulti-version concurrency control.\n\nVadim\n",
"msg_date": "Thu, 10 Dec 1998 16:05:00 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] TODO list elements"
},
{
"msg_contents": "> I studied the TODO list without finding these elements. How about\n> including then on the TODO list ?\n> \n> 1. Better money types. As it is now, it is unusable for large amounts,\n> because it really maps to an 4 byte integer. The best solution would be\n> to make the size configurable, but maybe it could do to make it 8 byte\n> integer?\n\nTODO list has:\n\n\t* Make MONEY/DECIMAL have a defined precision\n\n> \n> 2. An ALTER TABLE ALTER COLUMN new-type. Of course the will be a risk\n> of loosing information here, but that's the users problem if he eg.\n> want to convert a 50 character field to 30 character.\n\nAdded:\n\n\t* Add ALTER TABLE DROP/ALTER COLUMN feature\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": "Thu, 10 Dec 1998 12:15:00 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] TODO list elements"
},
{
"msg_contents": "\nMoved to -hackers...\n\nOn Thu, 10 Dec 1998, dustin sallings wrote:\n\n> On Thu, 10 Dec 1998, The Hermit Hacker wrote:\n> \n> // \tVadim has already starting throwing in code for this...his plan\n> // is for v6.5 release (~3months) for this to be in place...\n> \n> \tInteresting...will this help out vacuuming any?\n\n\tMy understanding that that is a seperate issue altogether, but I\nmay be wrong...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Thu, 10 Dec 1998 13:39:57 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] TODO list elements"
},
{
"msg_contents": "\n\n> -----Original Message-----\n> From: [email protected]\n> [mailto:[email protected]]On Behalf Of Vadim Mikheev\n> Sent: Thursday, December 10, 1998 6:05 PM\n> To: dustin sallings\n> Cc: The Hermit Hacker; Kaare Rasmussen; [email protected]\n> Subject: Re: [GENERAL] TODO list elements\n> \n> \n> dustin sallings wrote:\n> > \n> > On Thu, 10 Dec 1998, The Hermit Hacker wrote:\n> > \n> > // Vadim has already starting throwing in code for this...his plan\n> > // is for v6.5 release (~3months) for this to be in place...\n> > \n> > Interesting...will this help out vacuuming any?\n> \n> No. Vacuum will still exclusively lock relation. But\n> I'll fix vacuum to unlock pg_class (when vacuuming all tables\n> in database).\n>\n\n \n> And note - this will be not row level locking, but\n> multi-version concurrency control.\n>\n\nWhat does it mean ?\nLLL in 6.5 doesn't include row level locking ?\n\nThanks.\n\nHiroshi Inoue\[email protected] \n",
"msg_date": "Mon, 14 Dec 1998 10:51:08 +0900",
"msg_from": "\"Hiroshi Inoue\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [GENERAL] TODO list elements"
},
{
"msg_contents": "Hiroshi Inoue wrote:\n> \n> > And note - this will be not row level locking, but\n> > multi-version concurrency control.\n> >\n> \n> What does it mean ?\n> LLL in 6.5 doesn't include row level locking ?\n\nOne systems (Informix, Sybase) use locking for concurrency control, \nanother ones (Oracle, Interbase) use multi-versioning for this. \nI'm implementing multi-version concurrency control.\n\nVadim\n",
"msg_date": "Mon, 14 Dec 1998 15:03:12 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] TODO list elements"
},
{
"msg_contents": "> -----Original Message-----\n> From: [email protected] [mailto:[email protected]]On Behalf Of Vadim\n> Mikheev\n> Sent: Monday, December 14, 1998 5:03 PM\n> To: Hiroshi Inoue\n> Cc: The Hermit Hacker; [email protected]\n> Subject: Re: [GENERAL] TODO list elements\n> \n> \n> Hiroshi Inoue wrote:\n> > \n> > > And note - this will be not row level locking, but\n> > > multi-version concurrency control.\n> > >\n> > \n> > What does it mean ?\n> > LLL in 6.5 doesn't include row level locking ?\n> \n> One systems (Informix, Sybase) use locking for concurrency control, \n> another ones (Oracle, Interbase) use multi-versioning for this. \n> I'm implementing multi-version concurrency control.\n>\n\nMy words might be obscure.\n\nWhat I meant was\n \n How writers block other writers in LLL ?\n\nCertainly readers block no writers(readers) in LLL.\nBut writers block no writers or the same-row writers or the same-table \nwriters ?\n\nCurrently writers block the same-table writers(readers also) ?\n\nThanks.\n\nHiroshi Inoue\[email protected]\n",
"msg_date": "Mon, 14 Dec 1998 18:13:14 +0900",
"msg_from": "\"Hiroshi Inoue\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [GENERAL] TODO list elements"
},
{
"msg_contents": "Hiroshi Inoue wrote:\n> \n> My words might be obscure.\n> \n> What I meant was\n> \n> How writers block other writers in LLL ?\n> \n> Certainly readers block no writers(readers) in LLL.\n> But writers block no writers or the same-row writers or the same-table\n> writers ?\n> \n> Currently writers block the same-table writers(readers also) ?\n\nOnly same-row writers will be blocked.\n\nVadim\n",
"msg_date": "Mon, 14 Dec 1998 16:21:03 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] TODO list elements"
},
{
"msg_contents": "Anyone working on Row level locking and\non-line recovery?\n\nI have some ideas (and questions) in this area.\nAlso, I was wondering if there is a \"transaction \nserver\" project, i.e., a server that helps provide \na consistent client interface to both the database and\napplication servers. \n\nThank you!\n\nClark\n",
"msg_date": "Mon, 14 Dec 1998 16:46:30 +0000",
"msg_from": "Clark Evans <[email protected]>",
"msg_from_op": false,
"msg_subject": "Row Level Locking, On-line Recovery"
},
{
"msg_contents": "Is there anyone working on XML integration? \nServices include:\n\na) Import/Export to XML (DTD defines schema)\nb) Access (through stored procedures) to XML (PSAX anyone?)\nb) Emulated tables (that direct queries to XML queries)?\nc) Other fun XML items.\n\n:) Clark\n",
"msg_date": "Mon, 14 Dec 1998 16:51:22 +0000",
"msg_from": "Clark Evans <[email protected]>",
"msg_from_op": false,
"msg_subject": "XML Integration"
},
{
"msg_contents": "On Mon, 14 Dec 1998, Clark Evans wrote:\n\n> Anyone working on Row level locking and\n> on-line recovery?\n\n\trow level loocking shoudl be in place for v6.5...not sure about\non-line recovery, but I *believe* Jan is working on redo-log support...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 14 Dec 1998 16:32:54 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Row Level Locking, On-line Recovery"
},
{
"msg_contents": "On Mon, 14 Dec 1998, The Hermit Hacker wrote:\n\n// \trow level loocking shoudl be in place for v6.5...not sure about\n// on-line recovery, but I *believe* Jan is working on redo-log\n// support...\n\n\tOoh, is this the kind of log you can use for transaction\nreplication?\n\n--\nPrinciple Member Technical Staff, beyond.com The world is watching America,\npub 1024/3CAE01D5 1994/11/03 Dustin Sallings <[email protected]>\n| Key fingerprint = 87 02 57 08 02 D0 DA D6 C8 0F 3E 65 51 98 D8 BE \nL______________________________________________ and America is watching TV. __\n\n",
"msg_date": "Tue, 15 Dec 1998 19:00:28 -0800 (PST)",
"msg_from": "dustin sallings <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Row Level Locking, On-line Recovery"
}
] |
[
{
"msg_contents": "Thumbs up on the math changes; my Linux box behaves better than it used\nto (so our \"reference machine\" will match the expected results).\n\nSo, should we look at catching integer overflows while we are at it?\nRight now those overflow silently...\n\n - Tom\n",
"msg_date": "Fri, 04 Dec 1998 16:40:33 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "exp() changes"
},
{
"msg_contents": "> Thumbs up on the math changes; my Linux box behaves better than it used\n> to (so our \"reference machine\" will match the expected results).\n> \n> So, should we look at catching integer overflows while we are at it?\n> Right now those overflow silently...\n\nI would be surprised if you can catch them.\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, 4 Dec 1998 12:17:11 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] exp() changes"
}
] |
[
{
"msg_contents": "Is anyone interested in this problem? (At least add it to the TODO list!)\n\nThis was reported as a bug with the Debian package of 6.3.2; the same\nbehaviour is still present in 6.4. \n\nbray=> create table foo ( t text[]);\nCREATE\nbray=> insert into foo values ( '{\"a\"}');\nINSERT 201354 1\nbray=> insert into foo values ( '{\"a\",\"b\"}');\nINSERT 201355 1\nbray=> insert into foo values ( '{\"a\",\"b\",\"c\"}');\nINSERT 201356 1\nbray=> select * from foo;\nt \n-------------\n{\"a\"} \n{\"a\",\"b\"} \n{\"a\",\"b\",\"c\"}\n(3 rows)\n\nbray=> select t[1] from foo;\nERROR: type name lookup of t failed\nbray=> select * from foo;\nt \n-------------\n{\"a\"} \n{\"a\",\"b\"} \n{\"a\",\"b\",\"c\"}\n(3 rows)\n\nbray=> select foo.t[1] from foo;\nt\n-\na\na\na\n(3 rows)\n\nbray=> select count(foo.t[1]) from foo;\npqReadData() -- backend closed the channel unexpectedly.\n\n-- \nOliver Elphick [email protected]\nIsle of Wight http://www.lfix.co.uk/oliver\n PGP key from public servers; key ID 32B8FAA1\n ========================================\n \"Go ye therefore, and teach all nations, baptizing them\n in the name of the Father, and of the Son, and of the \n Holy Ghost; Teaching them to observe all things \n whatsoever I have commanded you; and, lo, I am with \n you alway, even unto the end of the world. Amen.\" \n Matthew 28:19,20 \n\n\n",
"msg_date": "Sat, 05 Dec 1998 08:54:00 +0000",
"msg_from": "\"Oliver Elphick\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Sorry to nag, but... [Failures with arrays]"
},
{
"msg_contents": "> Is anyone interested in this problem? (At least add it to the TODO list!)\n> \n> This was reported as a bug with the Debian package of 6.3.2; the same\n> behaviour is still present in 6.4. \n> \n> bray=> create table foo ( t text[]);\n> CREATE\n> bray=> insert into foo values ( '{\"a\"}');\n> INSERT 201354 1\n> bray=> insert into foo values ( '{\"a\",\"b\"}');\n> INSERT 201355 1\n> bray=> insert into foo values ( '{\"a\",\"b\",\"c\"}');\n> INSERT 201356 1\n> bray=> select * from foo;\n> t \n> -------------\n> {\"a\"} \n> {\"a\",\"b\"} \n> {\"a\",\"b\",\"c\"}\n> (3 rows)\n> \n> bray=> select t[1] from foo;\n> ERROR: type name lookup of t failed\n> bray=> select * from foo;\n> t \n> -------------\n> {\"a\"} \n> {\"a\",\"b\"} \n> {\"a\",\"b\",\"c\"}\n> (3 rows)\n> \n> bray=> select foo.t[1] from foo;\n> t\n> -\n> a\n> a\n> a\n> (3 rows)\n> \n> bray=> select count(foo.t[1]) from foo;\n> pqReadData() -- backend closed the channel unexpectedly.\n> \n\nI have added:\n\n\t* array index references without table name cause problems\n\t* aggregates on array indexes crash backend\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, 12 Dec 1998 23:25:33 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Sorry to nag, but... [Failures with arrays]"
},
{
"msg_contents": "Added to TODO.\n\n> Is anyone interested in this problem? (At least add it to the TODO list!)\n> \n> This was reported as a bug with the Debian package of 6.3.2; the same\n> behaviour is still present in 6.4. \n> \n> bray=> create table foo ( t text[]);\n> CREATE\n> bray=> insert into foo values ( '{\"a\"}');\n> INSERT 201354 1\n> bray=> insert into foo values ( '{\"a\",\"b\"}');\n> INSERT 201355 1\n> bray=> insert into foo values ( '{\"a\",\"b\",\"c\"}');\n> INSERT 201356 1\n> bray=> select * from foo;\n> t \n> -------------\n> {\"a\"} \n> {\"a\",\"b\"} \n> {\"a\",\"b\",\"c\"}\n> (3 rows)\n> \n> bray=> select t[1] from foo;\n> ERROR: type name lookup of t failed\n> bray=> select * from foo;\n> t \n> -------------\n> {\"a\"} \n> {\"a\",\"b\"} \n> {\"a\",\"b\",\"c\"}\n> (3 rows)\n> \n> bray=> select foo.t[1] from foo;\n> t\n> -\n> a\n> a\n> a\n> (3 rows)\n> \n> bray=> select count(foo.t[1]) from foo;\n> pqReadData() -- backend closed the channel unexpectedly.\n> \n> -- \n> Oliver Elphick [email protected]\n> Isle of Wight http://www.lfix.co.uk/oliver\n> PGP key from public servers; key ID 32B8FAA1\n> ========================================\n> \"Go ye therefore, and teach all nations, baptizing them\n> in the name of the Father, and of the Son, and of the \n> Holy Ghost; Teaching them to observe all things \n> whatsoever I have commanded you; and, lo, I am with \n> you alway, even unto the end of the world. Amen.\" \n> Matthew 28:19,20 \n> \n> \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": "Fri, 18 Dec 1998 13:03:43 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Sorry to nag, but... [Failures with arrays]"
}
] |
[
{
"msg_contents": "Can anyone give me an answer on this Debian bug report, please?:\n\nSize is still defined as unsigned int in 6.4. Paul is suggesting that this\nbe changed to size_t for compatibility with alpha.\n\nIs that correct?\n\n------- Forwarded Message\n\nDate: Wed, 16 Sep 1998 21:03:44 +0200\nFrom: Paul Slootman <[email protected]>\nTo: [email protected]\nSubject: Bug#26778: postgresql-dev: postgresql/c.h typedefs Size as 'unsigned i\n\t nt', should be 'size_t'\n\nPackage: postgresql-dev\nVersion: 6.3.2-11\n\nUsing the above wrong type for \"Size\" leads to conflicts with system\nprototypes, e.g. palloc() which is apparently a define for malloc().\nReplacing the \"typedef unsigned int Size;\" with \"typedef size_t Size;\"\nleads to a usable system again; at least, on Alpha, where \"unsigned int\"\nis NOT the same as \"unsigned long\", unlike i386.\n\nThanks,\nPaul Slootman\n\n- -- System Information\nDebian Release: 2.0\nKernel Version: Linux alf 2.0.35 #1 Tue Aug 11 11:09:24 CEST 1998 alpha unknown\n\n------- End of Forwarded Message\n\n-- \nOliver Elphick [email protected]\nIsle of Wight http://www.lfix.co.uk/oliver\n PGP key from public servers; key ID 32B8FAA1\n ========================================\n \"Go ye therefore, and teach all nations, baptizing them\n in the name of the Father, and of the Son, and of the \n Holy Ghost; Teaching them to observe all things \n whatsoever I have commanded you; and, lo, I am with \n you alway, even unto the end of the world. Amen.\" \n Matthew 28:19,20 \n\n\n",
"msg_date": "Sat, 05 Dec 1998 08:56:17 +0000",
"msg_from": "\"Oliver Elphick\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Nag: postgresql/c.h typedefs Size as 'unsigned int' on Alpha"
},
{
"msg_contents": "> Can anyone give me an answer on this Debian bug report, please?:\n> \n> Size is still defined as unsigned int in 6.4. Paul is suggesting that this\n> be changed to size_t for compatibility with alpha.\n> \n> Is that correct?\n\nWill be in 6.5.\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, 12 Dec 1998 23:26:03 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Nag: postgresql/c.h typedefs Size as 'unsigned int' on\n\tAlpha"
}
] |
[
{
"msg_contents": "subscribe\n\n",
"msg_date": "Sat, 5 Dec 1998 18:34:56 +0800",
"msg_from": "\"Cd Chen\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "None"
}
] |
[
{
"msg_contents": "Hello!\n\n Few weeks ago I've sent a little patch to pgsql-patches concerning small\nfix for locale test. The patch was not approved nor rejected. Something\ngoes wrong?\n I want to flush the patch before I'll start patching postgres on\ndate/time issue.\n\nOleg.\n---- \n Oleg Broytmann http://members.tripod.com/~phd2/ [email protected]\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Sat, 5 Dec 1998 19:40:45 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": true,
"msg_subject": "Locale fix"
}
] |
[
{
"msg_contents": "Tom Lane wrote:\n> \n> I have found one cause of sometimes-but-not-always coredumps in\n> pg_dump -z. Please apply the following patch and let me know whether\n> things get better for you.\n\nI have applied the patch and test it against the same database.\nNow it's ok, no more core dumps, thanks a lot.\n \n> The ACL code in pg_dump.c looks really butt-ugly ... I think I will\n> rewrite the whole thing this afternoon ... but this one bugfix may be\n> enough to get you going for now.\n\nIf you are going to rewrite some of the portions of pg_dump, I would\nlike to ask you to include that tiny patch that makes pg_dump write for\ntable structures just 1 field per line (inserting \\n\\t after every\nfield) in order to get a better look of the file. The current version is\ngiving a extremely long line for tables and it's somehow difficult to\nwatch or edit it.\n\nAll the best,\n\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Sat, 05 Dec 1998 20:44:19 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] pg_dump - segfault with -z option"
}
] |
[
{
"msg_contents": "Hi Tom,\n\nThis looks like a parser bug in 6.4. I found it while trying to run\npg_upgrade:\n\nbray=> create table junk (\"singleton\" \"char\");\nCREATE\nbray=> drop table junk;\nDROP\nbray=> create table junk (\"singleton\" \"char\" default 'M');\nERROR: DEFAULT: const type mismatched\nbray=> create table junk (\"singleton\" char default 'M');\nCREATE\n\npg_dump puts all type names in double quotes, which triggers this bug\nwhen pg_upgrade is run, with the result that the upgrade fails.\n\n[Machine is i686 Linux 2.1.125 with glibc 2.0.7u]\n\n-- \nOliver Elphick [email protected]\nIsle of Wight http://www.lfix.co.uk/oliver\n PGP key from public servers; key ID 32B8FAA1\n ========================================\n \"Go ye therefore, and teach all nations, baptizing them\n in the name of the Father, and of the Son, and of the \n Holy Ghost; Teaching them to observe all things \n whatsoever I have commanded you; and, lo, I am with \n you alway, even unto the end of the world. Amen.\" \n Matthew 28:19,20 \n\n\n",
"msg_date": "Sat, 05 Dec 1998 19:57:15 +0000",
"msg_from": "\"Oliver Elphick\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Parser bug (?)"
},
{
"msg_contents": "\"Oliver Elphick\" <[email protected]> writes:\n> This looks like a parser bug in 6.4. I found it while trying to run\n> pg_upgrade:\n> bray=> create table junk (\"singleton\" \"char\");\n> CREATE\n> bray=> drop table junk;\n> DROP\n> bray=> create table junk (\"singleton\" \"char\" default 'M');\n> ERROR: DEFAULT: const type mismatched\n> bray=> create table junk (\"singleton\" char default 'M');\n> CREATE\n\nOh my, that's interesting ... and it looks like it is closely related\nto my gripe a few days ago that type char is actually implemented as\nchar(1), or more accurately type bpchar(1).\n\nI did\n\tCREATE TABLE t1 (c1 char, c2 \"char\");\n\nand then looked at the info for the relation in pg_attribute:\n\nattname|atttypid|attlen|attnum|attnelems|atttypmod|attbyval|attalign\n-------+--------+------+------+---------+---------+--------+--------\nc1 | 1042| -1| 1| 0| 5|f |i \nc2 | 18| 1| 2| 0| -1|t |c \n\npg_type contains\n\n oid|typname|typowner|typlen|typprtlen|typbyval|typtype|typisdefined|typdelim|typrelid|typelem|typinput|typoutput|typreceive|typsend |typalign|typdefault\n----+-------+--------+------+---------+--------+-------+------------+--------+--------+-------+--------+---------+----------+---------+--------+----------\n 18|char | 256| 1| 1|t |b |t |, | 0| 0|charin |charout |charin |charout |c |\n1042|bpchar | 256| -1| -1|f |b |t |, | 0| 18|bpcharin|bpcharout|bpcharin |bpcharout|i |\n\n\nSo it appears that *something* (probably the parser) is converting the\nunquoted type name char to a bpchar (blank-padded char array),\nbut if it's quoted then you actually get the primitive char type.\n\nSeems to me that that's a bug: if you ask for char, and not char(1), you\nshould get char. At least I'd like it to be a bug, because I want to be\nable to use the unvarnished char type, and I don't want to depend on\nsome weird quoting behavior to get at it.\n\nThe second bug exhibited by Oliver's example is that a literal 'M'\nis not recognized as compatible with the primitive char type.\nThat won't do either.\n\nI tried coercing the 'M' to a char explicitly, just to see what would\nhappen, and got a coredump:\n\ncreate table junk (\"singleton\" \"char\" default 'M'::char);\npqReadData() -- backend closed the channel unexpectedly.\n This probably means the backend terminated abnormally before or while processing the request.\nWe have lost the connection to the backend, so further processing is impossible. Terminating.\n\nBacktrace is\n\n#0 0x800c4234 in _strlen ()\n#1 0xca87c in FlattenStringList (list=0x4008dd58) at gram.y:5047\n#2 0xc6380 in yyparse () at gram.y:845\n#3 0xcac8c in parser (\n str=0x7b034218 \"create table junk (\\\"singleton\\\" \\\"char\\\" default 'M'::char);\", typev=0x0, nargs=0) at parser.c:53\n#4 0xf1c0c in pg_parse_and_plan (\n query_string=0x7b034218 \"create table junk (\\\"singleton\\\" \\\"char\\\" default 'M'::char);\", typev=0x7b038550, nargs=0, queryListP=0x7b036358, dest=Remote,\n aclOverride=0) at postgres.c:420\n\nOoops. Looks like FlattenStringList is getting handed a node type it\nwas not expecting. The elements of the list it is handed are:\n\n$6 = {type = T_String, val = {str = 0x41b6c \"CAST\", ival = 269164,\n dval = 5.7116487507937656e-309}}\n$8 = {type = T_String, val = {str = 0x4008d910 \"'M'\", ival = 1074321680,\n dval = 3.105987548828125}}\n$10 = {type = T_String, val = {str = 0x41b74 \"AS\", ival = 269172,\n dval = 5.7118185104570428e-309}}\n$12 = {type = T_TypeName, val = {str = 0x5 <Address 0x5 out of bounds>,\n ival = 5, dval = 1.0609978954826362e-313}}\n\nSo that's another bug, though possibly unrelated.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 05 Dec 1998 19:27:33 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Parser bug (?) "
},
{
"msg_contents": "> Oh my, that's interesting ... and it looks like it is closely related\n> to my gripe a few days ago that type char is actually implemented as\n> char(1), or more accurately type bpchar(1).\n\nThat's a feature! Sorry I missed your gripe session, but it apparently\nflew by in the e-mail and I didn't spot it. Do you want to dredge it up\nagain, or did one gripe get you feeling better?\n\n> So it appears that *something* (probably the parser) is converting the\n> unquoted type name char to a bpchar (blank-padded char array),\n> but if it's quoted then you actually get the primitive char type.\n> Seems to me that that's a bug: if you ask for char, and not char(1), \n> you should get char.\n\nI'm pretty sure that any benefit to a visible \"unvarnished char\" are\noutweighed by the burden of fully implementing all of the \"uvchar\"\nsupport functions. It would need to be entirely transparent when mixed\nwith char(), varchar(), and text types, and it probably isn't at the\nmoment.\n\nAs a reminder to others (as I'm sure you are already aware of this, but\nwant the feature anyway :), an apparent space savings of 4 bytes (1\nbytes vs 5 bytes) for char vs char(1) is not the 80% savings it appears;\nthere is a large per-tuple overhead that pretty much swamps any tiny\ndata type.\n\n> The second bug exhibited by Oliver's example is that a literal 'M'\n> is not recognized as compatible with the primitive char type.\n> That won't do either.\n> I tried coercing the 'M' to a char explicitly, just to see what would\n> happen, and got a coredump:\n> So that's another bug, though possibly unrelated.\n\nWe should continue the discussion before I try for a fix, to make sure\nthat you don't get railroaded into *never* being able to use \"uvchar\" ;)\n\n - Tom\n",
"msg_date": "Sun, 06 Dec 1998 03:55:17 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Parser bug (?)"
},
{
"msg_contents": "\"Thomas G. Lockhart\" <[email protected]> writes:\n>> ... my gripe a few days ago that type char is actually implemented as\n>> char(1), or more accurately type bpchar(1).\n\n> That's a feature! Sorry I missed your gripe session, but it apparently\n> flew by in the e-mail and I didn't spot it.\n\nI think I sent it to pgsql-sql not the hackers list.\n\n> Do you want to dredge it up again, or did one gripe get you feeling\n> better?\n\nNo, I'm still unhappy.\n\nBasically, what I'm using \"plain char\" for is as a poor man's\nenumerated type. I have a lot of fields that have only half a dozen\nlegal values, and what I do is assign somewhat-mnemonic character\ncodes to each of the allowed values. This is compact, easy to process,\nand I can examine the raw data at need without straining my tired\nbrain cells very far to remember what the codes stand for. (It'd be\neven better to have a *real* enumerated-type facility, I suppose,\nbut this works fine and doesn't take a major amount of work to\nimplement...)\n\nOnly problem is that the fields aren't nearly as compact as I\nthought they were.\n\n> As a reminder to others (as I'm sure you are already aware of this, but\n> want the feature anyway :), an apparent space savings of 4 bytes (1\n> bytes vs 5 bytes) for char vs char(1) is not the 80% savings it appears;\n\nYou forgot the 4-byte alignment requirement of char(n). These things\nreally take 8 bytes each, not the 1 byte each that several \"uvchar\"s\nin a row would take.\n\n> there is a large per-tuple overhead that pretty much swamps any tiny\n> data type.\n\nWhen you have half a dozen of these per tuple (which I do), it starts\nto add up. Besides, the per-tuple overhead has redeeming social\nvalue --- all those fields have a defensible reason for being there.\nA 700% storage overhead with no defensible purpose is harder to\nswallow.\n\n> I'm pretty sure that any benefit to a visible \"unvarnished char\" are\n> outweighed by the burden of fully implementing all of the \"uvchar\"\n> support functions. It would need to be entirely transparent when mixed\n> with char(), varchar(), and text types, and it probably isn't at the\n> moment.\n\nWell, all I really give a darn about is whether I can insert, update,\nand select on equality. (Indeed, since I'm using these as enums,\nI have no need whatever for them to be transparently interchangeable\nwith char-string types. However, I do want them to display as \"char\"\nand not \"int1\", since small integer codes would lose the mnemonic\naspect...) A quick look at pg_operator.h indicates that char is\nreasonably well supplied with functions, anyway.\n\nOnce I found out that I could get to \"uvchar\" with this little quoting\nhack, I satisfied myself that it does everything I need done. But I\ndon't care for relying on what is presumably just a parser bug to get\nat a fundamental data type.\n\nI didn't object to removing char2, char4, and the other fixed-width\nspecial character types, but I think that plain char has applications\nthat justify leaving it in there. I'm willing to consider giving\nit a different name if you really insist that char should be char(1)\n(though I don't agree with that).\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 06 Dec 1998 00:46:35 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Parser bug (?) "
},
{
"msg_contents": "All points are well taken, but I will object to the \"700%\" number. Still\nmisleading imho. But that's not the point...\n\n> I didn't object to removing char2, char4, and the other fixed-width\n> special character types, but I think that plain char has applications\n> that justify leaving it in there. I'm willing to consider giving\n> it a different name if you really insist that char should be char(1)\n> (though I don't agree with that).\n\nThe SQL92 standard sez that char is shorthand for char(1). Trying to\nmake the behavior of char vs char() *completely* transparent as called\nfor in the standard is one reason why I squashed them together from a\nuser's point of view. I only left char in the mix because it is used\ninternally in Postgres system tables and I didn't want to open that can\nof worms (and don't intend to, so don't panic).\n\nI shouldn't hide behind the SQL92 standards argument, since there are\nsome parts of the standard which are so hideous that they should be\nignored. But I'm not certain this is one of them.\n\nAnother detail: we would need to figure out how to do locale and\nmultibyte support for this \"char\" type to allow equivalence with\nmultibyte char(1). Not sure how to do that since this type *is* used\ninternally and probably can't be resized like that.\n\nSpeaking of which, are you or Bruce (or anyone else) thinking of testing\nthe unsigned int vs size_t for the Size typedef recently mentioned by\nOliver? Can we count on all platforms to have size_t defined? If so, we\ncould consider adding it in for v6.4.1. I'm currently working on getting\nmin() and max() to work with string types, but could drop that to polish\nthings for a v6.4.1 release.\n\n - Tom\n",
"msg_date": "Sun, 06 Dec 1998 06:46:56 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Parser bug (?)"
},
{
"msg_contents": "\"Thomas G. Lockhart\" <[email protected]> writes:\n> All points are well taken, but I will object to the \"700%\" number. Still\n> misleading imho.\n\nOK, you understated for effect, I overstated for effect, we're even ;-)\n\n> The SQL92 standard sez that char is shorthand for char(1).\n\nUm. Well, what do you think of calling it \"char1\" or some such?\n\n> I only left char in the mix because it is used\n> internally in Postgres system tables and I didn't want to open that can\n> of worms (and don't intend to, so don't panic).\n\nNow that you mention it, I've noticed a few places where plain char is\nused in the system tables in *exactly* the way I'm talking about, ie,\nas a simple form of enumerated type. For example, the typtype and\ntypalign fields in pg_type.\n\nAll I'm asking for is a reliable way to get at that same functionality\nin a user table.\n\n> Another detail: we would need to figure out how to do locale and\n> multibyte support for this \"char\" type to allow equivalence with\n> multibyte char(1). Not sure how to do that since this type *is* used\n> internally and probably can't be resized like that.\n\nUrgh. Multibyte char support is the *last* thing I'm looking for\nin this context. How about we rename the type\n\"justoneplainvanillaasciichar\" and have done with it ;-) ?\n\n\n[ caution, topic drift ahead ]\n\n> Speaking of which, are you or Bruce (or anyone else) thinking of testing\n> the unsigned int vs size_t for the Size typedef recently mentioned by\n> Oliver? Can we count on all platforms to have size_t defined?\n\nI was looking at that. size_t exists on all platforms, the trouble\nis that pre-ANSI platforms are not very consistent about exactly\nwhich system header file(s) define it. If we make c.h (and c.h, not\npostgres.h, is what defines Size) depend on size_t then we may find\na few compilation failures due to .c files not pulling in all the\nright system headers before including c.h.\n\nOn the other hand, c.h unconditionally requires <stdlib.h> which\nitself is an ANSI-ism. And stdlib is one of the headers that ANSI\nspecifies to define size_t --- so any ANSI-compliant header fileset\n*will* support this change. It's only not-quite-ANSI systems that\nwe risk problems with here. So probably I'm being too conservative\nto worry at all. If you don't have a reasonably ANSI-conformant\ncompiler and header fileset you're gonna have a heck of an\nunpleasant time building Postgres anyway, I suspect.\n\nc.h says that Size is intended to represent the result type of\nsizeof, and that most certainly is size_t, *not* any other type,\naccording to the ANSI spec. So if Size is being used in the\ncode to represent the size of in-memory objects then it definitely\nought to be size_t.\n\nIn theory this change is absolutely correct, and my guess is we\nshould do it. But if we see a few glitches on obsolete platforms,\ndon't say I didn't warn you ;-).\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 06 Dec 1998 02:58:46 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Parser bug (?) "
},
{
"msg_contents": "> Speaking of which, are you or Bruce (or anyone else) thinking of testing\n> the unsigned int vs size_t for the Size typedef recently mentioned by\n> Oliver? Can we count on all platforms to have size_t defined? If so, we\n> could consider adding it in for v6.4.1. I'm currently working on getting\n> min() and max() to work with string types, but could drop that to polish\n> things for a v6.4.1 release.\n\nI think everyone has size_t. It is s POSIX requirement. Sounds like a\ngood change.\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": "Sun, 6 Dec 1998 11:12:07 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Parser bug (?)t"
},
{
"msg_contents": "> I think everyone has size_t. It is s POSIX requirement. Sounds like a\n> good change.\n\nTom Lane points out that some of our more esoteric platforms may be\nmissing it, so it sounds like it should be an optional v6.4.1 patch and\na v6.5 built-in feature (to allow platform testing). Who is testing now?\n\n - Tom\n",
"msg_date": "Sun, 06 Dec 1998 23:27:38 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Parser bug (?)t"
},
{
"msg_contents": "Is this fixed?\n\n\n> Hi Tom,\n> \n> This looks like a parser bug in 6.4. I found it while trying to run\n> pg_upgrade:\n> \n> bray=> create table junk (\"singleton\" \"char\");\n> CREATE\n> bray=> drop table junk;\n> DROP\n> bray=> create table junk (\"singleton\" \"char\" default 'M');\n> ERROR: DEFAULT: const type mismatched\n> bray=> create table junk (\"singleton\" char default 'M');\n> CREATE\n> \n> pg_dump puts all type names in double quotes, which triggers this bug\n> when pg_upgrade is run, with the result that the upgrade fails.\n> \n> [Machine is i686 Linux 2.1.125 with glibc 2.0.7u]\n> \n> -- \n> Oliver Elphick [email protected]\n> Isle of Wight http://www.lfix.co.uk/oliver\n> PGP key from public servers; key ID 32B8FAA1\n> ========================================\n> \"Go ye therefore, and teach all nations, baptizing them\n> in the name of the Father, and of the Son, and of the \n> Holy Ghost; Teaching them to observe all things \n> whatsoever I have commanded you; and, lo, I am with \n> you alway, even unto the end of the world. Amen.\" \n> Matthew 28:19,20 \n> \n> \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, 12 Dec 1998 23:29:41 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Parser bug (?)"
},
{
"msg_contents": "> Speaking of which, are you or Bruce (or anyone else) thinking of testing\n> the unsigned int vs size_t for the Size typedef recently mentioned by\n> Oliver? Can we count on all platforms to have size_t defined? If so, we\n> could consider adding it in for v6.4.1. I'm currently working on getting\n> min() and max() to work with string types, but could drop that to polish\n> things for a v6.4.1 release.\n\nAdded to current tree. 6.4.1 tree too likely to break.\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, 12 Dec 1998 23:31:22 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Parser bug (?)"
},
{
"msg_contents": "> Is this fixed?\n\nNo, not yet. Put it on the \"show stopper\" list for v6.4.1...\n\nThe explanation is that CHAR is a \"keyword\" so is passed from the\nscanner to the parser explicitly, which then makes sure that the various\nlegal forms of char type declarations get translated into the internal\nbpchar type. \n\nHowever, tokens surrounded by double quotes are passed as IDENTs, which\nbypasses this mechanism completely.\n\nI'll look at modifying pg_dump to leave out the double quotes on type\nfields unless there is mixed or upper case. Also, I'll look at having\nsome automatic conversions between bpchar and char (though this might\nnot be available for v6.4.1 since it may require catalog changes).\n\n - Tom\n\n> > This looks like a parser bug in 6.4. I found it while trying to run\n> > pg_upgrade:\n> > bray=> create table junk (\"singleton\" \"char\" default 'M');\n> > ERROR: DEFAULT: const type mismatched\n> > pg_dump puts all type names in double quotes, which triggers this \n> > bug when pg_upgrade is run, with the result that the upgrade fails.\n",
"msg_date": "Sun, 13 Dec 1998 13:31:38 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Parser bug (?)"
