Datasets:
wanshenl
/
pgsql-hackers

Dataset card Data Studio Files
xet
Community
threads
listlengths
1
2.99k
[ { "msg_contents": "> -----Original Message-----\n> From: Mark Pritchard [mailto:mark@tangent.net.au] \n> Sent: Monday, August 19, 2002 11:27 PM\n> To: Dann Corbit; Neil Conway\n> Cc: Justin Clift; Tom Lane; Christopher Kings-Lynne; \n> pgsql-hackers@postgresql.org\n> Subject: Re: [HACKERS] @(#) Mordred Labs advisory 0x0001: \n> Buffer overflow in\n> \n> \n> On Tue, 20 Aug 2002 15:35, Dann Corbit wrote:\n> > Most computer virus problems are caused by buffer overrun. Someone \n> > decided it wasn't very important.\n> >\n> > Some computer viruses have caused billions of dollars in damage. \n> > Sounds important to me.\n> >\n> > \"Please try our database. Someday, we hope to close off \n> all the virus \n> > entry points, but right now, we figure it isn't too important.\"\n> \n> This sounds a little hysterical to me...don't happen to have \n> a remotely \n> accessible database do you? :)\n\nI tend to be hyperbolic at times.\n \n> > Will you trust your multi-million dollar database to \n> someone who says \n> > the above? I think the priorities are upside down. Any *known* \n> > buffer-overrun _must_ be repaired, and as quickly as possible. And\n> \n> As always, feedback accepted in diff -c format. \n> \n> Seriously though, Oracle was unbreakable for what, two days? \n> Software has \n> bugs. I'm sure there are a stack more in PostgreSQL. \n> \n> You limit your exposure to bugs/defects/etc through the use \n> of multiple layers \n> of protection. If you leave your database out in the wild, \n> you deserve to be \n> hacked. \n\nNobody deserves to be hacked. Security should assume that each link in\nthe chain is the only way to bar the door. IMO-YMMV.\n\n> > potential overruns should be identified. A grep for \n> memcpy, strcpy, \n> > gets, etc. should hunt down most of them. A known buffer overrun \n> > should fill the designer of a product with abject terror. And I \n> > really mean that, literally. If you *know* of a buffer \n> overrun, and \n> > simply decide\n> \n> I'd be worried if my IT consultants experienced \"abject \n> terror\". I much prefer \n> them to be calm, safe in the knowledge that vulnerabilities \n> such as this will \n> not cause me any problems, because they had the forethought \n> to plan for \n> situations like this and limit their exposure.\n\nMy comment was meant to emphasize urgency, rather than irrational\nbehavior. Somewhat hyperbolic, obviously.\n \n> I worry about two pieces of software - Apache and OpenSSH. I \n> compile from \n> source, knowing that I can fix the issue (be it the recent \n> issues with either \n> piece of software) as soon as the fixed source becomes \n> available. I may be in \n> the minority, but at least I don't experience abject terror \n> too often (well, \n> unless I let my sister drive my car...but that is another story).\n> \n> Cheers\n> \n> Mark\n> \n", "msg_date": "Mon, 19 Aug 2002 23:34:56 -0700", "msg_from": "\"Dann Corbit\" <DCorbit@connx.com>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in" } ]
[ { "msg_contents": "Hi all,\n\nI get the following when I attempt to build outside of the source tree.\nAll goes well until....\n\nmake[4]: Entering directory\n`/home/jgray/postgres/pg-build/src/backend/utils/mb/conversion_procs/ascii_and_mic'\nMakefile:11: ../proc.mk: No such file or directory\nmake[4]: *** No rule to make target `../proc.mk'. Stop.\n\n\nAny suggestions. I've tried various things (cf. my previous message to\n-hackers but had no success -I really don't quite get all the details of\nthe build or I would send a patch...)\n\nRegards\n\nJohn\n\n-- \nJohn Gray\t\nAzuli IT\t\nwww.azuli.co.uk\t\n\n\n", "msg_date": "20 Aug 2002 12:06:58 +0100", "msg_from": "John Gray <jgray@azuli.co.uk>", "msg_from_op": true, "msg_subject": "Build failure in current CVS (src/backend/utils/mb/conversion_procs)" }, { "msg_contents": "John Gray <jgray@azuli.co.uk> writes:\n> I get the following when I attempt to build outside of the source tree.\n\nIt looks like the new stuff for loadable conversion modules is not\ncorrect for building outside the source tree; it's being careless about\nthe paths it uses to find files. Tatsuo, can you fix that, or ask Peter\nfor help?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 12:40:52 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS\n\t(src/backend/utils/mb/conversion_procs)" }, { "msg_contents": "Tom Lane writes:\n\n> John Gray <jgray@azuli.co.uk> writes:\n> > I get the following when I attempt to build outside of the source tree.\n>\n> It looks like the new stuff for loadable conversion modules is not\n> correct for building outside the source tree; it's being careless about\n> the paths it uses to find files. Tatsuo, can you fix that, or ask Peter\n> for help?\n\nTatsuo, why is there one directory per encoding combination with only one\nsource file each? Couldn't we put all the files in one directory?\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Tue, 20 Aug 2002 23:22:21 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS\n\t(src/backend/utils/mb/conversion_procs)" }, { "msg_contents": "> Tatsuo, why is there one directory per encoding combination with only one\n> source file each? Couldn't we put all the files in one directory?\n\neuc_tw_and_big5 has *two* files. Also Unicode mapping files could be\nmoved to each directory. So the Unicode related functions might have\none more file. IMO giving a directory to each conversion\nis more manageable than putting everything in single directory.\n--\nTatsuo Ishii\n", "msg_date": "Wed, 21 Aug 2002 10:09:13 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS" }, { "msg_contents": "Tatsuo Ishii <t-ishii@sra.co.jp> writes:\n> IMO giving a directory to each conversion\n> is more manageable than putting everything in single directory.\n\nI lean towards Peter's point here. One or two files per conversion\nis not unmanageable to keep in a single directory. And I have noticed\nthat \"make\" and \"make install\" are visibly slower than before because\nthey have to recurse into all these new subdirectories. I think the\naccumulated developer time spent watching \"make\" will soon exceed any\npossible benefit from splitting these files into subdirectories :-(\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 23:10:54 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS\n\t(src/backend/utils/mb/conversion_procs)" }, { "msg_contents": "> It looks like the new stuff for loadable conversion modules is not\n> correct for building outside the source tree; it's being careless about\n> the paths it uses to find files. Tatsuo, can you fix that, or ask Peter\n> for help?\n\nI have applied following changes and am getting:\n\nmake: *** No rule to make target `ascii_and_mic.o', needed by `libascii_and_mic.so.0.0'. Stop.\n\nunder one of a conversion directory. The weird thing is I do not get\nthis if I do a build \"inside\" the source tree. Any idea?\n\nIndex: proc.mk\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/backend/utils/mb/conversion_procs/proc.mk,v\nretrieving revision 1.2\ndiff -c -r1.2 proc.mk\n*** proc.mk\t8 Aug 2002 07:47:43 -0000\t1.2\n--- proc.mk\t21 Aug 2002 06:09:13 -0000\n***************\n*** 16,21 ****\n clean distclean maintainer-clean: clean-lib\n \t$(RM) $(OBJS)\n \n! include $(top_builddir)/src/Makefile.shlib\n \n all: $(shlib)\n--- 16,21 ----\n clean distclean maintainer-clean: clean-lib\n \t$(RM) $(OBJS)\n \n! include $(top_srcdir)/src/Makefile.shlib\n \n all: $(shlib)\nIndex: ascii_and_mic/Makefile\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/backend/utils/mb/conversion_procs/ascii_and_mic/Makefile,v\nretrieving revision 1.1\ndiff -c -r1.1 Makefile\n*** ascii_and_mic/Makefile\t14 Aug 2002 02:45:10 -0000\t1.1\n--- ascii_and_mic/Makefile\t21 Aug 2002 06:09:13 -0000\n***************\n*** 8,11 ****\n \n NAME\t\t:= ascii_and_mic\n \n! include ../proc.mk\n--- 8,12 ----\n \n NAME\t\t:= ascii_and_mic\n \n! include $(top_srcdir)/src/backend/utils/mb/conversion_procs/proc.mk\n! \n", "msg_date": "Wed, 21 Aug 2002 15:15:23 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS" }, { "msg_contents": "Tatsuo Ishii writes:\n\n> I have applied following changes and am getting:\n>\n> make: *** No rule to make target `ascii_and_mic.o', needed by `libascii_and_mic.so.0.0'. Stop.\n>\n> under one of a conversion directory. The weird thing is I do not get\n> this if I do a build \"inside\" the source tree. Any idea?\n\nThe following patch works, it just needs to be extrapolated to the other\ndirectories.\n\ndiff -ru cvs-pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic/Makefile pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic/Makefile\n--- cvs-pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic/Makefile\tWed Aug 14 04:45:10 2002\n+++ pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic/Makefile\tWed Aug 21 21:22:47 2002\n@@ -3,9 +3,10 @@\n # $Id: Makefile,v 1.1 2002/08/14 02:45:10 ishii Exp $\n #\n #-------------------------------------------------------------------------\n+subdir = src/backend/utils/mb/conversion_procs/ascii_and_mic\n top_builddir = ../../../../../..\n include $(top_builddir)/src/Makefile.global\n\n NAME\t\t:= ascii_and_mic\n\n-include ../proc.mk\n+include $(srcdir)/../proc.mk\ndiff -ru cvs-pgsql/src/backend/utils/mb/conversion_procs/proc.mk pgsql/src/backend/utils/mb/conversion_procs/proc.mk\n--- cvs-pgsql/src/backend/utils/mb/conversion_procs/proc.mk\tSat Aug 10 00:53:26 2002\n+++ pgsql/src/backend/utils/mb/conversion_procs/proc.mk\tWed Aug 21 21:12:29 2002\n@@ -16,6 +16,6 @@\n clean distclean maintainer-clean: clean-lib\n \t$(RM) $(OBJS)\n\n-include $(top_builddir)/src/Makefile.shlib\n+include $(top_srcdir)/src/Makefile.shlib\n\n all: $(shlib)\n===end\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Wed, 21 Aug 2002 22:48:07 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS" }, { "msg_contents": "\nOK, patch applied to all Makefiles, as outlined by Peter.