Datasets:
wanshenl
/
pgsql-hackers

Dataset card Data Studio Files
xet
Community
threads
listlengths
1
2.99k
[ { "msg_contents": "Hi all,\n\nVince asked me to forward this here.\n\nRegards and best wishes,\n\nJustin Clift\n\n-------- Original Message --------\nSubject: Re: [HACKERS] Call for platforms\nDate: Sun, 25 Mar 2001 19:45:37 -0500 (EST)\nFrom: Vince Vielhaber <vev@michvhf.com>\nTo: Justin Clift <aa2@bigpond.net.au>\n\nOn Mon, 26 Mar 2001, Justin Clift wrote:\n\n> Hi Vince,\n>\n> I'm wondering if it's worth putting something on the regresssion test\n> results page asking people to try ALL the options if possible?\n>\n> Like, I tested Solaris 8 SPARC with './configure --p=/opt/postgresql71\n> --enable-syslog', and that worked perfectly. BUT, when I started adding\n> in stuff like --with-openssl --with-python, there were nasty\n> interactions and things didn't work.\n>\n> It would be nice to know that on a Supported Platform that ALL of the\n> options worked (apart from the Perl option which we apparently can't\n> make 100% work).\n>\n> Regards and best wishes,\n>\n> Justin Clift\n>\n\nYou might want to pass that along to hackers. It'd be interesting to\nhear\ntheir thoughts.\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: vev@michvhf.com \nhttp://www.pop4.net\n 56K Nationwide Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n", "msg_date": "Mon, 26 Mar 2001 12:00:13 +1000", "msg_from": "Justin Clift <aa2@bigpond.net.au>", "msg_from_op": true, "msg_subject": "[Fwd: Re: Call for platforms]" } ]
[ { "msg_contents": "\nI'm seeing random failures in the parallel tests run by\n\"make check\". Sometimes (40% or so) all tests will succeed.\nThe rest of the time, I will get one or more failures.\n\nThe specific failures are connection failures:\n\n! psql: connectDBStart() -- connect() failed: Connection refused\n! Is the postmaster running locally\n! and accepting connections on Unix socket '/tmp/.s.PGSQL.65432'?\n\nI tried running pg_regress with --debug to see if it would\nshed some light on the matter. I didn't see anything in the\noutput indicating what was failing on the backend.\n\nI'm running on Solaris 8, with the 2nd update installed.\n\nI'm new to postgresql, could someone with some experience give\nme a few pointers on tracking this one down?\n\nSteve Nicolai\n\n", "msg_date": "Sun, 25 Mar 2001 21:47:43 -0600", "msg_from": "Steve Nicolai <snicolai@mac.com>", "msg_from_op": true, "msg_subject": "Random make check failures" }, { "msg_contents": "Steve Nicolai <snicolai@mac.com> writes:\n> I'm seeing random failures in the parallel tests run by\n> \"make check\". Sometimes (40% or so) all tests will succeed.\n> The rest of the time, I will get one or more failures.\n\n> The specific failures are connection failures:\n\n> ! psql: connectDBStart() -- connect() failed: Connection refused\n> ! Is the postmaster running locally\n> ! and accepting connections on Unix socket '/tmp/.s.PGSQL.65432'?\n\n> I'm running on Solaris 8, with the 2nd update installed.\n\nWe've heard reports that this goes away if you tweak the pg_regress\nscript to use TCP rather than Unix-socket connections (look for the\nbit that forces unix_sockets=no). Nobody seems to have any idea why\nSolaris has a problem with Unix-socket connections ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 26 Mar 2001 00:40:04 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Random make check failures " } ]
[ { "msg_contents": "\n> Mainly it's a schedule-driven thing. I don't want to take any risk that\n> a last-minute patch to work around AIX's broken compiler will break any\n> other platforms. If we had found this problem before beta cycle\n> started, I would be more willing to say \"let's try it and find out\n> whether it works everywhere\".\n> \n> Yeah, it's paranoia, but considering that the whole thing is an exercise\n> in covering up a \"shouldn't happen\" compiler bug, I think paranoia is\n> not so unreasonable.\n\nThank you for going to all that trouble !! The fix, as it is in RC1 works great \non AIX with both xlc and gcc.\n\nAndreas \n", "msg_date": "Mon, 26 Mar 2001 12:44:56 +0200", "msg_from": "Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at>", "msg_from_op": true, "msg_subject": "AW: AW: AW: Re: RELEASE STOPPER? nonportable int64 cons\n\ttant s in pg_crc.c" } ]
[ { "msg_contents": "Hi,\n\nThere's no bug with regexp, because \"*\" operator in\nregex match zero or more times. So if you want\nmatch exactly your patern try ~ 'ibd01.*' . \".\" match an\narbitrary character.\n\nCheers,\n\nPEJAC Pascal\n\n", "msg_date": "Mon, 26 Mar 2001 14:34:52 +0200 (CEST)", "msg_from": "<pejac@altern.org>", "msg_from_op": true, "msg_subject": "re : Prob with regexp" } ]
[ { "msg_contents": "PostgreSQL typically uses port 5432 for client-server communications.\nIt would be a good idea to register this with IANA. This will help to\navoid a clash with other services that might try to use the port.\nDB2, Interbase, MS SQL, MySQL, Oracle, Sybase, etc. are already\nregistered.\n\nMight someone with a reasonable grasp of the low level messages in\nPostgreSQL care to submit a registration?\n\n http://www.iana.org/\n http://www.iana.org/cgi-bin/usr-port-number.pl\n http://www.isi.edu/in-notes/iana/assignments/port-numbers\n\n-- \nPete Forman -./\\.- Disclaimer: This post is originated\nWesternGeco -./\\.- by myself and does not represent\npete.forman@westerngeco.com -./\\.- opinion of Schlumberger, Baker\nhttp://www.crosswinds.net/~petef -./\\.- Hughes or their divisions.\n", "msg_date": "Mon, 26 Mar 2001 15:58:53 +0100", "msg_from": "Pete Forman <pete.forman@westerngeco.com>", "msg_from_op": true, "msg_subject": "IANA registration" }, { "msg_contents": "Pete Forman <pete.forman@westerngeco.com> writes:\n> PostgreSQL typically uses port 5432 for client-server communications.\n> It would be a good idea to register this with IANA. This will help to\n> avoid a clash with other services that might try to use the port.\n> Might someone with a reasonable grasp of the low level messages in\n> PostgreSQL care to submit a registration?\n\n> http://www.iana.org/\n> http://www.iana.org/cgi-bin/usr-port-number.pl\n> http://www.isi.edu/in-notes/iana/assignments/port-numbers\n\nThis seems like a good idea. They seem to want a particular person's\nname as contact, so I suppose one of the core group should do it.\nHas anyone already initiated the process; if not, shall I?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 15:33:57 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: IANA registration " }, { "msg_contents": "> Pete Forman <pete.forman@westerngeco.com> writes:\n> > PostgreSQL typically uses port 5432 for client-server communications.\n> > It would be a good idea to register this with IANA. This will help to\n> > avoid a clash with other services that might try to use the port.\n> > Might someone with a reasonable grasp of the low level messages in\n> > PostgreSQL care to submit a registration?\n> \n> > http://www.iana.org/\n> > http://www.iana.org/cgi-bin/usr-port-number.pl\n> > http://www.isi.edu/in-notes/iana/assignments/port-numbers\n> \n> This seems like a good idea. They seem to want a particular person's\n> name as contact, so I suppose one of the core group should do it.\n> Has anyone already initiated the process; if not, shall I?\n\nThis seemed like a Marc job to me. Marc?\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 27 Mar 2001 16:33:17 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: IANA registration" }, { "msg_contents": "> > > PostgreSQL typically uses port 5432 for client-server communications.\n> > > It would be a good idea to register this with IANA. This will help to\n> > > avoid a clash with other services that might try to use the port.\n> > > Might someone with a reasonable grasp of the low level messages in\n> > > PostgreSQL care to submit a registration?\n> >\n> > > http://www.iana.org/\n> > > http://www.iana.org/cgi-bin/usr-port-number.pl\n> > > http://www.isi.edu/in-notes/iana/assignments/port-numbers\n> >\n> > This seems like a good idea. They seem to want a particular person's\n> > name as contact, so I suppose one of the core group should do it.\n> > Has anyone already initiated the process; if not, shall I?\n> \n> This seemed like a Marc job to me. Marc?\n\nThey ask for some (high level) details on the protocol features/packet\nformat. If Tom (or someone; could be me but I've been swamped) could\nhelp out with a few sentences, we'll get this done a bit faster ;)\n\nPresumably for port 5432 it is just the connection handshake stuff...\n\n - Thomas\n", "msg_date": "Tue, 27 Mar 2001 23:17:28 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": false, "msg_subject": "Re: IANA registration" }, { "msg_contents": "On Tue, 27 Mar 2001, Bruce Momjian wrote:\n\n> > Pete Forman <pete.forman@westerngeco.com> writes:\n> > > PostgreSQL typically uses port 5432 for client-server communications.\n> > > It would be a good idea to register this with IANA. This will help to\n> > > avoid a clash with other services that might try to use the port.\n> > > Might someone with a reasonable grasp of the low level messages in\n> > > PostgreSQL care to submit a registration?\n> >\n> > > http://www.iana.org/\n> > > http://www.iana.org/cgi-bin/usr-port-number.pl\n> > > http://www.isi.edu/in-notes/iana/assignments/port-numbers\n> >\n> > This seems like a good idea. They seem to want a particular person's\n> > name as contact, so I suppose one of the core group should do it.\n> > Has anyone already initiated the process; if not, shall I?\n>\n> This seemed like a Marc job to me. Marc?\n\nWill do ... *nod*\n\n\n", "msg_date": "Tue, 27 Mar 2001 20:07:59 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: IANA registration" }, { "msg_contents": "The Hermit Hacker <scrappy@hub.org> writes:\n>> This seemed like a Marc job to me. Marc?\n\n> Will do ... *nod*\n\nOk, fine here. BTW, I was thinking that rather than filling in their\nwebform in any detail, it'd make the most sense to just supply a link\nto our protocol documentation, presently\n\nhttp://www.postgresql.org/users-lounge/docs/7.0/postgres/protocol.htm\nor\nhttp://www.postgresql.org/devel-corner/docs/postgres/protocol.html\n\nEven though the latter is nominally unstable docs, the URL doesn't\nhave built-in obsolescence in it ;-)\n\nMaybe we should arrange for a more stable path to the \"current release\"\ndocs, ie a symlink without a version number?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 19:13:22 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: IANA registration " }, { "msg_contents": "On Tue, 27 Mar 2001, Tom Lane wrote:\n\n> Ok, fine here. BTW, I was thinking that rather than filling in their\n> webform in any detail, it'd make the most sense to just supply a link\n> to our protocol documentation, presently\n>\n> http://www.postgresql.org/users-lounge/docs/7.0/postgres/protocol.htm\n> or\n> http://www.postgresql.org/devel-corner/docs/postgres/protocol.html\n>\n> Even though the latter is nominally unstable docs, the URL doesn't\n> have built-in obsolescence in it ;-)\n\nIf it's in the devel-corner it does!\n\n> Maybe we should arrange for a more stable path to the \"current release\"\n> docs, ie a symlink without a version number?\n\nBut there is...\n\nhttp://www.postgresql.org/docs/special/protocol.html\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: vev@michvhf.com http://www.pop4.net\n 56K Nationwide Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n", "msg_date": "Tue, 27 Mar 2001 19:35:27 -0500 (EST)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": false, "msg_subject": "Re: IANA registration " }, { "msg_contents": "Vince Vielhaber <vev@michvhf.com> writes:\n>> Maybe we should arrange for a more stable path to the \"current release\"\n>> docs, ie a symlink without a version number?\n\n> But there is...\n\n> http://www.postgresql.org/docs/special/protocol.html\n\nNo, that's just the first page of that chapter. If the 'next' link\nworked ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 19:38:47 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: IANA registration " }, { "msg_contents": "On Tue, 27 Mar 2001, Tom Lane wrote:\n\n> Vince Vielhaber <vev@michvhf.com> writes:\n> >> Maybe we should arrange for a more stable path to the \"current release\"\n> >> docs, ie a symlink without a version number?\n>\n> > But there is...\n>\n> > http://www.postgresql.org/docs/special/protocol.html\n>\n> No, that's just the first page of that chapter. If the 'next' link\n> worked ...\n\nIt's what you provided:\n\nhttp://www.postgresql.org/users-lounge/docs/7.0/postgres/protocol.htm\n\nBut it's been changed.\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: vev@michvhf.com http://www.pop4.net\n 56K Nationwide Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n", "msg_date": "Tue, 27 Mar 2001 19:45:01 -0500 (EST)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": false, "msg_subject": "Re: IANA registration " }, { "msg_contents": "\nDid we get anywhere with this?\n\n> PostgreSQL typically uses port 5432 for client-server communications.\n> It would be a good idea to register this with IANA. This will help to\n> avoid a clash with other services that might try to use the port.\n> DB2, Interbase, MS SQL, MySQL, Oracle, Sybase, etc. are already\n> registered.\n> \n> Might someone with a reasonable grasp of the low level messages in\n> PostgreSQL care to submit a registration?\n> \n> http://www.iana.org/\n> http://www.iana.org/cgi-bin/usr-port-number.pl\n> http://www.isi.edu/in-notes/iana/assignments/port-numbers\n> \n> -- \n> Pete Forman -./\\.- Disclaimer: This post is originated\n> WesternGeco -./\\.- by myself and does not represent\n> pete.forman@westerngeco.com -./\\.- opinion of Schlumberger, Baker\n> http://www.crosswinds.net/~petef -./\\.- Hughes or their divisions.\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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 May 2001 18:23:22 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: IANA registration" }, { "msg_contents": "Bruce Momjian writes:\n > Did we get anywhere with this?\n\nMarc said on 2001-03-27 that he was going to make the submission.\n\nThe port assignments listing published by IANA was last updated on\n2001-05-03 and does not yet have an entry for 5432.\n-- \nPete Forman -./\\.- Disclaimer: This post is originated\nWesternGeco -./\\.- by myself and does not represent\npete.forman@westerngeco.com -./\\.- opinion of Schlumberger, Baker\nhttp://www.crosswinds.net/~petef -./\\.- Hughes or their divisions.\n", "msg_date": "Tue, 8 May 2001 09:03:34 +0100", "msg_from": "Pete Forman <pete.forman@westerngeco.com>", "msg_from_op": true, "msg_subject": "Re: IANA registration" } ]
[ { "msg_contents": "I posted this to the list a few days ago, but it was not addressed. Can\nsomeone please make the changes to the src/test/regress/resultmap?\n\nThanks.\n\n- brandon\n\n\nb. palmer, bpalmer@crimelabs.net\npgp: www.crimelabs.net/bpalmer.pgp5", "msg_date": "Mon, 26 Mar 2001 12:27:30 -0500 (EST)", "msg_from": "bpalmer <bpalmer@crimelabs.net>", "msg_from_op": true, "msg_subject": "OBSD resultmap changes needed. " }, { "msg_contents": "Applied. I thought there was some openbsd confusion results so I\nignored the previous patch.\n\n> I posted this to the list a few days ago, but it was not addressed. Can\n> someone please make the changes to the src/test/regress/resultmap?\n> \n> Thanks.\n> \n> - brandon\n> \n> \n> b. palmer, bpalmer@crimelabs.net\n> pgp: www.crimelabs.net/bpalmer.pgp5\n> \nContent-Description: \n\n[ Attachment, skipping... ]\n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 5: Have you checked our extensive FAQ?\n> \n> http://www.postgresql.org/users-lounge/docs/faq.html\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 26 Mar 2001 12:51:42 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: OBSD resultmap changes needed." }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Applied. I thought there was some openbsd confusion results so I\n> ignored the previous patch.\n\nYeah, I thought there was still some debate about whether we knew the\ncorrect resultmap for *bsd or not ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 26 Mar 2001 13:20:21 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: OBSD resultmap changes needed. " } ]
[ { "msg_contents": "Hi,\n\n\tI know I'm a bit early to submit stuff for 7.2, but attached is a patch\n(agains 7.1RC1) to add uint2 and uint4 as new types. I haven't put in\nall the possible combinations of signed and unsigned ints as arguments\nto operators -- I was going insane just doing it for all combinations of\nuint2 and uint4. If anything is missing, please let me know and I'd\nappreciate it if somebody could do some sanity checking on the patch, as\nthis is the first time I've actually inserted a new type into the\ncatalog.\n\nAlso, some other databases (Compaq Himalaya) use\n\n\tSMALLINT UNSIGNED\n\tINTEGER UNSIGNED\n\nfor these types. Dunno whether a type consisting of 2 words is going to\nbreak the parser completely. If there are any problems with the patch,\nplease let me know as well -- not sure I've built it in the correct way.\n\nRegards,\n\nAdriaan", "msg_date": "Mon, 26 Mar 2001 22:30:32 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Unsigned ints" }, { "msg_contents": "Uggh, this needs some help. I've got the problem that I can insert a\nvalue bigger than MAXINT into an int8 column, but when I try to do the\nsame for a uint4 column, the parser coerces it into an int4, as in:\n\ntest=# insert into tint8 values (3043140617);\nINSERT 30254 1\ntest=# insert into tuint4 values (3043140617);\nERROR: Attribute 'a' is of type 'uint4' but expression is of type\n'float8'\n You will need to rewrite or cast the expression\ntest=# insert into tuint4 values (3043140617::uint4);\nINSERT 30255 1\n\n\nApparently this happens in parse_target.c, from where it calls\nCoerceTargetExpr from where it calls routines in parse_coerce.c.\n\nAt this point I decided that somewhere in the definition of the type\nthere must be a way of specifying how values can be transformed. Can\nanybody explain to me what I need to change to make this work? Without\nthis ecpg cannot work with unsigned ints, so explicit casting is not an\noption.\n\nCheers!\n\nAdriaan\n", "msg_date": "Wed, 28 Mar 2001 18:41:18 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Re: Unsigned ints (Help)" }, { "msg_contents": "> At this point I decided that somewhere in the definition of the type\n> there must be a way of specifying how values can be transformed. Can\n> anybody explain to me what I need to change to make this work? Without\n> this ecpg cannot work with unsigned ints, so explicit casting is not an\n> option.\n\nThe large integer-like value is silently transformed into a float8 by\nthe scanner (very early in the parsing stage). You have not provided a\nfunction to transform float8 into uint4, which if you do so will fix\nyour problem. Do a\n\n create function uint4(float8)...\n\nand the type coersion code will understand how to convert one into the\nother.\n\n - Thomas\n", "msg_date": "Wed, 28 Mar 2001 15:53:08 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": false, "msg_subject": "Re: Unsigned ints (Help)" }, { "msg_contents": "Adriaan Joubert <a.joubert@albourne.com> writes:\n> Uggh, this needs some help. I've got the problem that I can insert a\n> value bigger than MAXINT into an int8 column, but when I try to do the\n> same for a uint4 column, the parser coerces it into an int4, as in:\n\nSee past discussions about appropriate handling of unlabeled numeric\nconstants. This is a tricky area that needs a thoroughgoing rethink.\nIf you go into it with only \"fix uint4\" in mind then you will almost\ncertainly make things worse.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 11:09:28 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Unsigned ints (Help) " }, { "msg_contents": "Tom Lane wrote:\n> See past discussions about appropriate handling of unlabeled numeric\n> constants. This is a tricky area that needs a thoroughgoing rethink.\n> If you go into it with only \"fix uint4\" in mind then you will almost\n> certainly make things worse.\n\nGosh, you were certainly right there! Now inserting 65535 works, but\ninserting 65536 gives me 0 unless I cast to uint4!\n\nSearching the mailing list is not working for the individual mailing\nlists (I get Error: File Not Found - Did you enter the correct domain\nname, or URL? from ReadySetNet), and searching through all the files at\nuni-erlangen.de turned up nothing. \n\nCan somebody tell me when that discussion took place? Or tell me where\nto look in the code?\n\nThanks!\n\nAdriaan\n", "msg_date": "Wed, 28 Mar 2001 22:27:55 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Re: Unsigned ints (Help)" } ]
[ { "msg_contents": "Hi there.\n\nI was just looking for the CVS tags for downloading the beta6 and the\nRC1 of 7.1 but there are only the following tags:\n\nREL_7_1_BETA2\nREL_7_1_BETA3\nREL_7_1\n\nAren't there tags for the versions I am looking for?\n\nRegards,\n Matthias\n\n", "msg_date": "Mon, 26 Mar 2001 23:39:16 +0200 (METDST)", "msg_from": "Matthias Juchem <matthias@euklid.math.uni-mannheim.de>", "msg_from_op": true, "msg_subject": "CVS tags for betas and release candidate" }, { "msg_contents": "On Mon, 26 Mar 2001, Matthias Juchem wrote:\n\n> Hi there.\n>\n> I was just looking for the CVS tags for downloading the beta6 and the\n> RC1 of 7.1 but there are only the following tags:\n>\n> REL_7_1_BETA2\n> REL_7_1_BETA3\n> REL_7_1\n>\n> Aren't there tags for the versions I am looking for?\n\nNope ... doing the tags didn't work as well as was hoped, so we've just\nbeen using date ranges instead ... release itself will be tag'd ...\n\n\n", "msg_date": "Tue, 27 Mar 2001 08:48:14 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: CVS tags for betas and release candidate" }, { "msg_contents": "* The Hermit Hacker <scrappy@hub.org> [010327 04:53] wrote:\n> On Mon, 26 Mar 2001, Matthias Juchem wrote:\n> \n> > Hi there.\n> >\n> > I was just looking for the CVS tags for downloading the beta6 and the\n> > RC1 of 7.1 but there are only the following tags:\n> >\n> > REL_7_1_BETA2\n> > REL_7_1_BETA3\n> > REL_7_1\n> >\n> > Aren't there tags for the versions I am looking for?\n> \n> Nope ... doing the tags didn't work as well as was hoped, so we've just\n> been using date ranges instead ... release itself will be tag'd ...\n\nYou know you can nuke tags right?\n\n-- \n-Alfred Perlstein - [bright@wintelcom.net|alfred@freebsd.org]\nDaemon News Magazine in your snail-mail! http://magazine.daemonnews.org/\n", "msg_date": "Tue, 27 Mar 2001 05:19:17 -0800", "msg_from": "Alfred Perlstein <bright@wintelcom.net>", "msg_from_op": false, "msg_subject": "Re: CVS tags for betas and release candidate" }, { "msg_contents": "On Tue, 27 Mar 2001, Alfred Perlstein wrote:\n\n> * The Hermit Hacker <scrappy@hub.org> [010327 04:53] wrote:\n> > On Mon, 26 Mar 2001, Matthias Juchem wrote:\n> >\n> > > Hi there.\n> > >\n> > > I was just looking for the CVS tags for downloading the beta6 and the\n> > > RC1 of 7.1 but there are only the following tags:\n> > >\n> > > REL_7_1_BETA2\n> > > REL_7_1_BETA3\n> > > REL_7_1\n> > >\n> > > Aren't there tags for the versions I am looking for?\n> >\n> > Nope ... doing the tags didn't work as well as was hoped, so we've just\n> > been using date ranges instead ... release itself will be tag'd ...\n>\n> You know you can nuke tags right?\n\nno, this one I didn't know ... how? :)\n\n", "msg_date": "Tue, 27 Mar 2001 09:31:41 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: CVS tags for betas and release candidate" }, { "msg_contents": "* The Hermit Hacker <scrappy@hub.org> [010327 05:31] wrote:\n> On Tue, 27 Mar 2001, Alfred Perlstein wrote:\n> \n> > * The Hermit Hacker <scrappy@hub.org> [010327 04:53] wrote:\n> > > On Mon, 26 Mar 2001, Matthias Juchem wrote:\n> > >\n> > > > Hi there.\n> > > >\n> > > > I was just looking for the CVS tags for downloading the beta6 and the\n> > > > RC1 of 7.1 but there are only the following tags:\n> > > >\n> > > > REL_7_1_BETA2\n> > > > REL_7_1_BETA3\n> > > > REL_7_1\n> > > >\n> > > > Aren't there tags for the versions I am looking for?\n> > >\n> > > Nope ... doing the tags didn't work as well as was hoped, so we've just\n> > > been using date ranges instead ... release itself will be tag'd ...\n> >\n> > You know you can nuke tags right?\n> \n> no, this one I didn't know ... how? :)\n\n\n If you use `cvs tag -d symbolic_tag...', the sym-\n bolic tag you specify is deleted instead of being\n added. Warning: Be very certain of your ground\n before you delete a tag; doing this effectively\n discards some historical information, which may\n later turn out to have been valuable.\n\n\n\n-- \n-Alfred Perlstein - [bright@wintelcom.net|alfred@freebsd.org]\nInstead of asking why a piece of software is using \"1970s technology,\"\nstart asking why software is ignoring 30 years of accumulated wisdom.\n", "msg_date": "Tue, 27 Mar 2001 05:35:15 -0800", "msg_from": "Alfred Perlstein <bright@wintelcom.net>", "msg_from_op": false, "msg_subject": "Re: CVS tags for betas and release candidate" } ]
[ { "msg_contents": "Hi there,\n\nI just run 'make check' on RC1 and got the output below. I've read about\nthe negative zeros, but I haven't seen someone reporting a failure during\nthe random test, so I thought it would be good to report this one.\n\nIf you need additional information, please contact me.\n\nRegards,\n Matthias\n\n\n*** ./expected/geometry-positive-zeros.out Tue Sep 12 23:07:16 2000\n--- ./results/geometry.out Tue Mar 27 01:23:47 2001\n***************\n*** 114,120 ****\n | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n | (-5,-12) | [(1,2),(3,4)] | (1,2)\n | (10,10) | [(1,2),(3,4)] | (3,4)\n! | (0,0) | [(0,0),(6,6)] | (0,0)\n | (-10,0) | [(0,0),(6,6)] | (0,0)\n | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n--- 114,120 ----\n | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n | (-5,-12) | [(1,2),(3,4)] | (1,2)\n | (10,10) | [(1,2),(3,4)] | (3,4)\n! | (0,0) | [(0,0),(6,6)] | (-0,0)\n | (-10,0) | [(0,0),(6,6)] | (0,0)\n | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n***************\n*** 224,233 ****\n twentyfour | rotation\n ------------+-----------------------------\n | (0,0),(0,0)\n! | (0,0),(-20,-20)\n! | (0,2),(-14,0)\n | (0,79.2),(-58.8,0)\n! | (14,0),(0,-34)\n | (0,40),(0,0)\n | (0,0),(0,0)\n | (-10,-10),(-30,-30)\n--- 224,233 ----\n twentyfour | rotation\n ------------+-----------------------------\n | (0,0),(0,0)\n! | (-0,0),(-20,-20)\n! | (-0,2),(-14,0)\n | (0,79.2),(-58.8,0)\n! | (14,-0),(0,-34)\n | (0,40),(0,0)\n | (0,0),(0,0)\n | (-10,-10),(-30,-30)\n***************\n*** 254,264 ****\n WHERE (p.f1 <-> point '(0,0)') >= 1;\n twenty | rotation\n\n--------+-----------------------------------------------------------------------------------\n! | (0,0),(-0.2,-0.2)\n | (-0.1,-0.1),(-0.3,-0.3)\n | (-0.25,-0.25),(-0.25,-0.35)\n | (-0.3,-0.3),(-0.3,-0.3)\n! | (0.08,0),(0,-0.56)\n | (0.12,-0.28),(0.04,-0.84)\n | (0.26,-0.7),(0.1,-0.82)\n | (0.12,-0.84),(0.12,-0.84)\n--- 254,264 ----\n WHERE (p.f1 <-> point '(0,0)') >= 1;\n twenty | rotation\n\n--------+-----------------------------------------------------------------------------------\n! | (0,-0),(-0.2,-0.2)\n | (-0.1,-0.1),(-0.3,-0.3)\n | (-0.25,-0.25),(-0.25,-0.35)\n | (-0.3,-0.3),(-0.3,-0.3)\n! | (0.08,-0),(0,-0.56)\n | (0.12,-0.28),(0.04,-0.84)\n | (0.26,-0.7),(0.1,-0.82)\n | (0.12,-0.84),(0.12,-0.84)\n***************\n*** 266,272 ****\n |\n(0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n |\n(0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n |\n(0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n! | (0,0.0828402366863905),(-0.201183431952663,0)\n |\n(-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n |\n(-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n |\n(-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n--- 266,272 ----\n |\n(0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n |\n(0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n |\n(0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n! | (-0,0.0828402366863905),(-0.201183431952663,0)\n |\n(-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n |\n(-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n |\n(-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n\n======================================================================\n\n*** ./expected/random.out Thu Jan 6 07:40:54 2000\n--- ./results/random.out Tue Mar 27 01:24:00 2001\n***************\n*** 31,35 ****\n WHERE random NOT BETWEEN 80 AND 120;\n random\n --------\n! (0 rows)\n\n--- 31,36 ----\n WHERE random NOT BETWEEN 80 AND 120;\n random\n --------\n! 139\n! (1 row)\n\n\n======================================================================\n\n\n", "msg_date": "Tue, 27 Mar 2001 01:36:51 +0200 (METDST)", "msg_from": "Matthias Juchem <matthias@euklid.math.uni-mannheim.de>", "msg_from_op": true, "msg_subject": "regression test result HP-UX 11.00, PA8500" } ]
[ { "msg_contents": "Hi,\n\nSemi-topical I hope ;) I've started using Postgres 7.1 (FreeBSD 4.2-S) and large objects via JDBC. (postmaster (PostgreSQL) 7.1beta5)\n\nEverything has been working nicely with storing/retrieving blobs, until last night during a vacuum of the database the backend process crashed with the messages added to the end of this email. I'm also using the 'vacuumlo' contributed code. The order of the cron jobs is:\n\n59 2 * * * postgres /usr/local/pgsql/bin/vacuumlo -v db1 db2 db3\n59 3 * * * postgres /usr/local/pgsql/bin/vacuumdb -z db1\n59 4 * * * postgres /usr/local/pgsql/bin/vacuumdb -z db2\n59 5 * * * postgres /usr/local/pgsql/bin/vacuumdb -z db3\n\nso I was wondering if there might be a bug in the vacuumlo code (though its vacuumdb dying)? Or I was thinking, because they're development db's, that frequent dropping/recreating of tables is maybe causing the prob? The same vacuum commands have run fine before, both from cron and the command line, the only difference was slightly heavier dropping/recreating yesterday.\n\nI'm yet to see if that particular database is stuffed as I can recreate and retest easily enough. Let me know if I can give any further info,\n\nRegards,\nJoe\n\n---\nNOTICE: Rel pg_attribute: TID 1/115: OID IS INVALID. TUPGONE 1.\n...\nNOTICE: Rel pg_attribute: TID 1/6087: OID IS INVALID. TUPGONE 1.\nNOTICE: Rel pg_attribute: TID 1/6111: OID IS INVALID. TUPGONE 1.\nNOTICE: Rel pg_attribute: TID 1/6112: OID IS INVALID. TUPGONE 1.\nNOTICE: Rel pg_attribute: TID 1/6136: OID IS INVALID. TUPGONE 1.\nNOTICE: Rel pg_attribute: TID 1/6137: OID IS INVALID. TUPGONE 1.\npqReadData() -- backend closed the channel unexpectedly.\n This probably means the backend terminated abnormally\n before or while processing the request.\nconnection to server was lost\nvacuumdb: vacuum db2 failed\n---\n\nwith ~500 of the NOTICE lines then the crash. About 1% give a TUPGONE 0 ending instead.\n\n", "msg_date": "Tue, 27 Mar 2001 10:37:35 +1000", "msg_from": "\"Joe Shevland\" <shevlandj@kpi.com.au>", "msg_from_op": true, "msg_subject": "Possible large object bug?" }, { "msg_contents": "\"Joe Shevland\" <shevlandj@kpi.com.au> writes:\n> Semi-topical I hope ;)\n\nCompletely irrelevant to JDBC as far as I can see. I've redirected to\npghackers.\n\n> NOTICE: Rel pg_attribute: TID 1/115: OID IS INVALID. TUPGONE 1.\n> ...\n> NOTICE: Rel pg_attribute: TID 1/6087: OID IS INVALID. TUPGONE 1.\n> NOTICE: Rel pg_attribute: TID 1/6111: OID IS INVALID. TUPGONE 1.\n> NOTICE: Rel pg_attribute: TID 1/6112: OID IS INVALID. TUPGONE 1.\n> NOTICE: Rel pg_attribute: TID 1/6136: OID IS INVALID. TUPGONE 1.\n> NOTICE: Rel pg_attribute: TID 1/6137: OID IS INVALID. TUPGONE 1.\n> pqReadData() -- backend closed the channel unexpectedly.\n> This probably means the backend terminated abnormally\n> before or while processing the request.\n\nUgh. It looks like something has clobbered your pg_attribute file.\nWas this the first sign of trouble? Can you provide an \"od -x\" dump of\nthat file? (It'd be $PGDATA/base/DB2OID/1249; you'll need to look in\npg_database to determine the OID of db2.)\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 26 Mar 2001 20:00:54 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [JDBC] Possible large object bug? " }, { "msg_contents": "At 10:37 27/03/01 +1000, Joe Shevland wrote:\n>Hi,\n>\n>Semi-topical I hope ;)\n\nYes semi ;-)\n\n> I've started using Postgres 7.1 (FreeBSD 4.2-S) and large objects via \n> JDBC. (postmaster (PostgreSQL) 7.1beta5)\n\nI'm forwarding this to the bugs list as it looks like something nasty in \nthe back end.\n\n\n>Everything has been working nicely with storing/retrieving blobs, until \n>last night during a vacuum of the database the backend process crashed \n>with the messages added to the end of this email. I'm also using the \n>'vacuumlo' contributed code. The order of the cron jobs is:\n>\n>59 2 * * * postgres /usr/local/pgsql/bin/vacuumlo -v db1 db2 db3\n>59 3 * * * postgres /usr/local/pgsql/bin/vacuumdb -z db1\n>59 4 * * * postgres /usr/local/pgsql/bin/vacuumdb -z db2\n>59 5 * * * postgres /usr/local/pgsql/bin/vacuumdb -z db3\n>\n>so I was wondering if there might be a bug in the vacuumlo code (though \n>its vacuumdb dying)? Or I was thinking, because they're development db's, \n>that frequent dropping/recreating of tables is maybe causing the prob? The \n>same vacuum commands have run fine before, both from cron and the command \n>line, the only difference was slightly heavier dropping/recreating yesterday.\n>\n>I'm yet to see if that particular database is stuffed as I can recreate \n>and retest easily enough. Let me know if I can give any further info,\n>\n>Regards,\n>Joe\n>\n>---\n>NOTICE: Rel pg_attribute: TID 1/115: OID IS INVALID. TUPGONE 1.\n>...\n>NOTICE: Rel pg_attribute: TID 1/6087: OID IS INVALID. TUPGONE 1.\n>NOTICE: Rel pg_attribute: TID 1/6111: OID IS INVALID. TUPGONE 1.\n>NOTICE: Rel pg_attribute: TID 1/6112: OID IS INVALID. TUPGONE 1.\n>NOTICE: Rel pg_attribute: TID 1/6136: OID IS INVALID. TUPGONE 1.\n>NOTICE: Rel pg_attribute: TID 1/6137: OID IS INVALID. TUPGONE 1.\n>pqReadData() -- backend closed the channel unexpectedly.\n> This probably means the backend terminated abnormally\n> before or while processing the request.\n>connection to server was lost\n>vacuumdb: vacuum db2 failed\n>---\n>\n>with ~500 of the NOTICE lines then the crash. About 1% give a TUPGONE 0 \n>ending instead.\n>\n>\n>---------------------------(end of broadcast)---------------------------\n>TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n\n", "msg_date": "Thu, 29 Mar 2001 09:52:05 +0100", "msg_from": "Peter Mount <peter@retep.org.uk>", "msg_from_op": false, "msg_subject": "Re: [JDBC] Possible large object bug?" }, { "msg_contents": "Hi,\n\nTom C. and I exchanged a few emails about the issue offlist; Tom spotted after a while that the PostgreSQL data had been clobbered by the output of my cron job (not possible in normal conditions I guess)... end result being he suspects its a hardware glitch or similar (also had 'wc' dump core that same night during the daily periodic run; I'm choosing cosmic radiation ;).\n\nI'm going to upgrade to the latest release and also run up another shinier box and see if things work out, I'm sure they will (digging out a tripwire floppy to make ultra sure nothings changed). As a side issue, could/should the vacuumlo functionality be merged with vacuum?\n\nCheers,\nJoe\n\n> -----Original Message-----\n> From: pgsql-jdbc-owner@postgresql.org\n> [mailto:pgsql-jdbc-owner@postgresql.org]On Behalf Of Peter Mount\n> Sent: Thursday, March 29, 2001 6:52 PM\n> To: Joe Shevland; pgsql-jdbc@postgresql.org\n> Cc: pgsql-bugs@postgresql.org\n> Subject: Re: [JDBC] Possible large object bug?\n> \n> \n> At 10:37 27/03/01 +1000, Joe Shevland wrote:\n> >Hi,\n> >\n> >Semi-topical I hope ;)\n> \n> Yes semi ;-)\n> \n> > I've started using Postgres 7.1 (FreeBSD 4.2-S) and large objects via \n> > JDBC. (postmaster (PostgreSQL) 7.1beta5)\n> \n> I'm forwarding this to the bugs list as it looks like something nasty in \n> the back end.\n> \n> \n> >Everything has been working nicely with storing/retrieving blobs, until \n> >last night during a vacuum of the database the backend process crashed \n> >with the messages added to the end of this email. I'm also using the \n> >'vacuumlo' contributed code. The order of the cron jobs is:\n> >\n> >59 2 * * * postgres /usr/local/pgsql/bin/vacuumlo -v db1 db2 db3\n> >59 3 * * * postgres /usr/local/pgsql/bin/vacuumdb -z db1\n> >59 4 * * * postgres /usr/local/pgsql/bin/vacuumdb -z db2\n> >59 5 * * * postgres /usr/local/pgsql/bin/vacuumdb -z db3\n> >\n> >so I was wondering if there might be a bug in the vacuumlo code (though \n> >its vacuumdb dying)? Or I was thinking, because they're \n> development db's, \n> >that frequent dropping/recreating of tables is maybe causing the \n> prob? The \n> >same vacuum commands have run fine before, both from cron and \n> the command \n> >line, the only difference was slightly heavier \n> dropping/recreating yesterday.\n> >\n> >I'm yet to see if that particular database is stuffed as I can recreate \n> >and retest easily enough. Let me know if I can give any further info,\n> >\n> >Regards,\n> >Joe\n> >\n> >---\n> >NOTICE: Rel pg_attribute: TID 1/115: OID IS INVALID. TUPGONE 1.\n> >...\n> >NOTICE: Rel pg_attribute: TID 1/6087: OID IS INVALID. TUPGONE 1.\n> >NOTICE: Rel pg_attribute: TID 1/6111: OID IS INVALID. TUPGONE 1.\n> >NOTICE: Rel pg_attribute: TID 1/6112: OID IS INVALID. TUPGONE 1.\n> >NOTICE: Rel pg_attribute: TID 1/6136: OID IS INVALID. TUPGONE 1.\n> >NOTICE: Rel pg_attribute: TID 1/6137: OID IS INVALID. TUPGONE 1.\n> >pqReadData() -- backend closed the channel unexpectedly.\n> > This probably means the backend terminated abnormally\n> > before or while processing the request.\n> >connection to server was lost\n> >vacuumdb: vacuum db2 failed\n> >---\n> >\n> >with ~500 of the NOTICE lines then the crash. About 1% give a TUPGONE 0 \n> >ending instead.\n> >\n> >\n> >---------------------------(end of broadcast)---------------------------\n> >TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n> \n\n", "msg_date": "Fri, 30 Mar 2001 08:05:48 +1000", "msg_from": "\"Joe Shevland\" <shevlandj@kpi.com.au>", "msg_from_op": true, "msg_subject": "RE: [JDBC] Possible large object bug?" } ]
[ { "msg_contents": "\n could anyone compress Postgresql package in bzip2 format?\nI am behind a slow line, to save me some bandwith, I would like\nto download a package named postgresql-7.1RC1.tar.bz2 .\n\nRegards,\nXuYifeng\n\n\n", "msg_date": "Tue, 27 Mar 2001 12:52:41 +0800", "msg_from": "Xu Yifeng <jamexu@telekbird.com.cn>", "msg_from_op": true, "msg_subject": "compress package in bzip2" } ]
[ { "msg_contents": "Any change of getting a 7.1 RC1 RPM? I'm using the beta4 RPMs at the moment\nbut don't seem to be any more recent ones.\n\nIt would seem dangerous to me to produce a 7.1 RPM without testing the RPM\nbuild process?\n\n-mike\n\n\n", "msg_date": "Tue, 27 Mar 2001 18:31:24 +1000", "msg_from": "\"Mike Cannon-Brookes\" <mcannon@internet.com>", "msg_from_op": true, "msg_subject": "7.1 RC1 RPM" }, { "msg_contents": "Mike Cannon-Brookes wrote:\n> \n> Any change of getting a 7.1 RC1 RPM? I'm using the beta4 RPMs at the moment\n> but don't seem to be any more recent ones.\n\nI'm building a quickie RC1-1 RPM right now. There are some other things\nI need to do on the RPMset before final release -- and I plan on working\na while this Saturday on that -- but my week is so loaded that I'm going\nto put out a rebuild of 7.1beta6->7.1RC1 as is -- once I get it to\nbuild.....\n\nStay tuned...\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Tue, 27 Mar 2001 12:13:10 -0500", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: 7.1 RC1 RPM" }, { "msg_contents": "Lamar Owen wrote:\n> \n> Mike Cannon-Brookes wrote:\n> >\n> > Any change of getting a 7.1 RC1 RPM? I'm using the beta4 RPMs at the moment\n> > but don't seem to be any more recent ones.\n> \n> I'm building a quickie RC1-1 RPM right now. There are some other things\n> I need to do on the RPMset before final release -- and I plan on working\n> a while this Saturday on that -- but my week is so loaded that I'm going\n> to put out a rebuild of 7.1beta6->7.1RC1 as is -- once I get it to\n> build.....\n> \n> Stay tuned...\n\nOk, 7.1RC1 RPM uploaded to ftp://ftp.postgresql.org/pub/dev/test-rpms as\nbefore.\n\nBIG NOTE: These are built on Red Hat _7.0_ NOT 6.2 as previous ones\nhave been. The ODBC client build on 6.2 has been broken -- it built at\nbeta4, but now gives this error set at beta6/RC1:\n\nar crs libpsqlodbc.a info.o bind.o columninfo.o connection.o convert.o\ndrvconn.o environ.o execute.o lobj.o misc.o options.o pgtypes.o\npsqlodbc.o qresult.o results.o socket.o parse.o statement.o gpps.o\ntuple.o tuplelist.o dlg_specific.o\nranlib libpsqlodbc.a\ngcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -O2 -m486\n-fno-strength-reduce -fpic -shared -Wl,-soname,libpsqlodbc.so.0\n-Wl,-Bsymbolic info.o bind.o columninfo.o connection.o convert.o\ndrvconn.o environ.o execute.o lobj.o misc.o options.o pgtypes.o\npsqlodbc.o qresult.o results.o socket.o parse.o statement.o gpps.o\ntuple.o tuplelist.o dlg_specific.o -lm -Wl,-rpath,/usr/lib -o\nlibpsqlodbc.so.0.26\nconnection.o: In function `CC_connect':\nconnection.o(.text+0x914): undefined reference to\n`check_client_encoding'\nconnection.o(.text+0x91a): undefined reference to\n`check_client_encoding'\nconvert.o: In function `copy_statement_with_parameters':\nconvert.o(.text+0x110f): undefined reference to `multibyte_init'\nconvert.o(.text+0x1191): undefined reference to `multibyte_char_check'\nconvert.o(.text+0x11c3): undefined reference to `multibyte_strchr'\nconvert.o: In function `convert_special_chars':\nconvert.o(.text+0x2949): undefined reference to `multibyte_init'\nconvert.o(.text+0x29c4): undefined reference to `multibyte_char_check'\nparse.o: In function `getNextToken':\nparse.o(.text+0x151): undefined reference to `multibyte_char_check'\ndlg_specific.o: In function `getDSNinfo':\ndlg_specific.o(.text+0xd8c): undefined reference to\n`check_client_encoding'\ndlg_specific.o(.text+0xd98): undefined reference to\n`check_client_encoding'\ncollect2: ld returned 1 exit status\nmake[3]: *** [libpsqlodbc.so.0.26] Error 1\nmake[3]: Leaving directory\n`/root/rpm-building/redhat/BUILD/postgresql-7.1RC1/src/interfaces/odbc'\nmake[2]: *** [all] Error 2\nmake[2]: Leaving directory\n`/root/rpm-building/redhat/BUILD/postgresql-7.1RC1/src/interfaces'\nmake[1]: *** [all] Error 2\nmake[1]: Leaving directory\n`/root/rpm-building/redhat/BUILD/postgresql-7.1RC1/src'\nmake: *** [all] Error 2\n\n\nThere are no other errors in the ODBC build that I could find. Again --\nthis builds on Red Hat 7, but not RH 6.2.\n\nNow, according to the cvsweb on postgresql.org, Bruce has just as of 14\nhours ago done an update in this dir. What updates were done between\nbeta4 and beta6? (Is that not when the flap about patches occurred?). \nWell, Red Hat 7 will build it -- and Red Hat 6.2 will now NOT build it\n-- but it did build beta4 successfully.\n\nWell, in any case, preliminary 7.1RC1 RPMS are up. There are some odd\nissues with the packaging that I am working on. Be sure to read\nREADME.rpm-dist -- attached to this message for your convenience.\n\nTHIS IS A PRERELEASE RPM. Please use for testing ONLY.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Tue, 27 Mar 2001 13:35:16 -0500", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: 7.1 RC1 RPM" }, { "msg_contents": "Lamar Owen <lamar.owen@wgcr.org> writes:\n> BIG NOTE: These are built on Red Hat _7.0_ NOT 6.2 as previous ones\n> have been. The ODBC client build on 6.2 has been broken -- it built at\n> beta4, but now gives this error set at beta6/RC1:\n\n> ar crs libpsqlodbc.a info.o bind.o columninfo.o connection.o convert.o\n> drvconn.o environ.o execute.o lobj.o misc.o options.o pgtypes.o\n> psqlodbc.o qresult.o results.o socket.o parse.o statement.o gpps.o\n> tuple.o tuplelist.o dlg_specific.o\n> ranlib libpsqlodbc.a\n> gcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -O2 -m486\n> -fno-strength-reduce -fpic -shared -Wl,-soname,libpsqlodbc.so.0\n> -Wl,-Bsymbolic info.o bind.o columninfo.o connection.o convert.o\n> drvconn.o environ.o execute.o lobj.o misc.o options.o pgtypes.o\n> psqlodbc.o qresult.o results.o socket.o parse.o statement.o gpps.o\n> tuple.o tuplelist.o dlg_specific.o -lm -Wl,-rpath,/usr/lib -o\n> libpsqlodbc.so.0.26\n> connection.o: In function `CC_connect':\n> connection.o(.text+0x914): undefined reference to\n> `check_client_encoding'\n> connection.o(.text+0x91a): undefined reference to\n> `check_client_encoding'\n\nIt would appear you have a conflict about whether MULTIBYTE is defined\nor not --- the code thinks so, but the makefile does not, since\nmultibyte.o is not seen in the link command.\n\nThe identical technique is used in libpq's makefile, so I'm not sure\nwhy you do not see a link failure in libpq as well.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 13:55:16 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: 7.1 RC1 RPM " }, { "msg_contents": "Lamar Owen wrote:\n> Well, in any case, preliminary 7.1RC1 RPMS are up. There are some odd\n> issues with the packaging that I am working on. Be sure to read\n> README.rpm-dist -- attached to this message for your convenience.\n\nForgot to attach the file. :-(.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11", "msg_date": "Tue, 27 Mar 2001 14:06:30 -0500", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: 7.1 RC1 RPM" }, { "msg_contents": "Tom Lane wrote:\n> It would appear you have a conflict about whether MULTIBYTE is defined\n> or not --- the code thinks so, but the makefile does not, since\n> multibyte.o is not seen in the link command.\n \n> The identical technique is used in libpq's makefile, so I'm not sure\n> why you do not see a link failure in libpq as well.\n\nHmmmm. Hiroshi committed an update to GNUmakefile to 'enable multibyte\nsupport' for ODBC. But that was only 33 hours ago -- meaning it wasn't\nupdated in time for RC1. Lessee..... I'm rebuilding RC1 with Hiroshi's\nGNUmakefile change as part of the RPMset patch -- and it succeeds.\n\nBut it succeeds on RH7 _without_ Hiroshi's patch. Odd. Lessee....\nExamining my logs of the build on RH7 shows that the same error occurs\n-- it just doesn't abort the build. Argh.\n\nThis means that the binary up on the ftp site right now for ODBC is\nbroken. I'll fix it tonight or tomorrow -- and the source RPM won't\nrebuild on RH 6.2. I'll upload a -2 set tonight or tomorrow to fix\nthat, and a few other issues I found while dinking with it today.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Tue, 27 Mar 2001 14:35:12 -0500", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: 7.1 RC1 RPM" }, { "msg_contents": "FYI: I successfully built a beta4 RPM for Mandrake, having to make only\nthe following changes:\n\n1) add -fno-fast-math to the CFLAGS set from RPM_OPT_FLAGS. This fixes\nthe date/time rounding problems. Would seem to be a non-invasive fix to\nput this in for every build, or to flag it as a Mandrake-only feature\n(but I have not gotten a suggestion on how to do that).\n\n2) explicitly remove Pg.bs from the list of files for the perl\ninstallation. On my system (with perl-5.6.0) that is a zero-length file\nwhich seems to disappear (or never gets copied) during the installation\nprocess. Not sure why it does not propagate, but it does not seem to be\nessential.\n\nComments?\n\n - Thomas\n", "msg_date": "Tue, 27 Mar 2001 21:38:09 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": false, "msg_subject": "Re: 7.1 RC1 RPM" }, { "msg_contents": "Thomas Lockhart wrote:\n> FYI: I successfully built a beta4 RPM for Mandrake, having to make only\n> the following changes:\n\n> 1) add -fno-fast-math to the CFLAGS set from RPM_OPT_FLAGS. This fixes\n> the date/time rounding problems. Would seem to be a non-invasive fix to\n> put this in for every build, or to flag it as a Mandrake-only feature\n> (but I have not gotten a suggestion on how to do that).\n\nI've got the code Tom and I came up with to deal with the LinuxPPC\nnonsense of a few releases ago... That or similar would be the ticket. \nA sed pipeline re-setting CFLAGS by removing 'unsafe' options from the\nflags would be useful.\n\nI'm working on a 'what distribution is this' deal for the spec file.\n \n> 2) explicitly remove Pg.bs from the list of files for the perl\n> installation. On my system (with perl-5.6.0) that is a zero-length file\n> which seems to disappear (or never gets copied) during the installation\n> process. Not sure why it does not propagate, but it does not seem to be\n> essential.\n\nInteresting. I have found a couple of other non-props -- including the\nREADME.rpm-dist, for some reason. I'm going to have to trace the build\nin detail. And the html docs tree is going to the wrong place... \n\nIn any case, a unified or context diff against the 7.1beta4 spec would\nbe useful.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Tue, 27 Mar 2001 21:53:57 -0500", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: Re: 7.1 RC1 RPM" }, { "msg_contents": "Lamar Owen wrote:\n> \n> Tom Lane wrote:\n> > It would appear you have a conflict about whether MULTIBYTE is defined\n> > or not --- the code thinks so, but the makefile does not, since\n> > multibyte.o is not seen in the link command.\n> \n> > The identical technique is used in libpq's makefile, so I'm not sure\n> > why you do not see a link failure in libpq as well.\n> \n> Hmmmm. Hiroshi committed an update to GNUmakefile to 'enable multibyte\n> support' for ODBC. But that was only 33 hours ago -- meaning it wasn't\n> updated in time for RC1. Lessee..... I'm rebuilding RC1 with Hiroshi's\n> GNUmakefile change as part of the RPMset patch -- and it succeeds.\n> \n\nOops I apologize for my mistake. I must have changed the\nGNUmakefile when I committed multibyte support for ODBC.\n\nregards,\nHiroshi Inoue\n", "msg_date": "Wed, 28 Mar 2001 14:53:10 +0900", "msg_from": "Hiroshi Inoue <Inoue@tpf.co.jp>", "msg_from_op": false, "msg_subject": "Re: 7.1 RC1 RPM" }, { "msg_contents": "> In any case, a unified or context diff against the 7.1beta4 spec would\n> be useful.\n\nOK, here is a context diff of the spec file (note only two lines\nchanged). This addresses the \"-fno-fast-math\" problem on Mandrake (at\nleast 7.2 and earlier), and the \"disappearing Pg.bs file\" problem on\nsame.\n\n - Thomas", "msg_date": "Wed, 28 Mar 2001 14:17:04 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": false, "msg_subject": "Re: 7.1 RC1 RPM" }, { "msg_contents": "Hiroshi Inoue wrote:\n> Lamar Owen wrote:\n> > Hmmmm. Hiroshi committed an update to GNUmakefile to 'enable multibyte\n> > support' for ODBC. But that was only 33 hours ago -- meaning it wasn't\n> > updated in time for RC1. Lessee..... I'm rebuilding RC1 with Hiroshi's\n> > GNUmakefile change as part of the RPMset patch -- and it succeeds.\n \n> Oops I apologize for my mistake. I must have changed the\n> GNUmakefile when I committed multibyte support for ODBC.\n\nI'm afraid I may have been misunderstood. The change you made _fixed_\nthe problem -- it didn't cause it. You made no error.\n--\nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Wed, 28 Mar 2001 10:02:41 -0500", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: 7.1 RC1 RPM" } ]