}
] |
[
{
"msg_contents": ">> (I am currently packaging 6.4 into an RPM for Red Hat 5.2).\n>\n>Is this an alternative to the RedHat packaging, or are you using the\n>same layout and RPM setup that the RedHat 6.3 package had? Just curious;\n>I've been thinking about trying to help with RPM support but would be\n>happy that someone else is looking after it.\n>\n>btw, we are currently looking at a date/time peculiarity in Postgres\n>with glibc2 which gives funny results for at least one test case. You\n>might want to stay in touch so we can fix that up for you also, once we\n>have a good workaround.\n>\n>Actually, we should also test with RH5.2 to make sure the date/time\n>problem is still present, since it may have been due to a problem in\n>glibc2 which has since been fixed. I'm only running RH4.2 and RH5.1 so\n>can't directly test that myself...\n\n\nI am running PG 6.4 on RH5.2. If you give me the details I can try a test.\n\n\n",
"msg_date": "Sun, 6 Dec 1998 19:38:18 -0500",
"msg_from": "\"Bryan White\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [INTERFACES] crypt not included when compiling libpgtcl !!!!!!!"
}
] |
[
{
"msg_contents": "Dear All\nwe have a simple problem: a straight select on one table blows the top \nof our 512MB computer. The table has arounf 800000 entries which we\nneed to dump to a file. We redirect the output to a file in psql\nwith \\o file.extract\nwhich is then follwoed by the select (no joins, or\naggregations). Instead of writing each entry write away to the file,\nPOSTGRES seems to collect everything in RAM: top shows that Postgres\nis getting bigger and bigger. At around 700MB it finally dies - with\nan empty file. I dont think that this is a memory leak, instead \nit seems to be a 'feature'. When doing a select * it also takes quite\na while before the records get spit out.\n\nAny ideas of what to do?\n\ngreetings\n\nEildert Groeneveld\n\n\n=========================================\nInstitute for\nAnimal Science and Animal Behaviour\nMariensee 31535 Neustadt Germany\nTel : (49)(0)5034 871155\nFax : (49)(0)5034 92579\nwww : http://www.tzv.fal.de/~eg/ \ne-mail: [email protected]\n=========================================\n",
"msg_date": "Mon, 7 Dec 1998 15:36:33 +0100 (CET)",
"msg_from": "Eildert Groeneveld <[email protected]>",
"msg_from_op": true,
"msg_subject": "memory leak?"
}
] |
[
{
"msg_contents": "After a couple of weeks of checking and testing, final 0.92 release of\nPgAccess is ready to be included in PostgreSQL 6.4.1 ! Please, will you\nbe so kind to add this last version to the current PostgreSQL snapshot.\n\nThere are a lot of minor improvements regarding window handling,\ndatabase accesses, error handling.\nIn \"Table information\" window, you can now rename columns or add new\nones.\nThe significant improvement is a new module for user management. You can\ncreate now new postgresql users, change passwords and so on.\n\nThe same download address : http://www.flex.ro/pgaccess or \nftp://ftp.flex.ro/pub/pgaccess/pgaccess-0.92.tar.gz\n\nBE CAREFUL : Version 0.92b is a beta , 0.92 is the latest stable one.\n\n\nAll the best,\n\nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n\nPS. A lot of people are asking me if there is a libpgtcl.dll for M$\nWindows. The pgtcl package, written in pure Tcl/Tk is working but it's\nvery slooooow. Is there any chance that someone could try to port the\nlibpgtcl library for Windows ?\n",
"msg_date": "Mon, 07 Dec 1998 17:26:15 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "PgAccess stable version 0.92 released !"
},
{
"msg_contents": "> PS. A lot of people are asking me if there is a libpgtcl.dll for M$\n> Windows. The pgtcl package, written in pure Tcl/Tk is working but it's\n> very slooooow. Is there any chance that someone could try to port the\n> libpgtcl library for Windows ?\nI tried but the functions\nTcl_CreateFileHandler(/usr/src/pgsql/src/interfaces/libpgtcl/pgtclId.c:679)\nand Tcl_DeleteFileHandler\n(/usr/src/pgsql/src/interfaces/libpgtcl/pgtclId.c:701) were removed for\nnon-Unix platforms as of tcl/tk8.0b1 (or thereabouts). I am no tcl expert so\nI will not try to change that but I got the advise to start with\nTcl_MakeFileChannel. If you (Constatin) can make the change I will\nincorporate that in the Windows NT port.\n\nJoost Kraaijeveld\n\nAskesis B.V.\nGroenewoudeseweg 46\n6524VB Nijmegen\ntel: 024-3888063\nfax: 024-3608416\nweb: www.askesis.nl\n\n\n",
"msg_date": "Tue, 8 Dec 1998 08:01:11 +0100",
"msg_from": "\"Joost Kraaijeveld\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [INTERFACES] PgAccess stable version 0.92 released !"
},
{
"msg_contents": "Joost Kraaijeveld wrote:\n> \n> > PS. A lot of people are asking me if there is a libpgtcl.dll for M$\n> > Windows. The pgtcl package, written in pure Tcl/Tk is working but it's\n> > very slooooow. Is there any chance that someone could try to port the\n> > libpgtcl library for Windows ?\n\n> I tried but the functions\n> Tcl_CreateFileHandler(/usr/src/pgsql/src/interfaces/libpgtcl/pgtclId.c:679)\n> and Tcl_DeleteFileHandler\n> (/usr/src/pgsql/src/interfaces/libpgtcl/pgtclId.c:701) were removed for\n> non-Unix platforms as of tcl/tk8.0b1 (or thereabouts). I am no tcl expert so\n> I will not try to change that but I got the advise to start with\n> Tcl_MakeFileChannel. If you (Constatin) can make the change I will\n> incorporate that in the Windows NT port.\n\nUnfortunately, (for Windows PgAccess users :-) but very convenient for\nme and my company) my company has completely switched from Microsoft\nWindows to Linux so I cannot test or compile libpgtcl for Windows with\nyour above suggestions.\n\nI don't know where that functions are used, but if they are used in\nfunctions concerning large objects, I think that we could just ignore\nthem and prepare a tiny libpgtcl.dll just for PgAccess use, taking into\naccount that PgAccess is not using large object functions.\n\nOn the other hand, Hiroshi Inoue [email protected] informed me that he\nsucceeded in compiling libgtcl for Windows in PostgreSQL 6.4 beta\ndistribution and he could probably do that for current 6.4 tree. If this\nwould be possible, we may expect that the next 6.4.1 distribution will\ncontain also PgAccess available for Windows users.\n\nAnyhow, I will just wait for a sign from you or from any other developer\nwho is able to compile and test libpgtcl.dll against PgAccess and send\nto me the libpq.dll and libpgtcl.dll in order to includ them in pgaccess\ndistribution kit.\n\nAll the best,\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Tue, 08 Dec 1998 09:38:11 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: libpgtcl.dll for Windows"
},
{
"msg_contents": "Constantin Teodorescu <[email protected]> writes:\n> Joost Kraaijeveld wrote:\n>>>> PS. A lot of people are asking me if there is a libpgtcl.dll for M$\n>>>> Windows. The pgtcl package, written in pure Tcl/Tk is working but it's\n>>>> very slooooow. Is there any chance that someone could try to port the\n>>>> libpgtcl library for Windows ?\n\n>> I tried but the functions\n>> Tcl_CreateFileHandler(/usr/src/pgsql/src/interfaces/libpgtcl/pgtclId.c:679)\n>> and Tcl_DeleteFileHandler\n>> (/usr/src/pgsql/src/interfaces/libpgtcl/pgtclId.c:701) were removed for\n>> non-Unix platforms as of tcl/tk8.0b1 (or thereabouts).\n\nOK, I'm to blame for that code...\n\n> I don't know where that functions are used, but if they are used in\n> functions concerning large objects,\n\nLarge objects aren't the issue; support for async NOTIFY is. The\ncomments at the head of pgtclId.c explain:\n\n Another headache is that Ousterhout keeps changing the Tcl I/O interfaces.\n libpgtcl currently claims to work with Tcl 7.5, 7.6, and 8.0, and each of\n 'em is different. Worse, the Tcl_File type went away in 8.0, which means\n there is no longer any platform-independent way of waiting for file ready.\n So we now have to use a Unix-specific interface. Grumble.\n\n In the current design, Pg_Notify_FileHandler is a file handler that\n we establish by calling Tcl_CreateFileHandler(). It gets invoked from\n the Tcl event loop whenever the underlying PGconn's socket is read-ready.\n We suck up any available data (to clear the OS-level read-ready condition)\n and then transfer any available PGnotify events into the Tcl event queue.\n Eventually these events will be dispatched to Pg_Notify_EventProc. When\n we do an ordinary PQexec, we must also transfer PGnotify events into Tcl's\n event queue, since libpq might have read them when we weren't looking.\n\nIf anyone can explain to me a platform-independent way of waiting for\nsocket-read-ready under Tcl 8.0, I'd gladly incorporate it.\n\nI suppose the other alternative is to insert a bunch of \"#ifdef WIN32\"\ncode to provide a platform-specific implementation of this functionality\nfor Windows. But I'm not sure what that would be, and I'm in no\nposition to write or test it anyway. Volunteers?\n\n(If you don't need NOTIFY functionality, it'd probably be possible\njust to #ifdef out the filehandler and associated setup code... but\nI don't see that as a reasonable solution for general use.)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 08 Dec 1998 09:58:46 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [INTERFACES] Re: libpgtcl.dll for Windows "
},
{
"msg_contents": "Tom Lane wrote:\n> \n> (If you don't need NOTIFY functionality, it'd probably be possible\n> just to #ifdef out the filehandler and associated setup code... but\n> I don't see that as a reasonable solution for general use.)\n\nI think it's the best way of delivering a quick fix for the problem so\nthat pgaccess could work on Windows.\nI didn't think that NOTIFY missing would be a disaster.\nInstead, we will gain a lot of Windows users who will feel the good\ntaste of PostgreSQL :-) , and sometimes, someone will find the answer of\nthat socket-read-ready problem.\n\nSo, just #ifdef it, and let them join the band :-)\n\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Tue, 08 Dec 1998 20:54:53 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [INTERFACES] Re: libpgtcl.dll for Windows"
}
] |
[
{
"msg_contents": "Attached are several files with info in how to compile postgres 6.4 on\nWindows NT.\nFor questions, feel free to ask.\n\nJoost Kraaijeveld\n\nAskesis B.V.\nGroenewoudeseweg 46\n6524VB Nijmegen\ntel: 024-3888063\nfax: 024-3608416\nweb: www.askesis.nl",
"msg_date": "Mon, 7 Dec 1998 16:30:54 +0100",
"msg_from": "\"Joost Kraaijeveld\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Compiling Postgres for Windows NT"
}
] |
[
{
"msg_contents": "I just got a note from Chris Williams to the effect that when he built\n6.4 on his HPUX setup, the install step failed. Seems that HP has\na largely incompatible program called \"install\" living in /etc, and\nconfigure tried to use it. (I hadn't ever noticed this because I don't\nrun with /etc in my path except when I'm root. HP does offer a\ncompatible flavor of install in /opt/imake/bin...)\n\nThis brings up something I've been thinking for a while, but hadn't\ngotten around to proposing. I have noticed that most other large\npackages that need \"install\" use an install script distributed\nwith the package; they all say there's too much variation in\nvendor-supplied install programs to risk depending on those.\nI say it's time we do the same. We do already include an install script\n(src/install-sh), so it's not at all clear what the value is of trying\nto use a locally provided install program. We're not saving space, and\nwe're wasting time during configure.\n\nI propose taking out configure's search for an install program and\njust letting it select install-sh always. Anyone who really wanted\nto use their local install could still override that decision by\nchanging Makefile.global.\n\nAnother alternative is to use the standard autoconf AC_PROG_INSTALL\nmacro, which contains some grotty tests to weed out incompatible\nversions of install. But I don't really see the reason for taking any\nrisk.\n\nComments? Objections?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 07 Dec 1998 11:34:09 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": true,
"msg_subject": "Proposed autoconf change: rip out search for 'install'"
},
{
"msg_contents": "On Mon, 7 Dec 1998, Tom Lane wrote:\n\n> I just got a note from Chris Williams to the effect that when he built\n> 6.4 on his HPUX setup, the install step failed. Seems that HP has\n> a largely incompatible program called \"install\" living in /etc, and\n> configure tried to use it. (I hadn't ever noticed this because I don't\n> run with /etc in my path except when I'm root. HP does offer a\n> compatible flavor of install in /opt/imake/bin...)\n> \n> This brings up something I've been thinking for a while, but hadn't\n> gotten around to proposing. I have noticed that most other large\n> packages that need \"install\" use an install script distributed\n> with the package; they all say there's too much variation in\n> vendor-supplied install programs to risk depending on those.\n> I say it's time we do the same. We do already include an install script\n> (src/install-sh), so it's not at all clear what the value is of trying\n> to use a locally provided install program. We're not saving space, and\n> we're wasting time during configure.\n> \n> I propose taking out configure's search for an install program and\n> just letting it select install-sh always. Anyone who really wanted\n> to use their local install could still override that decision by\n> changing Makefile.global.\n> \n> Another alternative is to use the standard autoconf AC_PROG_INSTALL\n> macro, which contains some grotty tests to weed out incompatible\n> versions of install. But I don't really see the reason for taking any\n> risk.\n> \n> Comments? Objections?\n\n\tAC_PROG_INSTALL looks like a good way to do it...go for it...I do\nobject to having it 'hardcoded', as it makes the assumption that the\nsystem administrator doesn't know what they are doing (ie. in setting\npaths)\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 7 Dec 1998 13:13:23 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Proposed autoconf change: rip out search for 'install'"
},
{
"msg_contents": "> I propose taking out configure's search for an install program and\n> just letting it select install-sh always. Anyone who really wanted\n> to use their local install could still override that decision by\n> changing Makefile.global.\n\n> \n> Comments? Objections?\n\nI have no problem with that.\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": "Mon, 7 Dec 1998 12:28:54 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Proposed autoconf change: rip out search for 'install'"
},
{
"msg_contents": "On Mon, 7 Dec 1998, Bruce Momjian wrote:\n\n> > I propose taking out configure's search for an install program and\n> > just letting it select install-sh always. Anyone who really wanted\n> > to use their local install could still override that decision by\n> > changing Makefile.global.\n> \n> > \n> > Comments? Objections?\n> \n> I have no problem with that.\n\n\tI do...I don't like it when we assume that we're right and the\nperson installing the software is wrong...usnig AC_PROG_INSTALL should be\nsufficient...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 7 Dec 1998 15:33:40 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Proposed autoconf change: rip out search for 'install'"
},
{
"msg_contents": "The Hermit Hacker <[email protected]> writes:\n>>>> I propose taking out configure's search for an install program and\n>>>> just letting it select install-sh always. Anyone who really wanted\n>>>> to use their local install could still override that decision by\n>>>> changing Makefile.global.\n\n>> I have no problem with that.\n\n> \tI do...I don't like it when we assume that we're right and the\n> person installing the software is wrong...usnig AC_PROG_INSTALL should be\n> sufficient...\n\nWell, not to put too fine a point on it, Chris' complaint was that the\nsoftware failed to do the Right Thing even though he had a peculiar\nPATH. And certainly we do not have any installation instruction that\ncautions you against having an incompatible \"install\" in your path.\n(I'd not want to try to write an instruction like that, either, because\nI don't know which ones aren't compatible.) So I'm not convinced by\nMarc's argument that the configure program should assume the user knows\nwhat he's doing in this respect.\n\nThe minimum-change, safest way to address the issue is to take out\nconfigure.in's search for an install and always use install-sh\nunless the user specifically overrides that. Using AC_PROG_INSTALL\nlooks to me like it introduces more risk; there's a possibility that\nit will settle on an install that isn't really compatible, which is\nthe same problem we've got now. On the other hand, there's a lot to\nbe said for using standardized solutions, and AC_PROG_INSTALL is\npresumably pretty well tested...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 07 Dec 1998 20:26:23 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Proposed autoconf change: rip out search for 'install' "
},
{
"msg_contents": "> I do...I don't like it when we assume that we're right and the\n> person installing the software is wrong...usnig AC_PROG_INSTALL should \n> be sufficient...\n\n... and if it isn't, then perhaps a bunch of \n\n#ifdef HPUX ...\nif (port ~ \"HPUX\") ...\n\nmight make it clearer why we made the choice. All of the platforms have\ntheir foibles, but some brands don't even give it the old college try on\nkeeping consistancy and standardization between releases and within a\nrelease. I'd always liked Dec machines because even through major\nreleases and processor revs they managed binary and environment\ncompatibility. 'Course, fat lot of good it did them *sniff*. HP has\nbinaries which won't run on different processors in the same line, and\nhas system inconsistancies even between v9.01 and v9.02! Who do they\nthink they are, PostgreSQL? :)\n\nAnyway, Tom, do you think that the AC_PROG_INSTALL function might help\non the HP? If so, we've probably stressed it pretty good...\n\n - Tom\n",
"msg_date": "Tue, 08 Dec 1998 01:29:17 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Proposed autoconf change: rip out search for 'install'"
},
{
"msg_contents": "\"Thomas G. Lockhart\" <[email protected]> writes:\n> Anyway, Tom, do you think that the AC_PROG_INSTALL function might help\n> on the HP? If so, we've probably stressed it pretty good...\n\nAC_PROG_INSTALL would solve the problem on HP --- one of the ad hoc\ntests that it uses is to ignore /etc/install and /usr/sbin/install,\nwhich are the two places that that program might live on HP. (BTW,\nAC_PROG_INSTALL's comments refer to this as SysV install, so I think\nyou are being unfairly hard on HP to blame them for the lack of\ncompatibility. They *are* being compatible ... with SysV. And\nnormal users don't put either of those directories into PATH.)\n\nI don't *know* of any cases where AC_PROG_INSTALL would fail, and\ncertainly it's pretty widely used. I'm just being paranoid because\nit has no way to directly test what the install program really does ---\nit is using a bunch of ad-hoc rules to guess whether a program it finds\nis likely to be BSD-compatible or not. That's not my idea of how a\nreliable autoconfiguration test ought to work.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 07 Dec 1998 20:37:37 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Proposed autoconf change: rip out search for 'install' "
},
{
"msg_contents": "On Mon, 7 Dec 1998, Tom Lane wrote:\n\n> I don't *know* of any cases where AC_PROG_INSTALL would fail, and\n\n\tThen sounds like a good solution to me...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Tue, 8 Dec 1998 00:29:46 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Proposed autoconf change: rip out search for 'install'"
},
{
"msg_contents": " The minimum-change, safest way to address the issue is to take out\n configure.in's search for an install and always use install-sh\n unless the user specifically overrides that. Using AC_PROG_INSTALL\n looks to me like it introduces more risk; there's a possibility that\n it will settle on an install that isn't really compatible, which is\n the same problem we've got now. On the other hand, there's a lot to\n be said for using standardized solutions, and AC_PROG_INSTALL is\n presumably pretty well tested...\n\nWhy not just go with AC_PROG_INSTALL since it gets tested by all the\nother gnu software installations anyway. If there are special cases\nthat it doesn't grok we can always put it inside a conditional based\non a --with-install=/your/favorite/install/program configure option.\nI doubt we really will need that, though.\n\nCheers,\nBrook\n",
"msg_date": "Tue, 8 Dec 1998 13:53:24 -0700 (MST)",
"msg_from": "Brook Milligan <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Proposed autoconf change: rip out search for 'install'"
},
{
"msg_contents": "Seems like this issue gets more than its fair share of attention. I\nstarted a round of discussion by pointing out that /usr/ucb/install worked\non Solaris while the default one didn't.\n\nIt's just not a complicated program and many other open-source packages\n(like gcc I think) find it easier to just bundle a shell script rather than\ndecipher which existing program (if any) do what they want.\n\nGo for it Tom (Lane).\n\nSignature failed Preliminary Design Review.\nFeasibility of a new signature is currently being evaluated.\[email protected], or [email protected]\n\n\n",
"msg_date": "Wed, 9 Dec 1998 00:00:47 -0800",
"msg_from": "\"Henry B. Hotz\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Proposed autoconf change: rip out search for 'install'"
}
] |
[
{
"msg_contents": "Recently, while testing the application compatibility of the ODBC driver\nin , we have come across a problem. Some applications, specifically\nMS Project, want to create tables using a bounded binary data type.\n(i.e. SQL_BINARY) Up to now, we have mapped ODBC's SQL_BINARY to\nPostgreSQL's BYTEA. The problem is that when the application tries to\ncreate the table. Example:\n\n CREATE TABLE foo (bar BYTEA(100));\n\nPostgreSQL errors, complaining about the \"(\". I have created a\npatch, which I am now testing, to allow this syntax for BYTEA. It also,\ntruncates to the specified precision in the same fashion as does VARCHAR\nwhile preserving its original behavior when a precision is not\nspecified.\n\n1. Are there any objections?\n\n2. It is very light weight and I would like to get the patch into\n6.4.1. Is this possible? (I do not know the plans for this 6.4.1)\n\n",
"msg_date": "Mon, 07 Dec 1998 13:52:48 -0500",
"msg_from": "David Hartwig <[email protected]>",
"msg_from_op": true,
"msg_subject": "Bytea(precision)"
}
] |
[
{
"msg_contents": "We are in the process of implementing several additional Extended Fetch\ncapabilities into the ODBC driver. In order to complete the full set of\nfunctions, we need to be able to FETCH to the end of the cursor, detect\nthe end, and then be able to move to prior positions in the the cursor.\nThe problem is that once the end of the cursor is reached the cursor is\nno longer usable.\n\nI think the cursor should still be valid after moving past the end. I\nhave tinkered around to see if the cursor can be preserved after\nmoving/fetching past the end without much luck. I am throwing this\nout to the list in the hopes that someone will respond \"Gee, Dave, all\nyou have to do is (*^%^$$%$#!)\".\n\nI would greatly appreciate any assistance in resolving this problem.\nMany ODBC components and applications make heavy use of these extended\nfeatures.\n\nTIA.\n\n",
"msg_date": "Mon, 07 Dec 1998 14:19:18 -0500",
"msg_from": "David Hartwig <[email protected]>",
"msg_from_op": true,
"msg_subject": "Cursor Movement - Past the End"
},
{
"msg_contents": "> We are in the process of implementing several additional Extended Fetch\n> capabilities into the ODBC driver. In order to complete the full set of\n> functions, we need to be able to FETCH to the end of the cursor, detect\n> the end, and then be able to move to prior positions in the the cursor.\n> The problem is that once the end of the cursor is reached the cursor is\n> no longer usable.\n> \n> I think the cursor should still be valid after moving past the end. I\n> have tinkered around to see if the cursor can be preserved after\n> moving/fetching past the end without much luck. I am throwing this\n> out to the list in the hopes that someone will respond \"Gee, Dave, all\n> you have to do is (*^%^$$%$#!)\".\n> \n> I would greatly appreciate any assistance in resolving this problem.\n> Many ODBC components and applications make heavy use of these extended\n> features.\n\nAdded to TODO:\n\n\t* make CURSOR valid even after you hit end of cursor\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, 12 Dec 1998 23:34:19 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Cursor Movement - Past the End"
}
] |
[
{
"msg_contents": "While investigating a user's complaint, I have found some memory\ndestructions in 6.4 source using purify.\n\n(1) parser/gram.y:fmtId()\n\nIt writes n+3 bytes into n+1 byte-long memory area if mixed case or\nnon-ascii identifiers given.\n\n(2) catalog/index.c:\n\nATTRIBUTE_TUPLE_SIZE bytes are allocated but\nsizeof(FormData_pg_attribute) bytes are written. Note that\nATTRIBUTE_TUPLE_SIZE is smaller than\nsizeof(FormData_pg_attribute). (for example, on solaris 2.6,\nATTRIBUTE_TUPLE_SIZE is 3 bytes smaller).\n\nAttached patches try to fix the problem. I do not check all of sources \nand there may be similar mistakes remained, however.\n--\nTatsuo Ishii\n----------------------------- cut here -----------------------------------\n*** postgresql-v6.4/src/backend/parser/gram.y.orig\tTue Dec 8 11:26:32 1998\n--- postgresql-v6.4/src/backend/parser/gram.y\tTue Dec 8 11:27:00 1998\n***************\n*** 5125,5131 ****\n \t\tif (! (islower(*cp) || isdigit(*cp) || (*cp == '_'))) break;\n \n \tif (*cp != '\\0') {\n! \t\tcp = palloc(strlen(rawid)+1);\n \t\tstrcpy(cp,\"\\\"\");\n \t\tstrcat(cp,rawid);\n \t\tstrcat(cp,\"\\\"\");\n--- 5125,5131 ----\n \t\tif (! (islower(*cp) || isdigit(*cp) || (*cp == '_'))) break;\n \n \tif (*cp != '\\0') {\n! \t\tcp = palloc(strlen(rawid)+3);\n \t\tstrcpy(cp,\"\\\"\");\n \t\tstrcat(cp,rawid);\n \t\tstrcat(cp,\"\\\"\");\n*** postgresql-v6.4/src/backend/catalog/index.c.orig\tTue Dec 8 11:41:20 1998\n--- postgresql-v6.4/src/backend/catalog/index.c\tTue Dec 8 14:14:29 1998\n***************\n*** 649,655 ****\n \tvalue[Anum_pg_attribute_attcacheoff - 1] = Int32GetDatum(-1);\n \n \tinit_tuple = heap_addheader(Natts_pg_attribute,\n! \t\t\t\t\t\t\t\tsizeof *(indexRelation->rd_att->attrs[0]),\n \t\t\t\t\t\t\t (char *) (indexRelation->rd_att->attrs[0]));\n \n \thasind = false;\n--- 649,655 ----\n \tvalue[Anum_pg_attribute_attcacheoff - 1] = Int32GetDatum(-1);\n \n \tinit_tuple = heap_addheader(Natts_pg_attribute,\n! \t\t\t\t ATTRIBUTE_TUPLE_SIZE,\n \t\t\t\t\t\t\t (char *) (indexRelation->rd_att->attrs[0]));\n \n \thasind = false;\n***************\n*** 689,695 ****\n \t\t */\n \t\tmemmove(GETSTRUCT(cur_tuple),\n \t\t\t\t(char *) indexTupDesc->attrs[i],\n! \t\t\t\tsizeof(FormData_pg_attribute));\n \n \t\tvalue[Anum_pg_attribute_attnum - 1] = Int16GetDatum(i + 1);\n \n--- 689,695 ----\n \t\t */\n \t\tmemmove(GETSTRUCT(cur_tuple),\n \t\t\t\t(char *) indexTupDesc->attrs[i],\n! \t\t\t ATTRIBUTE_TUPLE_SIZE);\n \n \t\tvalue[Anum_pg_attribute_attnum - 1] = Int16GetDatum(i + 1);\n \n",
"msg_date": "Tue, 08 Dec 1998 17:45:34 +0900",
"msg_from": "Tatsuo Ishii <[email protected]>",
"msg_from_op": true,
"msg_subject": "memory destruction in 6.4"
},
{
"msg_contents": "Tatsuo Ishii wrote:\n> \n> While investigating a user's complaint, I have found some memory\n> destructions in 6.4 source using purify.\n> \n> (1) parser/gram.y:fmtId()\n> \n> It writes n+3 bytes into n+1 byte-long memory area if mixed case or\n> non-ascii identifiers given.\n\nCould that be also the cause for the two bugs that I have been reported\nsome time ago occuring when object names contain spaces ?\n\nPostgreSQL 6.4 on RedHat Linux i386, 2.0.36 Kernel\n\nTHE FIRST ONE\n=============\ntest=>create table students (id int4, name text); \nCREATE\ntest=> create view \"my view\" as select * from students;\nCREATE\ntest=> select * from \"my view\";\nERROR: nodeRead: Bad type 0 \n\n\nTHE SECOND ONE\n==============\ntest=> create sequence student_id;\nCREATE\ntest=> create table \"my students\" (id int4 default\nnextval('student_id'), name text);\nERROR: my: Table does not exist.\n\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Tue, 08 Dec 1998 11:32:52 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] memory destruction in 6.4"
},
{
"msg_contents": ">Tatsuo Ishii wrote:\n>> \n>> While investigating a user's complaint, I have found some memory\n>> destructions in 6.4 source using purify.\n>> \n>> (1) parser/gram.y:fmtId()\n>> \n>> It writes n+3 bytes into n+1 byte-long memory area if mixed case or\n>> non-ascii identifiers given.\n>\n>Could that be also the cause for the two bugs that I have been reported\n>some time ago occuring when object names contain spaces ?\n>\n>PostgreSQL 6.4 on RedHat Linux i386, 2.0.36 Kernel\n>\n>THE FIRST ONE\n>=============\n>test=>create table students (id int4, name text); \n>CREATE\n>test=> create view \"my view\" as select * from students;\n>CREATE\n>test=> select * from \"my view\";\n>ERROR: nodeRead: Bad type 0 \n>\n>\n>THE SECOND ONE\n>==============\n>test=> create sequence student_id;\n>CREATE\n>test=> create table \"my students\" (id int4 default\n>nextval('student_id'), name text);\n>ERROR: my: Table does not exist.\n\nSorry, but my patches seem not to fix your problems.\n--\nTatsuo Ishii\n",
"msg_date": "Tue, 08 Dec 1998 18:45:14 +0900",
"msg_from": "Tatsuo Ishii <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] memory destruction in 6.4 "
},
{
"msg_contents": "Applied to both trees.\n\n\n\n> While investigating a user's complaint, I have found some memory\n> destructions in 6.4 source using purify.\n> \n> (1) parser/gram.y:fmtId()\n> \n> It writes n+3 bytes into n+1 byte-long memory area if mixed case or\n> non-ascii identifiers given.\n> \n> (2) catalog/index.c:\n> \n> ATTRIBUTE_TUPLE_SIZE bytes are allocated but\n> sizeof(FormData_pg_attribute) bytes are written. Note that\n> ATTRIBUTE_TUPLE_SIZE is smaller than\n> sizeof(FormData_pg_attribute). (for example, on solaris 2.6,\n> ATTRIBUTE_TUPLE_SIZE is 3 bytes smaller).\n> \n> Attached patches try to fix the problem. I do not check all of sources \n> and there may be similar mistakes remained, however.\n> --\n> Tatsuo Ishii\n> ----------------------------- cut here -----------------------------------\n> *** postgresql-v6.4/src/backend/parser/gram.y.orig\tTue Dec 8 11:26:32 1998\n> --- postgresql-v6.4/src/backend/parser/gram.y\tTue Dec 8 11:27:00 1998\n> ***************\n> *** 5125,5131 ****\n> \t\tif (! (islower(*cp) || isdigit(*cp) || (*cp == '_'))) break;\n> \n> \tif (*cp != '\\0') {\n> ! \t\tcp = palloc(strlen(rawid)+1);\n> \t\tstrcpy(cp,\"\\\"\");\n> \t\tstrcat(cp,rawid);\n> \t\tstrcat(cp,\"\\\"\");\n> --- 5125,5131 ----\n> \t\tif (! (islower(*cp) || isdigit(*cp) || (*cp == '_'))) break;\n> \n> \tif (*cp != '\\0') {\n> ! \t\tcp = palloc(strlen(rawid)+3);\n> \t\tstrcpy(cp,\"\\\"\");\n> \t\tstrcat(cp,rawid);\n> \t\tstrcat(cp,\"\\\"\");\n> *** postgresql-v6.4/src/backend/catalog/index.c.orig\tTue Dec 8 11:41:20 1998\n> --- postgresql-v6.4/src/backend/catalog/index.c\tTue Dec 8 14:14:29 1998\n> ***************\n> *** 649,655 ****\n> \tvalue[Anum_pg_attribute_attcacheoff - 1] = Int32GetDatum(-1);\n> \n> \tinit_tuple = heap_addheader(Natts_pg_attribute,\n> ! \t\t\t\t\t\t\t\tsizeof *(indexRelation->rd_att->attrs[0]),\n> \t\t\t\t\t\t\t (char *) (indexRelation->rd_att->attrs[0]));\n> \n> \thasind = false;\n> --- 649,655 ----\n> \tvalue[Anum_pg_attribute_attcacheoff - 1] = Int32GetDatum(-1);\n> \n> \tinit_tuple = heap_addheader(Natts_pg_attribute,\n> ! \t\t\t\t ATTRIBUTE_TUPLE_SIZE,\n> \t\t\t\t\t\t\t (char *) (indexRelation->rd_att->attrs[0]));\n> \n> \thasind = false;\n> ***************\n> *** 689,695 ****\n> \t\t */\n> \t\tmemmove(GETSTRUCT(cur_tuple),\n> \t\t\t\t(char *) indexTupDesc->attrs[i],\n> ! \t\t\t\tsizeof(FormData_pg_attribute));\n> \n> \t\tvalue[Anum_pg_attribute_attnum - 1] = Int16GetDatum(i + 1);\n> \n> --- 689,695 ----\n> \t\t */\n> \t\tmemmove(GETSTRUCT(cur_tuple),\n> \t\t\t\t(char *) indexTupDesc->attrs[i],\n> ! \t\t\t ATTRIBUTE_TUPLE_SIZE);\n> \n> \t\tvalue[Anum_pg_attribute_attnum - 1] = Int16GetDatum(i + 1);\n> \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, 12 Dec 1998 23:37:38 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] memory destruction in 6.4"
},
{
"msg_contents": "Added to TODO:\n\n\t* views with spaces in view name fail when referenced\n\n\n> Tatsuo Ishii wrote:\n> > \n> > While investigating a user's complaint, I have found some memory\n> > destructions in 6.4 source using purify.\n> > \n> > (1) parser/gram.y:fmtId()\n> > \n> > It writes n+3 bytes into n+1 byte-long memory area if mixed case or\n> > non-ascii identifiers given.\n> \n> Could that be also the cause for the two bugs that I have been reported\n> some time ago occuring when object names contain spaces ?\n> \n> PostgreSQL 6.4 on RedHat Linux i386, 2.0.36 Kernel\n> \n> THE FIRST ONE\n> =============\n> test=>create table students (id int4, name text); \n> CREATE\n> test=> create view \"my view\" as select * from students;\n> CREATE\n> test=> select * from \"my view\";\n> ERROR: nodeRead: Bad type 0 \n> \n> \n> THE SECOND ONE\n> ==============\n> test=> create sequence student_id;\n> CREATE\n> test=> create table \"my students\" (id int4 default\n> nextval('student_id'), name text);\n> ERROR: my: Table does not exist.\n> \n> -- \n> Constantin Teodorescu\n> FLEX Consulting Braila, ROMANIA\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, 12 Dec 1998 23:42:07 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] memory destruction in 6.4u"