\n\n---------------------------------------------------------------------------\n\nPeter Eisentraut wrote:\n> Tatsuo Ishii writes:\n> \n> > I have applied following changes and am getting:\n> >\n> > make: *** No rule to make target `ascii_and_mic.o', needed by `libascii_and_mic.so.0.0'. Stop.\n> >\n> > under one of a conversion directory. The weird thing is I do not get\n> > this if I do a build \"inside\" the source tree. Any idea?\n> \n> The following patch works, it just needs to be extrapolated to the other\n> directories.\n> \n> diff -ru cvs-pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic/Makefile pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic/Makefile\n> --- cvs-pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic/Makefile\tWed Aug 14 04:45:10 2002\n> +++ pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic/Makefile\tWed Aug 21 21:22:47 2002\n> @@ -3,9 +3,10 @@\n> # $Id: Makefile,v 1.1 2002/08/14 02:45:10 ishii Exp $\n> #\n> #-------------------------------------------------------------------------\n> +subdir = src/backend/utils/mb/conversion_procs/ascii_and_mic\n> top_builddir = ../../../../../..\n> include $(top_builddir)/src/Makefile.global\n> \n> NAME\t\t:= ascii_and_mic\n> \n> -include ../proc.mk\n> +include $(srcdir)/../proc.mk\n> diff -ru cvs-pgsql/src/backend/utils/mb/conversion_procs/proc.mk pgsql/src/backend/utils/mb/conversion_procs/proc.mk\n> --- cvs-pgsql/src/backend/utils/mb/conversion_procs/proc.mk\tSat Aug 10 00:53:26 2002\n> +++ pgsql/src/backend/utils/mb/conversion_procs/proc.mk\tWed Aug 21 21:12:29 2002\n> @@ -16,6 +16,6 @@\n> clean distclean maintainer-clean: clean-lib\n> \t$(RM) $(OBJS)\n> \n> -include $(top_builddir)/src/Makefile.shlib\n> +include $(top_srcdir)/src/Makefile.shlib\n> \n> all: $(shlib)\n> ===end\n> \n> -- \n> Peter Eisentraut peter_e@gmx.net\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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 17:34:18 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> OK, patch applied to all Makefiles, as outlined by Peter.\n\nI see this in current CVS:\n\nmake[3]: Entering directory `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs'\nmake[4]: Entering directory `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\nMakefile:1: *** missing separator. Stop.\nmake[4]: Leaving directory `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\nmake[3]: *** [all] Error 2\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "21 Aug 2002 19:12:41 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS" }, { "msg_contents": "I get the same - FreeBSD/Alpha.\n\nChris\n\n> -----Original Message-----\n> From: pgsql-hackers-owner@postgresql.org\n> [mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Neil Conway\n> Sent: Thursday, 22 August 2002 7:13 AM\n> To: Bruce Momjian\n> Cc: Peter Eisentraut; Tatsuo Ishii; tgl@sss.pgh.pa.us;\n> jgray@azuli.co.uk; pgsql-hackers@postgresql.org\n> Subject: Re: [HACKERS] Build failure in current CVS\n> \n> \n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > OK, patch applied to all Makefiles, as outlined by Peter.\n> \n> I see this in current CVS:\n> \n> make[3]: Entering directory \n> `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs'\n> make[4]: Entering directory \n> `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\n> Makefile:1: *** missing separator. Stop.\n> make[4]: Leaving directory \n> `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\n> make[3]: *** [all] Error 2\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\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> \n\n", "msg_date": "Thu, 22 Aug 2002 10:11:16 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS" }, { "msg_contents": "\nThanks. Fixed. I had a '[' on the first line of one of the makefiles.\n\n---------------------------------------------------------------------------\n\nNeil Conway wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > OK, patch applied to all Makefiles, as outlined by Peter.\n> \n> I see this in current CVS:\n> \n> make[3]: Entering directory `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs'\n> make[4]: Entering directory `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\n> Makefile:1: *** missing separator. Stop.\n> make[4]: Leaving directory `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\n> make[3]: *** [all] Error 2\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\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> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 22:18:58 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS" }, { "msg_contents": "I appreciate you and other guys who has been working for this\nproblem.\n--\nTatsuo Ishii\n\n> Thanks. Fixed. I had a '[' on the first line of one of the makefiles.\n\n> ---------------------------------------------------------------------------\n> \n> Neil Conway wrote:\n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > OK, patch applied to all Makefiles, as outlined by Peter.\n> > \n> > I see this in current CVS:\n> > \n> > make[3]: Entering directory `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs'\n> > make[4]: Entering directory `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\n> > Makefile:1: *** missing separator. Stop.\n> > make[4]: Leaving directory `/home/nconway/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\n> > make[3]: *** [all] Error 2\n> > \n> > Cheers,\n> > \n> > Neil\n> > \n> > -- \n> > Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\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> > \n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n> \n", "msg_date": "Thu, 22 Aug 2002 11:45:58 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": false, "msg_subject": "Re: Build failure in current CVS" } ]
[ { "msg_contents": "I just got the latest beta and it compiles ecpg grammar correctly! I had\nto make one change to my source though as bison no longer accepts a comma inside the token list.\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": "Tue, 20 Aug 2002 14:25:47 +0200", "msg_from": "Michael Meskes <meskes@postgresql.org>", "msg_from_op": true, "msg_subject": "bison news" }, { "msg_contents": "Michael Meskes <meskes@postgresql.org> writes:\n> I just got the latest beta and it compiles ecpg grammar correctly!\n\nThis is good. Any word on when it will go to an official release?\n\nBTW, I spent some time looking at the problem, and it seems the issue\nis not overrun of any bison internal table, but failure to compress the\nresulting \"action table\" into 32K entries. This means that the required\nexpansion from short to int is not just a cost paid while you run bison;\nthe actual table in the ecpg executable will double in size. I trust\nthey did not fix the problem in a way that causes *every* generated\nparser to use an int[] rather than short[] action table ...\n\nAlso, it seemed to me that the most leverage on the size of the\ncompressed action table would be gained by reducing the number of\nterminal symbols, more so than the number of rules. Dunno if there\nis a lot you can do about that, but it's a thought.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 11:10:01 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: bison news " }, { "msg_contents": "On Tue, Aug 20, 2002 at 11:10:01AM -0400, Tom Lane wrote:\n> BTW, I spent some time looking at the problem, and it seems the issue\n> is not overrun of any bison internal table, but failure to compress the\n> resulting \"action table\" into 32K entries. This means that the required\n\nOuch! This of course is not so much a problem for ecpg but for the\nbackend should we run into the problem there too.\n\n> ...\n> Also, it seemed to me that the most leverage on the size of the\n> compressed action table would be gained by reducing the number of\n> terminal symbols, more so than the number of rules. Dunno if there\n> is a lot you can do about that, but it's a thought.\n\nWill look at it.\n\nMichael\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n", "msg_date": "Tue, 20 Aug 2002 17:21:56 +0200", "msg_from": "Michael Meskes <meskes@postgresql.org>", "msg_from_op": true, "msg_subject": "Re: bison news" }, { "msg_contents": "Michael Meskes <meskes@postgresql.org> writes:\n> On Tue, Aug 20, 2002 at 11:10:01AM -0400, Tom Lane wrote:\n>> BTW, I spent some time looking at the problem, and it seems the issue\n>> is not overrun of any bison internal table, but failure to compress the\n>> resulting \"action table\" into 32K entries. This means that the required\n\n> Ouch! This of course is not so much a problem for ecpg but for the\n> backend should we run into the problem there too.\n\nAs of CVS tip a few days ago, the backend's action table was about 27K\nentries. So we have some breathing room, but certainly in the\nforeseeable future there will be a problem...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 11:32:52 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: bison news " }, { "msg_contents": "\nOK, now that _a_ bison exists that works, how does this effect our\nrelease? I don't see preproc.[ch] in CVS. Do we need this new bison\nversion on postgresql.org because Marc generates these as part of his\ninstall script?\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> Michael Meskes <meskes@postgresql.org> writes:\n> > On Tue, Aug 20, 2002 at 11:10:01AM -0400, Tom Lane wrote:\n> >> BTW, I spent some time looking at the problem, and it seems the issue\n> >> is not overrun of any bison internal table, but failure to compress the\n> >> resulting \"action table\" into 32K entries. This means that the required\n> \n> > Ouch! This of course is not so much a problem for ecpg but for the\n> > backend should we run into the problem there too.\n> \n> As of CVS tip a few days ago, the backend's action table was about 27K\n> entries. So we have some breathing room, but certainly in the\n> foreseeable future there will be a problem...