[ { "msg_contents": "Hi All...\n\nI've got a slight problem with pg_dump in Postgres v7.0.3, in basically \nduplicates all the data that it extracts\n\nI do a \ntestdatabase>CREATE TABLE bob (number int4,description text);\ntestdatabase>INSERT INTO TABLE bob VALUES (4453,'This is just a test of \npg_dump');\n\nthen \n\nkowalski@dagoba > pg_dump -t bob testdatabase \n\\connect - kowalski\nCREATE TABLE \"bob\" (\n \"number\" int4,\n \"description\" text\n);\nCREATE TABLE \"bob\" (\n \"number\" int4,\n \"description\" text\n);\nCOPY \"bob\" FROM stdin;\n4453 This is just a test of pg_dump\n\\.\nCOPY \"bob\" FROM stdin;\n4453 This is just a test of pg_dump\n\\.\n\nAs you can see the records are duplicated. I discovered this when I tried \nto migrate from 7.0.3 to 7.1 and found performance suddenly took a terrible \ndive. Is there a patch for pg_dump ??\n\nThanks for any help\nMarCin\n\n", "msg_date": "Tue, 27 Mar 2001 11:03:07 +0200", "msg_from": "Marcin Kowalski <kowalski@datrix.co.za>", "msg_from_op": true, "msg_subject": "pg_dump potential bug" }, { "msg_contents": "Hi\n\nI would have expected this problem to have been brought up long ago with \npeople doing database backups and restores. Anyway below are the details\n\nI am running PostgreSQL v 7.0.3 (upgraded yesterday from 7.0.2) on SuSE 6.3 \nwith 2.2.17(SMP) kernel with S/W raid patch. \nThe actual database is running on a 40 GIG Software RAID0 Ext2fs partition. \nIt has approximately 13million records in 9 tables with most of the data \nresiding in 2 tables(+- 6Mill Each). The machine is a Dual PII-350 with 256 \nmeg of Ram. Each table has two indices, both on two fields.\n\nIs it possible that one of the system tables has been corrupted and shows \nmultiple entries for the tables?? This makes me a bit worried about system \nintegrity. When I do a simple select * from tablename it works fine, what \ndoes pg_dump do that I don't ???\n\nThanks in ADvance\nMarCin\n\n\n> Marcin Kowalski <kowalski@datrix.co.za> writes:\n> > kowalski@dagoba > pg_dump -t bob testdatabase\n> > \\connect - kowalski\n> > CREATE TABLE \"bob\" (\n> > \"number\" int4,\n> > \"description\" text\n> > );\n> > CREATE TABLE \"bob\" (\n> > \"number\" int4,\n> > \"description\" text\n> > );\n> > COPY \"bob\" FROM stdin;\n> > 4453 This is just a test of pg_dump\n> > \\.\n> > COPY \"bob\" FROM stdin;\n> > 4453 This is just a test of pg_dump\n> > \\.\n>\n> Strange. I can't duplicate this (and neither can anyone else, or we'd\n> have heard about it long since). What platform are you on? How did you\n> build or obtain your executables?\n>\n> regards, tom lane\n\n\n-- \n-----------------------------\n Marcin Kowalski\n Linux/Perl Developer\n Datrix Solutions\n Cel. 082-400-7603\n ***Open Source Kicks Ass***\n-----------------------------\n", "msg_date": "Wed, 28 Mar 2001 09:37:21 +0200", "msg_from": "Marcin Kowalski <kowalski@datrix.co.za>", "msg_from_op": true, "msg_subject": "Re: pg_dump potential bug" }, { "msg_contents": "Marcin Kowalski <kowalski@datrix.co.za> writes:\n> Is it possible that one of the system tables has been corrupted and shows \n> multiple entries for the tables?\n\nCome to think of it, this is a fairly likely behavior if you have\nmultiple entries in pg_shadow with the same usesysid.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 10:22:56 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: pg_dump potential bug " }, { "msg_contents": "Hi\n\nThanks for the help, in fact that is Exactly what is wrong. The pg_shadow \ntable has duplicated entries in it, I think I'm going to create a unique \nindex on it.\nBTW I've migrated the entrie database to PG7.1RC1, running quite a big \nsearch on the database basically involving a huge amount of selects. \nCurrently I'm curising at 1250 selects per second (simple select, no Joins) \nfrom multiple tables with mutliple data, pretty fast I think. (But I still \nhave +- 10 Million to do :-) ).\n\nANyone have any ideas on how to improve performace, currently have indices \non key fields and am clustering (vacuum + vacuum analyze done)?? Are there \nany command line parameters I can try to increase performance..??\n\nThanks in ADvance\nMarCIn\n\nTom Lane wrote:\n\n> Marcin Kowalski <kowalski@datrix.co.za> writes:\n>> Is it possible that one of the system tables has been corrupted and shows\n>> multiple entries for the tables?\n> \n> Come to think of it, this is a fairly likely behavior if you have\n> multiple entries in pg_shadow with the same usesysid.\n> \n> regards, tom lane\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://www.postgresql.org/search.mpl\n> \n\n", "msg_date": "Thu, 29 Mar 2001 10:13:58 +0200", "msg_from": "Marcin Kowalski <kowalski@datrix.co.za>", "msg_from_op": true, "msg_subject": "Re: pg_dump potential bug" }, { "msg_contents": "Hi \n\nRegarding my previous post, I just successfully created a unique index on \npg_shadow. DON'T DO THIS!!!\n-------\nCREATE UNIQUE INDEX shadow_index ON pg_shadow (usename)\n-------\nI couldn't create at pg_shadow_index as the pg prefix is reserved for \nsystem tables. \n\nThis BROKE the database. At least I can't connect anymore with a:\n-------\ntemplate1=# \\c statements\nFATAL 1: Index 'pg_shadow_name_index' does not exist\nPrevious connection kept\ntemplate1=#\n-------\nIf I look at the error log I get :\n-------\nERROR: Illegal class name 'pg_shadow_index'\n The 'pg_' name prefix is reserved for system catalogs\nERROR: Index 'pg_shadow_name_index' does not exist\nERROR: SearchSysCache: recursive use of cache 23\nERROR: SearchSysCache: recursive use of cache 23\nERROR: SearchSysCache: recursive use of cache 23\nERROR: SearchSysCache: recursive use of cache 23 <-- quite psql here\nFATAL 1: Index 'pg_shadow_name_index' does not exist <-- restarted again\nFATAL 1: Index 'pg_shadow_name_index' does not exist\nFATAL 1: Index 'pg_shadow_name_index' does not exist\n-------\n\nWhat can I do??? I've got a non-trivial amount of data that I cannot afford \nto lose!! HELP!..\n\nRegards\nMArCin - Thanks\n\n", "msg_date": "Thu, 29 Mar 2001 10:35:06 +0200", "msg_from": "Marcin Kowalski <kowalski@datrix.co.za>", "msg_from_op": true, "msg_subject": "Re: pg_dump potential bug -UNIQUE INDEX on PG_SHADOW Dont!! HELP" }, { "msg_contents": "\nI can confirm with current sources:\n\n\ttest=> CREATE UNIQUE INDEX shadow_index ON pg_shadow (usename);\n\tCREATE\n\ttest=> select * from pg_shadow;\n\tERROR: Index 'pg_shadow_sysid_index' does not exist\n\ttest=> \\q\n\t$ psql test\n\tpsql: FATAL 1: Index 'pg_shadow_name_index' does not exist\n\t$\n\ngdb shows that the check in heap_create() is working because the\nindex name does not begin with pg_, just the base table:\n\n\tBreakpoint 1, heap_create (relname=0x838d1d0 \"shadow_index\", \n\t tupDesc=0x83915e4, istemp=0 '\\000', storage_create=0 '\\000', \n\t allow_system_table_mods=0) at heap.c:183\n\t183 bool nailme = false;\n\nFirst, should we allow user-specified indexes on system tables, and if\nso, why does this error happen?\n\nNotice the user wanted an index named shadow_index, but the error\nmentioned is pg_shadow_name_index.\n\n\n> Hi \n> \n> Regarding my previous post, I just successfully created a unique index on \n> pg_shadow. DON'T DO THIS!!!\n> -------\n> CREATE UNIQUE INDEX shadow_index ON pg_shadow (usename)\n> -------\n> I couldn't create at pg_shadow_index as the pg prefix is reserved for \n> system tables. \n> \n> This BROKE the database. At least I can't connect anymore with a:\n> -------\n> template1=# \\c statements\n> FATAL 1: Index 'pg_shadow_name_index' does not exist\n> Previous connection kept\n> template1=#\n> -------\n> If I look at the error log I get :\n> -------\n> ERROR: Illegal class name 'pg_shadow_index'\n> The 'pg_' name prefix is reserved for system catalogs\n> ERROR: Index 'pg_shadow_name_index' does not exist\n> ERROR: SearchSysCache: recursive use of cache 23\n> ERROR: SearchSysCache: recursive use of cache 23\n> ERROR: SearchSysCache: recursive use of cache 23\n> ERROR: SearchSysCache: recursive use of cache 23 <-- quite psql here\n> FATAL 1: Index 'pg_shadow_name_index' does not exist <-- restarted again\n> FATAL 1: Index 'pg_shadow_name_index' does not exist\n> FATAL 1: Index 'pg_shadow_name_index' does not exist\n> -------\n> \n> What can I do??? I've got a non-trivial amount of data that I cannot afford \n> to lose!! HELP!..\n> \n> Regards\n> MArCin - Thanks\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 3: if posting/reading through Usenet, please send an appropriate\n> subscribe-nomail command to majordomo@postgresql.org so that your\n> message can get through to the mailing list cleanly\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 30 Mar 2001 10:52:52 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [SQL] Re: pg_dump potential bug -UNIQUE INDEX on PG_SHADOW Dont!!\n\tHELP" }, { "msg_contents": "> Hi \n> \n> Regarding my previous post, I just successfully created a unique index on \n> pg_shadow. DON'T DO THIS!!!\n> -------\n> CREATE UNIQUE INDEX shadow_index ON pg_shadow (usename)\n> -------\n> I couldn't create at pg_shadow_index as the pg prefix is reserved for \n> system tables. \n> \n> This BROKE the database. At least I can't connect anymore with a:\n> -------\n> template1=# \\c statements\n> FATAL 1: Index 'pg_shadow_name_index' does not exist\n> Previous connection kept\n> template1=#\n> -------\n> If I look at the error log I get :\n> -------\n> ERROR: Illegal class name 'pg_shadow_index'\n> The 'pg_' name prefix is reserved for system catalogs\n> ERROR: Index 'pg_shadow_name_index' does not exist\n> ERROR: SearchSysCache: recursive use of cache 23\n> ERROR: SearchSysCache: recursive use of cache 23\n> ERROR: SearchSysCache: recursive use of cache 23\n> ERROR: SearchSysCache: recursive use of cache 23 <-- quite psql here\n> FATAL 1: Index 'pg_shadow_name_index' does not exist <-- restarted again\n> FATAL 1: Index 'pg_shadow_name_index' does not exist\n> FATAL 1: Index 'pg_shadow_name_index' does not exist\n> -------\n> \n> What can I do??? I've got a non-trivial amount of data that I cannot afford \n> to lose!! HELP!..\n\nFirst, here is a patch which will prevent this from happening in the\nfuture. Do people want this held for 7.2 or applied now? It disables\nthe creation of user indexes on system tables.\n\nThe user-defined indexes on system columns can not be made to work\neasily. Tom Lane pointed out to me in a phone call that code like:\n\n CatalogIndexInsert(irelations, Num_pg_class_indices, relrelation, reltup);\n\nassumes it knows the number of indexes on each system table, and a\nuser-defined one would not be updated by any system catalog change that\ndid not go through the executor.\n\nAs far as recovery, I am not sure. One issue is that pg_shadow is a\nglobal table, not local to the database. My guess is that the global\ntable is still fine, but the index is in the database where you created\nthe index. You can't remove the file because pg_index thinks the index\nis proper and exists.\n\nI am kind of stumped.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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/catalog/index.c\n===================================================================\nRCS file: /home/projects/pgsql/cvsroot/pgsql/src/backend/catalog/index.c,v\nretrieving revision 1.144\ndiff -c -r1.144 index.c\n*** src/backend/catalog/index.c\t2001/03/22 06:16:10\t1.144\n--- src/backend/catalog/index.c\t2001/03/30 22:55:54\n***************\n*** 864,869 ****\n--- 864,876 ----\n \t\tindexInfo->ii_NumKeyAttrs < 1)\n \t\telog(ERROR, \"must index at least one attribute\");\n \n+ \tif (heapRelationName && !allow_system_table_mods &&\n+ \t\tIsSystemRelationName(heapRelationName) && IsNormalProcessingMode())\n+ \t{\n+ \t\telog(ERROR, \"You can not create indexes on system tables: '%s'\",\n+ \t\t\t heapRelationName);\n+ \t}\n+ \n \t/*\n \t * get heap relation oid and open the heap relation\n \t */", "msg_date": "Fri, 30 Mar 2001 20:14:50 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [SQL] Re: pg_dump potential bug -UNIQUE INDEX on PG_SHADOW Dont!!\n\tHELP" }, { "msg_contents": "Marcin Kowalski <kowalski@datrix.co.za> writes:\n> DON'T DO THIS!!!\n> -------\n> CREATE UNIQUE INDEX shadow_index ON pg_shadow (usename)\n> -------\n\nIndeed, trying to create nonstandard indexes on system catalogs is a BAD\nidea. There probably ought to be a check to prevent you from trying.\n\n> What can I do??? I've got a non-trivial amount of data that I cannot afford \n> to lose!! HELP!..\n\nI think you'd be OK if you could drop the index and then do\n\nupdate pg_class set relhasindex = 'f' where relname = 'pg_shadow';\n\nThe trick is to be able to do that when the database is busted.\nI think you may be able to do this if you restart in \"ignore system\nindexes\" mode (use \"-o -P\" while starting postmaster). Worth a try\nanyway.\n\n\t\t\tregards, tom lane\n", "msg_date": "Sat, 31 Mar 2001 18:57:50 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [SQL] Re: pg_dump potential bug -UNIQUE INDEX on PG_SHADOW Dont!!\n\tHELP" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> \ttest=> CREATE UNIQUE INDEX shadow_index ON pg_shadow (usename);\n> \tCREATE\n> \ttest=> select * from pg_shadow;\n> \tERROR: Index 'pg_shadow_sysid_index' does not exist\n> \ttest=> \\q\n> \t$ psql test\n> \tpsql: FATAL 1: Index 'pg_shadow_name_index' does not exist\n> \t$\n\n> Notice the user wanted an index named shadow_index, but the error\n> mentioned is pg_shadow_name_index.\n\nWhat's failing is catcache lookups on pg_shadow. The catcache has table\nentries that claim that there are indexes on pg_shadow(usename) and\npg_shadow(usesysid). The system would not work at all, except that\ncatcache's use of these indexes is defeated by sanity-check code that\nnotices that relhasindex is FALSE for pg_shadow (line 880 of\ncatcache.c).\n\nAs soon as you create an index on pg_shadow, relhasindex becomes TRUE\nand catcache.c starts trying to use these nonexistent indexes for\nroutine operations like ACL permissions checks. So, nothing works\nanymore.\n\nWe ought to create those indexes someday ;-)\n\n\t\t\tregards, tom lane\n", "msg_date": "Sat, 31 Mar 2001 19:14:59 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Re: [SQL] Re: pg_dump potential bug -UNIQUE INDEX on PG_SHADOW\n\tDont!! HELP" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> First, here is a patch which will prevent this from happening in the\n> future. Do people want this held for 7.2 or applied now? It disables\n> the creation of user indexes on system tables.\n \n> + \tif (heapRelationName && !allow_system_table_mods &&\n> + \t\tIsSystemRelationName(heapRelationName) && IsNormalProcessingMode())\n> + \t{\n> + \t\telog(ERROR, \"You can not create indexes on system tables: '%s'\",\n> + \t\t\t heapRelationName);\n> + \t}\n> + \n\nI think it would be a real good idea to put in this safeguard, but\nI don't much like that error message. How about\n\n elog(ERROR, \"User-defined indexes on system catalogs are not supported\");\n\n\n\t\t\tregards, tom lane\n", "msg_date": "Sat, 31 Mar 2001 19:53:39 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Re: [SQL] Re: pg_dump potential bug -UNIQUE INDEX on PG_SHADOW\n\tDont!! HELP" }, { "msg_contents": "> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > First, here is a patch which will prevent this from happening in the\n> > future. Do people want this held for 7.2 or applied now? It disables\n> > the creation of user indexes on system tables.\n> \n> > + \tif (heapRelationName && !allow_system_table_mods &&\n> > + \t\tIsSystemRelationName(heapRelationName) && IsNormalProcessingMode())\n> > + \t{\n> > + \t\telog(ERROR, \"You can not create indexes on system tables: '%s'\",\n> > + \t\t\t heapRelationName);\n> > + \t}\n> > + \n> \n> I think it would be a real good idea to put in this safeguard, but\n> I don't much like that error message. How about\n> \n> elog(ERROR, \"User-defined indexes on system catalogs are not supported\");\n\nChange made to patch.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 31 Mar 2001 19:58:17 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Re: [SQL] Re: pg_dump potential bug -UNIQUE INDEX on PG_SHADOW\n\tDont!! HELP" }, { "msg_contents": "> Bruce Momjian writes:\n> \n> > + elog(ERROR, \"You can not create indexes on system tables: %s'\",\n> > + heapRelationName);\n> \n> One of these days we should decide on a spelling of \"indexes\" vs\n> \"indices\".\n\nYes. Added to TODO:\n\n\t* Decide on spelling of indexes/indices\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n", "msg_date": "Mon, 2 Apr 2001 11:07:09 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Re: [SQL] Re: pg_dump potential bug -UNIQUE INDEX on PG_SHADOW\n\tDont!! HELP" }, { "msg_contents": "Bruce Momjian writes:\n\n> + elog(ERROR, \"You can not create indexes on system tables: %s'\",\n> + heapRelationName);\n\nOne of these days we should decide on a spelling of \"indexes\" vs\n\"indices\".\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Mon, 2 Apr 2001 17:15:36 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: Re: [SQL] Re: pg_dump potential bug -UNIQUE INDEX on\n\tPG_SHADOW Dont!! HELP" }, { "msg_contents": "[Dept of swatting flies with sledgehammers]\n\nPeter Eisentraut <peter_e@gmx.net> writes:\n> One of these days we should decide on a spelling of \"indexes\" vs\n> \"indices\".\n\nI'd vote for \"indexes\", first on the practical grounds that it's a more\nsensible spelling, and secondly on the grounds that the Oxford English\nDictionary agrees. Its entry for the noun index has:\n\n\nIndex: PL indexes and indices. In current use the plural is indices in\nsenses 8, 9, and usually in other senses except 5, in which indexes is\nusual.\n\n1. The fore-finger. (Now chiefly Anat.)\n\n2. A piece of wood, metal, or the like which serves as a pointer.\n\n3. The hand of a clock, watch, or sundial. (Now rare)\n\n4. That which serves to direct ... a guiding principle.\n\n5. (a) A table of contents, preface, or prologue (Obs). (b) An\nalphabetical list, placed (usually) at the end of a book, of the names,\nsubjects, etc. occurring in it, with indication of the places in which\nthey occur.\n\n6. Spec. (short for Index librorum prohibitorum) The list of books which\nRoman Catholics are forbidden to read.\n\n7. A \"hand\" marker in printing. (Obs)\n\n8. Math. (a) a number placed above and to the right of another quantity\nto denote a power or root. (b) the integral part of a logarithm. (Obs)\n\n9. In various sciences, a number or formula expressing some property of\nthe thing in question. (ex. Index of refraction)\n\n\n(I've abbreviated the definitions other than sense 5b.)\n\nI'd say that the use of \"index\" in database work clearly falls under\nsense 5b, and so \"indexes\" is the usual plural according to the OED.\n\nThe habit of using \"indices\" in the Postgres documentation seems to go\nback to the Berkeley days. Possibly the Berkeley boys were familiar\nwith sense 8 and/or 9 and so tended to use that plural.\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 02 Apr 2001 12:47:55 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Indexes vs indices" }, { "msg_contents": "> [Dept of swatting flies with sledgehammers]\n> \n> Peter Eisentraut <peter_e@gmx.net> writes:\n> > One of these days we should decide on a spelling of \"indexes\" vs\n> > \"indices\".\n> \n> I'd vote for \"indexes\", first on the practical grounds that it's a more\n> sensible spelling, and secondly on the grounds that the Oxford English\n> Dictionary agrees. Its entry for the noun index has:\n> \n\nI never liked indices. I like indexes.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n", "msg_date": "Mon, 2 Apr 2001 12:58:52 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Indexes vs indices" }, { "msg_contents": "Tom Lane wrote:\n> [Dept of swatting flies with sledgehammers]\n> I'd say that the use of \"index\" in database work clearly falls under\n> sense 5b, and so \"indexes\" is the usual plural according to the OED.\n\nAs a volume of the OED is about the weight of a small sledgehammer, that\nfly is one dead puppy (to mix my metaphors). Although, sense 4 is also\ngermane, as our index does serve to direct the query executor to the\nappropriate tuples, and could be considered to be the directing\nprinciple for performance enhancement <duck>......\n\nBut in reality, it doesn't matter. 'Indexes' is just fine. It's\ncertainly a better plural than 'Vaxen' was in its time; although I am\nstill inclined to use 'boxen' when referring to more than one computer.\n\n--\nLamar Owen\nWGCR Internet Radio\nProfessor of English, Anchor Baptist Bible College --so I'm allowed to\nplay with the language.... :-)\n1 Peter 4:11\n", "msg_date": "Mon, 02 Apr 2001 13:20:43 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: Re: Indexes vs indices" } ]
[ { "msg_contents": "I've fixed this bug in 7.1RC1. It may have been causing core dumps by\ndereferencing a null function pointer. As it was the Sun compiler\nflagged it as an error. (SC5.0 on Solaris 2.6.)\n\n\n\n\nInterestingly other compilers on other platforms accept a function as\na controlling expression in an if statement without complaint. That\nexpression should have scalar type, i.e. either arithmetic or pointer.\nThis test program demonstrates Sun catching the error.\n\n\n\n\n$ cc funcptr.c\n\"funcptr.c\", line 13: controlling expressions must have scalar type\ncc: acomp failed for funcptr.c\n\n-- \nPete Forman -./\\.- Disclaimer: This post is originated\nWesternGeco -./\\.- by myself and does not represent\npete.forman@westerngeco.com -./\\.- opinion of Schlumberger, Baker\nhttp://www.crosswinds.net/~petef -./\\.- Hughes or their divisions.", "msg_date": "Tue, 27 Mar 2001 10:20:07 +0100", "msg_from": "Pete Forman <pete.forman@westerngeco.com>", "msg_from_op": true, "msg_subject": "Fix for possible pg_dump core dump" }, { "msg_contents": "At 10:20 27/03/01 +0100, Pete Forman wrote:\n>I've fixed this bug in 7.1RC1. It may have been causing core dumps by\n>dereferencing a null function pointer. As it was the Sun compiler\n>flagged it as an error. (SC5.0 on Solaris 2.6.)\n>\n\nThanks for this, but Tom applied a similar patch a few days ago.\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Tue, 27 Mar 2001 20:33:55 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": false, "msg_subject": "Re: Fix for possible pg_dump core dump" }, { "msg_contents": "Philip Warner writes:\n > At 10:20 27/03/01 +0100, Pete Forman wrote:\n > >I've fixed this bug in 7.1RC1. It may have been causing core\n > >dumps by dereferencing a null function pointer. As it was the Sun\n > >compiler flagged it as an error. (SC5.0 on Solaris 2.6.)\n > >\n > \n > Thanks for this, but Tom applied a similar patch a few days ago.\n\nI can see Steve Nicolai's patch now. I had looked in the archives\nbefore posting but got 190 hits on pg_backup_null with no option to\nsort by date. Trying again on Google Groups this time gets me\nstraight there.\n-- \nPete Forman -./\\.- Disclaimer: This post is originated\nWesternGeco -./\\.- by myself and does not represent\npete.forman@westerngeco.com -./\\.- opinion of Schlumberger, Baker\nhttp://www.crosswinds.net/~petef -./\\.- Hughes or their divisions.\n", "msg_date": "Tue, 27 Mar 2001 11:42:54 +0100", "msg_from": "Pete Forman <pete.forman@westerngeco.com>", "msg_from_op": true, "msg_subject": "Re: Fix for possible pg_dump core dump" } ]
[ { "msg_contents": "The following patch updates the FAQ_AIX and makes C++ code work with\nmore recent versions of the IBM C++ compiler (now called VisualAge C++).\nThe C++ part was previously broken (g++ and xlC), thus this is zero risk. \nOnly AIX specific parts are touched (1 Makefile.shlib line (link with $(COMPILER) instead \nof $(CC) and one shell script line (parameter -C to nm to not demangle C++ symbols for \n.exp file)). \n\nI thus ask you to please apply this patch before release.\n\nWith or without this patch RC1 on AIX 4.3.2 RS6000 passes \"gmake check\" for both the native \ncompiler vac.C 5.0.1 and gcc 2.95.2 :-)\n\nAndreas", "msg_date": "Tue, 27 Mar 2001 11:50:33 +0200", "msg_from": "Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at>", "msg_from_op": true, "msg_subject": "FAQ_AIX and AIX C++ patches (was: Call for platforms)" }, { "msg_contents": "\nApplied because it is AIX-specific changes. Thanks.\n\n\n[ Charset ISO-8859-1 unsupported, converting... ]\n> \n> The following patch updates the FAQ_AIX and makes C++ code work with\n> more recent versions of the IBM C++ compiler (now called VisualAge C++).\n> The C++ part was previously broken (g++ and xlC), thus this is zero risk. \n> Only AIX specific parts are touched (1 Makefile.shlib line (link with $(COMPILER) instead \n> of $(CC) and one shell script line (parameter -C to nm to not demangle C++ symbols for \n> .exp file)). \n> \n> I thus ask you to please apply this patch before release.\n> \n> With or without this patch RC1 on AIX 4.3.2 RS6000 passes \"gmake check\" for both the native \n> compiler vac.C 5.0.1 and gcc 2.95.2 :-)\n> \n> Andreas\n> \n\n[ Attachment, skipping... ]\n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 27 Mar 2001 09:56:27 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] FAQ_AIX and AIX C++ patches (was: Call for platforms)" } ]
[ { "msg_contents": "Table `job' is inherited by `manufactured_job' and `purchased_job'. This\nquery works on either inherited table but not on the whole hierarchy:\n\nbray=# select * from manufactured_job AS j where not exists (select * from \nprice where price.product = j.product);\n product | qty | carriage | qc_sheet | barcodes | instructions | description \n---------+-----+----------+----------+----------+--------------+-------------\n(0 rows)\n\nbray=# select * from purchased_job AS j where not exists (select * from price \nwhere price.product = j.product);\n product | qty | carriage | qc_sheet | supplier | specification | del_point | \nimport_licence | import_duty | terms | deliv_clear\n---------+-----+----------+----------+----------+---------------+-----------+--\n--------------+-------------+-------+-------------\n(0 rows)\n\nbray=# select * from job AS j where not exists (select * from price where \nprice.product = j.product);\npqReadData() -- backend closed the channel unexpectedly.\n\tThis probably means the backend terminated abnormally\n\tbefore or while processing the request.\nThe connection to the server was lost. Attempting reset: Failed.\n!# \n\nThe log shows:\nQUERY STATISTICS\n! system usage stats:\n!\t0.160325 elapsed 0.100000 user 0.010000 system sec\n!\t[0.380000 user 0.040000 sys total]\n!\t0/0 [0/0] filesystem blocks in/out\n!\t71/2 [1109/429] page faults/reclaims, 0 [0] swaps\n!\t0 [0] signals rcvd, 0/0 [0/0] messages rcvd/sent\n!\t0/0 [0/0] voluntary/involuntary context switches\n! postgres usage stats:\n!\tShared blocks: 150 read, 0 written, buffer hit rate = 97.53%\n!\tLocal blocks: 0 read, 0 written, buffer hit rate = 0.00%\n!\tDirect blocks: 0 read, 0 written\nServer process (pid 15994) exited with status 139 at Tue Mar 27 10:47:49 2001\n\n(segmentation fault: 128 + 11 the coredump backtrace is below).\n\nbray=# select version();\n version \n------------------------------------------------------------------\n PostgreSQL 7.1beta5 on i686-pc-linux-gnu, compiled by GCC 2.95.3\n(1 row)\n\nbray=# select count(*) from job;\n count \n-------\n 4766\n(1 row)\n\nbray=# select count(*) from only job;\n count \n-------\n 0\n(1 row)\n\nbray=# select count(*) from manufactured_job;\n count \n-------\n 3279\n(1 row)\n\nbray=# select count(*) from purchased_job;\n count \n-------\n 1487\n(1 row)\n\n\n\nCREATE TABLE job\n(\n product VARCHAR(10) CONSTRAINT product\n\t REFERENCES stock (product)\n ON UPDATE CASCADE\n ON DELETE NO ACTION,\n qty INTEGER NOT NULL\n CHECK (qty > 0),\n carriage NUMERIC(12,2) NOT NULL\n DEFAULT 0,\n qc_sheet BOOLEAN NOT NULL\n DEFAULT 'f',\n PRIMARY KEY (product)\n)\n;\n\n\nCREATE TABLE manufactured_job\n(\n barcodes BOOLEAN NOT NULL\n DEFAULT 'f',\n instructions TEXT,\n description TEXT,\n PRIMARY KEY (product)\n)\n INHERITS (job)\n;\n\n\nCREATE TABLE purchased_job\n(\n supplier VARCHAR(10) NOT NULL\n\t\t\t\t CONSTRAINT supplier\n REFERENCES supplier (id)\n ON UPDATE CASCADE\n ON DELETE NO ACTION,\n specification TEXT,\n del_point CHAR(3) NOT NULL\n\t\t\t\tCONSTRAINT del_point\n REFERENCES location (id)\n ON UPDATE CASCADE\n ON DELETE NO ACTION,\n import_licence BOOLEAN DEFAULT 'f',\n import_duty NUMERIC(12,2),\n terms CHAR(3),\n deliv_clear NUMERIC(12,2),\n\n CONSTRAINT product_supplier\n FOREIGN KEY (product, supplier)\n REFERENCES product_supplier (product, supplier)\n MATCH FULL\n ON UPDATE CASCADE\n ON DELETE NO ACTION\n\t\t\t\t\t\tDEFERRABLE,\n PRIMARY KEY (product)\n)\n INHERITS (job)\n;\n\n(gdb) bt full\n#0 0x814786c in pfree (pointer=0x83054a8) at mcxt.c:451\n\tpointer = (void *) 0x83054a8\n#1 0x814766f in MemoryContextDelete (context=0x83054a8) at mcxt.c:191\n\tcontext = 0x83054a8\n#2 0x80c31b5 in FreeExprContext (econtext=0x8344bd8) at execUtils.c:218\n\tecontext = (ExprContext *) 0x8344bd8\n#3 0x80c65e9 in ExecEndIndexScan (node=0x8338f00) at nodeIndexscan.c:468\n\tscanstate = (CommonScanState *) 0x8344998\n\tindexstate = (IndexScanState *) 0x8344a78\n\truntimeKeyInfo = (int **) 0x8344b60\n\tscanKeys = (ScanKey *) 0x8344b18\n\tnumScanKeys = (int *) 0x8344b00\n\tnumIndices = 1\n\ti = 137644664\n#4 0x80c1589 in ExecEndNode (node=0x8338f00, parent=0x8338f00)\n at execProcnode.c:501\n\tnode = (Plan *) 0x8338f00\n\tsubp = (List *) 0x0\n#5 0x80ca121 in ExecEndSubPlan (node=0x833bb50) at nodeSubplan.c:452\n\tnode = (SubPlan *) 0x833bb50\n#6 0x80c151b in ExecEndNode (node=0x833bf50, parent=0x833bff0)\n at execProcnode.c:471\n\tnode = (Plan *) 0x833bf50\n\tsubp = (List *) 0x833c2d8\n#7 0x80c41db in ExecEndAppend (node=0x833bff0) at nodeAppend.c:367\n\tnode = (Append *) 0x833bff0\n\tappendstate = (AppendState *) 0x8305418\n\tnplans = 3\n\tappendplans = (List *) 0x833be68\n\tinitialized = (bool *) 0x833c660 \"\\001\\001\\001\"\n\ti = 2\n#8 0x80c156b in ExecEndNode (node=0x833bff0, parent=0x833c078)\n at execProcnode.c:489\n\tnode = (Plan *) 0x833bff0\n\tsubp = (List *) 0x0\n#9 0x80c8187 in ExecEndResult (node=0x833c078) at nodeResult.c:278\n\tnode = (Result *) 0x833c078\n\tresstate = (ResultState *) 0x833c5f0\n#10 0x80c155d in ExecEndNode (node=0x833c078, parent=0x833c078)\n at execProcnode.c:485\n\tnode = (Plan *) 0x833c078\n\tsubp = (List *) 0x0\n#11 0x80c015b in EndPlan (plan=0x833c078, estate=0x833c308) at execMain.c:851\n\tplan = (Plan *) 0x833c078\n\testate = (EState *) 0x833c308\n\tresultRelInfo = (ResultRelInfo *) 0x833c078\n\ti = 137609968\n\tl = (List *) 0x833c078\n#12 0x80bf9b1 in ExecutorEnd (queryDesc=0x833c2f0, estate=0x833c308)\n at execMain.c:265\n\tqueryDesc = (QueryDesc *) 0x8305418\n\testate = (EState *) 0x833c308\n#13 0x81070da in ProcessQuery (parsetree=0x832afc0, plan=0x833c078, \n dest=Remote) at pquery.c:314\n\tparsetree = (Query *) 0x833c308\n\tplan = (Plan *) 0x833c2f0\n\toperation = 1\n\ttag = 0x81850e0 \"SELECT\"\n\tisRetrieveIntoPortal = 0 '\\000'\n\tisRetrieveIntoRelation = 0 '\\000'\n\tintoName = 0x83453c0 \"\\004\"\n\tportal = 0x0\n\toldContext = 0x0\n\tqueryDesc = (QueryDesc *) 0x833c2f0\n\tstate = (EState *) 0x833c308\n\tattinfo = 0x83453c0\n#14 0x8105b01 in pg_exec_query_string (\n query_string=0x832a958 \"select * from job AS j where not exists (select * from price where price.product = j.product);\", dest=Remote, \n parse_context=0x8304c30) at postgres.c:810\n\tplan = (Plan *) 0x833c078\n\tquerytree = (Query *) 0x832afc0\n\tparsetree = (Node *) 0x832afc0\n\tisTransactionStmt = 0 '\\000'\n\tquerytree_list = (List *) 0x832afc0\n\tquerytree_item = (List *) 0x832c848\n\txact_started = 1 '\\001'\n\toldcontext = 0x833c078\n\tparsetree_list = (List *) 0x832afc0\n\tparsetree_item = (List *) 0x832af60\n#15 0x8106b64 in PostgresMain (argc=6, argv=0xbfffef80, real_argc=7, \n real_argv=0xbffff944, username=0x82fac21 \"olly\") at postgres.c:1883\n\tflag = 81\n\tDBName = 0x82fad88 \"bray\"\n\tsecure = 0 '\\000'\n\terrs = 0\n\tfirstchar = 81\n\tparser_input = 0x832a940\n\tremote_host = 0x832a940 \"X�2\\b^\"\n\tremote_port = 2071\n\tpotential_DataDir = 0x0\n...\n\n--==_Exmh_-19388723140\nContent-Type: text/plain; charset=iso-8859-1\n\nOliver Elphick Oliver.Elphick@lfix.co.uk\nIsle of Wight http://www.lfix.co.uk/oliver\nPGP: 1024R/32B8FAA1: 97 EA 1D 47 72 3F 28 47 6B 7E 39 CC 56 E4 C1 47\nGPG: 1024D/3E1D0C1C: CA12 09E0 E8D5 8870 5839 932A 614D 4C34 3E1D 0C1C\n ========================================\n \"Cast all your anxiety on him, because he cares for \n you.\" I Peter 5:7 \n\n\n", "msg_date": "Tue, 27 Mar 2001 11:25:17 +0100", "msg_from": "\"Oliver Elphick\" <olly@lfix.co.uk>", "msg_from_op": true, "msg_subject": "Backend crash (segfault) on query with inheritance hierarchy" }, { "msg_contents": "\"Oliver Elphick\" <olly@lfix.co.uk> writes:\n> Table `job' is inherited by `manufactured_job' and `purchased_job'. This\n> query works on either inherited table but not on the whole hierarchy:\n\nI've committed a fix to CVS.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 13:03:32 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Backend crash (segfault) on query with inheritance hierarchy " } ]
[ { "msg_contents": "�What is this file?\n�what can I do? One of this file is of 2.6 Bbytes\nthe DB is of 50 Mb\nIs urgent.\nThanks you.\n", "msg_date": "Tue, 27 Mar 2001 12:41:42 +0200", "msg_from": "Juan Manuel =?iso-8859-1?Q?Garc=EDa?= Arias <desarrollo@centrored.com>", "msg_from_op": true, "msg_subject": "Error: files pg_psort.*" } ]
[ { "msg_contents": "�What is this file?\n�what can I do? One of this file is of 2.6 Gb\nthe DB is of 50 Mb\nIs urgent.\nThanks you.\n", "msg_date": "Tue, 27 Mar 2001 12:49:39 +0200", "msg_from": "Juan Manuel =?iso-8859-1?Q?Garc=EDa?= Arias <desarrollo@centrored.com>", "msg_from_op": true, "msg_subject": "Error: file pg_psort*" }, { "msg_contents": "Juan Manuel =?iso-8859-1?Q?Garc=EDa?= Arias <desarrollo@centrored.com> writes:\n> �What is this file?\n> �what can I do? One of this file is of 2.6 Gb\n\nFiles named like that are just temporary files for sorting; you can\nremove them if the creating process isn't around anymore.\n\nBTW, consider updating to a more recent version of Postgres. We haven't\nnamed the temp files like that for quite a while.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 11:00:37 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Error: file pg_psort* " } ]
[ { "msg_contents": "On Mon, 26 Mar 2001, Justin Clift wrote:\n> Out of curiosity, how many times are you running the tests?\n> \n> I've been building 7.1RC1 over the weekend, and from one compiled\n> version I ran the regression tests 5 times before getting things to\n> pass. No changes anywhere, just re-ran the tests.\n> \n> So... it might be just co-incidence that the tests passed for you after\n> the change you mentioned below.\n\n> Alexander Klimov wrote:\n> > \n> > On Thu, 22 Mar 2001, Tom Lane wrote:\n> > > What I see is a lot of\n> > >\n> > > ! psql: Backend startup failed\n> > >\n> > > which suggests a fork() failure. Look in the postmaster logfile to see\n> > > the exact kernel error code --- but probably you are out of swap space\n> > > or up against the kernel's limit on number of processes for one userid.\n> > Strange, but this solution *also* works: I raise in /etc/system from 64 to\n> > set maxuprc=256\n> > revert pg_regress.sh in original state (with unix sockets for solaris),\n> > and now all tests are passed.\n\nYes, it was really just incidence -- I try again, and 15 of 15 `make\ncheck' passed with TCP sockets, but only 3 of 15 passed with UNIX\nsockets. So, final decision is `Unix sockets are not relaible on Solaris'\n\nRegards,\nASK\n\n\n", "msg_date": "Tue, 27 Mar 2001 14:10:28 +0200 (IST)", "msg_from": "Alexander Klimov <ask@wisdom.weizmann.ac.il>", "msg_from_op": true, "msg_subject": "Re: [BUGS] Tests randomly failed" }, { "msg_contents": "Alexander Klimov <ask@wisdom.weizmann.ac.il> writes:\n> Yes, it was really just incidence -- I try again, and 15 of 15 `make\n> check' passed with TCP sockets, but only 3 of 15 passed with UNIX\n> sockets. So, final decision is `Unix sockets are not relaible on Solaris'\n\nSo, shall we change pg_regress.sh to not use Unix sockets on Solaris?\n\nThis would potentially cause problems for \"make installcheck\", if the\npostmaster was not started with -i. I suspect the socket problems are\nonly seen when many clients try to connect at the same time, so the\nparallel regression tests are more prone to trouble than serial.\nPerhaps for Solaris, go to TCP only if it's parallel mode?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 11:05:32 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Re: [BUGS] Tests randomly failed " }, { "msg_contents": "Hi Tom,\n\nMy guess is that it would be possible to insert a check to see if the\ninstalled Postmaster was started with -i, and then choose between Unix\ndomain sockets or TCP. BUT, whether trying to explain this in the\ninstallation document to the novice user who is setting up PostgreSQL\nfor about the 1st, 2nd or 3rd time is something to think about...\n\n???\n\nRegards and best wishes,\n\nJustin Clift\n\nTom Lane wrote:\n> \n> Alexander Klimov <ask@wisdom.weizmann.ac.il> writes:\n> > Yes, it was really just incidence -- I try again, and 15 of 15 `make\n> > check' passed with TCP sockets, but only 3 of 15 passed with UNIX\n> > sockets. So, final decision is `Unix sockets are not relaible on Solaris'\n> \n> So, shall we change pg_regress.sh to not use Unix sockets on Solaris?\n> \n> This would potentially cause problems for \"make installcheck\", if the\n> postmaster was not started with -i. I suspect the socket problems are\n> only seen when many clients try to connect at the same time, so the\n> parallel regression tests are more prone to trouble than serial.\n> Perhaps for Solaris, go to TCP only if it's parallel mode?\n> \n> regards, tom lane\n\n-- \n\"My grandfather once told me that there are two kinds of people: those\nwho work and those who take the credit. He told me to try to be in the\nfirst group; there was less competition there.\"\n - Indira Gandhi\n", "msg_date": "Wed, 28 Mar 2001 02:07:03 +1000", "msg_from": "Justin Clift <jclift@iprimus.com.au>", "msg_from_op": false, "msg_subject": "Re: Re: [BUGS] Tests randomly failed" }, { "msg_contents": "Tom Lane writes:\n\n> Alexander Klimov <ask@wisdom.weizmann.ac.il> writes:\n> > Yes, it was really just incidence -- I try again, and 15 of 15 `make\n> > check' passed with TCP sockets, but only 3 of 15 passed with UNIX\n> > sockets. So, final decision is `Unix sockets are not relaible on Solaris'\n\nWhat become up 'set maxuprc=256'? I thought that made it work. Could\nother people try it or has it been disproven?\n\n> So, shall we change pg_regress.sh to not use Unix sockets on Solaris?\n\nThis would hide problems during the test phase which would reappear in the\nproduction phase, no?\n\n> Perhaps for Solaris, go to TCP only if it's parallel mode?\n\nUnfortunately, it's not possible to detect this globally, only when you're\nactually parsing the schedule file and encouter a parallel group. This\nwould mean running some tests this way and some tests another way. That\nmight not be the worst of ideas, but it should be done on all platforms\nthen. Additionally, it don't think it will really fix things, because\nsome tests that failed were not in a parallel group (and I firmly recall\nthat some of those were *not* follow-up failures). I think it is more\nrelated to a \"high load\" situation.\n\n\nIf I were a Solaris user and had a bit more insight into this problem I\nwould probably vote for #undef HAVE_UNIX_SOCKETS. But I'm not...\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Tue, 27 Mar 2001 19:17:47 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: Re: [BUGS] Tests randomly failed " }, { "msg_contents": "On Tue, Mar 27, 2001 at 07:17:47PM +0200, Peter Eisentraut allegedly wrote:\n> Tom Lane writes:\n> \n> > Alexander Klimov <ask@wisdom.weizmann.ac.il> writes:\n> > > Yes, it was really just incidence -- I try again, and 15 of 15 `make\n> > > check' passed with TCP sockets, but only 3 of 15 passed with UNIX\n> > > sockets. So, final decision is `Unix sockets are not relaible on Solaris'\n> \n> What become up 'set maxuprc=256'? I thought that made it work. Could\n> other people try it or has it been disproven?\n\nI'm giving this a test now...\n\nCheers,\n\nMathijs\n-- \n$_='while(read+STDIN,$_,2048){$a=29;$c=142;if((@a=unx\"C*\",$_)[20]&48){$h=5;\n$_=unxb24,join\"\",@b=map{xB8,unxb8,chr($_^$a[--$h+84])}@ARGV;s/...$/1$&/;$d=\nunxV,xb25,$_;$b=73;$e=256|(ord$b[4])<<9|ord$b[3];$d=$d>>8^($f=($t=255)&($d\n>>12^$d>>4^$d^$d/8))<<17,$e=$e>>8^($t&($g=($q=$e>>14&7^$e)^$q*8^$q<<6))<<9\n,$_=(map{$_%16or$t^=$c^=($m=(11,10,116,100,11,122,20,100)[$_/16%8])&110;$t\n^=(72,@z=(64,72,$a^=12*($_%16-2?0:$m&17)),$b^=$_%64?12:0,@z)[$_%8]}(16..271))\n[$_]^(($h>>=8)+=$f+(~$g&$t))for@a[128..$#a]}print+x\"C*\",@a}';s/x/pack+/g;eval \n", "msg_date": "Wed, 28 Mar 2001 02:40:00 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: Re: [BUGS] Tests randomly failed" }, { "msg_contents": "On Wed, Mar 28, 2001 at 02:40:00AM +0200, Mathijs Brands allegedly wrote:\n> On Tue, Mar 27, 2001 at 07:17:47PM +0200, Peter Eisentraut allegedly wrote:\n> > Tom Lane writes:\n> > \n> > > Alexander Klimov <ask@wisdom.weizmann.ac.il> writes:\n> > > > Yes, it was really just incidence -- I try again, and 15 of 15 `make\n> > > > check' passed with TCP sockets, but only 3 of 15 passed with UNIX\n> > > > sockets. So, final decision is `Unix sockets are not relaible on Solaris'\n> > \n> > What become up 'set maxuprc=256'? I thought that made it work. Could\n> > other people try it or has it been disproven?\n> \n> I'm giving this a test now...\n\nNo luck :( Tests still randomly crash. (This is an Ultra 10 machine.)\n\n7.1RC1 on Solaris 7 SPARC does pass the regression tests (apart from the\nrandom test, which seems to be ignored on Solaris). (This is an Ultra\n420 machine.)\n\nCheers,\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Wed, 28 Mar 2001 02:56:10 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Solaris 7 SPARC passes tests (was Re: Re: [BUGS] Tests randomly\n\tfailed)" }, { "msg_contents": "Mathijs Brands <mathijs@ilse.nl> writes:\n> No luck :( Tests still randomly crash. (This is an Ultra 10 machine.)\n\nHow about if you change the pg_regress script to use TCP connections?\n(Look for the bit that forces unix_sockets=no for certain OSes, and\nadd solaris)\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 20:08:47 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Solaris 7 SPARC passes tests (was Re: Re: [BUGS] Tests randomly\n\tfailed)" }, { "msg_contents": "> On Tue, Mar 27, 2001 at 08:08:47PM -0500, Tom Lane wrote:\n> Mathijs Brands <mathijs@ilse.nl> writes:\n> > No luck :( Tests still randomly crash. (This is an Ultra 10 machine.)\n> \n> How about if you change the pg_regress script to use TCP connections?\n> (Look for the bit that forces unix_sockets=no for certain OSes, and\n> add solaris)\n> \n> \t\t\tregards, tom lane\n\nSomeone ran into this again yesterday with Solaris x86. The unix\nsocket problem is probably the same for both architectures, so why\nnot change pg_regress.sh to include *solaris* as part of the same\ncase statement that excludes QNX and BeOS for unix sockets? It is\nsafe to say that Solaris does have this problem.\n\nThe postmaster startup test could say something a bit more useful\nthis way too, as a standard \"make check\" does not report which type\nof sockets are being used (but it does when --temp-install=\"\").\nSome folks may want that to be recorded in the output consistently.\n\nA very small patch to do both of those things is attached.\n\nCheers,\n\n-Rick", "msg_date": "Wed, 28 Mar 2001 03:32:33 -0800", "msg_from": "Rick Robino <rrobino@wavedivision.com>", "msg_from_op": false, "msg_subject": "Re: Solaris 7 SPARC passes tests (was Re: Re: [BUGS] Tests randomly\n\tfailed)" }, { "msg_contents": "On 3/27/01 8:05 AM, \"Tom Lane\" <tgl@sss.pgh.pa.us> wrote:\n\n> Alexander Klimov <ask@wisdom.weizmann.ac.il> writes:\n>> Yes, it was really just incidence -- I try again, and 15 of 15 `make\n>> check' passed with TCP sockets, but only 3 of 15 passed with UNIX\n>> sockets. So, final decision is `Unix sockets are not relaible on Solaris'\n> \n> So, shall we change pg_regress.sh to not use Unix sockets on Solaris?\n\nDoh! I just submitted a patch to change pg_regress.sh before reading all of\ntoday's posts. Oh well, it was small.\n\n> This would potentially cause problems for \"make installcheck\", if the\n> postmaster was not started with -i. I suspect the socket problems are\n> only seen when many clients try to connect at the same time, so the\n> parallel regression tests are more prone to trouble than serial.\n> Perhaps for Solaris, go to TCP only if it's parallel mode?\n\nSo just ignore my patch, as it makes pg_regress never use sockets with any\nsolaris in any mode. However, if you don't want to go with Peter E's #ifdef\nsolution there may be an easier way which is not dangerous at all. Adding a\n--with-sockets option to pg_regress.sh would be trivial and allow one to\ntest either kind of socket on any platform.\n\nMost people who just want to feel good about the build aren't going to do\nmuch about fixing the unix sockets, and are probably not using them anyway.\nSince a default will have to be picked, the one that shows all tests passing\nwill save some noise.\n\nHowever, both sockets may need to be checked during regression to detect the\nproblems Peter mentions. Since there is also a conflict with OpenSSL and\nSolaris' crypt, I can see some sun folks opting for running the database\nlocally as a hedge for trust in a single tier (DMZ) app scenario.\n\nAnyway, after messing around with the script tonight I just wanted to chime\nin that pg_regress.sh could use some improvement:\n\n- A more specific postmaster startup for a normal make check which says\nwhether inet or unix sockets will be used. If you aren't aware of the\nproblem on solaris and don't check netstat, the message is generic and there\nis a socket file in /tmp regardless of what type of socket gets used. Kind\nof subtle.\n\n- Maybe consistency in the script itself. Judging by the different styles of\ntesting and the output between the --temp-install conditions, it appears as\nif each section was written by two different people. It could be cleaned up\npretty fast and quite safely.\n\nNot anything that important, but in the interest of making things easier to\nunderstand these changes could be helpful (IMO). At the very least a mention\nof the socket thing in regress/README or the Solaris FAQ would be handy. I'd\nbe happy to do any of the above if you think they are good ideas. If you're\nalready on it, nevermind and thank you.\n\nCheers,\n\n-- Rick\n\n\n\n", "msg_date": "Wed, 28 Mar 2001 04:36:12 -0800", "msg_from": "\"Richard T. Robino\" <rickspam@wavedivision.com>", "msg_from_op": false, "msg_subject": "Re: Re: [BUGS] Tests randomly failed " }, { "msg_contents": "Richard T. Robino writes:\n\n> - A more specific postmaster startup for a normal make check which says\n> whether inet or unix sockets will be used. If you aren't aware of the\n> problem on solaris and don't check netstat, the message is generic and there\n> is a socket file in /tmp regardless of what type of socket gets used. Kind\n> of subtle.\n\nCan be done. I guess the TCP vs Unix domain issue was never this\nimportant before. The difference is also that the \"installcheck\" mode can\nbe used against either kind of socket using the standard --host and --port\noptions, depending on the requirements of the running server, whereas the\ntemp install mode handles this issue internally -- and it never used to\nmake a difference.\n\n> - Maybe consistency in the script itself. Judging by the different styles of\n> testing and the output between the --temp-install conditions, it appears as\n> if each section was written by two different people. It could be cleaned up\n> pretty fast and quite safely.\n\nAlthough large portions of actual code were copied over from the two\nseparate predecessors to this script, the conventions and formatting\nshould tend to be fairly consistent. Just taking a quick glance now, I\nwould probably still write it this way, although some ideas for cosmetic\nchanges, such as the one above, may arise through actual use.\n\n> Not anything that important, but in the interest of making things easier to\n> understand these changes could be helpful (IMO). At the very least a mention\n> of the socket thing in regress/README or the Solaris FAQ would be handy. I'd\n> be happy to do any of the above if you think they are good ideas. If you're\n> already on it, nevermind and thank you.\n\nUnless we decide on any code measures, it will end up being documented in\nFAQ_Solaris.\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Wed, 28 Mar 2001 23:08:20 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: Re: [BUGS] Tests randomly failed " } ]