},
{
"msg_contents": "> Applied to both trees.\n> \n> \n> \n> > While investigating a user's complaint, I have found some memory\n> > destructions in 6.4 source using purify.\n> > \n> > (1) parser/gram.y:fmtId()\n> > \n> > It writes n+3 bytes into n+1 byte-long memory area if mixed case or\n> > non-ascii identifiers given.\n> > \n> > (2) catalog/index.c:\n> > \n> > ATTRIBUTE_TUPLE_SIZE bytes are allocated but\n> > sizeof(FormData_pg_attribute) bytes are written. Note that\n> > ATTRIBUTE_TUPLE_SIZE is smaller than\n> > sizeof(FormData_pg_attribute). (for example, on solaris 2.6,\n> > ATTRIBUTE_TUPLE_SIZE is 3 bytes smaller).\n> > \n> > Attached patches try to fix the problem. I do not check all of sources \n> > and there may be similar mistakes remained, however.\n> > --\n> > Tatsuo Ishii\n\nThank you for taking care of my patches.\n---\nTatsuo Ishii\n",
"msg_date": "Sun, 13 Dec 1998 20:14:28 +0900",
"msg_from": "Tatsuo Ishii <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] memory destruction in 6.4 "
}
] |
[
{
"msg_contents": "\nDue to continued power problems at home (ie: I have no electricity at all\nnow), any email sent to retep.org.uk has not been read since last Friday.\n\nI'm hoping to have the system back on line sometime in the next 2 days.\n\nPeter\n\n-- \nPeter Mount, IT Section\[email protected]\nAnything I write here are my own views, and cannot be taken as being the\nofficial words of Maidstone Borough Council\n\n\n",
"msg_date": "Tue, 8 Dec 1998 08:58:31 +0000 (GMT)",
"msg_from": "Peter T Mount <[email protected]>",
"msg_from_op": true,
"msg_subject": "Problems"
}
] |
[
{
"msg_contents": "Hi,\n\nI tried to compile 6.4 with latest egcs and have experienced\nproblem to compile several files on my Linux x86 box.\nSorry, I don't remember right now actually what files but\nwill reinstall egcs and recompile.\n\nThis problem is mentioned in egcs FAQ:\nhttp://egcs.cygnus.com/faq.html/#asmclobber\n\nEgcs becomes popular and I think it would be good if someone\ncould fix sources to be egcs 1.1.1 compilant :-)\n\n\tRegards,\n\n\t\tOleg\n_____________________________________________________________\nOleg Bartunov, sci.researcher, hostmaster of AstroNet,\nSternberg Astronomical Institute, Moscow University (Russia)\nInternet: [email protected], http://www.sai.msu.su/~megera/\nphone: +007(095)939-16-83, +007(095)939-23-83\n\n",
"msg_date": "Tue, 8 Dec 1998 18:00:21 +0300 (MSK)",
"msg_from": "Oleg Bartunov <[email protected]>",
"msg_from_op": true,
"msg_subject": "problem compiling with egcs 1.1.1"
},
{
"msg_contents": "> I tried to compile 6.4 with latest egcs and have experienced\n> problem to compile several files on my Linux x86 box.\n\nI've had success compiling with egcs on Linux in the recent past, using\nthis version from rpms:\n\nReading specs from\n/usr/local/lib/gcc-lib/i686-pc-linux-gnulibc1/egcs-2.91.57/specs\ngcc version egcs-2.91.57 19980901 (egcs-1.1 release)\n\nI don't use this compiler on a regular basis yet, since it has\nundesirable rounding behavior for some floating point operations (e.g.\nlots of trailing \"9\"s in some output which are not present in gcc-2.7.x\nwhich screws up the Postgres regression tests).\n\nBut I can not recall having any fundamental compilation problems, and\ncertainly did not have to rewrite assembler code as suggested by your\nurl reference.\n\nSend some specific examples of problems you are seeing and I can try\nthem on my libc5 machine.\n\n - Tom\n",
"msg_date": "Tue, 08 Dec 1998 15:25:04 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "> Egcs becomes popular and I think it would be good if someone\n> could fix sources to be egcs 1.1.1 compilant :-)\nI am using EGCS 1.1 for the port to Windows NT ( actually Cygnus B20.1 but\nthat is more or less EGCS 1.1 with some Windows related patches). As far as\nI can see there is just a little problem with the libpq++. For some reason I\ncannot link it.\n\nJoost Kraaijeveld\n\nAskesis B.V.\nGroenewoudeseweg 46\n6524VB Nijmegen\ntel: 024-3888063\nfax: 024-3608416\nweb: www.askesis.nl\n\n\n",
"msg_date": "Tue, 8 Dec 1998 16:40:53 +0100",
"msg_from": "\"Joost Kraaijeveld\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "Thomas,\n\ngcc version egcs-2.91.57 19980901 (egcs-1.1 release) has no problem\nto compile 6.4, and 1.1.1 release also compiles 6.4 fine.\nBut several snapshots between 1.1 and 1.1.1 definitely have the problem\nI described. After that I read the FAQ and returns to 1.1 release.\nToday I installed 1.1.1 and recompile 6.4 - no problem !\nProbably egcs people fixed the problem. \n\n\tRegards,\n\t\t\n\t\tOleg\nOn Tue, 8 Dec 1998, Thomas G. Lockhart wrote:\n\n> Date: Tue, 08 Dec 1998 15:25:04 +0000\n> From: \"Thomas G. Lockhart\" <[email protected]>\n> To: Oleg Bartunov <[email protected]>\n> Cc: [email protected]\n> Subject: Re: [HACKERS] problem compiling with egcs 1.1.1\n> \n> > I tried to compile 6.4 with latest egcs and have experienced\n> > problem to compile several files on my Linux x86 box.\n> \n> I've had success compiling with egcs on Linux in the recent past, using\n> this version from rpms:\n> \n> Reading specs from\n> /usr/local/lib/gcc-lib/i686-pc-linux-gnulibc1/egcs-2.91.57/specs\n> gcc version egcs-2.91.57 19980901 (egcs-1.1 release)\n> \n> I don't use this compiler on a regular basis yet, since it has\n> undesirable rounding behavior for some floating point operations (e.g.\n> lots of trailing \"9\"s in some output which are not present in gcc-2.7.x\n> which screws up the Postgres regression tests).\n> \n> But I can not recall having any fundamental compilation problems, and\n> certainly did not have to rewrite assembler code as suggested by your\n> url reference.\n> \n> Send some specific examples of problems you are seeing and I can try\n> them on my libc5 machine.\n> \n> - Tom\n> \n\n_____________________________________________________________\nOleg Bartunov, sci.researcher, hostmaster of AstroNet,\nSternberg Astronomical Institute, Moscow University (Russia)\nInternet: [email protected], http://www.sai.msu.su/~megera/\nphone: +007(095)939-16-83, +007(095)939-23-83\n\n",
"msg_date": "Wed, 9 Dec 1998 17:11:17 +0300 (MSK)",
"msg_from": "Oleg Bartunov <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "> gcc version egcs-2.91.57 19980901 (egcs-1.1 release) has no problem\n> to compile 6.4, and 1.1.1 release also compiles 6.4 fine.\n> Today I installed 1.1.1 and recompile 6.4 - no problem !\n> Probably egcs people fixed the problem.\n\nWhat optimization level do you use? Do you find that the newer version\nof egcs does a good job with floating point rounding, or are the\nregression tests filled with \".999999\" results? I much preferred the\ngcc-2.7.x behavior wrt graceful rounding, and hope that at some point\negcs will also have it...\n\n - Tom\n",
"msg_date": "Wed, 09 Dec 1998 15:26:52 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "> > gcc version egcs-2.91.57 19980901 (egcs-1.1 release) has no problem\n> > to compile 6.4, and 1.1.1 release also compiles 6.4 fine.\n> > Today I installed 1.1.1 and recompile 6.4 - no problem !\n> > Probably egcs people fixed the problem.\n> \n> What optimization level do you use? Do you find that the newer version\n> of egcs does a good job with floating point rounding, or are the\n> regression tests filled with \".999999\" results? I much preferred the\n> gcc-2.7.x behavior wrt graceful rounding, and hope that at some point\n> egcs will also have it...\n\nWhy do people use egcs? It looks like an experimental version of gcc2,\nand I am not big on experimental compilers. I want something that\nworks, 100% of the time.\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": "Wed, 9 Dec 1998 10:48:40 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "\"Thomas G. Lockhart\" <[email protected]> writes:\n> What optimization level do you use? Do you find that the newer version\n> of egcs does a good job with floating point rounding, or are the\n> regression tests filled with \".999999\" results? I much preferred the\n> gcc-2.7.x behavior wrt graceful rounding, and hope that at some point\n> egcs will also have it...\n\nAre you sure you are right to blame the compiler? I should think this\nwould be a C library issue, not the compiler's fault...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 09 Dec 1998 11:15:15 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1 "
},
{
"msg_contents": "> Are you sure you are right to blame the compiler? I should think this\n> would be a C library issue, not the compiler's fault...\n\nWell, I've *got* to blame something! ;)\n\nIn my limited testing, the only variable was the compiler. I did not\nchange the C library. So, one can point fingers at the compiler for not\nbehaving the same as the old compiler, or one can surmise that there is\na deeper story of older C library misbehavior which was covered up by\nthe older compiler in a great conspiracy. I leaned toward blaming the\ncompiler, on the assumption that for most simple math compilers probably\ngenerate inline code rather than going to a library. It may be that for\nany compiler at high optimization levels you tend to see rounding\nproblems since they don't bother cleaning up results.\n\nDon't know what the real story is, just that the \".9999\" behavior\nreminds me of my Unix boxes from a dozen years ago, not my recent ones.\n\nSo, has anyone tried a new egcs with a new glibc2? I figured that since\nOleg is a bleeding-edge kind of guy (new egcs, Linux kernel 2.1.1xx,\netc.) he probably has that new combination.\n\n - Tom\n",
"msg_date": "Wed, 09 Dec 1998 17:58:39 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "On Wed, 9 Dec 1998, Thomas G. Lockhart wrote:\n\n> Date: Wed, 09 Dec 1998 17:58:39 +0000\n> From: \"Thomas G. Lockhart\" <[email protected]>\n> To: Tom Lane <[email protected]>\n> Cc: [email protected]\n> Subject: Re: [HACKERS] problem compiling with egcs 1.1.1\n> \n> > Are you sure you are right to blame the compiler? I should think this\n> > would be a C library issue, not the compiler's fault...\n> \n> Well, I've *got* to blame something! ;)\n> \n> In my limited testing, the only variable was the compiler. I did not\n> change the C library. So, one can point fingers at the compiler for not\n> behaving the same as the old compiler, or one can surmise that there is\n> a deeper story of older C library misbehavior which was covered up by\n> the older compiler in a great conspiracy. I leaned toward blaming the\n> compiler, on the assumption that for most simple math compilers probably\n> generate inline code rather than going to a library. It may be that for\n> any compiler at high optimization levels you tend to see rounding\n> problems since they don't bother cleaning up results.\n> \n> Don't know what the real story is, just that the \".9999\" behavior\n> reminds me of my Unix boxes from a dozen years ago, not my recent ones.\n> \n> So, has anyone tried a new egcs with a new glibc2? I figured that since\n> Oleg is a bleeding-edge kind of guy (new egcs, Linux kernel 2.1.1xx,\n> etc.) he probably has that new combination.\n\nHey, I'm not very bleeding edge man :-) I'm still afraid glibc2\n(I'm slackware based Linuxoid). I'm using egcs almost everyday\nand the only problem I got was compilation of 6.4 with some snapshot\nof 6.4. I'm glad in 1.1.1 release the problem was gone.\n\nBut on production servers ( I have 6 ppros ) I even never installed\negcs. I hate to have additional problem but I want to test perspective\ntool for myself on my private computers to have own opinion.\n\n\tOleg\n> \n> - Tom\n> \n\n_____________________________________________________________\nOleg Bartunov, sci.researcher, hostmaster of AstroNet,\nSternberg Astronomical Institute, Moscow University (Russia)\nInternet: [email protected], http://www.sai.msu.su/~megera/\nphone: +007(095)939-16-83, +007(095)939-23-83\n\n",
"msg_date": "Wed, 9 Dec 1998 21:47:12 +0300 (MSK)",
"msg_from": "Oleg Bartunov <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "On Wed, 9 Dec 1998, Bruce Momjian wrote:\n\n> > > gcc version egcs-2.91.57 19980901 (egcs-1.1 release) has no problem\n> > > to compile 6.4, and 1.1.1 release also compiles 6.4 fine.\n> > > Today I installed 1.1.1 and recompile 6.4 - no problem !\n> > > Probably egcs people fixed the problem.\n> > \n> > What optimization level do you use? Do you find that the newer version\n> > of egcs does a good job with floating point rounding, or are the\n> > regression tests filled with \".999999\" results? I much preferred the\n> > gcc-2.7.x behavior wrt graceful rounding, and hope that at some point\n> > egcs will also have it...\n> \n> Why do people use egcs? It looks like an experimental version of gcc2,\n> and I am not big on experimental compilers. I want something that\n> works, 100% of the time.\n\n\tEGCS is about as 'experimental' as PostgreSQL, actually. From\nwhat I've read about EGCS, its basically GCC in a more open development\nmodel/scenario...those using and playing with it have anon-cvs access to\nthe source code, can follow along with the development, and have very open\nand visible access to the developers...\n\n\tAbout the only 'truly experimental' nature to EGCS is that there\nare alot more ppl whom have access to the source code at development\nstages, compoared to GCC where you basically see a release and that's\nit...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Wed, 9 Dec 1998 22:54:43 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "> On Wed, 9 Dec 1998, Bruce Momjian wrote:\n> \n> > > > gcc version egcs-2.91.57 19980901 (egcs-1.1 release) has no problem\n> > > > to compile 6.4, and 1.1.1 release also compiles 6.4 fine.\n> > > > Today I installed 1.1.1 and recompile 6.4 - no problem !\n> > > > Probably egcs people fixed the problem.\n> > > \n> > > What optimization level do you use? Do you find that the newer version\n> > > of egcs does a good job with floating point rounding, or are the\n> > > regression tests filled with \".999999\" results? I much preferred the\n> > > gcc-2.7.x behavior wrt graceful rounding, and hope that at some point\n> > > egcs will also have it...\n> > \n> > Why do people use egcs? It looks like an experimental version of gcc2,\n> > and I am not big on experimental compilers. I want something that\n> > works, 100% of the time.\n> \n> \tEGCS is about as 'experimental' as PostgreSQL, actually. From\n\nThat bad, huh. :-) Only joking.\n\n> what I've read about EGCS, its basically GCC in a more open development\n> model/scenario...those using and playing with it have anon-cvs access to\n> the source code, can follow along with the development, and have very open\n> and visible access to the developers...\n> \n> \tAbout the only 'truly experimental' nature to EGCS is that there\n> are alot more ppl whom have access to the source code at development\n> stages, compoared to GCC where you basically see a release and that's\n> it...\n\nI agree gcc2 took a long time to become usable, and it seemed to take\ntoo long, and the egcs people think the development model was part of\nthe cause perhaps. I can see that. We add stuff in PostgreSQL, and\nwith so many other people testing, we get good quick feedback that\nallows us to be more aggressive in adding features.\n\nHas anyone compared the code output or speed comparisons?\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": "Wed, 9 Dec 1998 23:01:58 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "On Wed, 9 Dec 1998, Bruce Momjian wrote:\n\n> I agree gcc2 took a long time to become usable, and it seemed to take\n> too long, and the egcs people think the development model was part of\n> the cause perhaps. I can see that. We add stuff in PostgreSQL, and\n> with so many other people testing, we get good quick feedback that\n> allows us to be more aggressive in adding features.\n\n\tAs do they...the egcs mailing list is pretty constant as far as\nbusy is concerned, with bug reports submitted daily on \"todays\" cvs\nrepository...its very scary :)\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Thu, 10 Dec 1998 02:01:41 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "On Wed, Dec 09, 1998 at 05:58:39PM +0000, Thomas G. Lockhart wrote:\n> In my limited testing, the only variable was the compiler. I did not\n> change the C library. So, one can point fingers at the compiler for not\n> behaving the same as the old compiler, or one can surmise that there is\n> a deeper story of older C library misbehavior which was covered up by\n> the older compiler in a great conspiracy. I leaned toward blaming the\n> compiler, on the assumption that for most simple math compilers probably\n> generate inline code rather than going to a library. It may be that for\n> any compiler at high optimization levels you tend to see rounding\n> problems since they don't bother cleaning up results.\n\nFWIW, I understand this to be a problem mostly on the x86 architecture, \nwhere the floating point registers have higher precision than memory\nstorage. Thus the results are somewhat unpredictable, depending on whether\nregisters are spilled into memory. The -ffloat-store will help with\nexplictly named variables, but not temporary intermediate results.\n\nWhile the egcs people may find a nice fix (the whole register spilling\nissue is being worked on), apparently the C standard is very clear that\nthe .9999 behaviour is allowed. So perhaps fixing the postgres code is\nthe best idea in the long run.\n\nAnthony\n",
"msg_date": "Mon, 14 Dec 1998 12:48:39 +0800",
"msg_from": "Anthony Heading <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "> On Wed, Dec 09, 1998 at 05:58:39PM +0000, Thomas G. Lockhart wrote:\n> > In my limited testing, the only variable was the compiler. I did not\n> > change the C library. So, one can point fingers at the compiler for not\n> > behaving the same as the old compiler, or one can surmise that there is\n> > a deeper story of older C library misbehavior which was covered up by\n> > the older compiler in a great conspiracy. I leaned toward blaming the\n> > compiler, on the assumption that for most simple math compilers probably\n> > generate inline code rather than going to a library. It may be that for\n> > any compiler at high optimization levels you tend to see rounding\n> > problems since they don't bother cleaning up results.\n> \n> FWIW, I understand this to be a problem mostly on the x86 architecture, \n> where the floating point registers have higher precision than memory\n> storage. Thus the results are somewhat unpredictable, depending on whether\n> registers are spilled into memory. The -ffloat-store will help with\n> explictly named variables, but not temporary intermediate results.\n\nYes 80-bits vs. 64-bits. If the computation hits memory, you loose\nprecision. A definate problem, and compiler-specific.\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": "Mon, 14 Dec 1998 00:12:31 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
},
{
"msg_contents": "> While the egcs people may find a nice fix (the whole register spilling\n> issue is being worked on), apparently the C standard is very clear \n> that the .9999 behaviour is allowed. So perhaps fixing the postgres \n> code is the best idea in the long run.\n\nThanks for the background. This is probably a good time to go slow, and\nsee if the compiler gets fixed up. Although the .9999 behavior may be\nallowed, afaik most modern compilers seem to clean things up better than\nthat. No point putting work into code to work around an evolving\ncompiler for a single platform, at least not until it is clear that it\nis a permanent feature.\n\n - Tom\n",
"msg_date": "Mon, 14 Dec 1998 05:35:32 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] problem compiling with egcs 1.1.1"
}
] |
[
{
"msg_contents": "This is a Global broadcast to postgreSQL folks. (don't flame me. I\nthought everyone\nMUST know this as it has effects in the future of postgreSQL). How about\nrunning\npostgreSQL on a Linux Super-computer??\n\nInternet is developing a very high performance, high speed 64-bit CPU\nchip to compete directly with Intel's Merced chip! Visit -\nUS East Coast http://f-cpu.tux.org/\nEurope http://euro.f-cpu.ml.org/\nUS West Coast http://f-cpu.dyn.ml.org/\n\nFreedom CPU can be used for standalone PCs, Network computers, SMP and\neven Super computers using Linux. Cost of Freedom CPU will be $30 US.\nIntel Merced CPU will cost about $900.\n\nCost of Linux super computer will be about $70,000 ($70K US dollars)\nLinux will scream on super-computer machines at speeds anywhere\nfrom 200 GIGA-flops to 1000 GIG-flops.\n1 GIGA-flops = 1000 million floating point operations per sec.\n\nWhen main frame computers do not give that kind of power, you MUST\ngo to Linux Super- computers!!\nI had compiled a Linux super computer kit, please\nvisit - http://aldev.8m.com\n\nI would say forget NT altogether and GO for LINUX!!\n\nAldev's First Law: Once a piece of source code is written by someone\nthan you\nmust NOT waste even one minute trying to re-create it!!\n\n\n",
"msg_date": "Tue, 08 Dec 1998 17:38:00 +0000",
"msg_from": "Dev_Al <[email protected]>",
"msg_from_op": true,
"msg_subject": "Participate in \"Freedom 64-bit CPU\" project (PostgreSQL CPU!!)"
}
] |
[
{
"msg_contents": "\n\n-----Original Message-----\nFrom: Dan Gowin \nSent: Tuesday, December 08, 1998 11:46 AM\nTo: 'Tom Lane'; Pgsql Core (E-mail)\nSubject: RE: [HACKERS] Re: [INTERFACES] Re: libpgtcl.dll for Windows \n\n\nTom,\n\tWe ran into the same problem your describing with the Tcl code\nand developed the following solution.\n\tAt my company we built a special library that converts the \"C\"\ndata types that the Postgres engine uses and converts them into Tcl\ndatatypes. We then used this base to build a listener that runs on\nthe UNIX machine that only talks Tcl to other Tcl interpreters. This\nway, the only thing we need to move around from platform to platform\nis Tcl library code for doing front ends.\n\n\tWhat does this all mean? We currently have Tcl/Tk interfaces\nfor our RakenSQL servers that come on a RakenSQL Client CD-ROM for\ninterface development. And we currently are developing a developers\nversion that we will give away for free (2Q 99) to the development \ncommunity. \n\tBut there is a problem. The Tcl extensions we use are not the\nsame as those chosen for \"libpgtcl\". Instead of using an interface\nthat resembles one of the \"ProC\" pre compilers for Oracle, were the\nSQL is passed as a string to a function. And a cursor is used to\nloop through the resulting data. We built SQL into our library as \nan extension to the Tcl language.\n\n\tWe built five main commands, with each having multiple command \noptions. login, logout, sql, query, and sys.\nFor example: \"login\" on a GUI system will display a login \nwindow, if no commands are passed to it. \"logout\" signs off that \nconnection thread. \"sql\" can be passed SQL queries with the result \nreturned as Tcl lists and the database column titles are the names of \nthe lists. \"query\" will return the results of a SQL query in ascii table\n\nformat for use on the Tcl/Tl command console. And \"sys\" is used for \nour remote management channel for doing remote commands like \"backup\",\n\"reboot\", \"shutdown\", \"ps\", \"netstat\", \"restore\" and \"datetime\". \n\tThe \"query\" command still has an interface that returns a\ncursor to the user. We put this in there just incase the user needs\nto have quicker access to larger data sets (It can take some time\nto form very large Tcl list on Win95 systems). \n\n\tUnfortunately, none of this works with pgAccess. We tried doing\nsome hacking on pgAccess, but even if we use an unmodified Tcl/Tk\ninterpreter and add simple blank lines after \"if\" statements, the\ninterpreter core dumps. Also, our GUI login routine is setup with\npasswords and the database that is being logged into. \nFor example: login postgres dd34rf8 master_server template1\nIs a login that does not use our built in GUI login window.\n\nWe are using the latest pgAccess and we develop with Tcl/Tk 8.1 with\nthe latest patches from John Ousterhout.\nmailto:[email protected] \n\n \nJust my two cents.\nDan Gowin\n\n\n\n-----Original Message-----\nFrom: Tom Lane [mailto:[email protected]]\nSent: Tuesday, December 08, 1998 9:59 AM\nTo: Constantin Teodorescu\nCc: Pgsql-Interfaces; PostgreSQL Hacker\nSubject: [HACKERS] Re: [INTERFACES] Re: libpgtcl.dll for Windows \n\n\nConstantin Teodorescu <[email protected]> writes:\n> Joost Kraaijeveld wrote:\n>>>> PS. A lot of people are asking me if there is a libpgtcl.dll for M$\n>>>> Windows. The pgtcl package, written in pure Tcl/Tk is working but\nit's\n>>>> very slooooow. Is there any chance that someone could try to port\nthe\n>>>> libpgtcl library for Windows ?\n\n>> I tried but the functions\n>>\nTcl_CreateFileHandler(/usr/src/pgsql/src/interfaces/libpgtcl/pgtclId.c:6\n79)\n>> and Tcl_DeleteFileHandler\n>> (/usr/src/pgsql/src/interfaces/libpgtcl/pgtclId.c:701) were removed\nfor\n>> non-Unix platforms as of tcl/tk8.0b1 (or thereabouts).\n\nOK, I'm to blame for that code...\n\n> I don't know where that functions are used, but if they are used in\n> functions concerning large objects,\n\nLarge objects aren't the issue; support for async NOTIFY is. The\ncomments at the head of pgtclId.c explain:\n\n Another headache is that Ousterhout keeps changing the Tcl I/O\ninterfaces.\n libpgtcl currently claims to work with Tcl 7.5, 7.6, and 8.0, and each\nof\n 'em is different. Worse, the Tcl_File type went away in 8.0, which\nmeans\n there is no longer any platform-independent way of waiting for file\nready.\n So we now have to use a Unix-specific interface. Grumble.\n\n In the current design, Pg_Notify_FileHandler is a file handler that\n we establish by calling Tcl_CreateFileHandler(). It gets invoked from\n the Tcl event loop whenever the underlying PGconn's socket is\nread-ready.\n We suck up any available data (to clear the OS-level read-ready\ncondition)\n and then transfer any available PGnotify events into the Tcl event\nqueue.\n Eventually these events will be dispatched to Pg_Notify_EventProc.\nWhen\n we do an ordinary PQexec, we must also transfer PGnotify events into\nTcl's\n event queue, since libpq might have read them when we weren't looking.\n\nIf anyone can explain to me a platform-independent way of waiting for\nsocket-read-ready under Tcl 8.0, I'd gladly incorporate it.\n\nI suppose the other alternative is to insert a bunch of \"#ifdef WIN32\"\ncode to provide a platform-specific implementation of this functionality\nfor Windows. But I'm not sure what that would be, and I'm in no\nposition to write or test it anyway. Volunteers?\n\n(If you don't need NOTIFY functionality, it'd probably be possible\njust to #ifdef out the filehandler and associated setup code... but\nI don't see that as a reasonable solution for general use.)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 8 Dec 1998 14:24:00 -0500 ",
"msg_from": "Dan Gowin <[email protected]>",
"msg_from_op": true,
"msg_subject": "FW: [HACKERS] Re: [INTERFACES] Re: libpgtcl.dll for Windows "
}
] |
[
{
"msg_contents": "Since I have upgraded to postgres 6.4, when I do a \"ps -ef | grep post\", I get a list of postmaster\nprocesses. With 6.3.2 of postgres, I would have one postmaster process and many different posgres\nprocesses. Is this normal 6.4? All I do is start the postmaster process with the -i option.\nAny help would be great\n---------\nChris Williams\nSterling Software\nRome, New York\nPhone: (315) 336-0500\nEmail: [email protected]\n\n",
"msg_date": "Tue, 8 Dec 1998 15:07:42 -0500",
"msg_from": "\"Chris Williams\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Question about postmaster processes"
},
{
"msg_contents": "[Charset iso-8859-1 unsupported, filtering to ASCII...]\n> Since I have upgraded to postgres 6.4, when I do a \"ps -ef |\n> grep post\", I get a list of postmaster processes. With 6.3.2\n> of postgres, I would have one postmaster process and many\n> different posgres processes. Is this normal 6.4? All I do is\n> start the postmaster process with the -i option. Any help would\n> be great\n\nWe no longer exec() new postgres backends, but just fork the\npostmaster. Faster, but your platform doesn't seem to support 'ps'\nargs changes that we have made in the backend. What platform are\nyou using?\n\nTo find the postmaster, just get the parent of all the postmaster-named\nprocesses.\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": "Tue, 8 Dec 1998 15:20:34 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Question about postmaster processes"
},
{
"msg_contents": "\nI observed the same situation with: SGI Origin 2000 and IRIX 6.5.\n\nHakan Kalyoncu\n\n\nOn Tue, 8 Dec 1998, Bruce Momjian wrote:\n\n> [Charset iso-8859-1 unsupported, filtering to ASCII...]\n> > Since I have upgraded to postgres 6.4, when I do a \"ps -ef |\n> > grep post\", I get a list of postmaster processes. With 6.3.2\n> > of postgres, I would have one postmaster process and many\n> > different posgres processes. Is this normal 6.4? All I do is\n> > start the postmaster process with the -i option. Any help would\n> > be great\n> \n> We no longer exec() new postgres backends, but just fork the\n> postmaster. Faster, but your platform doesn't seem to support 'ps'\n> args changes that we have made in the backend. What platform are\n> you using?\n> \n> To find the postmaster, just get the parent of all the postmaster-named\n> processes.\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> \n> \n\n",
"msg_date": "Tue, 8 Dec 1998 22:52:06 +0200 (EET)",
"msg_from": "Hakan Kalyoncu <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Question about postmaster processes"
}
] |
[
{
"msg_contents": "\nA while back there was some discussion about NULL. Right now I'm trying\nto do something that 'Practical SQL' seems to think is normal yet brings up\nthe questions from earlier conversations. I'm querying two tables where\nif the second table (specialprice) has nothing that matches sku, I don't \ncare. IIRC this is how I do it in Sybase (don't have the actual calls in\nfront of me so I don't know if I'm mistaken). The following select call:\n\nselect\nproducts.image,products.sizes,products.colors,products.weight,products.categor\ny,products.accessories,products.saleprice,products.ourcost,products.descriptio\nn,isnull(specialprice.specialprice,0.0) from products,specialprice where\nproducts.sku = '28434' and (specialprice.sku = '28434' or specialprice is\nnull);\n\nresults in this:\n\nERROR: No such function 'isnull' with the specified attributes\n\nI thought that isnull() was something that existed in PostgreSQL 6.4, am\nI wrong or is it implemented differently than Sybase? Or am I not making\nsense (it should be easy to explain if I'm not)?\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] flame-mail: /dev/null\n # include <std/disclaimers.h> TEAM-OS2\n Online Searchable Campground Listings http://www.camping-usa.com\n \"There is no outfit less entitled to lecture me about bloat\n than the federal government\" -- Tony Snow\n==========================================================================\n\n\n",
"msg_date": "Tue, 08 Dec 1998 19:56:47 -0500 (EST)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": true,
"msg_subject": "isnull() or is it?"
},
{
"msg_contents": "> I thought that isnull() was something that existed in PostgreSQL 6.4, am\n> I wrong or is it implemented differently than Sybase? Or am I not making\n> sense (it should be easy to explain if I'm not)?\n\nisnull() is on our TODO list.\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": "Tue, 8 Dec 1998 20:24:19 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "> Right now I'm trying to do something that 'Practical SQL' seems to \n> think is normal yet brings up the questions from earlier \n> conversations. I'm querying two tables where\n> if the second table (specialprice) has nothing that matches sku, I \n> don't care.\n\n>From your example query, it seems that you want to substitute 0.0 for\nthe \"specialprice\" field if an entry corresponding to a row in the first\ntable does not exist. That matches up with the capabilities of a left\nouter join, where you would use CASE or COALESCE() to substitute the\nzero for a null. Postgres does not have outer joins yet. \n\nYour example does *not* give you the effect you want, since the inner\njoin you are specifying will not match up non-existent rows. btw,\nneither of my reference/tutorial books mentions ifnull() (or at least\nthey don't have an index entry for it). I'm looking at \"A Guide to the\nSQL Standard\", Date and Darwen, and \"Understanding the New SQL\", Melton\nand Simon.\n\nIn the meantime I think you can mimic the effect using a union:\n\n select products.image, products.sizes, products.colors, \n products.weight, products.category, products.accessories, \n products.saleprice, products.ourcost, products.description, \n specialprice.specialprice\n from products,specialprice\n where (products.sku = '28434')\n and (products.sku = specialprice.sku)\n union\n select products.image, products.sizes, products.colors, \n products.weight, products.category, products.accessories, \n products.saleprice, products.ourcost, products.description, \n 0.0\n from products\n where products.sku = '28434'\n and (products.sku not in (select sku from specialprice);\n\nThat last clause could just have the constant rather than products.sku.\n\nPostgres does have ISNULL in the parser, but that is just equivalent to\nIS NULL.\n\nGood luck.\n\n - Tom\n",
"msg_date": "Wed, 09 Dec 1998 03:07:09 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] isnull() or is it?"