\n> \n> \t\t\tregards, tom lane\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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 20 Aug 2002 16:30:07 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: bison news" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> OK, now that _a_ bison exists that works, how does this effect our\n> release? I don't see preproc.[ch] in CVS. Do we need this new bison\n> version on postgresql.org because Marc generates these as part of his\n> install script?\n\nI don't think we want a beta bison on postgres.org. Let's see if we can\nhold out for a release...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 16:34:10 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: bison news " }, { "msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > OK, now that _a_ bison exists that works, how does this effect our\n> > release? I don't see preproc.[ch] in CVS. Do we need this new bison\n> > version on postgresql.org because Marc generates these as part of his\n> > install script?\n> \n> I don't think we want a beta bison on postgres.org. Let's see if we can\n> hold out for a release...\n\nWell, we had better get it on or it will get zero testing, and we _need_\nit for the 7.3 release of ecpg, because as I remember, we didn't have\nany other good backup plans. ;-)\n\nThis may be a case where we have to do some beta testing on our own. I\nwill grab the bison beta myself for my machine.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 20 Aug 2002 17:17:46 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: bison news" }, { "msg_contents": "Bruce Momjian writes:\n\n> This may be a case where we have to do some beta testing on our own. I\n> will grab the bison beta myself for my machine.\n\nI imagine that bison doesn't get a lot of beta testing, since people don't\nhave a whole bunch of production grammars lying around that they want to\nupgrade at the earliest possible moment.\n\nSo if we want to test it more, we could propagate it to our beta testers.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Wed, 21 Aug 2002 22:45:42 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: bison news" } ]
[ { "msg_contents": "\n> > with Perl and *using placeholders and bind values*, the application\n> > developer has not to worry about this. So, usually I don't check the\n> > values in my applications (e.g. if only values between 1 and 5 are\n> > allowed and under normal circumstances only these are possible), it's the\n> > task of the database (check constraint). \n> \n> That's the idea. It's the job of the database to guarantee data\n> integrety.\n\nYes, but what is currently missing is a protocol to the backend\nwhere a statement is prepared with placeholders and then executed\n(multiple times) with given values. Then there is no doubt what is a\nvalue, and what a part of the SQL.\n\nI think that this would be a wanted feature of the next\nprotocol version. iirc the backend side part is currently beeing \nimplemented.\n\nAndreas\n", "msg_date": "Tue, 20 Aug 2002 18:14:39 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: [SECURITY] DoS attack on backend possible" }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n\n> Yes, but what is currently missing is a protocol to the backend\n> where a statement is prepared with placeholders and then executed\n> (multiple times) with given values. Then there is no doubt what is a\n> value, and what a part of the SQL.\n\nThis wouldn't have helped in the current case. The bug is in the\ndatetime parser which translates strings to an external\nrepresentation, not in the SQL parser.\n\n-- \nFlorian Weimer \t Weimer@CERT.Uni-Stuttgart.DE\nUniversity of Stuttgart http://CERT.Uni-Stuttgart.DE/people/fw/\nRUS-CERT fax +49-711-685-5898\n", "msg_date": "Tue, 20 Aug 2002 18:31:28 +0200", "msg_from": "Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>", "msg_from_op": false, "msg_subject": "Re: [SECURITY] DoS attack on backend possible" } ]
[ { "msg_contents": "\n> The cash_out problem can't really be fixed until we do something about\n> subdividing type \"opaque\" into multiple pseudo-types with more carefully\n> defined meanings. cash_out is declared cash_out(opaque) which does not\n> really mean that it accepts any input type ... but one of the several\n> meanings of \"opaque\" is \"accepts any type\", so the parser \n> doesn't reject cash_out(2).\n\nWould it be possible to update the system tables, so that cash_out does not take\nopaque but really takes type money ?\nI mean the first thing cash_out does is PG_GETARG_CASH(0), so it really only copes \nwith a money type.\n\nI know the problem is that the cat chases its tail here, because of what comes first,\nthe type or the io functions. But couldn't this be overcome, at least for internal types ?\n\nAndreas\n", "msg_date": "Tue, 20 Aug 2002 18:36:58 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> Would it be possible to update the system tables, so that cash_out does not take\n> opaque but really takes type money ?\n\nThat is part of the solution, but only part: we have hundreds of\nfunctions that take \"opaque\" because we don't currently have any way\nto declare what they really take. (In particular, all the typinput\nfunctions are like that --- so fixing typoutput functions isn't plugging\neven half of the gap.)\n\nSee my proposal to make \"opaque\" obsolete.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 14:50:00 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " } ]
[ { "msg_contents": "Large file support is now compiled by default if available. (Use\n--disable-largefile to turn it off. That's what Autoconf gives us.)\n\nBut:\n\nThe zlib library uses unsigned ints and unsigned longs for file positions\nand offsets. Depending on how that is used in detail and depending on how\nzlib itself is compiled, this may or may not work.\n\nThe tar file format (POSIX and traditional) has an inherent limitation on\nthe size of the member files of 2^33 bytes (pg_dump currently only handles\n2^30). The result in that case continues to be a broken archive. The GNU\ntar format has an extension that would handle 2^89 bytes. This may be\nsomething interesting to work on.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Tue, 20 Aug 2002 20:03:18 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": true, "msg_subject": "Large file support available" }, { "msg_contents": "Peter Eisentraut <peter_e@gmx.net> writes:\n> Large file support is now compiled by default if available.\n\nI am now getting (on HPUX 10.20)\n\n/usr/include/sys/resource.h: In function `getrlimit':\n/usr/include/sys/resource.h:168: warning: implicit declaration of function `__getrlimit64'\n/usr/include/sys/resource.h: In function `setrlimit':\n/usr/include/sys/resource.h:170: warning: implicit declaration of function `__setrlimit64'\n\nfor essentially every file in the system. A little digging shows that\nthis is happening because _FILE64 is defined and _LARGEFILE64_SOURCE\nis not; this is evidently a Bad Idea on HPUX.\n\nFurther digging shows that noplace in the standard headers is\n_LARGEFILE64_SOURCE #define'd, so evidently one is supposed to supply it\nfrom user headers. Ugh. Please add this to the list of\nplatform-specific symbols that had better be turned on to support large\nfiles.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 15:38:24 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available " }, { "msg_contents": "Also, even with configure --disable-largefile, I find that pg_config.h\nstill contains\n\n/* Define to 1 to make fseeko visible on some hosts. */\n#define _LARGEFILE_SOURCE 1\n\n/* Define to 1 if fseeko (and presumably ftello) exists and is declared. */\n#define HAVE_FSEEKO 1\n\nThis strikes me as probably Not a Good Thing, although I haven't dug to\nsee what the implications are.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 15:49:59 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available " }, { "msg_contents": "> Large file support is now compiled by default if available. (Use\n> --disable-largefile to turn it off. That's what Autoconf gives us.)\n\nAre you sure that backend gains more performance than 1GB segmented\nfile (I mean large file support turn on LET_OS_MANAGE_FILESIZE)? I\nmyself have not tried yet, but a linux kernel hacker around me gave\nthis question sometime ago.\n--\nTatsuo Ishii\n", "msg_date": "Wed, 21 Aug 2002 10:09:26 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": false, "msg_subject": "Re: Large file support available" }, { "msg_contents": "Tatsuo Ishii writes:\n\n> Are you sure that backend gains more performance than 1GB segmented\n> file (I mean large file support turn on LET_OS_MANAGE_FILESIZE)?\n\nNo idea. My change only enables access to large files, it doesn't change\nthe segmentation logic in the backend. The main use at this point is for\npg_dump-related activities.\n\nIn fact, while the large file support API can handle 64-bit offsets, its\navailability and use don't guarantee that the file system will support any\nparticular file size. So the segmentation logic in the backend isn't\ngoing anywhere, as far as I'm concerned.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Wed, 21 Aug 2002 22:44:38 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": true, "msg_subject": "Re: Large file support available" }, { "msg_contents": "Tom Lane writes:\n\n> Also, even with configure --disable-largefile, I find that pg_config.h\n> still contains\n>\n> /* Define to 1 to make fseeko visible on some hosts. */\n> #define _LARGEFILE_SOURCE 1\n>\n> /* Define to 1 if fseeko (and presumably ftello) exists and is declared. */\n> #define HAVE_FSEEKO 1\n>\n> This strikes me as probably Not a Good Thing, although I haven't dug to\n> see what the implications are.