[ { "msg_contents": "I contributed the first working s_lock.c code for the SGI's over three\nyears ago (using the test_and_set library calls). It's been working\nfor me ever since in a heavy multi-user environment. Please don't\nchange it unless there's an overwhelming reason.\n\nAlso, the GNU compiler systems works poorly on SGI's. In this\nenvironment, you need to stick to the SGI provided compilers. Once you\nget used to them, they work fairly well.\n\nSincerely,\nBob\n\n+----------------------------------+------------------------------------+\n| Robert E. Bruccoleri, Ph.D. | Phone: 609 737 6383 |\n| President, Congenomics, Inc. | Fax: 609 737 7528 |\n| 114 W Franklin Ave, Suite K1,4,5 | email: bruc@acm.org |\n| P.O. Box 314 | URL: http://www.congen.com/~bruc |\n| Pennington, NJ 08534 | |\n+----------------------------------+------------------------------------+\n", "msg_date": "Tue, 27 Mar 2001 08:59:45 -0500 (EST)", "msg_from": "bruc@stone.congenomics.com (Robert E. Bruccoleri)", "msg_from_op": true, "msg_subject": "Spinlocks on SGI's" }, { "msg_contents": "On Tue, Mar 27, 2001 at 08:59:45AM -0500, Robert E. Bruccoleri allegedly wrote:\n> I contributed the first working s_lock.c code for the SGI's over three\n> years ago (using the test_and_set library calls). It's been working\n> for me ever since in a heavy multi-user environment. Please don't\n> change it unless there's an overwhelming reason.\n> \n> Also, the GNU compiler systems works poorly on SGI's. In this\n> environment, you need to stick to the SGI provided compilers. Once you\n> get used to them, they work fairly well.\n\nThe problem is that SGI wants a couple of hundred dollars for it :(\n\nRegards,\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Tue, 27 Mar 2001 16:02:49 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: Spinlocks on SGI's" }, { "msg_contents": "Dear Mathijs,\n> \n> > \n> > Also, the GNU compiler systems works poorly on SGI's. In this\n> > environment, you need to stick to the SGI provided compilers. Once you\n> > get used to them, they work fairly well.\n> \n> The problem is that SGI wants a couple of hundred dollars for it :(\n\nUnfortunately, the GNU compiler developers have not put much effort\ninto support for the SGI platform over the years, and therefore,\nit's not a good alternative for the SGI compiler system. For PostgreSQL,\nwe need to provide the right advice on how to build it so it works well.\n\nBTW, SGI is distributing its compilers for Linux/Intel under an open\nsource license. Depending on the progress of Intel chips versus MIPS\nchips, the cost issue of the compilers on SGI's will eventually\ndisappear.\n\nRegards, Bob\n\n+----------------------------------+------------------------------------+\n| Robert E. Bruccoleri, Ph.D. | Phone: 609 737 6383 |\n| President, Congenomics, Inc. | Fax: 609 737 7528 |\n| 114 W Franklin Ave, Suite K1,4,5 | email: bruc@acm.org |\n| P.O. Box 314 | URL: http://www.congen.com/~bruc |\n| Pennington, NJ 08534 | |\n+----------------------------------+------------------------------------+\n", "msg_date": "Tue, 27 Mar 2001 09:15:50 -0500 (EST)", "msg_from": "bruc@stone.congenomics.com (Robert E. Bruccoleri)", "msg_from_op": true, "msg_subject": "Re: Spinlocks on SGI's" }, { "msg_contents": "On Tue, 27 Mar 2001, Robert E. Bruccoleri wrote:\n\n> I contributed the first working s_lock.c code for the SGI's over three\n> years ago (using the test_and_set library calls). It's been working\n> for me ever since in a heavy multi-user environment. Please don't\n> change it unless there's an overwhelming reason.\n>\n> Also, the GNU compiler systems works poorly on SGI's. In this\n> environment, you need to stick to the SGI provided compilers. Once you\n> get used to them, they work fairly well.\n\nRobert, any possibility of getting pre-compiled binaries for SGI that we\ncan put up on the site? For those that can't afford the extra cost of the\ncompilers? :)\n\n", "msg_date": "Tue, 27 Mar 2001 10:22:30 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: Spinlocks on SGI's" }, { "msg_contents": "Dear Marc,\n> \n> On Tue, 27 Mar 2001, Robert E. Bruccoleri wrote:\n> \n> > I contributed the first working s_lock.c code for the SGI's over three\n> > years ago (using the test_and_set library calls). It's been working\n> > for me ever since in a heavy multi-user environment. Please don't\n> > change it unless there's an overwhelming reason.\n> >\n> > Also, the GNU compiler systems works poorly on SGI's. In this\n> > environment, you need to stick to the SGI provided compilers. Once you\n> > get used to them, they work fairly well.\n> \n> Robert, any possibility of getting pre-compiled binaries for SGI that we\n> can put up on the site? For those that can't afford the extra cost of the\n> compilers? :)\n\nYes. What exactly do you need from the Postgresql tree after a successful\ncompilation and test?\n\n--Bob\n\n+----------------------------------+------------------------------------+\n| Robert E. Bruccoleri, Ph.D. | Phone: 609 737 6383 |\n| President, Congenomics, Inc. | Fax: 609 737 7528 |\n| 114 W Franklin Ave, Suite K1,4,5 | email: bruc@acm.org |\n| P.O. Box 314 | URL: http://www.congen.com/~bruc |\n| Pennington, NJ 08534 | |\n+----------------------------------+------------------------------------+\n", "msg_date": "Tue, 27 Mar 2001 09:30:41 -0500 (EST)", "msg_from": "bruc@stone.congenomics.com (Robert E. Bruccoleri)", "msg_from_op": true, "msg_subject": "Re: Spinlocks on SGI's" }, { "msg_contents": "On Tue, 27 Mar 2001, Robert E. Bruccoleri wrote:\n\n> BTW, SGI is distributing its compilers for Linux/Intel under an\n> open source license. Depending on the progress of Intel chips\n> versus MIPS chips, the cost issue of the compilers on SGI's will\n> eventually disappear.\n\nDo you mean their ia64 compiler?\n\nThere's a reason that that one's GPLed -- it uses the\ngcc frontend.\n\nMatthew.\n\n", "msg_date": "Tue, 27 Mar 2001 15:52:25 +0100 (BST)", "msg_from": "Matthew Kirkwood <matthew@hairy.beasts.org>", "msg_from_op": false, "msg_subject": "Re: Spinlocks on SGI's" }, { "msg_contents": "On Tue, 27 Mar 2001, Robert E. Bruccoleri wrote:\n\n> Dear Marc,\n> >\n> > On Tue, 27 Mar 2001, Robert E. Bruccoleri wrote:\n> >\n> > > I contributed the first working s_lock.c code for the SGI's over three\n> > > years ago (using the test_and_set library calls). It's been working\n> > > for me ever since in a heavy multi-user environment. Please don't\n> > > change it unless there's an overwhelming reason.\n> > >\n> > > Also, the GNU compiler systems works poorly on SGI's. In this\n> > > environment, you need to stick to the SGI provided compilers. Once you\n> > > get used to them, they work fairly well.\n> >\n> > Robert, any possibility of getting pre-compiled binaries for SGI that we\n> > can put up on the site? For those that can't afford the extra cost of the\n> > compilers? :)\n>\n> Yes. What exactly do you need from the Postgresql tree after a successful\n> compilation and test?\n\neasiest, I think, unless SGI has a special packaging (pkg_add) format, is\na tar of /usr/local/pgsql ... that *should* have everything required ...\nbin, libs and includes ...\n\nif you could create one based on RC1 and upload that, then we can get some\ntesting on whether or not that works before the release?\n\nlogin info to follow ...\n\n\n", "msg_date": "Tue, 27 Mar 2001 11:06:20 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: Spinlocks on SGI's" }, { "msg_contents": "bruc@stone.congenomics.com (Robert E. Bruccoleri) writes:\n> I contributed the first working s_lock.c code for the SGI's over three\n> years ago (using the test_and_set library calls). It's been working\n> for me ever since in a heavy multi-user environment. Please don't\n> change it unless there's an overwhelming reason.\n\nI don't believe anyone was proposing changing the existing __sgi code.\nThe discussion was about adding support for non-SGI MIPS hardware.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 10:26:38 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Spinlocks on SGI's " }, { "msg_contents": "The Hermit Hacker writes:\n\n> Robert, any possibility of getting pre-compiled binaries for SGI that we\n> can put up on the site? For those that can't afford the extra cost of the\n> compilers? :)\n\nI think this\n\nhttp://freeware.sgi.com/Installable/postgresql-7.0.html\n\nis the canonical place to put these things, although I don't know how well\nit is maintained. At least one could try to use their package building\ninfrastructure.\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Tue, 27 Mar 2001 18:40:03 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: Spinlocks on SGI's" }, { "msg_contents": "On Tue, Mar 27, 2001 at 11:06:20AM -0400, The Hermit Hacker allegedly wrote:\n> On Tue, 27 Mar 2001, Robert E. Bruccoleri wrote:\n> > Dear Marc,\n> > > On Tue, 27 Mar 2001, Robert E. Bruccoleri wrote:\n> > >\n> > > > I contributed the first working s_lock.c code for the SGI's over three\n> > > > years ago (using the test_and_set library calls). It's been working\n> > > > for me ever since in a heavy multi-user environment. Please don't\n> > > > change it unless there's an overwhelming reason.\n> > > >\n> > > > Also, the GNU compiler systems works poorly on SGI's. In this\n> > > > environment, you need to stick to the SGI provided compilers. Once you\n> > > > get used to them, they work fairly well.\n> > >\n> > > Robert, any possibility of getting pre-compiled binaries for SGI that we\n> > > can put up on the site? For those that can't afford the extra cost of the\n> > > compilers? :)\n> >\n> > Yes. What exactly do you need from the Postgresql tree after a successful\n> > compilation and test?\n> \n> easiest, I think, unless SGI has a special packaging (pkg_add) format, is\n> a tar of /usr/local/pgsql ... that *should* have everything required ...\n> bin, libs and includes ...\n> \n> if you could create one based on RC1 and upload that, then we can get some\n> testing on whether or not that works before the release?\n\nWouldn't it be best to build IRIX 6.2 binaries? Those should run properly on\nIRIX 6.2 and anything newer. Of course, having both IRIX 6.2 and 6.5 binaries\nwouldn't really hurt...\n\nCheers,\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Wed, 28 Mar 2001 03:04:24 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: Spinlocks on SGI's" }, { "msg_contents": "Dear Mathijs,\n> \n> Wouldn't it be best to build IRIX 6.2 binaries? Those should run properly on\n> IRIX 6.2 and anything newer. Of course, having both IRIX 6.2 and 6.5 binaries\n> wouldn't really hurt...\n\nIt's not feasible to do this, nor is it worth it. Irix 6.5 is much\nmore reliable than Irix 6.2. --Bob\n\n+----------------------------------+------------------------------------+\n| Robert E. Bruccoleri, Ph.D. | Phone: 609 737 6383 |\n| President, Congenomics, Inc. | Fax: 609 737 7528 |\n| 114 W Franklin Ave, Suite K1,4,5 | email: bruc@acm.org |\n| P.O. Box 314 | URL: http://www.congen.com/~bruc |\n| Pennington, NJ 08534 | |\n+----------------------------------+------------------------------------+\n", "msg_date": "Tue, 27 Mar 2001 21:14:16 -0500 (EST)", "msg_from": "bruc@stone.congenomics.com (Robert E. Bruccoleri)", "msg_from_op": true, "msg_subject": "Re: Spinlocks on SGI's" } ]
[ { "msg_contents": "At 3/27/2001 03:21 AM, pgsql-hackers-owner+M6689@postgresql.org wrote:\n>This message was created automatically by mail delivery software.\n>A Majordomo message could not be delivered to the following addresses:\n>\n> martin@math.unl.edu.ar:\n> 450 4.7.1 <martin@math.unl.edu.ar>... Can not check MX records for \n> recipient host math.unl.edu.ar\n>\n>-- Original message omitted --\n\nI have received a ton of these during the last few days... can someone \nplease check to remove this address or correct the problem?\n\n\n", "msg_date": "Tue, 27 Mar 2001 08:46:16 -0600", "msg_from": "Thomas Swan <tswan-lst@ics.olemiss.edu>", "msg_from_op": true, "msg_subject": "Re: Majordomo Delivery Error" }, { "msg_contents": "gone\n\nOn Tue, 27 Mar 2001, Thomas Swan wrote:\n\n> At 3/27/2001 03:21 AM, pgsql-hackers-owner+M6689@postgresql.org wrote:\n> >This message was created automatically by mail delivery software.\n> >A Majordomo message could not be delivered to the following addresses:\n> >\n> > martin@math.unl.edu.ar:\n> > 450 4.7.1 <martin@math.unl.edu.ar>... Can not check MX records for\n> > recipient host math.unl.edu.ar\n> >\n> >-- Original message omitted --\n>\n> I have received a ton of these during the last few days... can someone\n> please check to remove this address or correct the problem?\n>\n>\n>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n>\n\nMarc G. Fournier ICQ#7615664 IRC Nick: Scrappy\nSystems Administrator @ hub.org\nprimary: scrappy@hub.org secondary: scrappy@{freebsd|postgresql}.org\n\n", "msg_date": "Tue, 27 Mar 2001 11:33:23 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: Re: Majordomo Delivery Error" } ]
[ { "msg_contents": "I've already reported this to the webpage, but I got a fail on the random\ntest:\n\n random ... failed (ignored)\n\nThis is on a stock RedHat 7.0 kernel box with the SMP kernel (but running a\nsingle processor):\n\n[pjm3@localhost regress]$ less regression.diffs\n*** ./expected/random.out Thu Jan 6 06:40:54 2000\n--- ./results/random.out Tue Mar 27 15:07:16 2001\n***************\n*** 25,31 ****\n GROUP BY random HAVING count(random) > 1;\n random | count\n --------+-------\n! (0 rows)\n\n SELECT random FROM RANDOM_TBL\n WHERE random NOT BETWEEN 80 AND 120;\n--- 25,32 ----\n GROUP BY random HAVING count(random) > 1;\n random | count\n --------+-------\n! 99 | 2\n! (1 row)\n\n SELECT random FROM RANDOM_TBL\n WHERE random NOT BETWEEN 80 AND 120;\n\n======================================================================\n\nRegards,\nPhil\n\n+----------------------------------+\n| Phil Mayers, Network Support |\n| Centre for Computing Services |\n| Imperial College |\n+----------------------------------+ \n", "msg_date": "Tue, 27 Mar 2001 16:58:21 +0100", "msg_from": "\"Mayers, Philip J\" <p.mayers@ic.ac.uk>", "msg_from_op": true, "msg_subject": "RE: Re: Call for platforms " }, { "msg_contents": "\"Mayers, Philip J\" <p.mayers@ic.ac.uk> writes:\n> I've already reported this to the webpage, but I got a fail on the random\n> test:\n> random ... failed (ignored)\n\nSee\nhttp://www.postgresql.org/devel-corner/docs/postgres/regress.html\nespecially the last item ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 11:43:14 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Re: Call for platforms " } ]
[ { "msg_contents": "======================\n All 76 tests passed.\n======================\n\nvlad: uname -a\nSunOS vlad 5.8 Generic_108528-06 sun4u sparc SUNW,Ultra-5_10\n\nReading specs from /opt/sfw/lib/gcc-lib/sparc-sun-solaris2.8/2.95.2/specs\ngcc version 2.95.2 19991024 (release)\n\nCFLAGS=-O6 -mcpu=v8 -Wa,-xarch=v8plusa\nCXXFLAGS=-O6 -felide-constructors -fno-exceptions -fno-rtti -mcpu=v8 -Wa,-xarch=v8plusa\n\nconfigure --prefix=/opt/pgsql \\\n --with-tcl \\\n --with-includes=/opt/sfw/include \\\n --with-libraries=/opt/sfw/lib \\\n --with-tclconfig=/opt/sfw/lib \\\n --with-CC=gcc \\\n --with-CXX=gcc\n\nThis was cvs checjout this am (3/27/2001)\n\nThanks.\n\n--\nBrian Millett\nEnterprise Consulting Group \"Shifts in paradigms\n(314) 205-9030 often cause nose bleeds.\"\nbpm@ec-group.com Greg Glenn\n\n\n\n", "msg_date": "Tue, 27 Mar 2001 10:34:13 -0600", "msg_from": "Brian P Millett <bpm@ec-group.com>", "msg_from_op": true, "msg_subject": "SPARC Solaris 8 OK" }, { "msg_contents": "> Erm, I've just tested 7.1RC1 on Solaris 8 x86 and SPARC. Solaris x86\n> works without any problems and passes the regression test (apart from\n> a few very small roundoff errors in the geometry tests).\n\nOK, I've got Solaris/x86 checked off. Thanks!\n\n> ... However, on\n> Solaris SPARC tests randomly fail (like someone else reported too). I've\n> tried running configure without any options and the options given above,\n> but the results were the same. I'm using gcc 2.95.3. I'll see if I can\n> give the cvs version and gcc 2.95.2 a try tomorrow. Same goes for\n> Solaris 7 SPARC and maybe Solaris 2.6 SPARC.\n\nRandom failures on Solaris seem to *always* be linked to Unix socket\nresource limits, per the ongoing discussion.\n\n - Thomas\n", "msg_date": "Tue, 27 Mar 2001 23:54:51 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": false, "msg_subject": "Re: Solaris problems (was Re: SPARC Solaris 8 OK)" }, { "msg_contents": "On Tue, Mar 27, 2001 at 10:34:13AM -0600, Brian P Millett allegedly wrote:\n> ======================\n> All 76 tests passed.\n> ======================\n> \n> vlad: uname -a\n> SunOS vlad 5.8 Generic_108528-06 sun4u sparc SUNW,Ultra-5_10\n> \n> Reading specs from /opt/sfw/lib/gcc-lib/sparc-sun-solaris2.8/2.95.2/specs\n> gcc version 2.95.2 19991024 (release)\n> \n> CFLAGS=-O6 -mcpu=v8 -Wa,-xarch=v8plusa\n> CXXFLAGS=-O6 -felide-constructors -fno-exceptions -fno-rtti -mcpu=v8 -Wa,-xarch=v8plusa\n> \n> configure --prefix=/opt/pgsql \\\n> --with-tcl \\\n> --with-includes=/opt/sfw/include \\\n> --with-libraries=/opt/sfw/lib \\\n> --with-tclconfig=/opt/sfw/lib \\\n> --with-CC=gcc \\\n> --with-CXX=gcc\n> \n> This was cvs checjout this am (3/27/2001)\n> \n> Thanks.\n\nErm, I've just tested 7.1RC1 on Solaris 8 x86 and SPARC. Solaris x86\nworks without any problems and passes the regression test (apart from\na few very small roundoff errors in the geometry tests). However, on\nSolaris SPARC tests randomly fail (like someone else reported too). I've\ntried running configure without any options and the options given above,\nbut the results were the same. I'm using gcc 2.95.3. I'll see if I can\ngive the cvs version and gcc 2.95.2 a try tomorrow. Same goes for\nSolaris 7 SPARC and maybe Solaris 2.6 SPARC.\n\nCheers,\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Wed, 28 Mar 2001 02:36:00 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Solaris problems (was Re: SPARC Solaris 8 OK)" } ]
[ { "msg_contents": "Hi all.\n\nAfter two days of strugle I have compiled pgsql on ultrix, patch is in\nattachment. I still can't run `gmake check', because of shell problems\nin install.log:\n\ninstall:\tno destination specified\ngmake[6]: *** [install-lib-shared] Error 1\ngmake[6]: Leaving directory\n`/tmp_mnt/hosts/wisdom/NewSoftware/Ask/build/pgsql/src/interfaces/libpq'\n\nWith the speed of the box it will take another two days to fix and\ncheck :-)\n\nFirst problem is sys/socket.h. Here it has no guards in it (like \n#ifdef _SOCKET_H\n#endif), so it could not be included twice, and I have to remove its\nunnecesarry inclusions.\n\nI have to add src/utils/strdup.o to linking in\nsrc/interfaces/ecpg/preproc and /src/bin/pg_passwd\nand I don't shure how to do it in patch (so, it is not there).\n\n.frame in s_lock.c cause error about duplication of .frame for one .ent,\nso I just remove it -- and I don't shure I am right. Anybody know how to\ndeal with it?\n\nsys/ipc.h included from miscadmin.h is needed for sys/sem.h, so I reorder\ntheir inclusion.\n\nThere is no dynamic libraries for Ultrix, so I have to download libdl, but\nit is not supported by configure (I guess), so I add\nLIBS += -L/home/ask/soft/build/libdl -ldl\nit is obviously wrong -- should be changed.\n\nBTW: Do anybody know about tool, helping to analize include structure, in\norder to eliminate several includes of one file (like case with\nsys/socket.h which is unneeded for everybody, who includes libpq-be.h)\n\nRegards,\nASK", "msg_date": "Tue, 27 Mar 2001 19:14:17 +0200 (IST)", "msg_from": "Alexander Klimov <ask@wisdom.weizmann.ac.il>", "msg_from_op": true, "msg_subject": "Ultrix port" }, { "msg_contents": "Alexander Klimov writes:\n\n> After two days of strugle I have compiled pgsql on ultrix, patch is in\n> attachment. I still can't run `gmake check', because of shell problems\n> in install.log:\n>\n> install:\tno destination specified\n> gmake[6]: *** [install-lib-shared] Error 1\n> gmake[6]: Leaving directory\n> `/tmp_mnt/hosts/wisdom/NewSoftware/Ask/build/pgsql/src/interfaces/libpq'\n\nYou either need to add\n\nenable_shared = no\n\nto Makefile.ultrix4, or inform Makefile.shlib about how to build shared\nlibraries.\n\n> With the speed of the box it will take another two days to fix and\n> check :-)\n\nIf you're using GCC then configure with --enable-depend to build\ndependencies, so you don't have to do make clean everytime.\n\n> First problem is sys/socket.h. Here it has no guards in it (like\n> #ifdef _SOCKET_H\n> #endif), so it could not be included twice, and I have to remove its\n> unnecesarry inclusions.\n\nSince there will be more unnecessary inclusions the next time somebody\ntouches a file, the ultimately better strategy might be to add such a\nwrapper manually.\n\n> I have to add src/utils/strdup.o to linking in\n> src/interfaces/ecpg/preproc and /src/bin/pg_passwd\n> and I don't shure how to do it in patch (so, it is not there).\n\nSee src/bin/psql/Makefile for examples of linking in helper .o files.\n\n> sys/ipc.h included from miscadmin.h is needed for sys/sem.h, so I reorder\n> their inclusion.\n\nBetter to explicitly include sys/ipc.h again where it's needed.\n\n> There is no dynamic libraries for Ultrix, so I have to download libdl, but\n> it is not supported by configure (I guess), so I add\n> LIBS += -L/home/ask/soft/build/libdl -ldl\n> it is obviously wrong -- should be changed.\n\nThere's\n\nAC_CHECK_LIB(dl, main)\n\nwhich does what you want, but you need to give your linker a hint where to\nfind it. See configure --with-libraries.\n\n> BTW: Do anybody know about tool, helping to analize include structure, in\n> order to eliminate several includes of one file (like case with\n> sys/socket.h which is unneeded for everybody, who includes libpq-be.h)\n\nMaybe the stuff under\n\nsrc/tools/pginclude/\n\nhelps.\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Tue, 27 Mar 2001 20:10:17 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: Ultrix port" }, { "msg_contents": "Alexander Klimov <ask@wisdom.weizmann.ac.il> writes:\n> First problem is sys/socket.h. Here it has no guards in it (like \n> #ifdef _SOCKET_H\n> #endif), so it could not be included twice, and I have to remove its\n> unnecesarry inclusions.\n\nAre you sure these are unnecessary? What are the odds this patch will\nbreak other platforms?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 13:14:40 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Ultrix port " }, { "msg_contents": "On Tue, 27 Mar 2001, Tom Lane wrote:\n\n> Alexander Klimov <ask@wisdom.weizmann.ac.il> writes:\n> > First problem is sys/socket.h. Here it has no guards in it (like \n> > #ifdef _SOCKET_H\n> > #endif), so it could not be included twice, and I have to remove its\n> > unnecesarry inclusions.\n> \n> Are you sure these are unnecessary? What are the odds this patch will\n> break other platforms?\nI almost shure it is not, because sys/socket.h already included in each of\nthe files but indirectly, thru libpq-be -> pqcomm, that include it on non\nwindows platforms. My guess is that tracking and elimination of double\nincludes of the same file is useful in general, because it at\nleast decrease compilation time.\n\nRegards,\nASK\n\n", "msg_date": "Thu, 29 Mar 2001 11:18:16 +0200 (IST)", "msg_from": "Alexander Klimov <ask@wisdom.weizmann.ac.il>", "msg_from_op": true, "msg_subject": "Re: Ultrix port " } ]
[ { "msg_contents": "I have just released pgmonitor 0.30. I have added a new 'set debug'\noption that should help with porting. I know some people are testing\nSolaris and this may help.\n\nREADME and CHANGES files attached.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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 P G M O N I T O R\n\npgmonitor, version 0.30\n\nThe main web site for pgmonitor is:\n\thttp://greatbridge.org/project/pgmonitor/projdisplay.php\n\nYou can download the most recent version from\n\tftp://ftp.greatbridge.org/pub/pgmonitor\n\nThis tool allows monitoring of PostgreSQL activity. It requires Tcl/Tk\n8.0 or later. It may require modification of the 'ps' flags for certain\nplatforms. It is known to run on *BSD, Linux, and HPUX.\n\nPgmonitor only works when run on the database server machine. To use it\nremotely, log into the remote machine, set the DISPLAY variable to point\nto your local X server, and start pgmonitor. Pgmonitor will then run on\nthe remote machine, but will display on your local machine.\n\nPgmonitor uses 'ps' to display backend process activity. It uses 'gdb'\nto display running queries, and 'kill' to cancel queries and terminate\ndatabase connections.\n\nPgmonitor stores your most recent refresh and sort settings in the file\n~/.pgmonitor. This file is used to reload your defaults every time\npgmonitor is started.\n\nIf you are running PostgreSQL 7.1.0 or earlier, the 'query' button will\nnot work unless you compile PostgreSQL with debug symbols (-g), or apply\nthe supplied patch 'query_display.diff' and recompile PostgreSQL. The\nlater method is recommended.\n\nFor porting assistance, there is a 'set debug' option in the script that\noutputs status information as pgmonitor is running.\n\nBruce Momjian <pgman@candle.pha.pa.us>\n\n Pgmonitor Release History\n =========================\n\nv0.30 (2001-03-27)\n-----------------\nRemove escape from ^\n\nv0.29 (2001-03-27)\n-----------------\nNew 'set debug' output\nFixes for window close hang\n\nv0.28 (2001-03-26)\n-----------------\nTry PGDATA directory ownership\nSolaris fixes for awk\n\nv0.27 (2001-03-25)\n-----------------\nAwk cleanup\n\nv0.26 (2001-03-24)\n-----------------\nAdd release-specific patches for query display\nCheck for lock file owner if socket does not exist\n\nv0.25 (2001-03-20)\n-----------------\nFix for PostgreSQL 7.0.X on some platforms.\n\nv0.23 (2001-03-19)\n-----------------\nFix tar packaging problem.\nFix wish version setting.\n\nv0.21 (2001-03-17)\n-----------------\nFix for HPUX and other SysV-based systems, from Tom Lane.\n\nv0.20 (2001-03-16)\n-----------------\nInitial public release", "msg_date": "Tue, 27 Mar 2001 12:54:02 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "pgmonitor 0.30 released" }, { "msg_contents": "On Tuesday, March, 2001-03-27 at 12:54:02, Bruce Momjian wrote:\n> I have just released pgmonitor 0.30. I have added a new 'set debug'\n> option that should help with porting. I know some people are testing\n> Solaris and this may help.\n\nwhy don't I see any output? PostgreSQL 7.0.2 on RedHat (the RH7.0\noriginal version).\n\n\nregards\n\n-- \nMarek P�tlicki <marpet@buy.pl>\n", "msg_date": "Tue, 27 Mar 2001 20:20:20 +0200", "msg_from": "=?iso-8859-2?Q?Marek_P=EAtlicki?= <marpet@buy.pl>", "msg_from_op": false, "msg_subject": "Re: pgmonitor 0.30 released" } ]
[ { "msg_contents": "The regression test is failing early on, during initdb. The core file\nindicates that there is a SIGBUS. Hopefully the bugs fixed as a\nresult of the \"More bogus alignment assumptions\" thread will sort\nthings out. Here are initdb.log and the stack trace. If needs be I\ncan recompile with -g but unless I hear otherwise I'll wait for RC2.\n\n\n\n\n\n-- \nPete Forman -./\\.- Disclaimer: This post is originated\nWesternGeco -./\\.- by myself and does not represent\npete.forman@westerngeco.com -./\\.- opinion of Schlumberger, Baker\nhttp://www.crosswinds.net/~petef -./\\.- Hughes or their divisions.", "msg_date": "Tue, 27 Mar 2001 19:36:46 +0100", "msg_from": "Pete Forman <pete.forman@westerngeco.com>", "msg_from_op": true, "msg_subject": "RC1 core dumps in initdb on Solaris 2.6" }, { "msg_contents": "Pete Forman <pete.forman@westerngeco.com> writes:\n> The regression test is failing early on, during initdb. The core file\n> indicates that there is a SIGBUS. Hopefully the bugs fixed as a\n> result of the \"More bogus alignment assumptions\" thread will sort\n> things out.\n\nSure looks like this is the same issue reported (and fixed) before.\nWould you try it with current CVS or last night's snapshot tarball?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 13:57:08 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: RC1 core dumps in initdb on Solaris 2.6 " }, { "msg_contents": "Tom Lane writes:\n > Pete Forman <pete.forman@westerngeco.com> writes:\n > > The regression test is failing early on, during initdb. The core\n > > file indicates that there is a SIGBUS. Hopefully the bugs fixed\n > > as a result of the \"More bogus alignment assumptions\" thread will\n > > sort things out.\n > \n > Sure looks like this is the same issue reported (and fixed) before.\n > Would you try it with current CVS or last night's snapshot tarball?\n\nYes, it is fixed in the snapshot of 2001-03-27, as is the earlier bug\nin pg_backup_null.c. All tests passed.\n\n\n\nI've registered the result via the web form though the report is not\naccurate. Can somebody update the Remarks or Version field to\nindicate that I was using a snapshot rather than RC1 OOTB.\n-- \nPete Forman -./\\.- Disclaimer: This post is originated\nWesternGeco -./\\.- by myself and does not represent\npete.forman@westerngeco.com -./\\.- opinion of Schlumberger, Baker\nhttp://www.crosswinds.net/~petef -./\\.- Hughes or their divisions.\n", "msg_date": "Wed, 28 Mar 2001 10:44:31 +0100", "msg_from": "Pete Forman <pete.forman@westerngeco.com>", "msg_from_op": true, "msg_subject": "Re: RC1 core dumps in initdb on Solaris 2.6 " }, { "msg_contents": "Pete Forman writes:\n > I've registered the result via the web form though the report is\n > not accurate. Can somebody update the Remarks or Version field to\n > indicate that I was using a snapshot rather than RC1 OOTB.\n\nFurther to that request, please ignore/delete the existing Remarks.\n\nI have looked at 12 Solaris systems running 2.5, 2.5.1, 2.6, 7, and 8.\nIt looks as if my machine only has had a non-standard symlink made\nfrom libld.so.2 to libld.so in /usr/lib. This will be removed. As\nfar as I can tell, libld.so.2 is used in a special way by ld rather\nthan being a normal shared library.\n-- \nPete Forman http://www.bedford.waii.com/wsdev/petef/PeteF_links.html\nWesternGeco http://www.crosswinds.net/~petef\nManton Lane, Bedford, mailto:pete.forman@westerngeco.com\nMK41 7PA, UK tel:+44-1234-224798 fax:+44-1234-224804\n", "msg_date": "Wed, 28 Mar 2001 14:26:10 +0100", "msg_from": "Pete Forman <pete.forman@westerngeco.com>", "msg_from_op": true, "msg_subject": "Re: RC1 core dumps in initdb on Solaris 2.6 " } ]
[ { "msg_contents": "Dear all,\n\nplease excuse me for posting out of the blue (I am no longer\nsubscribed) but I have been asked by my colleagues to send a message\nsince I've pretty much been hacking at this problem all day.\n\nTo summarise the issue briefly we were very confused regarding the SET\nTIMEZONE command which behaved differently on Linux and Tru64 Unix\n4.0F. We immediately blamed Postgres, as one normally does, and then\ndecided that since RC1 is out it would be better if we actually worked\nout what the really issue was.\n\nAs usual a good RTFM session provided the answer. The idea of this\nmessage is to provide a sort of \"tutorial\" on how it apparently\ntimezone changes are handled according to POSIX.1 and XPG.4.\n\n\nSET TIMEZONE is dealt with in src/backend/commands/variable.c:357 and\nthe following short program[1]:\n\n/* tzset-test.c */\n\n#include <stdio.h>\n#include <time.h>\n#include <stdlib.h>\n\nmain(int argc, char **argv)\n{\n char *tzone=\"TZ=GMT0\";\n extern long timezone;\n\n if ( argc > 1)\n putenv(argv[1]);\n else\n putenv(tzone);\n tzset();\n printf(\"daylight=%d\\ntimezone=%ld\\ntzname[0]=%s\\ntzname[1]=%s\\n\",\n daylight, timezone,\n (tzname[0] ? tzname[0] : \"NULL\"),\n (tzname[1] ? tzname[1] : \"NULL\"));\n exit(0);\n}\n\nsimulates the procedure used to change the timezone in\nparse_timezone(). In a few words what needs to be done is that the\nenvironment variable TZ is set to the required value and this is\n\"imported\" back into the program by using tzset().\n\nNow, the simple issue we were facing was that setting the timezone to\nGMT worked under Linux but not under Tru64 Unix. In particular\nsomeone on this mailing list replied something along the lines of\n\"well, you need to set it to something which the OS recognises\". It\nturned out that the statement is true but in a different sense than\nwhat we had expected. We were of the mistaken belief that the\ntimezone had to be set to something known in /etc/zoneinfo (Tru64 Unix\nnotation), i.e. one of:\n\nAustralia GMT GMT+7 GMT-6 GMT4 Japan Singapore\nBelfast GMT+0 GMT+8 GMT-7 GMT5 Libya SystemV\nBrazil GMT+1 GMT+9 GMT-8 GMT6 London Turkey\nCET GMT+10 GMT-0 GMT-9 GMT7 MET UCT\nCanada GMT+11 GMT-1 GMT0 GMT8 Mexico US\nChile GMT+12 GMT-10 GMT1 GMT9 NZ UTC\nCuba GMT+13 GMT-11 GMT10 Greenwich NZ-CHAT Universal\nDublin GMT+2 GMT-12 GMT11 Hongkong Navajo W-SU\nEET GMT+3 GMT-2 GMT12 Iceland PRC WET\nEgypt GMT+4 GMT-3 GMT13 Iran Poland Zulu\nFactory GMT+5 GMT-4 GMT2 Israel ROC localtime\nGB-Eire GMT+6 GMT-5 GMT3 Jamaica ROK sources\n\nIt actually turns out that this is not the case. The _correct_ value,\ni.e. the one mandated by the tzset(3) man page and, according to\nMr. Digital, \n\n``\n Interfaces documented on this reference page conform to industry standards\n as follows:\n\n tzset(): POSIX.1, XPG4, XPG4-UNIX\n''\n\nis in fact not \"GMT\" or \"Iceland\" but a string of the form:\n\n``\n When TZ appears in the environment and its value is not a null string, the\n value has one of three formats:\n\n :\n :pathname\n stdoffset[dst[offset] [,start[/time],end[/time]]]\n''\n\nwhere ':' means UTC, ':pathname' sends you to the zoneinfo file and\nthe last one is the string which should be used. In particular, where\nLinux accepts GMT and reads it to be GMT0, under Tru64 Unix the\ncorrect behaviour _requires_ the use of GMT0.\n\nExamples of this behaviour are (local timezone EET DST, GMT+3):\n\n[Tru64 Unix 4.0F (and 4.0G)]\n./tzset-test TZ=GMT\ndaylight=1\ntimezone=-7200\ntzname[0]=EET\ntzname[1]=EET DST\n\n[Debian GNU/Linux 2.2 (Kernel 2.2.18, glibc 2.1.3)]\n./tzset-test TZ=GMT\ndaylight=0\ntimezone=0\ntzname[0]=GMT\ntzname[1]=GMT\n\nwhereas the POSIXly \"correct\" (the use of quotes will become apparent\nlater) setting of TZ=GMT0 gives the \"expected\" result:\n\n[Tru64 Unix 4.0F (and 4.0G)]\n./tzset-test TZ=GMT0\ndaylight=0\ntimezone=0\ntzname[0]=GMT\ntzname[1]=\n\n[Debian GNU/Linux 2.2 (Kernel 2.2.18, glibc 2.1.3)]\n./tzset-test TZ=GMT0\ndaylight=0\ntimezone=0\ntzname[0]=GMT\ntzname[1]=GMT\n\nAs you can imagine the above discrepancy, seen from within Postgres\nbut not tested separately, had driven us to despair for our application\nwhich \"localised\" times depending on the remote user location.\n\nNow, this might be all closed but as a matter of fact we went a little\nfurther and discovered that, as long as you use the POSIXly defined\nformat you can specify the timezones to be anything you want! For\nexample[2]:\n\n./tzset-test TZ=PIPPO0PLUTO-2\ndaylight=1\ntimezone=0\ntzname[0]=PIPPO\ntzname[1]=PLUTO\n\nis perfectly valid. Not only, unless you specify also the \"change\ndates\" for DST, as specified earlier, you run the risk of wrong\nconversions. Any timezone used at the moment will always be assumed\nto be in DST if your local timezone is in DST! So, the call\n./tzset-test 'TZ=CET-1CET DST-2' is correct (Central European Time is\n1 hr East of GMT and 2 hrs East of GMT during DST) and changes to DST\ncorrectly only because EET on the local server changes to EET EST at\nthe same time. So far the only \"fix\" we have is to write the changes\nin full gory detail by taking them from the sources in\n/etc/zoneinfo/sources (Tru64 Unix location).\n\nNow, possibly a small buglet. The external variable \"timezone\"\nappears to be corrupted somewhere in Postgres. For reasons which are\nnot entirely clear to me when we added some debugging lines to\nparse_timezone() in an attempt to resolve the issue, the value of\ntimezone was never the expected one. After adding an explicit \n\n extern long timezone;\n\nto the function it was still happily there and values printed are\nobviously wrong (compare with the -7200 in one of the examples above,\nit is meant to be the number of seconds west of GMT, negative numbers\nindicating east).\n\nAs an example, the log output from the \"patched\" version of variable.c\nis:\n\nNew parse_timezone(value) call\ndefaultTZ = 0 (NULL)\ndefaultTZ = 0 (NULL)\ntzbuf = 14004efe0 (TZ=GMT0)\ntimezone = 3458201564533694440, dst = 0, tzname=GMT/\nEnd of parse_timezone()\n\nand the value of timezone is clearly rather incorrect.\n\nI have been unable to work out if this is because timezone is shadowed\n(the cause of my retracted post to tru64-unix-managers) or not as I\ncannot find anything which obviously is shadowing it (the macro\nTIMEZONE_GLOBAL appears to be used consistently and the only\noccurrence of a \"bool timezone\" is within a struct).\n\nPlease excuse the length of this post but given the amount of time\nspent today working all this out we all thought it made sense to\ndocument our effort.\n\nCiao,\n\nArrigo\n\n[1] You might recognise this code from my erroneous posting to the\n tru64-unix-managers which claimed the bug was in their library...\n[2] The names are the italian equivalent of foo and bar (see \"The\n New Hacker's Dictionary\" by Eric Raymond).\n\n-- \nArrigo Triulzi <arrigo@albourne.com>\nAlbourne Partners Ltd. - London, UK\n", "msg_date": "Tue, 27 Mar 2001 19:44:33 +0100", "msg_from": "Arrigo Triulzi <arrigo@albourne.com>", "msg_from_op": true, "msg_subject": "A few notes on timezones" } ]
[ { "msg_contents": "I'm having a problem with functions written in SQL. Specifically, they\ndon't seem to be adhering to Postgres locking rules. For the record, I'm \nusing postgres 7.0.3, installed from RPMs, on Red Hat 6.2. I got the same\nresults with postgres 7.1 beta 6, installed from sources.\n\nHere's what I'm seeing:\n(psql input represented by '<<'; output represented by '>>'.)\n\n\nsession1<< create table idseq\nsession1<< (\nsession1<< name varchar(32) not null,\nsession1<< id int8 not null default 0\nsession1<< );\nsession1>> CREATE\n\nsession1<< insert into idseq values ('myid');\nsession1>> INSERT 18734 1\n\n\nEach row in the table is supposed to represent a named numeric sequence,\nmuch like the sequences built into postgres. (Mine use an int8 though,\nso their values can be much higher.)\n\n\nsession1<< create function nextid( varchar(32)) returns int8 as '\nsession1<< select * from idseq where name = $1::text for update;\nsession1<< update idseq set id = id + 1 where name = $1::text;\nsession1<< select id from idseq where name = $1::text;\nsession1<< ' language 'sql';\nsession1>> CREATE\n\n\nThe idea here is that the select...for update within the nextid() function\nwill establish a row level lock, preventing two concurrent function calls\nfrom overlapping.\n\nNext, I test with two sessions as follows:\n\n\nsession1<< begin;\nsession1>> BEGIN\n\nsession2<< begin;\nsession2>> BEGIN\n\nsession1<< select nextid('myid');\nsession1>> nextid\nsession1>> --------\nsession1>> 1\nsession1>> (1 row)\n\nsession2<< select nextid('myid');\n\n(session2 blocks until session1 completes its transaction)\n\nsession1<< commit;\nsession1>> COMMIT\n\n(session2 resumes)\n\nsession2>> nextid\nsession2>> --------\nsession2>> 0\nsession2>> (1 row)\n\nWhat gives??? I expected the second call to nextid() to return 2!\n\nsession2<< commit;\nsession2>> COMMIT\n\nsession2<< select * from idseq;\nsession2>> name | id\nsession2>> ------+----\nsession2>> myid | 2\nsession2>> (1 row)\n\nsession1<< select * from idseq;\nsession1>> name | id\nsession1>> ------+----\nsession1>> myid | 2\nsession1>> (1 row)\n\nAs you can see, my nextid() function is not synchronized the way I hoped.\nI don't know why, though. Can someone help? I'm going to try out some of my \nSPI functions with 7.1 beta 6, to see if they exhibit a locking problem as \nwell.\n\nThanks,\n\nForest Wilkinson\n\n", "msg_date": "Tue, 27 Mar 2001 13:10:16 -0800", "msg_from": "Forest Wilkinson <fspam@home.com>", "msg_from_op": true, "msg_subject": "possible row locking bug in 7.0.3 & 7.1" }, { "msg_contents": "Forest Wilkinson <fspam@home.com> writes:\n> session1<< create function nextid( varchar(32)) returns int8 as '\n> session1<< select * from idseq where name = $1::text for update;\n> session1<< update idseq set id = id + 1 where name = $1::text;\n> session1<< select id from idseq where name = $1::text;\n> session1<< ' language 'sql';\n> [ doesn't work as expected in parallel transactions ]\n\nThis is a fairly interesting example. What I find is that at the final\nSELECT, the function can see both the tuple outdated by the other\ntransaction AND the new tuple it has inserted. (You can demonstrate\nthat by doing select count(id) instead of select id.) Whichever one\nhappens to be visited first is the one that gets returned by the\nfunction, and that's generally the older one in this example.\n\nMVCC seems to be operating as designed here, more or less. The outdated\ntuple is inserted by a known-committed transaction, and deleted by a\ntransaction that's also committed, but one that committed *since the\nstart of the current transaction*. So its effects should not be visible\nto the SELECT, and therefore the tuple should be visible. The anomalous\nbehavior is not really in the final SELECT, but in the earlier commands\nthat were able to see the effects of a transaction committed later than\nthe start of the second session's transaction.\n\nThe workaround for Forest is to make the final SELECT be a SELECT FOR\nUPDATE, so that it's playing by the same rules as the earlier commands.\nBut I wonder whether we ought to rethink the MVCC rules so that that's\nnot necessary. I have no idea how we might change the rules though.\nIf nothing else, we should document this issue better: SELECT and SELECT\nFOR UPDATE have different visibility rules, so you probably don't want\nto intermix them.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 18:14:02 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: possible row locking bug in 7.0.3 & 7.1 " }, { "msg_contents": "At 18:14 27/03/01 -0500, Tom Lane wrote:\n>Forest Wilkinson <fspam@home.com> writes:\n>> session1<< create function nextid( varchar(32)) returns int8 as '\n>> session1<< select * from idseq where name = $1::text for update;\n>> session1<< update idseq set id = id + 1 where name = $1::text;\n>> session1<< select id from idseq where name = $1::text;\n>> session1<< ' language 'sql';\n>> [ doesn't work as expected in parallel transactions ]\n>\n>What I find is that at the final\n>SELECT, the function can see both the tuple outdated by the other\n>transaction AND the new tuple it has inserted.\n\nSurely we should distinguish between real new tuples, and new tuple\nversions? I don't think it's ever reasonable behaviour to see two versions\nof the same row.\n\n\n>(You can demonstrate\n>that by doing select count(id) instead of select id.) Whichever one\n>happens to be visited first is the one that gets returned by the\n>function, and that's generally the older one in this example.\n>\n>MVCC seems to be operating as designed here, more or less. The outdated\n>tuple is inserted by a known-committed transaction, and deleted by a\n>transaction that's also committed, but one that committed *since the\n>start of the current transaction*. So its effects should not be visible\n>to the SELECT, and therefore the tuple should be visible. The anomalous\n>behavior is not really in the final SELECT, but in the earlier commands\n>that were able to see the effects of a transaction committed later than\n>the start of the second session's transaction.\n\nLooking at the docs, I see that 'SERIALIZABLE' has the same visibility\nrules as 'READ COMMITTED', which is very confusing. I expect that a Read\nCommitted TX should see committed changes for a TX that commits during the\nfirst TX (although this may need to be limited to TXs started before the\nfirst TX, but I'm not sure). If this is not the case, then we never get\nnon-repeatable reads, AFAICT:\n\n P2 (������Non-repeatable read������): SQL-transaction T1 reads a row. \n SQL-transaction T2 then modifies or deletes that row and performs \n a COMMIT. If T1 then attempts to reread the row, it may\n receive the modified value or discover that the row has been deleted.\n\nwhich is one of the differences between SERIALIZABLE and READ-COMMITTED.\n\n\n>The workaround for Forest is to make the final SELECT be a SELECT FOR\n>UPDATE, so that it's playing by the same rules as the earlier commands.\n\nEek. Does this seem good to you? I would expect that SELECT and\nSELECT...FOR UPDATE should return the same result set.\n\n\n>But I wonder whether we ought to rethink the MVCC rules so that that's\n>not necessary. I have no idea how we might change the rules though.\n\nDisallowing visibility of two versions of the same row would help.\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Wed, 28 Mar 2001 10:14:05 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: possible row locking bug in 7.0.3 &\n 7.1" }, { "msg_contents": "Philip Warner <pjw@rhyme.com.au> writes:\n>> The workaround for Forest is to make the final SELECT be a SELECT FOR\n>> UPDATE, so that it's playing by the same rules as the earlier commands.\n\n> Eek. Does this seem good to you?\n\nI did call it a workaround ;-)\n\nI don't think that we dare try to make any basic changes in MVCC for 7.1\nat this late hour, so Forest is going to have to live with that answer\nfor awhile. But I would like to see a cleaner answer in future\nreleases. As I've opined before, the whole EvalPlanQual mechanism\nstrikes me as essentially bogus in any case...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 19:30:16 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1 " }, { "msg_contents": "On Tuesday 27 March 2001 15:14, Tom Lane wrote:\n> Forest Wilkinson <fspam@home.com> writes:\n> > session1<< create function nextid( varchar(32)) returns int8 as '\n> > session1<< select * from idseq where name = $1::text for update;\n> > session1<< update idseq set id = id + 1 where name = $1::text;\n> > session1<< select id from idseq where name = $1::text;\n> > session1<< ' language 'sql';\n> > [ doesn't work as expected in parallel transactions ]\n[snip]\n> The workaround for Forest is to make the final SELECT be a SELECT FOR\n> UPDATE, so that it's playing by the same rules as the earlier commands.\n> But I wonder whether we ought to rethink the MVCC rules so that that's\n> not necessary. I have no idea how we might change the rules though.\n> If nothing else, we should document this issue better: SELECT and SELECT\n> FOR UPDATE have different visibility rules, so you probably don't want\n> to intermix them.\n\nMy, that's ugly. (But thanks for the workaround.)\n\nIf I remember correctly, UPDATE establishes a lock on the affected rows, \nwhich will block another UPDATE on the same rows for the duration of the \ntransaction. If that's true, shouldn't I be able to achieve my desired \nbehavior by removing the initial as follows:\n\ncreate function nextid( varchar(32)) returns int8 as '\n update idseq set id = id + 1 where name = $1::text;\n select id from idseq where name = $1::text;\n ' language 'sql';\n\nOr, would I still have to add FOR UPDATE to that final SELECT?\n\n\nForest\n", "msg_date": "Tue, 27 Mar 2001 22:08:00 -0800", "msg_from": "Forest Wilkinson <fspam@home.com>", "msg_from_op": true, "msg_subject": "Re: possible row locking bug in 7.0.3 & 7.1" }, { "msg_contents": "Forest Wilkinson <fspam@home.com> writes:\n> If I remember correctly, UPDATE establishes a lock on the affected rows, \n> which will block another UPDATE on the same rows for the duration of the \n> transaction. If that's true, shouldn't I be able to achieve my desired \n> behavior by removing the initial as follows:\n\n> create function nextid( varchar(32)) returns int8 as '\n> update idseq set id = id + 1 where name = $1::text;\n> select id from idseq where name = $1::text;\n> ' language 'sql';\n\n> Or, would I still have to add FOR UPDATE to that final SELECT?\n\nYou're right, the initial SELECT FOR UPDATE is a waste of cycles\nconsidering that you're not using the value it returns. But you'll\nstill need the last select to be FOR UPDATE so that it plays by the\nsame rules as the UPDATE does.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 01:11:08 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: possible row locking bug in 7.0.3 & 7.1 " }, { "msg_contents": "Tom Lane wrote:\n> \n> Philip Warner <pjw@rhyme.com.au> writes:\n> >> The workaround for Forest is to make the final SELECT be a SELECT FOR\n> >> UPDATE, so that it's playing by the same rules as the earlier commands.\n> \n> > Eek. Does this seem good to you?\n> \n> I did call it a workaround ;-)\n> \n> I don't think that we dare try to make any basic changes in MVCC for 7.1\n> at this late hour, so Forest is going to have to live with that answer\n> for awhile. But I would like to see a cleaner answer in future\n> releases.\n\nIs it the MVCC's restriction that each query inside a function\nmust use the same snapshot ?\n\n> As I've opined before, the whole EvalPlanQual mechanism\n> strikes me as essentially bogus in any case...\n> \n\nHow would you change it ? UPDATE/SELECT FOR UPDATE have to\nSELECT/UPDATE the latest tuples. I don't think of any simple\nway for 'SELECT FOR UPDATE' to have the same visibility as\nsimple SELECT.\n\nregards,\nHiroshi Inoue\n", "msg_date": "Wed, 28 Mar 2001 15:19:05 +0900", "msg_from": "Hiroshi Inoue <Inoue@tpf.co.jp>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" } ]