},
{
"msg_contents": "> isnull() is on our TODO list.\n\nIf the one-line description in the TODO list is accurate and complete,\nistm that the SQL92-standard function COALESCE() which I put into the\ncurrent development tree as part of the CASE feature does that and more.\n\nNeither function would allow an inner join to magically become an outer\njoin. Speaking of which, remember that you're planning on helping with\nouter joins for the next release, right? Or at least you haven't turned\nme down yet... :)\n\n - Tom\n",
"msg_date": "Wed, 09 Dec 1998 05:01:55 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "> > isnull() is on our TODO list.\n> \n> If the one-line description in the TODO list is accurate and complete,\n> istm that the SQL92-standard function COALESCE() which I put into the\n> current development tree as part of the CASE feature does that and more.\n> \n> Neither function would allow an inner join to magically become an outer\n> join. Speaking of which, remember that you're planning on helping with\n> outer joins for the next release, right? Or at least you haven't turned\n> me down yet... :)\n\nI reserve the right to fail, though.\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": "Wed, 9 Dec 1998 01:23:07 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] isnull() or is it?t"
}
] |
[
{
"msg_contents": "I noticed that there was an item in the TODO about min/max for OIDs.\nThought that maybe the new type coersion stuff would take care of it by\nconverting OID to int4. It probably did, but something else is unhappy:\n\npostgres=> select min(oid) from t;\npqReadData() -- backend closed the channel unexpectedly.\n\npostgres=> select min(x) from t;\nmin\n---\na\n(1 row)\n\npostgres=> select int4(oid) from t;\n?column?\n--------\n 18570\n 18571\n 18572\n 144716\n(4 rows)\n\npostgres=> select min(int4(oid)) from t;\npqReadData() -- backend closed the channel unexpectedly.\n\nOh well. I'll put it on my list of things to look at...\n\n - Tom\n",
"msg_date": "Wed, 09 Dec 1998 05:11:35 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "OID min/max functions"
}
] |
[
{
"msg_contents": "Got the latest Tcl/Tk 8.0.4 from scriptics, compiled it and installed it\nwith great success.\nRedHat 5.2 i386, Pentium machine, kernel 2.0.36\n\nAll my tcl/tk applications are working fine but :\n\nAfter compiling and installing pltcl packacge, trying to run the tests\ngot the error :\n\nQUERY: insert into T_pkey1 values (1, 'key1-1', 'test key');\nERROR: Load of file /usr/local/pgsql/lib/pltcl.so failed:\n/lib/libtcl8.0.so: undefined symbol: stat \n\nIs there any other library that have to be manually added when compiling\npltcl, same way as crypt was ?\nOr the error comes from tcl library ?\n\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Wed, 09 Dec 1998 09:41:23 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "plTcl test failed"
},
{
"msg_contents": "> Got the latest Tcl/Tk 8.0.4 from scriptics, compiled it and installed it\n> with great success.\n> RedHat 5.2 i386, Pentium machine, kernel 2.0.36\n> \n> All my tcl/tk applications are working fine but :\n> \n> After compiling and installing pltcl packacge, trying to run the tests\n> got the error :\n> \n> QUERY: insert into T_pkey1 values (1, 'key1-1', 'test key');\n> ERROR: Load of file /usr/local/pgsql/lib/pltcl.so failed:\n> /lib/libtcl8.0.so: undefined symbol: stat \n\nLooks like something from the standard C library.\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": "Wed, 9 Dec 1998 10:05:40 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] plTcl test failed"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> > After compiling and installing pltcl packacge, trying to run the tests\n> > got the error :\n> >\n> > QUERY: insert into T_pkey1 values (1, 'key1-1', 'test key');\n> > ERROR: Load of file /usr/local/pgsql/lib/pltcl.so failed:\n> > /lib/libtcl8.0.so: undefined symbol: stat\n> \n> Looks like something from the standard C library.\n\nI succeeded in compiling it but, as usual :-) , adding manually the\npath to libpq, a -lc and a -lcrypt when linking.\nNow pltcl works and even passed the included tests.\n\nHope that 6.4.1 would have less problems finding the needed libraries.\n\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Wed, 09 Dec 1998 17:41:36 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] plTcl test failed"
},
{
"msg_contents": "Constantin Teodorescu <[email protected]> writes:\n> I succeeded in compiling it but, as usual :-) , adding manually the\n> path to libpq, a -lc and a -lcrypt when linking.\n> Now pltcl works and even passed the included tests.\n\n> Hope that 6.4.1 would have less problems finding the needed libraries.\n\nI hope so too. But we won't find out unless someone who is seeing this\nproblem (I'm not, because libcrypt isn't a separate shlib on my machine)\npulls down the current sources and tests to see if the fix that's in\nthere actually works. Preferably before 6.4.1 goes out the door.\nHint hint...\n\n\t\t\tregards, tom lane\n\nPS: grabbing either the development sources or REL6_4 branch from the\nCVS server should do for testing this fix.\n\nPPS: Are we close enough to 6.4.1 that it'd make sense to put up a\nsnapshot on the FTP server?\n",
"msg_date": "Wed, 09 Dec 1998 11:19:46 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] plTcl test failed "
},
{
"msg_contents": "Tom Lane wrote:\n> \n> > Hope that 6.4.1 would have less problems finding the needed libraries.\n> \n> I hope so too. But we won't find out unless someone who is seeing this\n> problem (I'm not, because libcrypt isn't a separate shlib on my machine)\n> pulls down the current sources and tests to see if the fix that's in\n> there actually works. Preferably before 6.4.1 goes out the door.\n\nI can test it the current version on my RedHat 5.2.\n\n> PS: grabbing either the development sources or REL6_4 branch from the\n> CVS server should do for testing this fix.\n\nPlease, specify what .tar.gz and from where should I download in order\nto test it ?\nThe file postgresql.snapshot.tar.gz (5324902 bytes) from 9 Dec 8:01\nwould be fine ?\n\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Wed, 09 Dec 1998 18:30:49 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] plTcl test failed"
},
{
"msg_contents": "Constantin Teodorescu <[email protected]> writes:\n> Please, specify what .tar.gz and from where should I download in order\n> to test it ?\n> The file postgresql.snapshot.tar.gz (5324902 bytes) from 9 Dec 8:01\n> would be fine ?\n\nI didn't know we were still generating daily snaps. Probably that file\nwould work, although I imagine it's the development branch and not the\nREL6_4 branch.\n\nPerhaps we should be generating snapshots of the REL6_4 branch?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 09 Dec 1998 11:45:35 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] plTcl test failed "
},
{
"msg_contents": "Tom Lane wrote:\n> \n> Constantin Teodorescu <[email protected]> writes:\n> > I succeeded in compiling it but, as usual :-) , adding manually the\n> > path to libpq, a -lc when linking.\n> > Now pltcl works and even passed the included tests.\n> \n> > Hope that 6.4.1 would have less problems finding the needed libraries.\n> \n> I hope so too. But we won't find out unless someone who is seeing this\n> problem (I'm not, because libcrypt isn't a separate shlib on my machine)\n> pulls down the current sources and tests to see if the fix that's in\n> there actually works. Preferably before 6.4.1 goes out the door.\n\nGot postgresql.snapshot.tar.gz (5324902 bytes) from 9 Dec 8:01\n\nI have removed any trace of PostgreSQL 6.4 from my system (directories,\nlibraries).\n\nCompiled on my RedHat 5.2 i386 Pentium machine, kernel 2.0.36 with\n\n./configure --with-tcl\n\nI has compiled ok but the same errors occured :\n* for libpgtcl I have to add manually -lcrypt to SHLIBS in Makefile\n* for pltcl I have to add manually -lc\n\nThis time, the .so libraries checked with ldd show correct dynamic links\nto libraries. No more static links.\n\nGood news : one of the errors that I have reported (cannot create tables\nwith name containing spaces having an default sequence auto-increment\nfield) was solved.\n\nBad news : another error regarding view names containing spaces, still\npersist.\n\nI am apologising for reporting it again, but maybe someone will find the\nbug.\n------------------------------------------------------------------------\ntest=>create table students (id int4, name text); \nCREATE\ntest=> create view \"my view\" as select * from students;\nCREATE\ntest=> select * from \"my view\";\nERROR: nodeRead: Bad type 0 \n\nBest regards,\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Fri, 11 Dec 1998 17:42:09 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "Bad libraries again in latest snapshot , libpgtcl and pltcl"
},
{
"msg_contents": "Constantin Teodorescu <[email protected]> writes:\n> Compiled on my RedHat 5.2 i386 Pentium machine, kernel 2.0.36 with\n> ./configure --with-tcl\n> I has compiled ok but the same errors occured :\n> * for libpgtcl I have to add manually -lcrypt to SHLIBS in Makefile\n> * for pltcl I have to add manually -lc\n\nOK, the pltcl problem is easily fixed --- closer inspection of Tcl's\nmakefiles shows that we weren't using the results read from tclConfig.sh\nproperly.\n\nThe other problem turns out to be even nastier than I feared. It seems\nthat on some Unix platforms (evidently including teo's Linux version),\nthe link command that builds a shared library *must* mention any other\nshared libraries that that lib depends on.\n\nOn other Unix platforms, not only is it not necessary to mention\nshared-lib dependencies, but it may actually Not Work.\n\nI find this information in the configure.in file for Tcl, which is\nprobably a fairly reliable source since Tcl/Tk has the same problem\nwe do of one shlib depending on another.\n\n(Curiously, Tcl believes that on Linux dependent shlibs should *not* be\nmentioned. This leads me to think that there is something unusual or\nbroken about teo's system. I don't know what, though.)\n\nAt this point I have no way to know whether adding -lcrypt to the link\ncommand for libpgtcl.so would break things on more systems than it\nfixes ... and so I am not about to just check in that change.\n\nWe could try making it system-dependent, using the info in Tcl's\nconfigure file as a guide, but that doesn't strike me as something\nI want to try to stick in just a few days before 6.4.1 release either.\n\nThe more I learn about this stuff, the more I think that we should get\nout of the business of writing our own shlib build/install rules and\nlet libtool do the work instead.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 13 Dec 1998 19:40:55 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bad libraries again in latest snapshot ,\n\tlibpgtcl and pltcl"
}
] |
[
{
"msg_contents": "\n\t> isnull() is on our TODO list.\n\n\tIt should be called NVL() from \"Null VaLue\" like in Oracle and\nInformix.\n\tisnull() would suggest a boolean return value to me.\n\n\tAndreas\n",
"msg_date": "Wed, 9 Dec 1998 11:29:32 +0100 ",
"msg_from": "Zeugswetter Andreas IZ5 <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "Zeugswetter Andreas IZ5 wrote:\n> \n> > isnull() is on our TODO list.\n> \n> It should be called NVL() from \"Null VaLue\" like in Oracle and\n> Informix.\n> isnull() would suggest a boolean return value to me.\n> \n\n\nWith SyBase, IsNull(X,Y) returns X if X is not null, and Y\nif X is null.\n\nExample:\n\nSELECT reference, name, IsNull(price, 0.00)\nFROM t_items\n\n\nreference name price\n------------ --------------- -------\nW95 Windows95 190.00\nWNT400 Windows NT 4.0 490.00\nLX Linux 0.00\n\n\nrather than:\n\nSELECT reference, name, price,\nFROM t_items\n\n\nreference name price\n------------ --------------- -------\nW95 Windows95 190.00\nWNT400 Windows NT 4.0 490.00\nLX Linux NULL\n\n\n--\nH.Lefebvre\n",
"msg_date": "Wed, 09 Dec 1998 14:56:12 +0000",
"msg_from": "\"H.Lefebvre\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "> \n> \t> isnull() is on our TODO list.\n> \n> \tIt should be called NVL() from \"Null VaLue\" like in Oracle and\n> Informix.\n> \tisnull() would suggest a boolean return value to me.\n> \n> \tAndreas\n> \n> \n\nNVL added to TODO.\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": "Wed, 9 Dec 1998 10:07:23 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "> Zeugswetter Andreas IZ5 wrote:\n> > \n> > > isnull() is on our TODO list.\n> > \n> > It should be called NVL() from \"Null VaLue\" like in Oracle and\n> > Informix.\n> > isnull() would suggest a boolean return value to me.\n> > \n> \n> \n> With SyBase, IsNull(X,Y) returns X if X is not null, and Y\n> if X is null.\n\nWe should support both isnull() and nvl(). On the TODO list now.\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": "Wed, 9 Dec 1998 10:10:07 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "> With SyBase, IsNull(X,Y) returns X if X is not null, and Y\n> if X is null.\n\nWhich is identical in behavior to the SQL92-defined function\nCOALESCE(X,Y) (IsNull() is not in the standard). This is now in the\nPostgres development tree, to be available in the next release. Should\nwe also have the less capable IsNull() available too? COALESCE() has the\nnice feature that it takes an unlimited number of arguments, returning\nthe first non-null result.\n\n - Tom\n",
"msg_date": "Wed, 09 Dec 1998 15:21:53 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "> > With SyBase, IsNull(X,Y) returns X if X is not null, and Y\n> > if X is null.\n> \n> Which is identical in behavior to the SQL92-defined function\n> COALESCE(X,Y) (IsNull() is not in the standard). This is now in the\n> Postgres development tree, to be available in the next release. Should\n> we also have the less capable IsNull() available too? COALESCE() has the\n> nice feature that it takes an unlimited number of arguments, returning\n> the first non-null result.\n\nOh, NVL and isnull are not standard? Then let's just use coalesce. I\nwill remove them from the TODO list.\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": "Wed, 9 Dec 1998 10:47:08 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "On Wed, 9 Dec 1998, Bruce Momjian wrote:\n\n> > > With SyBase, IsNull(X,Y) returns X if X is not null, and Y\n> > > if X is null.\n> > \n> > Which is identical in behavior to the SQL92-defined function\n> > COALESCE(X,Y) (IsNull() is not in the standard). This is now in the\n> > Postgres development tree, to be available in the next release. Should\n> > we also have the less capable IsNull() available too? COALESCE() has the\n> > nice feature that it takes an unlimited number of arguments, returning\n> > the first non-null result.\n> \n> Oh, NVL and isnull are not standard? Then let's just use coalesce. I\n> will remove them from the TODO list.\n\nSurprises me too. I was under the (incorrect) impression that isnull\nwas standard. \n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] flame-mail: /dev/null\n # include <std/disclaimers.h> TEAM-OS2\n Online Searchable Campground Listings http://www.camping-usa.com\n \"There is no outfit less entitled to lecture me about bloat\n than the federal government\" -- Tony Snow\n==========================================================================\n\n\n\n",
"msg_date": "Wed, 9 Dec 1998 11:16:42 -0500 (EST)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "> Surprises me too. I was under the (incorrect) impression that isnull\n> was standard.\n\nIf it is, my books don't bother mentioning it. It would be pretty easy\nto implement, if it *is* standard, since it is just a subset of\nCOALESCE() behavior.\n\nBut I'm a bit worried about the recent trend toward coding in\nnon-standard extensions as *alternatives* to standards-based behavior\njust because a specific product has it. I know it's convenient for some,\nbut there are *lots* of products which have these kinds of things, each\nin a different way, and at some point we would collapse under the weight\nof trying to be all things at once.\n\nNot naming any names, but *$ products seem to be the worst violators of\nthe simplest standards. One might ascribe this to simple incompetence,\nexcept for their no-longer-in-doubt agenda of subverting standards in a\nso-far-successful drive for market dominance. For example, Access\nproduces code of the form\n val = NULL\nas an equivalent to the SQL92-standard\n val IS NULL\n\nWhere my Date book has a sentence:\n\"It is not possible to specify NULL explicitly as an operand of a\nconditional expression - e.g., 'WHERE X = NULL' is illegal.\"\n\nWe now have a parser which violates that, for the sake of Access\ncompatibility. Of course, we do support another extension to SQL92 which\nviolates another rule in the same area:\n\"It is not possible to specify NULL explicitly as a select-item -- e.g.,\n'SELECT NULL' is illegal.\"\n\nOh well. otoh, perhaps some of this would be legal for SQL3, which has a\nsomewhat more developed concept of NULL...\n\n - Tom\n",
"msg_date": "Wed, 09 Dec 1998 18:17:21 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "If it is interesting to someone, we can partially emulate COALESCE\nright now as:\n\ncreate function coalesce(integer) returns integer as\n'declare\n nonullo alias for $1;\nbegin\n if nonullo then\n return nonullo;\n else\n return 0;\n end if;\n end;\n' language 'plpgsql';\nCREATE\n\nselect *,coalesce(comm) from emp where comm is null;\n\nename|empno|job | hiredate|sal |comm|deptno|level| mgr|coalesce\n-----+-----+----------+----------+---------+----+------+-----+----+--------\nBLAKE| 7698|MANAGER |1981-05-01|$2,850.00| | 30| 3|7782| 0\nJONES| 7900|CLERK |1981-12-03|$950.00 | | 30| 2|7782| 0\nCLARK| 7844|SALESMAN |1981-09-08|$1,500.00| | 10| 2|7839| 0\n(3 rows)\n\n-Jose'-\n\n>> > > With SyBase, IsNull(X,Y) returns X if X is not null, and Y\n>> > > if X is null.\n>> > \n>> > Which is identical in behavior to the SQL92-defined function\n>> > COALESCE(X,Y) (IsNull() is not in the standard). This is now in the\n>> > Postgres development tree, to be available in the next release. Should\n>> > we also have the less capable IsNull() available too? COALESCE() has the\n>> > nice feature that it takes an unlimited number of arguments, returning\n>> > the first non-null result.\n>> \n>> Oh, NVL and isnull are not standard? Then let's just use coalesce. I\n>> will remove them from the TODO list.\n\nVV> Surprises me too. I was under the (incorrect) impression that isnull\nVV> was standard. \n\n\n",
"msg_date": "Thu, 10 Dec 1998 14:41:17 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re[2]: AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "On a somewhat related topic, the development parser also now supports\nthe SQL92-standard NULLIF(), which behaves in an opposite sense to the\ncommon IsNull() under discussion. That is, it tests the first argument\nagainst the second, and then returns NULL if they are equal. Weird.\n\n - Tom\n",
"msg_date": "Thu, 10 Dec 1998 16:45:08 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": ">\n> If it is interesting to someone, we can partially emulate COALESCE\n> right now as:\n>\n> create function coalesce(integer) returns integer as\n> 'declare\n> nonullo alias for $1;\n> begin\n> if nonullo then\n> return nonullo;\n> else\n> return 0;\n> end if;\n> end;\n> ' language 'plpgsql';\n> CREATE\n\n Pardon, but you still misuse the fact, that PL/pgSQL's IF\n expression is implicitly casted into a boolean. That's only\n possible for integer values.\n\n Please use\n\n IF nonullo ISNULL THEN\n RETURN 0;\n ELSE\n RETURN nonullo;\n END IF;\n\n instead, because this would work for other types (like text,\n varchar etc.) too.\n\n Since PL functions can be overloaded (like SQL functions), it\n would be possible, but currently not that performant :-(, to\n create such a function for all types required.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 10 Dec 1998 21:26:41 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: Re[2]: AW: [HACKERS] isnull() or is it?t"
}
] |
[
{
"msg_contents": "I have a table I created and it shows up when I do the \\dt from the psql prompt. But when I try to\ndo a select from it I get the following.\n\nselect * from jddattdata;\nERROR: cannot find attribute 1 of relation jddattdata\n\nI haven't inserted any data yet. I tried to drop the table and I get an error also. Any way I can\nclean some system table to help me fix this problem.\nThanks\n---------\nChris Williams\nSterling Software\nRome, New York\nPhone: (315) 336-0500\nEmail: [email protected]\n\n\n",
"msg_date": "Wed, 9 Dec 1998 08:44:46 -0500",
"msg_from": "\"Chris Williams\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Error with table that exists but postgres says it doesn't"
}
] |
[
{
"msg_contents": "\n\t>Zeugswetter Andreas IZ5 wrote:\n\t>> \n\t>> > isnull() is on our TODO list.\n\t>> \n\t>> It should be called NVL() from \"Null VaLue\" like in\nOracle and\n\t>> Informix.\n\t>> isnull() would suggest a boolean return value to me.\n\t>> \n\t>\n\t>\n\t>With SyBase, IsNull(X,Y) returns X if X is not null, and Y\n\t>if X is null.\n\n\tYes, the difference is only the name of the function.\n\n\tAndreas\n",
"msg_date": "Wed, 9 Dec 1998 14:59:20 +0100 ",
"msg_from": "Zeugswetter Andreas IZ5 <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: AW: [HACKERS] isnull() or is it?t"
},
{
"msg_contents": "> \n> \t>Zeugswetter Andreas IZ5 wrote:\n> \t>> \n> \t>> > isnull() is on our TODO list.\n> \t>> \n> \t>> It should be called NVL() from \"Null VaLue\" like in\n> Oracle and\n> \t>> Informix.\n> \t>> isnull() would suggest a boolean return value to me.\n> \t>> \n> \t>\n> \t>\n> \t>With SyBase, IsNull(X,Y) returns X if X is not null, and Y\n> \t>if X is null.\n> \n> \tYes, the difference is only the name of the function.\n\nisnull() is used by Ingres.\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": "Wed, 9 Dec 1998 10:08:12 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: AW: [HACKERS] isnull() or is it?t"
}
] |
[
{
"msg_contents": "\n\n-----Original Message-----\nFrom: Dan Gowin \nSent: Wednesday, December 09, 1998 10:05 AM\nTo: '[email protected]'\nSubject: RE: [HACKERS] plTcl test failed\n\n\nConstantin,\n\tDo a \"find\" for your Tcl libraries to find out if your Tcl\nwas installed correctly. Rerun the make for Tcl and do a \"make install\"\njust in case. You might also want to run \"ldconfig\" to make sure\nyour libraries are registered correctly. Also, in the library\ndirectories\ncheck to make sure that your Tcl library links are symbolically\nlinked to the correct library version. You can check it by \"cd\"ing\nto the directory and running \"ls -las\" to get the long listing with\nlinks and where they go. If the links are incorrect you could\ncorrect it with the following \"ln -s libtcl8.0.so libtcl.so\", assuming\n\"libtcl.so\" is the correct link and \"libtcl8.0.so\" exists.\n\n\tIt seams to me that \"libtcl.so\" symbolic link can't find the\nfile \"libtcl8.0.so\". And ether the \"libtcl8.0.so\" does not exist or\n\"libtcl.so\" is linked incorrectly.\n\nMy two cents.\nDan Gowin\n\n-----Original Message-----\nFrom: Constantin Teodorescu [mailto:[email protected]]\nSent: Wednesday, December 09, 1998 2:41 AM\nTo: PostgreSQL Hacker\nSubject: [HACKERS] plTcl test failed\n\n\nGot the latest Tcl/Tk 8.0.4 from scriptics, compiled it and installed it\nwith great success.\nRedHat 5.2 i386, Pentium machine, kernel 2.0.36\n\nAll my tcl/tk applications are working fine but :\n\nAfter compiling and installing pltcl packacge, trying to run the tests\ngot the error :\n\nQUERY: insert into T_pkey1 values (1, 'key1-1', 'test key');\nERROR: Load of file /usr/local/pgsql/lib/pltcl.so failed:\n/lib/libtcl8.0.so: undefined symbol: stat \n\nIs there any other library that have to be manually added when compiling\npltcl, same way as crypt was ?\nOr the error comes from tcl library ?\n\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Wed, 9 Dec 1998 10:38:30 -0500 ",
"msg_from": "Dan Gowin <[email protected]>",
"msg_from_op": true,
"msg_subject": "FW: [HACKERS] plTcl test failed"
}
] |
[
{
"msg_contents": "Sorry, if I am posting to the wrong group.\n\nI am using Unixware 7 and tring to compile postgresql 6.4\n\nI am getting this error:\ngmake[2]: Entering directory `/usr/src/pgsql/src/interfaces/libpq++'\nc++ -I../../backend -I../../include -I../../interfaces/libpq -I../../include -I\n../../backend -O -K i486,host,inline,loop_unroll,alloca -Dsvr4 -K PIC -c pg\nenv.cc -o pgenv.o \nc++: cannot specify -o with -c and multiple compilations\n\nthen it stops.\n\nThanks for all the help.\n\n",
"msg_date": "Wed, 09 Dec 1998 11:08:55 -0600",
"msg_from": "\"Jim Nicholson\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Trouble with compile of 6.4 under UW7"
}
] |
[
{
"msg_contents": "> I applied every patch but it doesn't work yet, I belive I have to\n> create min and max aggregates at this point. Am I right ?\n> If so, may you help me with this? Thanks\n\nAssuming that all the patches were packaged, sent, received, and applied\ncorrectly, then I would guess that you have not yet done a full\ndump/reload of your installation. The pg_aggregate table changed so a\nnew initdb must be done.\n\nYou may be able to work around this by declaring the aggregates, but\nit's probably more reliable to take the hit and upgrade the database\ncatalogs.\n\nIf this is not possible for you, then we can figure out the right\n\"CREATE AGGREGATE\" declaration to get you started.\n\n - Tom\n",
"msg_date": "Thu, 10 Dec 1998 16:35:33 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: MIN(text)/MAX(text)"
}
] |
[
{
"msg_contents": "Hello all,\n\nWith the precious help of \"Hiroshi Inoue\" <[email protected]> (that I\nwould like to thank for his help) who has worked on libpgtcl in order to\nrun under Microsoft Windows, I have the pleasure to announce the\navailability of PgAccess version 0.93 for both worlds, Unix and Windows.\nIt is supposed to work also under Microsoft Windows NT, I didn't have\none around here to check it but I expect that it would run fine.\n\nI have tested this version with Tcl/Tk 8.0.4 for Windows 95 & NT,\navailable from :\nftp://ftp.scriptics.com/pub/tcl/tcl8_0/tcl804.exe \n\nIt works fine, the same speed as on my Linux box. The most important\nchanges in 0.93 were made in order to preserve the same look & feel on\nboth platforms.\n\nAlso, 0.93 has the ability to define owner's preferences for widget\nfonts, solving some problems with national languages and different\nscreen resolutions.\n\nThe default font for \"fixed\" on Windows platform has been chosen\n\"Terminal 8\" but I would like more the font used for MS-Dos shell\nwindows style bitmap \"5 x 12\". I don't know how to declare that font to\nTcl so maybe someone with more skills in windows who will find the\nproper name, please send me a note.\n\nThe pgaccess-0.93.tar.gz has now 111 Kbytes and includes the current\nlibpgtcl.dll and libpq.dll needed for run PgAccess under Windows. In\nREADME file there is some information how to use them.\n\nThe package is available, as usual, from :\nhttp://www.flex.ro/pgaccess \n or\nftp://ftp.flex.ro/pgaccess/pub/pgaccess/pgaccess-0.93.tar.gz\n\nFor PostgreSQL developers : Please include this version in the current\ndevelopment tree for PostgreSQL 6.4.1\n\nHappy PostgreSQL-ing to all !\n\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Thu, 10 Dec 1998 20:58:00 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "PgAccess version 0.93 (for Unix and Windows) has been released!"
},
{
"msg_contents": "Installed in both release trees.\n\n> Hello all,\n> \n> With the precious help of \"Hiroshi Inoue\" <[email protected]> (that I\n> would like to thank for his help) who has worked on libpgtcl in order to\n> run under Microsoft Windows, I have the pleasure to announce the\n> availability of PgAccess version 0.93 for both worlds, Unix and Windows.\n> It is supposed to work also under Microsoft Windows NT, I didn't have\n> one around here to check it but I expect that it would run fine.\n> \n> I have tested this version with Tcl/Tk 8.0.4 for Windows 95 & NT,\n> available from :\n> ftp://ftp.scriptics.com/pub/tcl/tcl8_0/tcl804.exe \n> \n> It works fine, the same speed as on my Linux box. The most important\n> changes in 0.93 were made in order to preserve the same look & feel on\n> both platforms.\n> \n> Also, 0.93 has the ability to define owner's preferences for widget\n> fonts, solving some problems with national languages and different\n> screen resolutions.\n> \n> The default font for \"fixed\" on Windows platform has been chosen\n> \"Terminal 8\" but I would like more the font used for MS-Dos shell\n> windows style bitmap \"5 x 12\". I don't know how to declare that font to\n> Tcl so maybe someone with more skills in windows who will find the\n> proper name, please send me a note.\n> \n> The pgaccess-0.93.tar.gz has now 111 Kbytes and includes the current\n> libpgtcl.dll and libpq.dll needed for run PgAccess under Windows. In\n> README file there is some information how to use them.\n> \n> The package is available, as usual, from :\n> http://www.flex.ro/pgaccess \n> or\n> ftp://ftp.flex.ro/pgaccess/pub/pgaccess/pgaccess-0.93.tar.gz\n> \n> For PostgreSQL developers : Please include this version in the current\n> development tree for PostgreSQL 6.4.1\n> \n> Happy PostgreSQL-ing to all !\n> \n> -- \n> Constantin Teodorescu\n> FLEX Consulting Braila, ROMANIA\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, 12 Dec 1998 23:44:17 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] PgAccess version 0.93 (for Unix and Windows) has been\n\treleased!"
}
] |
[
{
"msg_contents": "When are we doing it? With two development trees, we don't seem to have\nany motivation to release it.\n\nI think we should release 6.4.1 soon.\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": "Thu, 10 Dec 1998 14:16:14 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "6.4.1 release"
},
{
"msg_contents": "Bruce Momjian <[email protected]> writes:\n> I think we should release 6.4.1 soon.\n\nWell, portability bug reports and fixes still seem to be flowing in\nat a great rate, so I don't think we can declare 6.4 frozen yet.\nMaybe a 6.4.1 shortly to get the known problems, and expect 6.4.2\nin another month or two?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 10 Dec 1998 21:43:00 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release "
},
{
"msg_contents": ">Bruce Momjian <[email protected]> writes:\n>> I think we should release 6.4.1 soon.\n>\n>Well, portability bug reports and fixes still seem to be flowing in\n>at a great rate, so I don't think we can declare 6.4 frozen yet.\n>Maybe a 6.4.1 shortly to get the known problems, and expect 6.4.2\n>in another month or two?\n\nI think at least large object stuffs should be fixed(just a \"select\nlo_import('/foo/bar')\" easily kills backend) before 6.4.1. I've been\nlooking into codes for sometime but have not found complete fixes yet.\n--\nTatsuo Ishii\[email protected]\n",
"msg_date": "Fri, 11 Dec 1998 11:58:46 +0900",
"msg_from": "Tatsuo Ishii <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release "
},
{
"msg_contents": "> >Bruce Momjian <[email protected]> writes:\n> >> I think we should release 6.4.1 soon.\n> >\n> >Well, portability bug reports and fixes still seem to be flowing in\n> >at a great rate, so I don't think we can declare 6.4 frozen yet.\n> >Maybe a 6.4.1 shortly to get the known problems, and expect 6.4.2\n> >in another month or two?\n> \n> I think at least large object stuffs should be fixed(just a \"select\n> lo_import('/foo/bar')\" easily kills backend) before 6.4.1. I've been\n> looking into codes for sometime but have not found complete fixes yet.\n\nI thought we already had a large object fix in the two trees already?\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": "Thu, 10 Dec 1998 23:16:36 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
},
{
"msg_contents": "On Thu, 10 Dec 1998, Tom Lane wrote:\n\n> Bruce Momjian <[email protected]> writes:\n> > I think we should release 6.4.1 soon.\n> \n> Well, portability bug reports and fixes still seem to be flowing in\n> at a great rate, so I don't think we can declare 6.4 frozen yet.\n> Maybe a 6.4.1 shortly to get the known problems, and expect 6.4.2\n> in another month or two?\n\n\tWorks for me...let's say next Friday?\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Fri, 11 Dec 1998 02:15:49 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release "
},
{
"msg_contents": "> On Thu, 10 Dec 1998, Tom Lane wrote:\n> \n> > Bruce Momjian <[email protected]> writes:\n> > > I think we should release 6.4.1 soon.\n> > \n> > Well, portability bug reports and fixes still seem to be flowing in\n> > at a great rate, so I don't think we can declare 6.4 frozen yet.\n> > Maybe a 6.4.1 shortly to get the known problems, and expect 6.4.2\n> > in another month or two?\n> \n> \tWorks for me...let's say next Friday?\n\nCool.\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": "Fri, 11 Dec 1998 01:38:25 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
},
{
"msg_contents": "> > I think at least large object stuffs should be fixed(just a \"select\n> > lo_import('/foo/bar')\" easily kills backend) before 6.4.1. I've been\n> > looking into codes for sometime but have not found complete fixes yet.\n> \n> I thought we already had a large object fix in the two trees already?\n\nSo you fixed inv_api.c? I got cvs header in REL6_4 tree (FreeBSD\n2.2.6-RELEASE). Is this the latest one?\n\n* $Header: /usr/local/cvsroot/pgsql/src/backend/storage/large_object/in\\\nv_api.c,v 1.41 1998/10/06 03:55:43 momjian Exp $\n\nFollowing is a backend-crashing example. Any idea?\n\n(/tmp/html.tar.gz is a 102458 bytes long file)\n\n> select lo_import('/tmp/html.tar.gz');\nblank\n 1: lo_import (typeid = 26, len = 4, typmod = -1, byval = t)\n ----\n\nProgram received signal SIGSEGV, Segmentation fault.\n0x9dc75 in ReleaseBuffer ()\n(gdb) where\n#0 0x9dc75 in ReleaseBuffer ()\n#1 0xa20b3 in inv_write ()\n#2 0x4605f in lo_import ()\n#3 0xd84a9 in fmgr_c ()\n#4 0x3b06a in ExecMakeFunctionResult ()\n#5 0x3b107 in ExecEvalFunc ()\n#6 0x3b3c9 in ExecEvalExpr ()\n#7 0x3b5e6 in ExecTargetList ()\n#8 0x3b79a in ExecProject ()\n#9 0x4139d in ExecResult ()\n#10 0x39c66 in ExecProcNode ()\n#11 0x392c6 in ExecutePlan ()\n#12 0x38cf1 in ExecutorRun ()\n#13 0xaad5b in ProcessQueryDesc ()\n#14 0xaadc6 in ProcessQuery ()\n#15 0xa8d8e in pg_exec_query_dest ()\n#16 0xa8c24 in pg_exec_query ()\n#17 0xaa598 in PostgresMain ()\n#18 0x4c16c in main ()\n(gdb)\n",
"msg_date": "Fri, 11 Dec 1998 23:43:27 +0900",
"msg_from": "Tatsuo Ishii <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release "
},
{
"msg_contents": "> > Works for me...let's say next Friday?\n\nBruce, would you be willing to maintain a Status list for this minor\nrelease? There are two or three patches floating around which would seem\nto be candidates, and I suspect that there may be one or two things\nsomeone (Tom Lane?) thinks I am looking at, but I'm not clear on what is\nresolved and what is not.\n\nSo, a starting list:\n\nTerry M.'s contrib addition\nTatsuo's Multi-Byte patches\nHiroshi's ParseComplexProject() patches\n (should include regression example?)\nConstantin's PgAccess patches\nPossible portability problems with PgAccess and Tcl-8.1\nSize int4 -> size_t in c.h (put off 'til v6.5?)\n\nAnything else?\n\n - Tom\n",
"msg_date": "Fri, 11 Dec 1998 15:21:27 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> > > Works for me...let's say next Friday?\n> \n> Bruce, would you be willing to maintain a Status list for this minor\n> release? There are two or three patches floating around which would seem\n> to be candidates, and I suspect that there may be one or two things\n> someone (Tom Lane?) thinks I am looking at, but I'm not clear on what is\n> resolved and what is not.\n> \n> So, a starting list:\n> \n> Terry M.'s contrib addition\n> Tatsuo's Multi-Byte patches\n> Hiroshi's ParseComplexProject() patches\n> (should include regression example?)\n> Constantin's PgAccess patches\n> Possible portability problems with PgAccess and Tcl-8.1\n> Size int4 -> size_t in c.h (put off 'til v6.5?)\n> \n> Anything else?\n\nI'd like to see Jans 2 patches, the LIMIT one and the one that \nmakes ORDER BY ose index if one is available (actually it omits the \nunnecessary sort node if possible).\n\nIf you are unable to include them without regression tests, I can \ntry to think out some;\n\nI guess the OR index patch is in CVS already.\n\n------------\nHannu\n",
"msg_date": "Fri, 11 Dec 1998 18:02:51 +0200",
"msg_from": "Hannu Krosing <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
},
{
"msg_contents": "> I'd like to see Jans 2 patches, the LIMIT one and the one that \n> makes ORDER BY ose index if one is available (actually it omits the \n> unnecessary sort node if possible).\n> \n> If you are unable to include them without regression tests, I can \n> try to think out some;\n> \n> I guess the OR index patch is in CVS already.\n\nYes, the OR fix is in there. I need to look through my mailbox to see\nif there is anything else that needs applying.\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, 11 Dec 1998 11:25:53 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
},
{
"msg_contents": "\n\nTatsuo Ishii wrote:\n\n> > > I think at least large object stuffs should be fixed(just a \"select\n> > > lo_import('/foo/bar')\" easily kills backend) before 6.4.1. I've been\n> > > looking into codes for sometime but have not found complete fixes yet.\n> >\n> > I thought we already had a large object fix in the two trees already?\n\nI thought I did. And in fact vanilla 6.4 (that include my patch; as I am not a\nregular PostgreSQL developper, I did not set up SUP here) does not suffer the\nspecified bug (system is Linux 2.1.123/GNU libc 2; test-data is 121225 bytes long):\n\nandre=> select lo_import('/tmp/test-data');\nlo_import\n---------\n 18465\n(1 row)\n\nandre=> select lo_unlink(18465);\nlo_unlink\n---------\n 1\n(1 row)\n\n\n> Following is a backend-crashing example. Any idea?\n\n[ snip : backend crash, identical to test case abose, but ... crashing ]\n\nThat's quite similar to the previous problem. Did someone modify buffer handling\nrecently?\n\nTatsuo Ishii, could you try this surrounded with begin/end statements (and email me\n\nthe result of course) ? The large object buffer problem I corrected was partly\nrelated to freed buffers at the automatic commit (at the end of the query path in\nthe backend, outside transactions). The fix was to release buffers earlier. It\nwould be helpful in order to understand the exact problem.\n\nThanks.\n\n--\nPascal ANDRE, Internet and Media Consulting\[email protected]\n\"Use the source, Luke. Be one with the Code.\" -- Linus Torvalds\n\n\n",
"msg_date": "Sat, 12 Dec 1998 09:44:18 +0100",
"msg_from": "\"Pascal Andr���\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
},
{
"msg_contents": "> > > > I think at least large object stuffs should be fixed(just a \"select\n> > > > lo_import('/foo/bar')\" easily kills backend) before 6.4.1. I've been\n> > > > looking into codes for sometime but have not found complete fixes yet.\n> > >\n> > > I thought we already had a large object fix in the two trees already?\n> \n> I thought I did. And in fact vanilla 6.4 (that include my patch; as I am not a\n> regular PostgreSQL developper, I did not set up SUP here) does not suffer the\n> specified bug (system is Linux 2.1.123/GNU libc 2; test-data is 121225 bytes long):\n\nLinux boxes does not seem to have any problem with large object. As\nfar as I know, the bug appears on FreeBSD and Solaris/Sparc. I'm not\nsure about other platforms.\n\n> Tatsuo Ishii, could you try this surrounded with begin/end statements (and email me\n> \n> the result of course) ? The large object buffer problem I corrected was partly\n> related to freed buffers at the automatic commit (at the end of the query path in\n> the backend, outside transactions). The fix was to release buffers earlier. It\n> would be helpful in order to understand the exact problem.\n> \n> Thanks.\n\nOk. Here are results.\n\no FreeBSD: surrounding with begin/end prevents backend-crash.\n\no Solaris: surrounding with begin/end does not help.\n\nI guess there may be another bug that is different from the one you\nmentioned. Solairs seems to be very \"sensitive\" to that kind of\nproblem?:-)\n--\nTatsuo Ishii\n",
"msg_date": "Sun, 13 Dec 1998 00:09:22 +0900",
"msg_from": "Tatsuo Ishii <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release "
},
{
"msg_contents": "On Fri, 11 Dec 1998, Hannu Krosing wrote:\n\n> Thomas G. Lockhart wrote:\n> > \n> > > > Works for me...let's say next Friday?\n> > \n> > Bruce, would you be willing to maintain a Status list for this minor\n> > release? There are two or three patches floating around which would seem\n> > to be candidates, and I suspect that there may be one or two things\n> > someone (Tom Lane?) thinks I am looking at, but I'm not clear on what is\n> > resolved and what is not.\n> > \n> > So, a starting list:\n> > \n> > Terry M.'s contrib addition\n> > Tatsuo's Multi-Byte patches\n> > Hiroshi's ParseComplexProject() patches\n> > (should include regression example?)\n> > Constantin's PgAccess patches\n> > Possible portability problems with PgAccess and Tcl-8.1\n> > Size int4 -> size_t in c.h (put off 'til v6.5?)\n> > \n> > Anything else?\n> \n> I'd like to see Jans 2 patches, the LIMIT one and the one that \n> makes ORDER BY ose index if one is available (actually it omits the \n> unnecessary sort node if possible).\n\n\tv6.4.1 is a bug fix release, no extra features will be added...Jan\nis planning on releasing a 'LIMIT-patch' after v6.4.1 so that if someone\nwants it, its there, but it will not be in the standard distribution until\nv6.5...\n\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sat, 12 Dec 1998 16:06:22 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
},
{
"msg_contents": "> > > I think at least large object stuffs should be fixed(just a \"select\n> > > lo_import('/foo/bar')\" easily kills backend) before 6.4.1. I've been\n> > > looking into codes for sometime but have not found complete fixes yet.\n> > \n> > I thought we already had a large object fix in the two trees already?\n> \n> So you fixed inv_api.c? I got cvs header in REL6_4 tree (FreeBSD\n> 2.2.6-RELEASE). Is this the latest one?\n> \n> * $Header: /usr/local/cvsroot/pgsql/src/backend/storage/large_object/in\\\n> v_api.c,v 1.41 1998/10/06 03:55:43 momjian Exp $\n> \n> Following is a backend-crashing example. Any idea?\n> \n> (/tmp/html.tar.gz is a 102458 bytes long file)\n> \n> > select lo_import('/tmp/html.tar.gz');\n> blank\n> 1: lo_import (typeid = 26, len = 4, typmod = -1, byval = t)\n> ----\n\nFixed. Since I re-designed the heap access API, the bug was crystal\nclear as soon as I looked at the code. Scarry when I can figure out the\nbackend code so quickly.\n\nPatch applied to both trees.\n\n---------------------------------------------------------------------------\n\n\n*** ./inv_api.c.orig\tSun Dec 13 00:05:01 1998\n--- ./inv_api.c\tSun Dec 13 00:06:51 1998\n***************\n*** 549,556 ****\n \t\t\t\ttuplen = inv_wrnew(obj_desc, buf, nbytes - nwritten);\n \t\t\telse\n \t\t\t\ttuplen = inv_wrold(obj_desc, buf, nbytes - nwritten, tuple, buffer);\n \t\t}\n- \t\tReleaseBuffer(buffer);\n \n \t\t/* move pointers past the amount we just wrote */\n \t\tbuf += tuplen;\n--- 549,556 ----\n \t\t\t\ttuplen = inv_wrnew(obj_desc, buf, nbytes - nwritten);\n \t\t\telse\n \t\t\t\ttuplen = inv_wrold(obj_desc, buf, nbytes - nwritten, tuple, buffer);\n+ \t\t\tReleaseBuffer(buffer);\n \t\t}\n \n \t\t/* move pointers past the amount we just wrote */\n \t\tbuf += tuplen;\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": "Sun, 13 Dec 1998 00:10:46 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
},
{
"msg_contents": "> > > Works for me...let's say next Friday?\n> \n> Bruce, would you be willing to maintain a Status list for this minor\n> release? There are two or three patches floating around which would seem\n> to be candidates, and I suspect that there may be one or two things\n> someone (Tom Lane?) thinks I am looking at, but I'm not clear on what is\n> resolved and what is not.\n\nOnly way to do it is cvs log, and run through my script, I get a nice\nlog of changes. I will do it soon, now that I have cleaned out my\nmailbox of patches people wanted applied.\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": "Sun, 13 Dec 1998 00:12:08 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
},
{
"msg_contents": "> > > > I think at least large object stuffs should be fixed(just a \"select\n> > > > lo_import('/foo/bar')\" easily kills backend) before 6.4.1. I've been\n> > > > looking into codes for sometime but have not found complete fixes yet.\n> > > \n> > > I thought we already had a large object fix in the two trees already?\n> > \n> > So you fixed inv_api.c? I got cvs header in REL6_4 tree (FreeBSD\n> > 2.2.6-RELEASE). Is this the latest one?\n> > \n> > * $Header: /usr/local/cvsroot/pgsql/src/backend/storage/large_object/in\\\n> > v_api.c,v 1.41 1998/10/06 03:55:43 momjian Exp $\n> > \n> > Following is a backend-crashing example. Any idea?\n> > \n> > (/tmp/html.tar.gz is a 102458 bytes long file)\n> > \n> > > select lo_import('/tmp/html.tar.gz');\n> > blank\n> > 1: lo_import (typeid = 26, len = 4, typmod = -1, byval = t)\n> > ----\n> \n> Fixed. Since I re-designed the heap access API, the bug was crystal\n> clear as soon as I looked at the code. Scarry when I can figure out the\n> backend code so quickly.\n> \n> Patch applied to both trees.\n\nThanks. Your patches definitely fix the problem on\nFreeBSD. Unfotunately it does not help Solaris/Sparc. I'll look into\nmore on that platform.\n---\nTatsuo Ishii\n",
"msg_date": "Sun, 13 Dec 1998 20:14:37 +0900",
"msg_from": "Tatsuo Ishii <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release "
},
{
"msg_contents": "Marc G. Fournier wrote:\n\n> On Fri, 11 Dec 1998, Hannu Krosing wrote:\n>\n> > I'd like to see Jans 2 patches, the LIMIT one and the one that\n> > makes ORDER BY ose index if one is available (actually it omits the\n> > unnecessary sort node if possible).\n>\n> v6.4.1 is a bug fix release, no extra features will be added...Jan\n> is planning on releasing a 'LIMIT-patch' after v6.4.1 so that if someone\n> wants it, its there, but it will not be in the standard distribution until\n> v6.5...\n\n As it is already in the v6.4-feature-patch you'll find in the\n patches directory on ftp.\n\n Does anyone have experience with the patch suppressing sort\n on ORDER BY? If so and experiences are positive, I would also\n give it a try and include it into the feature patch of\n v6.4.1.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 14 Dec 1998 10:23:40 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
},
{
"msg_contents": "The Hermit Hacker wrote:\n> \n> On Fri, 11 Dec 1998, Hannu Krosing wrote:\n> \n> > Thomas G. Lockhart wrote:\n> > >\n> > > > > Works for me...let's say next Friday?\n> > >\n> > > Bruce, would you be willing to maintain a Status list for this minor\n> > > release? There are two or three patches floating around which would seem\n> > > to be candidates, and I suspect that there may be one or two things\n> > > someone (Tom Lane?) thinks I am looking at, but I'm not clear on what is\n> > > resolved and what is not.\n> > >\n> > > So, a starting list:\n> > >\n> > > Terry M.'s contrib addition\n> > > Tatsuo's Multi-Byte patches\n> > > Hiroshi's ParseComplexProject() patches\n> > > (should include regression example?)\n> > > Constantin's PgAccess patches\n> > > Possible portability problems with PgAccess and Tcl-8.1\n> > > Size int4 -> size_t in c.h (put off 'til v6.5?)\n> > >\n> > > Anything else?\n> >\n> > I'd like to see Jans 2 patches, the LIMIT one and the one that\n> > makes ORDER BY ose index if one is available (actually it omits the\n> > unnecessary sort node if possible).\n> \n> v6.4.1 is a bug fix release, no extra features will be added...Jan\n> is planning on releasing a 'LIMIT-patch' after v6.4.1 so that if someone\n> wants it, its there, but it will not be in the standard distribution until\n> v6.5...\n\nAw... ;(p \n\nWhen not accepting the ORDER BY patch into 6.4 the explanation was \nthat soon after 6.4 there will be a performance-enchancement release \n(no new features) called 6.4.1 for things like this.\n\nI guessed this would be the one.\n\nAnd for me the ORDER BY patch is actually a bugfix - it enables me to \nget at my huge table in a predictable order. Without the patch the same \nquery used to kill the computer.\n\n(BTW, how could we test for such cases that exhaust some resources?\n we obviously can't include them in standard regression tests)\n\nAnd, I don't need the LIMIT patch nearly as much as ORDER BY, \nI can do declare_cursor-move-fetch myself.\n\nIt would still be very nice if moving past end would not render \nthe cursor unusable.\n\n-----\nHannu\n",
"msg_date": "Mon, 14 Dec 1998 20:49:22 +0200",
"msg_from": "Hannu Krosing <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 6.4.1 release"
}
] |
[
{
"msg_contents": "subscribe\n",
"msg_date": "Thu, 10 Dec 1998 21:10:21 +0100 (MET)",
"msg_from": "Frank Ridderbusch <[email protected]>",
"msg_from_op": true,
"msg_subject": "subscribe"
}
] |
[
{
"msg_contents": "Constantin,\n\tPgAccess work great on Windows NT 4.0 patch 3. Great thanks, it\nworks\nfine. It works fine on Windows 95 and Windows 98 on our test bed also.\nThis will really make the boyz in Redmond sweat.\n\nCongratulations a great piece of code.\nD. Gowin\n\n-----Original Message-----\nFrom: Constantin Teodorescu [mailto:[email protected]]\nSent: Thursday, December 10, 1998 1:58 PM\nTo: PostgreSQL Interfaces; PostgreSQL Hacker;\[email protected]; [email protected]\nSubject: [HACKERS] PgAccess version 0.93 (for Unix and Windows) has been\nreleased!\n\n\nHello all,\n\nWith the precious help of \"Hiroshi Inoue\" <[email protected]> (that I\nwould like to thank for his help) who has worked on libpgtcl in order to\nrun under Microsoft Windows, I have the pleasure to announce the\navailability of PgAccess version 0.93 for both worlds, Unix and Windows.\nIt is supposed to work also under Microsoft Windows NT, I didn't have\none around here to check it but I expect that it would run fine.\n\nI have tested this version with Tcl/Tk 8.0.4 for Windows 95 & NT,\navailable from :\nftp://ftp.scriptics.com/pub/tcl/tcl8_0/tcl804.exe \n\nIt works fine, the same speed as on my Linux box. The most important\nchanges in 0.93 were made in order to preserve the same look & feel on\nboth platforms.\n\nAlso, 0.93 has the ability to define owner's preferences for widget\nfonts, solving some problems with national languages and different\nscreen resolutions.\n\nThe default font for \"fixed\" on Windows platform has been chosen\n\"Terminal 8\" but I would like more the font used for MS-Dos shell\nwindows style bitmap \"5 x 12\". I don't know how to declare that font to\nTcl so maybe someone with more skills in windows who will find the\nproper name, please send me a note.\n\nThe pgaccess-0.93.tar.gz has now 111 Kbytes and includes the current\nlibpgtcl.dll and libpq.dll needed for run PgAccess under Windows. In\nREADME file there is some information how to use them.\n\nThe package is available, as usual, from :\nhttp://www.flex.ro/pgaccess \n or\nftp://ftp.flex.ro/pgaccess/pub/pgaccess/pgaccess-0.93.tar.gz\n\nFor PostgreSQL developers : Please include this version in the current\ndevelopment tree for PostgreSQL 6.4.1\n\nHappy PostgreSQL-ing to all !\n\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Thu, 10 Dec 1998 15:50:31 -0500",
"msg_from": "Dan Gowin <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] PgAccess version 0.93 (for Unix and Windows) has be\n\ten released!"