\n\nThis is harmless (until proven otherwise). fseeko() is identical to\nfseek() except that the offset argument uses off_t, and _LARGEFILE_SOURCE\nmakes fseeko() and friends visible in the headers. That's all. No large\nfiles involved.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Wed, 21 Aug 2002 22:44:59 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": true, "msg_subject": "Re: Large file support available " }, { "msg_contents": "Tom Lane writes:\n\n> /usr/include/sys/resource.h: In function `getrlimit':\n> /usr/include/sys/resource.h:168: warning: implicit declaration of function `__getrlimit64'\n> /usr/include/sys/resource.h: In function `setrlimit':\n> /usr/include/sys/resource.h:170: warning: implicit declaration of function `__setrlimit64'\n>\n> for essentially every file in the system. A little digging shows that\n> this is happening because _FILE64 is defined and _LARGEFILE64_SOURCE\n> is not; this is evidently a Bad Idea on HPUX.\n\nYou're supposed to define _LARGEFILE64_SOURCE if you want to use functions\nlike open64(), fseek64(), getrlimit64(), etc. in your source. We don't\nwant those, obviously.\n\nWhat is happening here is that evidently the system headers effectively\nredefine getrlimit() to point to getrlimit64() if FILE_OFFSET_BITS=64,\nwhich is the usual strategy for all the I/O functions. But you're not\nsupposed to have to define _LARGEFILE64_SOURCE for this, because the\nchange is supposed to be transparent.\n\nIf the {s|g}etrlimit warnings are indeed the only ones (i.e., none about\nopen, fseek, write, read, etc.) then this is either a bug or there's\nsomething wrong in the include file order or something like that. Which\nway is sys/resource.h included anyway?\n\nIf there's no way to fix it then we can add a definition of\n_LARGEFILE64_SOURCE to hpux.h and consider further action.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Wed, 21 Aug 2002 22:45:20 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": true, "msg_subject": "Re: Large file support available " }, { "msg_contents": "Peter Eisentraut wrote:\n> Tom Lane writes:\n> \n> > Also, even with configure --disable-largefile, I find that pg_config.h\n> > still contains\n> >\n> > /* Define to 1 to make fseeko visible on some hosts. */\n> > #define _LARGEFILE_SOURCE 1\n> >\n> > /* Define to 1 if fseeko (and presumably ftello) exists and is declared. */\n> > #define HAVE_FSEEKO 1\n> >\n> > This strikes me as probably Not a Good Thing, although I haven't dug to\n> > see what the implications are.\n> \n> This is harmless (until proven otherwise). fseeko() is identical to\n> fseek() except that the offset argument uses off_t, and _LARGEFILE_SOURCE\n> makes fseeko() and friends visible in the headers. That's all. No large\n> files involved.\n\nI am confused. fseeko() doesn't look standard to me. I though\nfgetpos/fsetpos() where the standard interfaces for large file support; \nfrom BSD/OS:\n\n The fgetpos(), fsetpos(), fseek(), ftell(), and rewind() functions con-\n form to ANSI C X3.159-1989 (``ANSI C '').\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 17:14:53 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available" }, { "msg_contents": "Peter Eisentraut <peter_e@gmx.net> writes:\n> If the {s|g}etrlimit warnings are indeed the only ones (i.e., none about\n> open, fseek, write, read, etc.) then this is either a bug or there's\n> something wrong in the include file order or something like that.\n\nNo such luck. Here's a more complete excerpt of one typical failure:\n\ngcc -O1 -Wall -Wmissing-prototypes -Wmissing-declarations -g -I../../../../src/include -c -o tuptoaster.o tuptoaster.c\nIn file included from /usr/include/sys/wait.h:83,\n from /usr/local/lib/gcc-lib/hppa2.0-hp-hpux10.20/2.95.3/include/stdlib.h:231,\n from ../../../../src/include/c.h:56,\n from ../../../../src/include/postgres.h:47,\n from tuptoaster.c:25:\n/usr/include/sys/resource.h: In function `getrlimit':\n/usr/include/sys/resource.h:168: warning: implicit declaration of function `__getrlimit64'\n/usr/include/sys/resource.h: In function `setrlimit':\n/usr/include/sys/resource.h:170: warning: implicit declaration of function `__setrlimit64'\nIn file included from /usr/include/unistd.h:11,\n from tuptoaster.c:27:\n/usr/include/sys/unistd.h: In function `truncate':\n/usr/include/sys/unistd.h:539: warning: implicit declaration of function `__truncate64'\n/usr/include/sys/unistd.h: In function `prealloc':\n/usr/include/sys/unistd.h:543: warning: implicit declaration of function `__prealloc64'\n/usr/include/sys/unistd.h: In function `lockf':\n/usr/include/sys/unistd.h:544: warning: implicit declaration of function `__lockf64'\n/usr/include/sys/unistd.h: In function `ftruncate':\n/usr/include/sys/unistd.h:545: warning: implicit declaration of function `__ftruncate64'\nIn file included from /usr/include/fcntl.h:9,\n from tuptoaster.c:28:\n/usr/include/sys/fcntl.h: In function `open':\n/usr/include/sys/fcntl.h:216: warning: implicit declaration of function `__open64'\n/usr/include/sys/fcntl.h: In function `creat':\n/usr/include/sys/fcntl.h:217: warning: implicit declaration of function `__creat64'\n\nAFAICT a *lot* of HPUX headers expect you to #define _LARGEFILE64_SOURCE\nif you want this stuff to work.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 00:55:11 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available " }, { "msg_contents": "\nPeter, I have received no reply to this question.\n\n---------------------------------------------------------------------------\n\nBruce Momjian wrote:\n> Peter Eisentraut wrote:\n> > Tom Lane writes:\n> > \n> > > Also, even with configure --disable-largefile, I find that pg_config.h\n> > > still contains\n> > >\n> > > /* Define to 1 to make fseeko visible on some hosts. */\n> > > #define _LARGEFILE_SOURCE 1\n> > >\n> > > /* Define to 1 if fseeko (and presumably ftello) exists and is declared. */\n> > > #define HAVE_FSEEKO 1\n> > >\n> > > This strikes me as probably Not a Good Thing, although I haven't dug to\n> > > see what the implications are.\n> > \n> > This is harmless (until proven otherwise). fseeko() is identical to\n> > fseek() except that the offset argument uses off_t, and _LARGEFILE_SOURCE\n> > makes fseeko() and friends visible in the headers. That's all. No large\n> > files involved.\n> \n> I am confused. fseeko() doesn't look standard to me. I though\n> fgetpos/fsetpos() where the standard interfaces for large file support; \n> from BSD/OS:\n> \n> The fgetpos(), fsetpos(), fseek(), ftell(), and rewind() functions con-\n> form to ANSI C X3.159-1989 (``ANSI C '').\n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 17:55:10 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available" }, { "msg_contents": "\nOK, with no one replying to this, I will take it upon myself to resolve\nthis. According to the Mac OSX fseek() manual page:\n\n The fgetpos(), fsetpos(), fseek(), ftell(), and rewind() functions\n conform to ANSI X3.159-1989 (``ANSI C'').\n\n The fseeko() and ftello() functions conform to Version 2 of the\n Single UNIX Specification (``SUSv2'').\n\nwhich basically says that we should be using fseek or preferably\nfseekpos, not fseeko. I realize that the advantage of fseeko is that it\nhas the same API as fseek, but if we are going to fix this, we may as\nwell do it right and use fgetpos if we have it.\n\nIs there anyone who has fseeko() but _not_ fsetpos()?\n\n---------------------------------------------------------------------------\n\nBruce Momjian wrote:\n> \n> Peter, I have received no reply to this question.\n> \n> ---------------------------------------------------------------------------\n> \n> Bruce Momjian wrote:\n> > Peter Eisentraut wrote:\n> > > Tom Lane writes:\n> > > \n> > > > Also, even with configure --disable-largefile, I find that pg_config.h\n> > > > still contains\n> > > >\n> > > > /* Define to 1 to make fseeko visible on some hosts. */\n> > > > #define _LARGEFILE_SOURCE 1\n> > > >\n> > > > /* Define to 1 if fseeko (and presumably ftello) exists and is declared. */\n> > > > #define HAVE_FSEEKO 1\n> > > >\n> > > > This strikes me as probably Not a Good Thing, although I haven't dug to\n> > > > see what the implications are.\n> > > \n> > > This is harmless (until proven otherwise). fseeko() is identical to\n> > > fseek() except that the offset argument uses off_t, and _LARGEFILE_SOURCE\n> > > makes fseeko() and friends visible in the headers. That's all. No large\n> > > files involved.\n> > \n> > I am confused. fseeko() doesn't look standard to me. I though\n> > fgetpos/fsetpos() where the standard interfaces for large file support; \n> > from BSD/OS:\n> > \n> > The fgetpos(), fsetpos(), fseek(), ftell(), and rewind() functions con-\n> > form to ANSI C X3.159-1989 (``ANSI C '').\n> > \n> > -- \n> > Bruce Momjian | http://candle.pha.pa.us\n> > pgman@candle.pha.pa.us | (610) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Fri, 23 Aug 2002 23:44:03 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Is there anyone who has fseeko() but _not_ fsetpos()?\n\nAFAICT this is completely irrelevant to large files.\n\n\t\t\tregards, tom lane\n", "msg_date": "Sat, 24 Aug 2002 00:25:38 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available " }, { "msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Is there anyone who has fseeko() but _not_ fsetpos()?\n> \n> AFAICT this is completely irrelevant to large files.\n\nPlease explain. At:\n\n\thttp://man.dnswatch.com/cgi-bin/htmlman?fseek+3\n\nI see:\n\n The fseeko() function is identical to fseek(), except it takes an off_t\n argument instead of a long. Likewise, the ftello() function is identical\n to ftell(), except it returns an off_t.\n\nwhile fsetpos() is:\n\n fsetpos(FILE *stream, const fpos_t *pos);\n\nand presumably fpos_t handles long files too.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sat, 24 Aug 2002 00:36:08 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> I see:\n> The fseeko() function is identical to fseek(), except it takes an off_t\n> argument instead of a long. Likewise, the ftello() function is identical\n> to ftell(), except it returns an off_t.