[ { "msg_contents": "I believe I have found the cause of the pgmonitor problem on Solaris.\n\npgmonitor assumes you have a working 'ps' status display, which shows\nuser/database/connection, and query type. I have learned from a Great\nBridge employee that his Solaris box is not updating the ps display for\n7.1beta, and I assume the same trouble exists for 7.0.3.\n\nCan someone confirm that 'ps' status display doesn't work on Solaris,\nand if it doesn't, can someone come up with a fix that we can put into\n7.1.1? You can look in /pg/backend/utils/misc/ps_status.c for all the\nps status code. It has many options for ps status updating.\n\nThanks.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 27 Mar 2001 18:41:23 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "pgmonitor and Solaris" }, { "msg_contents": "Bruce Momjian writes:\n\n> Can someone confirm that 'ps' status display doesn't work on Solaris,\n\nYou may need to use /usr/ucb/ps. Last I checked there was no way to\nchange the display of /usr/bin/ps.\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Wed, 28 Mar 2001 22:55:58 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "> Bruce Momjian writes:\n> \n> > Can someone confirm that 'ps' status display doesn't work on Solaris,\n> \n> You may need to use /usr/ucb/ps. Last I checked there was no way to\n> change the display of /usr/bin/ps.\n\nI can do that, but I am told that /usr/ucb/ps does not allow me to\nrestrict the display to a specific user. Without that, ps shows all\nprocesses, which is pretty slow to run regularly, no? Can you check for\na user restriction on /usr/ucb/ps. I can certainly code in a check for\nits existance and use that instead.\n\nDoes 'ps' status display work with PostgreSQL and /usr/ucb/ps?\n\nLet me know. Thanks.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 28 Mar 2001 16:31:08 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "whats wrong with /usr/ucb/ps auxw | grep $PGUSER \n\nto get only the processes for PG?\n\nLER\n\n-- \nLarry Rosenman http://www.lerctr.org/~ler/\nPhone: +1 972 414 9812 E-Mail: ler@lerctr.org\nUS Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749 US\n\n>>>>>>>>>>>>>>>>>> Original Message <<<<<<<<<<<<<<<<<<\n\nOn 3/28/01, 3:31:08 PM, Bruce Momjian <pgman@candle.pha.pa.us> wrote \nregarding [HACKERS] Re: [PORTS] pgmonitor and Solaris:\n\n\n> > Bruce Momjian writes:\n> >\n> > > Can someone confirm that 'ps' status display doesn't work on Solaris,\n> >\n> > You may need to use /usr/ucb/ps. Last I checked there was no way to\n> > change the display of /usr/bin/ps.\n\n> I can do that, but I am told that /usr/ucb/ps does not allow me to\n> restrict the display to a specific user. Without that, ps shows all\n> processes, which is pretty slow to run regularly, no? Can you check for\n> a user restriction on /usr/ucb/ps. I can certainly code in a check for\n> its existance and use that instead.\n\n> Does 'ps' status display work with PostgreSQL and /usr/ucb/ps?\n\n> Let me know. Thanks.\n\n> --\n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (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> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n", "msg_date": "Wed, 28 Mar 2001 21:51:50 GMT", "msg_from": "Larry Rosenman <ler@lerctr.org>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "[ Charset ISO-8859-1 unsupported, converting... ]\n> whats wrong with /usr/ucb/ps auxw | grep $PGUSER \n> \n> to get only the processes for PG?\n\nI can do that if there is no other option, but on my BSDI machine,\nrestricting ps to a specific user is much faster than a ps on the whole\nsystem. Seeing that 'ps' is run by default every 5 seconds, this could\nbe a performance issue.\n\nIf I have to use 'grep' I will, but I was hoping for a real user\nrestriction.\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 28 Mar 2001 16:56:15 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "On Wed, Mar 28, 2001 at 04:56:15PM -0500, Bruce Momjian allegedly wrote:\n> [ Charset ISO-8859-1 unsupported, converting... ]\n> > whats wrong with /usr/ucb/ps auxw | grep $PGUSER \n> > \n> > to get only the processes for PG?\n> \n> I can do that if there is no other option, but on my BSDI machine,\n> restricting ps to a specific user is much faster than a ps on the whole\n> system. Seeing that 'ps' is run by default every 5 seconds, this could\n> be a performance issue.\n> \n> If I have to use 'grep' I will, but I was hoping for a real user\n> restriction.\n\nHow about the following:\n\nilsefe2:~$ /usr/ucb/ps -aux|head -1\nUSER PID %CPU %MEM SZ RSS TT S START TIME COMMAND\nBroken Pipe\nilsefe2:~$ /usr/ucb/ps -aux|grep mathijs\nmathijs 7255 0.1 0.2 2432 1816 pts/0 S 00:12:41 0:00 -bash\nmathijs 7775 0.1 0.1 960 728 pts/0 S 00:24:29 0:00 grep mathijs\nmathijs 7344 0.0 0.1 1064 912 pts/0 T 00:14:56 0:00 man ps\nmathijs 7359 0.0 0.1 1040 808 pts/0 T 00:14:57 0:00 sh -c more -s /tmp\nmathijs 7360 0.0 0.1 1184 968 pts/0 T 00:14:57 0:00 more -s /tmp/mp3Ha\nilsefe2:~$ ps -U mathijs -o user,pid,pcpu,pmem,vsz,rss,tty,s,stime=START -o time,comm\n USER PID %CPU %MEM VSZ RSS TT S START TIME COMMAND\n mathijs 7359 0.0 0.1 1040 808 pts/0 T 00:14:57 0:00 sh\n mathijs 7255 0.1 0.2 2432 1816 pts/0 S 00:12:41 0:00 -bash\n root 7816 0.1 0.1 1080 840 pts/0 O 00:28:13 0:00 ps\n mathijs 7344 0.0 0.1 1064 912 pts/0 T 00:14:56 0:00 man\n mathijs 7360 0.0 0.1 1184 968 pts/0 T 00:14:57 0:00 more\n\n\nIt doesn't use any extra extra programs, nor the somewhat deprecated /usr/ucb/ps\ncommand. The only problem I see may be the alignment of some fields. The ps\ncommand itself is listed because the user mathijs was running it...\n\nRegards,\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Thu, 29 Mar 2001 00:30:06 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "> How about the following:\n> \n> ilsefe2:~$ /usr/ucb/ps -aux|head -1\n> USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND\n> Broken Pipe\n> ilsefe2:~$ /usr/ucb/ps -aux|grep mathijs\n> mathijs 7255 0.1 0.2 2432 1816 pts/0 S 00:12:41 0:00 -bash\n> mathijs 7775 0.1 0.1 960 728 pts/0 S 00:24:29 0:00 grep mathijs\n> mathijs 7344 0.0 0.1 1064 912 pts/0 T 00:14:56 0:00 man ps\n> mathijs 7359 0.0 0.1 1040 808 pts/0 T 00:14:57 0:00 sh -c more -s /tmp\n> mathijs 7360 0.0 0.1 1184 968 pts/0 T 00:14:57 0:00 more -s /tmp/mp3Ha\n> ilsefe2:~$ ps -U mathijs -o user,pid,pcpu,pmem,vsz,rss,tty,s,stime=START -o time,comm\n> USER PID %CPU %MEM VSZ RSS TT S START TIME COMMAND\n> mathijs 7359 0.0 0.1 1040 808 pts/0 T 00:14:57 0:00 sh\n> mathijs 7255 0.1 0.2 2432 1816 pts/0 S 00:12:41 0:00 -bash\n> root 7816 0.1 0.1 1080 840 pts/0 O 00:28:13 0:00 ps\n> mathijs 7344 0.0 0.1 1064 912 pts/0 T 00:14:56 0:00 man\n> mathijs 7360 0.0 0.1 1184 968 pts/0 T 00:14:57 0:00 more\n> \n> \n> It doesn't use any extra extra programs, nor the somewhat deprecated /usr/ucb/ps\n> command. The only problem I see may be the alignment of some fields. The ps\n> command itself is listed because the user mathijs was running it...\n\nSure 'ps -U' will work, but it was reported that on Solaris, plain ps\ncan't show the postgres status display, while ucb/ps can. I don't need\nspecific columns. What I need is the postgres status parameters, and if\npossible, a user restriction to ps for performance reasons.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 28 Mar 2001 17:33:05 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "On Thu, Mar 29, 2001 at 12:30:06AM +0200, Mathijs Brands allegedly wrote:\n> On Wed, Mar 28, 2001 at 04:56:15PM -0500, Bruce Momjian allegedly wrote:\n> > [ Charset ISO-8859-1 unsupported, converting... ]\n> > > whats wrong with /usr/ucb/ps auxw | grep $PGUSER \n> > > \n> > > to get only the processes for PG?\n> > \n> > I can do that if there is no other option, but on my BSDI machine,\n> > restricting ps to a specific user is much faster than a ps on the whole\n> > system. Seeing that 'ps' is run by default every 5 seconds, this could\n> > be a performance issue.\n> > \n> > If I have to use 'grep' I will, but I was hoping for a real user\n> > restriction.\n> \n> How about the following:\n> \n> ilsefe2:~$ /usr/ucb/ps -aux|head -1\n> USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND\n> Broken Pipe\n> ilsefe2:~$ /usr/ucb/ps -aux|grep mathijs\n> mathijs 7255 0.1 0.2 2432 1816 pts/0 S 00:12:41 0:00 -bash\n> mathijs 7775 0.1 0.1 960 728 pts/0 S 00:24:29 0:00 grep mathijs\n> mathijs 7344 0.0 0.1 1064 912 pts/0 T 00:14:56 0:00 man ps\n> mathijs 7359 0.0 0.1 1040 808 pts/0 T 00:14:57 0:00 sh -c more -s /tmp\n> mathijs 7360 0.0 0.1 1184 968 pts/0 T 00:14:57 0:00 more -s /tmp/mp3Ha\n> ilsefe2:~$ ps -U mathijs -o user,pid,pcpu,pmem,vsz,rss,tty,s,stime=START -o time,comm\n> USER PID %CPU %MEM VSZ RSS TT S START TIME COMMAND\n> mathijs 7359 0.0 0.1 1040 808 pts/0 T 00:14:57 0:00 sh\n> mathijs 7255 0.1 0.2 2432 1816 pts/0 S 00:12:41 0:00 -bash\n> root 7816 0.1 0.1 1080 840 pts/0 O 00:28:13 0:00 ps\n> mathijs 7344 0.0 0.1 1064 912 pts/0 T 00:14:56 0:00 man\n> mathijs 7360 0.0 0.1 1184 968 pts/0 T 00:14:57 0:00 more\n\nDamn! Small correction:\n\nps -U mathijs -o user,pid,pcpu,pmem,vsz=SZ -o rss,tty,s,stime=START -o time,comm\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Thu, 29 Mar 2001 00:34:53 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "On Wed, Mar 28, 2001 at 05:33:05PM -0500, Bruce Momjian allegedly wrote:\n> Sure 'ps -U' will work, but it was reported that on Solaris, plain ps\n> can't show the postgres status display, while ucb/ps can. I don't need\n> specific columns. What I need is the postgres status parameters, and if\n> possible, a user restriction to ps for performance reasons.\n\nMy mistake. Have a look at this snippet from the ps manpage:\n\n| args The command with all its arguments as a string. The\n| implementation may truncate this value to the field\n| width; it is implementation-dependent whether any\n| further truncation occurs. It is unspecified whether\n| the string represented is a version of the argument\n| list as it was passed to the command when it started,\n| or is a version of the arguments as they may have been\n| modified by the application. Applications cannot\n| depend on being able to modify their argument list and\n| having that modification be reflected in the output of\n| ps. The Solaris implementation limits the string to\n| 80 bytes; the string is the version of the argument\n| list as it was passed to the command when it started.\n\nNote the last line...\n\nThe following must also seem familiar ;)\n\n| The Hermit Hacker <scrappy@hub.org> writes:\n| > On Wed, 29 Apr 1998, Bruce Momjian wrote:\n| >> No reason for the exec(). I believe the only advantage is that it gives\n| >> us a separate process name in the 'ps' listing. I have looked into\n| >> simulating this.\n| > \tUnder FreeBSD, there is:\n| > setproctitle(3) - set the process title for ps 1\n| > \tThis isn't available under Solaris though, last I checked...\n| \n| Setting the process title from C is messy, but there is a readily\n| available reference. The Berkeley sendmail distribution includes code\n| to emulate setproctitle on practically every platform. See conf.h and\n| conf.c in any recent sendmail release. Warning: it's grotty enough to\n| make a strong man weep. Don't read near mealtime ;-)\n| \n| \t\t\tregards, tom lane\n\nRegards,\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Thu, 29 Mar 2001 00:51:26 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "> On Wed, Mar 28, 2001 at 05:33:05PM -0500, Bruce Momjian allegedly wrote:\n> > Sure 'ps -U' will work, but it was reported that on Solaris, plain ps\n> > can't show the postgres status display, while ucb/ps can. I don't need\n> > specific columns. What I need is the postgres status parameters, and if\n> > possible, a user restriction to ps for performance reasons.\n> \n> My mistake. Have a look at this snippet from the ps manpage:\n> \n> | args The command with all its arguments as a string. The\n> | implementation may truncate this value to the field\n> | width; it is implementation-dependent whether any\n> | further truncation occurs. It is unspecified whether\n> | the string represented is a version of the argument\n> | list as it was passed to the command when it started,\n> | or is a version of the arguments as they may have been\n> | modified by the application. Applications cannot\n> | depend on being able to modify their argument list and\n> | having that modification be reflected in the output of\n> | ps. The Solaris implementation limits the string to\n> | 80 bytes; the string is the version of the argument\n> | list as it was passed to the command when it started.\n> \n> Note the last line...\n\nOK, I need someone on Solaris to test ps and /ucb/ps with regard to user\nrestriction inside ps, and display of PostgreSQL status display.\n\nI have uploaded a new pgmonitor 0.33 version that has a show_all\nconfiguration parameter. This will show all PostgreSQL-owned processes\nfor use on operating systems that don't have PostgreSQL status display.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 28 Mar 2001 18:07:46 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "> On Wed, Mar 28, 2001 at 05:33:05PM -0500, Bruce Momjian allegedly wrote:\n>> Sure 'ps -U' will work, but it was reported that on Solaris, plain ps\n>> can't show the postgres status display, while ucb/ps can. I don't need\n>> specific columns. What I need is the postgres status parameters, and if\n>> possible, a user restriction to ps for performance reasons.\n\nLooking at the sendmail code, it seems they use SPT_REUSEARGV (what we\ncall PS_USE_CLOBBER_ARGV) technique on Solaris. Possibly the problem is\nsimply that line 65 in src/backend/utils/misc/ps_status.c fails to\ncover Solaris as one of the possible options:\n\n#elif defined(__linux__) || defined(_AIX4) || defined(_AIX3) || defined(__sgi) || (defined(sun) && !defined(BSD)) || defined(ultrix) || defined(__ksr__) || defined(__osf__) || defined(__QNX__) || defined(__svr4__) || defined(__svr5__)\n#define PS_USE_CLOBBER_ARGV\n\nCan someone check whether adding an appropriate Solaris symbol (which\none?) fixes the problem?\n\nThe coding technique embodied in lines 63-67 pretty much sucks anyway,\nsince this platform-specific knowledge ought to be out in the port.h\nfiles rather than hidden in the guts of the system. I don't want to\ntouch it right now, but I think we ought to have things like\n\n\t#define PS_STATUS_TECHNIQUE PS_USE_CLOBBER_ARGV\n\nin the port.h files, rather than these massive #ifdefs.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 18:23:16 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris " }, { "msg_contents": "\nFYI, the WU-FTPD code (2.6.0 or better) has a couple of more platforms \nincluding UnixWare. The UnixWare code will need /dev/kmem permission to \nchange it's stuff, so I don't know whether we want to do this or not, but \nif people are looking at the ps stuff, please look at this as well.\n\nLarry Rosenman\n-- \nLarry Rosenman http://www.lerctr.org/~ler/\nPhone: +1 972 414 9812 E-Mail: ler@lerctr.org\nUS Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749 US\n>>>>>>>>>>>>>>>>>> Original Message <<<<<<<<<<<<<<<<<<\n\nOn 3/28/01, 5:23:16 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote regarding Re: \n[HACKERS] Re: [PORTS] pgmonitor and Solaris :\n\n\n> > On Wed, Mar 28, 2001 at 05:33:05PM -0500, Bruce Momjian allegedly wrote:\n> >> Sure 'ps -U' will work, but it was reported that on Solaris, plain ps\n> >> can't show the postgres status display, while ucb/ps can. I don't need\n> >> specific columns. What I need is the postgres status parameters, and if\n> >> possible, a user restriction to ps for performance reasons.\n\n> Looking at the sendmail code, it seems they use SPT_REUSEARGV (what we\n> call PS_USE_CLOBBER_ARGV) technique on Solaris. Possibly the problem is\n> simply that line 65 in src/backend/utils/misc/ps_status.c fails to\n> cover Solaris as one of the possible options:\n\n> #elif defined(__linux__) || defined(_AIX4) || defined(_AIX3) || \ndefined(__sgi) || (defined(sun) && !defined(BSD)) || defined(ultrix) || \ndefined(__ksr__) || defined(__osf__) || defined(__QNX__) || \ndefined(__svr4__) || defined(__svr5__)\n> #define PS_USE_CLOBBER_ARGV\n\n> Can someone check whether adding an appropriate Solaris symbol (which\n> one?) fixes the problem?\n\n> The coding technique embodied in lines 63-67 pretty much sucks anyway,\n> since this platform-specific knowledge ought to be out in the port.h\n> files rather than hidden in the guts of the system. I don't want to\n> touch it right now, but I think we ought to have things like\n\n> #define PS_STATUS_TECHNIQUE PS_USE_CLOBBER_ARGV\n\n> in the port.h files, rather than these massive #ifdefs.\n\n> regards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 23:36:30 GMT", "msg_from": "Larry Rosenman <ler@lerctr.org>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris " }, { "msg_contents": "On Wed, Mar 28, 2001 at 06:23:16PM -0500, Tom Lane allegedly wrote:\n> > On Wed, Mar 28, 2001 at 05:33:05PM -0500, Bruce Momjian allegedly wrote:\n> >> Sure 'ps -U' will work, but it was reported that on Solaris, plain ps\n> >> can't show the postgres status display, while ucb/ps can. I don't need\n> >> specific columns. What I need is the postgres status parameters, and if\n> >> possible, a user restriction to ps for performance reasons.\n> \n> Looking at the sendmail code, it seems they use SPT_REUSEARGV (what we\n> call PS_USE_CLOBBER_ARGV) technique on Solaris. Possibly the problem is\n> simply that line 65 in src/backend/utils/misc/ps_status.c fails to\n> cover Solaris as one of the possible options:\n> \n> #elif defined(__linux__) || defined(_AIX4) || defined(_AIX3) || defined(__sgi) || (defined(sun) && !defined(BSD)) || defined(ultrix) || defined(__ksr__) || defined(__osf__) || defined(__QNX__) || defined(__svr4__) || defined(__svr5__)\n> #define PS_USE_CLOBBER_ARGV\n> \n> Can someone check whether adding an appropriate Solaris symbol (which\n> one?) fixes the problem?\n\nPS_USE_CLOBBER_ARGV -is- used on Solaris. I just checked.\n\nilsedb:~$ uname -a\nSunOS ilsedb 5.7 Generic_106541-11 sun4u sparc SUNW,Ultra-80\nilsedb:~$ cat test.c\n#include <stdio.h>\n\nmain()\n{\n printf(\"%d\\n\", __svr4__);\n}\nilsedb:~$ gcc -o test test.c\nilsedb:~$ ./test\n1\n\nRegards,\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Thu, 29 Mar 2001 01:38:42 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "Mathijs Brands <mathijs@ilse.nl> writes:\n> PS_USE_CLOBBER_ARGV -is- used on Solaris. I just checked.\n\nHm. But 7.1 postgres backends fail to change their ps display?\nDoes sendmail change its ps display on your machine?\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 18:48:27 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris " }, { "msg_contents": "Larry Rosenman <ler@lerctr.org> writes:\n> FYI, the WU-FTPD code (2.6.0 or better) has a couple of more platforms \n> including UnixWare. The UnixWare code will need /dev/kmem permission to \n> change it's stuff, so I don't know whether we want to do this or not, but \n> if people are looking at the ps stuff, please look at this as well.\n\nWell, *I* sure wouldn't run Postgres with write permission on /dev/kmem.\nEspecially not just to make ps_status work...\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 18:51:13 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris " }, { "msg_contents": "I tend to agree, but wanted someone else's opinion.\n\nLER\n\n\n>>>>>>>>>>>>>>>>>> Original Message <<<<<<<<<<<<<<<<<<\n\nOn 3/28/01, 5:51:13 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote regarding Re: \n[ADMIN] Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris :\n\n\n> Larry Rosenman <ler@lerctr.org> writes:\n> > FYI, the WU-FTPD code (2.6.0 or better) has a couple of more platforms\n> > including UnixWare. The UnixWare code will need /dev/kmem permission to\n> > change it's stuff, so I don't know whether we want to do this or not, but\n> > if people are looking at the ps stuff, please look at this as well.\n\n> Well, *I* sure wouldn't run Postgres with write permission on /dev/kmem.\n> Especially not just to make ps_status work...\n\n> regards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 23:54:13 GMT", "msg_from": "Larry Rosenman <ler@lerctr.org>", "msg_from_op": false, "msg_subject": "Re: Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris " }, { "msg_contents": "On Wed, Mar 28, 2001 at 06:48:27PM -0500, Tom Lane allegedly wrote:\n> Mathijs Brands <mathijs@ilse.nl> writes:\n> > PS_USE_CLOBBER_ARGV -is- used on Solaris. I just checked.\n> \n> Hm. But 7.1 postgres backends fail to change their ps display?\n> Does sendmail change its ps display on your machine?\n\ntemplate1=# \\set\nVERSION = 'PostgreSQL 7.0.3 on sparc-sun-solaris2.7, compiled by gcc 2.95.2'\nDBNAME = 'template1'\nUSER = 'mathijs'\nPORT = '5432'\nENCODING = 'SQL_ASCII'\nPROMPT1 = '%/%R%# '\nPROMPT2 = '%/%R%# '\nPROMPT3 = '>> '\nHISTSIZE = '500'\ntemplate1=#\n[1]+ Stopped psql template1\njumpstart.l3.ilse.nl:/export/home/mathijs >/usr/ucb/ps auxww|grep post\nmathijs 297 0.1 0.2 6120 3352 pts/1 S 01:59:18 0:00 /opt/pgsql/bin/./postgres mathijs localhost template1 idle\nmathijs 319 0.1 0.1 944 696 pts/3 S 01:59:45 0:00 grep post\nmathijs 25810 0.0 0.1 5736 2160 pts/1 S 20:44:34 0:00 /opt/pgsql/bin/./postmaster \n\n\ntemplate1=# \\set\nVERSION = 'PostgreSQL 7.1RC1 on sparc-sun-solaris2.8, compiled by GCC 2.95.3'\nDBNAME = 'template1'\nUSER = 'mathijs'\nPORT = '5432'\nENCODING = 'SQL_ASCII'\nPROMPT1 = '%/%R%# '\nPROMPT2 = '%/%R%# '\nPROMPT3 = '>> '\ntemplate1=# ^Z[1] + Stopped (SIGTSTP) psql template1\n$ /usr/ucb/ps auxww|grep post\nmathijs 8072 0.1 1.4 5552 3352 pts/2 S 01:01:43 0:00 ./postmaster -D /export/home/mathijs/pgtest/data\nmathijs 8074 0.1 0.3 976 672 pts/2 S 01:01:55 0:00 grep post\nmathijs 8067 0.1 0.8 4912 1984 pts/2 S 01:01:37 0:00 ./postmaster -D /export/home/mathijs/pgtest/data\n\nI guess it doesn't work :( Of course, it's also possible it doesn't\nwork on Solaris 8, but does on 7. I'll check this and if this is the\ncase, I'll post it.\n\nRegards,\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Thu, 29 Mar 2001 02:07:39 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "* Tom Lane <tgl@sss.pgh.pa.us> [010328 16:07] wrote:\n> Larry Rosenman <ler@lerctr.org> writes:\n> > FYI, the WU-FTPD code (2.6.0 or better) has a couple of more platforms \n> > including UnixWare. The UnixWare code will need /dev/kmem permission to \n> > change it's stuff, so I don't know whether we want to do this or not, but \n> > if people are looking at the ps stuff, please look at this as well.\n> \n> Well, *I* sure wouldn't run Postgres with write permission on /dev/kmem.\n> Especially not just to make ps_status work...\n\nWow... is this all for \"pgmonitor\"?\n\nsorry, just my opinion...\n\nIf it for pgmonitor then you guys ought to just mark it broken on\nthese platforms, the non-\"ps based\" solution could have been\nimplemented with all the time wasted trying to get the \"ps based\"\nhack working. :(\n\n-- \n-Alfred Perlstein - [bright@wintelcom.net|alfred@freebsd.org]\nDaemon News Magazine in your snail-mail! http://magazine.daemonnews.org/\n", "msg_date": "Wed, 28 Mar 2001 16:10:52 -0800", "msg_from": "Alfred Perlstein <bright@wintelcom.net>", "msg_from_op": false, "msg_subject": "Re: Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "Mathijs Brands <mathijs@ilse.nl> writes:\n> mathijs 297 0.1 0.2 6120 3352 pts/1 S 01:59:18 0:00 /opt/pgsql/bin/./postgres mathijs localhost template1 idle\n\nThe interesting point about this is that the 7.0.3-on-2.7 installation\n*is* managing to change its PS display. So either Solaris 2.8\nretrogressed (different predefined symbols maybe?), or we broke the code\nsince 7.0.3.\n\nAnyway I think the right thing to look at is why the 7.1 install is not\nmanaging to update the display.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 19:12:39 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris " }, { "msg_contents": "> > On Wed, Mar 28, 2001 at 05:33:05PM -0500, Bruce Momjian allegedly wrote:\n> >> Sure 'ps -U' will work, but it was reported that on Solaris, plain ps\n> >> can't show the postgres status display, while ucb/ps can. I don't need\n> >> specific columns. What I need is the postgres status parameters, and if\n> >> possible, a user restriction to ps for performance reasons.\n> \n> Looking at the sendmail code, it seems they use SPT_REUSEARGV (what we\n> call PS_USE_CLOBBER_ARGV) technique on Solaris. Possibly the problem is\n> simply that line 65 in src/backend/utils/misc/ps_status.c fails to\n> cover Solaris as one of the possible options:\n> \n> #elif defined(__linux__) || defined(_AIX4) || defined(_AIX3) || defined(__sgi) || (defined(sun) && !defined(BSD)) || defined(ultrix) || defined(__ksr__) || defined(__osf__) || defined(__QNX__) || defined(__svr4__) || defined(__svr5__)\n> #define PS_USE_CLOBBER_ARGV\n> \n> Can someone check whether adding an appropriate Solaris symbol (which\n> one?) fixes the problem?\n> \n> The coding technique embodied in lines 63-67 pretty much sucks anyway,\n> since this platform-specific knowledge ought to be out in the port.h\n> files rather than hidden in the guts of the system. I don't want to\n> touch it right now, but I think we ought to have things like\n> \n> \t#define PS_STATUS_TECHNIQUE PS_USE_CLOBBER_ARGV\n> \n> in the port.h files, rather than these massive #ifdefs.\n\nAdded to TODO:\n\n * Move platform-specific ps status display info from ps_status.c to ports\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 28 Mar 2001 19:12:45 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "> Mathijs Brands <mathijs@ilse.nl> writes:\n> > PS_USE_CLOBBER_ARGV -is- used on Solaris. I just checked.\n> \n> Hm. But 7.1 postgres backends fail to change their ps display?\n> Does sendmail change its ps display on your machine?\n\nSeems we could easily use the sendmail display to populate the ps\ndisplay defines/ports.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 28 Mar 2001 19:14:00 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "Alfred Perlstein <bright@wintelcom.net> writes:\n> If it for pgmonitor then you guys ought to just mark it broken on\n> these platforms, the non-\"ps based\" solution could have been\n> implemented with all the time wasted trying to get the \"ps based\"\n> hack working. :(\n\nMy thoughts exactly ;-) ... I'm willing to give advice about this,\nbut not to do any legwork myself.\n\nPS status display does have some utility with or without pgmonitor,\nbut not enough to do a lot of work for.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 19:14:27 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris " }, { "msg_contents": "> Larry Rosenman <ler@lerctr.org> writes:\n> > FYI, the WU-FTPD code (2.6.0 or better) has a couple of more platforms \n> > including UnixWare. The UnixWare code will need /dev/kmem permission to \n> > change it's stuff, so I don't know whether we want to do this or not, but \n> > if people are looking at the ps stuff, please look at this as well.\n> \n> Well, *I* sure wouldn't run Postgres with write permission on /dev/kmem.\n> Especially not just to make ps_status work...\n\nAgreed. If that is what is required, I think you can forget ps status\non that platform.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 28 Mar 2001 19:14:35 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "> Mathijs Brands <mathijs@ilse.nl> writes:\n> > mathijs 297 0.1 0.2 6120 3352 pts/1 S 01:59:18 0:00 /opt/pgsql/bin/./postgres mathijs localhost template1 idle\n> \n> The interesting point about this is that the 7.0.3-on-2.7 installation\n> *is* managing to change its PS display. So either Solaris 2.8\n> retrogressed (different predefined symbols maybe?), or we broke the code\n> since 7.0.3.\n> \n> Anyway I think the right thing to look at is why the 7.1 install is not\n> managing to update the display.\n\n[ CC just to ports]\n\nAlso, does /bin/ps show that status too, or just ucb/ps?\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 28 Mar 2001 19:22:29 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] Re: pgmonitor and Solaris" }, { "msg_contents": "On Wed, Mar 28, 2001 at 07:12:39PM -0500, Tom Lane allegedly wrote:\n> Mathijs Brands <mathijs@ilse.nl> writes:\n> > mathijs 297 0.1 0.2 6120 3352 pts/1 S 01:59:18 0:00 /opt/pgsql/bin/./postgres mathijs localhost template1 idle\n> \n> The interesting point about this is that the 7.0.3-on-2.7 installation\n> *is* managing to change its PS display. So either Solaris 2.8\n> retrogressed (different predefined symbols maybe?), or we broke the code\n> since 7.0.3.\n> \n> Anyway I think the right thing to look at is why the 7.1 install is not\n> managing to update the display.\n> \n> \t\t\tregards, tom lane\n\nHere's 7.1:\n\ntemplate1=# \\set\nVERSION = 'PostgreSQL 7.1RC1 on sparc-sun-solaris2.7, compiled by GCC 2.8.1'\nDBNAME = 'template1'\nUSER = 'mathijs'\nPORT = '5432'\nENCODING = 'SQL_ASCII'\nPROMPT1 = '%/%R%# '\nPROMPT2 = '%/%R%# '\nPROMPT3 = '>> '\ntemplate1=# ^Z\n[1]+ Stopped ./psql template1\nilsedb:~/pgtest/bin$ /usr/ucb/ps auxww|grep post\nmathijs 29830 0.1 0.2 6328 3976 pts/4 S 02:18:16 0:00 ./postmaster -D /export/home/mathijs/pgtest/data\nmathijs 29832 0.1 0.1 960 704 pts/4 S 02:18:28 0:00 grep post\nmathijs 29823 0.0 0.1 5696 2176 pts/4 S 02:18:03 0:00 ./postmaster -D /export/home/mathijs/pgtest/data\n\nNo go...\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Thu, 29 Mar 2001 02:24:39 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "On Wed, Mar 28, 2001 at 04:10:52PM -0800, Alfred Perlstein wrote:\n> * Tom Lane <tgl@sss.pgh.pa.us> [010328 16:07] wrote:\n> > Larry Rosenman <ler@lerctr.org> writes:\n> > > FYI, the WU-FTPD code (2.6.0 or better) has a couple of more platforms \n> > > including UnixWare. The UnixWare code will need /dev/kmem permission to \n> > > change it's stuff, so I don't know whether we want to do this or not, but \n> > > if people are looking at the ps stuff, please look at this as well.\n> > \n> > Well, *I* sure wouldn't run Postgres with write permission on /dev/kmem.\n> > Especially not just to make ps_status work...\n> \n> Wow... is this all for \"pgmonitor\"?\n> \n> sorry, just my opinion...\n> \n> If it for pgmonitor then you guys ought to just mark it broken on\n> these platforms, the non-\"ps based\" solution could have been\n> implemented with all the time wasted trying to get the \"ps based\"\n> hack working. :(\n\n If is still a mind for (IMHO) terrible work with 'ps'... not sure if \nit's interesting, but exist cross-platforms library for processes \ninformation loading - the libgtop. Supports:\n\n\t* All versions of Linux\n\t* FreeBSD, NetBSD and OpenBSD\n\t* BSD/OS\n\t* Digital Unix\n\t* Solaris\n\t* ..and may be others\n\n Sure, it's in C :-)\n\n\t\t\tKarel\n\n-- \n Karel Zak <zakkr@zf.jcu.cz>\n http://home.zf.jcu.cz/~zakkr/\n \n C, PostgreSQL, PHP, WWW, http://docs.linux.cz, http://mape.jcu.cz\n", "msg_date": "Thu, 29 Mar 2001 11:20:02 +0200", "msg_from": "Karel Zak <zakkr@zf.jcu.cz>", "msg_from_op": false, "msg_subject": "Re: [ADMIN] Re: Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "Tom Lane previously wrote:\n > Looking at the sendmail code, it seems they use SPT_REUSEARGV (what\n > we call PS_USE_CLOBBER_ARGV) technique on Solaris. Possibly the\n > problem is simply that line 65 in\n > src/backend/utils/misc/ps_status.c fails to cover Solaris as one of\n > the possible options:\n >\n > #elif defined(__linux__) || defined(_AIX4) || defined(_AIX3)\n > || defined(__sgi) || (defined(sun) && !defined(BSD))\n > || defined(ultrix) || defined(__ksr__) || defined(__osf__)\n > || defined(__QNX__) || defined(__svr4__) || defined(__svr5__)\n > #define PS_USE_CLOBBER_ARGV\n\n> Can someone check whether adding an appropriate Solaris symbol\n> (which one?) fixes the problem?\n\n\nTom Lane writes:\n > The interesting point about this is that the 7.0.3-on-2.7\n > installation *is* managing to change its PS display. So either\n > Solaris 2.8 retrogressed (different predefined symbols maybe?), or\n > we broke the code since 7.0.3.\n > \n > Anyway I think the right thing to look at is why the 7.1 install is\n > not managing to update the display.\n\nThe identifier sun is not set if the compiler is in -Xc mode. It\nwould be safer to use to use __sun which is defined in all compiler\nmodes. The symbols defined in all modes are __sun, __unix,\n__SUNPRO_C=0x500 (or 400, 420, etc.), __`uname -s`_`uname -r`\n(e.g. __SunOS_5_8), __sparc (SPARC), __sparcv9 (SPARC with\n-xarch=v9|v9a), __i386 (x86), __BUILTIN_VA_ARG_INCR, __SVR4.\n\nThat applies to the native compiler. Perhaps someone else could\nestablish what Solaris specific defines are available in gcc.\n-- \nPete Forman -./\\.- Disclaimer: This post is originated\nWesternGeco -./\\.- by myself and does not represent\npete.forman@westerngeco.com -./\\.- opinion of Schlumberger, Baker\nhttp://www.crosswinds.net/~petef -./\\.- Hughes or their divisions.\n", "msg_date": "Thu, 29 Mar 2001 11:05:52 +0100", "msg_from": "Pete Forman <pete.forman@westerngeco.com>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: pgmonitor and Solaris " }, { "msg_contents": "Bruce Momjian wrote:\n> > Larry Rosenman <ler@lerctr.org> writes:\n> > > FYI, the WU-FTPD code (2.6.0 or better) has a couple of more platforms\n> > > including UnixWare. The UnixWare code will need /dev/kmem permission to\n> > > change it's stuff, so I don't know whether we want to do this or not, but\n> > > if people are looking at the ps stuff, please look at this as well.\n> >\n> > Well, *I* sure wouldn't run Postgres with write permission on /dev/kmem.\n> > Especially not just to make ps_status work...\n>\n> Agreed. If that is what is required, I think you can forget ps status\n> on that platform.\n\nBruce,\n\n have you taken a look at my statistics collector stuff yet?\n The view pgstat_activity shows datid, datname, procpid,\n usesysid, usename and current_query per backend.\n\n The code still needs some more work, but as it is it's ready\n to apply as soon as 7.1 is out the door. All that is totally\n OS independant, restricts the display of querystrings to\n postgres superusers and works remotely.\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#================================================== JanWieck@Yahoo.com #\n\n\n\n_________________________________________________________\nDo You Yahoo!?\nGet your free @yahoo.com address at http://mail.yahoo.com\n\n", "msg_date": "Thu, 29 Mar 2001 07:12:34 -0500 (EST)", "msg_from": "Jan Wieck <JanWieck@Yahoo.com>", "msg_from_op": false, "msg_subject": "Re: Re: [HACKERS] Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "> > Agreed. If that is what is required, I think you can forget ps status\n> > on that platform.\n> \n> Bruce,\n> \n> have you taken a look at my statistics collector stuff yet?\n> The view pgstat_activity shows datid, datname, procpid,\n> usesysid, usename and current_query per backend.\n> \n> The code still needs some more work, but as it is it's ready\n> to apply as soon as 7.1 is out the door. All that is totally\n> OS independant, restricts the display of querystrings to\n> postgres superusers and works remotely.\n> \n\n[ Sent only to ports]\n\nYes, that is what I will do for 7.2, but that may be a year away.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 29 Mar 2001 08:32:36 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [ADMIN] Re: [HACKERS] Re: pgmonitor and Solaris" }, { "msg_contents": "My apologies for all the pgmonitor crossposts. All future discussion\nshould take place on the ports list. Thanks.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 29 Mar 2001 08:36:48 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] Re: pgmonitor and Solaris" }, { "msg_contents": "Tom Lane writes:\n\n> Mathijs Brands <mathijs@ilse.nl> writes:\n> > mathijs 297 0.1 0.2 6120 3352 pts/1 S 01:59:18 0:00 /opt/pgsql/bin/./postgres mathijs localhost template1 idle\n>\n> The interesting point about this is that the 7.0.3-on-2.7 installation\n> *is* managing to change its PS display. So either Solaris 2.8\n> retrogressed (different predefined symbols maybe?), or we broke the code\n> since 7.0.3.\n\nI think I broke it.\n\nThe deal on Solaris is this: There's SysV-style ps at /usr/bin/ps, and\nthere's BSD-style ps at /usr/ucb/ps. The most obvious differences are the\ndifferent argument style and the different output format. In true BSD\nstyle, the /usr/ucb/ps display can be changed by assigning argv[x] =\n\"whatever\". In true SysV style, the /usr/bin/ps display can (presumably)\nbe changed with strcpy(argv[0], \"whatever\").\n\nThe old (pre-7.1) code only worked for BSD and had an exception case for\nLinux. The new code is mostly adopted from Sendmail. So in pre-7.1 you\ncould get /usr/ucb/ps to work, while Sendmail apparently tried to get the\nSysV-style ps to work -- and failed(?). (This is generally the right\ndirection, because the /usr/ucb stuff is obsolescent on Solaris, only for\ncompatibility with SunOS 4.)\n\nThe consequence should be:\n\n1. check if sendmail works with /usr/bin/ps\n\n2. a) if yes, figure out what got lost in PostgreSQL\n\n2. b) if no, make the Solaris case in ps_status.c use the BSD approach\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Thu, 29 Mar 2001 18:49:42 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: Re: [PORTS] pgmonitor and Solaris " }, { "msg_contents": "> Tom Lane writes:\n> \n> > Mathijs Brands <mathijs@ilse.nl> writes:\n> > > mathijs 297 0.1 0.2 6120 3352 pts/1 S 01:59:18 0:00 /opt/pgsql/bin/./postgres mathijs localhost template1 idle\n> >\n> > The interesting point about this is that the 7.0.3-on-2.7 installation\n> > *is* managing to change its PS display. So either Solaris 2.8\n> > retrogressed (different predefined symbols maybe?), or we broke the code\n> > since 7.0.3.\n> \n> I think I broke it.\n> \n> The deal on Solaris is this: There's SysV-style ps at /usr/bin/ps, and\n> there's BSD-style ps at /usr/ucb/ps. The most obvious differences are the\n> different argument style and the different output format. In true BSD\n> style, the /usr/ucb/ps display can be changed by assigning argv[x] =\n> \"whatever\". In true SysV style, the /usr/bin/ps display can (presumably)\n> be changed with strcpy(argv[0], \"whatever\").\n> \n> The old (pre-7.1) code only worked for BSD and had an exception case for\n> Linux. The new code is mostly adopted from Sendmail. So in pre-7.1 you\n> could get /usr/ucb/ps to work, while Sendmail apparently tried to get the\n> SysV-style ps to work -- and failed(?). (This is generally the right\n> direction, because the /usr/ucb stuff is obsolescent on Solaris, only for\n> compatibility with SunOS 4.)\n> \n> The consequence should be:\n> \n> 1. check if sendmail works with /usr/bin/ps\n> \n> 2. a) if yes, figure out what got lost in PostgreSQL\n> \n> 2. b) if no, make the Solaris case in ps_status.c use the BSD approach\n\n[ I had tried to move this discussion to ports, but hackers is OK too.]\n\nWell, this is very interesting. I am glad to hear you based the current\nps_status code on sendmail, which I think is the perfect way to go. \n\nI seem to remember the Solaris manual page stating it doesn't update the\nps display, but I may be mixing that up with something else. I agree\n/usr/bin/ps is the better option, if we can get it working. Most people\nwill not remember to use /usr/ucb/ps.\n\nI just uploaded a 0.34 version that allows the 'ps' command to be\ncustomized by editing the script.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 29 Mar 2001 12:02:32 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "* Karel Zak <zakkr@zf.jcu.cz> [010329 03:10] wrote:\n> On Wed, Mar 28, 2001 at 04:10:52PM -0800, Alfred Perlstein wrote:\n> > * Tom Lane <tgl@sss.pgh.pa.us> [010328 16:07] wrote:\n> > > Larry Rosenman <ler@lerctr.org> writes:\n> > > > FYI, the WU-FTPD code (2.6.0 or better) has a couple of more platforms \n> > > > including UnixWare. The UnixWare code will need /dev/kmem permission to \n> > > > change it's stuff, so I don't know whether we want to do this or not, but \n> > > > if people are looking at the ps stuff, please look at this as well.\n> > > \n> > > Well, *I* sure wouldn't run Postgres with write permission on /dev/kmem.\n> > > Especially not just to make ps_status work...\n> > \n> > Wow... is this all for \"pgmonitor\"?\n> > \n> > sorry, just my opinion...\n> > \n> > If it for pgmonitor then you guys ought to just mark it broken on\n> > these platforms, the non-\"ps based\" solution could have been\n> > implemented with all the time wasted trying to get the \"ps based\"\n> > hack working. :(\n> \n> If is still a mind for (IMHO) terrible work with 'ps'... not sure if \n> it's interesting, but exist cross-platforms library for processes \n> information loading - the libgtop. Supports:\n> \n> \t* All versions of Linux\n> \t* FreeBSD, NetBSD and OpenBSD\n> \t* BSD/OS\n> \t* Digital Unix\n> \t* Solaris\n> \t* ..and may be others\n> \n> Sure, it's in C :-)\n\nAnd infected with the GPV, not even the LGPV... not worth it. :)\n\n-- \n-Alfred Perlstein - [bright@wintelcom.net|alfred@freebsd.org]\nDaemon News Magazine in your snail-mail! http://magazine.daemonnews.org/\n", "msg_date": "Thu, 29 Mar 2001 11:57:30 -0800", "msg_from": "Alfred Perlstein <bright@wintelcom.net>", "msg_from_op": false, "msg_subject": "Re: [ADMIN] Re: Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "Bruce Momjian writes:\n > > Tom Lane writes:\n > > The consequence should be:\n > > \n > > 1. check if sendmail works with /usr/bin/ps\n > > \n > > 2. a) if yes, figure out what got lost in PostgreSQL\n > > \n > > 2. b) if no, make the Solaris case in ps_status.c use the BSD\n > > approach\n > \n > [ I had tried to move this discussion to ports, but hackers is OK\n > too.]\n > \n > Well, this is very interesting. I am glad to hear you based the\n > current ps_status code on sendmail, which I think is the perfect\n > way to go.\n > \n > I seem to remember the Solaris manual page stating it doesn't\n > update the ps display, but I may be mixing that up with something\n > else. I agree /usr/bin/ps is the better option, if we can get it\n > working. Most people will not remember to use /usr/ucb/ps.\n\nI've done a quick survey of the Suns available to me to see whether\nsendmail updates the ps display. The summary is:\n\n 1) Require \"/usr/ucb/ps w\" at least. /usr/bin/ps has no options\n that I can find to display the status of sendmail.\n\n 2) Older versions of sendmail do not update the status. 8.6 does\n not update, 8.8.8 does.\n\n 3) Solaris 2.5 and 2.5.1 have sendmail 8.6, 7 has 8.9.1, 8 has\n 8.9.3. Some 2.6 have 8.6, others have 8.8.8. Presumably\n patches have beed applied.\n\nOn other OSs, AIX 4.1, 4.2, 4.3 have sendmail 8.8.4, 8.8.6, 8.8.8 or\n8.9.3. They all display the status in both SysV and BSD modes.\nOn AIX there is one ps command which handles both styles.\n\nIRIX 6.2 and 6.5.4m through 6.5.10m have sendmail 8.8.8, 8.9.1, 8.9.3.\nNo status is available. There do not appear to be any BSD-ish ps\noptions.\n-- \nPete Forman -./\\.- Disclaimer: This post is originated\nWesternGeco -./\\.- by myself and does not represent\npete.forman@westerngeco.com -./\\.- opinion of Schlumberger, Baker\nhttp://www.crosswinds.net/~petef -./\\.- Hughes or their divisions.\n", "msg_date": "Fri, 30 Mar 2001 11:07:25 +0100", "msg_from": "Pete Forman <pete.forman@westerngeco.com>", "msg_from_op": false, "msg_subject": "Re: Re: [PORTS] pgmonitor and Solaris" }, { "msg_contents": "On Fri, Mar 30, 2001 at 11:07:25AM +0100, Pete Forman allegedly wrote:\n> I've done a quick survey of the Suns available to me to see whether\n> sendmail updates the ps display. The summary is:\n> \n> 1) Require \"/usr/ucb/ps w\" at least. /usr/bin/ps has no options\n> that I can find to display the status of sendmail.\n> \n> 2) Older versions of sendmail do not update the status. 8.6 does\n> not update, 8.8.8 does.\n> \n> 3) Solaris 2.5 and 2.5.1 have sendmail 8.6, 7 has 8.9.1, 8 has\n> 8.9.3. Some 2.6 have 8.6, others have 8.8.8. Presumably\n> patches have beed applied.\n> \n> On other OSs, AIX 4.1, 4.2, 4.3 have sendmail 8.8.4, 8.8.6, 8.8.8 or\n> 8.9.3. They all display the status in both SysV and BSD modes.\n> On AIX there is one ps command which handles both styles.\n> \n> IRIX 6.2 and 6.5.4m through 6.5.10m have sendmail 8.8.8, 8.9.1, 8.9.3.\n> No status is available. There do not appear to be any BSD-ish ps\n> options.\n\nThe way /usr/ucb/ps in Solaris extracts the status for a process requires\nroot rights and is (in my opinion) pretty gross. What it does is read the\npseudo-file /proc/<pid/as, which is actually the memory used by the\nprogram. /usr/bin/ps only accesses /proc/<pid>/psinfo, which contains the\nreal parameters supplied to the program and not the clobbered version.\n\nRegards,\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Fri, 30 Mar 2001 12:17:01 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: Re: [PORTS] pgmonitor and Solaris" } ]
[ { "msg_contents": "\n(I know that everyone is focusing on 7.1. This is a question about how\nhard it would be to plan a feature for 7.2 or later)\n\nOne of the small annoyances in PG is that I use many functions to handle\nsmall details, and these functions are called by views. If I want to\nimprove a function, I have to drop and recreate those views (and all views\nthat depend on them, etc.)\n\nIf I understand everything, this is because the function OID is used to\ncall the function, not its name.\n\nHow difficult would it be to either\n\n. allow creation of a function using a specific OID? (CREATE FUNCTION\nUSING OID xxxxx ...)\n\nor\n\n. add a command like ALTER FUNCTION foo(text) TO foo(text) returns text as\n....\n\nThen, (presumably?) the calling functions and views could just find the\nfunction by its old oid.\n\nThis would seem to require that the new function would take the same\nparameters (and return the same?) as the old function. This could be\nhandled either as a runtime check by the ALTER command, or it could be\nleft as caveat functioner.\n\nPerhaps this is tricky, or perhaps there's already a great solution. (If \nso, please tell!) If I'm making an idiot of myself, do let me know.\n\nOtherwise, is this feasible? I could try my rusty hand at C, but I'd need\nsomeone to give me a 2-paragraph game plan on where to start playing with\nthe code.\n\nThanks!\n\n\n(of course, this raises the same question for VIEWs... dropping and\ncreating them is a pain b/c of the same reasons...)\n\n\n-- \nJoel Burton <jburton@scw.org>\nDirector of Information Systems, Support Center of Washington\n\n", "msg_date": "Tue, 27 Mar 2001 19:38:59 -0500 (EST)", "msg_from": "Joel Burton <jburton@scw.org>", "msg_from_op": true, "msg_subject": "Feature Request: ALTER FUNCTION (or something like that)" }, { "msg_contents": "Joel Burton <jburton@scw.org> writes:\n> . add a command like ALTER FUNCTION foo(text) TO foo(text) returns text as\n> ....\n\nThis is on the TODO list already, I believe.\n\n> This would seem to require that the new function would take the same\n> parameters (and return the same?) as the old function.\n\nIf it doesn't take the same parameters then it's not the same function\nat all, so that part is a nonissue. We'd have to disallow change of\nreturn type as well.\n\n> Perhaps this is tricky,\n\nUpdating pg_proc wouldn't be hard. What's missing is a notification\nmechanism that would cause cached copies of the function to get\nreplaced. A crude first cut could just ignore that issue and still be\nextremely useful for development ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Mar 2001 19:47:47 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Feature Request: ALTER FUNCTION (or something like that) " }, { "msg_contents": "On Tue, 27 Mar 2001, Tom Lane wrote:\n\n> Joel Burton <jburton@scw.org> writes:\n> > . add a command like ALTER FUNCTION foo(text) TO foo(text) returns text as\n> > ....\n> \n> This is on the TODO list already, I believe.\n\nYikes. I should have read it more carefully. My apologies. There are so\nmany good things on it. Sadly, no one has claimed this item.\n\n(I still didn't see ALTER VIEW, though, which would seem just as nice.)\n\n> > This would seem to require that the new function would take the same\n> > parameters (and return the same?) as the old function.\n> \n> If it doesn't take the same parameters then it's not the same function\n> at all, so that part is a nonissue. We'd have to disallow change of\n> return type as well.\n> \n> > Perhaps this is tricky,\n> \n> Updating pg_proc wouldn't be hard. What's missing is a notification\n> mechanism that would cause cached copies of the function to get\n> replaced. A crude first cut could just ignore that issue and still be\n> extremely useful for development ...\n\nA crude first cut would be useful for development... even if I had to\nrestart postmaster, it's still much easier than dumping/restoring the\nwhole database, which is the usual solution around here (trying to\nindividually fix each view/table using the function becomes so tricky as\none has to track every dependency after that, etc.) Dumping/restoring can\nbe great (three cheers especially for the new formats using the -F\nswitch!), but in the past, I've had one or two complicated views that\nwould dump, but couldn't be restored from the dump. (As was discussed\nat the time in the list; the culprit was that pg_dump produced a\ndeeply complex expresion with a bazillion nested parentheses; if one\nremoved some of these, pg_restore did just fine.) This has disappeared\nin the 7.1 code, but I'm still slightly skitting about the dump-restore so\nvery frequently cycle.\n\nIf I have your ear on the subject, tgl, is there any ugly-but-working hack\nto update the function by modifying the system tables directly?\n\nAs always, thanks,\n-- \nJoel Burton <jburton@scw.org>\nDirector of Information Systems, Support Center of Washington\n\n", "msg_date": "Tue, 27 Mar 2001 22:55:45 -0500 (EST)", "msg_from": "Joel Burton <jburton@scw.org>", "msg_from_op": true, "msg_subject": "Re: Feature Request: ALTER FUNCTION (or something like that) " }, { "msg_contents": "Actually,\n\nThis reminds me of something I have been meaning to ask. All the new\nperformance features are cool, but what I really need are all the ALTER\nTABLE ... functions implemented. In 7.0.x you could only add columns and\nforeign keys. You couldn't drop anything or add stuff like CHECK\nconstraints. Has this situation changed for 7.1?\n\nRegards,\n\nChris\n\n-----Original Message-----\nFrom: pgsql-hackers-owner@postgresql.org\n[mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Joel Burton\nSent: Wednesday, 28 March 2001 8:39 AM\nTo: pgsql-hackers@postgresql.org\nSubject: [HACKERS] Feature Request: ALTER FUNCTION (or something like\nthat)\n\n\n\n(I know that everyone is focusing on 7.1. This is a question about how\nhard it would be to plan a feature for 7.2 or later)\n\nOne of the small annoyances in PG is that I use many functions to handle\nsmall details, and these functions are called by views. If I want to\nimprove a function, I have to drop and recreate those views (and all views\nthat depend on them, etc.)\n\nIf I understand everything, this is because the function OID is used to\ncall the function, not its name.\n\nHow difficult would it be to either\n\n. allow creation of a function using a specific OID? (CREATE FUNCTION\nUSING OID xxxxx ...)\n\nor\n\n. add a command like ALTER FUNCTION foo(text) TO foo(text) returns text as\n....\n\nThen, (presumably?) the calling functions and views could just find the\nfunction by its old oid.\n\nThis would seem to require that the new function would take the same\nparameters (and return the same?) as the old function. This could be\nhandled either as a runtime check by the ALTER command, or it could be\nleft as caveat functioner.\n\nPerhaps this is tricky, or perhaps there's already a great solution. (If\nso, please tell!) If I'm making an idiot of myself, do let me know.\n\nOtherwise, is this feasible? I could try my rusty hand at C, but I'd need\nsomeone to give me a 2-paragraph game plan on where to start playing with\nthe code.\n\nThanks!\n\n\n(of course, this raises the same question for VIEWs... dropping and\ncreating them is a pain b/c of the same reasons...)\n\n\n--\nJoel Burton <jburton@scw.org>\nDirector of Information Systems, Support Center of Washington\n\n\n---------------------------(end of broadcast)---------------------------\nTIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n\n", "msg_date": "Wed, 28 Mar 2001 13:00:41 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "RE: Feature Request: ALTER FUNCTION (or something like that)" }, { "msg_contents": "Joel Burton <jburton@scw.org> writes:\n> If I have your ear on the subject, tgl, is there any ugly-but-working hack\n> to update the function by modifying the system tables directly?\n\nFor interpreted function languages,\n\nUPDATE pg_proc SET prosrc = 'new body' WHERE proname = '...'\n\nwill work as long as the function name is unique. (If not, you'd need\nto also mention argument types in the WHERE.)\n\nAgain, this won't do anything to update cached copies, so backend\nrestarts might be needed to get the change to take effect.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 00:32:54 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Feature Request: ALTER FUNCTION (or something like that) " }, { "msg_contents": "> Actually,\n> \n> This reminds me of something I have been meaning to ask. All the new\n> performance features are cool, but what I really need are all the ALTER\n> TABLE ... functions implemented. In 7.0.x you could only add columns and\n> foreign keys. You couldn't drop anything or add stuff like CHECK\n> constraints. Has this situation changed for 7.1?\n> \n\nNot done in 7.1. I hope for 7.2.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 28 Mar 2001 07:20:09 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Feature Request: ALTER FUNCTION (or something like that)" }, { "msg_contents": "On Wed, 28 Mar 2001, Bruce Momjian wrote:\n\n> > Actually,\n> > \n> > This reminds me of something I have been meaning to ask. All the new\n> > performance features are cool, but what I really need are all the ALTER\n> > TABLE ... functions implemented. In 7.0.x you could only add columns and\n> > foreign keys. You couldn't drop anything or add stuff like CHECK\n> > constraints. Has this situation changed for 7.1?\n> > \n> \n> Not done in 7.1. I hope for 7.2.\n\nIf you're curious, Christopher, on the TODO list is a whole page on the\nDROP COLUMN bit. There's been some headscratching about how we want to\nDROP COLUMNs (whether to hide them but keep them there, which may be\neasier to undo, and probably easier to develop, and doesn't require any\ntime to process), or whether to actually recreate w/o the offending\ncolumn.\n\nIIRC, you can change whether a field can be NULL or not by twiddling a\nfield in pg_class.\n\n\n-- \nJoel Burton <jburton@scw.org>\nDirector of Information Systems, Support Center of Washington\n\n", "msg_date": "Wed, 28 Mar 2001 10:52:28 -0500 (EST)", "msg_from": "Joel Burton <jburton@scw.org>", "msg_from_op": true, "msg_subject": "Re: Feature Request: ALTER FUNCTION (or something like that)" }, { "msg_contents": "At 3/28/2001 09:52 AM, Joel Burton wrote:\n>On Wed, 28 Mar 2001, Bruce Momjian wrote:\n>\n> > > Actually,\n> > >\n> > > This reminds me of something I have been meaning to ask. All the new\n> > > performance features are cool, but what I really need are all the ALTER\n> > > TABLE ... functions implemented. In 7.0.x you could only add columns and\n> > > foreign keys. You couldn't drop anything or add stuff like CHECK\n> > > constraints. Has this situation changed for 7.1?\n> > >\n> >\n> > Not done in 7.1. I hope for 7.2.\n>\n>If you're curious, Christopher, on the TODO list is a whole page on the\n>DROP COLUMN bit. There's been some headscratching about how we want to\n>DROP COLUMNs (whether to hide them but keep them there, which may be\n>easier to undo, and probably easier to develop, and doesn't require any\n>time to process), or whether to actually recreate w/o the offending\n>column.\n\nYou could hide and then delete on a vacuum.\n\nOn a personal note, one \"work around\" for altering a table column is to \nrename the column, then do an update where you set the new columns value to \nthe renamed columns value thus copying its data over to the new column with \nany casting if necessary.\n\nIt doesn't always work, but it's one way to try.\n\nBut then you're left with the renamed column dangling...\n\nSelecting the table(with your preferred columns) into a new table can clean it\nDelete the old table and rename the new table to the old name...\n\nYou lose the constraints/triggers you had in place though...\n\nAnd you must pray that no one is altering the DB while your doing this...\n\n", "msg_date": "Wed, 28 Mar 2001 10:55:35 -0600", "msg_from": "Thomas Swan <tswan-lst@ics.olemiss.edu>", "msg_from_op": false, "msg_subject": "Re: Feature Request: ALTER FUNCTION (or something like that)" } ]
[ { "msg_contents": "set nomail\n\n", "msg_date": "Wed, 28 Mar 2001 13:24:45 +0800", "msg_from": "\"Brian Baquiran\" <brianb@edsamail.com>", "msg_from_op": true, "msg_subject": "" } ]
[ { "msg_contents": "Hi \n\nI am new to PostgreSQL, and I would like to install it on a Linux box. I wish to create and edit databases programmatically using Java. My problem is that I don't know where to get the drivers for PostgreSQL. I will be glad if you could refer me to a few sites that you know of.\n\nregards\n\ngilmour\n\nps: please send your response to grankoe@mail.co.za, since I will be leaving my current job at the end of this week.\n\n\n\n\n\nHi \n \nI am new to PostgreSQL, and I would like to install it on a Linux box. I \nwish to create and edit databases programmatically using Java. My problem is \nthat I don't know where to get the drivers for PostgreSQL. I will be glad if you \ncould refer me to a few sites that you know of.\n \nregards\n \ngilmour\n \nps: please send your response to grankoe@mail.co.za, since I will be leaving \nmy current job at the end of this week.", "msg_date": "Wed, 28 Mar 2001 09:48:18 +0200", "msg_from": "\"Gilmour Rankoe\" <GRankoe@csir.co.za>", "msg_from_op": true, "msg_subject": "drivers for postgresql" } ]
[ { "msg_contents": "Hi,\n\n\tI've been doing some tests with writing int8 to the database using\necpg. This does not work if the long long variable is in a structure, as\nit is not recognised as a simple type.\n\nThe patch is attached, but is not very satisfactory. Moving the\ndefinition of \n\n ECPGt_long_long, ECPGt_unsigned_long_long\n\nup to before ECPGt_varchar causes the resulting application to crash, as\nit finds an ECPGt_union, where it should haven found an ECPGt_EOIT (I\nthink), and I could not figure out where the static offset comes from.\n\nAnother problem with ecpg is that it dies on the alpha for long long\nvariables unless \n\n#define HAVE_LONG_LONG_INT_64\n\nis set in config.h. It is not set by default.\n\nWithout fixing these int8 values cannot be written to the database.\n\nRegards,\n\nAdriaan", "msg_date": "Wed, 28 Mar 2001 12:17:51 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "ecpg bug and patch" }, { "msg_contents": "Ooops, patch was the wrong way round. Here is a better one.\n\nSorry,\n\nAdriaan", "msg_date": "Wed, 28 Mar 2001 12:37:15 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Re: ecpg bug and patch" }, { "msg_contents": "On Wed, Mar 28, 2001 at 12:37:15PM +0300, Adriaan Joubert wrote:\n> Ooops, patch was the wrong way round. Here is a better one.\n\nApplied. I did add another set of braces for easier understanding though.\n:-)\n\nTo be able to move ECPGt_long_long up the list we would have to bump the\nsoname of the shared library as it would make an incompatible change. So\neveryone would have to recompile all their programs. I don't think would be\nworth the effort.\n\nMichael\n\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n", "msg_date": "Thu, 29 Mar 2001 12:01:41 +0200", "msg_from": "Michael Meskes <meskes@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: ecpg bug and patch" } ]