}
] |
[
{
"msg_contents": "Hi all\n\nI made dll's(libpq.dll,libpgtcl.dll) only for pgaccess.\nIt is a temporary solution for quick fix and the solution for general use \nis necessary.\n\nI made libpgtcl.dll ignoring NOTIFY functionality(not used in pgaccess).\nSorry,I'm not a Tcl expert and can't provide proper code for NOTIFY \nfunctionality in Windows. \n\nAnother problem exists between libpq and libpgtcl in Windows.\nLibpgtcl references global data pgresStatus[] in libpq when using \npg_result -status command. \nBut in Windows libpgtcl doesn't reference pgresStatus[] in the same \nway as UNIX and pg_result -status returns improper values.\n\nAny ideas ?\n\nIn my temporary solution for pgaccess,I used a function interface \nbetween libpq and libpgtcl instead of global data. \n\nThanks\n\nHiroshi Inoue\[email protected]\n\n\n\n",
"msg_date": "Fri, 11 Dec 1998 11:22:36 +0900",
"msg_from": "\"Hiroshi Inoue\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "libpq and libpgtcl in Windows"
},
{
"msg_contents": "Hi all.\nI have changed the subject from \"libpq and libpgtcl in Windows\".\n\nMy main point is\n\n The treatment of pgresStatus[] in libpq must be changed ?\n\nIn case of Windows,global data interface between dll's is \ndifferent from UNIX.\nIf fact,libpgtcl references global data pgresStatus[] in libpq \nand pg_result -status command does not work properly for \nWindows. \n\nI think it is preferable to use function interfaces than gobal \ndata references.\n\nAny ideas ?\n\nThanks\n\nHiroshi Inoue\[email protected]\n\nPS.\n I accepted a lot of mails about libpq and libpgtcl.\n Seems many people thought I am good at libpq,Tcl etc. \n But my work is only the compilation of libpq and libpgtcl.\n\n LIBPQ for WIN32 is the work of Magnus Hagander.\n\n I changed libpgtcl and libpq just a little for pgaccess. \n\n The dll's I provieded are only for pgaccess not for general use.\n",
"msg_date": "Mon, 14 Dec 1998 19:12:23 +0900",
"msg_from": "\"Hiroshi Inoue\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "The problem between libpq and libpgtcl for WIN32"
},
{
"msg_contents": "Hello all.\n\n> -----Original Message-----\n> From: [email protected]\n> [mailto:[email protected]]On Behalf Of Hiroshi Inoue\n> Subject: [INTERFACES] The problem between libpq and libpgtcl for WIN32\n>\n>\n> The treatment of pgresStatus[] in libpq must be changed ?\n>\n> In case of Windows,global data interface between dll's is\n> different from UNIX.\n> If fact,libpgtcl references global data pgresStatus[] in libpq\n> and pg_result -status command does not work properly for\n> Windows.\n>\n> I think it is preferable to use function interfaces than gobal\n> data references.\n>\n> Any ideas ?\n\n I solved this problem by the change in the following.\n\n======== cut here ========\n*** libpq-fe.h.dist Tue Oct 6 23:16:48 1998\n--- libpq-fe.h Tue Dec 15 03:52:34 1998\n***************\n*** 50,56 ****\n--- 50,60 ----\n } ExecStatusType;\n\n /* String descriptions of the ExecStatusTypes */\n+ #ifdef WIN32\n+ __declspec(dllimport) const char * const pgresStatus[];\n+ #else\n extern const char * const pgresStatus[];\n+ #endif\n\n /* PGconn encapsulates a connection to the backend.\n * The contents of this struct are not supposed to be known to\napplications.\n*** libpqdll.def.dist Tue Oct 6 23:05:02 1998\n--- libpqdll.def Tue Dec 15 03:52:44 1998\n***************\n*** 63,68 ****\n lo_unlink @ 60\n lo_import @ 61\n lo_export @ 62\n! pgresStatus @ 63\n PQmblen @ 64\n PQresultErrorMessage @ 65\n--- 63,68 ----\n lo_unlink @ 60\n lo_import @ 61\n lo_export @ 62\n! pgresStatus @ 63 DATA\n PQmblen @ 64\n PQresultErrorMessage @ 65\n======== cut here ========\n\n \"pg_result -status\" operates correctly by this method. Moreover, it is\nthought that this method is easy because the change part is in libpq.\n\n--\nHiroki Kataoka\[email protected]\n\n",
"msg_date": "Wed, 16 Dec 1998 21:27:47 +0900",
"msg_from": "\"Hiroki Kataoka\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [INTERFACES] The problem between libpq and libpgtcl for WIN32"
}
] |
[
{
"msg_contents": "Well, I've started looking through my books for info on joins. The cross\njoin was pretty easy:\n\npostgres=> select * from (a cross join b);\ni| j|i| k\n-+----+-+--\n1|10.1|1|-1\n2|20.2|1|-1\n4| |1|-1\n<snip>\n\nwhich I've put into my copy of the parser.\n\nDoes anyone have a commercial installation which has good support for\nSQL92 joins? I'd like to send some small test cases to verify that I\nunderstand what the behavior should be.\n\nAlso, if anyone has worked with join syntax, outer joins especially, it\nwould be great to get some test case contributions...\n\n - Tom\n",
"msg_date": "Fri, 11 Dec 1998 06:35:39 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "JOIN syntax. Examples?"
},
{
"msg_contents": "On Fri, 11 Dec 1998, Thomas G. Lockhart wrote:\n\n> Well, I've started looking through my books for info on joins. The cross\n> join was pretty easy:\n> \n> postgres=> select * from (a cross join b);\n> i| j|i| k\n> -+----+-+--\n> 1|10.1|1|-1\n> 2|20.2|1|-1\n> 4| |1|-1\n> <snip>\n> \n> which I've put into my copy of the parser.\n> \n> Does anyone have a commercial installation which has good support for\n> SQL92 joins? I'd like to send some small test cases to verify that I\n> understand what the behavior should be.\n> \n> Also, if anyone has worked with join syntax, outer joins especially, it\n> would be great to get some test case contributions...\n\nI have access to several different versions of Oracle at work...if you\nwant to send me your test cases, go for it...work email is\n'[email protected]'...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Fri, 11 Dec 1998 02:52:24 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] JOIN syntax. Examples?"
},
{
"msg_contents": "On Fri, Dec 11, 1998 at 06:35:39AM +0000, Thomas G. Lockhart wrote:\n> Does anyone have a commercial installation which has good support for\n> SQL92 joins? I'd like to send some small test cases to verify that I\n> understand what the behavior should be.\n\nWould Oracle qualify? I hope to get the Linux trial version next week.\n\n> Also, if anyone has worked with join syntax, outer joins especially, it\n> would be great to get some test case contributions...\n\nHmm Oracle's version of outer joins is completely different from Informix.\n\nMichael\n-- \nDr. Michael Meskes, Manager of the Western Branch Office, Datenrevision GmbH\nwork: Cuxhavener Str. 36, D-21149 Hamburg, [email protected]\nhome: Th.-Heuss-Str. 61, D-41812 Erkelenz, [email protected]\nGo SF49ers! Go Rhein Fire! Use Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Fri, 11 Dec 1998 09:24:01 +0100",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] JOIN syntax. Examples?"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> Well, I've started looking through my books for info on joins. The cross\n> join was pretty easy:\n> \n> postgres=> select * from (a cross join b);\n> i| j|i| k\n> -+----+-+--\n> 1|10.1|1|-1\n> 2|20.2|1|-1\n> 4| |1|-1\n> <snip>\n> \n> which I've put into my copy of the parser.\n> \n> Does anyone have a commercial installation which has good support for\n> SQL92 joins? I'd like to send some small test cases to verify that I\n> understand what the behavior should be.\n> \n> Also, if anyone has worked with join syntax, outer joins especially, it\n> would be great to get some test case contributions...\n\nYou could use MS Access.\n\nIt has terrible outer join syntax (compared to Oracle), but I quess it \nis more or less what SQL92 standard prescribes ;(\n\nTo make the queries, just open the query builder and join the tables, \nthen double-click on the join and set the property not to require all \nfrom one of the tables and only matching from the other.\n\nHere is whar it produced for me:\n\nSELECT account.account_nr, domestic_po.bank_order_id\nFROM account LEFT JOIN domestic_po ON account.account_nr =\ndomestic_po.account_nr;\n\nthis requires all fields from account and matching fields from\ndomestic_po\n\nthe oracle equivalent would be:\n\nSELECT account.account_nr, domestic_po.bank_order_id\nFROM account \nWHERE account.account_nr = (+) domestic_po.account_nr;\n\nBTW, I do think Oracle syntax to be much clearer, but I'm not sure \nif we should allow it as and additional syntax.\n\n----------------\nHannu\n",
"msg_date": "Fri, 11 Dec 1998 13:07:10 +0200",
"msg_from": "Hannu Krosing <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] JOIN syntax. Examples?"
},
{
"msg_contents": "> BTW, I do think Oracle syntax to be much clearer, but I'm not sure\n> if we should allow it as and additional syntax.\n\nDo you know of any place I can look up Oracle's syntax? Or if not, could\nI send you a (small) regression test for joins and have you translate\nthat to Oracle's syntax?\n\nThis brings up a question: I would guess that Oracle claims to conform\nto SQL92 (or to be compliant, or to be an extended subset, or ??). afaik\nthis means for many companies that they conform to the \"Entry Level\"\npart of SQL92. What are Oracle's claims on the subject? I'd like to know\nwhere we stand on this relative to them; I've claimed that we are an\n\"extended subset\" but perhaps we are closer than we think...\n\n - Tom\n",
"msg_date": "Fri, 11 Dec 1998 13:52:57 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] JOIN syntax. Examples?"
}
] |
[
{
"msg_contents": "\t>> Does anyone have a commercial installation which has good support\nfor\n\t>> SQL92 joins? I'd like to send some small test cases to verify\nthat I\n\t>> understand what the behavior should be.\n\t>I have access to several different versions of Oracle at work...if\nyou\n>want to send me your test cases, go for it...work email is\n\nOracle is not a good Candidate for SQL92 Syntax.\nThe outer join is completely different. \nA little better is Informix 7.30, and most conformant is probably DB2 V5.\nOracle uses (+) in the join quals like a.id (+) = b.id where table a has\nmissing entries.\nIn Informix you write outer before each table that can have missing entries.\nDB2 has the left outer, right outer, and full outer join syntax.\n\nExample:\nOracle: /* size and free space of all tablespaces (no row in dba_free_space\nif nothing free for this tablespace) */\nselect s.tablespace_name as \"tbsname\", sum(s.bytes)/1024 as \"size\",\n nvl(sum(f.bytes)/1024,0) as \"free\"\nfrom dba_data_files s, dba_free_space f\n where f.TABLESPACE_NAME (+) = s.TABLESPACE_NAME\n group by s.TABLESPACE_NAME;\n\nInformix: /* all tables + indexes if available */\nselect tabname, idxname \nfrom systables t, outer sysindexes i\nwhere t.tabid = i.tabid\n\nDB2: /* all tables + indexes if available */\nSELECT T.NAME, T.CREATOR , I.NAME, I.CREATOR\nFROM SYSIBM.SYSTABLES T LEFT OUTER JOIN SYSIBM.SYSINDEXES I\nON T.NAME = I.TBNAME AND T.CREATOR = I.TBCREATOR\n\n\n\tAndreas\n",
"msg_date": "Fri, 11 Dec 1998 10:26:19 +0100",
"msg_from": "Zeugswetter Andreas IZ5 <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: [HACKERS] JOIN syntax. Examples?"
},
{
"msg_contents": "> Oracle is not a good Candidate for SQL92 Syntax.\n> The outer join is completely different.\n> A little better is Informix 7.30, and most conformant is probably DB2 V5.\n\nYou can now download DB2 5.2 beta for linux from the IBM web-site. Somebody\nwith a big enough computer (and probably net connection, I suppose it's a long\ndownload:) could pick it up to do some testing.\n\nMaarten\n\n",
"msg_date": "Fri, 11 Dec 1998 15:24:57 +0100",
"msg_from": "\"Maarten Boekhold\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] JOIN syntax. Examples?"
}
] |
[
{
"msg_contents": "Announce: Release of PyGreSQL version 2.2\n===============================================\n\nPyGreSQL v2.2 has been released.\nIt is available at: ftp://ftp.druid.net/pub/distrib/PyGreSQL-2.2.tgz.\n\nPostgreSQL is a database system derived from Postgres4.2. It conforms\nto (most of) ANSI SQL and offers many interesting capabilities (C\ndynamic linking for functions or type definition, etc.). This package\nis copyright by the Regents of the University of California, and is\nfreely distributable.\n\nPython is an interpreted programming language. It is object oriented,\nsimple to use (light syntax, simple and straightforward statements), and\nhas many extensions for building GUIs, interfacing with WWW, etc. An\nintelligent web browser (HotJava like) is currently under development\n(November 1995), and this should open programmers many doors. Python is\ncopyrighted by Stichting S Mathematisch Centrum, Amsterdam, The\nNetherlands, and is freely distributable.\n\nPyGreSQL is a python module that interfaces to a PostgreSQL database. It\nembeds the PostgreSQL query library to allow easy use of the powerful\nPostgreSQL features from a Python script.\n\nPyGreSQL 2.2 was developed and tested on a NetBSD 1.3_BETA system. It\nis based on the PyGres95 code written by Pascal Andre,\[email protected]. I changed the version to 2.0 and updated the\ncode for Python 1.5 and PostgreSQL 6.2.1. While I was at it I upgraded\nthe code to use full ANSI style prototypes and changed the order of\narguments to connect. Later versions are fixes and enhancements to that.\n\nImportant changes from PyGreSQL 2.1 to PyGreSQL 2.2:\n - Added user and password support thanks to Ng Pheng Siong <[email protected]>\n - Insert queries return the inserted oid\n - Add new pg wrapper (C module renamed to _pg)\n - Wrapped database connection in a class.\n - Cleaned up some of the tutorial. (More work needed.)\n - Added version and __version__. Thanks to [email protected] for\n the suggestion.\n\nImportant changes from PyGreSQL 2.0 to PyGreSQL 2.1:\n - return fields as proper Python objects for field type\n - Cleaned up pgext.py\n - Added dictresult method\n\nImportant changes from Pygres95 1.0b to PyGreSQL 2.0:\n - Updated code for PostgreSQL 6.2.1 and Python 1.5.\n - Reformatted code and converted to ANSI .\n - Changed name to PyGreSQL (from PyGres95.)\n - Changed order of arguments to connect function.\n - Created new type pgqueryobject and moved certain methods to it.\n - Added a print function for pgqueryobject\n - Various code changes - mostly stylistic.\n\nFor more information about each package, please have a look to their\nweb pages:\n - Python : http://www.python.org/\n - PostgreSQL : http://www.PostgreSQL.org/\n - PyGreSQL : http://www.druid.net/pygresql/\n\n\nD'Arcy J.M. Cain\[email protected]\n\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 424 2871 (DoD#0082) (eNTP) | what's for dinner.\n",
"msg_date": "Fri, 11 Dec 1998 07:34:58 -0500 (EST)",
"msg_from": "\"D'Arcy\" \"J.M.\" Cain <[email protected]>",
"msg_from_op": true,
"msg_subject": "Announce: PyGreSQL 2.2"
},
{
"msg_contents": "Do we need to install this in the main source trees?\n\n\n> Announce: Release of PyGreSQL version 2.2\n> ===============================================\n> \n> PyGreSQL v2.2 has been released.\n> It is available at: ftp://ftp.druid.net/pub/distrib/PyGreSQL-2.2.tgz.\n> \n> PostgreSQL is a database system derived from Postgres4.2. It conforms\n> to (most of) ANSI SQL and offers many interesting capabilities (C\n> dynamic linking for functions or type definition, etc.). This package\n> is copyright by the Regents of the University of California, and is\n> freely distributable.\n> \n> Python is an interpreted programming language. It is object oriented,\n> simple to use (light syntax, simple and straightforward statements), and\n> has many extensions for building GUIs, interfacing with WWW, etc. An\n> intelligent web browser (HotJava like) is currently under development\n> (November 1995), and this should open programmers many doors. Python is\n> copyrighted by Stichting S Mathematisch Centrum, Amsterdam, The\n> Netherlands, and is freely distributable.\n> \n> PyGreSQL is a python module that interfaces to a PostgreSQL database. It\n> embeds the PostgreSQL query library to allow easy use of the powerful\n> PostgreSQL features from a Python script.\n> \n> PyGreSQL 2.2 was developed and tested on a NetBSD 1.3_BETA system. It\n> is based on the PyGres95 code written by Pascal Andre,\n> [email protected]. I changed the version to 2.0 and updated the\n> code for Python 1.5 and PostgreSQL 6.2.1. While I was at it I upgraded\n> the code to use full ANSI style prototypes and changed the order of\n> arguments to connect. Later versions are fixes and enhancements to that.\n> \n> Important changes from PyGreSQL 2.1 to PyGreSQL 2.2:\n> - Added user and password support thanks to Ng Pheng Siong <[email protected]>\n> - Insert queries return the inserted oid\n> - Add new pg wrapper (C module renamed to _pg)\n> - Wrapped database connection in a class.\n> - Cleaned up some of the tutorial. (More work needed.)\n> - Added version and __version__. Thanks to [email protected] for\n> the suggestion.\n> \n> Important changes from PyGreSQL 2.0 to PyGreSQL 2.1:\n> - return fields as proper Python objects for field type\n> - Cleaned up pgext.py\n> - Added dictresult method\n> \n> Important changes from Pygres95 1.0b to PyGreSQL 2.0:\n> - Updated code for PostgreSQL 6.2.1 and Python 1.5.\n> - Reformatted code and converted to ANSI .\n> - Changed name to PyGreSQL (from PyGres95.)\n> - Changed order of arguments to connect function.\n> - Created new type pgqueryobject and moved certain methods to it.\n> - Added a print function for pgqueryobject\n> - Various code changes - mostly stylistic.\n> \n> For more information about each package, please have a look to their\n> web pages:\n> - Python : http://www.python.org/\n> - PostgreSQL : http://www.PostgreSQL.org/\n> - PyGreSQL : http://www.druid.net/pygresql/\n> \n> \n> D'Arcy J.M. Cain\n> [email protected]\n> \n> \n> -- \n> D'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\n> http://www.druid.net/darcy/ | and a sheep voting on\n> +1 416 424 2871 (DoD#0082) (eNTP) | what's for dinner.\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, 12 Dec 1998 23:44:59 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Announce: PyGreSQL 2.2"
},
{
"msg_contents": "Thus spake Bruce Momjian\n> Do we need to install this in the main source trees?\n\nI would say so. The older version is in there.\n\n> > Announce: Release of PyGreSQL version 2.2\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 424 2871 (DoD#0082) (eNTP) | what's for dinner.\n",
"msg_date": "Sun, 13 Dec 1998 07:56:14 -0500 (EST)",
"msg_from": "\"D'Arcy\" \"J.M.\" Cain <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [GENERAL] Announce: PyGreSQL 2.2"
},
{
"msg_contents": "Applied.\n\n> Thus spake Bruce Momjian\n> > Do we need to install this in the main source trees?\n> \n> I would say so. The older version is in there.\n> \n> > > Announce: Release of PyGreSQL version 2.2\n> \n> -- \n> D'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\n> http://www.druid.net/darcy/ | and a sheep voting on\n> +1 416 424 2871 (DoD#0082) (eNTP) | what's for dinner.\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": "Fri, 18 Dec 1998 13:04:23 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Announce: PyGreSQL 2.2"
}
] |
[
{
"msg_contents": "Hiroshi,\n\tI've got a bug report for you. On a Win95 machine running\nTcl/Tk 8.1, the dll does not load correctly. A \"load libpgtcl.dll\"\nat the \"wish81\" console gives a :\n<Couldn't load file \"libpgtcl.dll\": invalid argument>. \nBut it works file on Tcl/Tk 8.0 for Windows. Apparently the DLL \ninterface in Tcl/Tk 8.1 is a little different, I'm not sure what \nit is. Are you using Microsoft Visual C++ or a Borland Compiler? \nThe DLLs that Borland puts out are a little different but for the \nmost part I have'nt to many problems with them. The Tcl that \nScriptics puts out is compiled in Microsoft Visual C++.\n\nI'll look at your code this weekend.\n\nD.\n\n-----Original Message-----\nFrom: Hiroshi Inoue [mailto:[email protected]]\nSent: Thursday, December 10, 1998 9:23 PM\nTo: pgsql-hackers; [email protected]\nSubject: [HACKERS] libpq and libpgtcl in Windows\n\n\nHi all\n\nI made dll's(libpq.dll,libpgtcl.dll) only for pgaccess.\nIt is a temporary solution for quick fix and the solution for general\nuse \nis necessary.\n\nI made libpgtcl.dll ignoring NOTIFY functionality(not used in pgaccess).\nSorry,I'm not a Tcl expert and can't provide proper code for NOTIFY \nfunctionality in Windows. \n\nAnother problem exists between libpq and libpgtcl in Windows.\nLibpgtcl references global data pgresStatus[] in libpq when using \npg_result -status command. \nBut in Windows libpgtcl doesn't reference pgresStatus[] in the same \nway as UNIX and pg_result -status returns improper values.\n\nAny ideas ?\n\nIn my temporary solution for pgaccess,I used a function interface \nbetween libpq and libpgtcl instead of global data. \n\nThanks\n\nHiroshi Inoue\[email protected]\n\n\n\n",
"msg_date": "Fri, 11 Dec 1998 09:16:43 -0500",
"msg_from": "Dan Gowin <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] libpq and libpgtcl in Windows"
},
{
"msg_contents": "Hi all\n\n> -----Original Message-----\n> From: Dan Gowin [mailto:[email protected]]\n> Sent: Friday, December 11, 1998 11:17 PM\n> To: 'Hiroshi Inoue'; PGSQL HACKERS (E-mail);\n> '[email protected]'\n> Subject: RE: [HACKERS] libpq and libpgtcl in Windows\n> \n> \n> Hiroshi,\n> \tI've got a bug report for you. On a Win95 machine running\n> Tcl/Tk 8.1, the dll does not load correctly. A \"load libpgtcl.dll\"\n> at the \"wish81\" console gives a :\n> <Couldn't load file \"libpgtcl.dll\": invalid argument>. \n> But it works file on Tcl/Tk 8.0 for Windows. Apparently the DLL \n> interface in Tcl/Tk 8.1 is a little different, I'm not sure what \n> it is.\n\nIf the version of Tcl is different,the name is different(tcl76.dll,\ntcl80.dll,tcl81.dll etc).\nThe dll I provided is only for Tcl8.0(tcl80.dll).\nFor Tcl8.1 another dll is necessary.\nThe dll for Tcl8.1 is necessary ?\n\n Are you using Microsoft Visual C++ or a Borland Compiler? \n> The DLLs that Borland puts out are a little different but for the \n> most part I have'nt to many problems with them. The Tcl that \n> Scriptics puts out is compiled in Microsoft Visual C++.\n> \n> I'll look at your code this weekend.\n> \n> D.\n> \n\nThanks\n\nHiroshi Inoue\[email protected] \n",
"msg_date": "Mon, 14 Dec 1998 09:49:34 +0900",
"msg_from": "\"Hiroshi Inoue\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] libpq and libpgtcl in Windows"
},
{
"msg_contents": "Hiroshi Inoue wrote:\n> \n> If the version of Tcl is different,the name is different(tcl76.dll,\n> tcl80.dll,tcl81.dll etc).\n> The dll I provided is only for Tcl8.0(tcl80.dll).\n> For Tcl8.1 another dll is necessary.\n> The dll for Tcl8.1 is necessary ?\n\nTcl/Tk 8.1 is currently in beta stage, but a lot of programmers choose\nto \"taste\" the new flavor of Tcl/Tk 8.1\nIf it's not so hard, maybe it would be better to deliver DLL for all\ncurrently available versions.\n\n\nBest regards,\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Mon, 14 Dec 1998 09:34:48 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [INTERFACES] RE: [HACKERS] libpq and libpgtcl in Windows"
}
] |
[
{
"msg_contents": "Hi all\n\n6.4 release version on RedHat 4.2 w/ many upgrades including flex.\n\nCheck this out, the 'password' field was an after thought, and was added\nusing the alter table command:\nAlso, I found a spelling error in a field name, so ran these two commands\nin this order:\n\nalter table listings rename lanme to lname;\n\nalter table listings add column password char(10);\n\nAll seemed fine untill the project got to the point where password came\ninto use:\n\nlse=> \\d listings\n\nTable = listings\n+------------------------+----------------------------------+-------+\n| Field | Type | Length|\n+------------------------+----------------------------------+-------+\n| title | char() not null | 30 |\n| discription | char() not null | 150 |\n| url | char() not null | 100 |\n| lname | char() not null | 20 |\n| fname | char() not null | 20 |\n| email | char() not null | 100 |\n| ent_date | datetime not null default dateti | 8 |\n| mod_date | datetime not null default dateti | 8 |\n| approved | bool not null default 'False' | 1 |\n| item_id | int4 not null default nextval ( | 4 |\n| password | char() | 10 |\n+------------------------+----------------------------------+-------+\nIndices: listings_pkey\n listings_url_key\nlse=> select * from listings where password = '093b7a6b';\nERROR: RestrictionClauseSelectivity: bad value 2.164003\nlse=> select password from listings where item_id = 1;\npassword\n----------\n093b7a6b\n(1 row)\n\nlse=>\n\nAny ideas what that is all about?\n\nThanks, have a great day\nTerry Mackintosh <[email protected]> http://www.terrym.com\nsysadmin/owner Please! No MIME encoded or HTML mail, unless needed.\n\nProudly powered by R H Linux 4.2, Apache 1.3, PHP 3, PostgreSQL 6.4\n-------------------------------------------------------------------\nSuccess Is A Choice ... book by Rick Patino, get it, read it!\n\n",
"msg_date": "Fri, 11 Dec 1998 09:25:34 -0500 (EST)",
"msg_from": "Terry Mackintosh <[email protected]>",
"msg_from_op": true,
"msg_subject": "alter table bug???"