\n\nIndeed. Notice the complete lack of any commitment about the size of\noff_t ...\n\n> while fsetpos() is:\n> fsetpos(FILE *stream, const fpos_t *pos);\n\n... or the size of fpos_t.\n\nYou might find it illuminating to read this random extract from the\nHPUX 10.20 man pages:\n\nNAME\n fgetpos64(), fopen64(), freopen64(), fseeko64(), fsetpos64(),\n fstatvfsdev64(), ftello64(), ftw64(), nftw64(), statvfsdev64(),\n tmpfile64() - non-POSIX standard API interfaces to support large\n files.\n\nDESCRIPTION\n New API's to support large files. These API interfaces are not a part\n of the POSIX standard and may be removed in the future.\n\n fgetpos64() The fgetpos64() function is identical to\n fgetpos() except that fgetpos64() returns the\n position in a fpos64_t instead of a fpos_t. All\n other functional behaviors, returns, and errors\n are identical.\n\n ... etc ...\n\nI don't see any reason to believe that fgetpos buys us anything but\nnotational inconvenience. It certainly doesn't buy large file support,\nat least not without the same behind-the-scenes redefinitions needed for\nfseek/fseeko and friends...\n\n\t\t\tregards, tom lane\n", "msg_date": "Sat, 24 Aug 2002 00:45:46 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available " }, { "msg_contents": "Bruce Momjian wrote:\n\n>OK, with no one replying to this, I will take it upon myself to resolve\n>this. According to the Mac OSX fseek() manual page:\n>\n> The fgetpos(), fsetpos(), fseek(), ftell(), and rewind() functions\n> conform to ANSI X3.159-1989 (``ANSI C'').\n>\n> The fseeko() and ftello() functions conform to Version 2 of the\n> Single UNIX Specification (``SUSv2'').\n> \n>\n\nI might be veering *slightly* off the topic here, but since I got bitten \nby this recently I thought I would mention it:\n\nOn Linux, and found that I needed\n\n<#include asm/fcntl.h>\ninstead of\n<#include fcntl.h>\n\nwhen using lseek. I had expected defining _FILE_OFFSET_BITS=64 to sort \nthis (which it did not).\n\nI think that this will only be an issue if folk want relation files to \nbe chunked at > 2G (or want to define LET_OS_MANAGE_FILES).\n\nbest wishes\n\nMark\n\n", "msg_date": "Sat, 24 Aug 2002 16:57:07 +1200", "msg_from": "Mark Kirkwood <markir@slingshot.co.nz>", "msg_from_op": false, "msg_subject": "Re: Large file support available" }, { "msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > I see:\n> > The fseeko() function is identical to fseek(), except it takes an off_t\n> > argument instead of a long. Likewise, the ftello() function is identical\n> > to ftell(), except it returns an off_t.\n> \n> Indeed. Notice the complete lack of any commitment about the size of\n> off_t ...\n> \n> > while fsetpos() is:\n> > fsetpos(FILE *stream, const fpos_t *pos);\n> \n> ... or the size of fpos_t.\n> \n> You might find it illuminating to read this random extract from the\n> HPUX 10.20 man pages:\n...\n> I don't see any reason to believe that fgetpos buys us anything but\n> notational inconvenience. It certainly doesn't buy large file support,\n> at least not without the same behind-the-scenes redefinitions needed for\n> fseek/fseeko and friends...\n\nClearly there is the issues that fseek uses long, which isn't enough for\nlarge file support. On BSD/OS, we have fsetpos, which is the way we do\nlarge file support:\n\n int\n fseek(FILE *stream, long offset, int whence);\n\n int\n fsetpos(FILE *stream, const fpos_t *pos);\n\nMy point is that it seems fsetpos is the approved way of accessing large\nfiles, rather than fseeko. In fact, I don't have fseeko here but I do\nhave fsetpos, and it does handle large files because my includes have\nthis:\n\t\n\ttypedef off_t fpos_t\n\ttypedef quad_t off_t;\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sat, 24 Aug 2002 07:43:24 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available" }, { "msg_contents": "Bruce Momjian writes:\n\n> My point is that it seems fsetpos is the approved way of accessing large\n> files, rather than fseeko. In fact, I don't have fseeko here but I do\n> have fsetpos, and it does handle large files because my includes have\n> this:\n>\n> \ttypedef off_t fpos_t\n> \ttypedef quad_t off_t;\n\nInteresting. In general, you can't rely on fpos_t being an integral type,\nwhich indeed on my machine it isn't. But for pg_dump we need an integral\ntype because we do offset arithmetic.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Tue, 27 Aug 2002 20:27:17 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": true, "msg_subject": "Re: Large file support available" }, { "msg_contents": "Peter Eisentraut wrote:\n> Bruce Momjian writes:\n> \n> > My point is that it seems fsetpos is the approved way of accessing large\n> > files, rather than fseeko. In fact, I don't have fseeko here but I do\n> > have fsetpos, and it does handle large files because my includes have\n> > this:\n> >\n> > \ttypedef off_t fpos_t\n> > \ttypedef quad_t off_t;\n> \n> Interesting. In general, you can't rely on fpos_t being an integral type,\n> which indeed on my machine it isn't. But for pg_dump we need an integral\n> type because we do offset arithmetic.\n\nOh, is that why fsetpos is always SEEK_SET and not SET_CURR or offset\nstuff. Strange I don't have fseeko and do have large file support. \nBSD/OS has had it for years. I guess they just do off_t arithmetic, but\nthat isn't portable.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 14:32:03 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Large file support available" } ]
[ { "msg_contents": "Okay, I'm tired of hearing about this, and I've decided I can make the problem\ngo away with a day or so's concentrated work.\n\nHere is the plan:\n\nCreate several \"pseudo types\" (typtype 'p' in pg_type). We already have one\npseudo-type (RECORD), plus the logic in heap.c to reject any attempt to create\na table column that's of a pseudo-type. I think we need the following\npseudotypes:\n\n\tcstring\t\t-- represents a null-terminated C string\n\tanytype\t\t-- for count(*), nullvalue, nonnullvalue *ONLY*\n\tanyarraytype\t-- for array_eq, array_dims\n\tvoid\t\t-- result type for functions with no useful result\n\ttuple\t\t-- result type for BEFORE triggers\n\tinternal\t-- catchall for non-SQL internal data structures\n\nI am not by any means wedded to the above type names; does anyone have\nsuggestions for better names? (In particular, I am wondering if \"tuple\"\nand \"internal\" would be better named \"pg_tuple\" and \"pg_internal\". We\nmight also consider making a type specifically named \"trigger\" and using\nthat to declare trigger functions, rather than \"tuple\".)\n\nI am also thinking of creating a pseudotype for \"opaque\" itself, so that\nwe can get rid of the kluge of using type OID 0 in places where a valid type\nis expected. We cannot remove \"opaque\" completely (yet) because of backwards\ncompatibility with existing user I/O functions and triggers; but we can see\nto it that no built-in or contrib function is declared with \"opaque\".\n\nAbout I/O behavior: the pg_type entries for these pseudo-types will have to\nhave typinput and typoutput functions. In general these I/O routines must\njust throw errors. Otherwise you could break the intended type safety by\nsupplying user-written constants. For instance, the present definition of\nRECORD is wrong (sorry Joe) because it uses oidin and oidout; so I could\nwrite \"SELECT foo('42'::record)\" and thereby crash a function expecting\nRECORD. Not that there will be any such function, but the analogous case\nwith, say, INTERNAL would be bad news.\n\nAn exception is that void_out should succeed and just return an empty string;\nthis allows functions-returning-void to be called by SELECT and behave\nreasonably. Less obviously, void_in should succeed (and return nothing\ninteresting, probably just a zero datum; it can ignore its input). This\nallows plpgsql functions to be defined to return VOID.\n\nI am also considering allowing cstring_out to succeed (and, of course,\njust return a copy of what it's given). That would allow explicit invocation\nof an output function to work, viz \"SELECT cash_out('42'::money)\" would\nactually do what the user expects. This is not really necessary though.\nWe can't allow cstring_in to succeed, although that looks like a no-brainer,\nbecause the system isn't ready to support CSTRING Datums in general contexts.\n\nTrigger functions will now be expected to take no arguments and return\neither tuple (or trigger if we call it that) or opaque. It would also be\nsensible to allow VOID in the case of AFTER triggers, but I'm inclined not\nto do so: I think it's better that a trigger function be declared in a way\nthat makes it clear it's supposed to be a trigger. If CREATE TRIGGER accepts\nfunctions returning void then I think you lose some useful error checking.\n\nShould we throw a NOTICE stating that opaque is deprecated if a trigger\nis declared with opaque? Or should we wait a release or two for that?\n\nSimilarly, the preferred signature for I/O functions now uses cstring\nand the function's actual datatype, rather than OPAQUE (and again we\ncould consider throwing a NOTICE).\n\nOne of the original concerns about this was how to handle the circularity\nproblem for user-defined I/O functions. If we don't do anything special,\nthen it will still work as long as people define the input function first:\n\tcreate function foo_in(cstring) returns foo as ' ... ';\n\t-- you'll get a NOTICE here about \"type foo is not yet defined\"\n\tcreate function foo_out(foo) returns cstring as '...';\n\t-- you'll get another NOTICE here about \"foo is only a shell\"\n\tcreate type foo (input = foo_in, output = foo_out, ...);\nAre the notices annoying enough to be a problem? Is there a way around\nthem?