[ { "msg_contents": "Hi guys,\n\nI don't want to do a patch for a one character error. Yet, that's a HUGE\nerror and it really needs to be fixed.\n\nOnce you have a pg_database file which is over 1 page, it CRASHES. That's\nwhat it did on me. Yes! I have many databases or rather, I often destroy\nmy databases to regenerate them from scratch (that's for me the easiest\nway to do it). Because of that, the pg_database is now two pages.\n\nThe utils/misc/database.c has a function called GetRawDatabaseInfo()\nwhich reads that file \"on its own\". There is a HUGE bug in there, and\nit may not always crash a system, but it really needs to be fixed.\n\nAt line #183, you have a for() loop which looks like this (since at\nleast V6.5.0 and still present in V7.0.3):\n\n for (i = 0; i <= max; i++)\n\nAll the other such loops start with an index of 1, not zero. And\ntherefore you want the <=. In this special case loop (or are all\nthe others special cases?!?) you need to use the following:\n\n for(i = 0; i < max; i++)\n\nPlease, I know it's easier when you get a patch, but FIX IT. It's\nnot fun to try to access your database and have the backend crash\nbecause of such a tiny bug!\n\nThank you for all your work.\n\n\n\nAlexis Wilke\nDirector\nMade to Order Software, Ltd\n\ne-mail: alexis@m2sow.com\n\nWeb Page: http://www.m2osw.com\nCompany e-mail: contact@m2osw.com\nPhone: 020 8748 9898 +(44) 20 8748 9898\nFax: 020 8748 4250 +(44) 20 8748 4250\nAddress: Britannia House\n 1-11 Glenthorne Road\n Hammersmith\n London W6 0LF\n United Kingdom\n\n\n__________________________________________________\nDo You Yahoo!?\nGet email at your own domain with Yahoo! Mail. \nhttp://personal.mail.yahoo.com/?.refer=text\n", "msg_date": "Wed, 28 Mar 2001 02:28:45 -0800 (PST)", "msg_from": "Alexis Wilke <alexis_wilke@yahoo.com>", "msg_from_op": true, "msg_subject": "HUGE BUG - Please fix!!!" }, { "msg_contents": "\nThis is fixed in current 7.1RC1 sources.\n\n> Hi guys,\n> \n> I don't want to do a patch for a one character error. Yet, that's a HUGE\n> error and it really needs to be fixed.\n> \n> Once you have a pg_database file which is over 1 page, it CRASHES. That's\n> what it did on me. Yes! I have many databases or rather, I often destroy\n> my databases to regenerate them from scratch (that's for me the easiest\n> way to do it). Because of that, the pg_database is now two pages.\n> \n> The utils/misc/database.c has a function called GetRawDatabaseInfo()\n> which reads that file \"on its own\". There is a HUGE bug in there, and\n> it may not always crash a system, but it really needs to be fixed.\n> \n> At line #183, you have a for() loop which looks like this (since at\n> least V6.5.0 and still present in V7.0.3):\n> \n> for (i = 0; i <= max; i++)\n> \n> All the other such loops start with an index of 1, not zero. And\n> therefore you want the <=. In this special case loop (or are all\n> the others special cases?!?) you need to use the following:\n> \n> for(i = 0; i < max; i++)\n> \n> Please, I know it's easier when you get a patch, but FIX IT. It's\n> not fun to try to access your database and have the backend crash\n> because of such a tiny bug!\n> \n> Thank you for all your work.\n> \n> \n> \n> Alexis Wilke\n> Director\n> Made to Order Software, Ltd\n> \n> e-mail: alexis@m2sow.com\n> \n> Web Page: http://www.m2osw.com\n> Company e-mail: contact@m2osw.com\n> Phone: 020 8748 9898 +(44) 20 8748 9898\n> Fax: 020 8748 4250 +(44) 20 8748 4250\n> Address: Britannia House\n> 1-11 Glenthorne Road\n> Hammersmith\n> London W6 0LF\n> United Kingdom\n> \n> \n> __________________________________________________\n> Do You Yahoo!?\n> Get email at your own domain with Yahoo! Mail. \n> http://personal.mail.yahoo.com/?.refer=text\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 3: if posting/reading through Usenet, please send an appropriate\n> subscribe-nomail command to majordomo@postgresql.org so that your\n> message can get through to the mailing list cleanly\n> \n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 30 Mar 2001 10:15:28 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: HUGE BUG - Please fix!!!" } ]
[ { "msg_contents": "> > I don't think that we dare try to make any basic changes in \n> > MVCC for 7.1 at this late hour, so Forest is going to have\n> > to live with that answer for awhile. But I would like to see\n> > a cleaner answer in future releases.\n> \n> Is it the MVCC's restriction that each query inside a function\n> must use the same snapshot ?\n\nNo. MVCC restricts what is visible to query itself. Current\nfunction' behaviour is like Oracle' one.\nStrictly speaking queries inside function don't use the same\nsnapshot - they see changes made by other queries of this\nfunction. Should we allow them see changes made by other\ntransactions? I'm not sure. Maybe by special means like\nCREATE SNAPSHOT S;\nSELECT FROM foo WITH SNAPSHOT S;\n?\n\nFor this particular case - concurrent UPDATE then\nUPDATE/DELETE + SELECT - there is simple solution: meeting\ntuple updated by concurrent *committed* transaction\nSELECT (in READ COMMITTED mode) should look in newer tuple\nversions. If some of newer tuples is invalidated (updated/deleted)\nby *this* transaction and this invalidation is *visible*\nto SELECT (older CommandId) then old tuple version must not\nbe returned (newer tuple version will be returned of course).\nReported problem is caused by bug (only one tuple version must be\nreturned by SELECT) and this is way to fix it.\n\nBut note that for the case of concurrent DELETE then\nINSERT + SELECT two tuples will be returned anyway and\nI don't think that this is bug.\n\n> > As I've opined before, the whole EvalPlanQual mechanism\n> > strikes me as essentially bogus in any case...\n> \n> How would you change it ? UPDATE/SELECT FOR UPDATE have to\n> SELECT/UPDATE the latest tuples. I don't think of any simple\n> way for 'SELECT FOR UPDATE' to have the same visibility as\n> simple SELECT.\n\nYes, I also don't understand what's wrong with EvalPlanQual.\n\nVadim\n", "msg_date": "Wed, 28 Mar 2001 09:58:43 -0800", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" } ]
[ { "msg_contents": "> Looking at the docs, I see that 'SERIALIZABLE' has the same visibility\n> rules as 'READ COMMITTED', which is very confusing. I expect \n\nHm, you're right:\nhttp://www.postgresql.org/devel-corner/docs/postgres/xact-read-committed.htm\nl\n\n\"Read Committed is the default isolation level in Postgres. When\na transaction runs on this isolation level, a SELECT query sees only\ndata committed before the transaction began...\"\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nMust be \"committed before the *query* began\" as it was in 6.5 docs.\nAny way to fix it before release?\n\nVadim\n", "msg_date": "Wed, 28 Mar 2001 10:10:02 -0800", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: Re: [SQL] possible row locking bug in 7.0.3 & 7.1 " }, { "msg_contents": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM> writes:\n> Hm, you're right:\n> http://www.postgresql.org/devel-corner/docs/postgres/xact-read-committed.html\n\n> \"Read Committed is the default isolation level in Postgres. When\n> a transaction runs on this isolation level, a SELECT query sees only\n> data committed before the transaction began...\"\n> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n> Must be \"committed before the *query* began\" as it was in 6.5 docs.\n\nWill fix this shortly...\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Mar 2001 15:24:29 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Re: [SQL] possible row locking bug in 7.0.3 & 7.1 " } ]
[ { "msg_contents": "Hi all,\n\nThe OpenSSL guys are working on a solution to the conflicting\ndes_encrypt() function names on Solaris and Unixware.\n\nWhilst this bodes well for future versions, how are we going to work\naround conflicts with existing versions?\n\nRegards and best wishes,\n\nJustin Clift\n\n-------- Original Message --------\nSubject: Delayed release of OpenSSL 0.9.6a beta 3\nDate: Wed, 28 Mar 2001 15:02:15 +0200 (MET DST)\nFrom: Richard Levitte - VMS Whacker <levitte@stacken.kth.se>\nReply-To: openssl-dev@openssl.org\nTo: openssl-dev@openssl.org,\nopenssl-users@openssl.org,openssl-announce@openssl.org\n\nThe release of OpenSSL 0.9.6a beta3 and the true release have been\ndelayed. The reason for the delay is mainly due to the following\npaper and the need to add a fix for the problem it describes:\n\nhttp://crypto.stanford.edu/~dabo/abstracts/faults.html\n\nThe problem is now fixed in the current source tree. At the same\ntime, we're trying to solve the last small known problem (the reported\nclash with des_encrypt() on some systems), to make sure that beta 3\nhas a fair chance of becoming the actual release.\n\nThanks for your patience.\n\n-- \nRichard Levitte \\ Spannv�gen 38, II \\ LeViMS@stacken.kth.se\nChairman@Stacken \\ S-168 35 BROMMA \\ T: +46-8-26 52 47\nRedakteur@Stacken \\ SWEDEN \\ or +46-709-50 36 10\nProcurator Odiosus Ex Infernis -- poei@bofh.se\nMember of the OpenSSL development team: http://www.openssl.org/\nSoftware Engineer, Celo Communications: http://www.celocom.com/\n\nUnsolicited commercial email is subject to an archival fee of $400.\nSee <http://www.stacken.kth.se/~levitte/mail/> for more info.\n______________________________________________________________________\nOpenSSL Project http://www.openssl.org\nDevelopment Mailing List openssl-dev@openssl.org\nAutomated List Manager majordomo@openssl.org\n", "msg_date": "Thu, 29 Mar 2001 11:46:55 +1000", "msg_from": "Justin Clift <jclift@iprimus.com.au>", "msg_from_op": true, "msg_subject": "[Fwd: Delayed release of OpenSSL 0.9.6a beta 3]" } ]
[ { "msg_contents": "Subject: Importing data from Informix to PostgreSQL.\n Error in the date field (WITH NULL value)\n\nHello!\nI'll try to explain my little problem. Well, I have this table\n\ncreate table mytable\n ( codice char(16) not null,\n dt_inizio date,\n dt_fine date,\n tipo_operazione char(1),\n causa_operazione integer\n );\n\nok!\n\nI find out that pgsql:\n\nIn my example '' is the NULL value exported from Informix... ! (an ASCII file)\n\n INFORMIX PostgreSQL\nchar(16) '' ------> blank \nstring (I think it's ok! )\nchar(1) '' ------> blank \nstring (I think it's ok)\ninteger '' ------> 0 (is \nit an error? )\ndate '' ------> ERROR! Bad date \nexternal representation ''\n\n\n >> select * from mytable ;\n\n codice | dt_inizio | dt_fine | \ntipo_operazione | causa_operazione\n\n ABCEDEFFFFFFFFF | \n2001-03-28 | | | 0\n XXXYYYAAA23C957Y | \n2001-03-28 | | | 0\n\n\nclinica=# insert into mytable values ( '','03/28/2001', '' , '' , '' );\nERROR: Bad date external representation '' ^^^\n\nPostgreSQL doesn't want '' as an input of a date with NULL value:\nit's necessary to use this kind of insert:\n\n >> insert into mytable values ( '','03/28/2001',null,'','');\n ^^^^^^\n\nNow there is a new line in the table:\n\n | \n2001-03-28 | | | 0\n\n\nMy question:\nHow can I resolv my problem? I have a big data file to import where\nin the 2nd date field there is '' instead of null ....\n\nHow can I \"binds\" PostgreSQL to consider '' as null ?\n\nMany thanks for any suggestions!\n\nCIAO!\nMAURIZIO\n\n*******************************************\n** Happy surfing on THE NET !! **\n** Ciao by **\n** C R I X 98 **\n*******************************************\nAntiSpam: rimuovere il trattino basso\n dall'indirizzo per scrivermi...\n(delete the underscore from the e-mail address to reply)\n\n", "msg_date": "Wed, 28 Mar 2001 20:50:56 -0800", "msg_from": "Maurizio Ortolan <crix98__@tin.it>", "msg_from_op": true, "msg_subject": "Error in the date field (with NULL value...).Thanks!" }, { "msg_contents": "> How can I \"binds\" PostgreSQL to consider '' as null ?\n\nYou can modify src/backend/utils/{datetime,date,timestamp}.c to accept\nan empty string as a null value. But imho it is better to fix this in\nyour input file, perhaps using sed:\n\n sed \"s/''/NULL/g\" < informix.dump > pg.dump\n\n - Thomas\n", "msg_date": "Fri, 30 Mar 2001 06:22:21 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": false, "msg_subject": "Re: [PATCHES] Error in the date field (with NULL value...).Thanks!" }, { "msg_contents": "Just do a search-replace on your source file and replace all occurrences of\n'' with NULL.\n\nChris\n\n-----Original Message-----\nFrom: pgsql-hackers-owner@postgresql.org\n[mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Maurizio Ortolan\nSent: Thursday, 29 March 2001 12:51 PM\nTo: pgsql-hackers@postgresql.org; pgsql-sql@postgresql.org;\npgsql-general@postgresql.org; pgsql-ports@postgresql.org;\npgsql-admin@postgresql.org; pgsql-bugs@postgresql.org;\npgsql-patches@postgresql.org\nSubject: [HACKERS] Error in the date field (with NULL value...).Thanks!\n\n\nSubject: Importing data from Informix to PostgreSQL.\n Error in the date field (WITH NULL value)\n\nHello!\nI'll try to explain my little problem. Well, I have this table\n\ncreate table mytable\n ( codice char(16) not null,\n dt_inizio date,\n dt_fine date,\n tipo_operazione char(1),\n causa_operazione integer\n );\n\nok!\n\nI find out that pgsql:\n\nIn my example '' is the NULL value exported from Informix... ! (an ASCII\nfile)\n\n INFORMIX PostgreSQL\nchar(16) '' ------> blank\nstring (I think it's ok! )\nchar(1) '' ------> blank\nstring (I think it's ok)\ninteger '' ------> 0 (is\nit an error? )\ndate '' ------> ERROR! Bad date\nexternal representation ''\n\n\n >> select * from mytable ;\n\n codice | dt_inizio | dt_fine |\ntipo_operazione | causa_operazione\n\n ABCEDEFFFFFFFFF |\n2001-03-28 | | | 0\n XXXYYYAAA23C957Y |\n2001-03-28 | | | 0\n\n\nclinica=# insert into mytable values ( '','03/28/2001', '' , '' , '' );\nERROR: Bad date external representation '' ^^^\n\nPostgreSQL doesn't want '' as an input of a date with NULL value:\nit's necessary to use this kind of insert:\n\n >> insert into mytable values ( '','03/28/2001',null,'','');\n ^^^^^^\n\nNow there is a new line in the table:\n\n |\n2001-03-28 | | | 0\n\n\nMy question:\nHow can I resolv my problem? I have a big data file to import where\nin the 2nd date field there is '' instead of null ....\n\nHow can I \"binds\" PostgreSQL to consider '' as null ?\n\nMany thanks for any suggestions!\n\nCIAO!\nMAURIZIO\n\n*******************************************\n** Happy surfing on THE NET !! **\n** Ciao by **\n** C R I X 98 **\n*******************************************\nAntiSpam: rimuovere il trattino basso\n dall'indirizzo per scrivermi...\n(delete the underscore from the e-mail address to reply)\n\n\n---------------------------(end of broadcast)---------------------------\nTIP 3: if posting/reading through Usenet, please send an appropriate\nsubscribe-nomail command to majordomo@postgresql.org so that your\nmessage can get through to the mailing list cleanly\n\n", "msg_date": "Mon, 2 Apr 2001 09:31:43 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "RE: [HACKERS] Error in the date field (with NULL value...).Thanks!" } ]
[ { "msg_contents": "hi,\n\nI come back in office after a long period out for work.\nYesterday I have downloaded the last snapshot.\n\nWhen I execute configure I have an error compiling conftest.\nThe last version I have checked was 7.1 b3 and all works fine.\n\nAttached is the config.log file.\nHas someone any suggestion ?\n\nThanks\n\n\n\nMaurizio Cauci\nDREAMTECH di Cauci Maurizio\nVia Ronchetti, 2 - 21013 Gallarate (VA)\nwww.dreamtech-it.com", "msg_date": "Thu, 29 Mar 2001 10:53:53 +0200", "msg_from": "\"Maurizio\" <maurizio.c@libero.it>", "msg_from_op": true, "msg_subject": "testing last sanpshot in QNX platform" } ]
[ { "msg_contents": "Hi,\n\n\tI've got the weird behaviour in an ecpg programme that when I issue\n\n EXEC SQL SET AUTOCOMMIT TO ON;\n\ntransactions behave as one would expect, i.e. I can use BEGIN\nWORK/COMMIT/ROLLBACK, but when I set \n\n\n EXEC SQL SET AUTOCOMMIT TO OFF;\n\nevery statement is wrapped in its own transaction, whether I have issued\na BEGIN WORK or not. This seems a tad counter-intuitive. Previous\ndiscussions on the mailing list seem to suggest that it should be the\nother way round? \n\nThis is on 7.1RC1 on Linux.\n\nCheers,\n\nAdriaan\n", "msg_date": "Thu, 29 Mar 2001 11:56:44 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "ecpg autocommit weirdness" } ]
[ { "msg_contents": "Hi there,\n\nI have found a small but annoying bug. I have created a view. The\nSQL-statement in the view contains a GROUP BY statement. Then I compose a\nSQL-statement using this view and another GROUP BY statement and a COUNT(*)\nstatement. The count(*) statement doesn't count the amount of grouped\nrecord's of the view, but it count's the amount of grouped records of the\nGROUP BY in the view and of the GROUP BY in the select statement. It counts\nall the records grouped instead of only the records grouped from the view.\nThis is wrong (IMHO). When I use a temporary table instead of a view all\nthings work OK. IMHO views shouldn't differ from temporary tables.\n\nTo make things a bit more clear I have add an SQL-attachment. You can run\nthe attachment in an empty database form psql to have a look at the problem.\n\nI sometimes get another <<database.sql>> error too while executing these\ngroup statements: 'My bit's blew right of the end of the world'. (This is\nwhen i am using an ODBC link to my db.)\n\nMighty thanks in advance,\n\nJeroen Eitjes\nj.eitjes<nospam>@chem.leidenuniv.nl\neitjes<nospam>@walras.nl", "msg_date": "Thu, 29 Mar 2001 15:30:37 +0200", "msg_from": "Jeroen Eitjes <Eitjes@WalRas.nl>", "msg_from_op": true, "msg_subject": "Problem with group by in conjuction with Views" }, { "msg_contents": "Jeroen Eitjes <Eitjes@WalRas.nl> writes:\n> I have found a small but annoying bug. I have created a view. The\n> SQL-statement in the view contains a GROUP BY statement. Then I compose a\n> SQL-statement using this view and another GROUP BY statement and a COUNT(*)\n> statement.\n\nThis sort of thing will not work at all in Postgres releases earlier\nthan 7.1. Sorry.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 29 Mar 2001 11:13:03 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Problem with group by in conjuction with Views " } ]
[ { "msg_contents": "This seems to work for me. I used the snapshot from 3/28 on Solaris 8\n\nSELECT service, count(*) AS GebruikersAantal\nFROM tbtrouble GROUP BY service;\n service | gebruikersaantal \n-----------+------------------\n Service 1 | 2\n Service 3 | 2\n Service 4 | 1\n(3 rows)\n\nSELECT service, count(*) AS GebruikersAantal\nFROM vwtrouble GROUP BY service;\n service | gebruikersaantal \n-----------+------------------\n Service 1 | 2\n Service 3 | 2\n Service 4 | 1\n(3 rows)\n\n\n\n> This message is in MIME format. Since your mail reader does not\nunderstand\n> this format, some or all of this message may not be legible.\n> \n> \n> Hi there,\n> \n> I have found a small but annoying bug. I have created a view. The\n> SQL-statement in the view contains a GROUP BY statement. Then I\ncompose a\n> SQL-statement using this view and another GROUP BY statement and a\nCOUNT(*)\n> statement. The count(*) statement doesn't count the amount of grouped\n> record's of the view, but it count's the amount of grouped records of\nthe\n> GROUP BY in the view and of the GROUP BY in the select statement. It\ncounts\n> all the records grouped instead of only the records grouped from the\nview.\n> This is wrong (IMHO). When I use a temporary table instead of a view\nall\n> things work OK. IMHO views shouldn't differ from temporary tables.\n> \n> To make things a bit more clear I have add an SQL-attachment. You can\nrun\n> the attachment in an empty database form psql to have a look at the\nproblem.\n> \n> I sometimes get another <<database.sql>> error too while executing\nthese\n> group statements: 'My bit's blew right of the end of the world'. (This\nis\n> when i am using an ODBC link to my db.)\n> \n> Mighty thanks in advance,\n> \n> Jeroen Eitjes\n> j.eitjes<nospam>@chem.leidenuniv.nl\n> eitjes<nospam>@walras.nl\n> \n> \n\n\n", "msg_date": "Thu, 29 Mar 2001 10:00:26 -0500 (EST)", "msg_from": "\"Jim Buttafuoco\" <jim@/etc/mail/ok>", "msg_from_op": true, "msg_subject": "Re: Problem with group by in conjuction with Views" } ]
[ { "msg_contents": "\nINSERT INTO t1 SELECT a FROM t2 ORDER BY a LIMIT 10;\n\nIs it possible in 7.1?\n\nDirk\n", "msg_date": "Thu, 29 Mar 2001 17:31:42 +0200 (CEST)", "msg_from": "Dirk Lutzebaeck <lutzeb@aeccom.com>", "msg_from_op": true, "msg_subject": "INSERT/SELECT with ORDER BY and LIMIT in 7.1?" }, { "msg_contents": "Dirk Lutzebaeck wrote:\n >\n >INSERT INTO t1 SELECT a FROM t2 ORDER BY a LIMIT 10;\n >\n >Is it possible in 7.1?\n \nYes (I tried it).\n\n-- \nOliver Elphick Oliver.Elphick@lfix.co.uk\nIsle of Wight http://www.lfix.co.uk/oliver\nPGP: 1024R/32B8FAA1: 97 EA 1D 47 72 3F 28 47 6B 7E 39 CC 56 E4 C1 47\nGPG: 1024D/3E1D0C1C: CA12 09E0 E8D5 8870 5839 932A 614D 4C34 3E1D 0C1C\n ========================================\n \"Trust in the Lord with all your heart and lean not on \n your own understanding; in all your ways acknowledge \n him, and he will direct your paths.\" Proverbs 3:5,6 \n\n\n", "msg_date": "Thu, 29 Mar 2001 22:47:13 +0100", "msg_from": "\"Oliver Elphick\" <olly@lfix.co.uk>", "msg_from_op": false, "msg_subject": "Re: INSERT/SELECT with ORDER BY and LIMIT in 7.1? " } ]
[ { "msg_contents": "I'm at the moment installing Postgresql-7.1RC1 on Solaris 7 and 8 over \nUltraSPARC.\nWhat do I have to do to report a regression test?\nWhere should I look for info on this.\n\nSaludos... :-)\n\n-- \nEl mejor sistema operativo es aquel que te da de comer.\nCuida tu dieta.\n-----------------------------------------------------------------\nMartin Marques | mmarques@unl.edu.ar\nProgramador, Administrador | Centro de Telematica\n Universidad Nacional\n del Litoral\n-----------------------------------------------------------------\n", "msg_date": "Thu, 29 Mar 2001 18:58:56 +0300", "msg_from": "=?iso-8859-1?q?Mart=EDn=20Marqu=E9s?= <martin@bugs.unl.edu.ar>", "msg_from_op": true, "msg_subject": "regression test" } ]
[ { "msg_contents": "Hi,\n\n\tI finally seem to have my unsigned int2/int4 types working correctly,\nbut will wait until 7.1 is out of the door, and test a bit more, before\nresubmitting.\n\nA question though: \n\nI've put in functions (as copied from the int2/int4 implementation) that\nimplement operators for differently typed arguments, e.g. uint2*uint4.\nThis saves the type conversions, but adds to the number of functions in\nthe system.\n\nWhen sorting out the constant problems, I realised that (uint2,uint4)\ncombinations will probably be very rarely used, while (int4,uint4)\ncombinations will be much more common, i.e. when there are constants\ninvolved. \n\nQuestion is: should I add these functions? Are we looking at too much\nbloat, i.e. should I replace the (uint2,uint4) combinations with\n(int4,uint2) and (int4,uint4)? Lots of combinations are possible, but I\ndo not have a good feel for the trade-offs. \n\nI only wanted unsigned ints, so that we could develop and test stuff on\npostgres before moving it onto Tandem. So please let me know what you\nthink the correct trade-offs are and I will implement it and resubmit\nthe patch.\n\nCheers,\n\nAdriaan\n", "msg_date": "Thu, 29 Mar 2001 22:16:09 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Unsigned int functions" }, { "msg_contents": "Adriaan Joubert <a.joubert@albourne.com> writes:\n> Question is: should I add these functions? Are we looking at too much\n> bloat, i.e. should I replace the (uint2,uint4) combinations with\n> (int4,uint2) and (int4,uint4)? Lots of combinations are possible, but I\n> do not have a good feel for the trade-offs. \n\nMy guess is that we ought to avoid bloating the system with\ncross-datatype functions. I know there are some already for int2*int4\nand so forth, but I'd like to see those go away in favor of a smarter\ntype promotion scheme --- ie, the parser should be able to figure out\nthat it ought to do int2_var * uint4_var as\n\tuint4_mul(uint4(int2_var), uint4_var)\nA cross-datatype function ought to exist only if it can usefully do\nsomething different from an implicit promotion.\n\nAside from bloating the system, providing a plethora of functions also\ntends to confuse the ambiguous-function-call resolution mechanism.\nSee discussion of a few days ago wherein the parser could resolve an\nambiguous situation involving varchar, but could not resolve the same\nsituation with text, because there are too many possibilities for\ncoercion of text to something else.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 29 Mar 2001 14:51:40 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Unsigned int functions " }, { "msg_contents": "> Adriaan Joubert <a.joubert@albourne.com> writes:\n> > Question is: should I add these functions? Are we looking at too much\n> > bloat, i.e. should I replace the (uint2,uint4) combinations with\n> > (int4,uint2) and (int4,uint4)? Lots of combinations are possible, but I\n> > do not have a good feel for the trade-offs. \n> \n> My guess is that we ought to avoid bloating the system with\n> cross-datatype functions. I know there are some already for int2*int4\n> and so forth, but I'd like to see those go away in favor of a smarter\n> type promotion scheme --- ie, the parser should be able to figure out\n> that it ought to do int2_var * uint4_var as\n> \tuint4_mul(uint4(int2_var), uint4_var)\n> A cross-datatype function ought to exist only if it can usefully do\n> something different from an implicit promotion.\n\nA larger question is whether unsigned types really add much to the\nsystem vs. the bloat. We already have unsigned int4 as oid. Also,\nunsigned doubles the space of the type, but if a value doesn't fit in\n32k, what are the odds it will fit in 64k. I am not sure unsigned\noptimzations for space really are significant in SQL.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 29 Mar 2001 15:03:27 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Unsigned int functions" }, { "msg_contents": "Bruce Momjian wrote:\n> \n> > Adriaan Joubert <a.joubert@albourne.com> writes:\n> > > Question is: should I add these functions? Are we looking at too much\n> > > bloat, i.e. should I replace the (uint2,uint4) combinations with\n> > > (int4,uint2) and (int4,uint4)? Lots of combinations are possible, but I\n> > > do not have a good feel for the trade-offs.\n> >\n> > My guess is that we ought to avoid bloating the system with\n> > cross-datatype functions. I know there are some already for int2*int4\n> > and so forth, but I'd like to see those go away in favor of a smarter\n> > type promotion scheme --- ie, the parser should be able to figure out\n> > that it ought to do int2_var * uint4_var as\n> > uint4_mul(uint4(int2_var), uint4_var)\n> > A cross-datatype function ought to exist only if it can usefully do\n> > something different from an implicit promotion.\n> \n> A larger question is whether unsigned types really add much to the\n> system vs. the bloat. We already have unsigned int4 as oid. Also,\n> unsigned doubles the space of the type, but if a value doesn't fit in\n> 32k, what are the odds it will fit in 64k. I am not sure unsigned\n> optimzations for space really are significant in SQL.\n\nA fair question. As I said, I only implemented them to simplify porting\napplications between database systems. Personally I think it is good to\nsupport types that make porting easier.\n\nOn the other hand the arguments about bloat are strong. It seems to me\nthat all cross-datatype functions should be removed, to reduce the\nnumber of functions for the unsigned data types to a minimum. \n\nWould this be a reasonable compromise? \n\nIf general opinion is that unsigned types should not be part of\npostgres, I'll have to look at turning them into a contrib type. Please\nlet me know.\n\nCheers,\n\nAdriaan\n", "msg_date": "Fri, 30 Mar 2001 08:36:42 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Re: Unsigned int functions" }, { "msg_contents": "> If general opinion is that unsigned types should not be part of\n> postgres, I'll have to look at turning them into a contrib type. Please\n> let me know.\n\nProviding them as a contrib/ package will allow you to provide the\n*full* complement of cross-type conversion and operator functions\nwithout worrying about bloat. Tom Lane has shown how to use entry points\non shared libraries to start us thinking about how to provide better\npackage integration, which should allow us to reduce the distinction\nbetween contrib/ and standard features.\n\nThis would be a great package to develop these additional package\nsupport features for 7.2 (hint hint ;)\n\n - Thomas\n", "msg_date": "Fri, 30 Mar 2001 06:17:31 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": false, "msg_subject": "Re: Unsigned int functions" } ]
[ { "msg_contents": "\nJust a heads up for anyone that might have something outstanding ... I'm\ngoing to package her early evening (~18:30AST) and announce it to both\npgsql-hackers and pgsql-announce when done ...\n\nOnce RC2 goes out, its meant to be a \"this is what we'd release if docs\nwere completely ready to go\" ...\n\n\nMarc G. Fournier ICQ#7615664 IRC Nick: Scrappy\nSystems Administrator @ hub.org\nprimary: scrappy@hub.org secondary: scrappy@{freebsd|postgresql}.org\n\n", "msg_date": "Thu, 29 Mar 2001 16:21:58 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "RC2 schedualed for Tomorrow evening ..." }, { "msg_contents": "The Hermit Hacker writes:\n\n> Just a heads up for anyone that might have something outstanding ... I'm\n> going to package her early evening (~18:30AST) and announce it to both\n> pgsql-hackers and pgsql-announce when done ...\n>\n> Once RC2 goes out, its meant to be a \"this is what we'd release if docs\n> were completely ready to go\" ...\n\nAFAICT, we need to close out the platform list, re-generate INSTALL, and\nthat's it. Bruce mentioned something about additional changes in HISTORY,\nthough.\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Fri, 30 Mar 2001 18:56:17 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: RC2 schedualed for Tomorrow evening ..." }, { "msg_contents": "On Fri, 30 Mar 2001, Peter Eisentraut wrote:\n\n> The Hermit Hacker writes:\n>\n> > Just a heads up for anyone that might have something outstanding ... I'm\n> > going to package her early evening (~18:30AST) and announce it to both\n> > pgsql-hackers and pgsql-announce when done ...\n> >\n> > Once RC2 goes out, its meant to be a \"this is what we'd release if docs\n> > were completely ready to go\" ...\n>\n> AFAICT, we need to close out the platform list, re-generate INSTALL, and\n> that's it. Bruce mentioned something about additional changes in HISTORY,\n> though.\n\nright, that falls under the \"this is what we'd release if docs were\ncompletely ready to go\" :)\n\n\n", "msg_date": "Fri, 30 Mar 2001 13:46:05 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: RC2 schedualed for Tomorrow evening ..." }, { "msg_contents": "> The Hermit Hacker writes:\n> \n> > Just a heads up for anyone that might have something outstanding ... I'm\n> > going to package her early evening (~18:30AST) and announce it to both\n> > pgsql-hackers and pgsql-announce when done ...\n> >\n> > Once RC2 goes out, its meant to be a \"this is what we'd release if docs\n> > were completely ready to go\" ...\n> \n> AFAICT, we need to close out the platform list, re-generate INSTALL, and\n> that's it. Bruce mentioned something about additional changes in HISTORY,\n> though.\n\nI just need to grovel through the CVS logs again to update them. I will\ndo that today.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 30 Mar 2001 13:44:49 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: RC2 schedualed for Tomorrow evening ..." } ]
[ { "msg_contents": "At 09:58 28/03/01 -0800, Mikheev, Vadim wrote:\n>\n>Reported problem is caused by bug (only one tuple version must be\n>returned by SELECT) and this is way to fix it.\n>\n\nI assume this is not possible in 7.1?\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Fri, 30 Mar 2001 12:49:11 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": true, "msg_subject": "RE: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" } ]
[ { "msg_contents": "[ Replies set to hackers.]\n\nI have started coding a user/group administration tool that allows you\nto add/modify/delete users and groups. I should have something working\nin a week. I will look similar to my pgmonitor tool.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 29 Mar 2001 22:12:45 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "User administration tool" }, { "msg_contents": "Bruce Momjian writes:\n\n> I have started coding a user/group administration tool that allows you\n> to add/modify/delete users and groups. I should have something working\n> in a week. I will look similar to my pgmonitor tool.\n\nPgaccess already does part of this. If you're going to write it in Tcl/Tk\nanyway, I think you might as well integrate it there.\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Fri, 30 Mar 2001 17:24:30 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: [ADMIN] User administration tool" }, { "msg_contents": "> Bruce Momjian writes:\n> \n> > I have started coding a user/group administration tool that allows you\n> > to add/modify/delete users and groups. I should have something working\n> > in a week. I will look similar to my pgmonitor tool.\n> \n> Pgaccess already does part of this. If you're going to write it in Tcl/Tk\n> anyway, I think you might as well integrate it there.\n\nWow, I see. I never suspected it did that too. :-) Seems I don't need\nto write anything, except perhaps add group capabilities to pgaccess.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 30 Mar 2001 10:48:54 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: User administration tool" }, { "msg_contents": "On Fri, Mar 30, 2001 at 10:48:54AM -0500, Bruce Momjian wrote:\n> > Bruce Momjian writes:\n> > \n> > > I have started coding a user/group administration tool that allows you\n> > > to add/modify/delete users and groups. I should have something working\n> > > in a week. I will look similar to my pgmonitor tool.\n> > \n> > Pgaccess already does part of this. If you're going to write it in Tcl/Tk\n> > anyway, I think you might as well integrate it there.\n> \n> Wow, I see. I never suspected it did that too. :-) Seems I don't need\n> to write anything, except perhaps add group capabilities to pgaccess.\n\nIsn't phpPgAdmin yet another tool of this type? I haven't tried it myself,\n(no need, myself being the only user...) but the web page\n(http://www.greatbridge.org/project/phppgadmin/projdisplay.php) says:\n\n Features include: \n\n * create and drop databases \n * create, copy, drop and alter\n tables/views/sequences/functions/indicies/triggers \n * edit and add fields (to the extent Postgres allows) \n * execute any SQL-statement, even batch-queries \n * manage primary and unique keys \n * create and read dumps of tables \n * administer one single database \n * administer multiple servers \n * administer postgres users and groups \n\nGreetinx, Jan\n-- \n +- Jan T. Kim -------------------------------------------------------+\n | *NEW* --> email: kim@inb.mu-luebeck.de |\n | *NEW* --> WWW: http://www.inb.mu-luebeck.de/staff/kim.html |\n *-----=< hierarchical systems are for files, not for humans >=-----*\n", "msg_date": "Fri, 30 Mar 2001 21:20:49 +0200", "msg_from": "\"Jan T. Kim\" <kim@inb.mu-luebeck.de>", "msg_from_op": false, "msg_subject": "Re: [ADMIN] User administration tool" }, { "msg_contents": "> > Wow, I see. I never suspected it did that too. :-) Seems I don't need\n> > to write anything, except perhaps add group capabilities to pgaccess.\n> \n> Isn't phpPgAdmin yet another tool of this type? I haven't tried it myself,\n> (no need, myself being the only user...) but the web page\n> (http://www.greatbridge.org/project/phppgadmin/projdisplay.php) says:\n\nYes, it is. Some people prefer a non-web interaface.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 30 Mar 2001 14:23:31 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [ADMIN] User administration tool" }, { "msg_contents": "\n<moved to -interfaces, where it belongs ... >\n\nOn Fri, 30 Mar 2001, Bruce Momjian wrote:\n\n> > > Wow, I see. I never suspected it did that too. :-) Seems I don't need\n> > > to write anything, except perhaps add group capabilities to pgaccess.\n> >\n> > Isn't phpPgAdmin yet another tool of this type? I haven't tried it myself,\n> > (no need, myself being the only user...) but the web page\n> > (http://www.greatbridge.org/project/phppgadmin/projdisplay.php) says:\n>\n> Yes, it is. Some people prefer a non-web interaface.\n\nI could understand this sentiment if you were doing something from the\ncommand line, but if you are doing an interface in tcl/tk, I think a web\ninterface fairs better ...\n\ntcl/tk, you have to be on the same machine or risk \"network problems\"\nlosing your link ... web, its a stateless connection, so each time you do\nsomething, it re-establishes the connection and performs the operation ...\n\nIMHO, a web interface will always be appropriate in a network environment,\nunless you have the luxury of sitting on the same machine you are running\nthe X app (tcl/tk) on ... *shrug*\n\n\n\n", "msg_date": "Fri, 30 Mar 2001 15:44:23 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [ADMIN] User administration tool" }, { "msg_contents": "The Hermit Hacker writes:\n\n> > Yes, it is. Some people prefer a non-web interaface.\n>\n> I could understand this sentiment if you were doing something from the\n> command line, but if you are doing an interface in tcl/tk, I think a web\n> interface fairs better ...\n>\n> tcl/tk, you have to be on the same machine or risk \"network problems\"\n> losing your link ...\n\nA Tcl/Tk applications doesn't risk any more network problems than any\nother interface to the database, unless that interface is run on the same\nmachine as the database, which is by no means a given.\n\nIf you are under a constant risk of \"network problems\" then you better fix\nyour network. A web interface won't magically bridge network problems.\n\n> web, its a stateless connection, so each time you do something, it\n> re-establishes the connection and performs the operation ...\n\nIt makes the whole thing slower and won't let you preserve state. This\nseems to be a disadvantage to me.\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Fri, 30 Mar 2001 23:45:39 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [ADMIN] User administration tool" }, { "msg_contents": "On Fri, 30 Mar 2001, Peter Eisentraut wrote:\n\n> The Hermit Hacker writes:\n>\n> > > Yes, it is. Some people prefer a non-web interaface.\n> >\n> > I could understand this sentiment if you were doing something from the\n> > command line, but if you are doing an interface in tcl/tk, I think a web\n> > interface fairs better ...\n> >\n> > tcl/tk, you have to be on the same machine or risk \"network problems\"\n> > losing your link ...\n>\n> A Tcl/Tk applications doesn't risk any more network problems than any\n> other interface to the database, unless that interface is run on the same\n> machine as the database, which is by no means a given.\n>\n> If you are under a constant risk of \"network problems\" then you better fix\n> your network. A web interface won't magically bridge network problems.\n\nmy \"network problem\" happens to be a database server half way across the\ncountry from me ... running a tcl/tk interface would be useless, while a\nweb interface does exactly what I need ...\n\n\n", "msg_date": "Fri, 30 Mar 2001 17:48:47 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: [ADMIN] User administration tool" } ]
[ { "msg_contents": "> >Reported problem is caused by bug (only one tuple version must be\n> >returned by SELECT) and this is way to fix it.\n> >\n> \n> I assume this is not possible in 7.1?\n\nJust looked in heapam.c - I can fix it in two hours.\nThe question is - should we do this now?\nComments?\n\nVadim\n", "msg_date": "Thu, 29 Mar 2001 19:14:39 -0800", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" } ]
[ { "msg_contents": "At 19:14 29/03/01 -0800, Mikheev, Vadim wrote:\n>> >Reported problem is caused by bug (only one tuple version must be\n>> >returned by SELECT) and this is way to fix it.\n>> >\n>> \n>> I assume this is not possible in 7.1?\n>\n>Just looked in heapam.c - I can fix it in two hours.\n>The question is - should we do this now?\n>Comments?\n\nIt's a bug; how confident are you of the fix?\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Fri, 30 Mar 2001 14:02:28 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": true, "msg_subject": "RE: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" }, { "msg_contents": "Philip Warner wrote:\n> \n> At 19:14 29/03/01 -0800, Mikheev, Vadim wrote:\n> >> >Reported problem is caused by bug (only one tuple version must be\n> >> >returned by SELECT) and this is way to fix it.\n> >> >\n> >>\n> >> I assume this is not possible in 7.1?\n> >\n> >Just looked in heapam.c - I can fix it in two hours.\n> >The question is - should we do this now?\n> >Comments?\n> \n> It's a bug; how confident are you of the fix?\n> \n\nI doubt if it's a bug of SELECT. Well what\n'concurrent UPDATE then SELECT FOR UPDATE +\nSELECT' return ?\n\nregards,\nHiroshi Inoue\n", "msg_date": "Fri, 30 Mar 2001 13:16:16 +0900", "msg_from": "Hiroshi Inoue <Inoue@tpf.co.jp>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" }, { "msg_contents": "At 13:16 30/03/01 +0900, Hiroshi Inoue wrote:\n>Philip Warner wrote:\n>> \n>> At 19:14 29/03/01 -0800, Mikheev, Vadim wrote:\n>> >> >Reported problem is caused by bug (only one tuple version must be\n>> >> >returned by SELECT) and this is way to fix it.\n>> >> >\n>> >>\n>> >> I assume this is not possible in 7.1?\n>> >\n>> >Just looked in heapam.c - I can fix it in two hours.\n>> >The question is - should we do this now?\n>> >Comments?\n>> \n>> It's a bug; how confident are you of the fix?\n>> \n>\n>I doubt if it's a bug of SELECT. \n\nNo idea where the bug is, but SELECT should never return two versions of\nthe *same* row.\n\n\n>'Well what\n>'concurrent UPDATE then SELECT FOR UPDATE +\n>SELECT' return ?\n\nNo idea, maybe Vadim or Tom can help?\n\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Fri, 30 Mar 2001 16:14:54 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" }, { "msg_contents": "> Just looked in heapam.c - I can fix it in two hours.\n> The question is - should we do this now?\n\nThis scares the hell out of me.\n\nI do NOT think we should be making quick-hack changes in fundamental\nsystem semantics at this point of the release cycle.\n\nThe problem went unnoticed for two full release cycles --- therefore,\nit can wait another cycle for a fix that has been considered, reviewed,\nand tested. Let's not risk making things worse by releasing a new\nbehavior we might find out is also wrong.\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 30 Mar 2001 01:15:15 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1 " }, { "msg_contents": "On Thursday 29 March 2001 22:15, Tom Lane wrote:\n> > Just looked in heapam.c - I can fix it in two hours.\n> > The question is - should we do this now?\n>\n> This scares the hell out of me.\n>\n> I do NOT think we should be making quick-hack changes in fundamental\n> system semantics at this point of the release cycle.\n\nAlthough I'm the one who is being bit by this bug, I tend to agree.\n\n> The problem went unnoticed for two full release cycles \n\nI first reported the problem on 25 September 2000, on the pgsql-sql list, \nmessage subject \"SQL functions not locking properly?\" I was using 7.0.2 at \nthe time. Also, I seem to remember that a problem of this nature bit me in \n6.5.x as well.\n\n> it can wait another cycle for a fix that has been considered, reviewed,\n> and tested. Let's not risk making things worse by releasing a new\n> behavior we might find out is also wrong.\n\nGood point. How long is the next cycle likely to take?\n\nForest\n", "msg_date": "Fri, 30 Mar 2001 15:47:22 -0800", "msg_from": "Forest Wilkinson <fspam@home.com>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" }, { "msg_contents": "Forest Wilkinson <fspam@home.com> writes:\n> Good point. How long is the next cycle likely to take?\n\nGood question. I'd like to say 4 to 6 months, but that was how long 7.1\nwas supposed to take, and we're pushing a year now.\n\nWhat might make the most sense is to develop and test a fix in the early\npart of the 7.2 development cycle, and then back-patch it into a 7.1.x\nrelease perhaps 2 or 3 months from now.\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 30 Mar 2001 19:01:27 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1 " }, { "msg_contents": "I see postgres 7.1.1 is out now. Was the fix for this problem included in \nthe new release?\n\n\nOn Thursday 29 March 2001 20:02, Philip Warner wrote:\n> At 19:14 29/03/01 -0800, Mikheev, Vadim wrote:\n> >> >Reported problem is caused by bug (only one tuple version must be\n> >> >returned by SELECT) and this is way to fix it.\n> >>\n> >> I assume this is not possible in 7.1?\n> >\n> >Just looked in heapam.c - I can fix it in two hours.\n> >The question is - should we do this now?\n> >Comments?\n>\n> It's a bug; how confident are you of the fix?\n", "msg_date": "Wed, 16 May 2001 19:18:51 -0700", "msg_from": "Forest Wilkinson <fspam@home.com>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" } ]
[ { "msg_contents": "> I have started coding a user/group administration tool that allows you\n> to add/modify/delete users and groups. I should have something working\n> in a week. I will look similar to my pgmonitor tool.\n> \n\tsemi related to this, I have always thought that the way postgresql\nhandles the deletion of users and groups to be flawed. If I create a user,\ngrant permissions on a table and then drop the user, permissions now exist\non that table for a user that does not exist. I see this as a possible\nsecurity flaw since a new user can then be created with the user id of the\nID user and have all the permissions that might have ever been assigned to\nthat old user. When a user is deleted, shouldn't all permissions associated\nwith that user be deleted also, I would think this could be handled with a\nPK/ FK cascading delete type setup.\n\n\tmy 2�\n\n\tMatt O'Connor\n", "msg_date": "Thu, 29 Mar 2001 22:56:19 -0600", "msg_from": "Matthew <matt@ctlno.com>", "msg_from_op": true, "msg_subject": "RE: User administration tool" }, { "msg_contents": "Matthew writes:\n\n> \tsemi related to this, I have always thought that the way postgresql\n> handles the deletion of users and groups to be flawed. If I create a user,\n> grant permissions on a table and then drop the user, permissions now exist\n> on that table for a user that does not exist.\n\nUnfortunately it is not possible to prevent this with anything approaching\nease, in the same way that userdel on Unix can't scan all file systems for\nsome to-be-stale files before removing users.\n\n> I see this as a possible security flaw since a new user can then be\n> created with the user id of the ID user and have all the permissions\n> that might have ever been assigned to that old user.\n\nThis will be fixed in 7.2 when Oids will be used as user ids. Of course\nOids can wrap, but that's another days project...\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Fri, 30 Mar 2001 17:28:33 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "RE: User administration tool" } ]
[ { "msg_contents": "> > >> I assume this is not possible in 7.1?\n> > >\n> > >Just looked in heapam.c - I can fix it in two hours.\n> > >The question is - should we do this now?\n> > >Comments?\n> > \n> > It's a bug; how confident are you of the fix?\n\n95% -:)\n\n> I doubt if it's a bug of SELECT. Well what\n> 'concurrent UPDATE then SELECT FOR UPDATE +\n> SELECT' return ?\n\nI'm going to add additional check to heapgettup and\nheap_fetch:\n\nHeapTupleSatisfies(T) is TRUE:\n\nIF XactIsoLevel is READ_COMMITTED\nand snapshot != SnapshotDirty\nand !(T->t_data->t_infomask & HEAP_XMAX_INVALID)\nand T->t_data->t_infomask & HEAP_XMAX_COMMITTED\nand T->t_self != T->t_data->t_ctid\n{\n FOR ( ; ; )\n {\n fetch tuple->t_data->t_ctid tuple\n IF t_infomask & (HEAP_XMAX_INVALID | HEAP_MARKED_FOR_UPDATE)\n break; -- and return T\n IF t_infomask & HEAP_XMAX_COMMITTED\n {\n IF t_self != ctid\t-- updated\n continue;\n break; -- deleted, return T\n }\n -- uncommitted update/delete\n IF t_xmax != CurrentTransactionID\n break; -- and return T\n -- changed by current TX!\n IF changed *BEFORE* this query began\n {\n -- DELETE + SELECT: nothing to return\n -- UPDATE + SELECT: newer tuple version\n -- will be/was returned by query\n return NULL;\n }\n continue;\n }\n}\n\nVadim\n", "msg_date": "Thu, 29 Mar 2001 21:17:50 -0800", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" }, { "msg_contents": "\"Mikheev, Vadim\" wrote:\n> \n> > > >> I assume this is not possible in 7.1?\n> > > >\n> > > >Just looked in heapam.c - I can fix it in two hours.\n> > > >The question is - should we do this now?\n> > > >Comments?\n> > >\n> > > It's a bug; how confident are you of the fix?\n> \n> 95% -:)\n> \n> > I doubt if it's a bug of SELECT. Well what\n> > 'concurrent UPDATE then SELECT FOR UPDATE +\n> > SELECT' return ?\n> \n> I'm going to add additional check to heapgettup and\n> heap_fetch:\n>\n\nSELECT seems to be able to return a different result\nfrom that of preceding SELECT FOR UPDATE even after\napplying your change.\nSELECT doesn't seem guilty but the result is far \nfrom intuitive.\nIt seems impossoble for all queires inside such\na function to use a common snapshot.\n\nregards,\nHiroshi Inoue\n", "msg_date": "Fri, 30 Mar 2001 19:15:22 +0900", "msg_from": "Hiroshi Inoue <Inoue@tpf.co.jp>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" } ]
[ { "msg_contents": "\n> > A larger question is whether unsigned types really add much to the\n> > system vs. the bloat. We already have unsigned int4 as oid. Also,\n> > unsigned doubles the space of the type, but if a value doesn't fit in\n> > 32k, what are the odds it will fit in 64k. I am not sure unsigned\n> > optimzations for space really are significant in SQL.\n\n> If general opinion is that unsigned types should not be part of\n> postgres, I'll have to look at turning them into a contrib type. Please\n> let me know.\n\nImho this would be better. These types are not generally available (e.g. not in Oracle).\nOne other approach to the porting issue would probably simply involve moving \nfrom int2 to integer with an appropriate check constraint \n[int xx check (xx between 0 and 65536)], and from int4 to decimal(9).\n\nAndreas\n", "msg_date": "Fri, 30 Mar 2001 09:03:55 +0200", "msg_from": "Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at>", "msg_from_op": true, "msg_subject": "AW: Unsigned int functions" } ]