},
{
"msg_contents": "> Indices: listings_pkey\n> listings_url_key\n> lse=> select * from listings where password = '093b7a6b';\n> ERROR: RestrictionClauseSelectivity: bad value 2.164003\n> lse=> select password from listings where item_id = 1;\n> password\n> ----------\n> 093b7a6b\n> (1 row)\n> \n> lse=>\n> \n> Any ideas what that is all about?\n\nVACUUM ANALYZE the table and see if that fixes it.\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, 11 Dec 1998 11:22:46 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] alter table bug???"
}
] |
[
{
"msg_contents": "Hi all\n\nAs 6.4.1 is near release it seemed a good time to ask if the modification\ndatetime stamp function has been or will be contributed to the\n/contrib/spi/ stuff? \n\nIt seemed to me a handy thing to have if any one wants to keep track of\nwhen a record was last updated. In fact I've already used it in two\nprojects.\n\nI did not write it from scratch, but modified what was already there.\nThat is why it seems to me logical to add it to the existing spi examples.\n\nThere was no response to the last 2 submitions.\nShould such things be sent else where?\n\nThanks\nTerry Mackintosh <[email protected]> http://www.terrym.com\nsysadmin/owner Please! No MIME encoded or HTML mail, unless needed.\n\nProudly powered by R H Linux 4.2, Apache 1.3, PHP 3, PostgreSQL 6.4\n-------------------------------------------------------------------\nSuccess Is A Choice ... book by Rick Patino, get it, read it!\n\n",
"msg_date": "Fri, 11 Dec 1998 09:42:54 -0500 (EST)",
"msg_from": "Terry Mackintosh <[email protected]>",
"msg_from_op": true,
"msg_subject": "Inclusion of spi mod. date. function?"
}
] |
[
{
"msg_contents": "I have posted following patches on Nov 12 and do not see them in the\nREL6_4 tree. Please let me know the patches are acceptable or not.\nIf ok, please apply them to 6.5 tree also.\n\n> Included patches should fix following problems in the muti-byte\n> enabled PostgreSQL 6.4.\n> \n> o binary cursor does not work\n> o pg_dumpall produces incorrect create database statemnt\n> \n> Tatsuo Ishii\n> [email protected]\n> -------------------------------------------------------------------------\n> *** postgresql-v6.4/src/backend/access/common/printtup.c~\tSat Nov 7 11:47:47 1998\n> --- postgresql-v6.4/src/backend/access/common/printtup.c\tSat Nov 7 11:48:37 1998\n> ***************\n> *** 289,300 ****\n> \t\t\t\t/* variable length, assume a varlena structure */\n> \t\t\t\tlen = VARSIZE(attr) - VARHDRSZ;\n> \n> - #ifdef MULTIBYTE\n> - \t\t\t\tpq_putncharlen(VARDATA(attr), len);\n> - #else\n> \t\t\t\tpq_putint(len, VARHDRSZ);\n> \t\t\t\tpq_putnchar(VARDATA(attr), len);\n> ! #endif\n> #ifdef IPORTAL_DEBUG\n> \t\t\t\t{\n> \t\t\t\t\tchar\t *d = VARDATA(attr);\n> --- 289,297 ----\n> \t\t\t\t/* variable length, assume a varlena structure */\n> \t\t\t\tlen = VARSIZE(attr) - VARHDRSZ;\n> \n> \t\t\t\tpq_putint(len, VARHDRSZ);\n> \t\t\t\tpq_putnchar(VARDATA(attr), len);\n> ! \n> #ifdef IPORTAL_DEBUG\n> \t\t\t\t{\n> \t\t\t\t\tchar\t *d = VARDATA(attr);\n> *** postgresql-v6.4/src/bin/pg_dump/pg_dumpall~\tMon Aug 24 10:14:06 1998\n> --- postgresql-v6.4/src/bin/pg_dump/pg_dumpall\tSat Nov 7 11:51:12 1998\n> ***************\n> *** 59,65 ****\n> \n> \tif createdb -help|grep encoding >/dev/null\n> \tthen\n> ! \t\techo \"create database with encoding='`pg_encoding $ENCODING`' $DATABASE;\"\n> \telse\n> \t\techo \"create database $DATABASE;\"\n> \tfi\n> --- 59,65 ----\n> \n> \tif createdb -help|grep encoding >/dev/null\n> \tthen\n> ! \t\techo \"create database $DATABASE with encoding='`pg_encoding $ENCODING`';\"\n> \telse\n> \t\techo \"create database $DATABASE;\"\n> \tfi\n> \n\n",
"msg_date": "Fri, 11 Dec 1998 23:53:58 +0900",
"msg_from": "Tatsuo Ishii <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [PATCHES] [REL6_4] pg_dumpall & binary cursor fix in MB support "
}
] |
[
{
"msg_contents": "All,\n\tBug report:\n\tServer:\tRed Hat 5.1 (Pentium II, 300 Mhz)\n\t\tPostgreSQL v6.4 (latest)\n\t\tServer name: zeplin (199.211.199.2)\n\t\tUser: dbx1 (non dba privileges)\n\t\tPort:5432\n\t\tpg_hba.conf: host\t199.211.199.6\t255.255.255.0\ntrust\n\t\tDatabases: template1, system, test\n\t\tTest database table: test.t1\n\t\t\tcreate table t1 (name varchar(30), sal int);\n\t\t\tinsert into t1 (name, sal) values ('Dan',\n23456);\n\t\t\tinsert into t1 (name, sal) values ('Jake',\n52345);\n\t\t\tinsert into t1 (name, sal) values ('John', 345);\n\n\tClient:\tWindows NT 4.0 patch 3 (Pentium II, 300Mhz)\n\t\tMicrosoft Access 97\n\t\tClient name: nttest (199.211.199.6)\n\t\tPost ODBC\tv6.30.0246\tODBC Driver\n\t\tDescription: Raken 1\n\t\tDatabase: test\n\t\tServer: zeplin\n\t\tUser: dbx1\n\t\tAccess: Read Only\n\nProblem: In Microsoft Access when making a ODBC-Link to a table from\nwithin\nan \".mdb\" the server authenticates correctly. But when the link is\nactually\nbeing made to \"t1\" a:\n\t<MSACCESS.exe - Application Error\n\tException: access violation (0xc0000005) Address:0x100025b0>\n\n\nD. Gowin\n\t\t\n",
"msg_date": "Fri, 11 Dec 1998 10:21:57 -0500",
"msg_from": "Dan Gowin <[email protected]>",
"msg_from_op": true,
"msg_subject": "PostgreODBC error"
}
] |
[
{
"msg_contents": "I run three HP minicomputers and two Sun Ultra 3000 all with \nOracle 7.3 and one with Oracle 8.0 .\n\nSend it to me.\n\nD.\n\n\n-----Original Message-----\nFrom: Thomas G. Lockhart [mailto:[email protected]]\nSent: Friday, December 11, 1998 1:36 AM\nTo: Postgres Hackers List\nSubject: [HACKERS] JOIN syntax. Examples?\n\n\nWell, I've started looking through my books for info on joins. The cross\njoin was pretty easy:\n\npostgres=> select * from (a cross join b);\ni| j|i| k\n-+----+-+--\n1|10.1|1|-1\n2|20.2|1|-1\n4| |1|-1\n<snip>\n\nwhich I've put into my copy of the parser.\n\nDoes anyone have a commercial installation which has good support for\nSQL92 joins? I'd like to send some small test cases to verify that I\nunderstand what the behavior should be.\n\nAlso, if anyone has worked with join syntax, outer joins especially, it\nwould be great to get some test case contributions...\n\n - Tom\n",
"msg_date": "Fri, 11 Dec 1998 10:25:40 -0500",
"msg_from": "Dan Gowin <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] JOIN syntax. Examples?"
}
] |
[
{
"msg_contents": "I did\n\nlse=> vacuum analyze;\nVACUUM\n\nand now it works fine.\n\nNote that before that I did just a plain 'vacuum;' and it had no effect.\n\nHave a great day, and thanks\nTerry Mackintosh <[email protected]> http://www.terrym.com\nsysadmin/owner Please! No MIME encoded or HTML mail, unless needed.\n\nProudly powered by R H Linux 4.2, Apache 1.3, PHP 3, PostgreSQL 6.4\n-------------------------------------------------------------------\nSuccess Is A Choice ... book by Rick Patino, get it, read it!\n\n",
"msg_date": "Fri, 11 Dec 1998 10:36:30 -0500 (EST)",
"msg_from": "Terry Mackintosh <[email protected]>",
"msg_from_op": true,
"msg_subject": "follow up to: alter table bug???"
}
] |
[
{
"msg_contents": "Maybe you could include on the TODO list the following bug \n\n\n\nAlter table add column doesn't save defaults options :\n\ntest=> create table students (id int4, name text);\nCREATE\ntest=> insert into students values (1,'bill');\nINSERT 39338 1\ntest=> alter table students add column age int2 default 20;\nADD\ntest=> insert into students values (2,'tom');\nINSERT 39340 1\ntest=> select * from students;\nid|name|age\n--+----+---\n 1|bill|\n 2|tom |\n(2 rows)\n\nThe default for age does not work !\n\n\nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Fri, 11 Dec 1998 18:48:17 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "For the TODO list"
},
{
"msg_contents": "> Maybe you could include on the TODO list the following bug \n> \n\nAdded:\n\n\t* ALTER TABLE ADD COLUMN does not honor DEFAULT\n\n> \n> \n> Alter table add column doesn't save defaults options :\n> \n> test=> create table students (id int4, name text);\n> CREATE\n> test=> insert into students values (1,'bill');\n> INSERT 39338 1\n> test=> alter table students add column age int2 default 20;\n> ADD\n> test=> insert into students values (2,'tom');\n> INSERT 39340 1\n> test=> select * from students;\n> id|name|age\n> --+----+---\n> 1|bill|\n> 2|tom |\n> (2 rows)\n> \n> The default for age does not work !\n> \n> \n> Constantin Teodorescu\n> FLEX Consulting Braila, ROMANIA\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": "Fri, 11 Dec 1998 12:11:40 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] For the TODO list"
},
{
"msg_contents": "Already done. See TODO copy on web page.\n\n\n\n> Maybe you could include on the TODO list the following bug \n> \n> \n> \n> Alter table add column doesn't save defaults options :\n> \n> test=> create table students (id int4, name text);\n> CREATE\n> test=> insert into students values (1,'bill');\n> INSERT 39338 1\n> test=> alter table students add column age int2 default 20;\n> ADD\n> test=> insert into students values (2,'tom');\n> INSERT 39340 1\n> test=> select * from students;\n> id|name|age\n> --+----+---\n> 1|bill|\n> 2|tom |\n> (2 rows)\n> \n> The default for age does not work !\n> \n> \n> Constantin Teodorescu\n> FLEX Consulting Braila, ROMANIA\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": "Sun, 13 Dec 1998 00:15:36 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: For the TODO list"
}
] |
[
{
"msg_contents": "Hi all\n\nIn looking up the substr() function I just noticed this:\n(man pgbuiltin)\n\n text ltrim(text,text) left trim characters from\ntext\n text position(text,text) extract specified substring\n--->>>--------------------------cut and past error? ^^^^^^^^^^^^^^^^^^^^^\n text rpad(text,int,text) right pad string to \nspecified length\n text rtrim(text,text) right trim characters from\ntext\n text substr(text,int[,int]) extract specified substring\n--->>>--- same as this -------------------------^^^^^^^^^^^^^^^^^^^^^^\n text upper(text) convert text to upper case\n\nHave a great day\nTerry Mackintosh <[email protected]> http://www.terrym.com\nsysadmin/owner Please! No MIME encoded or HTML mail, unless needed.\n\nProudly powered by R H Linux 4.2, Apache 1.3, PHP 3, PostgreSQL 6.4\n-------------------------------------------------------------------\nSuccess Is A Choice ... book by Rick Patino, get it, read it!\n\n",
"msg_date": "Fri, 11 Dec 1998 13:18:51 -0500 (EST)",
"msg_from": "Terry Mackintosh <[email protected]>",
"msg_from_op": true,
"msg_subject": "Error in man page"
},
{
"msg_contents": "pgbuitin manual page is gone from 6.4.\n\n\n> Hi all\n> \n> In looking up the substr() function I just noticed this:\n> (man pgbuiltin)\n> \n> text ltrim(text,text) left trim characters from\n> text\n> text position(text,text) extract specified substring\n> --->>>--------------------------cut and past error? ^^^^^^^^^^^^^^^^^^^^^\n> text rpad(text,int,text) right pad string to \n> specified length\n> text rtrim(text,text) right trim characters from\n> text\n> text substr(text,int[,int]) extract specified substring\n> --->>>--- same as this -------------------------^^^^^^^^^^^^^^^^^^^^^^\n> text upper(text) convert text to upper case\n> \n> Have a great day\n> Terry Mackintosh <[email protected]> http://www.terrym.com\n> sysadmin/owner Please! No MIME encoded or HTML mail, unless needed.\n> \n> Proudly powered by R H Linux 4.2, Apache 1.3, PHP 3, PostgreSQL 6.4\n> -------------------------------------------------------------------\n> Success Is A Choice ... book by Rick Patino, get it, read it!\n> \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": "Sun, 13 Dec 1998 00:18:17 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Error in man page"
}
] |
[
{
"msg_contents": "\nI have added stdio.h to getrusage.c to fix this problem, in both trees.\n\n> Bruce,\n> Regarding your inquiry, I looked into stdio.h and (lo and behold) this is the\n> definition.\n> \n> #ifndef NULL\n> #define NULL\t0\n> #endif\n> \n> I'm not sure why I had to add it to the Postgres source file to avoid\n> compilation errors, since the definition in stdio.h is identical.\n> \n> Troy Hanson\n> \n> Bruce Momjian wrote:\n> > \n> > Please find the NULL define in your system header, and tell me what it\n> > is so I can include that in the file. Could be in stdio.h?\n> > \n> > > On SCO OpenServer 3.2.5, I get an error when compiling\n> > > pgsql/src/backend/port/getrusage.c . NULL is undefined:\n> > >\n> > > make -C port all\n> > > make[2]: Entering directory `/home/postgres/pgsql/src/backend/port'\n> > > cc -b elf -I../../include -I../../backend -dy -I.. -c getrusage.c\n> > > \"getrusage.c\", line 26: error: undefined symbol: NULL\n> > > make[2]: *** [getrusage.o] Error 1\n> > >\n> > > The simple fix is to add these three 3 lines to the file:\n> > >\n> > > #ifndef NULL\n> > > #define NULL 0\n> > > #endif\n> > >\n> > > Troy Hanson\n> > > [email protected]\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> \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, 12 Dec 1998 14:58:12 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [PATCHES] Patch for PostgreSQL 6.4-BETA5:\n\tpgsql/src/backend/port/getrusage.c"
}
] |
[
{
"msg_contents": "Bruce Momjian wrote:\n> \n> I believe these are all fixed in 6.4.1, due out on Friday.\n\nYes. They are fixed. Tried them on 9 Dec snapshot.\n \n> Constantin Teodorescu wrote: \n> > Found today two ugly bugs in pg_dump :\n> >\n> > Environment : Linux i386 RedHat 5.2, 2.0.36 kernel, Pentium 233 MMX\n> > Pg version : PostgreSQL 6.4\n\n-- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n",
"msg_date": "Sun, 13 Dec 1998 11:30:09 +0200",
"msg_from": "Constantin Teodorescu <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Two pg_dump ugly bugs :-("
}
] |
[
{
"msg_contents": "\nWhat is currently *not* working in it, for what operating systems?\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sun, 13 Dec 1998 21:33:54 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "vnsprintf()"
},
{
"msg_contents": "The Hermit Hacker <[email protected]> writes:\n> What is currently *not* working in it, for what operating systems?\n\nThe vsnprintf code we have might well work fine everywhere. I only\nobjected to starting to depend on it a few days before a major release,\nwhen we hadn't been using it before. I see no reason not to check in\nthe trace changes to the development tree and see whether anything\nbreaks for anyone... if problems show up, we have time to work on it\nfor 6.5...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 13 Dec 1998 20:54:43 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] vnsprintf() "
},
{
"msg_contents": "On Sun, 13 Dec 1998, Tom Lane wrote:\n\n> The Hermit Hacker <[email protected]> writes:\n> > What is currently *not* working in it, for what operating systems?\n> \n> The vsnprintf code we have might well work fine everywhere. I only\n> objected to starting to depend on it a few days before a major release,\n> when we hadn't been using it before. I see no reason not to check in\n> the trace changes to the development tree and see whether anything\n> breaks for anyone... if problems show up, we have time to work on it\n> for 6.5...\n\n\tGo for it...I'm diving into the sprintf()->snprintf() changes\ntonight, which was why I was askign about the vnsprintf()...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sun, 13 Dec 1998 21:57:44 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] vnsprintf() "
}
] |
[
{
"msg_contents": "Hi all\n\nAm I going insane, I've been going around with this for a few hours now,\nit could be lack of sleep:)\n\nclaimsnet=> select * from roster where state = 'TX' or state = 'SD';\nBackend message type 0x44 arrived while idle\nBackend message type 0x44 arrived while idle\nWe have lost the connection to the backend, so further processing is\nimpossible. Terminating.\n[postgres@laptop ~]$\n\nI can not seem to 'or' any char() field, int4 is OK. I've vacuum'ed with\nand with out analyze, I've pg_dumped'ed, destroydb'ed and recreated the\ndatabase and reloaded, same thing.\n\nTo make matter even crazyer, it only happens with this one database,\nothers work fine. AND it is doing this on 2 differant boxes, my laptop\nand my server, both Red Hat 4.2 with PostgreSQL 6.4.\n\nI just reloaded new data, so it must be the data, but what about the data\nwhould do that? Its only 1281 rows, so it's not like it's a lot of data.\n\nI'm going to start splitting up the data and see if I can isolate it.\n\nThanks\nTerry Mackintosh <[email protected]> http://www.terrym.com\nsysadmin/owner Please! No MIME encoded or HTML mail, unless needed.\n\nProudly powered by R H Linux 4.2, Apache 1.3, PHP 3, PostgreSQL 6.4\n-------------------------------------------------------------------\nSuccess Is A Choice ... book by Rick Patino, get it, read it!\n\n",
"msg_date": "Sun, 13 Dec 1998 21:57:04 -0500 (EST)",
"msg_from": "Terry Mackintosh <[email protected]>",
"msg_from_op": true,
"msg_subject": "Can't 'or' things?"
},
{
"msg_contents": "> Hi all\n> \n> Am I going insane, I've been going around with this for a few hours now,\n> it could be lack of sleep:)\n> \n> claimsnet=> select * from roster where state = 'TX' or state = 'SD';\n> Backend message type 0x44 arrived while idle\n> Backend message type 0x44 arrived while idle\n> We have lost the connection to the backend, so further processing is\n> impossible. Terminating.\n> [postgres@laptop ~]$\n> \n> I can not seem to 'or' any char() field, int4 is OK. I've vacuum'ed with\n> and with out analyze, I've pg_dumped'ed, destroydb'ed and recreated the\n> database and reloaded, same thing.\n> \n> To make matter even crazyer, it only happens with this one database,\n> others work fine. AND it is doing this on 2 differant boxes, my laptop\n> and my server, both Red Hat 4.2 with PostgreSQL 6.4.\n> \n> I just reloaded new data, so it must be the data, but what about the data\n> whould do that? Its only 1281 rows, so it's not like it's a lot of data.\n> \n> I'm going to start splitting up the data and see if I can isolate it.\n\nBug in 6.4. Fixed in both current trees, and 6.4.1, to be released on\nFriday.\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": "Sun, 13 Dec 1998 22:58:32 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Can't 'or' things?"
}
] |
[
{
"msg_contents": "\nCan someone check the following '#if 0'd code and remove, if its useless?\n\nFile: catalog/heap.c\nLine: 1474\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sun, 13 Dec 1998 23:31:06 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "catalog/heap.c..."
},
{
"msg_contents": "> \n> Can someone check the following '#if 0'd code and remove, if its useless?\n> \n> File: catalog/heap.c\n> Line: 1474\n\nI don't see any if 0 in catalog/heap.c.\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": "Sun, 13 Dec 1998 23:28:39 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] catalog/heap.c..."
},
{
"msg_contents": "> > Can someone check the following '#if 0'd code and remove, if its \n> > useless?\n> > File: catalog/heap.c\n> > Line: 1474\n> I don't see any if 0 in catalog/heap.c.\n\nI just made changes there, and want to leave the old code until people\nhave had a chance to test. Will clean up later...\n\n - Tom\n",
"msg_date": "Mon, 14 Dec 1998 05:25:46 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] catalog/heap.c..."
},
{
"msg_contents": "On Mon, 14 Dec 1998, Thomas G. Lockhart wrote:\n\n> > > Can someone check the following '#if 0'd code and remove, if its \n> > > useless?\n> > > File: catalog/heap.c\n> > > Line: 1474\n> > I don't see any if 0 in catalog/heap.c.\n> \n> I just made changes there, and want to leave the old code until people\n> have had a chance to test. Will clean up later...\n\n\tOops...;) Its gone now...we always have 'cvs diff' to fall back\nto, so it isn't like we've lost anything...I love CVS :)\n\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 14 Dec 1998 02:22:11 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] catalog/heap.c..."
},
{
"msg_contents": "> Oops...;) Its gone now...we always have 'cvs diff' to fall back\n> to, so it isn't like we've lost anything...I love CVS :)\n\nYeah, well, maybe the new code actually works? ;)\n\n - Tom\n",
"msg_date": "Mon, 14 Dec 1998 06:25:15 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] catalog/heap.c..."
},
{
"msg_contents": "On Mon, 14 Dec 1998, Thomas G. Lockhart wrote:\n\n> > Oops...;) Its gone now...we always have 'cvs diff' to fall back\n> > to, so it isn't like we've lost anything...I love CVS :)\n> \n> Yeah, well, maybe the new code actually works? ;)\n\n\tI have confidence in you *grin*\n\n\tJust wait, if anyone is going to break something, the changes I\njust made should pretty much blow up the whole system...just about to do a\ntest build with it now :)\n\n\tI've changed appendStringInfo so that it is 'sprintf' like, in\nthat it takes a format/args...right now, there are areas of the code where\nwe sprintf() to a buffer, and then pass that down into appendStringInfo...\nthis eliminates all those sprintf()s...\n\n\tI knew I should never have gotten started :( Have ot be at work\nin 5 hrs *sigh* *grin*\n\n Marc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 14 Dec 1998 03:17:38 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] catalog/heap.c..."
}
] |
[
{
"msg_contents": "\nin commands/cluster.c, in function cluster, we define NewIndexName as:\n\nchar NewIndexName[NAMEDATALEN]; /* line 93 */\n\nin function copy_index, we define it as:\n\nchar *NewIndexName; /* line 246 */\n\nAnd then palloc(NAMEDATALEN) before it gets used...\n\nNow, which we use doesn't much matter to me, but I would think some sort\nof consistency would be in order...or am I missing something as far as\neach are concerned? Is one method inheriently faster then another, or do\nthey have about the same performance characteristics?\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sun, 13 Dec 1998 23:42:02 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "palloc() vs static define?"
},
{
"msg_contents": "> \n> in commands/cluster.c, in function cluster, we define NewIndexName as:\n> \n> char NewIndexName[NAMEDATALEN]; /* line 93 */\n> \n> in function copy_index, we define it as:\n> \n> char *NewIndexName; /* line 246 */\n> \n> And then palloc(NAMEDATALEN) before it gets used...\n> \n> Now, which we use doesn't much matter to me, but I would think some sort\n> of consistency would be in order...or am I missing something as far as\n> each are concerned? Is one method inheriently faster then another, or do\n> they have about the same performance characteristics?\n\nUh, O, Marc is coding. :-)\n\ncluster.c was written by a hack, and has been cleaned up over time. You\ncan use either method. \n\n> char NewIndexName[NAMEDATALEN]; /* line 93 */\n\nAllocated stack space on function entry, and releases it on function\nexit.\n\n> And then palloc(NAMEDATALEN) before it gets used...\n\npalloc() allocates at function call time, and you have to pfree it, or\nwait for transaction to pfree it. If you are starting/stopping\ntransactions between palloc() and pfree() you could loose the memory\nunless you change to the 'cache' memory context before doing the palloc\nand pfree. See other command/*.c files for examples of this.\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": "Sun, 13 Dec 1998 23:01:27 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] palloc() vs static define?"
},
{
"msg_contents": "The Hermit Hacker <[email protected]> writes:\n> in commands/cluster.c, in function cluster, we define NewIndexName as:\n> char NewIndexName[NAMEDATALEN]; /* line 93 */\n> in function copy_index, we define it as:\n> char *NewIndexName; /* line 246 */\n> And then palloc(NAMEDATALEN) before it gets used...\n\n> Now, which we use doesn't much matter to me, but I would think some sort\n> of consistency would be in order...or am I missing something as far as\n> each are concerned? Is one method inheriently faster then another, or do\n> they have about the same performance characteristics?\n\nThe first method (char name[SIZE]) is certainly far faster, since it is\njust allocating local variable space during function entry. In fact,\nit's probably effectively *free*, zero cycles. If you allocate any\nlocal variable space in a function, then allocating more just means\nsubtracting a larger constant from the stack pointer. palloc() is going\nto go through the memory management code which is certainly not cheap\nby comparison.\n\nOn the other hand, palloc easily supports asking for a variable amount\nof space, whereas local array variables have to have a compile-time-\nconstant size. (We ignore GNU-only language extensions here ;-).)\nFor this particular situation that's not a concern, but other places\nyou might have to use palloc.\n\nThe really critical issue is what lifetime the allocated space needs\nto have. A local array var will go away automatically at function\nexit; palloc'd space lives until you pfree it or it gets cleaned up\nduring transaction exit.\n\nIn this particular situation, the local array var approach looks better\nto me, but I wonder whether index_create is expecting to be handed a\npointer to persistent storage...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 14 Dec 1998 11:00:33 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] palloc() vs static define? "
}
] |
[
{
"msg_contents": "\nMore unused code that can be removed?\n\nLine: 88\n\n\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sun, 13 Dec 1998 23:49:31 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Please check: commands/dbcommands.c"
},
{
"msg_contents": "> \n> More unused code that can be removed?\n> \n> Line: 88\n\nYes.\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": "Sun, 13 Dec 1998 23:02:17 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Please check: commands/dbcommands.c"
}
] |
[
{
"msg_contents": "\nEverything is in CVS, if I oops, its recoverable...am removing all code\nthat has either:\n\n#if 0\n\nor \n\n#ifdef FALSE\n\nsurrounding it, since it isn't used...we have the logs and diffs in the\nCVS repository if I make a mistake, but since it isn't used, can't see how\nI can make a mistake :)\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sun, 13 Dec 1998 23:54:37 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Please check...ignore them..."
},
{
"msg_contents": "> \n> Everything is in CVS, if I oops, its recoverable...am removing all code\n> that has either:\n> \n> #if 0\n> \n> or \n> \n> #ifdef FALSE\n> \n> surrounding it, since it isn't used...we have the logs and diffs in the\n> CVS repository if I make a mistake, but since it isn't used, can't see how\n> I can make a mistake :)\n\nYep, except there is some code if'defed out that could be useful some\nday. The one from dbcommand.c clearly would never be useful. \nBasically, if you can figure out what the ifdef 0 did, and you see a\nreplacement in the code, you can remove it. That is the guide I use.\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": "Sun, 13 Dec 1998 23:03:39 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Please check...ignore them..."
}
] |
[
{
"msg_contents": "\nIts empty...*raised eyebrow*\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 14 Dec 1998 00:03:36 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Reason for commands/user.c?"
},
{
"msg_contents": "> \n> Its empty...*raised eyebrow*\n\nNot here.\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": "Sun, 13 Dec 1998 23:27:09 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Reason for commands/user.c?"
},
{
"msg_contents": "On Sun, 13 Dec 1998, Bruce Momjian wrote:\n\n> > \n> > Its empty...*raised eyebrow*\n> \n> Not here.\n\nNow, that's odd...its not here anymore either *shake head* guess that's\nthe last beer for tonight, eh? :) back to the code...:)\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 14 Dec 1998 00:37:54 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Reason for commands/user.c?"
}
] |
[
{
"msg_contents": "\nOkay, I'm brave, but not *that* brave...commands/version.c, the whole file\nconsists of functions, each individually '#ifdef NOT_USED'...tempted to\njust remove it, since it isn't used, but figured I'd say something on the\nlist before I got that brave...\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 14 Dec 1998 00:06:11 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Alot of code...NOT_USED..."
},
{
"msg_contents": "> \n> Okay, I'm brave, but not *that* brave...commands/version.c, the whole file\n> consists of functions, each individually '#ifdef NOT_USED'...tempted to\n> just remove it, since it isn't used, but figured I'd say something on the\n> list before I got that brave...\n\nGood news. version.c is no longer compiled into the system. Just there\nfor reference if we ever try and do something like that. Perhaps it\nshould be moved into an 'unused/' directory, with some other *.c files\nthat are not used anymore.\n\nI am on IRC for a little while.\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": "Sun, 13 Dec 1998 23:26:41 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Alot of code...NOT_USED..."
}
] |
[
{
"msg_contents": "\nexecutor/nodeTee.c, line 174? I can't find anywhere where space was\nallocated...?\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 14 Dec 1998 00:15:36 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "sprintf() of char* value, no memory allocation?"
},
{
"msg_contents": "> \n> executor/nodeTee.c, line 174? I can't find anywhere where space was\n> allocated...?\n\nYep. However, I think Tee may never be used as a node by any query, so\nit may not be a problem. Search the archives. You may find\nsomeone(Vadim?) saying that Tee is never used.\n\nStill, should be cleaned up.\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": "Sun, 13 Dec 1998 23:24:56 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] sprintf() of char* value, no memory allocation?"
}
] |
[
{
"msg_contents": "\nchar * \ncrypt_getpwdfilename()\n{ \n \n static char *pfnam = NULL;\n \n if (!pfnam)\n { \n int bufsize;\n bufsize = strlen(DataDir) + strlen(CRYPT_PWD_FILE) + 2;\n pfnam = (char *) palloc(bufsize); \n spprintf(pfnam, bufsize, \"%s/%s\", DataDir, CRYPT_PWD_FILE);\n }\n \n return pfnam;\n}\n\nWhy the check for '!ipfnam'? Seems useless since we are setting it to\nNULL the line before...no?\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 14 Dec 1998 02:21:07 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Does this make sense:"
},
{
"msg_contents": "> Why the check for '!ipfnam'? Seems useless since we are setting it to\n> NULL the line before...no?\n\nDoes a static char get initialized each time through, or just once at\nallocation? Is this setting a \"persistant value\" for pfnam? I would\nguess so, but ymmv...\n\n - Tom\n",
"msg_date": "Mon, 14 Dec 1998 07:29:51 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Does this make sense:"
},
{
"msg_contents": "> \n> char * \n> crypt_getpwdfilename()\n> { \n> \n> static char *pfnam = NULL;\n> \n> if (!pfnam)\n> { \n> int bufsize;\n> bufsize = strlen(DataDir) + strlen(CRYPT_PWD_FILE) + 2;\n> pfnam = (char *) palloc(bufsize); \n> spprintf(pfnam, bufsize, \"%s/%s\", DataDir, CRYPT_PWD_FILE);\n> }\n> \n> return pfnam;\n> }\n> \n> Why the check for '!ipfnam'? Seems useless since we are setting it to\n> NULL the line before...no?\n\nActually, no. We are declaring it as static, so the first time the\nfunction is called, it is set to NULL. After that, it is not\ninitialized for each function call because a static local variable's\nvalue is kept between function calls. It is like a global variable in\nits duration, but in local scope.\n\nThis is an old trick to run the initialization code only the first time\nthe function is called.\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": "Mon, 14 Dec 1998 09:55:56 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Does this make sense:"
},
{
"msg_contents": "Then <[email protected]> spoke up and said:\n> \n> char * \n> crypt_getpwdfilename()\n> { \n> \n> static char *pfnam = NULL;\n> \n> if (!pfnam)\n> { \n> int bufsize;\n> bufsize = strlen(DataDir) + strlen(CRYPT_PWD_FILE) + 2;\n> pfnam = (char *) palloc(bufsize); \n> spprintf(pfnam, bufsize, \"%s/%s\", DataDir, CRYPT_PWD_FILE);\n> }\n> \n> return pfnam;\n> }\n> \n> Why the check for '!ipfnam'? Seems useless since we are setting it to\n> NULL the line before...no?\n\nstatic variables should be thought of as globals with only local\nscope. This variable is allocated in the BSS area rather than on the\nstack at runtime. The initialization is performed only once (usually\nat compile time), and the value is retained from execution to\nexecution. Basically, it always returns the previously computed\nfilename except for the very first call.\n\n-- \n=====================================================================\n| JAVA must have been developed in the wilds of West Virginia. |\n| After all, why else would it support only single inheritance?? |\n=====================================================================\n| Finger [email protected] for my public key. |\n=====================================================================",
"msg_date": "14 Dec 1998 10:21:12 -0500",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Does this make sense:"
},
{
"msg_contents": ">\n> >\n> > char *\n> > crypt_getpwdfilename()\n> > {\n> >\n> > static char *pfnam = NULL;\n> >\n> > if (!pfnam)\n> > {\n> > int bufsize;\n> > bufsize = strlen(DataDir) + strlen(CRYPT_PWD_FILE) + 2;\n> > pfnam = (char *) palloc(bufsize);\n> > spprintf(pfnam, bufsize, \"%s/%s\", DataDir, CRYPT_PWD_FILE);\n> > }\n> >\n> > return pfnam;\n> > }\n> >\n> > Why the check for '!ipfnam'? Seems useless since we are setting it to\n> > NULL the line before...no?\n>\n> Actually, no. We are declaring it as static, so the first time the\n> function is called, it is set to NULL. After that, it is not\n> initialized for each function call because a static local variable's\n> value is kept between function calls. It is like a global variable in\n> its duration, but in local scope.\n>\n> This is an old trick to run the initialization code only the first time\n> the function is called.\n\n But is it good then to use palloc() instead of malloc()?\n Anything palloc()'d is thrown away when the memory context in\n which it is made get's destroyed. So you have to care about\n the memory context in which the call is made. If under some\n (but not all) circumstances the FIRST call is made in the\n wrong mcxt, the pointer maybe get's corrupted later.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 14 Dec 1998 17:20:50 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Does this make sense:"
},
{
"msg_contents": "> >\n> > >\n> > > char *\n> > > crypt_getpwdfilename()\n> > > {\n> > >\n> > > static char *pfnam = NULL;\n> > >\n> > > if (!pfnam)\n> > > {\n> > > int bufsize;\n> > > bufsize = strlen(DataDir) + strlen(CRYPT_PWD_FILE) + 2;\n> > > pfnam = (char *) palloc(bufsize);\n> > > spprintf(pfnam, bufsize, \"%s/%s\", DataDir, CRYPT_PWD_FILE);\n> > > }\n> > >\n> > > return pfnam;\n> > > }\n> > >\n> > > Why the check for '!ipfnam'? Seems useless since we are setting it to\n> > > NULL the line before...no?\n> >\n> > Actually, no. We are declaring it as static, so the first time the\n> > function is called, it is set to NULL. After that, it is not\n> > initialized for each function call because a static local variable's\n> > value is kept between function calls. It is like a global variable in\n> > its duration, but in local scope.\n> >\n> > This is an old trick to run the initialization code only the first time\n> > the function is called.\n> \n> But is it good then to use palloc() instead of malloc()?\n> Anything palloc()'d is thrown away when the memory context in\n> which it is made get's destroyed. So you have to care about\n> the memory context in which the call is made. If under some\n> (but not all) circumstances the FIRST call is made in the\n> wrong mcxt, the pointer maybe get's corrupted later.\n\nOops, yep, I didn't see that. palloc is bad to use in this context. \nMalloc should be used, Marc.\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": "Mon, 14 Dec 1998 11:43:38 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Does this make sense:"
}
] |
[
{
"msg_contents": "\nThomas...\n\n\tNot sure what you were attempting with this, but it looks like at\nline 74, under FreeBSD, you are assigning something that hasn't been\ndefined? tzn? \n\n\t\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Mon, 14 Dec 1998 04:07:30 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "nabstime.c changes..."