\n\nCurrently, most of the PL languages explicitly check for type opaque as a\nfunction argument or result type, and reject it. This should be generalized\nto \"reject any pseudo-type except those explicitly supported\" (which will\nprobably be nothing except VOID).\n\nComments?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 14:08:00 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Proposal: make \"opaque\" obsolete" }, { "msg_contents": "> \tanyarraytype\t-- for array_eq, array_dims\n\nWill this allow generic array iterator functions in the future?\n\n", "msg_date": "20 Aug 2002 14:19:29 -0400", "msg_from": "Rod Taylor <rbt@zort.ca>", "msg_from_op": false, "msg_subject": "Re: Proposal: make \"opaque\" obsolete" }, { "msg_contents": "Rod Taylor <rbt@zort.ca> writes:\n>> anyarraytype\t-- for array_eq, array_dims\n\n> Will this allow generic array iterator functions in the future?\n\nHm. Not directly; there's still the issue of how to tell what element\ntype the array is. array_dims doesn't care, and I think that we have\nsome kluge for array_eq, but in general it'd be a problem for generic\nfunctions.\n\nI had been thinking that as long as we are going to break datafile\ncompatibility (due to Manfred's tuple-header changes) this would be\na good time to try to clean up the representation of arrays. It's\nbothered me for a long while that the array code is not doing alignment\ncorrectly --- it seems not to matter for any standard type, but arrays\nof, say, interval are not aligned the way pg_type says they should be.\n\nThe reason I bring this up is that if we are changing the internal\nrepresentation of arrays, we could add type OID and perhaps typmod to\nthe array header, thus making an array value interpretable without any\noutside info. Then you could actually do something interesting with\na function taking anyarraytype.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 14:31:24 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Re: Proposal: make \"opaque\" obsolete " }, { "msg_contents": "On Tue, 20 Aug 2002, Tom Lane wrote:\n\nIn general I think it sounds good, so I'm only responding to places where\nI want to say something specific.\n\n> I am not by any means wedded to the above type names; does anyone have\n> suggestions for better names? (In particular, I am wondering if \"tuple\"\n> and \"internal\" would be better named \"pg_tuple\" and \"pg_internal\". We\n> might also consider making a type specifically named \"trigger\" and using\n> that to declare trigger functions, rather than \"tuple\".)\n\nI like something with trigger better, makes it very obvious that it's a\ntrigger function.\n\n> An exception is that void_out should succeed and just return an empty string;\n> this allows functions-returning-void to be called by SELECT and behave\n> reasonably. Less obviously, void_in should succeed (and return nothing\n> interesting, probably just a zero datum; it can ignore its input). This\n> allows plpgsql functions to be defined to return VOID.\n\nDoes this require additional work to the plpgsql grammar? The\nnatural way to return from such a function (return;) doesn't seem like\nit'd work without some changes. In any case I don't think this would\nbe necessary for 7.3.\n\n> I am also considering allowing cstring_out to succeed (and, of course,\n> just return a copy of what it's given). That would allow explicit invocation\n> of an output function to work, viz \"SELECT cash_out('42'::money)\" would\n> actually do what the user expects. This is not really necessary though.\n\nI like the idea of cstring_out working, but I wonder if we should stop\nusers from calling the I/O functions directly anyway even if they were\nmade to be safe.\n\n> Should we throw a NOTICE stating that opaque is deprecated if a trigger\n> is declared with opaque? Or should we wait a release or two for that?\n> Similarly, the preferred signature for I/O functions now uses cstring\n> and the function's actual datatype, rather than OPAQUE (and again we\n> could consider throwing a NOTICE).\n\nI think we should throw the notices right away, although this makes me\nwonder in general about upgrade path. Are we ever planning to make that\nan error, and if so, how are we going to handle functions that are coming\nfrom previous versions where it was okay? It's relatively easy to do fix\nones that are currently used as trigger functions or type i/o functions,\nbut what about ones that aren't being used at dump time? Do we even need\nto do anything?\n\n> One of the original concerns about this was how to handle the circularity\n> problem for user-defined I/O functions. If we don't do anything special,\n> then it will still work as long as people define the input function first:\n> \tcreate function foo_in(cstring) returns foo as ' ... ';\n> \t-- you'll get a NOTICE here about \"type foo is not yet defined\"\n> \tcreate function foo_out(foo) returns cstring as '...';\n> \t-- you'll get another NOTICE here about \"foo is only a shell\"\n> \tcreate type foo (input = foo_in, output = foo_out, ...);\n> Are the notices annoying enough to be a problem? Is there a way around\n> them?\n\nI personally don't think it's a big deal, although I'm alot less annoyed\nby notices than alot of people.\n\n\n", "msg_date": "Tue, 20 Aug 2002 12:39:40 -0700 (PDT)", "msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>", "msg_from_op": false, "msg_subject": "Re: Proposal: make \"opaque\" obsolete" }, { "msg_contents": "Stephan Szabo <sszabo@megazone23.bigpanda.com> writes:\n> On Tue, 20 Aug 2002, Tom Lane wrote:\n>> Less obviously, void_in should succeed (and return nothing\n>> interesting, probably just a zero datum; it can ignore its input). This\n>> allows plpgsql functions to be defined to return VOID.\n\n> Does this require additional work to the plpgsql grammar?\n\nI suspect you'd need to say \"return 0\" (or return anything-at-all,\npretty much) to make it fly with the current plpgsql sources. This\nis a tad ugly but I think we can live with it until someone wants to\nfix it. If we have type void then for sure people will want to use\nit for plpgsql functions; there are plenty of cases where you run a\nplpgsql function just for side-effects.\n\n> I think we should throw the notices right away, although this makes me\n> wonder in general about upgrade path. Are we ever planning to make that\n> an error, and if so, how are we going to handle functions that are coming\n> from previous versions where it was okay?\n\nWe can't make it an error until sufficiently far down the road that we\ndon't care about forward compatibility from 7.2-or-before dump files.\nThat'll be a long while, probably.\n\nThrowing a notice right away is okay with me personally, but I wanted to\nsee what other people thought...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 15:46:51 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Re: Proposal: make \"opaque\" obsolete " }, { "msg_contents": "Tom Lane wrote:\n> > I think we should throw the notices right away, although this makes me\n> > wonder in general about upgrade path. Are we ever planning to make that\n> > an error, and if so, how are we going to handle functions that are coming\n> > from previous versions where it was okay?\n> \n> We can't make it an error until sufficiently far down the road that we\n> don't care about forward compatibility from 7.2-or-before dump files.\n> That'll be a long while, probably.\n> \n> Throwing a notice right away is okay with me personally, but I wanted to\n> see what other people thought...\n\nNOTICE seems good. We will have all this mentioned in the release notes\ntoo. In fact, we can point to the release notes in the NOTICE if\ndesired.\n\nI like the array storage format change too.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 20 Aug 2002 16:43:27 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposal: make \"opaque\" obsolete" }, { "msg_contents": "> Trigger functions will now be expected to take no arguments and return\n> either tuple (or trigger if we call it that) or opaque. It would also be\n> sensible to allow VOID in the case of AFTER triggers, but I'm inclined not\n> to do so: I think it's better that a trigger function be declared in a way\n> that makes it clear it's supposed to be a trigger. If CREATE\n> TRIGGER accepts\n> functions returning void then I think you lose some useful error checking.\n>\n> Should we throw a NOTICE stating that opaque is deprecated if a trigger\n> is declared with opaque? Or should we wait a release or two for that?\n\nI think a NOTICE at creation time is fine.\n\n> Comments?\n\nSounds really good.\n\nChris\n\n", "msg_date": "Wed, 21 Aug 2002 09:43:48 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: Proposal: make \"opaque\" obsolete" }, { "msg_contents": "Tom Lane wrote:\n> About I/O behavior: the pg_type entries for these pseudo-types will have to\n> have typinput and typoutput functions. In general these I/O routines must\n> just throw errors. Otherwise you could break the intended type safety by\n> supplying user-written constants. For instance, the present definition of\n> RECORD is wrong (sorry Joe) because it uses oidin and oidout; so I could\n> write \"SELECT foo('42'::record)\" and thereby crash a function expecting\n> RECORD. Not that there will be any such function, but the analogous case\n> with, say, INTERNAL would be bad news.\n\nSorry I've been unable to be very involved today. Anything you want me \nto do here?\n\n\n> An exception is that void_out should succeed and just return an empty string;\n> this allows functions-returning-void to be called by SELECT and behave\n> reasonably. Less obviously, void_in should succeed (and return nothing\n> interesting, probably just a zero datum; it can ignore its input). This\n> allows plpgsql functions to be defined to return VOID.\n\nThis will be useful if/when we want to implement \"CALL stored_proc;\"\n\n\n> Should we throw a NOTICE stating that opaque is deprecated if a trigger\n> is declared with opaque? Or should we wait a release or two for that?