[ { "msg_contents": "hi,\n\nSorry if you receive this message again . I just sent it yesterday with attached the config file but probably was too large.\n\nI come back in office after a long period out for work.\nYesterday I have downloaded the last snapshot.\n\nWhen I execute configure I have an error compiling conftest.\nSeems that non found some librarys but in the previus lines (when execute the check write yes)\n\nThe last version I have checked was 7.1 b3 and all works fine.\n\nThis is only the last few lines of the config.log file.\n-----\nconfigure:6840: checking test program\nconfigure:6849: gcc -o conftest.map conftest.c -lz -lunix -lresolv -lPW -lgen -lBSD -lcompat -lld -ldld -llc -lIPC -lipc -lnsl -lsocket -ldl -lm -lbsd -lsfio -lunix 1>&5\ncc warning: cc: cannot find library 'resolv'\ncc warning: cc: cannot find library 'PW'\ncc warning: cc: cannot find library 'gen'\ncc warning: cc: cannot find library 'BSD'\ncc warning: cc: cannot find library 'compat'\ncc warning: cc: cannot find library 'ld'\ncc warning: cc: cannot find library 'dld'\ncc warning: cc: cannot find library 'lc'\ncc warning: cc: cannot find library 'IPC'\ncc warning: cc: cannot find library 'ipc'\ncc warning: cc: cannot find library 'nsl'\ncc warning: cc: cannot find library 'dl'\ncc warning: cc: cannot find library 'bsd'\ncc warning: cc: cannot find library 'sfio'\ncc warning: cc: cannot find library 'resolv'\ncc warning: cc: cannot find library 'PW'\ncc warning: cc: cannot find library 'gen'\ncc warning: cc: cannot find library 'BSD'\ncc warning: cc: cannot find library 'compat'\ncc warning: cc: cannot find library 'ld'\ncc warning: cc: cannot find library 'dld'\ncc warning: cc: cannot find library 'lc'\ncc warning: cc: cannot find library 'IPC'\ncc warning: cc: cannot find library 'ipc'\ncc warning: cc: cannot find library 'nsl'\ncc warning: cc: cannot find library 'dl'\ncc warning: cc: cannot find library 'bsd'\ncc warning: cc: cannot find library 'sfio'\nconfigure: failed program was:\n#line 6845 \"configure\"\n#include \"confdefs.h\"\nint main() { return 0; }\n \n\n-----\nHas someone any suggestion ?\n\nThanks\n\n\nMaurizio Cauci\nDREAMTECH di Cauci Maurizio\nVia Ronchetti, 2 - 21013 Gallarate (VA)\nwww.dreamtech-it.com\n\n\n\n\n\n\n\n\nhi,\n \nSorry if you receive this message again . I just sent it yesterday \nwith attached the config file but probably was too large.\n \nI come back in office after a long period out for \nwork.\nYesterday I have downloaded  the last \nsnapshot.\n \nWhen I execute configure I have an error compiling \nconftest.\nSeems that non found some librarys but in the \nprevius lines (when execute the check write yes)\n \nThe last version I have checked was 7.1 b3 and all \nworks fine.\n \nThis is only the last few lines of the \nconfig.log file.\n-----\n\nconfigure:6840: checking test \nprogramconfigure:6849: gcc -o conftest.map      conftest.c -lz \n-lunix -lresolv -lPW -lgen -lBSD -lcompat -lld -ldld -llc -lIPC -lipc -lnsl \n-lsocket -ldl -lm -lbsd -lsfio -lunix 1>&5cc warning: cc: cannot find \nlibrary 'resolv'cc warning: cc: cannot find library 'PW'cc warning: cc: \ncannot find library 'gen'cc warning: cc: cannot find library 'BSD'cc \nwarning: cc: cannot find library 'compat'cc warning: cc: cannot find library \n'ld'cc warning: cc: cannot find library 'dld'cc warning: cc: cannot find \nlibrary 'lc'cc warning: cc: cannot find library 'IPC'cc warning: cc: \ncannot find library 'ipc'cc warning: cc: cannot find library 'nsl'cc \nwarning: cc: cannot find library 'dl'cc warning: cc: cannot find library \n'bsd'cc warning: cc: cannot find library 'sfio'cc warning: cc: cannot \nfind library 'resolv'cc warning: cc: cannot find library 'PW'cc warning: \ncc: cannot find library 'gen'cc warning: cc: cannot find library 'BSD'cc \nwarning: cc: cannot find library 'compat'cc warning: cc: cannot find library \n'ld'cc warning: cc: cannot find library 'dld'cc warning: cc: cannot find \nlibrary 'lc'cc warning: cc: cannot find library 'IPC'cc warning: cc: \ncannot find library 'ipc'cc warning: cc: cannot find library 'nsl'cc \nwarning: cc: cannot find library 'dl'cc warning: cc: cannot find library \n'bsd'cc warning: cc: cannot find library 'sfio'configure: failed program \nwas:#line 6845 \"configure\"#include \"confdefs.h\"int main() { return \n0; } \n-----\nHas someone any suggestion ?\n \nThanks\n \n \nMaurizio CauciDREAMTECH di Cauci \nMaurizioVia Ronchetti, 2 - 21013 Gallarate (VA)www.dreamtech-it.com", "msg_date": "Fri, 30 Mar 2001 11:13:08 +0200", "msg_from": "\"Maurizio\" <maurizio.c@libero.it>", "msg_from_op": true, "msg_subject": "testing last sanpshot in QNX platform" }, { "msg_contents": "Maurizio writes:\n\n> configure:6840: checking test program\n> configure:6849: gcc -o conftest.map conftest.c -lz -lunix -lresolv -lPW -lgen -lBSD -lcompat -lld -ldld -llc -lIPC -lipc -lnsl -lsocket -ldl -lm -lbsd -lsfio -lunix 1>&5\n> cc warning: cc: cannot find library 'resolv'\n> cc warning: cc: cannot find library 'PW'\n[etc]\n\nThis means that earlier in configure it was determined that these\nlibraries existed (see \"checking for main in -lxxx\") but now it doesn't\nwork anymore. Not sure why this could happen, given that people have used\nQNX previously. (At least they got past this point.)\n\nWhat's curious here is that it wants to name the output program\n\"conftest.map\", which looks like it detected \".map\" as the executable\nextension (ordinarily only used for \".exe\" on Windows). What's also\ncurious is that the error message doesn't look like anything \"gcc\" would\nproduce. I think this might be a case of a messed up compiler\ninstallation and/or a case of a user cheating with configure to cover up\nfor that fact. ;-) Some more information about your compiler setup and a\npeek into config.log near the compiler detection tests could shed some\nlight onto the problem.\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Fri, 30 Mar 2001 17:38:31 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: testing last sanpshot in QNX platform" }, { "msg_contents": "Strange. The compiler work with all my old project and also with the old\nversion of postgresql.\nHowever I will search the problem and I will post the risult.\n\n\nMaurizio\n\n.\n----- Original Message -----\nFrom: \"Peter Eisentraut\" <peter_e@gmx.net>\nTo: \"Maurizio\" <maurizio.c@libero.it>\nCc: <pgsql-hackers@postgresql.org>\nSent: Friday, March 30, 2001 5:38 PM\nSubject: Re: [HACKERS] testing last sanpshot in QNX platform\n\n\n> Maurizio writes:\n>\n> > configure:6840: checking test program\n> > configure:6849: gcc -o conftest.map\n\n\n\n\n\nconftest.c -lz -lunix -lresolv -lPW -lgen -lBSD -lcompat -lld -ldld -llc -lI\nPC -lipc -lnsl -lsocket -ldl -lm -lbsd -lsfio -lunix 1>&5\n> > cc warning: cc: cannot find library 'resolv'\n> > cc warning: cc: cannot find library 'PW'\n> [etc]\n>\n> This means that earlier in configure it was determined that these\n> libraries existed (see \"checking for main in -lxxx\") but now it doesn't\n> work anymore. Not sure why this could happen, given that people have used\n> QNX previously. (At least they got past this point.)\n>\n> What's curious here is that it wants to name the output program\n> \"conftest.map\", which looks like it detected \".map\" as the executable\n> extension (ordinarily only used for \".exe\" on Windows). What's also\n> curious is that the error message doesn't look like anything \"gcc\" would\n> produce. I think this might be a case of a messed up compiler\n> installation and/or a case of a user cheating with configure to cover up\n> for that fact. ;-) Some more information about your compiler setup and a\n> peek into config.log near the compiler detection tests could shed some\n> light onto the problem.\n>\n> --\n> Peter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n>\n\n", "msg_date": "Sat, 31 Mar 2001 17:48:14 +0200", "msg_from": "\"Maurizio\" <maurizio.c@libero.it>", "msg_from_op": true, "msg_subject": "Re: testing last sanpshot in QNX platform" } ]
[ { "msg_contents": "\n> > > I doubt if it's a bug of SELECT. Well what\n> > > 'concurrent UPDATE then SELECT FOR UPDATE +\n> > > SELECT' return ?\n> > \n> > I'm going to add additional check to heapgettup and\n> > heap_fetch:\n> >\n> \n> SELECT seems to be able to return a different result\n> from that of preceding SELECT FOR UPDATE even after\n> applying your change.\n\nOnly if you left this cursor position without doing an actual update\n(i.e. after fetch next). The select for update is only supposed to guard \nthe current cursor position. Once you leave without modification\nanother session can be allowed to update. \nThis is how it is supposed to react in read committed\nmode. If you don't like this you need repeatable read.\n\nThe example given is of questionable value, since a select for update \nwithout a cursor in read committed mode does not need to behave any different \nthan a simple select without for update.\n\n> SELECT doesn't seem guilty but the result is far \n> from intuitive.\n\nIt is intuitive. The bug was iirc, that you saw 2 versions of the same row\nin the second select statement (= 2 rows returned by second select).\nVadim's patch will let you see only the newer row.\n\n> It seems impossoble for all queires inside such\n> a function to use a common snapshot.\n\nIn read committed they are not required to !\n\nIt looks like a lot of people on the list are absolute fans\nof repeatable read isolation :-) Not me, I know a lot of applications\nwhere committed read, or even read uncommitted makes a lot more \nsense.\n\nAndreas\n", "msg_date": "Fri, 30 Mar 2001 15:13:14 +0200", "msg_from": "Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at>", "msg_from_op": true, "msg_subject": "AW: Re: [SQL] possible row locking bug in 7.0.3 & 7.1" }, { "msg_contents": "Zeugswetter Andreas SB <ZeugswetterA@Wien.Spardat.at> writes:\n> It is intuitive. The bug was iirc, that you saw 2 versions of the same row\n> in the second select statement (= 2 rows returned by second select).\n\nI think we should be extremely wary of assuming that we have a clear\ncharacterization of \"what the bug is\", let alone \"how to fix it\".\nThe real issue here is that SELECT has different MVCC visibility rules\nfrom UPDATE and SELECT FOR UPDATE. I suspect that that *must* be so\nin any mode that allows more concurrency than full serializable mode.\nThus, the question we are really facing is how we might alter the\nvisibility rules in a way that will make the results more intuitive\nand/or useful while still allowing concurrency.\n\nThis will take thought, research and discussion. A quick fix is the\nlast thing that should be on our minds.\n\nA first question: where did the MVCC rules come from originally, anyway?\nIs there any academic research to look at?\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 30 Mar 2001 10:05:40 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: AW: Re: [SQL] possible row locking bug in 7.0.3 & 7.1 " } ]
[ { "msg_contents": "\nHi,\n\nI've written an extension to PGSQL that in some cases has to abort the\ncurrent transaction. It calls AbortCurrentTransaction() in that case, but\nthe problem is that the client doesn't get notified.\n\nIs there a way to detect such an abort, or do I have to make a\nmodification the the client libs ??\n\n\n\n\tRegards,\n\n\n\t\tIgmar\n\n", "msg_date": "Fri, 30 Mar 2001 16:42:34 +0200 (CEST)", "msg_from": "Igmar Palsenberg <maillist@jdimedia.nl>", "msg_from_op": true, "msg_subject": "client notification of AbortTransaction()" } ]
[ { "msg_contents": "Unreported or problem platforms:\n\nLinux 2.0.x MIPS 7.0 2000-04-13 (Tatsuo has lost machine)\nmklinux PPC750 7.0 2000-04-13, Tatsuo Ishii\nNetBSD m68k 7.0 2000-04-10 (Henry has lost machine)\nNetBSD Sparc 7.0 2000-04-13, Tom I. Helbekkmo\nQNX 4.25 x86 7.0 2000-04-01, Dr. Andreas Kardos\nUltrix MIPS 7.1 2001-??-??, Alexander Klimov\n\nmklinux has failed Tatsuo's testing afaicr. Demote to unsupported?\n\nAny NetBSD partisans who can do testing or solicit testing from the\nNetBSD crowd? Same for OpenBSD?\n\nQNX is known to have problems with 7.1. Any hope of fixing for 7.1.1? Is\nthere anyone able to work on it? If not, I'll move to the unsupported\nlist.\n\n\nAnd here are the up-to-date platforms; thanks for the reports:\n\nAIX 4.3.3 RS6000 7.1 2001-03-21, Gilles Darold\nBeOS 5.0.3 x86 7.1 2000-12-18, Cyril Velter\nBSDI 4.01 x86 7.1 2001-03-19, Bruce Momjian\nCompaq Tru64 4.0g Alpha 7.1 2001-03-19, Brent Verner\nFreeBSD 4.3 x86 7.1 2001-03-19, Vince Vielhaber\nHPUX PA-RISC 7.1 2001-03-19, 10.20 Tom Lane, 11.00 Giles Lean\nIRIX 6.5.11 MIPS 7.1 2001-03-22, Robert Bruccoleri\nLinux 2.2.x Alpha 7.1 2001-01-23, Ryan Kirkpatrick\nLinux 2.2.x armv4l 7.1 2001-03-22, Mark Knox\nLinux 2.2.18 PPC750 7.1 2001-03-19, Tom Lane\nLinux 2.2.x S/390 7.1 2000-11-17, Neale Ferguson\nLinux 2.2.15 Sparc 7.1 2001-01-30, Ryan Kirkpatrick\nLinux 2.2.16 x86 7.1 2001-03-19, Thomas Lockhart\nMacOS X Darwin PPC 7.1 2000-12-11, Peter Bierman\nNetBSD 1.5 alpha 7.1 2001-03-22, Giles Lean\nNetBSD 1.5E arm32 7.1 2001-03-21, Patrick Welche\nNetBSD 1.5S x86 7.1 2001-03-21, Patrick Welche\nOpenBSD 2.8 x86 7.1 2001-03-22, Brandon Palmer\nSCO OpenServer 5 x86 7.1 2001-03-13, Billy Allie\nSCO UnixWare 7.1.1 x86 7.1 2001-03-19, Larry Rosenman\nSolaris 2.7 Sparc 7.1 2001-03-22, Marc Fournier\nSolaris x86 7.1 2001-03-27, Mathijs Brands\nSunOS 4.1.4 Sparc 7.1 2001-03-23, Tatsuo Ishii\nWindows/Win32 x86 7.1 2001-03-26, Magnus Hagander (clients only)\nWinNT/Cygwin x86 7.1 2001-03-16, Jason Tishler\n", "msg_date": "Fri, 30 Mar 2001 15:17:06 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": true, "msg_subject": "Third call for platform testing" }, { "msg_contents": "On Fri, Mar 30, 2001 at 03:17:06PM +0000, Thomas Lockhart allegedly wrote:\n> And here are the up-to-date platforms; thanks for the reports:\n\n<SNIP>\n\n> Solaris 2.7 Sparc 7.1 2001-03-22, Marc Fournier\n\nMarc, was this done without unix sockets?\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Fri, 30 Mar 2001 19:25:12 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: Third call for platform testing" }, { "msg_contents": "On Fri, 30 Mar 2001, Mathijs Brands wrote:\n\n> On Fri, Mar 30, 2001 at 03:17:06PM +0000, Thomas Lockhart allegedly wrote:\n> > And here are the up-to-date platforms; thanks for the reports:\n>\n> <SNIP>\n>\n> > Solaris 2.7 Sparc 7.1 2001-03-22, Marc Fournier\n>\n> Marc, was this done without unix sockets?\n\nnope, purely default ... it was only the x86 platform that I had a bugger\nwith getting a clean regress working on ...\n\n\n", "msg_date": "Fri, 30 Mar 2001 13:50:53 -0400 (AST)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: Third call for platform testing" }, { "msg_contents": "I still don't see an entry for Linux 2.4.x\n\nCheers.\n\nThomas Lockhart wrote:\n\n> Unreported or problem platforms:\n>\n> Linux 2.0.x MIPS 7.0 2000-04-13 (Tatsuo has lost machine)\n> mklinux PPC750 7.0 2000-04-13, Tatsuo Ishii\n> NetBSD m68k 7.0 2000-04-10 (Henry has lost machine)\n> NetBSD Sparc 7.0 2000-04-13, Tom I. Helbekkmo\n> QNX 4.25 x86 7.0 2000-04-01, Dr. Andreas Kardos\n> Ultrix MIPS 7.1 2001-??-??, Alexander Klimov\n>\n> mklinux has failed Tatsuo's testing afaicr. Demote to unsupported?\n>\n> Any NetBSD partisans who can do testing or solicit testing from the\n> NetBSD crowd? Same for OpenBSD?\n>\n> QNX is known to have problems with 7.1. Any hope of fixing for 7.1.1? Is\n> there anyone able to work on it? If not, I'll move to the unsupported\n> list.\n>\n> And here are the up-to-date platforms; thanks for the reports:\n>\n> AIX 4.3.3 RS6000 7.1 2001-03-21, Gilles Darold\n> BeOS 5.0.3 x86 7.1 2000-12-18, Cyril Velter\n> BSDI 4.01 x86 7.1 2001-03-19, Bruce Momjian\n> Compaq Tru64 4.0g Alpha 7.1 2001-03-19, Brent Verner\n> FreeBSD 4.3 x86 7.1 2001-03-19, Vince Vielhaber\n> HPUX PA-RISC 7.1 2001-03-19, 10.20 Tom Lane, 11.00 Giles Lean\n> IRIX 6.5.11 MIPS 7.1 2001-03-22, Robert Bruccoleri\n> Linux 2.2.x Alpha 7.1 2001-01-23, Ryan Kirkpatrick\n> Linux 2.2.x armv4l 7.1 2001-03-22, Mark Knox\n> Linux 2.2.18 PPC750 7.1 2001-03-19, Tom Lane\n> Linux 2.2.x S/390 7.1 2000-11-17, Neale Ferguson\n> Linux 2.2.15 Sparc 7.1 2001-01-30, Ryan Kirkpatrick\n> Linux 2.2.16 x86 7.1 2001-03-19, Thomas Lockhart\n> MacOS X Darwin PPC 7.1 2000-12-11, Peter Bierman\n> NetBSD 1.5 alpha 7.1 2001-03-22, Giles Lean\n> NetBSD 1.5E arm32 7.1 2001-03-21, Patrick Welche\n> NetBSD 1.5S x86 7.1 2001-03-21, Patrick Welche\n> OpenBSD 2.8 x86 7.1 2001-03-22, Brandon Palmer\n> SCO OpenServer 5 x86 7.1 2001-03-13, Billy Allie\n> SCO UnixWare 7.1.1 x86 7.1 2001-03-19, Larry Rosenman\n> Solaris 2.7 Sparc 7.1 2001-03-22, Marc Fournier\n> Solaris x86 7.1 2001-03-27, Mathijs Brands\n> SunOS 4.1.4 Sparc 7.1 2001-03-23, Tatsuo Ishii\n> Windows/Win32 x86 7.1 2001-03-26, Magnus Hagander (clients only)\n> WinNT/Cygwin x86 7.1 2001-03-16, Jason Tishler\n>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 2: you can get off all lists at once with the unregister command\n> (send \"unregister YourEmailAddressHere\" to majordomo@postgresql.org)\n\n", "msg_date": "Sat, 31 Mar 2001 12:02:35 +1200", "msg_from": "Franck Martin <franck@sopac.org>", "msg_from_op": false, "msg_subject": "Re: Third call for platform testing (linux 2.4.x)" }, { "msg_contents": "> I still don't see an entry for Linux 2.4.x\n\nMy (uncommitted) updates to the real list show 2.4.2 in the comments\nsection. I may remove all mention of versions, since it seems that most\nreleased versions of x86 Linux run PostgreSQL successfully.\n\nComments?\n\n - Thomas\n\n> > Unreported or problem platforms:\n> >\n> > Linux 2.0.x MIPS 7.0 2000-04-13 (Tatsuo has lost machine)\n> > mklinux PPC750 7.0 2000-04-13, Tatsuo Ishii\n> > NetBSD m68k 7.0 2000-04-10 (Henry has lost machine)\n> > NetBSD Sparc 7.0 2000-04-13, Tom I. Helbekkmo\n> > QNX 4.25 x86 7.0 2000-04-01, Dr. Andreas Kardos\n> > Ultrix MIPS 7.1 2001-??-??, Alexander Klimov\n> >\n> > mklinux has failed Tatsuo's testing afaicr. Demote to unsupported?\n> >\n> > Any NetBSD partisans who can do testing or solicit testing from the\n> > NetBSD crowd? Same for OpenBSD?\n> >\n> > QNX is known to have problems with 7.1. Any hope of fixing for 7.1.1? Is\n> > there anyone able to work on it? If not, I'll move to the unsupported\n> > list.\n> >\n> > And here are the up-to-date platforms; thanks for the reports:\n> >\n> > AIX 4.3.3 RS6000 7.1 2001-03-21, Gilles Darold\n> > BeOS 5.0.3 x86 7.1 2000-12-18, Cyril Velter\n> > BSDI 4.01 x86 7.1 2001-03-19, Bruce Momjian\n> > Compaq Tru64 4.0g Alpha 7.1 2001-03-19, Brent Verner\n> > FreeBSD 4.3 x86 7.1 2001-03-19, Vince Vielhaber\n> > HPUX PA-RISC 7.1 2001-03-19, 10.20 Tom Lane, 11.00 Giles Lean\n> > IRIX 6.5.11 MIPS 7.1 2001-03-22, Robert Bruccoleri\n> > Linux 2.2.x Alpha 7.1 2001-01-23, Ryan Kirkpatrick\n> > Linux 2.2.x armv4l 7.1 2001-03-22, Mark Knox\n> > Linux 2.2.18 PPC750 7.1 2001-03-19, Tom Lane\n> > Linux 2.2.x S/390 7.1 2000-11-17, Neale Ferguson\n> > Linux 2.2.15 Sparc 7.1 2001-01-30, Ryan Kirkpatrick\n> > Linux 2.2.16 x86 7.1 2001-03-19, Thomas Lockhart\n> > MacOS X Darwin PPC 7.1 2000-12-11, Peter Bierman\n> > NetBSD 1.5 alpha 7.1 2001-03-22, Giles Lean\n> > NetBSD 1.5E arm32 7.1 2001-03-21, Patrick Welche\n> > NetBSD 1.5S x86 7.1 2001-03-21, Patrick Welche\n> > OpenBSD 2.8 x86 7.1 2001-03-22, Brandon Palmer\n> > SCO OpenServer 5 x86 7.1 2001-03-13, Billy Allie\n> > SCO UnixWare 7.1.1 x86 7.1 2001-03-19, Larry Rosenman\n> > Solaris 2.7 Sparc 7.1 2001-03-22, Marc Fournier\n> > Solaris x86 7.1 2001-03-27, Mathijs Brands\n> > SunOS 4.1.4 Sparc 7.1 2001-03-23, Tatsuo Ishii\n> > Windows/Win32 x86 7.1 2001-03-26, Magnus Hagander (clients only)\n> > WinNT/Cygwin x86 7.1 2001-03-16, Jason Tishler\n> >\n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 2: you can get off all lists at once with the unregister command\n> > (send \"unregister YourEmailAddressHere\" to majordomo@postgresql.org)\n", "msg_date": "Sat, 31 Mar 2001 00:50:48 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": true, "msg_subject": "Re: Third call for platform testing (linux 2.4.x)" }, { "msg_contents": "On Sat, Mar 31, 2001 at 12:02:35PM +1200, Franck Martin allegedly wrote:\n> I still don't see an entry for Linux 2.4.x\n> \n> Cheers.\n\nThis should fix that:\n\n======================\n All 76 tests passed.\n======================\n\nrm regress.o\nmake[2]: Leaving directory `/usr/exp/tmp/postgresql-7.1RC1/src/test/regress'\nmake[1]: Leaving directory `/usr/exp/tmp/postgresql-7.1RC1/src/test'\npc11j:/usr/exp/tmp/postgresql-7.1RC1$ uname -a\nLinux pc11j 2.4.2 #8 Fri Mar 2 16:40:13 CET 2001 i686 unknown\n\nMathijs\n-- \nIt's not that perl programmers are idiots, it's that the language\nrewards idiotic behavior in a way that no other language or tool has\never done.\n Erik Naggum\n", "msg_date": "Sat, 31 Mar 2001 03:05:12 +0200", "msg_from": "Mathijs Brands <mathijs@ilse.nl>", "msg_from_op": false, "msg_subject": "Re: Third call for platform testing (linux 2.4.x)" }, { "msg_contents": "\nI just built and tested RC1 on Linux 2.4.2, with glibc-2.2.2 and\ngcc-2.95.2 on a Debian 2.2+ x86 system. (\"+\" implying some packages\nfrom \"unstable\".)\n\nI configured it --with-perl --with-openssl --with-CXX.\nIt built without errors, but with a few warnings.\n\nThis one seemed (portably) odd:\n----------\n In file included from gram.y:43:\n lex.plpgsql_yy.c: In function `plpgsql_yylex':\n lex.plpgsql_yy.c:972: warning: label `find_rule' defined but not used\n----------\n\nAnd this:\n----------\n ar crs libpq.a `lorder fe-auth.o fe-connect.o fe-exec.o fe-misc.o fe-print.o fe-lobj.o pqexpbuffer.o dllist.o pqsignal.o | tsort`\n tsort: -: input contains a loop:\n\n tsort: dllist.o\n----------\n\nAnd this:\n----------\n ar crs libecpg.a `lorder execute.o typename.o descriptor.o data.o error.o prepare.o memory.o connect.o misc.o | tsort`\n tsort: -: input contains a loop:\n\n tsort: connect.o\n tsort: execute.o\n tsort: data.o\n----------\n\nAnd this:\n\n----------\n ar crs libplpgsql.a `lorder pl_parse.o pl_handler.o pl_comp.o pl_exec.o pl_funcs.o | tsort`\n tsort: -: input contains a loop:\n\n tsort: pl_comp.o\n tsort: pl_parse.o\n----------\n\nI ran \"make check\". It said:\n\n----------\n All 76 tests passed. \n----------\n\nNathan Myers\nncm@zembu.com\n\nOn Sat, Mar 31, 2001 at 12:02:35PM +1200, Franck Martin wrote:\n> I still don't see an entry for Linux 2.4.x\n> \n> Cheers.\n> \n> Thomas Lockhart wrote:\n> \n> > Unreported or problem platforms:\n> >\n> > Linux 2.0.x MIPS 7.0 2000-04-13 (Tatsuo has lost machine)\n> > mklinux PPC750 7.0 2000-04-13, Tatsuo Ishii\n> > NetBSD m68k 7.0 2000-04-10 (Henry has lost machine)\n> > NetBSD Sparc 7.0 2000-04-13, Tom I. Helbekkmo\n> > QNX 4.25 x86 7.0 2000-04-01, Dr. Andreas Kardos\n> > Ultrix MIPS 7.1 2001-??-??, Alexander Klimov\n> >\n> > mklinux has failed Tatsuo's testing afaicr. Demote to unsupported?\n> >\n> > Any NetBSD partisans who can do testing or solicit testing from the\n> > NetBSD crowd? Same for OpenBSD?\n> >\n> > QNX is known to have problems with 7.1. Any hope of fixing for 7.1.1? Is\n> > there anyone able to work on it? If not, I'll move to the unsupported\n> > list.\n> >\n> > And here are the up-to-date platforms; thanks for the reports:\n> >\n> > AIX 4.3.3 RS6000 7.1 2001-03-21, Gilles Darold\n> > BeOS 5.0.3 x86 7.1 2000-12-18, Cyril Velter\n> > BSDI 4.01 x86 7.1 2001-03-19, Bruce Momjian\n> > Compaq Tru64 4.0g Alpha 7.1 2001-03-19, Brent Verner\n> > FreeBSD 4.3 x86 7.1 2001-03-19, Vince Vielhaber\n> > HPUX PA-RISC 7.1 2001-03-19, 10.20 Tom Lane, 11.00 Giles Lean\n> > IRIX 6.5.11 MIPS 7.1 2001-03-22, Robert Bruccoleri\n> > Linux 2.2.x Alpha 7.1 2001-01-23, Ryan Kirkpatrick\n> > Linux 2.2.x armv4l 7.1 2001-03-22, Mark Knox\n> > Linux 2.2.18 PPC750 7.1 2001-03-19, Tom Lane\n> > Linux 2.2.x S/390 7.1 2000-11-17, Neale Ferguson\n> > Linux 2.2.15 Sparc 7.1 2001-01-30, Ryan Kirkpatrick\n> > Linux 2.2.16 x86 7.1 2001-03-19, Thomas Lockhart\n> > MacOS X Darwin PPC 7.1 2000-12-11, Peter Bierman\n> > NetBSD 1.5 alpha 7.1 2001-03-22, Giles Lean\n> > NetBSD 1.5E arm32 7.1 2001-03-21, Patrick Welche\n> > NetBSD 1.5S x86 7.1 2001-03-21, Patrick Welche\n> > OpenBSD 2.8 x86 7.1 2001-03-22, Brandon Palmer\n> > SCO OpenServer 5 x86 7.1 2001-03-13, Billy Allie\n> > SCO UnixWare 7.1.1 x86 7.1 2001-03-19, Larry Rosenman\n> > Solaris 2.7 Sparc 7.1 2001-03-22, Marc Fournier\n> > Solaris x86 7.1 2001-03-27, Mathijs Brands\n> > SunOS 4.1.4 Sparc 7.1 2001-03-23, Tatsuo Ishii\n> > Windows/Win32 x86 7.1 2001-03-26, Magnus Hagander (clients only)\n> > WinNT/Cygwin x86 7.1 2001-03-16, Jason Tishler\n> >\n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 2: you can get off all lists at once with the unregister command\n> > (send \"unregister YourEmailAddressHere\" to majordomo@postgresql.org)\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 2: you can get off all lists at once with the unregister command\n> (send \"unregister YourEmailAddressHere\" to majordomo@postgresql.org)\n", "msg_date": "Fri, 30 Mar 2001 17:43:51 -0800", "msg_from": "ncm@zembu.com (Nathan Myers)", "msg_from_op": false, "msg_subject": "Re: Third call for platform testing (linux 2.4.x)" }, { "msg_contents": "> Unreported or problem platforms:\n> \n> Linux 2.0.x MIPS 7.0 2000-04-13 (Tatsuo has lost machine)\n> mklinux PPC750 7.0 2000-04-13, Tatsuo Ishii\n> NetBSD m68k 7.0 2000-04-10 (Henry has lost machine)\n> NetBSD Sparc 7.0 2000-04-13, Tom I. Helbekkmo\n> QNX 4.25 x86 7.0 2000-04-01, Dr. Andreas Kardos\n> Ultrix MIPS 7.1 2001-??-??, Alexander Klimov\n> \n> mklinux has failed Tatsuo's testing afaicr. Demote to unsupported?\n\nYes. But you'd better to change mklinux -> MkLinux DR1. There may be\na chance that latest MkLinux or gcc successfully runs 7.1...\n\n> \n> Any NetBSD partisans who can do testing or solicit testing from the\n> NetBSD crowd? Same for OpenBSD?\n> \n> QNX is known to have problems with 7.1. Any hope of fixing for 7.1.1? Is\n> there anyone able to work on it? If not, I'll move to the unsupported\n> list.\n> \n> \n> And here are the up-to-date platforms; thanks for the reports:\n> \n> AIX 4.3.3 RS6000 7.1 2001-03-21, Gilles Darold\n> BeOS 5.0.3 x86 7.1 2000-12-18, Cyril Velter\n> BSDI 4.01 x86 7.1 2001-03-19, Bruce Momjian\n> Compaq Tru64 4.0g Alpha 7.1 2001-03-19, Brent Verner\n> FreeBSD 4.3 x86 7.1 2001-03-19, Vince Vielhaber\n> HPUX PA-RISC 7.1 2001-03-19, 10.20 Tom Lane, 11.00 Giles Lean\n> IRIX 6.5.11 MIPS 7.1 2001-03-22, Robert Bruccoleri\n> Linux 2.2.x Alpha 7.1 2001-01-23, Ryan Kirkpatrick\n> Linux 2.2.x armv4l 7.1 2001-03-22, Mark Knox\n> Linux 2.2.18 PPC750 7.1 2001-03-19, Tom Lane\n> Linux 2.2.x S/390 7.1 2000-11-17, Neale Ferguson\n> Linux 2.2.15 Sparc 7.1 2001-01-30, Ryan Kirkpatrick\n> Linux 2.2.16 x86 7.1 2001-03-19, Thomas Lockhart\n> MacOS X Darwin PPC 7.1 2000-12-11, Peter Bierman\n> NetBSD 1.5 alpha 7.1 2001-03-22, Giles Lean\n> NetBSD 1.5E arm32 7.1 2001-03-21, Patrick Welche\n> NetBSD 1.5S x86 7.1 2001-03-21, Patrick Welche\n> OpenBSD 2.8 x86 7.1 2001-03-22, Brandon Palmer\n> SCO OpenServer 5 x86 7.1 2001-03-13, Billy Allie\n> SCO UnixWare 7.1.1 x86 7.1 2001-03-19, Larry Rosenman\n> Solaris 2.7 Sparc 7.1 2001-03-22, Marc Fournier\n> Solaris x86 7.1 2001-03-27, Mathijs Brands\n> SunOS 4.1.4 Sparc 7.1 2001-03-23, Tatsuo Ishii\n> Windows/Win32 x86 7.1 2001-03-26, Magnus Hagander (clients only)\n> WinNT/Cygwin x86 7.1 2001-03-16, Jason Tishler\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 2: you can get off all lists at once with the unregister command\n> (send \"unregister YourEmailAddressHere\" to majordomo@postgresql.org)\n> \n", "msg_date": "Tue, 03 Apr 2001 19:58:07 +0900", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": false, "msg_subject": "Re: Third call for platform testing" }, { "msg_contents": "> > mklinux has failed Tatsuo's testing afaicr. Demote to unsupported?\n> Yes. But you'd better to change mklinux -> MkLinux DR1. There may be\n> a chance that latest MkLinux or gcc successfully runs 7.1...\n\nOK. So we are close to a final tally of supported machines. The\n\"unsupported machines\" are listed at the end, and those include QNX and\nUltrix, which both may end up supported in the very near future. I've\nleft NetBSD/m68k and NetBSD/Sparc on the supported list since we have no\nreason to think that they *won't* work for 7.1, but they may get bumped\nto unsupported if we do not have maintainers with access to working\nmachines. That will be especially problematic for the m68k-based\nmachines, since they are no longer in (large scale) production.\n\nI may not have correct info for SCO OpenServer. Can someone verify that\n7.1 works on this platform? With the UDK compiler set, it *should* be\nidentical wrt support to UnixWare with the same compiler set, right? If\nwe don't get testing done, I'll revert the description to that for 7.0,\nbut leave it as a supported platform.\n\nSince Windows (not NT) is supported on the client side only, should I\nmove it to \"unsupported\"? I think I will, but leave the comments that\nclients have been tested.\n\nIf the scorecard does not change, we are on 29 distinct platforms, the\nlargest number we have *ever* been qualified on. And simultaneously for\nall on the day of release, which is simply not done in the closed source\nworld.\n\nThanks to everyone for the great support on porting and testing!\n\n - Thomas\n\nHere are the up-to-date platforms:\n\nAIX 4.3.3 RS6000 7.1 2001-03-21, Gilles Darold\nBeOS 5.0.4 x86 7.1 2000-12-18, Cyril Velter\nBSDI 4.01 x86 7.1 2001-03-19, Bruce Momjian\nCompaq Tru64 4.0g Alpha 7.1 2001-03-19, Brent Verner\nFreeBSD 4.3 x86 7.1 2001-03-19, Vince Vielhaber\nHPUX PA-RISC 7.1 2001-03-19, 10.20 Tom Lane, 11.00 Giles Lean\nIRIX 6.5.11 MIPS 7.1 2001-03-22, Robert Bruccoleri\nLinux 2.2.x Alpha 7.1 2001-01-23, Ryan Kirkpatrick\nLinux 2.2.x armv4l 7.1 2001-03-22, Mark Knox\nLinux 2.0.x MIPS 7.1 2001-03-30, Dominic Eidson\nLinux 2.2.18 PPC74xx 7.1 2001-03-19, Tom Lane\nLinux 2.2.x S/390 7.1 2000-11-17, Neale Ferguson\nLinux 2.2.15 Sparc 7.1 2001-01-30, Ryan Kirkpatrick\nLinux 2.2.16 x86 7.1 2001-03-19, Thomas Lockhart\nMacOS X Darwin PPC 7.1 2000-12-11, Peter Bierman\nNetBSD 1.5 Alpha 7.1 2001-03-22, Giles Lean\nNetBSD 1.5E arm32 7.1 2001-03-21, Patrick Welche\nNetBSD m68k 7.0 2000-04-10 (Henry has lost machine)\nNetBSD Sparc 7.0 2000-04-13, Tom I. Helbekkmo\nNetBSD VAX 7.1 2001-03-30, Tom I. Helbekkmo\nNetBSD 1.5 x86 7.1 2001-03-23, Giles Lean\nOpenBSD 2.8 Sparc 7.1 2001-03-23, Brandon Palmer\nOpenBSD 2.8 x86 7.1 2001-03-22, Brandon Palmer\nSCO OpenServer 5 x86 7.1 2001-03-13, Billy Allie\nSCO UnixWare 7.1.1 x86 7.1 2001-03-19, Larry Rosenman\nSolaris 2.7-8 Sparc 7.1 2001-03-22, Marc Fournier\nSolaris x86 7.1 2001-03-27, Mathijs Brands\nSunOS 4.1.4 Sparc 7.1 2001-03-23, Tatsuo Ishii\nWinNT/Cygwin x86 7.1 2001-03-16, Jason Tishler\n\nAnd the \"unsupported platforms\":\n\nDGUX m88k\nMkLinux DR1 PPC750 7.0 2000-04-13, Tatsuo Ishii\nNextStep x86\nQNX 4.25 x86 7.0 2000-04-01, Dr. Andreas Kardos\nSystem V R4 m88k\nSystem V R4 MIPS\nUltrix MIPS 7.1 2001-03-26, Alexander Klimov\nWindows/Win32 x86 7.1 2001-03-26, Magnus Hagander (clients only)\n", "msg_date": "Tue, 03 Apr 2001 15:31:25 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": true, "msg_subject": "Final call for platform testing" }, { "msg_contents": "On Tue, 3 Apr 2001, Thomas Lockhart wrote:\n\n[Snip]\n\n> Linux 2.0.x MIPS 7.1 2001-03-30, Dominic Eidson\n\nI just ran the \"make check\" (paralell regression tests) - instead of the\n\"make installcheck\" that I'd run previously...\n\n[nobody@web-cache regress]$ grep 'FAILED' regression.out\ntest geometry ... FAILED\ntest horology ... FAILED\n\nThe relevant diff for horology seem to be:\n\n[nobody@web-cache regress]$ diff -c ./expected/horology.out ./results/horology.out \n*** ./expected/horology.out Sun Dec 3 08:51:11 2000\n--- ./results/horology.out Tue Apr 3 11:38:27 2001\n***************\n*** 122,128 ****\n SELECT time with time zone '01:30' + interval '02:01' AS \"03:31:00-08\";\n 03:31:00-08 \n -------------\n! 03:31:00-08\n (1 row)\n \n SELECT time with time zone '01:30-08' - interval '02:01' AS \"23:29:00-08\";\n--- 122,128 ----\n SELECT time with time zone '01:30' + interval '02:01' AS \"03:31:00-08\";\n 03:31:00-08 \n -------------\n! 03:31:00-07\n (1 row)\n \n SELECT time with time zone '01:30-08' - interval '02:01' AS \"23:29:00-08\";\n***************\n*** 140,146 ****\n SELECT time with time zone '03:30' + interval '1 month 04:01' AS \"07:31:00-08\";\n 07:31:00-08 \n -------------\n! 07:31:00-08\n (1 row)\n \n SELECT interval '04:30' - time with time zone '01:02' AS \"+03:28\";\n--- 140,146 ----\n SELECT time with time zone '03:30' + interval '1 month 04:01' AS \"07:31:00-08\";\n 07:31:00-08 \n -------------\n! 07:31:00-07\n (1 row)\n \n SELECT interval '04:30' - time with time zone '01:02' AS \"+03:28\";\n\nSeems there's a couple of off-by-one errors on this platform (according to\nMarc, the same was the case for the geometry stuff...)\n\n\n-- \nDominic J. Eidson\n \"Baruk Khazad! Khazad ai-menu!\" - Gimli\n-------------------------------------------------------------------------------\nhttp://www.the-infinite.org/ http://www.the-infinite.org/~dominic/\n\n", "msg_date": "Tue, 3 Apr 2001 10:56:30 -0500 (CDT)", "msg_from": "\"Dominic J. Eidson\" <sauron@the-infinite.org>", "msg_from_op": false, "msg_subject": "Re: Final call for platform testing" }, { "msg_contents": ">\n> I just ran the \"make check\" (paralell regression tests) - instead of the\n> \"make installcheck\" that I'd run previously...\n>\n> [nobody@web-cache regress]$ grep 'FAILED' regression.out\n> test geometry ... FAILED\n> test horology ... FAILED\n>\n> The relevant diff for horology seem to be:\n\nI can't speak to the geo test failures, but the horology failures have to\ndo with the change from daylight saving change. Since we lost an hour\nthey will be off. Is this something to be looked at? The failure of the\ntest is to be expected, but it will cause some to worry when their\nregression tests fail.\n\nThoughts?\n\n- Brandon\n\n\nb. palmer, bpalmer@crimelabs.net\npgp: www.crimelabs.net/bpalmer.pgp5\n\n\n", "msg_date": "Tue, 3 Apr 2001 12:15:29 -0400 (EDT)", "msg_from": "bpalmer <bpalmer@crimelabs.net>", "msg_from_op": false, "msg_subject": "Re: Final call for platform testing" }, { "msg_contents": "\nHi all...\n\nI just installed 7.1RC1 on an sun enterprise 250 runnning debian 2.2r2\nlinux and everything went ok..\n\nI compiled with --enable-local and --with-java among other but it seems\nthat the ja created doesn't like international characters. I use the\npostgresql.jar found in /usr/local/postgresql/share/java/. When running\nusing this one i get ? instead of for example �,� or �. Changing to\njdbc7.0-1.2.jar from an rpminstall (mandrake 7.2) made everything work\njust fine.\n\nDid I do something wrong or miss something?\n\nThanks in advance!\n\nFredrik\n\n", "msg_date": "Tue, 3 Apr 2001 18:23:52 +0200 (MEST)", "msg_from": "Fredrik Hultkrantz <fjutt@blink.se>", "msg_from_op": false, "msg_subject": "International characters and the jdbcdriver" }, { "msg_contents": "> Thoughts?\n\nBesides \"Thomas is an idiot\"? :)\n\nNot sure how to test the \"implicit time zone\" feature of \"TIME WITH TIME\nZONE\" without risking the same kinds of trouble. Maybe the test should\nbe recast to using only comparisons with other date/time types which\nhave been shown to behave themselves across time zone boundaries. But\nI'm not sure that this would continue to really test the feature.\n\n - Thomas\n", "msg_date": "Tue, 03 Apr 2001 16:51:53 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": true, "msg_subject": "Re: Final call for platform testing" }, { "msg_contents": "\nOn Tue, 3 Apr 2001 14:10:12 -0700, Nathan Myers alluded:\n\n> I saw three separate reports of successful builds on Linux 2.4.2 on x86\n> (including mine), but it isn't listed here. \n\n[jeff@cairhien pronto]$ /var/postgresql/bin/psql -V\npsql (PostgreSQL) 7.1RC1\ncontains history support\n..\n\n[jeff@cairhien pronto]$ uname -a\nLinux cairhien 2.4.3 #1 Tue Apr 3 01:08:19 EDT 2001 i686 unknown\n\nJeff\n\n-- \n Jeff Duffy\n jeff@alanne.com\n\n", "msg_date": "03 Apr 2001 17:40:41 EDT", "msg_from": "\"Jeff Duffy\" <jeff@alanne.com>", "msg_from_op": false, "msg_subject": "Re: Final call for platform testing" }, { "msg_contents": "Thomas Lockhart <lockhart@alumni.caltech.edu> writes:\n> Not sure how to test the \"implicit time zone\" feature of \"TIME WITH TIME\n> ZONE\" without risking the same kinds of trouble. Maybe the test should\n> be recast to using only comparisons with other date/time types which\n> have been shown to behave themselves across time zone boundaries. But\n> I'm not sure that this would continue to really test the feature.\n\nI suspect we have no choice but to eliminate this particular test from\nthe regression suite. A test that fails for a few days around DST\nboundaries is one thing, but a test that fails for six months out of the\nyear is another.\n\nBTW, the reason HPUX was failing to fail is that its mktime() is picky\nabout tm_year; we were forgetting to subtract off 1900 in that\nparticular code path. Fixed. Now I get a failure like everyone else ;-)\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 03 Apr 2001 14:17:05 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Final call for platform testing " }, { "msg_contents": "On Tue, Apr 03, 2001 at 03:31:25PM +0000, Thomas Lockhart wrote:\n> \n> OK. So we are close to a final tally of supported machines.\n> ...\n> Here are the up-to-date platforms:\n> \n> AIX 4.3.3 RS6000 7.1 2001-03-21, Gilles Darold\n> BeOS 5.0.4 x86 7.1 2000-12-18, Cyril Velter\n> BSDI 4.01 x86 7.1 2001-03-19, Bruce Momjian\n> Compaq Tru64 4.0g Alpha 7.1 2001-03-19, Brent Verner\n> FreeBSD 4.3 x86 7.1 2001-03-19, Vince Vielhaber\n> HPUX PA-RISC 7.1 2001-03-19, 10.20 Tom Lane, 11.00 Giles Lean\n> IRIX 6.5.11 MIPS 7.1 2001-03-22, Robert Bruccoleri\n> Linux 2.2.x Alpha 7.1 2001-01-23, Ryan Kirkpatrick\n> Linux 2.2.x armv4l 7.1 2001-03-22, Mark Knox\n> Linux 2.0.x MIPS 7.1 2001-03-30, Dominic Eidson\n> Linux 2.2.18 PPC74xx 7.1 2001-03-19, Tom Lane\n> Linux 2.2.x S/390 7.1 2000-11-17, Neale Ferguson\n> Linux 2.2.15 Sparc 7.1 2001-01-30, Ryan Kirkpatrick\n> Linux 2.2.16 x86 7.1 2001-03-19, Thomas Lockhart\n> MacOS X Darwin PPC 7.1 2000-12-11, Peter Bierman\n> NetBSD 1.5 Alpha 7.1 2001-03-22, Giles Lean\n> NetBSD 1.5E arm32 7.1 2001-03-21, Patrick Welche\n> NetBSD m68k 7.0 2000-04-10 (Henry has lost machine)\n> NetBSD Sparc 7.0 2000-04-13, Tom I. Helbekkmo\n> NetBSD VAX 7.1 2001-03-30, Tom I. Helbekkmo\n> NetBSD 1.5 x86 7.1 2001-03-23, Giles Lean\n> OpenBSD 2.8 Sparc 7.1 2001-03-23, Brandon Palmer\n> OpenBSD 2.8 x86 7.1 2001-03-22, Brandon Palmer\n> SCO OpenServer 5 x86 7.1 2001-03-13, Billy Allie\n> SCO UnixWare 7.1.1 x86 7.1 2001-03-19, Larry Rosenman\n> Solaris 2.7-8 Sparc 7.1 2001-03-22, Marc Fournier\n> Solaris x86 7.1 2001-03-27, Mathijs Brands\n> SunOS 4.1.4 Sparc 7.1 2001-03-23, Tatsuo Ishii\n> WinNT/Cygwin x86 7.1 2001-03-16, Jason Tishler\n> \n> And the \"unsupported platforms\":\n> \n> DGUX m88k\n> MkLinux DR1 PPC750 7.0 2000-04-13, Tatsuo Ishii\n> NextStep x86\n> QNX 4.25 x86 7.0 2000-04-01, Dr. Andreas Kardos\n> System V R4 m88k\n> System V R4 MIPS\n> Ultrix MIPS 7.1 2001-03-26, Alexander Klimov\n> Windows/Win32 x86 7.1 2001-03-26, Magnus Hagander (clients only)\n\nI saw three separate reports of successful builds on Linux 2.4.2 on x86\n(including mine), but it isn't listed here. \n\n-- \nNathan Myers\nncm@zembu.com\n", "msg_date": "Tue, 3 Apr 2001 14:10:12 -0700", "msg_from": "ncm@zembu.com (Nathan Myers)", "msg_from_op": false, "msg_subject": "Re: Final call for platform testing" }, { "msg_contents": "> I saw three separate reports of successful builds on Linux 2.4.2 on x86\n> (including mine), but it isn't listed here.\n\nIt is listed in the comments in the real docs. At least one report was\nfor an extensively patched 2.4.2, and I'm not sure of the true lineage\nof the others.\n\nI *could* remove the version info from the x86 listing, and mention both\n2.2.x and 2.4.x in the comments.\n\n - Thomas\n", "msg_date": "Tue, 03 Apr 2001 23:19:04 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": true, "msg_subject": "Re: Final call for platform testing" }, { "msg_contents": "On Tue, Apr 03, 2001 at 11:19:04PM +0000, Thomas Lockhart wrote:\n> > I saw three separate reports of successful builds on Linux 2.4.2 on x86\n> > (including mine), but it isn't listed here.\n> \n> It is listed in the comments in the real docs. At least one report was\n> for an extensively patched 2.4.2, and I'm not sure of the true lineage\n> of the others.\n\nYou could ask. Just to ignore reports that you have asked for is not \npolite. My report was based on a virgin, unpatched 2.4.2 kernel, and \n(as noted) the Debian-packaged glibc-2.2.2. \n\nIf you are trying to trim your list, would be reasonable to drop \nLinux-2.0.x, because that version is not being maintained any more.\n\n> I *could* remove the version info from the x86 listing, and mention both\n> 2.2.x and 2.4.x in the comments.\n\nLinux-2.2 and Linux-2.4 are different codebases. It is worth noting,\nbesides, the glibc-version tested along with each Linux kernel version.\n\nNathan Myers\nncm@zembu.com\n", "msg_date": "Tue, 3 Apr 2001 17:54:05 -0700", "msg_from": "ncm@zembu.com (Nathan Myers)", "msg_from_op": false, "msg_subject": "Re: Re: Final call for platform testing" }, { "msg_contents": "\n> Compaq Tru64 4.0g Alpha 7.1 2001-03-19, Brent Verner\n\nWe ran these regression tests with both native cc and gcc -- worth\nmentioning that both work.\n\nAdriaan\n", "msg_date": "Wed, 04 Apr 2001 08:14:47 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": false, "msg_subject": "Re: Final call for platform testing" }, { "msg_contents": ">\n>Here are the up-to-date platforms:\n>\n>AIX 4.3.3 RS6000 7.1 2001-03-21, Gilles Darold\n>BeOS 5.0.4 x86 7.1 2000-12-18, Cyril Velter\n\n I just checked RC2 on BeOS and everything is OK except the Horlogy test \n(regarding previous discussions, it seems to be normal ?)\n\n cyril\n\n\n\nselect version();\n version\n--------------------------------------------------------------------\n PostgreSQL 7.1RC2 on i586-pc-beos, compiled by GCC 2.9-beos-991026\n(1 row) \n\n\n\n$ more regression.diffs\n*** ./expected/horology.out Sun Dec 3 15:51:11 2000\n--- ./results/horology.out Wed Apr 4 09:24:27 2001\n***************\n*** 122,128 ****\n SELECT time with time zone '01:30' + interval '02:01' AS \"03:31:00-08\";\n 03:31:00-08\n -------------\n! 03:31:00-08\n (1 row)\n\n SELECT time with time zone '01:30-08' - interval '02:01' AS \"23:29:00-08\";\n--- 122,128 ----\n SELECT time with time zone '01:30' + interval '02:01' AS \"03:31:00-08\";\n 03:31:00-08\n -------------\n! 03:31:00-07\n (1 row)\n\n SELECT time with time zone '01:30-08' - interval '02:01' AS \"23:29:00-08\";\n***************\n*** 140,146 ****\n SELECT time with time zone '03:30' + interval '1 month 04:01' AS \"07:31:00-\n08\";\n 07:31:00-08\n -------------\n! 07:31:00-08\n (1 row)\n\n SELECT interval '04:30' - time with time zone '01:02' AS \"+03:28\";\n--- 140,146 ----\n SELECT time with time zone '03:30' + interval '1 month 04:01' AS \"07:31:00-\n08\";\n 07:31:00-08\n -------------\n! 07:31:00-07\n (1 row)\n\n SELECT interval '04:30' - time with time zone '01:02' AS \"+03:28\";\n\n====================================================================== \n\n\n\n", "msg_date": "Wed, 04 Apr 2001 09:39:37 +0200", "msg_from": "Cyril VELTER <cyril.velter@libertysurf.fr>", "msg_from_op": false, "msg_subject": "Re: Final call for platform testing" }, { "msg_contents": " > Solaris 2.7-8 Sparc 7.1 2001-03-22, Marc Fournier\n\nI've reported Solaris 2.6 Sparc as working on a post-RC1 snapshot.\n-- \nPete Forman -./\\.- Disclaimer: This post is originated\nWesternGeco -./\\.- by myself and does not represent\npete.forman@westerngeco.com -./\\.- opinion of Schlumberger, Baker\nhttp://www.crosswinds.net/~petef -./\\.- Hughes or their divisions.\n", "msg_date": "Wed, 4 Apr 2001 11:29:47 +0100", "msg_from": "Pete Forman <pete.forman@westerngeco.com>", "msg_from_op": false, "msg_subject": "Re: Final call for platform testing" }, { "msg_contents": "On Tue, 3 Apr 2001, Nathan Myers wrote:\n\n> On Tue, Apr 03, 2001 at 11:19:04PM +0000, Thomas Lockhart wrote:\n> > > I saw three separate reports of successful builds on Linux 2.4.2 on x86\n> > > (including mine), but it isn't listed here.\n> >\n> > It is listed in the comments in the real docs. At least one report was\n> > for an extensively patched 2.4.2, and I'm not sure of the true lineage\n> > of the others.\n>\n> You could ask. Just to ignore reports that you have asked for is not\n> polite. My report was based on a virgin, unpatched 2.4.2 kernel, and\n> (as noted) the Debian-packaged glibc-2.2.2.\n\nI just looked at the regression test database and didn't see any report\nfrom you. Did you not submit it?\n\n http://www.postgresql.org/~vev/regress/\n\nVince.\n-- \n==========================================================================\nVince Vielhaber -- KA8CSH email: vev@michvhf.com http://www.pop4.net\n 56K Nationwide Dialup from $16.00/mo at Pop4 Networking\n Online Campground Directory http://www.camping-usa.com\n Online Giftshop Superstore http://www.cloudninegifts.com\n==========================================================================\n\n\n\n", "msg_date": "Wed, 4 Apr 2001 06:47:45 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": false, "msg_subject": "Re: Re: Final call for platform testing" }, { "msg_contents": "> You could ask. Just to ignore reports that you have asked for is not\n> polite. My report was based on a virgin, unpatched 2.4.2 kernel, and\n> (as noted) the Debian-packaged glibc-2.2.2.\n\nThe simple fact is that the vast majority of Linux boxes are not yet\nrunning 2.4 kernels, so imho it is not a big issue. I've mentioned 2.4\nas well as 2.2 kernels in the (very short) tabular summary, so any\ninterested user would know that both will work for them. 2.0.x kernels\nwork as well, so I may just drop version info from the first field in\nthe summary altogether.\n\nYou can inspect the current info in the developer's docs, posted on the\nweb site.\n\nThe Right Place for more extensive information might be FAQ_Linux, which\nwas last updated three years ago. Would someone have time to go through\nthat and scrub the info?\n\n - Thomas\n", "msg_date": "Wed, 04 Apr 2001 14:33:27 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": true, "msg_subject": "Re: Final call for platform testing" }, { "msg_contents": "On Wednesday 04 April 2001 13:29, Pete Forman wrote:\n> > Solaris 2.7-8 Sparc 7.1 2001-03-22, Marc Fournier\n>\n> I've reported Solaris 2.6 Sparc as working on a post-RC1 snapshot.\n\nSame for Solaris 8 Sparc, but only tested with RC1.\n\n-- \nEl mejor sistema operativo es aquel que te da de comer.\nCuida tu dieta.\n-----------------------------------------------------------------\nMartin Marques | mmarques@unl.edu.ar\nProgramador, Administrador | Centro de Telematica\n Universidad Nacional\n del Litoral\n-----------------------------------------------------------------\n", "msg_date": "Wed, 4 Apr 2001 18:37:10 +0300", "msg_from": "=?iso-8859-1?q?Mart=EDn=20Marqu=E9s?= <martin@bugs.unl.edu.ar>", "msg_from_op": false, "msg_subject": "Re: Re: Final call for platform testing" } ]
[ { "msg_contents": "\n> > It is intuitive. The bug was iirc, that you saw 2 versions of the same row\n> > in the second select statement (= 2 rows returned by second select).\n> \n> I think we should be extremely wary of assuming that we have a clear\n> characterization of \"what the bug is\", let alone \"how to fix it\".\n> The real issue here is that SELECT has different MVCC visibility rules\n> from UPDATE and SELECT FOR UPDATE. I suspect that that *must* be so\n> in any mode that allows more concurrency than full serializable mode.\n\nYes, definitely.\n\n> Thus, the question we are really facing is how we might alter the\n> visibility rules in a way that will make the results more intuitive\n> and/or useful while still allowing concurrency.\n> \n> This will take thought, research and discussion. A quick fix is the\n> last thing that should be on our minds.\n\n From my latest tests( see following post), I tend to agree, that this is \nextremely sensitive :-( \nI do however think that Vadim's patch description was the correct thing to do.\n\nThe problem case seems to be when the function is not executed inside a txn.\nI was not able to reproduce any failure, when inside txns, since the first update \nor select for update blocks the rest.\n\nAndreas\n", "msg_date": "Fri, 30 Mar 2001 17:25:39 +0200", "msg_from": "Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at>", "msg_from_op": true, "msg_subject": "AW: AW: Re: [SQL] possible row locking bug in 7.0.3 & 7\n\t.1" } ]
[ { "msg_contents": "\n> If I remember correctly, UPDATE establishes a lock on the affected rows, \n> which will block another UPDATE on the same rows for the duration of the \n> transaction. If that's true, shouldn't I be able to achieve my desired \n> behavior by removing the initial as follows:\n> \n> create function nextid( varchar(32)) returns int8 as '\n> update idseq set id = id + 1 where name = $1::text;\n> select id from idseq where name = $1::text;\n> ' language 'sql';\n\nYes, better, but be sure, to only use this function from inside a transaction.\nIf you use it in autocommit mode (no begin work) you might in theory read a row, \nthat another session modified between the two lines. \n\n> Or, would I still have to add FOR UPDATE to that final SELECT?\n\nNow, this certainly looks very funny. You actually get reasonable results only \nif you do include the \"for update\" with RC1 sources .\n\nTo the rest on the list:\nTry the above example by adding a lock between the two lines:\n\ncreate function nextid( varchar(32)) returns int8 as '\n update idseq set id = id + 1 where name = $1::text;\n select * from lock1;\n select id from idseq where name = $1::text for update;\n ' language 'sql';\n\n\t\t\t\t\t\tsession1:\n\t\t\t\t\t\t\tbegin work;\n\t\t\t\t\t\t\tlock table lock1 in access exclusive mode;\nsession 2:\n\tnot in txn: select nextid('one'); // this blocks\n\t\t\t\t\t\t\tselect nextid('one');\n\t\t\t\t\t\t\tcommit work;\n\nAnd stare at the results you get with and without for update :-(\nSomething is definitely fishy with the visibility of SELECT here. \n\nAndreas\n", "msg_date": "Fri, 30 Mar 2001 17:42:08 +0200", "msg_from": "Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at>", "msg_from_op": true, "msg_subject": "AW: Re: [SQL] possible row locking bug in 7.0.3 & 7.1" } ]
[ { "msg_contents": "\n> To the rest on the list:\n> Try the above example by adding a lock between the two lines:\n> \n> create function nextid( varchar(32)) returns int8 as '\n> update idseq set id = id + 1 where name = $1::text;\n> select * from lock1;\n> select id from idseq where name = $1::text for update;\n> ' language 'sql';\n> \n> \t\t\t\t\t\tsession1:\n> \t\t\t\t\t\t\tbegin work;\n> \t\t\t\t\t\t\tlock table lock1 in access exclusive mode;\n> session 2:\n> \tnot in txn: select nextid('one'); // this blocks\n> \t\t\t\t\t\t\tselect nextid('one');\n> \t\t\t\t\t\t\tcommit work;\n> \n> And stare at the results you get with and without for update :-(\n> Something is definitely fishy with the visibility of SELECT here. \n\nWithout \"for update\" I see a tuple in session2 from before session1 began.\nAfter both complete, the net result is correct (id is incremented by 2).\nThis is very interesting, unfortunately I must leave Internet access until \nmonday since my daughter called me home, and mail is so dead slow,\nthat I did not even receive my last mails yet :-(\n\nAndreas\n", "msg_date": "Fri, 30 Mar 2001 18:04:08 +0200", "msg_from": "Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at>", "msg_from_op": true, "msg_subject": "AW: Re: [SQL] possible row locking bug in 7.0.3 & 7.1" } ]
[ { "msg_contents": "\n( I deleted the email accidentially)\n\nI have a Cobalt 2.0.x MIPS box that is currently compiling the latest CVS\nof PostgreSQL ... I'll let you know in a few hours how it went.\n\n# uname -a\nLinux web-cache 2.0.34C52_SK #1 Tue Nov 30 18:14:40 PST 1999 mips unknown\n\n\n-- \nDominic J. Eidson\n \"Baruk Khazad! Khazad ai-menu!\" - Gimli\n-------------------------------------------------------------------------------\nhttp://www.the-infinite.org/ http://www.the-infinite.org/~dominic/\n\n", "msg_date": "Fri, 30 Mar 2001 11:31:33 -0600 (CST)", "msg_from": "\"Dominic J. Eidson\" <sauron@the-infinite.org>", "msg_from_op": true, "msg_subject": "Re: third call for platforms..." }, { "msg_contents": "On Fri, 30 Mar 2001, Dominic J. Eidson wrote:\n\n> I have a Cobalt 2.0.x MIPS box that is currently compiling the latest CVS\n> of PostgreSQL ... I'll let you know in a few hours how it went.\n\nCompiled fine, and passed all but the geometry regression test:\n\n[root@web-cache regress]# more regression.out\ntest boolean ... ok\ntest char ... ok\ntest name ... ok\ntest varchar ... ok\ntest text ... ok\ntest int2 ... ok\ntest int4 ... ok\ntest int8 ... ok\ntest oid ... ok\ntest float4 ... ok\ntest float8 ... ok\ntest bit ... ok\ntest numeric ... ok\ntest strings ... ok\ntest numerology ... ok\ntest point ... ok\ntest lseg ... ok\ntest box ... ok\ntest path ... ok\ntest polygon ... ok\ntest circle ... ok\ntest date ... ok\ntest time ... ok\ntest timestamp ... ok\ntest interval ... ok\ntest abstime ... ok\ntest reltime ... ok\ntest tinterval ... ok\ntest inet ... ok\ntest comments ... ok\ntest oidjoins ... ok\ntest type_sanity ... ok\ntest opr_sanity ... ok\ntest geometry ... FAILED\ntest horology ... ok\ntest create_function_1 ... ok\ntest create_type ... ok\ntest create_table ... ok\ntest create_function_2 ... ok\ntest copy ... ok\ntest constraints ... ok\ntest triggers ... ok\ntest create_misc ... ok\ntest create_aggregate ... ok\ntest create_operator ... ok\ntest create_index ... ok\ntest inherit ... ok\ntest create_view ... ok\ntest sanity_check ... ok\ntest errors ... ok\ntest select ... ok\ntest select_into ... ok\ntest select_distinct ... ok\ntest select_distinct_on ... ok\ntest select_implicit ... ok\ntest select_having ... ok\ntest subselect ... ok\ntest union ... ok\ntest case ... ok\ntest join ... ok\ntest aggregates ... ok\ntest transactions ... ok\ntest random ... ok\ntest portals ... ok\ntest arrays ... ok\ntest btree_index ... ok\ntest hash_index ... ok\ntest misc ... ok\ntest select_views ... ok\ntest alter_table ... ok\ntest portals_p2 ... ok\ntest rules ... ok\ntest foreign_key ... ok\ntest limit ... ok\ntest plpgsql ... ok\ntest temp ... ok\n\nAttached, find the regression.diffs file.\n\n\n-- \nDominic J. Eidson\n \"Baruk Khazad! Khazad ai-menu!\" - Gimli\n-------------------------------------------------------------------------------\nhttp://www.the-infinite.org/ http://www.the-infinite.org/~dominic/", "msg_date": "Fri, 30 Mar 2001 14:20:17 -0600 (CST)", "msg_from": "\"Dominic J. Eidson\" <sauron@the-infinite.org>", "msg_from_op": true, "msg_subject": "Re: Re: third call for platforms..." }, { "msg_contents": "> I have a Cobalt 2.0.x MIPS box that is currently compiling the latest CVS\n> of PostgreSQL ... I'll let you know in a few hours how it went.\n\nGreat!\n\n - Thomas\n", "msg_date": "Fri, 30 Mar 2001 20:46:07 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": false, "msg_subject": "Re: third call for platforms..." } ]