},
{
"msg_contents": "> Not sure what you were attempting with this, but it looks like at\n> line 74, under FreeBSD, you are assigning something that hasn't been\n> defined? tzn?\n\nOops. Cut and pasted from another place, and my local machine doesn't\ncompile that conditional code so I didn't catch it.\n\nTry this patch (take out the check for a null pointer, and substitute\nCTZName for tzn, all around line 74), and let me know if it works...\n\n - Tom",
"msg_date": "Mon, 14 Dec 1998 14:53:26 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] nabstime.c changes..."
}
] |
[
{
"msg_contents": "Hi Hackers,\n\nWith the following query from the opr_sanity regression test I'm\ngetting a backend crash.\n\nSELECT p1.oid, p1.oprcode, p2.oid, p2.oprcode\nFROM pg_operator AS p1, pg_operator AS p2\nWHERE p1.oprlsortop = p2.oid AND\n (p1.oprname != '=' OR\n p1.oprkind != 'b' OR p2.oprkind != 'b' OR\n p1.oprleft != p2.oprleft OR\n p1.oprright != p2.oprright OR\n p1.oprresult != 16 OR\n p2.oprresult != 16 OR\n p1.oprrsortop = 0);\n\nThis is on the latest CVS, although I believe it happened on earlier\nsnapshots.\n\nPlatform is SPARC/Linux Redhat 4.2 SPARCstation IPX.\n\nI've not been able to investigate further.\n\nKeith.\n\nHere's a backtrace of the process:-\n \n\nProgram received signal SIGSEGV, Segmentation fault.\n0x76a44 in ExecScanHashBucket (hjstate=0x209b10, bucket=0x214d4c, curtuple=0x0, \nhjclauses=0x20add0,\n econtext=0x20bc90) at nodeHash.c:734\n734 heapTuple->t_data = (HeapTupleHeader)\n(gdb) bt\n#0 0x76a44 in ExecScanHashBucket (hjstate=0x209b10, bucket=0x214d4c, \ncurtuple=0x0, hjclauses=0x20add0,\n econtext=0x20bc90) at nodeHash.c:734\n#1 0x770e8 in ExecHashJoin (node=0x209c10) at nodeHashjoin.c:283\n#2 0x7164c in ExecProcNode (node=0x209c10, parent=0x209c10) at \nexecProcnode.c:315\n#3 0x70ac4 in ExecutePlan (estate=0x209b90, plan=0x209c10, parseTree=0x1a5090, \noperation=CMD_SELECT,\n numberTuples=0, direction=ForwardScanDirection, printfunc=0x28d08 \n<printtup>) at execMain.c:733\n#4 0x705a8 in ExecutorRun (queryDesc=0x209c10, estate=0x209b90, feature=3, \ncount=0) at execMain.c:236\n#5 0xd1130 in ProcessQueryDesc (queryDesc=0x20a8f0) at pquery.c:333\n#6 0xd11a0 in ProcessQuery (parsetree=0x1a5090, plan=0x209c10, dest=Remote) at \npquery.c:376\n#7 0xcf8a0 in pg_exec_query_dest (\n query_string=0xefffd1a0 \"SELECT p1.oid, p1.oprcode, p2.oid, p2.oprcode\\nFROM \npg_operator AS p1, pg_operator AS p2\\nWHERE p1.oprlsortop = p2.oid AND\\n \n(p1.oprname != '=' OR\\n p1.oprkind != 'b' OR p2.oprkind != 'b' OR\\n \np\"..., dest=Remote, aclOverride=0 '\\000') at postgres.c:800\n#8 0xcf730 in pg_exec_query (\n query_string=0xefffd1a0 \"SELECT p1.oid, p1.oprcode, p2.oid, p2.oprcode\\nFROM \npg_operator AS p1, pg_operator AS p2\\nWHERE p1.oprlsortop = p2.oid AND\\n \n(p1.oprname != '=' OR\\n p1.oprkind != 'b' OR p2.oprkind != 'b' OR\\n \np\"...) at postgres.c:699\n#9 0xd0a6c in PostgresMain (argc=-8192, argv=0x15dc00, real_argc=10, \nreal_argv=0xeffffd84) at postgres.c:1645\n#10 0xb34e8 in DoBackend (port=0x135400) at postmaster.c:1532\n#11 0xb2f74 in BackendStartup (port=0x199c00) at postmaster.c:1303\n#12 0xb23dc in ServerLoop () at postmaster.c:757\n#13 0xb1f24 in PostmasterMain (argc=0, argv=0xeffffd84) at postmaster.c:563\n#14 0x83580 in main (argc=10, argv=0xeffffd84) at main.c:93 \n\n",
"msg_date": "Mon, 14 Dec 1998 09:48:14 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "Backend crash in opr_sanity regression test."
},
{
"msg_contents": "Keith Parks <[email protected]> writes:\n> With the following query from the opr_sanity regression test I'm\n> getting a backend crash.\n> [ details snipped ]\n> This is on the latest CVS, although I believe it happened on earlier\n> snapshots.\n> Platform is SPARC/Linux Redhat 4.2 SPARCstation IPX.\n\nFWIW, I saw a backend crash in opr_sanity in an otherwise reasonably\nsuccessful attempt to build/test Postgres on SPARC/SunOS 4.1.4. This\nwas with REL6_4 CVS sources around the start of December. I didn't have\ntime to investigate it closely, but it does seem there's a platform-\ndependent problem lurking in there somewhere. (Wonder if it could be\nspecific to the SPARC hardware?)\n\nCan any of the backend experts make any guesses from Keith's backtrace?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 14 Dec 1998 20:18:10 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Backend crash in opr_sanity regression test. "
},
{
"msg_contents": "Tom Lane wrote:\n> \n> Keith Parks <[email protected]> writes:\n> > With the following query from the opr_sanity regression test I'm\n> > getting a backend crash.\n> > [ details snipped ]\n> > This is on the latest CVS, although I believe it happened on earlier\n> > snapshots.\n> > Platform is SPARC/Linux Redhat 4.2 SPARCstation IPX.\n> \n> FWIW, I saw a backend crash in opr_sanity in an otherwise reasonably\n> successful attempt to build/test Postgres on SPARC/SunOS 4.1.4. This\n> was with REL6_4 CVS sources around the start of December. I didn't have\n ^^^^^^^^^^^^^^^\nIt's probably caused by new HeapTuple stuff, but I didn't\nput it in 6_4 branch...\n\n> time to investigate it closely, but it does seem there's a platform-\n> dependent problem lurking in there somewhere. (Wonder if it could be\n> specific to the SPARC hardware?)\n> \n> Can any of the backend experts make any guesses from Keith's backtrace?\n\nVadim\n",
"msg_date": "Tue, 15 Dec 1998 09:14:56 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Backend crash in opr_sanity regression test."
},
{
"msg_contents": "> Hi Hackers,\n> \n> With the following query from the opr_sanity regression test I'm\n> getting a backend crash.\n> \n> SELECT p1.oid, p1.oprcode, p2.oid, p2.oprcode\n> FROM pg_operator AS p1, pg_operator AS p2\n> WHERE p1.oprlsortop = p2.oid AND\n> (p1.oprname != '=' OR\n> p1.oprkind != 'b' OR p2.oprkind != 'b' OR\n> p1.oprleft != p2.oprleft OR\n> p1.oprright != p2.oprright OR\n> p1.oprresult != 16 OR\n> p2.oprresult != 16 OR\n> p1.oprrsortop = 0);\n\nI tried the query in the current development tree, and it returned zero\nrows and did not crash.\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, 16 Jan 1999 22:35:52 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Backend crash in opr_sanity regression test."
}
] |
[
{
"msg_contents": "> -----Original Message-----\n> Thomas G. Lockhart wrote:\n> > \n> > > The book \"The Practical SQL Handbook\", which is often \n> recommended on\n> > > these lists, uses the syntax `*=' and `=*' for left and \n> right outer\n> > > joins (page 211). I think we ought to support this \n> syntax as well,\n> > > since it will save new users from confusion.\n> > \n> > This one conflicts with Postgres' operator extensibility \n> features, since\n> > it would look just like a legal operator.\n> \n> so does =\n> \n> Could it be possible to extend the operator extensibility features \n> to achieve the behaviour of outer/cross joins ?\n> \n> > The two books I have at hand (besides my old Ingres docs) \n> are A Guide to\n> > the SQL Standard by Date and Darwen and Understanding the New SQL by\n> > Melton and Simon. Both focus on SQL standard syntax, and \n> neither mention\n> > the various outer join syntaxes accepted by Oracle, \n> Informix, or Sybase.\n> \n> Has anybody tried out DB2 ?\n> \n> I have downloaded it (for linux) but have not yet tried it.\n> \n> > An explanation for the lack of standards compliance by the big three\n> > probably involves the fact that they predate the standard by a\n> > significant number of years.\n> \n> Not to mention that both =* and =(+) are more concise and easier to \n> follow, at least for one with my headshape.\n> \n> The standard is probably the 'worst common denominator' or something \n> like that :(\n> \n\nMicrosoft SQL Server v6.5 also support the Sybase syntax =* and *=, but\nI found that syntax for more limiting than the SQL92 Standard syntax.\nI'm not sure how it would be implemented in Postgres but in both Sybase\nand M$ SQL you couldn't perform an Outer Join on an Outer Joined table\nusing the =* *= syntax. Also, we'd need an extension for cross join,\n*=*. \n\nMy vote is to implement the SQL92 Standard, it's far more flexible and\nwe don't have to shift with the tide when Oracle decides to change to %=\n=% to be it's join operators. \n\nAnd if it makes anyone feel better; the M$ SQL v6.5 docs state that *=\n=* would be depreciated in M$ SQL v7.0. Haven't seen 7.0 so don't know\nif they went through with it (or will).\n\t-DEJ\n",
"msg_date": "Mon, 14 Dec 1998 15:48:25 -0600",
"msg_from": "\"Jackson, DeJuan\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] JOIN syntax. Examples?"
}
] |
[
{
"msg_contents": "Hi all,\n\nI tried to divide 123.45 by 123.00 but PostgreSQL gives me a wrong\nresult:\n\nhygea=> select 123.45 - 123.00;\n ?column?\n-----------------\n0.450000000000003\n(1 row)\n\n-Jose'-\n\n\n",
"msg_date": "Tue, 15 Dec 1998 13:41:26 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": true,
"msg_subject": "123.45 - 123 = 0.45"
},
{
"msg_contents": "Hello!\n\nOn Tue, 15 Dec 1998, Sferacarta Software wrote:\n> I tried to divide 123.45 by 123.00 but PostgreSQL gives me a wrong\n> result:\n> \n> hygea=> select 123.45 - 123.00;\n> ?column?\n> -----------------\n> 0.450000000000003\n> (1 row)\n\n I cannot understand anything. You said \"to divide\" and then \"123.45 - 123.00\".\nAre you trying to divide or to substract?\n\nOleg.\n---- \n Oleg Broytmann National Research Surgery Centre http://sun.med.ru/~phd/\n Programmers don't die, they just GOSUB without RETURN.\n\n",
"msg_date": "Tue, 15 Dec 1998 16:08:48 +0300 (MSK)",
"msg_from": "Oleg Broytmann <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45"
},
{
"msg_contents": "> Hi all,\n> \n> I tried to divide 123.45 by 123.00 but PostgreSQL gives me a wrong\n> result:\n> \n> hygea=> select 123.45 - 123.00;\n> ?column?\n> -----------------\n> 0.450000000000003\n> (1 row)\n> \n\nWow, I get the same thing here.\n\nEven this doesn't work:\n\n\ttest=> select float8(123.45) - float8(123.00);\n\t ?column?\n\t-----------------\n\t0.450000000000003\n\t(1 row)\n\nNow constants are automatically promoted to float8, so I expected the\nsame results, but what is going on here?\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": "Tue, 15 Dec 1998 09:27:03 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45"
},
{
"msg_contents": "Try this:\n\n#include <stdio.h>\n\nmain()\n{\ndouble f1 = 123.45;\ndouble f2 = 123.00;\ndouble r;\n\n r = f1 - f2;\n printf(\"%0.15f %0.15f %0.15f\\n\", f1, f2, r);\n}\n\nInternal representation of 123.45 is not exact. In the conversion to\nbinary, an irrational number is created which is truncated to 64 bits.\n\n\nBruce Momjian wrote:\n\n> > Hi all,\n> >\n> > I tried to divide 123.45 by 123.00 but PostgreSQL gives me a wrong\n> > result:\n> >\n> > hygea=> select 123.45 - 123.00;\n> > ?column?\n> > -----------------\n> > 0.450000000000003\n> > (1 row)\n> >\n>\n> Wow, I get the same thing here.\n>\n> Even this doesn't work:\n>\n> test=> select float8(123.45) - float8(123.00);\n> ?column?\n> -----------------\n> 0.450000000000003\n> (1 row)\n>\n> Now constants are automatically promoted to float8, so I expected the\n> same results, but what is going on here?\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\n",
"msg_date": "Tue, 15 Dec 1998 10:18:21 -0500",
"msg_from": "David Hartwig <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45"
},
{
"msg_contents": "On Tue, 15 Dec 1998, Sferacarta Software wrote:\n\n> Hi all,\n> \n> I tried to divide 123.45 by 123.00 but PostgreSQL gives me a wrong\n> result:\n> \n> hygea=> select 123.45 - 123.00;\n> ?column?\n> -----------------\n> 0.450000000000003\n> (1 row)\n\nAre you trying to subtract or divide? Your select is subtraction,\nyour question was division.\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] flame-mail: /dev/null\n # include <std/disclaimers.h> TEAM-OS2\n Online Searchable Campground Listings http://www.camping-usa.com\n \"There is no outfit less entitled to lecture me about bloat\n than the federal government\" -- Tony Snow\n==========================================================================\n\n\n\n\n",
"msg_date": "Tue, 15 Dec 1998 10:34:42 -0500 (EST)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45"
},
{
"msg_contents": "Hello Oleg,\n\nmarted�, 15 dicembre 98, you wrote:\n\nOB> Hello!\n\nOB> On Tue, 15 Dec 1998, Sferacarta Software wrote:\n>> I tried to divide 123.45 by 123.00 but PostgreSQL gives me a wrong\n>> result:\n>> \n>> hygea=> select 123.45 - 123.00;\n>> ?column?\n>> -----------------\n>> 0.450000000000003\n>> (1 row)\n\nOB> I cannot understand anything. You said \"to divide\" and then \"123.45 - 123.00\".\nOB> Are you trying to divide or to substract?\n\nSorry I want to say subtract.\n\n-Jose'-\n\n\n",
"msg_date": "Tue, 15 Dec 1998 16:35:04 +0100",
"msg_from": "Sferacarta Software <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re[2]: [HACKERS] 123.45 - 123 = 0.45"
},
{
"msg_contents": "> Try this:\n> \n> #include <stdio.h>\n> \n> main()\n> {\n> double f1 = 123.45;\n> double f2 = 123.00;\n> double r;\n> \n> r = f1 - f2;\n> printf(\"%0.15f %0.15f %0.15f\\n\", f1, f2, r);\n> }\n> \n> Internal representation of 123.45 is not exact. In the conversion to\n> binary, an irrational number is created which is truncated to 64 bits.\n\nYes, someone pointed this out to me in private e-mail, and I wrote a C\nprogram to confirm it. I just had never seen such rounding on such\nsmall non-irrational numbers.\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": "Tue, 15 Dec 1998 11:20:15 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45"
},
{
"msg_contents": "Bruce Momjian <[email protected]> writes:\n> [ 123.45 - 123.00 = 0.450000000000003 ]\n\n> Now constants are automatically promoted to float8, so I expected the\n> same results, but what is going on here?\n\nPlain old, garden-variety, floating point roundoff error. Do the\nsame calculation in any other program and you'll get the same result\n(on the same hardware anyway), if the other program insists on showing\n16 digits of precision.\n\nIEEE 64-bit floats only have about 17 decimal digits of accuracy.\nSo float8(123.45) is good to about 14 digits after the decimal point.\nSubtract off the 123, and print what's left with 16 digits, and\nby golly you find out that the original number wasn't exactly 123.45,\njust an approximation to it. Along about the 15th digit after the\ndecimal point, you start finding crud.\n\nIn short: no surprises here for anyone who's used float math for any\nlength of time.\n\nSooner or later we ought to try to implement true fixed-point\narbitrary-precision numeric data types per the SQL spec. That'll be\na lot slower than hardware float math, but its roundoff properties will\nbe less surprising to novices.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 15 Dec 1998 11:43:08 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45 "
},
{
"msg_contents": "Tom Lane wrote:\n\n> Sooner or later we ought to try to implement true fixed-point\n> arbitrary-precision numeric data types per the SQL spec. That'll be\n> a lot slower than hardware float math, but its roundoff properties will\n> be less surprising to novices.\n\n I think the string math done in bc(1) could be a good point\n to start from. In the old version (1.3 if I remember right)\n of the minix sources (yepp - still have them), there are the\n algorithm's to get sine, logarithm and the like in any\n precision as bc functions. Internally, only the four base\n operations, (think think) pow() and sqrt() are implemented in\n C. Anything else is done with them on the higher level.\n\n We need to define what the precision of a result should be,\n if it is not assigned to a column (where the precision can be\n the atttypmod). Is there any standard defined for? If not,\n what about this:\n\n Internal representation holds different precision for DISPLAY\n and CALC.\n\n On any operation, the DISPLAY precision is set to the higher\n of the two operands.\n\n On add/subtract, the CALC precision becomes the higher of the\n two.\n\n On multiply, the CALC precision is adjusted to hold the exact\n result up to a (variable settable?) maximum.\n\n On divide, the CALC precision is set to max and after it to\n the number of used digits.\n\n If the result get's assigned to an attribute, it is rounded\n to it's atttypmod and both precisions set to that.\n\n The types output function rounds it to the DISPLAY precision.\n\n The input function sets both precisions to the number of\n digits present after decimal point.\n\n Needless to say that there will be special functions to round\n explicitly and set the precisions.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Tue, 15 Dec 1998 21:15:11 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45"
},
{
"msg_contents": "> We need to define what the precision of a result should be,\n> if it is not assigned to a column (where the precision can be\n> the atttypmod). Is there any standard defined for? If not,\n> what about this:\n> \n> Internal representation holds different precision for DISPLAY\n> and CALC.\n> \n> On any operation, the DISPLAY precision is set to the higher\n> of the two operands.\n> \n> On add/subtract, the CALC precision becomes the higher of the\n> two.\n> \n> On multiply, the CALC precision is adjusted to hold the exact\n> result up to a (variable settable?) maximum.\n> \n> On divide, the CALC precision is set to max and after it to\n> the number of used digits.\n> \n> If the result get's assigned to an attribute, it is rounded\n> to it's atttypmod and both precisions set to that.\n> \n> The types output function rounds it to the DISPLAY precision.\n> \n> The input function sets both precisions to the number of\n> digits present after decimal point.\n> \n> Needless to say that there will be special functions to round\n> explicitly and set the precisions.\n\nI have a routine that does the necessary rounding on 8 byte floating points to a\nprecision up to 8 decimal places. Not exactly based on higher math but it\ndoes the job in many financial applications on all assignments, comparisons \nand when displayed.\n\nI never use it on arithmetic operations.\n\nThe interface is as follows\n\ndouble RoundDouble(double value, int decimals);\n\nI am happy to submit it for use in the postgres code.\n\nRegards\nTheo\n",
"msg_date": "Wed, 16 Dec 1998 07:22:41 +0200 (SAST)",
"msg_from": "Theo Kramer <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45"
}
] |
[
{
"msg_contents": "> Yes, someone pointed this out to me in private e-mail, and I wrote a C\n> program to confirm it. I just had never seen such rounding on such\n> small non-irrational numbers.\nBut, the problem is that it is irrational in base 2 (or at least so many\ndigits that it won't fit).\n\t-DEJ\n",
"msg_date": "Tue, 15 Dec 1998 10:31:29 -0600",
"msg_from": "\"Jackson, DeJuan\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] 123.45 - 123 = 0.45"
},
{
"msg_contents": "> > Yes, someone pointed this out to me in private e-mail, and I wrote a C\n> > program to confirm it. I just had never seen such rounding on such\n> > small non-irrational numbers.\n> But, the problem is that it is irrational in base 2 (or at least so many\n> digits that it won't fit).\n\nAh, I figured that out after I sent the message.\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": "Tue, 15 Dec 1998 11:53:26 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45"
},
{
"msg_contents": "\"Jackson, DeJuan\" <[email protected]> writes:\n> But, the problem is that it is irrational in base 2 (or at least so many\n> digits that it won't fit).\n\nIt's not irrational in any base. 123.45 = 12345/100 is a ratio of\nintegers, ie, rational. The problem is that the exact representation\nof this number as a base-2 fraction is longer than the 52 or so bits\navailable in float8 format. (I think it's an infinite repeating\nfraction, in the same way that 1/9 = 0.111111111111111111111111111...\nis a repeating fraction in base 10. But even a terminating fraction\nwill get rounded off as a float8 if it takes more than 52 bits.)\n\nIrrational numbers are those which are not expressible as the ratio of\nany two integers, eg, square root of 2. (The ancient Greeks became\n*very* unhappy when they realized that there were such things.) A\nsubcategory is transcendentals, which are not expressible as the\nsolution of any algebraic equation, eg, pi.\n\nRational numbers have either terminating or repeating expansions in\nany base you care to use; irrationals never do.\n\nSorry, I'll get off my math-pedant soapbox now. Just tend to get\nannoyed when people misuse technical terms.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 15 Dec 1998 19:21:15 -0500",
"msg_from": "Tom Lane <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45 "
},
{
"msg_contents": "> Rational numbers have either terminating or repeating expansions in\n> any base you care to use; irrationals never do.\n> \n> Sorry, I'll get off my math-pedant soapbox now. Just tend to get\n> annoyed when people misuse technical terms.\n\nAnd I was a math minor in college. I am embarassed at what I have\nforgotten.\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": "Tue, 15 Dec 1998 19:29:32 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] 123.45 - 123 = 0.45"
}
] |
[
{
"msg_contents": "> > > Oops, sorry...meant to commit the patch from Thomas for \n> > > tzn->CTZName\n> > So it worked?\n> Like a charm...I'm trying to finish the outfuncs.c cleanups before\n> I do my next regression test/build, butthe compile went through \n> clean...\n\nOK, just keep in the back of your mind that this is a *very* simple fix\nwhich will either work or will cause problems, some of which subtle.\n\nIf we test on linux/glibc2 and FreeBSD that will make me feel more\nconfident about it for the v6.4.1 release.\n\n - Tom\n",
"msg_date": "Tue, 15 Dec 1998 19:55:00 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [COMMITTERS] 'pgsql/src/backend/utils/adt nabstime.c'"
}
] |
[
{
"msg_contents": "FYI, here is an announcement and url for the IBM DB2 beta. There is a CD\n(4-6 weeks delivery :( ) or a 'net download (~65MB!).\n\nThis is glibc2 only, which I have running at work, but haven't had a\nchance to try it.\n\n - Tom\nhi all.\nlast friday I got e-mail news from IBM.\nmaybe interested to you.\njohn\n\n\n\nDate: Fri, 11 Dec 1998 17:05:05 GMT\nSubject: IBM e-business and Software News Alert, 1998 Issue 34\nDOWNLOAD: Beta of IBM DB2 UniversalDeveloper's Edition Version 5.2 for Linux\n-----------------------------------------\nWith DB2 Universal Developer's Edition Version 5.2\nfor Linux, developers can combine the power of\nLinux with the robustness of DB2 to create\ne-business applications. DB2 for Linux features\nthe DB2 UDB Web Control Center for doing database\nadministration tasks over the Web, and it supports\nJava database applications. Download this beta,\nand beta code for Net.Data* Version 2 for Linux.\n http://www.ibm.com/software/data/db2/linux/",
"msg_date": "Tue, 15 Dec 1998 20:04:44 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "[Fwd: SGVLLUG DOWNLOAD: Beta of IBM DB2 5.2 for Linux]"
}
] |
[
{
"msg_contents": "(taken on-list so this is indexed in mhonarc)\n\n> Looking thru the freebsd sgml primer, they specify that \n> SGML_CATALOG_FILES needs to be set.\n> I find this to be true if I want to use PSGML mode in\n> emacs, and if I want to run nsgmls when validating within emacs, \n> otherwise it can't find the proper rules for parsing.\n> \n> Anyway the locations for the catalog files as per the primer:\n> /usr/local/share/sgml/docbook/3.0/catalog\n> /usr/local/share/sgml/html/catalog\n> /usr/local/share/sgml/iso8879/catalog\n> /usr/local/share/sgml/jade/catalog\n\nBingo. I can build docs on postgresql.org (a FreeBSD machine) using the\nFreeBSD packages of the tools if I set an environment variable,\nspecifying these four paths delimited with colons:\n\nsetenv SGML_CATALOG_FILES\n'/usr/local/share/sgml/docbook/3.0/catalog:...'\n\nOn my linux/rpm machine, I don't need to set this environment variable\nat all (except for emacs parsing) since jade was built with the correct\ndefault route to catalogs. Don't know why FreeBSD does not do this;\nperhaps this should be written up as a \"feature deficit\" or bug?\n\n - Tom\n",
"msg_date": "Tue, 15 Dec 1998 20:32:20 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "[Fwd: catalog files]"
},
{
"msg_contents": "\nOn 15-Dec-98 Thomas G. Lockhart wrote:\n> (taken on-list so this is indexed in mhonarc)\n> \n>> Looking thru the freebsd sgml primer, they specify that \n>> SGML_CATALOG_FILES needs to be set.\n>> I find this to be true if I want to use PSGML mode in\n>> emacs, and if I want to run nsgmls when validating within emacs, \n>> otherwise it can't find the proper rules for parsing.\n>> \n>> Anyway the locations for the catalog files as per the primer:\n>> /usr/local/share/sgml/docbook/3.0/catalog\n>> /usr/local/share/sgml/html/catalog\n>> /usr/local/share/sgml/iso8879/catalog\n>> /usr/local/share/sgml/jade/catalog\n> \n> Bingo. I can build docs on postgresql.org (a FreeBSD machine) using the\n> FreeBSD packages of the tools if I set an environment variable,\n> specifying these four paths delimited with colons:\n> \n> setenv SGML_CATALOG_FILES\n> '/usr/local/share/sgml/docbook/3.0/catalog:...'\n> \n> On my linux/rpm machine, I don't need to set this environment variable\n> at all (except for emacs parsing) since jade was built with the correct\n> default route to catalogs. Don't know why FreeBSD does not do this;\n> perhaps this should be written up as a \"feature deficit\" or bug?\n> \n> - Tom\n\nI'm building most of 'em anyway. I'm getting this after it successfully \nbuilds admin, postgres and programmer:\n\n\njade -D ref -D ../graphics -V %use-id-as-filename% -d\n/home/vev/db124/docbook/html/docbook.dsl -t sgml tutorial.sgml\njade:intro.sgml:71:1:E: general entity \"history\" not defined and no default\nentity\njade:intro.sgml:73:1:E: general entity \"about\" not defined and no default\nentity\njade:intro.sgml:75:1:E: general entity \"info\" not defined and no default\nentity\njade:intro.sgml:77:1:E: general entity \"notation\" not defined and no default\nentity\njade:intro.sgml:79:1:E: general entity \"y2k\" not defined and no default entity\njade:intro.sgml:81:1:E: general entity \"legal\" not defined and no default\nentity\njade:start.sgml:229:26:X: reference to non-existent ID \"LIBPQ\"\njade:start.sgml:229:42:X: reference to non-existent ID \"LIBPQ\"\njade:/home/vev/db124/docbook/html/dblink.dsl:184:1:E: XRef LinkEnd to missing\nID 'LIBPQ'\ngmake[2]: *** [tutorial.html] Error 1\n\n\nOne more thing I'm missing maybe?\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] flame-mail: /dev/null\n # include <std/disclaimers.h> TEAM-OS2\n Online Searchable Campground Listings http://www.camping-usa.com\n \"There is no outfit less entitled to lecture me about bloat\n than the federal government\" -- Tony Snow\n==========================================================================\n\n\n",
"msg_date": "Tue, 15 Dec 1998 15:40:29 -0500 (EST)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [Fwd: catalog files]"
},
{
"msg_contents": "> I'm building most of 'em anyway. I'm getting this after it \n> successfully builds admin, postgres and programmer:\n<snip>\n> One more thing I'm missing maybe?\n\nNot at all. I get the same thing :)\n\nI did not rebuild the tutorial for the last release, but had\nrestructured the generic introduction used in several docs to separate\nout the legaleze, etc. tutorial.sgml is just missing a few definitions\nfor those entities...\n\nfwiw, if you build postgres.html you will find all the information from\nall the other standalone documents. I did that because in the html\nformat it is nice to be able to click around all of the info, rather\nthan having to open another document.\n\nSo, we have FreeBSD nailed for building documents. Vince, can you jot\ndown some notes on exactly what you did to get things working? There is\na section in docguide.sgml (Appendix A in postgres.html) which discusses\ninstalling tools, and the only two methods described so far are linux\nrpm and from-sources.\n\nAlso, if anyone has tried installing a relatively fresh sgml-tools\npackage, it would be great to write up that one too. btw, the sgml-tools\nfolks have taken over the Unix-side issues for jade, which was always a\ntroublesome spot.\n\n - Tom\n",
"msg_date": "Tue, 15 Dec 1998 22:41:55 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] RE: [Fwd: catalog files]"
},
{
"msg_contents": "\nOn 15-Dec-98 Thomas G. Lockhart wrote:\n\n> So, we have FreeBSD nailed for building documents. Vince, can you jot\n> down some notes on exactly what you did to get things working? There is\n> a section in docguide.sgml (Appendix A in postgres.html) which discusses\n> installing tools, and the only two methods described so far are linux\n> rpm and from-sources.\n\nWorkin on it now.\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: [email protected] flame-mail: /dev/null\n # include <std/disclaimers.h> TEAM-OS2\n Online Searchable Campground Listings http://www.camping-usa.com\n \"There is no outfit less entitled to lecture me about bloat\n than the federal government\" -- Tony Snow\n==========================================================================\n\n\n",
"msg_date": "Tue, 15 Dec 1998 18:31:36 -0500 (EST)",
"msg_from": "Vince Vielhaber <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] RE: [Fwd: catalog files]"
}
] |
[
{
"msg_contents": "Your (postgresql.org) DNS entry seems to be suspended. \n\nAlso, It mentions that the -i option for the postmaster\nwill allow TCP/IP connections, however, it seems to\nput it in interactive mode. How can I get Java to work\nwith postgresql? \n\nThanks. \nClark. Evans\n\n",
"msg_date": "Tue, 15 Dec 1998 18:37:39 -0500 (EST)",
"msg_from": "Clark Evans <[email protected]>",
"msg_from_op": true,
"msg_subject": "None"
}
] |
[
{
"msg_contents": "Hey, check it out:\n\npostgres=> select case_tbl.*;\n i| f\n---+-----\n -3| 10.1\n -6| 20.2\n -9|-30.3\n-12|\n(4 rows)\n\nNotes in the parser say this was allowed with POSTQUEL, so was retained\nwhen implementing SQL...\n",
"msg_date": "Wed, 16 Dec 1998 01:39:33 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Old parser feature"
},
{
"msg_contents": "> Hey, check it out:\n> \n> postgres=> select case_tbl.*;\n> i| f\n> ---+-----\n> -3| 10.1\n> -6| 20.2\n> -9|-30.3\n> -12|\n> (4 rows)\n> \n> Notes in the parser say this was allowed with POSTQUEL, so was retained\n> when implementing SQL...\n\nYes, nice feature. QUEL allowed:\n\n\tretrieve (case_tbl.all)\n\nIn fact, QUEL never had a FROM clause in the SQL sense.\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": "Tue, 15 Dec 1998 20:46:50 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Old parser feature"
}
] |
[
{
"msg_contents": "... in SELECT * FROM street;\n\nVadim\n",
"msg_date": "Wed, 16 Dec 1998 18:00:56 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": true,
"msg_subject": "CURRENT: crash in select_view regression test..."
}
] |
[
{
"msg_contents": "CVS is just updated...\n\nPlease try concurrent writes/reads...\n\nIt'll take some time to implement READ COMMITTED mode...\n\nNote:\n\n1. Vacuum is not updated yet and so will remove deleted tuples\n though some xactions would like to see them in selects...\n\n2. LOCK TABLE modes are not yet implemented...\n\n3. As well as SELECT ... FOR UPDATE: I need in advice here!\n\n Do Oracle, Informix etc support SELECT FOR UPDATE if\n there are aggregates, group by, distinct in SELECT ???\n Is SELECT FOR UPDATE allowed in subqueries ???\n\n4. Rtree & gist indices use _relation level_ locking -\n no wish, no time to change them...\n Executor locks them in exclusive mode (on relation level)\n for update/insert and so prevents deadlocks...\n These locks released after _query_ is done - not so bad\n for concurrency...\n\n5. Hash & btree indices use _page level_ locking and so\n are \"opened\" for deadlocks -:))\n (In the case of concurrent inserts/updates when\n indices are used for scans).\n \n I hope to change btree - it's possible, -\n but not hashes...\n\n6. I'm not happy with current deadlock detection code!\n It seems that backend does DeadLockCheck each time\n when timer expired - shouldn't this be done _once_,\n before backend is going to sleep ?!\n\n7. As I read in Sybase documentation the default \n standard transaction mode is _chained_ - all\n queries before explicit COMMIT/ABORT are run in _single\n transaction_... But we have to use BEGIN/END to get it!\n This was not so bad for system with relation level locks,\n but now only the same row writes block one other and so\n chained mode seems more appropriate...\n \n Shouldn't we change default transaction mode now?\n And use option/SET TRANSACTION MODE to switch to\n un-chained mode if one like it?\n\nVadim\n",
"msg_date": "Wed, 16 Dec 1998 20:00:54 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": true,
"msg_subject": "MVCC works in serialized mode!"