\n\nI'd throw the NOTICE now.\n\n\nJoe\n\n", "msg_date": "Tue, 20 Aug 2002 22:30:32 -0700", "msg_from": "Joe Conway <mail@joeconway.com>", "msg_from_op": false, "msg_subject": "Re: Proposal: make \"opaque\" obsolete" }, { "msg_contents": "Tom Lane wrote:\n> The reason I bring this up is that if we are changing the internal\n> representation of arrays, we could add type OID and perhaps typmod to\n> the array header, thus making an array value interpretable without any\n> outside info. Then you could actually do something interesting with\n> a function taking anyarraytype.\n\nThis sounds very cool. I'd vote for that.\n\nJoe\n\n\n", "msg_date": "Tue, 20 Aug 2002 22:32:44 -0700", "msg_from": "Joe Conway <mail@joeconway.com>", "msg_from_op": false, "msg_subject": "Re: Proposal: make \"opaque\" obsolete" }, { "msg_contents": "Joe Conway <mail@joeconway.com> writes:\n> Tom Lane wrote:\n>> ... Then you could actually do something interesting with\n>> a function taking anyarraytype.\n\n> This sounds very cool. I'd vote for that.\n\nUm, am I hearing a volunteer to make it happen? I have other problems\nI need to deal with ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 02:14:48 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Re: Proposal: make \"opaque\" obsolete " }, { "msg_contents": "Tom Lane wrote:\n>>>... Then you could actually do something interesting with\n>>>a function taking anyarraytype.\n> \n>>This sounds very cool. I'd vote for that.\n> \n> Um, am I hearing a volunteer to make it happen? I have other problems\n> I need to deal with ...\n> \n\nHmmm. I guess I should be careful what I wish for -- and plan to take \nsome time-off from my \"day job\" between now and September 1st ;-)\n\nI'll give it a shot, but a crude gameplan/general guidance to get me \nstarted would be much appreciated.\n\nJoe\n\n", "msg_date": "Tue, 20 Aug 2002 23:29:07 -0700", "msg_from": "Joe Conway <mail@joeconway.com>", "msg_from_op": false, "msg_subject": "Re: Proposal: make \"opaque\" obsolete" }, { "msg_contents": "On Wed, 21 Aug 2002, Tom Lane wrote:\n\n> Joe Conway <mail@joeconway.com> writes:\n> > Tom Lane wrote:\n> >> ... Then you could actually do something interesting with\n> >> a function taking anyarraytype.\n> \n> > This sounds very cool. I'd vote for that.\n> \n> Um, am I hearing a volunteer to make it happen? I have other problems\n> I need to deal with ...\n\n\nTom,\n\nI saw something in the other thread suggesting that you might be working on\nthis. Is that so?\n\nIf not I have had a little poke around the cash type but I'm no where near up\nto speed on the internals. Your proposal is that cstring etc. get entries like\nrecord on pg_type? That presumably means we'd need in and out functions defined\nfor these, which in the case of cstring would just be copying the input to\noutput?\n\n(As you can see I may not be the best person to work on this if it is to be\navailable for the beta)\n\n\n-- \nNigel J. Andrews\nDirector\n\n---\nLogictree Systems Limited\nComputer Consultants\n\n", "msg_date": "Wed, 21 Aug 2002 10:22:24 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": false, "msg_subject": "Re: Proposal: make \"opaque\" obsolete" }, { "msg_contents": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> I saw something in the other thread suggesting that you might be working on\n> this. Is that so?\n\nI am working on the main proposal to create pseudo-types. I was hoping\nto offload the bit about changing array representation, though.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 09:47:36 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Re: Proposal: make \"opaque\" obsolete " }, { "msg_contents": "Joe Conway <mail@joeconway.com> writes:\n> I'll give it a shot, but a crude gameplan/general guidance to get me \n> started would be much appreciated.\n\nAFAIK, the only code you should need to touch is in\n\tsrc/include/utils/array.h\n\tsrc/backend/utils/adt/arrayfuncs.c\n\tsrc/backend/utils/adt/arrayutils.c\n\nLook around for other references to ArrayType, but I don't *think* there\nis anything else that processes arrays directly; everything else\nshould be calling construct_array() or deconstruct_array().\n\nAdding an element-type OID field to the array header (ArrayType struct)\nought to be a pretty straightforward exercise. We need to debate\nwhether to store a typmod as well; I'm not sure that the space for it\nwould be justified.\n\nThe other thing that was bothering me was that the code is sloppy about\nalignment: although the overall start of the data area is correctly\nMAXALIGN'd, the offsets of individual data items in the array aren't\nnecessarily made to be multiples of the element data type's alignment\nspec. This would probably result in core dumps for datatypes whose size\nis not a multiple of their alignment requirement. (The only standard\none is INTERVAL. For reasons I've never quite figured out, an array\nof INTERVAL doesn't provoke core dumps; seems like it should, at least\non machines where doubles actually require 8-byte alignment.) There are\na dozen or two places in arrayfuncs.c that understand the alignment\nconventions for array elements, and they all need to be changed. The\natt_align macro in src/include/access/tupmacs.h is probably the thing\nto be using; look at the code in heaptuple.c to see how we position\nfield values inside a tuple, and do likewise.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 10:33:08 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Re: Proposal: make \"opaque\" obsolete " }, { "msg_contents": "Tom Lane wrote:\n> Joe Conway <mail@joeconway.com> writes:\n> \n>>I'll give it a shot, but a crude gameplan/general guidance to get me \n>>started would be much appreciated.\n> \n> \n> AFAIK, the only code you should need to touch is in\n> \tsrc/include/utils/array.h\n> \tsrc/backend/utils/adt/arrayfuncs.c\n> \tsrc/backend/utils/adt/arrayutils.c\n> \n> Look around for other references to ArrayType, but I don't *think* there\n> is anything else that processes arrays directly; everything else\n> should be calling construct_array() or deconstruct_array().\n\nOK. I should have a bit more time today and tonight. I'll see what I can \nget done with this.\n\nThanks,\n\nJoe\n\n\n", "msg_date": "Wed, 21 Aug 2002 08:24:29 -0700", "msg_from": "Joe Conway <mail@joeconway.com>", "msg_from_op": false, "msg_subject": "Re: Proposal: make \"opaque\" obsolete" } ]
[ { "msg_contents": "\n> \"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> > Would it be possible to update the system tables, so that cash_out does not take\n> > opaque but really takes type money ?\n> \n> That is part of the solution, but only part: we have hundreds of\n> functions that take \"opaque\" because we don't currently have any way\n> to declare what they really take.\n\nSo the idea is, that input functions take cstring type, and output functions \ntake the explicit type they are designed for ?\nIs there anything that can be done vs that types only exist after the functions ?\ne.g. create it in one tx with constraints deferred ? Or is that no issue ?\n\n> (In particular, all the typinput\n> functions are like that --- so fixing typoutput functions isn't plugging\n> even half of the gap.)\n> \n> See my proposal to make \"opaque\" obsolete.\n\nYes, it is great that you will look into this !!\nAbout the names, would it be good to use SQL99 reserved words ? (e.g. ROW for tuple) \nnice url: http://developer.mimer.se/validator/sql-reserved-words.tml\n\ncount(*) \t\t--> anynumeric :-) (two flies with one strike)\nNULL/NONNULL\t--> null|nullvalue|anynull ? We only need this internally, no ?\n\nHard to say what is good for those names imho, don't like \"anytype\" :-(\n(maybe even leave that opaque for now)\n\nI like \"cstring\", \"void\" and \"internal\". \nMaybe \"anyarray\" instead of \"anyarraytype\".\nAnd I would prefer \"row\" instead of \"tuple\".\n\nAndreas\n", "msg_date": "Tue, 20 Aug 2002 21:43:11 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> Hard to say what is good for those names imho, don't like \"anytype\" :-(\n\nHow about \"any\"? It's a reserved word per SQL99, I think.\n\n> I like \"cstring\", \"void\" and \"internal\". \n\nOkay.\n\n> Maybe \"anyarray\" instead of \"anyarraytype\".\n\nThat would match with \"any\".\n\n> And I would prefer \"row\" instead of \"tuple\".\n\nI'm leaning towards agreeing with Stephan: we should use typename\n\"trigger\" to declare triggers. \"Tuple\" (or \"row\") is strictly correct\nonly for BEFORE triggers, not AFTER triggers, so it's a bit of a\nmisnomer for triggers anyhow.\n\nI'm now also toying with inventing a pseudotype just for procedural\nlanguage handlers, which are currently \"foo() returns opaque\". If we\nwant the type system to catch misuses of trigger functions, we should\nwant it for handlers too. Maybe name this type \"language_handler\"?\n(I had thought we could declare handlers to return \"internal\", but we\ncan't do that without breaking type safety. We don't want *any* way\nfor an SQL construct to look like it returns type \"internal\".)\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 15:58:03 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " } ]
[ { "msg_contents": "SRA/Japan, my employer and a big PostgreSQL support company in Japan,\nhas announced they have purchased TurboLinux:\n\n\thttp://www.turbolinux.com/news/pr/020820.html\n\nHere is a summary from someone on Slashdot:\n\n * Turbolinux Inc. sold Turbolinux Japan K.K. (its Japanese\nsubsidary) to SRA. (This is the $1 mil. transaction according to\nSlashdot Japan [slashdot.jp] (in Japanese))\n\n * Turbolinux Japan K.K. will become the new Turbolinux Inc.\n\n * Turbolinux Inc. also sold all its Linux distribution business,\nlogo, trademarks to SRA, but the price is not yet disclosed.\n\n * SRA is also planning to buy the Chinese and Korean joint ventures\nbetween Turbolinux Inc. and local companies.\n\n * The old Turbolinux Inc. will change its name to CenterRex and\nfocus on software it developed like PowerCockpit or EnFusion.\n\nNot sure how this effects PostgreSQL usage, but it certainly could.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 20 Aug 2002 15:50:21 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "SRA purchases TurboLinux" } ]