[ { "msg_contents": "> > > I doubt if it's a bug of SELECT. Well what\n> > > 'concurrent UPDATE then SELECT FOR UPDATE +\n> > > SELECT' return ?\n> > \n> > I'm going to add additional check to heapgettup and\n> > heap_fetch:\n> \n> SELECT seems to be able to return a different result\n> from that of preceding SELECT FOR UPDATE even after\n> applying your change.\n\nOh, you're right. Well, if we really want that SELECT\nreturns the same result as SELECT FOR UPDATE *in functions*\n(out of functions results are already same) then we have to\nadd some modifications to fix proposed:\n\n1. If newer version of visible tuple T is marked for update by\nus *before* query began then do not return T.\n\n2. If tuple T1 is *not visible* because of it was inserted by\nconcurrent committed TX then check if it's marked for update\nby current TX *before* query began and return *this* tuple\nversion if yes.\n\nThis will be in accordance with standard which requires\nus return committed (whenever) rows in READ COMMITTED mode.\nIn fact, in this mode our SELECTs provide higher isolation\nthan required by standard returning rows committed *before*\nquery began. Why? Because of SELECT doesn't lock rows and\nthe same row may be visited by SELECT in join queries\nmany times - so we have to be protected against concurrent\nupdates. SELECT FOR UPDATE protects us BUT if query itself\ncalls some functions which updates queried table then currently\nwe may lose information that tuple was marked for update before\nquery began - so updating tuple inserted by concurrent committed\nTX and marked for update by us we would have to save its t_cmax\nin t_cmin (and either add new flag to t_infomask or don't turn\nOFF HEAP_MARKED_FOR_UPDATE in this case).\n\nThis is not what I would like to do in 7.1\n\n> SELECT doesn't seem guilty but the result is far \n> from intuitive.\n\nI think that SELECT is guilty. At least returning two versions\nof the same row! (One that could be fixed easy).\n\nVadim\n", "msg_date": "Fri, 30 Mar 2001 10:54:48 -0800", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: [HACKERS] Re: possible row locking bug in 7.0.3 & 7.1" } ]
[ { "msg_contents": "Tricia Holben of Great Bridge just pointed out to me a rather nasty\nproblem that's exposed by trying to pg_dump and reload the regression\ntest database. The regression tests include\n\nCREATE FUNCTION widget_in(opaque)\n RETURNS widget\n AS '/home/postgres/pgsql/src/test/regress/regress.sl'\n LANGUAGE 'c';\nNOTICE: ProcedureCreate: type 'widget' is not yet defined\n\nCREATE FUNCTION widget_out(opaque)\n RETURNS opaque\n AS '/home/postgres/pgsql/src/test/regress/regress.sl'\n LANGUAGE 'c';\n\nCREATE TYPE widget (\n internallength = 24, \n input = widget_in,\n output = widget_out,\n alignment = double\n);\n\nwhich is considered a correct approach to defining I/O procedures for\nuser-defined types; notice that the code goes out of its way to allow\ntype \"widget\" to be referenced before it is defined.\n\nUnfortunately, since the shell pg_type entry for type widget is created\nbefore the pg_proc entry for widget_in is, the OID assignment sequence\nis: widget, widget_in, widget_out. When pg_dump dumps these objects in\nOID order, it dumps the CREATE TYPE command first --- an ordering that\nwill fail upon reload.\n\n7.0.* and before do not have this problem because they dump type\ndefinitions after function definitions, regardless of OIDs.\n\nI can think of a couple of ways to deal with this, the simplest being\nto say \"don't do that\" --- ie, define widget_in with result type\n\"opaque\" rather than \"widget\". That's pretty ugly and will likely\nbreak people's 7.0 dump scripts all by itself. A more promising idea\nis to hack function creation so that the OID assigned to the function\nis lower than the OIDs assigned to any shell types created when the\nfunction is defined. Or we could try to hack pg_dump to fix this,\nbut that doesn't seem appetizing.\n\nThere may be similar problems with other shell-catalog-entry cases;\nhaven't looked yet.\n\nIs this a release stopper? I'm inclined to think it is.\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 30 Mar 2001 14:55:04 -0500", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "7.1 pg_dump fails for user-defined types (release stopper?)" }, { "msg_contents": "> A more promising idea is to hack function creation\n> so that the OID assigned to the function is lower\n> than the OIDs assigned to any shell types created\n> when the function is defined. Or we could try to\n> hack pg_dump to fix this, but that doesn't seem\n> appetizing.\n\nRequiring OID ordering would open up a new can of worms.\n\nWhat happens if the user does a drop/create on the function after\ncreating it?\n\nThe function could potentially be recreated with a higher OID and then\nthe user would be in the same situation.\n\nIf the system requires (or works around) creation ordering when creating\nfunctions and types, ISTM that pg_dump should have to do the same.\n\ndarrenk\n\n", "msg_date": "Fri, 30 Mar 2001 15:30:41 -0500", "msg_from": "\"Darren King\" <darrenk@insightdist.com>", "msg_from_op": false, "msg_subject": "RE: 7.1 pg_dump fails for user-defined types (release stopper?)" }, { "msg_contents": "At 14:55 30/03/01 -0500, Tom Lane wrote:\n>\n>A more promising idea\n>is to hack function creation so that the OID assigned to the function\n>is lower than the OIDs assigned to any shell types created when the\n>function is defined.\n\nThis seems hard; would it be better to have the CREATE TYPE use a new OID,\nand fixup the refs?\n\n\n>Or we could try to hack pg_dump to fix this,\n>but that doesn't seem appetizing.\n\nThis *may* not be all that hard; there is a currently unused (always NULL)\nparameter on the pg_dump ArchiveEntry calls intended for extra\ndependencies. For UDTs, we could set the this to be the max OID that\nreferences the type (or a list of OIDs, if we had to), then modify the\npg_restore sort code to check these values if not NULL. ie.\n\n (TOC2 > TOC1) \n iff (Max(TOC2.OID, TOC2.DEPS) > Max(TOC1.OID, TOC1.DEPS))\n OR ( Max(TOC2.OID, TOC2.DEPS) = Max(TOC1.OID, TOC1.DEPS)\n And TOC1.OID = Max(TOC2.DEPS)\n )\n\n Where DEPS is a list of OIDs the TOC entry depends on.\n\n(I *think* that's right...).\n\nSince this will only be used when the args is non-null, this code would\nonly be activated in the current broken case.\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Sat, 31 Mar 2001 15:49:26 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": false, "msg_subject": "Re: 7.1 pg_dump fails for user-defined types (release stopper?)" }, { "msg_contents": "At 15:49 31/03/01 +1000, Philip Warner wrote:\n>\n> (TOC2 > TOC1) \n> iff (Max(TOC2.OID, TOC2.DEPS) > Max(TOC1.OID, TOC1.DEPS))\n> OR ( Max(TOC2.OID, TOC2.DEPS) = Max(TOC1.OID, TOC1.DEPS)\n> And TOC1.OID = Max(TOC2.DEPS)\n> )\n>\n> Where DEPS is a list of OIDs the TOC entry depends on.\n>\n>(I *think* that's right...).\n>\n\nThis will of course not handle multi-level dependencies. But for the simple\nordering we are talking about, I think it will work. It can be extended\nlater when we want to walk a complete dependency tree.\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Sat, 31 Mar 2001 15:58:04 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": false, "msg_subject": "Re: Re: 7.1 pg_dump fails for user-defined types (release stopper?)" } ]
[ { "msg_contents": "> > It is intuitive. The bug was iirc, that you saw 2 versions \n> > of the same row in the second select statement (= 2 rows\n> > returned by second select).\n> \n> I think we should be extremely wary of assuming that we have a clear\n> characterization of \"what the bug is\", let alone \"how to fix it\".\n> The real issue here is that SELECT has different MVCC visibility rules\n> from UPDATE and SELECT FOR UPDATE. I suspect that that *must* be so\n\nThis is not correct - SELECT has same rules. Are you able to reproduce\nthis bad behaviour without running queries in functions? I assume\nthe answer is NO. I just overlooked function case two years ago.\nBut SELECT/UPDATE visibility rules are same!\nEver wonder why in SERIALIZABLE mode UPDATE/SELECT_FOR_UPDATE cause\nrollback in the event of concurrent modification? Because of concurrent\nmodifications make visibility of SELECT and UPDATE different and this\nmeans *unconsistent* view of database for applications.\nIn READ COMMITTED mode a query must see changes made by previous\nqueries - the only one rule we have to follow to provide consistent\nresult for applications.\n\n> in any mode that allows more concurrency than full serializable mode.\n> Thus, the question we are really facing is how we might alter the\n> visibility rules in a way that will make the results more intuitive\n> and/or useful while still allowing concurrency.\n> \n> This will take thought, research and discussion. A quick fix is the\n> last thing that should be on our minds.\n\nI agreed to leave it as Known Bug for 7.1.\n\n> A first question: where did the MVCC rules come from \n> originally, anyway?\n\n From the fact that I've used Oracle before Postgres'95,\nliked it and had time to read its documentation -:)\n\n> Is there any academic research to look at?\n\nThere is academic Theorem of Serializability but it's\ndifferent from SERIALIZABLE mode definitions in standard.\nProbably, this difference was caused by lobbying from\nOracle...\n\nVadim\n", "msg_date": "Fri, 30 Mar 2001 12:02:34 -0800", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: AW: Re: [SQL] possible row locking bug in 7.0.3 & 7\n\t.1" }, { "msg_contents": "> -----Original Message-----\n> From: Mikheev, Vadim [mailto:vmikheev@SECTORBASE.COM]\n> \n> > > It is intuitive. The bug was iirc, that you saw 2 versions \n> > > of the same row in the second select statement (= 2 rows\n> > > returned by second select).\n> > \n> > I think we should be extremely wary of assuming that we have a clear\n> > characterization of \"what the bug is\", let alone \"how to fix it\".\n> > The real issue here is that SELECT has different MVCC visibility rules\n> > from UPDATE and SELECT FOR UPDATE. I suspect that that *must* be so\n> \n> This is not correct - SELECT has same rules. Are you able to reproduce\n> this bad behaviour without running queries in functions? I assume\n> the answer is NO. I just overlooked function case two years ago.\n> But SELECT/UPDATE visibility rules are same!\n\nYes, there seems to be a confusion about visibility.\nEach query in SERIALIZABLE isolation level uses a common snapshot\nfor a TX. Each query in READ COMMITTED isolation level uses its own\nsnapshot. It seems the only difference between SERIALZABLE and READ\nCOMMITTED. But there's a sort of SERIALIZABLE world inside functions\neven under READ COMMITTED mode.\n\nregards,\nHiroshi Inoue\n \n", "msg_date": "Sat, 31 Mar 2001 10:15:59 +0900", "msg_from": "\"Hiroshi Inoue\" <Inoue@tpf.co.jp>", "msg_from_op": false, "msg_subject": "RE: AW: Re: [SQL] possible row locking bug in 7.0.3 & 7.1 " } ]
[ { "msg_contents": "> > This will take thought, research and discussion. A quick fix is the\n> > last thing that should be on our minds.\n> \n> From my latest tests( see following post), I tend to agree,\n> that this is extremely sensitive :-( \n> I do however think that Vadim's patch description was the\n> correct thing to do.\n\nTo avoid double tuple versions return - maybe.\nTo get same results from SELECT and SELECT FOR UPDATE in functions -\nno time for 7.1.\n\n> The problem case seems to be when the function is not \n> executed inside a txn.\n\nAny query is executed inside TX. All queries of a function\nare executed in the same TX.\n\nVadim\n", "msg_date": "Fri, 30 Mar 2001 12:08:44 -0800", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: AW: Re: [SQL] possible row locking bug in 7.0.3 & 7\n\t .1" } ]
[ { "msg_contents": "> I can think of a couple of ways to deal with this, the simplest being\n> to say \"don't do that\" --- ie, define widget_in with result type\n> \"opaque\" rather than \"widget\". That's pretty ugly and will likely\n\nWhy is it ugly? Why not update RETURNS type for XXX_in function when\ncreating type?\n\n> break people's 7.0 dump scripts all by itself. A more promising idea\n\nIs 7.1 pg_dump able to dump 7.0 database?..\n\n> is to hack function creation so that the OID assigned to the function\n> is lower than the OIDs assigned to any shell types created when the\n> function is defined.\n\nHow much lower? How to guarantee that OID of XXX_out created sometime\nafter XXX_in will be lower than XXX' OID?\n\n> Or we could try to hack pg_dump to fix this,\n> but that doesn't seem appetizing.\n\nIt looks like also right way to follow - pg_dump should care about\nsystem dependancies.\n\n> There may be similar problems with other shell-catalog-entry cases;\n> haven't looked yet.\n> \n> Is this a release stopper? I'm inclined to think it is.\n\nYes, looks like that one -:(\n\nVadim\n", "msg_date": "Fri, 30 Mar 2001 12:27:53 -0800", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: 7.1 pg_dump fails for user-defined types (release s\n\ttopper?)" } ]
[ { "msg_contents": "Anything dependent upon the original function in your example is busted \nanyhow, regardless of a dump/reload:\n\nCREATE function test() returns int4 AS 'SELECT 1' LANGUAGE 'SQL';\nCREATE\n\nSELECT test()\n test\n------\n 1\n(1 row)\n\nCREATE VIEW test_view AS SELECT test();\nCREATE\n\nSELECT * FROM test_view;\n test\n------\n 1\n(1 row)\n\nDROP FUNCTION test();\nDROP\n\nCREATE function test() returns int4 AS 'SELECT 1' LANGUAGE 'SQL';\nCREATE\n\nSELECT * FROM test_view;\nERROR: Function OID 387520 does not exist\n\nSo dumping OID order is least of the users' problems. Its hard to come up \nwith examples where an object dependent upon another in a *working* \ndatabase has a lesser OID. So the regression suite really did its job in \nthis case.\n\nMike Mascari\nmascarm@mascari.com\n\n-----Original Message-----\nFrom:\tDarren King [SMTP:darrenk@insightdist.com]\nSent:\tFriday, March 30, 2001 3:31 PM\nTo:\tPgsql-Hackers@Postgresql. Org\nSubject:\tRE: [HACKERS] 7.1 pg_dump fails for user-defined types (release \nstopper?)\n\n> A more promising idea is to hack function creation\n> so that the OID assigned to the function is lower\n> than the OIDs assigned to any shell types created\n> when the function is defined. Or we could try to\n> hack pg_dump to fix this, but that doesn't seem\n> appetizing.\n\nRequiring OID ordering would open up a new can of worms.\n\nWhat happens if the user does a drop/create on the function after\ncreating it?\n\nThe function could potentially be recreated with a higher OID and then\nthe user would be in the same situation.\n\nIf the system requires (or works around) creation ordering when creating\nfunctions and types, ISTM that pg_dump should have to do the same.\n\ndarrenk\n\n", "msg_date": "Fri, 30 Mar 2001 15:50:02 -0500", "msg_from": "Mike Mascari <mascarm@mascari.com>", "msg_from_op": true, "msg_subject": "RE: 7.1 pg_dump fails for user-defined types (release stopper?)" } ]
[ { "msg_contents": "Well, once I figured out a few things about how to use the Unix part \nof MacOS X it seemed to work without a hitch in the regression \ntesting. I need to figure out how to make the system start it up \nautomatically at boot time still, but then I need to figure out how \nto do double sided printing on an HP 5si and several other things too.\n\nPrintout at end. Details available if needed.\n\nBTW the offer of testing on NetBSD/mac68k stands. 250MB is plenty if \nyou don't use X, and I have some extra disks anyway. I'll probably \ntry NetBSD/macppc sometime in the next week or so.\n\n>At 5:14 PM +0000 3/26/01, Thomas Lockhart wrote:\n>>NetBSD m68k 7.0 2000-04-10, Henry B. Hotz\n>\n>I no longer have a 68k machine that's fast enough to reasonably test \n>PG on. I have a IIcx that sometimes serves as a router, but I'm \n>using some second-generation powermac's mostly now. (You still \n>have that Centris in your closet Tom?)\n>\n>I *did* just get MacOS X this weekend though and if I get it working \n>on my work G4 maybe I could give it a try there.\n\nScreen output from regression test.\n\n>[localhost:src/test/regress] hotz% make check\n>sed -e 's,@bindir@,/usr/local/pgsql/bin,g' \\\n> -e 's,@libdir@,/usr/local/pgsql/lib,g' \\\n> -e 's,@datadir@,/usr/local/pgsql/share,g' \\\n> -e 's/@VERSION@/7.1RC1/g' \\\n> -e 's/@host_tuple@/powerpc-apple-darwin1.3/g' \\\n> -e 's,@GMAKE@,make,g' \\\n> -e 's/@enable_shared@/yes/g' \\\n> -e 's/@GCC@/yes/g' \\\n> pg_regress.sh >pg_regress\n>chmod a+x pg_regress\n>cc -traditional-cpp -g -O2 -Wall -Wmissing-prototypes \n>-Wmissing-declarations -bundle -undefined suppress \n>-I../../../src/interfaces/libpq -I../../../src/include -c -o \n>regress.o regress.c\n>cc -traditional-cpp -g -O2 -Wall -Wmissing-prototypes \n>-Wmissing-declarations -bundle -undefined suppress -bundle \n>-undefined suppress -o regress.so regress.o\n>sed -e \n>'s,@abs_srcdir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress,g' \n>-e \n>'s,@abs_builddir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress, \n>g' -e 's/@DLSUFFIX@/.so/g' input/copy.source >sql/copy.sql\n>sed -e \n>'s,@abs_srcdir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress,g' \n>-e \n>'s,@abs_builddir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress, \n>g' -e 's/@DLSUFFIX@/.so/g' \n>input/create_function_1.source >sql/create_function_1.sql\n>sed -e \n>'s,@abs_srcdir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress,g' \n>-e \n>'s,@abs_builddir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress, \n>g' -e 's/@DLSUFFIX@/.so/g' \n>input/create_function_2.source >sql/create_function_2.sql\n>sed -e \n>'s,@abs_srcdir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress,g' \n>-e \n>'s,@abs_builddir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress, \n>g' -e 's/@DLSUFFIX@/.so/g' input/misc.source >sql/misc.sql\n>sed -e \n>'s,@abs_srcdir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress,g' \n>-e \n>'s,@abs_builddir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress, \n>g' -e 's/@DLSUFFIX@/.so/g' \n>input/constraints.source >sql/constraints.sql\n>sed -e \n>'s,@abs_srcdir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress,g' \n>-e \n>'s,@abs_builddir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress, \n>g' -e 's/@DLSUFFIX@/.so/g' output/copy.source >expected/copy.out\n>sed -e \n>'s,@abs_srcdir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress,g' \n>-e \n>'s,@abs_builddir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress, \n>g' -e 's/@DLSUFFIX@/.so/g' \n>output/create_function_1.source >expected/create_function_1.out\n>sed -e \n>'s,@abs_srcdir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress,g' \n>-e \n>'s,@abs_builddir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress, \n>g' -e 's/@DLSUFFIX@/.so/g' \n>output/create_function_2.source >expected/create_function_2.out\n>sed -e \n>'s,@abs_srcdir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress,g' \n>-e \n>'s,@abs_builddir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress, \n>g' -e 's/@DLSUFFIX@/.so/g' output/misc.source >expected/misc.out\n>sed -e \n>'s,@abs_srcdir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress,g' \n>-e \n>'s,@abs_builddir@,/Users/hotz/dist/postgresql-7.1RC1/src/test/regress, \n>g' -e 's/@DLSUFFIX@/.so/g' \n>output/constraints.source >expected/constraints.out\n>make -C ../../../contrib/spi REFINT_VERBOSE=1 refint.so autoinc.so\n>cc -traditional-cpp -g -O2 -Wall -Wmissing-prototypes \n>-Wmissing-declarations -bundle -undefined suppress -I. \n>-I../../src/include -DREFINT_VERBOSE -c -o refint.o refint.c\n>cc -traditional-cpp -g -O2 -Wall -Wmissing-prototypes \n>-Wmissing-declarations -bundle -undefined suppress -bundle \n>-undefined suppress -o refint.so refint.o\n>cc -traditional-cpp -g -O2 -Wall -Wmissing-prototypes \n>-Wmissing-declarations -bundle -undefined suppress -I. \n>-I../../src/include -DREFINT_VERBOSE -c -o autoinc.o autoinc.c\n>cc -traditional-cpp -g -O2 -Wall -Wmissing-prototypes \n>-Wmissing-declarations -bundle -undefined suppress -bundle \n>-undefined suppress -o autoinc.so autoinc.o\n>rm refint.o autoinc.o\n>/bin/sh ./pg_regress --temp-install --top-builddir=../../.. \n>--schedule=./parallel_schedule --multibyte=\n>============== creating temporary installation ==============\n>============== initializing database system ==============\n>============== starting postmaster ==============\n>running on port 65432 with pid 1407\n>============== creating database \"regression\" ==============\n>CREATE DATABASE\n>============== installing PL/pgSQL ==============\n>============== running regression test queries ==============\n>parallel group (13 tests): text name varchar int2 boolean int8 oid \n>float4 int4 char float8 bit numeric\n> boolean ... ok\n> char ... ok\n> name ... ok\n> varchar ... ok\n> text ... ok\n> int2 ... ok\n> int4 ... ok\n> int8 ... ok\n> oid ... ok\n> float4 ... ok\n> float8 ... ok\n> bit ... ok\n> numeric ... ok\n>test strings ... ok\n>test numerology ... ok\n>parallel group (18 tests): box point comments interval path time \n>date tinterval lseg circle polygon reltime abstime inet timestamp \n>type_sanity oidjoins opr_sanity\n> point ... ok\n> lseg ... ok\n> box ... ok\n> path ... ok\n> polygon ... ok\n> circle ... ok\n> date ... ok\n> time ... ok\n> timestamp ... ok\n> interval ... ok\n> abstime ... ok\n> reltime ... ok\n> tinterval ... ok\n> inet ... ok\n> comments ... ok\n> oidjoins ... ok\n> type_sanity ... ok\n> opr_sanity ... ok\n>test geometry ... ok\n>test horology ... ok\n>test create_function_1 ... ok\n>test create_type ... ok\n>test create_table ... ok\n>test create_function_2 ... ok\n>test copy ... ok\n>parallel group (7 tests): create_operator create_aggregate inherit \n>triggers constraints create_misc create_index\n> constraints ... ok\n> triggers ... ok\n> create_misc ... ok\n> create_aggregate ... ok\n> create_operator ... ok\n> create_index ... ok\n> inherit ... ok\n>test create_view ... ok\n>test sanity_check ... ok\n>test errors ... ok\n>test select ... ok\n>parallel group (16 tests): select_distinct_on arrays select_having \n>transactions subselect select_distinct union random portals \n>select_into select_implicit aggregates hash_index case join \n>btree_index\n> select_into ... ok\n> select_distinct ... ok\n> select_distinct_on ... ok\n> select_implicit ... ok\n> select_having ... ok\n> subselect ... ok\n> union ... ok\n> case ... ok\n> join ... ok\n> aggregates ... ok\n> transactions ... ok\n> random ... ok\n> portals ... ok\n> arrays ... ok\n> btree_index ... ok\n> hash_index ... ok\n>test misc ... ok\n>parallel group (5 tests): portals_p2 alter_table foreign_key rules \n>select_views\n> select_views ... ok\n> alter_table ... ok\n> portals_p2 ... ok\n> rules ... ok\n> foreign_key ... ok\n>parallel group (3 tests): temp limit plpgsql\n> limit ... ok\n> plpgsql ... ok\n> temp ... ok\n>============== shutting down postmaster ==============\n>\n>======================\n> All 76 tests passed.\n>======================\n>\n>rm regress.o\n\n\nSignature held pending an ISO 9000 compliant\nsignature design and approval process.\nh.b.hotz@jpl.nasa.gov, or hbhotz@oxy.edu\n", "msg_date": "Fri, 30 Mar 2001 14:08:28 -0800", "msg_from": "\"Henry B. Hotz\" <hotz@jpl.nasa.gov>", "msg_from_op": true, "msg_subject": "MacOS X OK, was: Call for platforms" }, { "msg_contents": "> Well, once I figured out a few things about how to use the Unix part\n> of MacOS X it seemed to work without a hitch in the regression\n> testing. I need to figure out how to make the system start it up\n> automatically at boot time still, but then I need to figure out how\n> to do double sided printing on an HP 5si and several other things too.\n\nOK, that confirms that MacOS X is supported.\n\n> BTW the offer of testing on NetBSD/mac68k stands. 250MB is plenty if\n> you don't use X, and I have some extra disks anyway. I'll probably\n> try NetBSD/macppc sometime in the next week or so.\n\nI'm always tempted to snare another platform \"win\", but will draw the\nline at dragging my own machines out of the closet ;) If there is no\npotential user who cares about it, we should let it die a graceful\ndeath.\n\nFor NetBSD/ppc (and any other platform not on the list) the sooner the\nbetter; if you can bang it out at the beginning of the week we should be\nable to get the info into the release.\n\n - Thomas\n", "msg_date": "Sat, 31 Mar 2001 00:57:04 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": false, "msg_subject": "Re: MacOS X OK, was: Call for platforms" } ]
[ { "msg_contents": "\nif i use type \"crypt\", the backend assumes that the client is\nhanding it an already encrypted passwd, and then compares it to an encrypted\nversion of pg_shadow->passwd.\n\nand if i use type \"password filename\", the backend assumes a clear text\npassword from the client, and then compares an encrypted version of that\nto the normal contents of the second field of \"filename\".\n\nhowever, if i use type \"password\", it just does a clear text comparison\nof the password from the client and the password in pg_shadow.\n\nattached are patches which allow for a special case type \"password pg_shadow\",\nwhich similar to supplying a filename, actually encrypts the cleartext\npassword from the client, and compares it to the normal contents of pg_shadow.\n\nthis allows the storage of encrypted passwords in pg_shadow.\n\ni was unable to determine any other way of not storing clear text passwords\nin pg_shadow.\n\ni implemented this in such a way that it will not impact existing\ninstallations.\n\n-- \n[ Jim Mercer jim@pneumonoultramicroscopicsilicovolcanoconiosis.ca ]\n[ Reptilian Research -- Longer Life through Colder Blood ]\n[ aka jim@reptiles.org +1 416 410-5633 ]\n\n\n*** auth.c.orig\tFri Mar 30 19:37:08 2001\n--- auth.c\tFri Mar 30 19:28:20 2001\n***************\n*** 695,701 ****\n static int\n checkPassword(Port *port, char *user, char *password)\n {\n! \tif (port->auth_method == uaPassword && port->auth_arg[0] != '\\0')\n \t\treturn verify_password(port->auth_arg, user, password);\n \n \treturn crypt_verify(port, user, password);\n--- 695,702 ----\n static int\n checkPassword(Port *port, char *user, char *password)\n {\n! \tif (port->auth_method == uaPassword && port->auth_arg[0] != '\\0'\n! \t\t\t&& strcmp(port->auth_arg, \"pg_shadow\") != 0)\n \t\treturn verify_password(port->auth_arg, user, password);\n \n \treturn crypt_verify(port, user, password);\n*** crypt.c.orig\tFri Mar 30 19:38:26 2001\n--- crypt.c\tFri Mar 30 19:39:07 2001\n***************\n*** 280,287 ****\n \t * authentication method being used for this connection.\n \t */\n \n! \tcrypt_pwd =\n! \t\t(port->auth_method == uaCrypt ? crypt(passwd, port->salt) : passwd);\n \n \tif (!strcmp(pgpass, crypt_pwd))\n \t{\n--- 280,294 ----\n \t * authentication method being used for this connection.\n \t */\n \n! \tif (port->auth_method == uaCrypt)\n! \t\tcrypt_pwd = crypt(passwd, port->salt);\n! \telse\n! \t{\n! \t\t/* if port->auth_arg, encrypt password from client before compare */\n! \t\tif (port->auth_arg[0] != 0)\n! \t\t\tpgpass = crypt(pgpass, passwd);\n! \t\tcrypt_pwd = passwd;\n! \t}\n \n \tif (!strcmp(pgpass, crypt_pwd))\n \t{\n", "msg_date": "Fri, 30 Mar 2001 19:51:12 -0500", "msg_from": "Jim Mercer <jim@reptiles.org>", "msg_from_op": true, "msg_subject": "pg_hba.conf \"password\" authentication broken?" } ]
[ { "msg_contents": "I never liked the fact that the TODO web page pulls up a flat file copy\nof the TODO.detail mailbox file.\n\nI hacked together a little script that automatically MHonarc's\nthe TODO.detail file and displays it to the user.\n\nYou can see it if you go to main TODO web page:\n\n\thttp://www.postgresql.org/docs/todo.html\n\nClick on any item in red.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 30 Mar 2001 20:07:13 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Updated TODO.detail" } ]
[ { "msg_contents": "At 15:49 31/03/01 +1000, Philip Warner wrote:\n>\n>(I *think* that's right...).\n>\n\nI've sent a version of this to patches. I have verified it dumps & restores\nthe regression db, excluding the f_star data, which is I *hope* a the one\nacceptable failing of pg_dump (ie. reordered attrs in tables).\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Sun, 01 Apr 2001 00:50:50 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": true, "msg_subject": "Re: Re: 7.1 pg_dump fails for user-defined types (release stopper?)" } ]
[ { "msg_contents": "I have added the recent elog and internationalization email thread to\nTODO.detail. (Wow, I spelled internationalization right the first\ntime.)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 31 Mar 2001 12:26:22 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "elog/internationalization thread" } ]
[ { "msg_contents": "\nHi,\n\nI'm in search of a hack that will allow me to clear the on-row commit \nstatus bits. Any assistance would greatly be appreciated.\n\nThanks,\nDave Perkins\n\n", "msg_date": "Sat, 31 Mar 2001 13:27:52 -0600", "msg_from": "Dave Perkins <drp@shore.net>", "msg_from_op": true, "msg_subject": "clearing on-row commit status bits" } ]
[ { "msg_contents": "Hi Everybody:\n\n How can i set permission for any user (in a database) in the entire database (all tables) .... ???.... with GRANT is just one table (or list), but i want , for example, create a user and set his permissions in all tables , but just SELECT for example, (without set superuser or createdb , etc)...????\n\nThanks!!!!\n\n\n\n\n\n\n\nHi Everybody:\n \n        How can \ni set permission for any user (in a database) in the entire database (all \ntables) .... ???.... with GRANT is just one table (or list), but i want , for \nexample, create a user and set his permissions in all tables , but just SELECT \nfor example, (without set superuser or createdb , etc)...????\n \nThanks!!!!", "msg_date": "Sat, 31 Mar 2001 19:23:53 -0400", "msg_from": "\"Johnny Cristensen\" <jcristen@ingenierias.com>", "msg_from_op": true, "msg_subject": "Complete Database Permission" } ]
[ { "msg_contents": "Hi EveryBody:\n\n How can i get the structure of a table (Fields names, data types, etc)???\n\n\nThanks!!!!\n\n\n\n\n\n\n\nHi EveryBody:\n \n    How can i get the structure of a \ntable (Fields names, data types, etc)???\n \n \nThanks!!!!", "msg_date": "Sat, 31 Mar 2001 19:30:02 -0400", "msg_from": "\"Johnny Cristensen\" <jcristen@ingenierias.com>", "msg_from_op": true, "msg_subject": "Table Structure" }, { "msg_contents": ">\n>Hi EveryBody:\n>\n> How can i get the structure of a table (Fields names, data types, etc)\n\ntry this:\n\n\\d table\n\n\n\n", "msg_date": "1 Apr 2001 15:06:17 GMT", "msg_from": "missive@frontiernet.net (Lee Harr)", "msg_from_op": false, "msg_subject": "Re: Table Structure" } ]
[ { "msg_contents": "Hi EveryBody!!!\n\n How can i set permission for any user (SELECT for example) in all tables (any database) without list all tables in GRANT stament?????\n\nThanks!!!!!\n\n\n\n\n\n\n\nHi EveryBody!!!\n \n    How can i set permission for any \nuser (SELECT for example) in all tables (any database) without list all tables \nin GRANT stament?????\n \nThanks!!!!!", "msg_date": "Sat, 31 Mar 2001 19:32:14 -0400", "msg_from": "\"Johnny Cristensen\" <jcristen@ingenierias.com>", "msg_from_op": true, "msg_subject": "Permissions on All Tables" } ]
[ { "msg_contents": "Hi EveryBody\n\n When i connect to my database in windows98 (ODBC) to the server in Linux, i can execute SELECT staments , but no INSERT or UPDATE staments... the error 'Key Violation ..Connection is readonly , only select staments are allowed' appears..... the user was created with CREATEUSER (SuperUser) but nothing.. i can't write to my tables.......what is the problem???\n\nThanks!!!1\n\n\n\n\n\n\n\nHi EveryBody\n \n        \n    When i connect to my database in windows98 (ODBC) to the \nserver in Linux, i can execute SELECT staments , but no INSERT or UPDATE \nstaments... the error 'Key Violation ..Connection is readonly , only select \nstaments are allowed' appears..... the user was created with CREATEUSER \n(SuperUser) but nothing.. i can't write to my tables.......what is the \nproblem???\n \nThanks!!!1", "msg_date": "Sat, 31 Mar 2001 20:11:49 -0400", "msg_from": "\"Johnny Cristensen\" <jcristen@ingenierias.com>", "msg_from_op": true, "msg_subject": "ODBC Problems" } ]
[ { "msg_contents": "\nCan I get a comment on this patch. Hold for 7.2 or apply?\n\ninitdb works with the patch. People who create indexes on global tables\nget a failed database, while people who create indexes on non-global\nsystem tables get unreliable indexes. This prevents such index\ncreation.\n\nI don't know how many people are creating their own system indexes.\n\n\n> > Hi \n> > \n> > Regarding my previous post, I just successfully created a unique index on \n> > pg_shadow. DON'T DO THIS!!!\n> > -------\n> > CREATE UNIQUE INDEX shadow_index ON pg_shadow (usename)\n> > -------\n> > I couldn't create at pg_shadow_index as the pg prefix is reserved for \n> > system tables. \n> > \n> > This BROKE the database. At least I can't connect anymore with a:\n> > -------\n> > template1=# \\c statements\n> > FATAL 1: Index 'pg_shadow_name_index' does not exist\n> > Previous connection kept\n> > template1=#\n> > -------\n> > If I look at the error log I get :\n> > -------\n> > ERROR: Illegal class name 'pg_shadow_index'\n> > The 'pg_' name prefix is reserved for system catalogs\n> > ERROR: Index 'pg_shadow_name_index' does not exist\n> > ERROR: SearchSysCache: recursive use of cache 23\n> > ERROR: SearchSysCache: recursive use of cache 23\n> > ERROR: SearchSysCache: recursive use of cache 23\n> > ERROR: SearchSysCache: recursive use of cache 23 <-- quite psql here\n> > FATAL 1: Index 'pg_shadow_name_index' does not exist <-- restarted again\n> > FATAL 1: Index 'pg_shadow_name_index' does not exist\n> > FATAL 1: Index 'pg_shadow_name_index' does not exist\n> > -------\n> > \n> > What can I do??? I've got a non-trivial amount of data that I cannot afford \n> > to lose!! HELP!..\n> \n> First, here is a patch which will prevent this from happening in the\n> future. Do people want this held for 7.2 or applied now? It disables\n> the creation of user indexes on system tables.\n> \n> The user-defined indexes on system columns can not be made to work\n> easily. Tom Lane pointed out to me in a phone call that code like:\n> \n> CatalogIndexInsert(irelations, Num_pg_class_indices, relrelation, reltup);\n> \n> assumes it knows the number of indexes on each system table, and a\n> user-defined one would not be updated by any system catalog change that\n> did not go through the executor.\n> \n> As far as recovery, I am not sure. One issue is that pg_shadow is a\n> global table, not local to the database. My guess is that the global\n> table is still fine, but the index is in the database where you created\n> the index. You can't remove the file because pg_index thinks the index\n> is proper and exists.\n> \n> I am kind of stumped.\n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (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> Index: src/backend/catalog/index.c\n> ===================================================================\n> RCS file: /home/projects/pgsql/cvsroot/pgsql/src/backend/catalog/index.c,v\n> retrieving revision 1.144\n> diff -c -r1.144 index.c\n> *** src/backend/catalog/index.c\t2001/03/22 06:16:10\t1.144\n> --- src/backend/catalog/index.c\t2001/03/30 22:55:54\n> ***************\n> *** 864,869 ****\n> --- 864,876 ----\n> \t\tindexInfo->ii_NumKeyAttrs < 1)\n> \t\telog(ERROR, \"must index at least one attribute\");\n> \n> + \tif (heapRelationName && !allow_system_table_mods &&\n> + \t\tIsSystemRelationName(heapRelationName) && IsNormalProcessingMode())\n> + \t{\n> + \t\telog(ERROR, \"You can not create indexes on system tables: '%s'\",\n> + \t\t\t heapRelationName);\n> + \t}\n> + \n> \t/*\n> \t * get heap relation oid and open the heap relation\n> \t */\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (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, 31 Mar 2001 19:31:38 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [SQL] Re: pg_dump potential bug -UNIQUE INDEX on PG_SHADOW Dont!!\n\tHELP" }, { "msg_contents": "\nPatch applied, with wording modifications by Tom Lane.\n\n> > Hi \n> > \n> > Regarding my previous post, I just successfully created a unique index on \n> > pg_shadow. DON'T DO THIS!!!\n> > -------\n> > CREATE UNIQUE INDEX shadow_index ON pg_shadow (usename)\n> > -------\n> > I couldn't create at pg_shadow_index as the pg prefix is reserved for \n> > system tables. \n> > \n> > This BROKE the database. At least I can't connect anymore with a:\n> > -------\n> > template1=# \\c statements\n> > FATAL 1: Index 'pg_shadow_name_index' does not exist\n> > Previous connection kept\n> > template1=#\n> > -------\n> > If I look at the error log I get :\n> > -------\n> > ERROR: Illegal class name 'pg_shadow_index'\n> > The 'pg_' name prefix is reserved for system catalogs\n> > ERROR: Index 'pg_shadow_name_index' does not exist\n> > ERROR: SearchSysCache: recursive use of cache 23\n> > ERROR: SearchSysCache: recursive use of cache 23\n> > ERROR: SearchSysCache: recursive use of cache 23\n> > ERROR: SearchSysCache: recursive use of cache 23 <-- quite psql here\n> > FATAL 1: Index 'pg_shadow_name_index' does not exist <-- restarted again\n> > FATAL 1: Index 'pg_shadow_name_index' does not exist\n> > FATAL 1: Index 'pg_shadow_name_index' does not exist\n> > -------\n> > \n> > What can I do??? I've got a non-trivial amount of data that I cannot afford \n> > to lose!! HELP!..\n> \n> First, here is a patch which will prevent this from happening in the\n> future. Do people want this held for 7.2 or applied now? It disables\n> the creation of user indexes on system tables.\n> \n> The user-defined indexes on system columns can not be made to work\n> easily. Tom Lane pointed out to me in a phone call that code like:\n> \n> CatalogIndexInsert(irelations, Num_pg_class_indices, relrelation, reltup);\n> \n> assumes it knows the number of indexes on each system table, and a\n> user-defined one would not be updated by any system catalog change that\n> did not go through the executor.\n> \n> As far as recovery, I am not sure. One issue is that pg_shadow is a\n> global table, not local to the database. My guess is that the global\n> table is still fine, but the index is in the database where you created\n> the index. You can't remove the file because pg_index thinks the index\n> is proper and exists.\n> \n> I am kind of stumped.\n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (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> Index: src/backend/catalog/index.c\n> ===================================================================\n> RCS file: /home/projects/pgsql/cvsroot/pgsql/src/backend/catalog/index.c,v\n> retrieving revision 1.144\n> diff -c -r1.144 index.c\n> *** src/backend/catalog/index.c\t2001/03/22 06:16:10\t1.144\n> --- src/backend/catalog/index.c\t2001/03/30 22:55:54\n> ***************\n> *** 864,869 ****\n> --- 864,876 ----\n> \t\tindexInfo->ii_NumKeyAttrs < 1)\n> \t\telog(ERROR, \"must index at least one attribute\");\n> \n> + \tif (heapRelationName && !allow_system_table_mods &&\n> + \t\tIsSystemRelationName(heapRelationName) && IsNormalProcessingMode())\n> + \t{\n> + \t\telog(ERROR, \"You can not create indexes on system tables: '%s'\",\n> + \t\t\t heapRelationName);\n> + \t}\n> + \n> \t/*\n> \t * get heap relation oid and open the heap relation\n> \t */\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n", "msg_date": "Mon, 2 Apr 2001 10:34:41 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [SQL] Re: pg_dump potential bug -UNIQUE INDEX on PG_SHADOW Dont!!\n\tHELP" } ]
[ { "msg_contents": "\tI think somebody has the \"owner\" address forwarding to the list.\nReally annoying. Would somebody please stop that?\n\n\tThanks,\n\n\t-Roberto\n\n----- Forwarded message from pgsql-hackers-owner+M6959@postgresql.org -----\n\nFrom: pgsql-hackers-owner+M6959@postgresql.org\nSubject: Majordomo Delivery Error\nTo: pgsql-hackers-owner+M6959@postgresql.org\nX-VMS-To: IN%\"pgsql-hackers-owner+M6959@postgresql.org\"\n\nThis message was created automatically by mail delivery software.\nA Majordomo message could not be delivered to the following addresses:\n\n biff@htmlhost.net:\n 450 4.7.1 <biff@htmlhost.net>... Can not check MX records for recipient host htmlhost.net\n\n-- Original message omitted --\n\n----- End forwarded message -----\n\n-- \n+----| http://fslc.usu.edu USU Free Software & GNU/Linux Club|------+\n Roberto Mello - Computer Science, USU - http://www.brasileiro.net \n http://www.sdl.usu.edu - Space Dynamics Lab, Developer \nA seminar on Time Travel will be held two weeks ago\n", "msg_date": "Sat, 31 Mar 2001 22:15:36 -0700", "msg_from": "Roberto Mello <rmello@cc.usu.edu>", "msg_from_op": true, "msg_subject": "[pgsql-hackers-owner+M6959@postgresql.org: Majordomo Delivery Error]" }, { "msg_contents": "Roberto Mello <rmello@cc.usu.edu> writes:\n> \tI think somebody has the \"owner\" address forwarding to the list.\n\nNo, certainly not --- otherwise we'd all be drowning in bounce messages.\nThe Postgres mailing lists are correctly configured. However, there are\ncertain sites that run broken mail software that sends bounce messages\nback to the author of the individual message being bounced, rather than\nsending 'em to the list owner address like it should.\n\nIf you get a bounce message for mailing list traffic that you authored,\nthen forward it to Marc, who otherwise may not find out that the dead\naddress needs to be removed from the lists.\n\nYou might also try complaining to the postmaster at the offending site,\nbut I've found that's generally a waste of time; if the admin had a clue\nyou'd never have seen any bounce anyway.\n\n\t\t\tregards, tom lane\n", "msg_date": "Sun, 01 Apr 2001 15:10:52 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [pgsql-hackers-owner+M6959@postgresql.org: Majordomo Delivery\n\tError]" } ]
[ { "msg_contents": "I think I remember something about PL/SQL having trouble when passing null to \na function. But searchinf for pl/sql and null doesn't bring any informative \nlinks.\n\nI can go through the steps of a function one by one in psql and everything \nworks fine.\n\nBut when I call the function with nulls for some of the parameters, it \ndoesn't work.\n\nWhat puzzles me more is that I tried to pass 0 (zero) instead and then let \nthe function subsitute it with null (simple IF param = 0 then v_param := null \nthing). Still no luck.\n\nI'm using 7.0.3. \n\n-- \nKaare Rasmussen --Linux, spil,-- Tlf: 3816 2582\nKaki Data tshirts, merchandize Fax: 3816 2501\nHowitzvej 75 �ben 14.00-18.00 Email: kar@webline.dk\n2000 Frederiksberg L�rdag 11.00-17.00 Web: www.suse.dk\n", "msg_date": "Sun, 1 Apr 2001 10:32:07 +0200", "msg_from": "Kaare Rasmussen <kar@webline.dk>", "msg_from_op": true, "msg_subject": "PL/SQL and null" }, { "msg_contents": "On Sun, Apr 01, 2001 at 10:32:07AM +0200, Kaare Rasmussen wrote:\n> \n> What puzzles me more is that I tried to pass 0 (zero) instead and then let \n> the function subsitute it with null (simple IF param = 0 then v_param := null \n> thing). Still no luck.\n> \n> I'm using 7.0.3. \n\n\tUpgrade to 7.1. This has been fixed.\n\n\t-Roberto\n-- \n+----| http://fslc.usu.edu USU Free Software & GNU/Linux Club|------+\n Roberto Mello - Computer Science, USU - http://www.brasileiro.net \n http://www.sdl.usu.edu - Space Dynamics Lab, Web Developer \n*8 *8 *8 <- Tribbles and mamas out for a stroll.\n", "msg_date": "Sun, 1 Apr 2001 16:13:18 -0600", "msg_from": "Roberto Mello <rmello@cc.usu.edu>", "msg_from_op": false, "msg_subject": "Re: PL/SQL and null" } ]
[ { "msg_contents": "Hi,\n\n\tIn response to comments made here, I have been rewriting the unsigned\ntypes as externally loadable. Using the same routines that worked fine\nwhen linked statically into the backend gives me core-dumps only.\nCreating only a single uint2 type with I/O routines, I get\n\ntest=# create table u2 ( u uint2);\nCREATE\ntest=# insert into u2 values (12::uint2);\npqReadData() -- backend closed the channel unexpectedly.\n This probably means the backend terminated abnormally\n before or while processing the request.\n\nRunning this under gdb (I tried this on alpha as well)\n\nbackend> insert into u2 values (12::uint2);\n(no debugging symbols found)...\nProgram received signal SIGSEGV, Segmentation fault.\n0x40115573 in memcpy () from /lib/libc.so.6\n(gdb) where\n#0 0x40115573 in memcpy () from /lib/libc.so.6\n#1 0x80cfb92 in _copyConst ()\n#2 0x80d25d9 in copyObject ()\n#3 0x80ebad9 in expression_tree_mutator ()\n#4 0x80eb407 in eval_const_expressions_mutator ()\n#5 0x80ebe42 in expression_tree_mutator ()\n#6 0x80eb407 in eval_const_expressions_mutator ()\n#7 0x80ebdf2 in expression_tree_mutator ()\n#8 0x80eb407 in eval_const_expressions_mutator ()\n#9 0x80eaf87 in eval_const_expressions ()\n#10 0x80e6d2a in preprocess_expression ()\n#11 0x80e6751 in subquery_planner ()\n#12 0x80e66c0 in planner ()\n#13 0x81036e7 in pg_plan_query ()\n#14 0x81038d9 in pg_exec_query_string ()\n#15 0x81049d4 in PostgresMain ()\n#16 0x80ce884 in main ()\n#17 0x400d8a42 in __libc_start_main () from /lib/libc.so.6\n(gdb)\n\nIt never seems to get to my code. So either I've defined something\nincorrectly or there is a bug. I'd appreciate it if somebody more\nknowledgable than I could have a look at it. I've included a tar with\nthe definitions.\n\nBTW it may be good to update the complex example to the new C-calling\ninterface, as there is no example of creating a type with the new\ncalling interface.\n\nCheers,\n\nAdriaan", "msg_date": "Mon, 02 Apr 2001 10:53:37 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Bug in user-defined types?" }, { "msg_contents": "Adriaan Joubert <a.joubert@albourne.com> writes:\n> \tIn response to comments made here, I have been rewriting the unsigned\n> types as externally loadable. Using the same routines that worked fine\n> when linked statically into the backend gives me core-dumps only.\n\nSeems unlikely that that code could have worked either way, since you\nforgot to mark type uint2 as PASSEDBYVALUE...\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 02 Apr 2001 10:24:30 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Bug in user-defined types? " }, { "msg_contents": "Tom Lane wrote:\n> \n> Seems unlikely that that code could have worked either way, since you\n> forgot to mark type uint2 as PASSEDBYVALUE...\n> \n\nAargh! Thanks! Yes, when implementing it in the backend, that was just a\nfield to fill in, so I did it there. All seems well now.\n\nOne ends up with a vast number of combinations of types combinations for\ndifferent operators. As C takes care of the conversions, I wrote a\n30-line perl script to generate me nearly 1600 lines of C for all the\ntype combinations (+ ~1700 lines of sql to define the\nfunctions/operators). I cannot help feeling that that is not the right\nway: if it can be done in a few lines of perl and relies on C cross-type\noperations underneath anyway, it seems wrong to have to generate all\nthis code. \n\nThe problem is that there is not a clean hierarchy of SQL types, but for\nmany cases one could either convert the operands to int4 or float8 and\nthen numeric(?) and then convert back. At least the conversion operators\ncheck for overflow, which is better than the current situation. And\nprecision wise it cannot be much worse: after all, large integer\nconstants already end up as floats. Is the SQL standard pedantic about\nthis?\n\nBTW I could not find the discussion on entry-points to shared libraries\nthat Thomas mentioned. I've got some rushed dead-lines at the moment, so\nI will not be able to look at anything for the next 3-4 weeks though.\n\nAdriaan\n", "msg_date": "Tue, 03 Apr 2001 08:29:25 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Re: Bug in user-defined types?" }, { "msg_contents": "> The problem is that there is not a clean hierarchy of SQL types, but for\n> many cases one could either convert the operands to int4 or float8 and\n> then numeric(?) and then convert back. At least the conversion operators\n> check for overflow, which is better than the current situation. And\n> precision wise it cannot be much worse: after all, large integer\n> constants already end up as floats. Is the SQL standard pedantic about\n> this?\n\nYes. The implicit \"big integer\" -> float8 done in the scanner is an\nexpedient hack to keep from rejecting the large number entirely, but is\nlikely not in the spirit of the SQL standard.\n\nThe Right Way would have us set the large integer string to int8 and/or\nnumeric, but the scanner does not know about those types at the moment,\nmostly for historical reasons.\n\nIf we made the scanner aware of integers larger than int4, we would have\nto choose between int8 (not supported on all platforms, but mostly OK)\nand numeric, which is markedly slower to process and handle. I recall\nthat Tom Lane had the proposal to use a more loosely categorized \"some\nkind of numeric type\" in the scanner, postponing the hard assignment of\ntype to later in the parser. That might get around the performance\nissues, since numeric would only be used if the context required it.\n\n - Thomas\n", "msg_date": "Tue, 03 Apr 2001 14:32:26 +0000", "msg_from": "Thomas Lockhart <lockhart@alumni.caltech.edu>", "msg_from_op": false, "msg_subject": "Re: Bug in user-defined types?" }, { "msg_contents": "Thomas Lockhart <lockhart@alumni.caltech.edu> writes:\n> If we made the scanner aware of integers larger than int4, we would have\n> to choose between int8 (not supported on all platforms, but mostly OK)\n> and numeric, which is markedly slower to process and handle. I recall\n> that Tom Lane had the proposal to use a more loosely categorized \"some\n> kind of numeric type\" in the scanner, postponing the hard assignment of\n> type to later in the parser. That might get around the performance\n> issues, since numeric would only be used if the context required it.\n\nYes, I was thinking of treating numerical literals more like we already\ntreat unknown-type string literals: keep the value in string format\nuntil we deduce from context which type it should be, then convert.\nInternally this already happens for literals that don't fit in int4,\nbut we still prejudge the type sooner than I think we should.\n\nIIRC, the main reason this isn't done yet was that we hadn't come to\na conclusion about the appropriate type promotion hierarchy for\nnumeric values.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 03 Apr 2001 10:45:45 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Re: Bug in user-defined types? " } ]
[ { "msg_contents": "Will current CVS commits make it into 7.1? Or do I have to use a different\nbranch. I just committed a minor patch to keep the parsers in sync and also\ncommitted a bug fix last week. Both should be in 7.1.\n\nMichael\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n", "msg_date": "Mon, 2 Apr 2001 10:19:24 +0200", "msg_from": "Michael Meskes <meskes@postgresql.org>", "msg_from_op": true, "msg_subject": "CVS commits" }, { "msg_contents": "* Michael Meskes <meskes@postgresql.org> [010402 04:41] wrote:\n> Will current CVS commits make it into 7.1? Or do I have to use a different\n> branch. I just committed a minor patch to keep the parsers in sync and also\n> committed a bug fix last week. Both should be in 7.1.\n\nYou should be able to check that via the cvsweb interface off the\ndeveloper's corner on the postgresql website. Just find your files\nand see if there's a new tag for 7.1 and whether or not your code\nis against HEAD or against the tag (if it exists) for 7.1.\n\n-- \n-Alfred Perlstein - [bright@wintelcom.net|alfred@freebsd.org]\nDaemon News Magazine in your snail-mail! http://magazine.daemonnews.org/\n", "msg_date": "Mon, 2 Apr 2001 04:43:23 -0700", "msg_from": "Alfred Perlstein <bright@wintelcom.net>", "msg_from_op": false, "msg_subject": "Re: CVS commits" }, { "msg_contents": "\nwill still get into v7.1 *nod*\n\nOn Mon, 2 Apr 2001, Michael Meskes wrote:\n\n> Will current CVS commits make it into 7.1? Or do I have to use a different\n> branch. I just committed a minor patch to keep the parsers in sync and also\n> committed a bug fix last week. Both should be in 7.1.\n>\n> Michael\n> --\n> Michael Meskes\n> Michael@Fam-Meskes.De\n> Go SF 49ers! Go Rhein Fire!\n> Use Debian GNU/Linux! Use PostgreSQL!\n>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n>\n\nMarc G. Fournier ICQ#7615664 IRC Nick: Scrappy\nSystems Administrator @ hub.org\nprimary: scrappy@hub.org secondary: scrappy@{freebsd|postgresql}.org\n\n", "msg_date": "Mon, 2 Apr 2001 09:27:48 -0300 (ADT)", "msg_from": "The Hermit Hacker <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: CVS commits" } ]