},
{
"msg_contents": "> 6. I'm not happy with current deadlock detection code!\n> It seems that backend does DeadLockCheck each time\n> when timer expired - shouldn't this be done _once_,\n> before backend is going to sleep ?!\n\nNot sure. Now that I think of it, it makes sense that if I go to sleep,\na deadlock is not sudenly going to appear while I am asleep. If a new\nprocess causes a deadlock, the new process that causes it will see it.\n\nI did not check when I went to sleep because I thought it may be too\ncpu-intensive to do checking on every sleep, but now that I remember it,\nit may be very trivial in cpu time to do the check on every sleep.\n\nI recommend changing it to do it just before every sleep. Let me know\nif you want me to make the change.\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": "Wed, 16 Dec 1998 13:00:34 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> > 6. I'm not happy with current deadlock detection code!\n> > It seems that backend does DeadLockCheck each time\n> > when timer expired - shouldn't this be done _once_,\n> > before backend is going to sleep ?!\n> \n> Not sure. Now that I think of it, it makes sense that if I go to sleep,\n> a deadlock is not sudenly going to appear while I am asleep. If a new\n> process causes a deadlock, the new process that causes it will see it.\n> \n> I did not check when I went to sleep because I thought it may be too\n> cpu-intensive to do checking on every sleep, but now that I remember it,\n> it may be very trivial in cpu time to do the check on every sleep.\n> \n> I recommend changing it to do it just before every sleep. Let me know\n> if you want me to make the change.\n\nMay be we could just reduce first sleep time (60 sec is too long),\ndo DeadLockCheck _only once_, after first SIGALARM, and after that\njust sleep forever ?\nWhy do DeadLockCheck many times ?\n\nLet's think more...\n\nAnd please consider when lock conflict occures:\n\n1. One process tries update row being updated by other.\n2. When reading/writing hashes (I hope to change btrees to\n use new buffer context lock code, as heap access methods\n do, - this is short term locking without deadlocks and so -\n without using lockmanager).\n\nVadim\n",
"msg_date": "Thu, 17 Dec 1998 13:14:39 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "> Bruce Momjian wrote:\n> > \n> > > 6. I'm not happy with current deadlock detection code!\n> > > It seems that backend does DeadLockCheck each time\n> > > when timer expired - shouldn't this be done _once_,\n> > > before backend is going to sleep ?!\n> > \n> > Not sure. Now that I think of it, it makes sense that if I go to sleep,\n> > a deadlock is not sudenly going to appear while I am asleep. If a new\n> > process causes a deadlock, the new process that causes it will see it.\n> > \n> > I did not check when I went to sleep because I thought it may be too\n> > cpu-intensive to do checking on every sleep, but now that I remember it,\n> > it may be very trivial in cpu time to do the check on every sleep.\n> > \n> > I recommend changing it to do it just before every sleep. Let me know\n> > if you want me to make the change.\n> \n> May be we could just reduce first sleep time (60 sec is too long),\n\nYes, much too long. 5 or 15 seconds sounds good.\n\n> do DeadLockCheck _only once_, after first SIGALARM, and after that\n> just sleep forever ?\n> Why do DeadLockCheck many times ?\n\nYes, no need to do many times.\n\nOK. Should I make the change?\n\n> \n> Let's think more...\n> \n> And please consider when lock conflict occures:\n> \n> 1. One process tries update row being updated by other.\n> 2. When reading/writing hashes (I hope to change btrees to\n> use new buffer context lock code, as heap access methods\n> do, - this is short term locking without deadlocks and so -\n> without using lockmanager).\n\nOh. I would think even a 1 second deadlock detection would work well. \nMost short-term locks don't last that long.\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, 18 Dec 1998 01:49:09 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "> Bruce Momjian wrote:\n> > \n> > > 6. I'm not happy with current deadlock detection code!\n> > > It seems that backend does DeadLockCheck each time\n> > > when timer expired - shouldn't this be done _once_,\n> > > before backend is going to sleep ?!\n> > \n> > Not sure. Now that I think of it, it makes sense that if I go to sleep,\n> > a deadlock is not sudenly going to appear while I am asleep. If a new\n> > process causes a deadlock, the new process that causes it will see it.\n> > \n> > I did not check when I went to sleep because I thought it may be too\n> > cpu-intensive to do checking on every sleep, but now that I remember it,\n> > it may be very trivial in cpu time to do the check on every sleep.\n> > \n> > I recommend changing it to do it just before every sleep. Let me know\n> > if you want me to make the change.\n> \n> May be we could just reduce first sleep time (60 sec is too long),\n> do DeadLockCheck _only once_, after first SIGALARM, and after that\n> just sleep forever ?\n> Why do DeadLockCheck many times ?\n> \n> Let's think more...\n> \n> And please consider when lock conflict occures:\n> \n> 1. One process tries update row being updated by other.\n> 2. When reading/writing hashes (I hope to change btrees to\n> use new buffer context lock code, as heap access methods\n> do, - this is short term locking without deadlocks and so -\n> without using lockmanager).\n> \n> Vadim\n> \n\nI have applied the following patch. It causes only one deadlock check\nafter a sleep of one second. Applied only to the CURRENT tree.\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\nIndex: src/backend/parser/scan.c\n===================================================================\nRCS file: /usr/local/cvsroot/pgsql/src/backend/parser/scan.c,v\nretrieving revision 1.31\ndiff -c -r1.31 scan.c\n*** scan.c\t1998/10/13 17:26:50\t1.31\n--- scan.c\t1998/12/18 19:30:54\n***************\n*** 1,7 ****\n /* A lexical scanner generated by flex */\n \n /* Scanner skeleton version:\n! * $Header: /usr/local/cvsroot/pgsql/src/backend/parser/scan.c,v 1.31 1998/10/13 17:26:50 scrappy Exp $\n */\n \n #define FLEX_SCANNER\n--- 1,7 ----\n /* A lexical scanner generated by flex */\n \n /* Scanner skeleton version:\n! * /master/usr.bin/lex/skel.c,v 1.3 1997/09/25 00:10:23 jch Exp\n */\n \n #define FLEX_SCANNER\n***************\n*** 556,562 ****\n *\n *\n * IDENTIFICATION\n! *\t $Header: /usr/local/cvsroot/pgsql/src/backend/parser/scan.c,v 1.31 1998/10/13 17:26:50 scrappy Exp $\n *\n *-------------------------------------------------------------------------\n */\n--- 556,562 ----\n *\n *\n * IDENTIFICATION\n! *\t $Header: /usr/local/cvsroot/pgsql/src/backend/parser/scan.l,v 1.44 1998/10/08 18:29:51 momjian Exp $\n *\n *-------------------------------------------------------------------------\n */\nIndex: src/backend/storage/lmgr/proc.c\n===================================================================\nRCS file: /usr/local/cvsroot/pgsql/src/backend/storage/lmgr/proc.c,v\nretrieving revision 1.43\ndiff -c -r1.43 proc.c\n*** proc.c\t1998/09/01 04:32:02\t1.43\n--- proc.c\t1998/12/18 19:30:59\n***************\n*** 77,83 ****\n #include \"storage/proc.h\"\n #include \"utils/trace.h\"\n \n! static void HandleDeadLock(int sig);\n static PROC *ProcWakeup(PROC *proc, int errType);\n \n #define DeadlockCheckTimer pg_options[OPT_DEADLOCKTIMEOUT]\n--- 77,83 ----\n #include \"storage/proc.h\"\n #include \"utils/trace.h\"\n \n! static void HandleDeadLock(void);\n static PROC *ProcWakeup(PROC *proc, int errType);\n \n #define DeadlockCheckTimer pg_options[OPT_DEADLOCKTIMEOUT]\n***************\n*** 154,161 ****\n \t * Routine called if deadlock timer goes off. See ProcSleep()\n \t * ------------------\n \t */\n- \tpqsignal(SIGALRM, HandleDeadLock);\n- \n \tSpinAcquire(ProcStructLock);\n \n \t/* attach to the free list */\n--- 154,159 ----\n***************\n*** 449,457 ****\n \t\t TransactionId xid)\t/* needed by user locks, see below */\n {\n \tint\t\t\ti;\n \tPROC\t *proc;\n! \tstruct itimerval timeval,\n! \t\t\t\tdummy;\n \n \t/*\n \t * If the first entries in the waitQueue have a greater priority than\n--- 447,455 ----\n \t\t TransactionId xid)\t/* needed by user locks, see below */\n {\n \tint\t\t\ti;\n+ \tbool\t\tdeadlock_checked = false;\n \tPROC\t *proc;\n! \tstruct timeval timeval;\n \n \t/*\n \t * If the first entries in the waitQueue have a greater priority than\n***************\n*** 523,539 ****\n \t * to 0.\n \t * --------------\n \t */\n! \tMemSet(&timeval, 0, sizeof(struct itimerval));\n! \ttimeval.it_value.tv_sec = \\\n \t\t(DeadlockCheckTimer ? DeadlockCheckTimer : DEADLOCK_CHECK_TIMER);\n \n \tdo\n \t{\n \t\tMyProc->errType = NO_ERROR;\t\t/* reset flag after deadlock check */\n \n! \t\tif (setitimer(ITIMER_REAL, &timeval, &dummy))\n \t\t\telog(FATAL, \"ProcSleep: Unable to set timer for process wakeup\");\n \n \t\t/* --------------\n \t\t * if someone wakes us between SpinRelease and IpcSemaphoreLock,\n \t\t * IpcSemaphoreLock will not block. The wakeup is \"saved\" by\n--- 521,546 ----\n \t * to 0.\n \t * --------------\n \t */\n! \tMemSet(&timeval, 0, sizeof(struct timeval));\n! \ttimeval.tv_sec = \\\n \t\t(DeadlockCheckTimer ? DeadlockCheckTimer : DEADLOCK_CHECK_TIMER);\n \n \tdo\n \t{\n+ \t\tint expire;\n+ \t\t\n \t\tMyProc->errType = NO_ERROR;\t\t/* reset flag after deadlock check */\n \n! \t\tif ((expire = select(0, NULL, NULL, NULL,\n! \t\t\t(deadlock_checked == false) ? &timeval : NULL)) == -1)\n \t\t\telog(FATAL, \"ProcSleep: Unable to set timer for process wakeup\");\n \n+ \t\tif (expire == 0 /* timeout reached */ && deadlock_checked == false)\n+ \t\t{\n+ \t\t\tHandleDeadLock();\n+ \t\t\tdeadlock_checked = true;\n+ \t\t}\n+ \t\t\n \t\t/* --------------\n \t\t * if someone wakes us between SpinRelease and IpcSemaphoreLock,\n \t\t * IpcSemaphoreLock will not block. The wakeup is \"saved\" by\n***************\n*** 545,558 ****\n \t} while (MyProc->errType == STATUS_NOT_FOUND);\t\t/* sleep after deadlock\n \t\t\t\t\t\t\t\t\t\t\t\t\t\t * check */\n \n- \t/* ---------------\n- \t * We were awoken before a timeout - now disable the timer\n- \t * ---------------\n- \t */\n- \ttimeval.it_value.tv_sec = 0;\n- \tif (setitimer(ITIMER_REAL, &timeval, &dummy))\n- \t\telog(FATAL, \"ProcSleep: Unable to diable timer for process wakeup\");\n- \n \t/* ----------------\n \t * We were assumed to be in a critical section when we went\n \t * to sleep.\n--- 552,557 ----\n***************\n*** 695,701 ****\n * --------------------\n */\n static void\n! HandleDeadLock(int sig)\n {\n \tLOCK\t *mywaitlock;\n \n--- 694,700 ----\n * --------------------\n */\n static void\n! HandleDeadLock()\n {\n \tLOCK\t *mywaitlock;\n \nIndex: src/pl/plpgsql/src/gram.c\n===================================================================\nRCS file: /usr/local/cvsroot/pgsql/src/pl/plpgsql/src/gram.c,v\nretrieving revision 1.1\ndiff -c -r1.1 gram.c\n*** gram.c\t1998/10/28 17:07:17\t1.1\n--- gram.c\t1998/12/18 19:31:12\n***************\n*** 65,71 ****\n *\t\t\t procedural language\n *\n * IDENTIFICATION\n! * $Header: /usr/local/cvsroot/pgsql/src/pl/plpgsql/src/gram.c,v 1.1 1998/10/28 17:07:17 momjian Exp $\n *\n * This software is copyrighted by Jan Wieck - Hamburg.\n *\n--- 65,71 ----\n *\t\t\t procedural language\n *\n * IDENTIFICATION\n! * $Header: /usr/local/cvsroot/pgsql/src/pl/plpgsql/src/gram.y,v 1.1 1998/08/24 19:14:47 momjian Exp $\n *\n * This software is copyrighted by Jan Wieck - Hamburg.\n *\nIndex: src/pl/plpgsql/src/scan.c\n===================================================================\nRCS file: /usr/local/cvsroot/pgsql/src/pl/plpgsql/src/scan.c,v\nretrieving revision 1.1\ndiff -c -r1.1 scan.c\n*** scan.c\t1998/10/28 17:07:17\t1.1\n--- scan.c\t1998/12/18 19:31:21\n***************\n*** 635,641 ****\n *\t\t\t procedural language\n *\n * IDENTIFICATION\n! * $Header: /usr/local/cvsroot/pgsql/src/pl/plpgsql/src/scan.c,v 1.1 1998/10/28 17:07:17 momjian Exp $\n *\n * This software is copyrighted by Jan Wieck - Hamburg.\n *\n--- 635,641 ----\n *\t\t\t procedural language\n *\n * IDENTIFICATION\n! * $Header: /usr/local/cvsroot/pgsql/src/pl/plpgsql/src/scan.l,v 1.1 1998/08/24 19:14:49 momjian Exp $\n *\n * This software is copyrighted by Jan Wieck - Hamburg.\n *",
"msg_date": "Fri, 18 Dec 1998 14:49:40 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> I have applied the following patch. It causes only one deadlock check\n> after a sleep of one second. Applied only to the CURRENT tree.\n\nNice, at least for now.\n\nVadim\n",
"msg_date": "Sat, 19 Dec 1998 18:47:40 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "Hi all,\n\n> -----Original Message-----\n> From: [email protected]\n> [mailto:[email protected]]On Behalf Of Vadim Mikheev\n> Sent: Wednesday, December 16, 1998 10:01 PM\n> To: [email protected]\n> Subject: [HACKERS] MVCC works in serialized mode!\n> \n> \n> CVS is just updated...\n> \n> Please try concurrent writes/reads...\n>\n\nI happend to enjoy MVCC a little in (v6.4.1) .\n\nReaders are never blocked and writers are blocked only by \nsame row writers as I expected.\nIt's so comfortable.\n\nBut I have a question.\nOnce transactions are blocked,it takes so long time to resume \nafter blocks were removed.\nWhy ?\nCurrently blocked transactions resume immediately after blocks \nwere removed.\n\nThanks.\n\nHiroshi Inoue\[email protected]\n",
"msg_date": "Fri, 25 Dec 1998 13:04:38 +0900",
"msg_from": "\"Hiroshi Inoue\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "[Charset iso-8859-1 unsupported, filtering to ASCII...]\n> Hi all,\n> \n> > -----Original Message-----\n> > From: [email protected]\n> > [mailto:[email protected]]On Behalf Of Vadim Mikheev\n> > Sent: Wednesday, December 16, 1998 10:01 PM\n> > To: [email protected]\n> > Subject: [HACKERS] MVCC works in serialized mode!\n> > \n> > \n> > CVS is just updated...\n> > \n> > Please try concurrent writes/reads...\n> >\n> \n> I happend to enjoy MVCC a little in (v6.4.1) .\n\nThat's funny.\n\n> \n> Readers are never blocked and writers are blocked only by \n> same row writers as I expected.\n> It's so comfortable.\n\nWow, that is cool.\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": "Thu, 24 Dec 1998 23:07:04 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "Hiroshi Inoue wrote:\n> \n> >\n> > CVS is just updated...\n> >\n> > Please try concurrent writes/reads...\n> >\n> \n> I happend to enjoy MVCC a little in (v6.4.1) .\n> \n> Readers are never blocked and writers are blocked only by\n> same row writers as I expected.\n> It's so comfortable.\n> \n> But I have a question.\n> Once transactions are blocked,it takes so long time to resume\n> after blocks were removed.\n> Why ?\n> Currently blocked transactions resume immediately after blocks\n> were removed.\n\nWhat do you mean?\nExample please...\n\nVadim\n",
"msg_date": "Fri, 25 Dec 1998 12:39:32 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "> -----Original Message-----\n> From: [email protected] [mailto:[email protected]]On Behalf Of Vadim\n> Mikheev\n> Sent: Friday, December 25, 1998 2:40 PM\n> To: Hiroshi Inoue\n> Cc: [email protected]\n> Subject: Re: [HACKERS] MVCC works in serialized mode!\n>\n>\n> Hiroshi Inoue wrote:\n> >\n> > >\n> > > CVS is just updated...\n> > >\n> > > Please try concurrent writes/reads...\n> > >\n> >\n> > I happend to enjoy MVCC a little in (v6.4.1) .\n> >\n> > Readers are never blocked and writers are blocked only by\n> > same row writers as I expected.\n> > It's so comfortable.\n> >\n> > But I have a question.\n> > Once transactions are blocked,it takes so long time to resume\n> > after blocks were removed.\n> > Why ?\n> > Currently blocked transactions resume immediately after blocks\n> > were removed.\n>\n> What do you mean?\n> Example please...\n>\n\ninoue=> create table t1 (key int,a int);\ninoue=> insert into t1 values (1,0);\n\n<Session-1> <Session-2>\n\ninoue=> begin; inoue=> begin;\nBEGIN BEGIN\ninoue=> update t1 set a=1 where key=1;\nUPDATE 1\n inoue=> update\nt1 set a=2 where key=1;\n\n\n [ blocked ]\ninoue=> end/abort;\nEND/ABORT\n [ after\nlong time .... ]\n ERROR: Can't\nserialize access due to concurrent update\n / UPDATE 1\n\n\nThanks.\n\nHiroshi Inoue\[email protected]\n\n",
"msg_date": "Fri, 25 Dec 1998 15:09:04 +0900",
"msg_from": "\"Hiroshi Inoue\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "Hiroshi Inoue wrote:\n> \n> > >\n> > > But I have a question.\n> > > Once transactions are blocked,it takes so long time to resume\n> > > after blocks were removed.\n> > > Why ?\n> > > Currently blocked transactions resume immediately after blocks\n> > > were removed.\n> >\n> > What do you mean?\n> > Example please...\n> >\n> [ after long time .... ]\n> ERROR: Can't serialize access due to concurrent update\n\nJust tried (with CURRENT) - no problems...\n\nVadim\n",
"msg_date": "Fri, 25 Dec 1998 15:06:40 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "> \n> 2. LOCK TABLE modes are not yet implemented...\n\nImplemented, as well as SET TRANSACTION ISOLATION LEVEL...\n\n> \n> 3. As well as SELECT ... FOR UPDATE: I need in advice here!\n> \n> Do Oracle, Informix etc support SELECT FOR UPDATE if\n> there are aggregates, group by, distinct in SELECT ???\n\nAs I see in Oracle documentation FOR UPDATE is disallowed\nin these cases... No objections ?\n\n> Is SELECT FOR UPDATE allowed in subqueries ???\n\nCould someone test this?\n\n> \n> 7. As I read in Sybase documentation the default\n> standard transaction mode is _chained_ - all\n> queries before explicit COMMIT/ABORT are run in _single\n> transaction_... But we have to use BEGIN/END to get it!\n> This was not so bad for system with relation level locks,\n> but now only the same row writes block one other and so\n> chained mode seems more appropriate...\n> \n> Shouldn't we change default transaction mode now?\n> And use option/SET TRANSACTION MODE to switch to\n> un-chained mode if one like it?\n\nNo comments on this?\nI would like to make BEGIN implicit...\nObjections?\n\nVadim\n",
"msg_date": "Fri, 25 Dec 1998 17:32:44 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "> > \n> > 7. As I read in Sybase documentation the default\n> > standard transaction mode is _chained_ - all\n> > queries before explicit COMMIT/ABORT are run in _single\n> > transaction_... But we have to use BEGIN/END to get it!\n> > This was not so bad for system with relation level locks,\n> > but now only the same row writes block one other and so\n> > chained mode seems more appropriate...\n> > \n> > Shouldn't we change default transaction mode now?\n> > And use option/SET TRANSACTION MODE to switch to\n> > un-chained mode if one like it?\n> \n> No comments on this?\n> I would like to make BEGIN implicit...\n> Objections?\n\nInformix requires BEGIN. Ingres does not, but has 'set autocommit'\nmode, so each SQL statement is its own transaction, like we have now.\n\nI always felt that if I go in to psql and run a query, I want it\napplied. I don't want to have to commit every query I type.\n\nI know I can use SET TRANSACTION MODE to change this.\n\nWhat do others think?\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, 25 Dec 1998 06:29:52 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "Bruce Momjian wrote:\n> > > 7. As I read in Sybase documentation the default\n> > > standard transaction mode is _chained_ - all\n> > > queries before explicit COMMIT/ABORT are run in _single\n> > > transaction_... But we have to use BEGIN/END to get it!\n> > > This was not so bad for system with relation level locks,\n> > > but now only the same row writes block one other and so\n> > > chained mode seems more appropriate...\n> > >\n> > > Shouldn't we change default transaction mode now?\n> > > And use option/SET TRANSACTION MODE to switch to\n> > > un-chained mode if one like it?\n> >\n> > No comments on this?\n> > I would like to make BEGIN implicit...\n> > Objections?\n> \n> Informix requires BEGIN. Ingres does not, but has 'set autocommit'\n> mode, so each SQL statement is its own transaction, like we have now.\n> \n> I always felt that if I go in to psql and run a query, I want it\n> applied. I don't want to have to commit every query I type.\n> \n> I know I can use SET TRANSACTION MODE to change this.\n> \n> What do others think?\n\nAs long as we can switch. Interactive applications typically update more\nthan one objects per transaction.\n\n--------\nRegards\nTheo\n",
"msg_date": "Mon, 28 Dec 1998 16:04:56 +0200",
"msg_from": "Theo Kramer <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "> > No comments on this?\n> > I would like to make BEGIN implicit...\n> > Objections?\n> Informix requires BEGIN. Ingres does not, but has 'set autocommit'\n> mode, so each SQL statement is its own transaction, like we have now.\n\nimho we should have an Ingres-like feature for \"autocommit\". We can have\nthe default be \"set autocommit on\" (probably with an equals sign like\nour other \"set\" variables) and we can have it be a run-time option like\nDATESTYLE and other settable parameters. So you can configure your\nserver or your client environment to always behave the way you prefer.\n\nI'd be happy to implement the syntax and the environment variable parts;\njust need a global boolean value to check and set...\n\n - Tom\n",
"msg_date": "Mon, 28 Dec 1998 15:39:31 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "Hello!\n\nI want to make a filtered selection on a table. This can return with many\ntuples. The download of the query results may take a few minutes. I'd like\nto show the progress status. How can i do this?\nIf i run this in two query (first i count the number of lines, than the real\nquery), than the backend calculates the query two times.\nIf i use cursor, i can not get the line count before i fetch the whole table.\n\n--\n So long and thanx for all the fish.\n NeKo@(kva.hu|kornel.szif.hu)\n\n\n",
"msg_date": "Mon, 4 Jan 1999 17:01:48 +0100 (NFT)",
"msg_from": "\"Vazsonyi Peter[ke]\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Q: CURSOR tuples count"
},
{
"msg_contents": "\"Thomas G. Lockhart\" wrote:\n> \n> > > No comments on this?\n> > > I would like to make BEGIN implicit...\n> > > Objections?\n> > Informix requires BEGIN. Ingres does not, but has 'set autocommit'\n> > mode, so each SQL statement is its own transaction, like we have now.\n> \n> imho we should have an Ingres-like feature for \"autocommit\". We can have\n> the default be \"set autocommit on\" (probably with an equals sign like\n> our other \"set\" variables) and we can have it be a run-time option like\n> DATESTYLE and other settable parameters. So you can configure your\n> server or your client environment to always behave the way you prefer.\n> \n> I'd be happy to implement the syntax and the environment variable parts;\n> just need a global boolean value to check and set...\n\nNice! TIA -:)\n\nCould you also change configure stuff, to choose default mode\nbefore compiling?\n\nVadim\n",
"msg_date": "Mon, 11 Jan 1999 00:12:13 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
},
{
"msg_contents": "> > imho we should have an Ingres-like feature for \"autocommit\".\n> Nice! TIA -:)\n\nSure, but if I do too much work on this I'll need you to help on outer\njoins :) (Actually, that's just an excuse; I'm flailing on it trying to\nunderstand things you already know.)\n\nIs there a global boolean \"autocommit flag\" already there to work with,\nor could you add one? I'd like to (if possible) keep from getting sucked\ninto non-parser/non-docs work until I've worked out some of the outer\njoin stuff. But I can do the parser and \"set autocommit\" stuff. Also,\nI'd like to go through the gram.y code for the new MVCC support commands\nand have it use parser keywords rather than \"Ident\" entities for the\nsyntax. OK?\n\n> Could you also change configure stuff, to choose default mode\n> before compiling?\n\nI'll set things up for it in the code, and perhaps Tom Lane or someone\ncan help with the configure support. We can ask when it's time...\n\n - Tom\n",
"msg_date": "Sun, 10 Jan 1999 17:42:29 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] MVCC works in serialized mode!"
}
] |
[
{
"msg_contents": "> Update of /usr/local/cvsroot/pgsql/src/backend/storage/ipc\n> In directory hub.org:/tmp/cvs-serv14856/storage/ipc\n> \n> Modified Files:\n> \tshmem.c \n> Log Message:\n> Serialized mode works!\n\nCongradulations on another giant leap for PostgreSQL.\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": "Wed, 16 Dec 1998 10:15:36 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [COMMITTERS] 'pgsql/src/backend/storage/ipc shmem.c'"
}
] |
[
{
"msg_contents": " Is PostgreSQL having any ideas to support GNU gettext package for\nlet's to suitable i18n ??\n\n--\n.....=======............................. Cd Chen, (嚙踝蕭嚙碼嚙踝蕭)\n..// 嚙編 嚙編 |............................ ===========================\n..|| 嚙踝蕭 嚙踝蕭 <............................ I am a PowerLinuxer !!\n..|< > |............................ mailto:[email protected]\n..| | \\___/ |............................ http://linux.ntcic.edu.tw/~cdchen\n.. |\\______/............................. ICQ UIN:3345768\n\n",
"msg_date": "Thu, 17 Dec 1998 05:22:47 +0800",
"msg_from": "Cd Chen <[email protected]>",
"msg_from_op": true,
"msg_subject": "Have any ideas to support GNU gettext package ??"
},
{
"msg_contents": "> Is PostgreSQL having any ideas to support GNU gettext package for\n> let's to suitable i18n ??\n\nafaik no one has proposed that yet. It sounds like a Good Thing in\nprinciple, but I don't know what would be involved in internationalizing\nthe message strings. It might substantially raise the level of effort\nfor development, since some of the developers are not adept at languages\nother than English and hence others would have to maintain the message\ncatalogs. Minor changes in messages would become more difficult I\nsuppose.\n\nWe would also need to stay compatible with the licensing for the rest of\nPostgres, but that may not be an issue here. I assume that there are no\nruntime restrictions on the code and files gettext produces?\n\n - Tom\n",
"msg_date": "Thu, 17 Dec 1998 03:32:57 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Have any ideas to support GNU gettext package ??"
},
{
"msg_contents": "\"Thomas G. Lockhart\" <[email protected]> writes:\n\n| It sounds like a Good Thing in principle, but I don't know what\n| would be involved in internationalizing the message strings. It\n| might substantially raise the level of effort for development, since\n| some of the developers are not adept at languages other than English\n| and hence others would have to maintain the message catalogs. Minor\n| changes in messages would become more difficult I suppose.\n\nThe developers should not be bothered by the localisations (l10n). It\nmight be the best to use Fran�ois Pinard's \"Robot\"; let me quote\nFran�ois' description (note: \"PO files\" are the message catalog files):\n\n======================================================================>\n\nThe Translation Project robot (or TP-Robot) is an email service which\ntakes care of PO file submissions. It checks if the file is receivable,\nthat is, if a translator has filled her disclaimer, and if her team allows\nher to do the work. It also calls `msgfmt' to see if the PO file is healthy.\n\nYou do not really ought to use the robot, but when you do, you spare some\ntime for the translation coordinator to do more useful things (thanks!).\nTo use the robot, your Subject line should look like:\n\n TP-Robot PACKAGE-VERSION.TEAM.po\n\nand the whole thing should be sent to <[email protected]>.\nYou may expect a reply within minutes. If you need help to resolve\nquestions, or if you plainly suspect the robot behaves poorly, you may\ndirectly write the translation coordinator. However, if you do so, be\ncareful _not_ to start your message Subject with TP-Robot! :-)\n\nThe robot is a bit dumb and does not understand MIME yet. So, for a\nwhile, please package your PO file with `uuencode' (either old `begin'\nor new `begin-base64' format) rather than MIME or `shar'. You may\nalso use a plain text copy, if you are real sure that we have 8-bit\nclean email and if you avoid adding your signature after the PO file.\nIf the Robot cannot decipher your packaging, it will merely forward it\nto the coordinator for manual unpacking.\n======================================================================<\n\nYou'll find detailed info about i18n/l10n at Fran�ois' web pages:\n\n http://www.iro.umontreal.ca/~pinard/\n\nAnd of course, it'd be nice to have frontend with NLS (\"native language\nsupport\")!\n\n| We would also need to stay compatible with the licensing for the rest of\n| Postgres, but that may not be an issue here. I assume that there are no\n| runtime restrictions on the code and files gettext produces?\n\nYes, you're right. The libintl is covered by the LGPL, like other GNU\nlibraries.\n\n-- \nKarl Eichwalder\n",
"msg_date": "17 Dec 1998 11:13:59 +0100",
"msg_from": "Karl Eichwalder <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: Have any ideas to support GNU gettext package ??"
}
] |
[
{
"msg_contents": "\nVadim Mikheev <[email protected]>\n> \n> ... in SELECT * FROM street;\n> \n\nNo crash just:-\n\nregression=> select * from street;\nERROR: nodeRead: Bad type 0\nregression=> \n\nKeith.\n\n",
"msg_date": "Wed, 16 Dec 1998 22:51:21 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] CURRENT: crash in select_view regression test..."
},
{
"msg_contents": "Keith Parks wrote:\n> \n> Vadim Mikheev <[email protected]>\n> >\n> > ... in SELECT * FROM street;\n> >\n> \n> No crash just:-\n> \n> regression=> select * from street;\n> ERROR: nodeRead: Bad type 0\n> regression=>\n\nIt seems platform dependent...\nBugs are in readfuncs.c\n\nVadim\n",
"msg_date": "Thu, 17 Dec 1998 11:00:58 +0700",
"msg_from": "Vadim Mikheev <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] CURRENT: crash in select_view regression test..."
}
] |
[
{
"msg_contents": "-----BEGIN PGP SIGNED MESSAGE-----\n\nIs anyone working on this? The two \"big\" features for me are this and\nthe outer join code. I'm willing to work on the alter table code, but\nexpect it will take a while for me to get up to speed....\n\nroland\n\n-----BEGIN PGP SIGNATURE-----\nVersion: 2.6.2\nComment: Processed by Mailcrypt 3.4, an Emacs/PGP interface\n\niQCVAwUBNnh0BeoW38lmvDvNAQE6EgP+M1kP+fMSXU0umP8DKZLbKJXUObawCMud\n86CweB6Q2Ac/18prtj7wTXA8GtMunE31LyETFVNMhr5bV3vYA2ZTeOmeFsf9/sbv\n2ULlYEbatZ4TB8QfUwU96bIBhaGvk7yJDr5EQpJl7l7I6vylGkQIfkn6IKy2+S/y\nJgqXlDEDixI=\n=pHyG\n-----END PGP SIGNATURE-----\n-- \nRoland B. Roberts, PhD Custom Software Solutions\[email protected] 101 West 15th St #4NN\n New York, NY 10011\n",
"msg_date": "16 Dec 1998 22:01:28 -0500",
"msg_from": "Roland Roberts <[email protected]>",
"msg_from_op": true,
"msg_subject": "alter table drop/alter column?"
}
] |
[
{
"msg_contents": "Vadim Mikheev <[email protected]>\n> \n> Keith Parks wrote:\n> > \n> > Vadim Mikheev <[email protected]>\n> > >\n> > > ... in SELECT * FROM street;\n> > >\n> > \n> > No crash just:-\n> > \n> > regression=> select * from street;\n> > ERROR: nodeRead: Bad type 0\n> > regression=>\n> \n> It seems platform dependent...\n> Bugs are in readfuncs.c\n> \n\nI think I've found it, a simple typo in outfuncs.c.\n\nLooks like :vartypmod got transmuted to %vartypmod in an editing session.\n\nHere's the patch,\n\nKeith.\n\n*** src/backend/nodes/outfuncs.c.orig Thu Dec 17 12:01:02 1998\n--- src/backend/nodes/outfuncs.c Thu Dec 17 12:01:22 1998\n***************\n*** 634,640 ****\n _outVar(StringInfo str, Var *node)\n {\n appendStringInfo(str,\n! \" VAR :varno %d :varattno %d :vartype %u %vartypmod %d \",\n node->varno,\n node->varattno,\n node->vartype,\n--- 634,640 ----\n _outVar(StringInfo str, Var *node)\n {\n appendStringInfo(str,\n! \" VAR :varno %d :varattno %d :vartype %u :vartypmod %d \",\n node->varno,\n node->varattno,\n node->vartype, \n\n",
"msg_date": "Thu, 17 Dec 1998 11:56:14 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] CURRENT: crash in select_view regression test..."
},
{
"msg_contents": "\nMy fault...patch applied...sorry about that :(\n\n\nOn Thu, 17 Dec 1998, Keith Parks wrote:\n\n> Vadim Mikheev <[email protected]>\n> > \n> > Keith Parks wrote:\n> > > \n> > > Vadim Mikheev <[email protected]>\n> > > >\n> > > > ... in SELECT * FROM street;\n> > > >\n> > > \n> > > No crash just:-\n> > > \n> > > regression=> select * from street;\n> > > ERROR: nodeRead: Bad type 0\n> > > regression=>\n> > \n> > It seems platform dependent...\n> > Bugs are in readfuncs.c\n> > \n> \n> I think I've found it, a simple typo in outfuncs.c.\n> \n> Looks like :vartypmod got transmuted to %vartypmod in an editing session.\n> \n> Here's the patch,\n> \n> Keith.\n> \n> *** src/backend/nodes/outfuncs.c.orig Thu Dec 17 12:01:02 1998\n> --- src/backend/nodes/outfuncs.c Thu Dec 17 12:01:22 1998\n> ***************\n> *** 634,640 ****\n> _outVar(StringInfo str, Var *node)\n> {\n> appendStringInfo(str,\n> ! \" VAR :varno %d :varattno %d :vartype %u %vartypmod %d \",\n> node->varno,\n> node->varattno,\n> node->vartype,\n> --- 634,640 ----\n> _outVar(StringInfo str, Var *node)\n> {\n> appendStringInfo(str,\n> ! \" VAR :varno %d :varattno %d :vartype %u :vartypmod %d \",\n> node->varno,\n> node->varattno,\n> node->vartype, \n> \n> \n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Thu, 17 Dec 1998 09:10:06 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] CURRENT: crash in select_view regression test..."
}
] |
[
{
"msg_contents": "Our server went down unexpectedly. I restarted postgres using\n\"postmaster -i &\" but now I can't access any existing databases\nor create new ones. When I run \"psql blc\" I get the following\nerror messages:\n\n$ psql blc\nIpcSemaphoreCreate: semget failed (Invalid argument) key=5432114,\nnum=16, permis\nsion=600\nConnection to database 'blc' failed.\npqReadData() -- backend closed the channel unexpectedly.\n This probably means the backend terminated abnormally before or\nwhile pr\nocessing the request.\n\nAny help would be much appreciated. I posted elsewhere first but never\ngot a response.\n\n-Anne Rose\n Human-Computer Interaction Lab\n University of Maryland\n\n",
"msg_date": "Thu, 17 Dec 1998 10:12:18 -0500",
"msg_from": "Anne Rose <[email protected]>",
"msg_from_op": true,
"msg_subject": "HELP restarting postmaster"
}
] |
[
{
"msg_contents": "> I have ported PostgreSQL 6.4 to Cobalt Qube running Linux 2.0.33 with\n> MIPS cpu (I think it's R4000). I have tested the patches on LinuxPPC\n> and FreeBSD. I believe they do not harm other platforms.\n\nShall we include this in the v6.4.1 release? It looks pretty harmless\nwrt other systems. I'll update the docs...\n\n - Tom\n",
"msg_date": "Thu, 17 Dec 1998 15:43:40 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [PATCHES] Cobalt/Linux porting patch"
},
{
"msg_contents": "On Thu, 17 Dec 1998, Thomas G. Lockhart wrote:\n\n> > I have ported PostgreSQL 6.4 to Cobalt Qube running Linux 2.0.33 with\n> > MIPS cpu (I think it's R4000). I have tested the patches on LinuxPPC\n> > and FreeBSD. I believe they do not harm other platforms.\n> \n> Shall we include this in the v6.4.1 release? It looks pretty harmless\n> wrt other systems. I'll update the docs...\n\n\tYou willing to stake your reputation on it? :)\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Thu, 17 Dec 1998 14:16:12 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [PATCHES] Cobalt/Linux porting patch"
},
{
"msg_contents": "> You willing to stake your reputation on it? :)\n\n*rofl*\n\nSo what's the rep: takes two patches to make anything work? :))\n\n - Tom\n",
"msg_date": "Thu, 17 Dec 1998 18:36:28 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [PATCHES] Cobalt/Linux porting patch"
},
{
"msg_contents": "On Thu, 17 Dec 1998, Thomas G. Lockhart wrote:\n\n> > You willing to stake your reputation on it? :)\n> \n> *rofl*\n> \n> So what's the rep: takes two patches to make anything work? :))\n\n\t*slap forehead*\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Thu, 17 Dec 1998 14:39:23 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [PATCHES] Cobalt/Linux porting patch"
}
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.