[ { "msg_contents": "Hi all\n\nAs I posted few days ago I started checking postgresql 7.1RC1 in QNX 4.25. The last I checked was 7.1b3.\n\nthe problem is :\n\nwhen I execute configure it recognize the executable suffix as .map and this is not right. And the test program fails.\nIf I try the same in 7.1b3 all works good.\n\nThen I tried to modify configure : \nline 1472\n *.c | *.o | *.obj | *.map) ;; // I added *.map\n\nWith this all works. I can compile but when I run initdb it crashes. this is the output file\n---------------\nRunning with debug mode on.\n\nInitdb variables:\n PGDATA=/usr/local/pgsql/data1\n datadir=/usr/local/pgsql/share\n PGPATH=//1/usr/local/pgsql/bin\n TEMPFILE=/tmp/initdb.25146\n MULTIBYTE=\n MULTIBYTEID=0\n POSTGRES_SUPERUSERNAME=maurizio\n POSTGRES_SUPERUSERID=100\n TEMPLATE1_BKI=/usr/local/pgsql/share/template1.bki\n GLOBAL_BKI=/usr/local/pgsql/share/global.bki\n TEMPLATE1_DESCR=/usr/local/pgsql/share/template1.description\n GLOBAL_DESCR=/usr/local/pgsql/share/global.description\n POSTGRESQL_CONF_SAMPLE=/usr/local/pgsql/share/postgresql.conf.sample\n PG_HBA_SAMPLE=/usr/local/pgsql/share/pg_hba.conf.sample\n PG_IDENT_SAMPLE=/usr/local/pgsql/share/pg_ident.conf.sample\nThis database system will be initialized with username \"maurizio\".\nThis user will own all the data files and must also own the server process.\n\nCreating directory /usr/local/pgsql/data1\nCreating directory /usr/local/pgsql/data1/base\nCreating directory /usr/local/pgsql/data1/global\nCreating directory /usr/local/pgsql/data1/pg_xlog\nCreating template1 database in /usr/local/pgsql/data1/base/1\nRunning: //1/usr/local/pgsql/bin/postgres -boot -x1 -C -F -D/usr/local/pgsql/data1 -d template1\nDEBUG: database system was shut down at 2001-04-02 10:59:31 cest\nDEBUG: CheckPoint record at (0, 8)\nDEBUG: Redo record at (0, 8); Undo record at (0, 8); Shutdown TRUE\nDEBUG: NextTransactionId: 514; NextOid: 16384\nDEBUG: database system is in production state\n<proname name> \n<proowner int4> \n<prolang oid> \n<proisinh bool> \n<proistrusted bool> \n<proiscachable bool> \n<proisstrict bool> \n<pronargs int2> \n<proretset bool> \n<prorettype oid> \n<proargtypes oidvector> \n<probyte_pct int4> \n<properbyte_cpu int4> \n<propercall_cpu int4> \n<prooutin_ratio int4> \n<prosrc text> \n<probin bytea> \n\n> creating bootstrap relation\nbootstrap relation created ok\n> Commit End\ntuple 1242<Inserting value: 'boolin'\nTyp == NULL, typeindex = 3 idx = 0\nboolin End InsertValue\nInserting value: '100'\nTyp == NULL, typeindex = 6 idx = 1\n100 End InsertValue\nInserting value: '12'\nTyp == NULL, typeindex = 9 idx = 2\n12 End InsertValue\nInserting value: 'f'\nTyp == NULL, typeindex = 0 idx = 3\nf End InsertValue\nInserting value: 't'\nTyp == NULL, typeindex = 0 idx = 4\nt End InsertValue\nInserting value: 't'\nTyp == NULL, typeindex = 0 idx = 5\nt End InsertValue\nInserting value: 't'\nTyp == NULL, typeindex = 0 idx = 6\nt End InsertValue\nInserting value: '1'\nTyp == NULL, typeindex = 4 idx = 7\n1 End InsertValue\nInserting value: 'f'\nTyp == NULL, typeindex = 0 idx = 8\nf End InsertValue\nInserting value: '16'\nTyp == NULL, typeindex = 9 idx = 9\n16 End InsertValue\nInserting value: '0'\nTyp == NULL, typeindex = 13 idx = 10\n End InsertValue\nInserting value: '100'\nTyp == NULL, typeindex = 6 idx = 11\n100 End InsertValue\nInserting value: '0'\nTyp == NULL, typeindex = 6 idx = 12\n0 End InsertValue\nInserting value: '0'\nTyp == NULL, typeindex = 6 idx = 13\n0 End InsertValue\nInserting value: '100'\nTyp == NULL, typeindex = 6 idx = 14\n100 End InsertValue\nInserting value: 'boolin'\nTyp == NULL, typeindex = 8 idx = 15\nboolin End InsertValue\nInserting value: '-'\nTyp == NULL, typeindex = 1 idx = 16\n- End InsertValue\nInsert Begin\nInsertOneTuple oid 1242, 17 attrs\n---------------------\n\n\nI don't know how configure works but from 7.1b3 to 7.1RC1 something was changed in it and I think this is the problem in QNX.\n\nI also checked my coimpiler but I can compile all but the last postgresql version.\n\n\nCould You help me ?\n\n\n\nThanks.\n\n\n\nMaurizio Cauci\nDREAMTECH di Cauci Maurizio\nVia Ronchetti, 2 - 21013 Gallarate (VA)\nwww.dreamtech-it.com\n\n\n\n\n\n\n\nHi all\n \nAs I posted few days ago I started checking \npostgresql 7.1RC1 in QNX 4.25. The last I checked was 7.1b3.\n \nthe problem is :\n \nwhen I execute configure it recognize the \nexecutable suffix as .map and this is not right. And the test program \nfails.\nIf I try the same in 7.1b3 all works \ngood.\n \nThen I tried to modify configure : \nline 1472\n *.c | *.o | *.obj | *.map) \n;;            // I added \n*.map\n \nWith this all works. I can compile but when I run \ninitdb it crashes. this is the output file\n---------------\nRunning with debug mode \non.Initdb variables:  \nPGDATA=/usr/local/pgsql/data1  \ndatadir=/usr/local/pgsql/share  \nPGPATH=//1/usr/local/pgsql/bin  \nTEMPFILE=/tmp/initdb.25146  \nMULTIBYTE=  \nMULTIBYTEID=0  \nPOSTGRES_SUPERUSERNAME=maurizio  \nPOSTGRES_SUPERUSERID=100  \nTEMPLATE1_BKI=/usr/local/pgsql/share/template1.bki  \nGLOBAL_BKI=/usr/local/pgsql/share/global.bki  \nTEMPLATE1_DESCR=/usr/local/pgsql/share/template1.description  \nGLOBAL_DESCR=/usr/local/pgsql/share/global.description  \nPOSTGRESQL_CONF_SAMPLE=/usr/local/pgsql/share/postgresql.conf.sample  \nPG_HBA_SAMPLE=/usr/local/pgsql/share/pg_hba.conf.sample  \nPG_IDENT_SAMPLE=/usr/local/pgsql/share/pg_ident.conf.sampleThis \ndatabase system will be initialized with username \"maurizio\".This user will \nown all the data files and must also own the server process.Creating \ndirectory /usr/local/pgsql/data1Creating directory \n/usr/local/pgsql/data1/baseCreating directory \n/usr/local/pgsql/data1/globalCreating directory \n/usr/local/pgsql/data1/pg_xlogCreating template1 database in \n/usr/local/pgsql/data1/base/1Running: //1/usr/local/pgsql/bin/postgres -boot \n-x1 -C -F -D/usr/local/pgsql/data1 -d template1DEBUG:  database system was shut down at \n2001-04-02 10:59:31 cestDEBUG:  \nCheckPoint record at (0, 8)DEBUG:  Redo record at (0, 8); Undo record at \n(0, 8); Shutdown TRUEDEBUG:  \nNextTransactionId: 514; NextOid: 16384DEBUG:  database system is in production \nstate<proname name> <proowner int4> <prolang oid> \n<proisinh bool> <proistrusted bool> <proiscachable \nbool> <proisstrict bool> <pronargs int2> \n<proretset bool> <prorettype oid> <proargtypes \noidvector> <probyte_pct int4> <properbyte_cpu int4> \n<propercall_cpu int4> <prooutin_ratio int4> <prosrc \ntext> <probin bytea> > creating bootstrap \nrelationbootstrap relation created ok> Commit Endtuple \n1242<Inserting value: 'boolin'Typ == NULL, typeindex = 3 idx = \n0boolin End InsertValueInserting value: '100'Typ == NULL, typeindex \n= 6 idx = 1100 End InsertValueInserting value: '12'Typ == NULL, \ntypeindex = 9 idx = 212 End InsertValueInserting value: 'f'Typ == \nNULL, typeindex = 0 idx = 3f End InsertValueInserting value: 't'Typ \n== NULL, typeindex = 0 idx = 4t End InsertValueInserting value: \n't'Typ == NULL, typeindex = 0 idx = 5t End InsertValueInserting \nvalue: 't'Typ == NULL, typeindex = 0 idx = 6t End \nInsertValueInserting value: '1'Typ == NULL, typeindex = 4 idx = 71 \nEnd InsertValueInserting value: 'f'Typ == NULL, typeindex = 0 idx = \n8f End InsertValueInserting value: '16'Typ == NULL, typeindex = 9 \nidx = 916 End InsertValueInserting value: '0'Typ == NULL, typeindex \n= 13 idx = 10 End \nInsertValueInserting value: '100'Typ == NULL, typeindex = 6 idx = \n11100 End InsertValueInserting value: '0'Typ == NULL, typeindex = 6 \nidx = 120 End InsertValueInserting value: '0'Typ == NULL, typeindex \n= 6 idx = 130 End InsertValueInserting value: '100'Typ == NULL, \ntypeindex = 6 idx = 14100 End InsertValueInserting value: \n'boolin'Typ == NULL, typeindex = 8 idx = 15boolin End \nInsertValueInserting value: '-'Typ == NULL, typeindex = 1 idx = 16- \nEnd InsertValueInsert BeginInsertOneTuple oid 1242, 17 \nattrs---------------------\n \n \nI don't know how configure works \nbut from 7.1b3 to 7.1RC1 something was changed in it and I think this is the \nproblem in QNX.\n \nI also checked my coimpiler but I \ncan compile all but the last postgresql version.\n \n \nCould You help me ?\n \nThanks.\n \nMaurizio CauciDREAMTECH di Cauci \nMaurizioVia Ronchetti, 2 - 21013 Gallarate (VA)www.dreamtech-it.com", "msg_date": "Mon, 2 Apr 2001 11:29:05 +0200", "msg_from": "\"Maurizio\" <maurizio.c@libero.it>", "msg_from_op": true, "msg_subject": "QNX : POSSIBLE BUG IN CONFIGURE ?" }, { "msg_contents": "Maurizio writes:\n\n> the problem is :\n>\n> when I execute configure it recognize the executable suffix as .map\n> and this is not right. And the test program fails.\n\nThis is a known (to me) bug in Autoconf. Maybe there's a way to prevent\nthe .map files to be generated? Fixing this isn't too hard, but I don't\nfeel urgent about it when there are more problems with the QNX port still\ndown the line.\n\n-- \nPeter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n\n", "msg_date": "Mon, 2 Apr 2001 18:38:50 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: QNX : POSSIBLE BUG IN CONFIGURE ?" }, { "msg_contents": "OK,\nI compiled postgresql RC2. I have not error nor warnings so I hope it's all\nright.\nI also changed something in os.h --> port/qnx4.h\nand in s_lock.c\n\nI will post the changes until the end of the week.\n\nI executed initdb and all works fine.\nI executed postmaster and the proces run OK.\n\nWhen I run psql template0 I have an error. I am not expert walking throu\npostgresql sources.\ncould You tell me if there some change from beta 6 to RC1 or RC2 that can\ngive this problem in QNX so I can try to check all?\n\nAttached is the server.log file with the SIGSEGV.\n\nThanks\n\n----- Original Message -----\nFrom: \"Peter Eisentraut\" <peter_e@gmx.net>\nTo: \"Maurizio\" <maurizio.c@libero.it>\nCc: <pgsql-hackers@postgresql.org>\nSent: Monday, April 02, 2001 6:38 PM\nSubject: Re: [HACKERS] QNX : POSSIBLE BUG IN CONFIGURE ?\n\n\n> Maurizio writes:\n>\n> > the problem is :\n> >\n> > when I execute configure it recognize the executable suffix as .map\n> > and this is not right. And the test program fails.\n>\n> This is a known (to me) bug in Autoconf. Maybe there's a way to prevent\n> the .map files to be generated? Fixing this isn't too hard, but I don't\n> feel urgent about it when there are more problems with the QNX port still\n> down the line.\n>\n> --\n> Peter Eisentraut peter_e@gmx.net http://yi.org/peter-e/\n>\n>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org", "msg_date": "Tue, 3 Apr 2001 18:31:38 +0200", "msg_from": "\"Maurizio\" <maurizio.c@libero.it>", "msg_from_op": true, "msg_subject": "Re: QNX : POSSIBLE BUG IN CONFIGURE ?" } ]
[ { "msg_contents": "On Sat, Mar 31, 2001 at 11:45:13PM -0500, Andrew Bosma wrote:\n> \n> Hello all\n> \n> A couple of weeks ago I received an email from Albert Langer inquiring\n> about the status of the python language module I had written for\n> postgresql. I told him I could have the port to postgresql 7.1 done\n> by the middle of this week (march 25-31). Well, it's the end of this\n> week, but I've finished it. Besides the conversion to the new style\n> function manager, I've implemented a complete SPI interface. (The 7.0\n> module couldn't execute saved plans.) If you are interested in\n> experimenting with the module it is available at\n> \n> \"http://users.ids.net/~bosma\"\n> \n> download the link \"tarball for postgresql 7.1\"\n> \n> comments, bug reports and suggestions are appreciated.\n\n Sure :-)\n\n It's great news that anyone works on PL/Python. Why you not say something\nabout it in hackers list? (I ask about this several time!).\n\n I hope we will see it in 7.2 and will be possible sending paches. \n\n I see the code and it's probably not bad, but needs some changes (remove \nmalloc(), ..etc :-)\n\n\t\t\tKarel \n\n-- \n Karel Zak <zakkr@zf.jcu.cz>\n http://home.zf.jcu.cz/~zakkr/\n \n C, PostgreSQL, PHP, WWW, http://docs.linux.cz, http://mape.jcu.cz\n", "msg_date": "Mon, 2 Apr 2001 12:08:07 +0200", "msg_from": "Karel Zak <zakkr@zf.jcu.cz>", "msg_from_op": true, "msg_subject": "Re: plpython for postgres 7.1" }, { "msg_contents": "On Mon, 2 Apr 2001, Karel Zak wrote:\n\n> > A couple of weeks ago I received an email from Albert Langer inquiring\n> > about the status of the python language module I had written for\n> > postgresql. I told him I could have the port to postgresql 7.1 done\n> > by the middle of this week (march 25-31). Well, it's the end of this\n> > week, but I've finished it. Besides the conversion to the new style\n> > function manager, I've implemented a complete SPI interface. (The 7.0\n> > module couldn't execute saved plans.) If you are interested in\n> > experimenting with the module it is available at\n> > \n> > \"http://users.ids.net/~bosma\"\n> > \n> > download the link \"tarball for postgresql 7.1\"\n> > \n> > comments, bug reports and suggestions are appreciated.\n\nThis is *great news* -- we use Python in our office for many things, and\nwith Python embedded into the DB server, it makes our Zope-PostgreSQL\nconnection ever tighter.\n\nI'm afraid I can't give much feedback about the code (I'm just not that\nfamiliar w/the PG internals), but, externally it seems to work great. I'm\nexcited about the SD[] and GD[] dicts -- they're a nice addition for us.\n\nFor those of you considering installing this, it was a very easy install\n(Linux-Madrake 7.2 (Linux 2.2.x) / Python 1.5.2). Run one diff against the\nPG sources, recompile, edit a Makefile for one- or two- library locations,\nand that's it. Worth playing with.\n\n> I hope we will see it in 7.2 ...\n\nIndeed.\n\nFor the deep gurus: what's the downside of adding PLs to our PG\nserver? (Of course, adding alpha- or beta- quality PLs has clear problems,\nI mean when this becomes production quality). Does each new PL bloat the\nPG server? Does each new PL slow it down?\n\nThanks!\n-- \nJoel Burton <jburton@scw.org>\nDirector of Information Systems, Support Center of Washington\n\n", "msg_date": "Tue, 3 Apr 2001 18:35:06 -0400 (EDT)", "msg_from": "Joel Burton <jburton@scw.org>", "msg_from_op": false, "msg_subject": "Re: plpython for postgres 7.1" } ]
[ { "msg_contents": "Binaries for PostgreSQL 7.1 RC1 are now available in\n\nftp://ftp.postgresql.org/pub/binary/v7.1/IRIX_6.5.7/\n\nAll regression tests pass except that geometry is different\nby very small amounts (< 10^14) and join gives the same rows in\na different order.\n\n+----------------------------------+------------------------------------+\n| Robert E. Bruccoleri, Ph.D. | Phone: 609 737 6383 |\n| President, Congenomics, Inc. | Fax: 609 737 7528 |\n| 114 W Franklin Ave, Suite K1,4,5 | email: bruc@acm.org |\n| P.O. Box 314 | URL: http://www.congen.com/~bruc |\n| Pennington, NJ 08534 | |\n+----------------------------------+------------------------------------+\n", "msg_date": "Mon, 2 Apr 2001 12:15:19 -0400 (EDT)", "msg_from": "bruc@stone.congenomics.com (Robert E. Bruccoleri)", "msg_from_op": true, "msg_subject": "Irix binaries of 7.1 RC1 are available" } ]
[ { "msg_contents": "While thinking over Jeremy Radlow's recent problem report in\npgsql-general, it occurs to me that it's probably wrong to implement\nreferential integrity actions like ON CASCADE DELETE in AFTER triggers.\nSeems to me that this breaks the fundamental rule of referential\nintegrity: if B references A then there must always be a matching A\nrow for every B row. Therefore, if we delete a row from A we should\ndelete the matching B row(s) before, not after, we delete from A.\nOtherwise the remainder of the transaction sees an illegal state of\nthe database.\n\nComments? How about ON UPDATE actions?\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 02 Apr 2001 12:23:50 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Shouldn't ON UPDATE/DELETE triggers be BEFORE triggers?" }, { "msg_contents": "> While thinking over Jeremy Radlow's recent problem report in\n> pgsql-general, it occurs to me that it's probably wrong to implement\n> referential integrity actions like ON CASCADE DELETE in AFTER triggers.\n> Seems to me that this breaks the fundamental rule of referential\n> integrity: if B references A then there must always be a matching A\n> row for every B row. Therefore, if we delete a row from A we should\n> delete the matching B row(s) before, not after, we delete from A.\n> Otherwise the remainder of the transaction sees an illegal state of\n> the database.\nIf we're right in how we read the spec, then this isn't an illegal\nstate except for non-deferred constraints and then only for the\nperiod between the delete and the after trigger running. Note:\nI think we may be misinterpreting the spec here (more below),\nbut if our interpretation, deferred actions occur at end of transaction,\nis correct, then the \"invalid\" state is valid to see for the rest of\ntransaction in that case.\n\n > Comments? How about ON UPDATE actions?\n\nHowever, I think that the intention was to have actions (obviously\nother than NO ACTION) occur immediately even on deferred\nconstraints. I say this because the sections on figuring matching\nand uniquely matching rows makes little sense if the action could be \ndeferred and IIRC it says things like \"if a row is marked for removal\"\nrather than \"at the time of checking a constraint, if a row was marked\nfor removal.\"\n\nWhen I tried this in Oracle a while ago, this was also what they\ndid. A deferred constraint with a cascade would kill the\nreferencing rows after the delete so you wouldn't see them for the\nrest of the transaction.\n\n\n\n_________________________________________________________\nDo You Yahoo!?\nGet your free @yahoo.com address at http://mail.yahoo.com\n\n", "msg_date": "Mon, 2 Apr 2001 14:11:15 -0700", "msg_from": "\"Stephan Szabo\" <acroyear_07030@yahoo.com>", "msg_from_op": false, "msg_subject": "Re: Shouldn't ON UPDATE/DELETE triggers be BEFORE triggers?" } ]
[ { "msg_contents": "> While thinking over Jeremy Radlow's recent problem report in\n> pgsql-general, it occurs to me that it's probably wrong to implement\n> referential integrity actions like ON CASCADE DELETE in AFTER \n> triggers. Seems to me that this breaks the fundamental rule of\n> referential integrity: if B references A then there must always be\n> a matching A row for every B row. Therefore, if we delete a row from\n> A we should delete the matching B row(s) before, not after, we delete\n> from A. Otherwise the remainder of the transaction sees an illegal\n> state of the database.\n> \n> Comments? How about ON UPDATE actions?\n\nNo - AFTER is right place where to do action: don't forget\nabout DEFERRED RI constraints for which we have to perform\ncheck/action sometime later *after* we changed PK/FK rows.\n\nVadim\n", "msg_date": "Mon, 2 Apr 2001 10:05:30 -0700 ", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: Shouldn't ON UPDATE/DELETE triggers be BEFORE trigg\n\ters?" } ]
[ { "msg_contents": "I have updated HISTORY/release.sgml to contain the most recent changes\nfor 7.1.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 853-3000\n + If your life is a hard drive, | 830 Blythe Avenue\n + Christ can be your backup. | Drexel Hill, Pennsylvania 19026\n", "msg_date": "Mon, 2 Apr 2001 13:59:11 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Update HISTORY/release.sgml" } ]
[ { "msg_contents": "Do we have any supported platforms where dereferencing a null pointer\ndoesn't trigger coredump?\n\nI'm wondering about this after noticing the likely side effects of\nfd.c's failure to check for null result from malloc(): it'll try to\nstrcpy() filenames to location zero. If it succeeds, you could end up\nwith multiple VFDs sharing the same filename string. Which could lead\nto, eg, writing on or even deleting one file under the delusion that\nwe were writing/deleting another.\n\nWith sufficient suspension of disbelief about how long a backend\ncould run at zero free memory before elog'ing, this might explain\nthe two recent reports of Postgres apparently deleting a file it\nshouldn't have. (I'm not sure I really believe that, but given\nthe way palloc works it's not out of the question. I've added\nappropriate checks to fd.c, just in case.)\n\nAFAIK, null pointer deref -> SIGSEGV is standard behavior on most\nplatforms these days, and we take steps to select that behavior on\nsome nonconformists like HPUX. But I'm wondering if there are any\nplatforms we could select it on and have forgotten to. I think it\nwould be a real good idea to turn on null pointer crash anywhere\nwe can.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 03 Apr 2001 00:43:18 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Do we have any platforms that allow null pointer dereference?" } ]
[ { "msg_contents": "At 23:57 2/04/01 -0400, Tom Lane wrote:\n>\n>NOT NULL on a child field would only force it to be dumped if none\n>of the parents say NOT NULL. Type name really is not an issue since\n>it will have to be the same in all the tables anyway; I wouldn't bother\n>expending any code there.\n>\n\nDone & applied to CVS. Only dumps changed info, so:\n\n create table p3_def1(f1 int default 1, f2 int);\n create table c5(f1 int not null, f3 int) inherits(p3_def1);\n\nwill get dumped as:\n\n CREATE TABLE \"c5\" (\n \"f1\" integer NOT NULL,\n \"f3\" integer\n )\n inherits (\"p3_def1\");\n\nie. without the DEFAULT.\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Tue, 03 Apr 2001 18:55:36 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": true, "msg_subject": "Re: Re: Changing the default value of an inherited\n column" } ]
[ { "msg_contents": "\n> will get dumped as:\n> \n> CREATE TABLE \"c5\" (\n> \"f1\" integer NOT NULL,\n> \"f3\" integer\n> )\n> inherits (\"p3_def1\");\n\nAs an aside answer without considerable importance:\n\nWhy do people tend to write SQL keywords in all capitals ?\nPostgreSQL converts everything to lower case (which I like).\nSo why not output lowercase ? Above example even mixes \ncase, why ?\n\nAndreas\n", "msg_date": "Tue, 3 Apr 2001 12:04:17 +0200 ", "msg_from": "Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at>", "msg_from_op": true, "msg_subject": "Re: capitalized sql (was: Re: Changing the default valu\n\te of an inherited column)" }, { "msg_contents": "Thus spake Zeugswetter Andreas SB\n> > will get dumped as:\n> > \n> > CREATE TABLE \"c5\" (\n> > \"f1\" integer NOT NULL,\n> > \"f3\" integer\n> > )\n> > inherits (\"p3_def1\");\n> \n> As an aside answer without considerable importance:\n> \n> Why do people tend to write SQL keywords in all capitals ?\n> PostgreSQL converts everything to lower case (which I like).\n> So why not output lowercase ? Above example even mixes \n> case, why ?\n\nI do it for maintenance purposes. All of my code uses caps for keywords\nso that I can find them without getting them confused with language\nconstructs and variables.\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 425 1212 (DoD#0082) (eNTP) | what's for dinner.\n", "msg_date": "Tue, 3 Apr 2001 07:06:43 -0400 (EDT)", "msg_from": "darcy@druid.net (D'Arcy J.M. Cain)", "msg_from_op": false, "msg_subject": "Re: capitalized sql (was: Re: Changing the default valu e\n\tof an inherited column)" } ]
[ { "msg_contents": "At 12:04 3/04/01 +0200, Zeugswetter Andreas SB wrote:\n>\n>> will get dumped as:\n>> \n>> CREATE TABLE \"c5\" (\n>> \"f1\" integer NOT NULL,\n>> \"f3\" integer\n>> )\n>> inherits (\"p3_def1\");\n>\n>As an aside answer without considerable importance:\n>\n>Why do people tend to write SQL keywords in all capitals ?\n\nIt's just one of those things that people do; some people find it easier to\nread (I actually prefer proper case keywords for SQL, unquoted upper case\nfor names etc):\n\n Create Table C5 (\n F1 Integer Not Null,\n F3 Integer\n )\n Inherits (P3_DEF1);\n\nAs to why it's in pg_dump, I think it reflects the many & varied\ncontributors as opposed to careful design. There is some proper-case stuff\nin there now (from me), which I will try to remember to remove. The upper\ncase stuff is there since 6.x at least, and I'm not sure why INHERITS is in\nlower case. Type names are now formatted (mostly) using formatType (from\nPeter E), so they will at least be consistent (and lower case).\n\n\n\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Tue, 03 Apr 2001 22:10:40 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": true, "msg_subject": "Re: capitalized sql (was: Re: Changing the default\n\tvalue of an inherited column)" } ]
[ { "msg_contents": "Hi,\n\n\twe had a problem on Alpha that in interfaces/ecpg/lib/typename.c we\nhave\nHAVE_LONG_INT_64 defined, but not HAVE_LONG_LONG_INT_64. Consequently no\ncode is included for long ints and typename calls *abort*. I put in a\nfew lines that check for HAVE_LONG_INT_64 and seem to generate the right\ncode. I've got a new version of typename.c attached. It would be good if\nMichael could review and get this into 7.1.\n\nCheers,\n\nAdriaan", "msg_date": "Tue, 03 Apr 2001 18:32:25 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "ecpg long int problem on alpha + fix" }, { "msg_contents": "On Tue, Apr 03, 2001 at 06:32:25PM +0300, Adriaan Joubert wrote:\n> \twe had a problem on Alpha that in interfaces/ecpg/lib/typename.c we\n> have\n> HAVE_LONG_INT_64 defined, but not HAVE_LONG_LONG_INT_64. Consequently no\n\nSure since that means your long int and not your long long int is 64 bits.\n\n> code is included for long ints and typename calls *abort*. I put in a\n> few lines that check for HAVE_LONG_INT_64 and seem to generate the right\n\nWhy is this needed? What you do is use \"long long\" as variable type for 64\nbits integer. But on the alpha you do not need \"long long\", just use \"long\nint\" instead.\n\nOr did I misunderstand something?\n\nMichael\n\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n", "msg_date": "Wed, 4 Apr 2001 14:02:53 +0200", "msg_from": "Michael Meskes <meskes@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: ecpg long int problem on alpha + fix" }, { "msg_contents": "Michael Meskes wrote:\n> \n> On Tue, Apr 03, 2001 at 06:32:25PM +0300, Adriaan Joubert wrote:\n> > we had a problem on Alpha that in interfaces/ecpg/lib/typename.c we\n> > have\n> > HAVE_LONG_INT_64 defined, but not HAVE_LONG_LONG_INT_64. Consequently no\n> \n> Sure since that means your long int and not your long long int is 64 bits.\n> \n> > code is included for long ints and typename calls *abort*. I put in a\n> > few lines that check for HAVE_LONG_INT_64 and seem to generate the right\n> \n> Why is this needed? What you do is use \"long long\" as variable type for 64\n> bits integer. But on the alpha you do not need \"long long\", just use \"long\n> int\" instead.\n> \n> Or did I misunderstand something?\n\nOK, I see. Problem is that without the fix ecpg aborts when writing to a\ntable with an int8 column using valid code.\n\nlong int\nlong long\nlong long int\n\nall exist on alpha and are all 64 bits, but HAVE_LONG_LONG_INT_64 is not\ndefined, so ecpg cannot handle ECPGt_long_long types. It is not clear to\nme what the best thing is to fix here -- possibly configure needs to set\nHAVE_LONG_LONG_INT_64 (which solves the problem on alpha as well), but I\ndo not know what the consequences of that are.\n\nCheers,\n\nAdriaan\n", "msg_date": "Wed, 04 Apr 2001 15:35:34 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Re: ecpg long int problem on alpha + fix" }, { "msg_contents": "On Wed, Apr 04, 2001 at 03:35:34PM +0300, Adriaan Joubert wrote:\n> OK, I see. Problem is that without the fix ecpg aborts when writing to a\n> table with an int8 column using valid code.\n\nSorry, I still don't seem to understand that. Data between ecpg and the\nbackend is tranfered in ascii only. What exactly happens?\n\n> all exist on alpha and are all 64 bits, but HAVE_LONG_LONG_INT_64 is not\n> defined, so ecpg cannot handle ECPGt_long_long types. It is not clear to\n\nI see. I was under the impression that HAVE_LONG_LONG_INT_64 should be\ndefined if long long int is 64 bit integer which of course it is on the\nalpha.\n\n> me what the best thing is to fix here -- possibly configure needs to set\n> HAVE_LONG_LONG_INT_64 (which solves the problem on alpha as well), but I\n> do not know what the consequences of that are.\n\nI would think that this is the correct solution. After all there is a\n#ifdef HAVE_LONG_LONG_INT_64 in the backend as well.\n\nAnyone else with more knowledge about HAVE_LONG_LONG_INT_64 dare to comment?\n\nMichael\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n", "msg_date": "Wed, 4 Apr 2001 15:34:02 +0200", "msg_from": "Michael Meskes <meskes@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: ecpg long int problem on alpha + fix" }, { "msg_contents": "Michael Meskes <meskes@postgresql.org> writes:\n>> all exist on alpha and are all 64 bits, but HAVE_LONG_LONG_INT_64 is not\n>> defined, so ecpg cannot handle ECPGt_long_long types. It is not clear to\n\n> I see. I was under the impression that HAVE_LONG_LONG_INT_64 should be\n> defined if long long int is 64 bit integer which of course it is on the\n> alpha.\n\nNo! Look at the configure test! It doesn't bother to investigate long\nlong int if it finds that long int will serve. At most one of the two\nsymbols will be defined. I believe that that is the correct way, at\nleast for the backend's purposes, since we only want to compile one set\nof int64-related code.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 04 Apr 2001 10:04:46 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: ecpg long int problem on alpha + fix " }, { "msg_contents": "Michael Meskes wrote:\n> \n> On Wed, Apr 04, 2001 at 03:35:34PM +0300, Adriaan Joubert wrote:\n> > OK, I see. Problem is that without the fix ecpg aborts when writing to a\n> > table with an int8 column using valid code.\n> \n> Sorry, I still don't seem to understand that. Data between ecpg and the\n> backend is tranfered in ascii only. What exactly happens?\n\nThis has nothing to do with the backend. ecpg itself core-dumps after\ncalling abort() at the end of the switch statement in typename.c, when\nprocessing a .pgc file. As people complained to me about ecpg\ncore-dumping I tried to find out why and then found that it called the\nabort() at the end of this switch ;-)\n\nI have not looked at ecpg in any detail, but I expect that the types in\ntypename.c are derived from the host variables in some way. If we have\nan int8 column in a table, we need to use a 64 bit type, i.e. a 'long\nlong', and as HAVE_LONG_LONG_INT_64 is not true, no such type is\ncompiled into the switch. As HAVE_LONG_INT_64 is defined on alpha, my\nfix fixes this for alpha, and any other platform where HAVE_LONG_INT_64\nis defined but not HAVE_LONG_LONG_INT_64. \n\n From Tom's mail I gather that it is not an option to define\nHAVE_LONG_LONG_INT_64 on alpha, so I think this patch, or something\nsimilar, is necessary.\n\nApologies for not being clear enough initially.\n\nCheers,\n\nAdriaan\n", "msg_date": "Wed, 04 Apr 2001 17:47:50 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Re: ecpg long int problem on alpha + fix" }, { "msg_contents": "On Wed, Apr 04, 2001 at 05:47:50PM +0300, Adriaan Joubert wrote:\n> This has nothing to do with the backend. ecpg itself core-dumps after\n> calling abort() at the end of the switch statement in typename.c, when\n> processing a .pgc file. As people complained to me about ecpg\n\nYes, I did understand that. What I do not understand is the problem with\nlong long int. On an alpha the normal long int should be sufficient for\nstoring int8.\n\n> core-dumping I tried to find out why and then found that it called the\n> abort() at the end of this switch ;-)\n\nThis abort() is used to make sure ecpg stops if a type id is found that does\nnot exist. Normally this must not happen. Ah, now I understand, the parser\naccepts long long but typename.c does not like that.\n\nIt should be sufficient to just remove the #ifdef in typename.c since\nnothing serious is happening there. The reason for this precompiler define\nis to make sure that no long long code is generated that the C compiler\ncannot cope with. But in typename.c there is no code generated other than\nthe creation of a string.\n\nCould you please try to just remove the cpp flag? Also I wonder why you are\nusing \"long long int\" instead of just \"long int\" in your C program. Well\nthat is the people who complained to you.\n\n> an int8 column in a table, we need to use a 64 bit type, i.e. a 'long\n> long', and as HAVE_LONG_LONG_INT_64 is not true, no such type is\n\nI thought \"long int\" is 64 bits too, so that should be sufficient.\n\n> From Tom's mail I gather that it is not an option to define\n> HAVE_LONG_LONG_INT_64 on alpha, so I think this patch, or something\n> similar, is necessary.\n\nThanks to Tom btw for setting this straight.\n\n> Apologies for not being clear enough initially.\n\nI think you were pretty clear from teh get go, but I simply misunderstood\nyou.\n\nMichael\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n", "msg_date": "Wed, 4 Apr 2001 20:43:07 +0200", "msg_from": "Michael Meskes <meskes@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: ecpg long int problem on alpha + fix" }, { "msg_contents": "> Could you please try to just remove the cpp flag? Also I wonder why you are\n> using \"long long int\" instead of just \"long int\" in your C program. Well\n> that is the people who complained to you.\n\nYes, dropping the CPP flags solves the problem for us. I assume all\nplatforms have long long now?\n\nWe used long long as this seems to be pretty consistently 64 bits on\ndifferent platforms, and our code runs on Tru64, PC linux and openBSD.\nIt also agrees with the CORBA type naming for 64 bit ints, so it makes\nthe type naming more consistent. \n\nThanks,\n\nAdriaan\n", "msg_date": "Thu, 05 Apr 2001 08:41:08 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Re: ecpg long int problem on alpha + fix" }, { "msg_contents": "Adriaan Joubert <a.joubert@albourne.com> writes:\n> Yes, dropping the CPP flags solves the problem for us. I assume all\n> platforms have long long now?\n\nSurely you jest.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 05 Apr 2001 01:59:27 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: ecpg long int problem on alpha + fix " }, { "msg_contents": "Tom Lane wrote:\n> \n> Adriaan Joubert <a.joubert@albourne.com> writes:\n> > Yes, dropping the CPP flags solves the problem for us. I assume all\n> > platforms have long long now?\n> \n> Surely you jest.\n\nYep, it was a rhetorical question. \n\nI think we probably do need the CPP defines from my patch in there, so\nthat it really doesn't work on platforms that do not have int8. But I do\nnot know the ins-and-outs of the ecpg code, so that's Michael's call. As\nlong as it works on my shiny alpha-boxes at the end of it all I'm happy\n;-)\n\nAdriaan\n", "msg_date": "Thu, 05 Apr 2001 09:13:49 +0300", "msg_from": "Adriaan Joubert <a.joubert@albourne.com>", "msg_from_op": true, "msg_subject": "Re: ecpg long int problem on alpha + fix" }, { "msg_contents": "On Thu, Apr 05, 2001 at 09:13:49AM +0300, Adriaan Joubert wrote:\n> I think we probably do need the CPP defines from my patch in there, so\n\nNot exactly. The test in typename.c does not make sense at all. It will be\nremoved. But there are other places where it is needed. Or can I safely\nassume that if HAVE_LONG_INT_64 is defined then HAVE_LONG_LONG_INT_64 also\nis true, although not defined?\n\nHmm, thinking about it, how about using long instead of long long if\nHAVE_LONG_LONG_INT_64 is undefined?\n\nMichael\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n", "msg_date": "Thu, 5 Apr 2001 10:01:54 +0200", "msg_from": "Michael Meskes <meskes@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: ecpg long int problem on alpha + fix" } ]
[ { "msg_contents": "> Hm, that sounds like some sort of conflict with a temp table. Is it\n> possible that you have been using a temp table name that matches the\n> sequence name? Are there any pg_class entries whose names begin with\n> pg_temp, and if so could we see details on those too?\n\nSome more input from Konstantin (his answer to my message bounced\nfrom bug-list -:)):\n\n> > How much time passed after sequence creation till crash?\n>\n> About 5-10 seconds. I opened a transaction, created a sequence,\n> created a temporary table with one column having NEXTVAL(seq) as\n ^^^^^^^^^^^^^^^^^^^^^^^^^\n> default, inserted some data into the table, committed the transaction.\n> After that I ran my function, which crashed the backend after \n> 3-4 seconds of work.\n\nVadim\n", "msg_date": "Tue, 3 Apr 2001 09:20:29 -0700 ", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: [BUGS] Loosing files after backend crash " }, { "msg_contents": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM> writes:\n> Some more input from Konstantin (his answer to my message bounced\n> from bug-list -:)):\n\n> How much time passed after sequence creation till crash?\n>> \n>> About 5-10 seconds. I opened a transaction, created a sequence,\n>> created a temporary table with one column having NEXTVAL(seq) as\n> ^^^^^^^^^^^^^^^^^^^^^^^^^\n>> default, inserted some data into the table, committed the transaction.\n>> After that I ran my function, which crashed the backend after \n>> 3-4 seconds of work.\n\nHmm. Maybe the WAL redo is messing things up??\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 03 Apr 2001 12:34:07 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [BUGS] Loosing files after backend crash " } ]
[ { "msg_contents": "> > How much time passed after sequence creation till crash?\n> >> \n> >> About 5-10 seconds. I opened a transaction, created a sequence,\n> >> created a temporary table with one column having NEXTVAL(seq) as\n> > ^^^^^^^^^^^^^^^^^^^^^^^^^\n> >> default, inserted some data into the table, committed the \n> >> transaction.\n> >> After that I ran my function, which crashed the backend after \n> >> 3-4 seconds of work.\n> \n> Hmm. Maybe the WAL redo is messing things up??\n\nIt could mess up pg_class content, but it never deletes\nfiles (currently).\n\nVadim\n", "msg_date": "Tue, 3 Apr 2001 09:59:43 -0700 ", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: [BUGS] Loosing files after backend crash " }, { "msg_contents": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM> writes:\n>> Hmm. Maybe the WAL redo is messing things up??\n\n> It could mess up pg_class content, but it never deletes\n> files (currently).\n\nI'm not convinced that any files have really been deleted. Maybe it's\njust that the pg_class entries are wrong, or even more likely that the\nindexes on pg_class are messed up (pointing at the wrong pg_class rows).\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 03 Apr 2001 13:03:35 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [BUGS] Loosing files after backend crash " } ]
[ { "msg_contents": "I notice that pg_dump now dumps primary-key indexes in the style\n\nCREATE TABLE ... (\n\t\"dest_index\" integer DEFAULT ...,\n\tConstraint \"dest_addresses_pkey\" Primary Key (\"dest_index\")\n);\n\n...\n\nCOPY ... FROM stdin;\n-- load data\n\\.\n\n-- create other indexes for table\n\nIsn't this pretty darn stupid? Previously, we created indexes after\nloading the data. We're going to take a huge performance hit to do it\nthis way.\n\nIMHO it would be better to reach in and set the \"primary key\" flag on\nthe index after creating it normally.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 03 Apr 2001 14:33:33 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "pg_dump performance lossage for primary keys" }, { "msg_contents": "At 14:33 3/04/01 -0400, Tom Lane wrote:\n>I notice that pg_dump now dumps primary-key indexes in the style\n>\n>CREATE TABLE ... (\n>\t\"dest_index\" integer DEFAULT ...,\n>\tConstraint \"dest_addresses_pkey\" Primary Key (\"dest_index\")\n>);\n>\n>Isn't this pretty darn stupid?\n\nYep.\n\n>Previously, we created indexes after\n>loading the data. We're going to take a huge performance hit to do it\n>this way.\n\nMy 7.0 dumps PK in table definitions as well, AFAICT (but it may have been\npatched) - can you check yours? \n\nThe first time PK-in-table was implemented was in rev 1.124/5, but it may\nhave been taken out afterwards.\n\nWe really need ALTER TABLE ADD CONSTRAINT for PK.\n\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n", "msg_date": "Wed, 04 Apr 2001 05:18:32 +1000", "msg_from": "Philip Warner <pjw@rhyme.com.au>", "msg_from_op": false, "msg_subject": "Re: pg_dump performance lossage for primary keys" }, { "msg_contents": "Philip Warner <pjw@rhyme.com.au> writes:\n> At 14:33 3/04/01 -0400, Tom Lane wrote:\n>> I notice that pg_dump now dumps primary-key indexes in the style\n>> \n>> CREATE TABLE ... (\n>> \"dest_index\" integer DEFAULT ...,\n>> Constraint \"dest_addresses_pkey\" Primary Key (\"dest_index\")\n>> );\n\n> My 7.0 dumps PK in table definitions as well, AFAICT (but it may have been\n> patched) - can you check yours? \n\nAh, you are right. My mistake --- the lossage is of longer standing\nthan I thought.\n\n> We really need ALTER TABLE ADD CONSTRAINT for PK.\n\nThat would be a cleaner way to do it, all right ... but for now, you can\njust reach in and set the indisprimary flag in pg_index after creating\nthe index. I'm visualizing\n\n\tCREATE TABLE table\n\t( field int NOT NULL, ...);\n\n\tload data\n\n\tCREATE UNIQUE INDEX table_pkey ON table(field);\n\n\tUPDATE pg_index SET indisprimary = true WHERE indexrelid =\n\t(SELECT oid FROM pg_class WHERE relname = 'table_pkey');\n\nOn the other hand, that would fall over if executed by a non-superuser.\nDrat. Okay, I guess we just have to leave this as a TODO item for now.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 03 Apr 2001 15:34:51 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Re: pg_dump performance lossage for primary keys " }, { "msg_contents": "On Tue, Apr 03, 2001 at 03:34:51PM -0400, Tom Lane wrote:\n> Philip Warner <pjw@rhyme.com.au> writes:\n> \n> > We really need ALTER TABLE ADD CONSTRAINT for PK.\n> \n> That would be a cleaner way to do it, all right ... but for now, you can\n> just reach in and set the indisprimary flag in pg_index after creating\n> the index. I'm visualizing\n> \n<snip>\n> \n> On the other hand, that would fall over if executed by a non-superuser.\n> Drat. Okay, I guess we just have to leave this as a TODO item for now.\n\nThis is one of those 'dual roles of pg_dump' problems: Philip has been\nslowly migrating it from being a 'quickest possible backup dump' tool\nto a 'recover my db in as human friendly (and SQL standards compliant)\na format as possible' tool. Which, not coincidently, has dramatically\nreduced the version fragility of the dump output.\n\nAdding implementation specific performance hacks back in is probably\na necessary evil, but should probably be protected by a '--fastdump'\nswitch or some such.\n\nRoss\n", "msg_date": "Tue, 3 Apr 2001 15:01:00 -0500", "msg_from": "\"Ross J. Reedstrom\" <reedstrm@rice.edu>", "msg_from_op": false, "msg_subject": "Re: pg_dump performance lossage for primary keys" } ]
[ { "msg_contents": "\nOk, I believe the postgres.h and all the files in the include/utils\ndirectory need to be copied over during the installation, it got most of\nthe others, but missed those ones and as I've spent a good chunk trying to\nget PHP and a few other utils built, they kinda needed them.\n\n7.1 release candidate 2 has this issue.\n\n Chris Bowlby,\n -----------------------------------------------------\n Web Developer @ Hub.org.\n excalibur@hub.org\n www.hub.org\n 1-902-542-3657\n -----------------------------------------------------\n\n", "msg_date": "Tue, 3 Apr 2001 17:17:43 -0400 (EDT)", "msg_from": "Chris Bowlby <excalibur@hub.org>", "msg_from_op": true, "msg_subject": "Missing include files." }, { "msg_contents": "> Ok, I believe the postgres.h and all the files in the include/utils\n> directory need to be copied over during the installation, it got most of\n> the others, but missed those ones and as I've spent a good chunk trying to\n> get PHP and a few other utils built, they kinda needed them.\n>\n> 7.1 release candidate 2 has this issue.\n\nThis has been discussed quite a few times recently. From Tom Lane:\n\n\"Change php_pgsql.h to refer to postgres_fe.h not postgres.h.\"\n\n", "msg_date": "Tue, 3 Apr 2001 17:26:23 -0400", "msg_from": "\"Gregory Wood\" <gregw@com-stock.com>", "msg_from_op": false, "msg_subject": "Re: Missing include files." }, { "msg_contents": "See my reports earlier in the release cycle in the archives.\n\nYou need to change PHP to pick up the NEW header.\n\nPHP 4.0.5 should have this fixed. \n\nLER\n\n\n>>>>>>>>>>>>>>>>>> Original Message <<<<<<<<<<<<<<<<<<\n\nOn 4/3/01, 4:17:43 PM, Chris Bowlby <excalibur@hub.org> wrote regarding \n[GENERAL] Missing include files.:\n\n\n> Ok, I believe the postgres.h and all the files in the include/utils\n> directory need to be copied over during the installation, it got most of\n> the others, but missed those ones and as I've spent a good chunk trying \nto\n> get PHP and a few other utils built, they kinda needed them.\n\n> 7.1 release candidate 2 has this issue.\n\n> Chris Bowlby,\n> -----------------------------------------------------\n> Web Developer @ Hub.org.\n> excalibur@hub.org\n> www.hub.org\n> 1-902-542-3657\n> -----------------------------------------------------\n\n\n> ---------------------------(end of broadcast)---------------------------\n> TIP 5: Have you checked our extensive FAQ?\n\n> http://www.postgresql.org/users-lounge/docs/faq.html\n", "msg_date": "Tue, 03 Apr 2001 21:27:08 GMT", "msg_from": "Larry Rosenman <ler@lerctr.org>", "msg_from_op": false, "msg_subject": "Re: Missing include files." }, { "msg_contents": "\nahh....cool, thanks\n\n Chris Bowlby,\n -----------------------------------------------------\n Web Developer @ Hub.org.\n excalibur@hub.org\n www.hub.org\n 1-902-542-3657\n -----------------------------------------------------\n\nOn Tue, 3 Apr 2001, Gregory Wood wrote:\n\n> > Ok, I believe the postgres.h and all the files in the include/utils\n> > directory need to be copied over during the installation, it got most of\n> > the others, but missed those ones and as I've spent a good chunk trying to\n> > get PHP and a few other utils built, they kinda needed them.\n> >\n> > 7.1 release candidate 2 has this issue.\n>\n> This has been discussed quite a few times recently. From Tom Lane:\n>\n> \"Change php_pgsql.h to refer to postgres_fe.h not postgres.h.\"\n>\n>\n\n", "msg_date": "Tue, 3 Apr 2001 17:34:03 -0400 (EDT)", "msg_from": "Chris Bowlby <excalibur@hub.org>", "msg_from_op": true, "msg_subject": "Re: Missing include files." } ]
[ { "msg_contents": "> >> Hmm. Maybe the WAL redo is messing things up??\n> \n> > It could mess up pg_class content, but it never deletes\n> > files (currently).\n> \n> I'm not convinced that any files have really been deleted. Maybe it's\n> just that the pg_class entries are wrong, or even more likely that the\n> indexes on pg_class are messed up (pointing at the wrong \n> pg_class rows).\n\n1. Indices could be recreated with REINDEX or pg_class could be queried\nwith seq scan (something like where relname like '%seq_i___data_buffer%')...\nKonstantin?\n\n2. While trying to reproduce this case (without success yet) I've noticed\nthat in the event of crash just after creation sequence would miss\nmagic number and so nextval would abort. Fixed. Looks like not related\nto reported problem, though.\n\n3. Could you help us reproduce this bug, Konstantin?\nWhat exactly did you do after sequence creation?\nDoes your function reference temp table you've mentioned?\nWhat cause crash? Maybe crash is related somehow...\nCould you try to reproduce failure with wal_debug = 1 and\npost me postmaster' log?\n\nRegards,\nVadim\n", "msg_date": "Tue, 3 Apr 2001 16:36:48 -0700 ", "msg_from": "\"Mikheev, Vadim\" <vmikheev@SECTORBASE.COM>", "msg_from_op": true, "msg_subject": "RE: [BUGS] Loosing files after backend crash " } ]
[ { "msg_contents": "\nDear all,\n\nI've migrated from RedHat6.2/PHP3.0/PostgreSQL6.5 to\nMandrake/PHP4.0/Postgres7.0.2 successfully as far as\npg_dump database_name is concerned.\n\nI am still running BOTH versions on two computers.\n\nPostgreSQL6.5 does not produce any error using math function \"integer\n(float_expression)\"\n(even \"int(float_expression\" is OK with 6.5)\nand 7.0.2 complains saying:\n\n\"function integer(float8) is not found for specified types\"\n\nThe query is used in something like this:\n\nSELECT\n attr-1,\n attr-2,\n\n ...\n\n attr-k,\ninteger(attr-2 * $php_variable_1 + atrr-3 * $php_variable_2) AS attr-N\n\nFROM\n table_1, table_2\nWHERE\n clause_1 AND\n clause_2\nORDER BY\n attr-2, attr-4\n\nIf I made an error HERE with the sintax, I apologise - I am not close to\nmy both Linux machines,\nand sintax is OK in my real SELECT queries.\n\n\"$php_xxx\"s are PHP variables and \"attr-xxx\"s are SQL table attributes.\nThe \"integer(float_expression)\"\nis more complicated then written above BUT it all DOES work in 6.5 and\nDOES NOT in 7.0.2.\n\nAny ideas, please?\n\n\nMuch obliged,\n\nSteven.\n\n\n--\n***********************************************\n\nSteven Vajdic (BSc/Hon, MSc)\nSenior Software Engineer\nMotorola Australia Software Centre (MASC)\n2 Second Avenue, Technology Park\nAdelaide, South Australia 5095\nemail: Steven.Vajdic@motorola.com\nemail: svajdic@asc.corp.mot.com\nPh.: +61-8-8168-3435\nFax: +61-8-8168-3501\nFront Office (Ph): +61-8-8168-3500\n\n----------------------------------------\nmobile: +61 (0)419 860 903\nAFTER WORK email: steven_vajdic@ivillage.com\n----------------------------------------\n\n***********************************************\n\n\n", "msg_date": "Wed, 04 Apr 2001 13:28:35 +0930", "msg_from": "Steven Vajdic <svajdic@asc.corp.mot.com>", "msg_from_op": true, "msg_subject": "Is this a bug in 7.0.2" } ]
[ { "msg_contents": "Hi,\n\nI just tried to go to RC2 on Solaris 7. However the configure fails at the opt-\nreset-test because the linker can not find the library to resolve it.\nThe configureline was:\n./configure --with-perl --with-openssl=/usr/local/ssl --enable-odbc --with-CXX --enable-locale\n\nCould someone give me a hint how to either set LD_LIBRARY_PATH correctly or\nwhat not to include in the configure to get this working?\n\nKonstantin\n-- \nDipl-Inf. Konstantin Agouros aka Elwood Blues. Internet: elwood@agouros.de\nOtkerstr. 28, 81547 Muenchen, Germany. Tel +49 89 69370185\n----------------------------------------------------------------------------\n\"Captain, this ship will not sustain the forming of the cosmos.\" B'Elana Torres\n", "msg_date": "4 Apr 2001 11:26:01 +0200", "msg_from": "elwood@agouros.de (Konstantinos Agouros)", "msg_from_op": true, "msg_subject": "optreset test failed in configure on Solaris 7.1RC2" }, { "msg_contents": "On Wed, Apr 04, 2001 at 11:26:01AM +0200, Konstantinos Agouros wrote:\n> Hi,\n> \n> I just tried to go to RC2 on Solaris 7. However the configure fails at the opt-\n> reset-test because the linker can not find the library to resolve it.\n> The configureline was:\n> ./configure --with-perl --with-openssl=/usr/local/ssl --enable-odbc --with-CXX --enable-locale\n> \n> Could someone give me a hint how to either set LD_LIBRARY_PATH correctly or\n> what not to include in the configure to get this working?\n\nLD_LIBRARY_PATH should set something like:\n\n$ LD_LIBRARY_PATH=YOURLIBPATH:$LD_LIBRARY_PATH\n$ export LD_LIBRARY_PATH\n\nWhere YOURLIBPATH is the path to other libraries. If you still have problems,\nplease send me the output with the errors...\n\nFrank\n", "msg_date": "Wed, 4 Apr 2001 08:14:39 -0500", "msg_from": "Frank Byrum <byrum@greatbridge.com>", "msg_from_op": false, "msg_subject": "Re: optreset test failed in configure on Solaris 7.1RC2" } ]