[ { "msg_contents": "\nHere's yet another. He claims malicious code can be run on the server\nwith this one.\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n---------- Forwarded message ----------\nDate: Tue, 20 Aug 2002 14:28:49 +0000\nFrom: Sir Mordred The Traitor <mordred@s-mail.com>\nTo: bugtraq@securityfocus.com\nSubject: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL\n\n//@(#)Mordred Labs advisory 0x0003\n\nRelease data: 20/08/02\nName: Buffer overflow in PostgreSQL\nVersions affected: all versions\nRisk: high\n\n--[ Description:\n\n...PostgreSQL is a sophisticated Object-Relational DBMS,\nsupporting almost all SQL constructs, including subselects,\ntransactions, and user-defined types and functions. It is the\nmost advanced open-source database available anywhere...blah...blah...\nFor more info check out this link:\nhttp://www.postgresql.org/idocs/index.php?preface.html#INTRO-WHATIS\n\nThere exists a heap buffer overflow in a repeat(text, integer) function,\nwhich\nallows an attacker to execute malicious code.\n\n--[ Details:\n\nUpon invoking a repeat() function, a\nsrc/backend/utils/adt/oracle_compat.c::repeat() function\nwill gets called which suffers from a buffer overflow.\n\n--[ How to reproduce:\n\npsql> select repeat('xxx',1431655765);\npqReadData() -- backend closed the channel unexpectedly.\n This probably means the backend terminated abnormally\n before or while processing the request.\nThe connection to the server was lost. Attempting reset: Failed.\n\n--[ Solution\n\nDo you still running postgresql? ...Can't believe that...\nIf so, execute the following command as a root: \"killall -9 postmaster\",\nand wait until the patch will be available.\n\n\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n\n", "msg_date": "Tue, 20 Aug 2002 15:52:52 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": true, "msg_subject": "@(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL (fwd)" }, { "msg_contents": "Vince Vielhaber <vev@michvhf.com> writes:\n> Here's yet another. He claims malicious code can be run on the server\n> with this one.\n\nregression=# select repeat('xxx',1431655765);\nserver closed the connection unexpectedly\n\nThis is probably caused by integer overflow in calculating the size of\nthe repeat's result buffer. It'd take some considerable doing to create\nan arbitrary-code exploit, but perhaps could be done. Anyone want to\ninvestigate a patch? I think we likely need something like\n\n\tbufsize = input_length * repeats;\n+\t/* check for overflow in multiplication */\n+\tif (bufsize / repeats != input_length)\n+\t\telog(ERROR, \"repeat result too long\");\n\nbut I haven't thought it through in detail.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 16:05:11 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL\n\t(fwd)" }, { "msg_contents": "On Tue, 20 Aug 2002, Tom Lane wrote:\n\n> Vince Vielhaber <vev@michvhf.com> writes:\n> > Here's yet another. He claims malicious code can be run on the server\n> > with this one.\n>\n> regression=# select repeat('xxx',1431655765);\n> server closed the connection unexpectedly\n>\n> This is probably caused by integer overflow in calculating the size of\n> the repeat's result buffer. It'd take some considerable doing to create\n> an arbitrary-code exploit, but perhaps could be done. Anyone want to\n> investigate a patch? I think we likely need something like\n>\n> \tbufsize = input_length * repeats;\n> +\t/* check for overflow in multiplication */\n> +\tif (bufsize / repeats != input_length)\n> +\t\telog(ERROR, \"repeat result too long\");\n>\n> but I haven't thought it through in detail.\n\nWhere do we check that this:\n\n\t\tresult = (text *) palloc(tlen);\n\nis even successful? Is it in one of the macros (VARATT_SIZEP or VARDATA)?\nIt appears like it goes into the count and copy regardless.\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n", "msg_date": "Tue, 20 Aug 2002 16:11:26 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": true, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "Vince Vielhaber <vev@michvhf.com> writes:\n> Where do we check that this:\n> \t\tresult = (text *) palloc(tlen);\n> is even successful?\n\npalloc elogs if it can't allocate the space; it's unlike malloc in that\nrespect. I believe it also has a guard to reject requests > 1Gb, so\nI think it's reasonably proof against internal arithmetic overflows.\n\nThis problem is strictly repeat's error, not palloc's.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 16:15:01 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL\n\t(fwd)" }, { "msg_contents": "Tom Lane <tgl@sss.pgh.pa.us> writes:\n\n> Vince Vielhaber <vev@michvhf.com> writes:\n> > Here's yet another. He claims malicious code can be run on the server\n> > with this one.\n> \n> regression=# select repeat('xxx',1431655765);\n> server closed the connection unexpectedly\n> \n> This is probably caused by integer overflow in calculating the size of\n> the repeat's result buffer. It'd take some considerable doing to create\n> an arbitrary-code exploit, but perhaps could be done. Anyone want to\n> investigate a patch?\n\nThis seems to fix the problem:\n\nnconway=# select repeat('xxx',1431655765);\nERROR: Requested buffer is too large.\n\nIt uses the logic you suggested. Just a quick and dirty fix, I may\nhave missed something... The patch applies cleanly to both CVS HEAD\nand the 7.2 stable branch.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "20 Aug 2002 16:31:50 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL\n\t(fwd)" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> + \t/* Check for integer overflow */\n> + \tif (tlen / slen != count)\n> + \t\telog(ERROR, \"Requested buffer is too large.\");\n\nWhat about slen == 0?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 16:36:09 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL\n\t(fwd)" }, { "msg_contents": "Vince Vielhaber <vev@michvhf.com> writes:\n> Here's yet another.\n\nShould someone from the core team perhaps get in contact with this guy\nand ask if he could get in contact with the development team before\npublicizing any further security holes? AFAIK that is standard\noperating procedure in most cases...\n\nSecond, it might be worth pushing a 7.2.2 release containing the fix\nfor this bug, as well as the datetime problem. If that sounds\nreasonable to the people who have to do the most work on a new release\n(e.g. Marc), I can volunteer to backport a fix for the datetime\nproblem.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "20 Aug 2002 16:43:31 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL\n\t(fwd)" }, { "msg_contents": "On 20 Aug 2002, Neil Conway wrote:\n\n> Vince Vielhaber <vev@michvhf.com> writes:\n> > Here's yet another.\n>\n> Should someone from the core team perhaps get in contact with this guy\n> and ask if he could get in contact with the development team before\n> publicizing any further security holes? AFAIK that is standard\n> operating procedure in most cases...\n\nI just asked him/her that on bugtraq.\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n", "msg_date": "Tue, 20 Aug 2002 16:50:45 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": true, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "Vince Vielhaber wrote:\n> \n> Here's yet another. He claims malicious code can be run on the server\n> with this one.\n\n> --[ Solution\n> \n> Do you still running postgresql? ...Can't believe that...\n> If so, execute the following command as a root: \"killall -9 postmaster\",\n> and wait until the patch will be available.\n\nYou've got to love this last part. ;-)\n\nThis has the attitude of \"Oh, look at me, I found a bug. Let me\npublicize it and get more attention. Aren't you proud of me?\"\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 20 Aug 2002 16:50:46 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL" }, { "msg_contents": "Tom Lane <tgl@sss.pgh.pa.us> writes:\n> Neil Conway <neilc@samurai.com> writes:\n> > + \t/* Check for integer overflow */\n> > + \tif (tlen / slen != count)\n> > + \t\telog(ERROR, \"Requested buffer is too large.\");\n> \n> What about slen == 0?\n\nGood point -- that wouldn't cause incorrect results or a security\nproblem, but it would reject input that we should really accept.\n\nRevised patch is attached.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "20 Aug 2002 16:54:23 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL\n\t(fwd)" }, { "msg_contents": "\nThis is all an indication of our increasing usage. Several PostgreSQL\narticles have appeared in the past week in _major_ media outlets, not\njust the open-source press (eweek, Bloomburg), a major PostgreSQL\nsupport company bought a Linux distribution (SRA-Turbolinux), and we\nhave independent projects auditing our code. It is only going to get\nworse. ;-)\n\n---------------------------------------------------------------------------\n\nVince Vielhaber wrote:\n> On 20 Aug 2002, Neil Conway wrote:\n> \n> > Vince Vielhaber <vev@michvhf.com> writes:\n> > > Here's yet another.\n> >\n> > Should someone from the core team perhaps get in contact with this guy\n> > and ask if he could get in contact with the development team before\n> > publicizing any further security holes? AFAIK that is standard\n> > operating procedure in most cases...\n> \n> I just asked him/her that on bugtraq.\n> \n> Vince.\n> -- \n> ==========================================================================\n> Vince Vielhaber -- KA8CSH email: vev@michvhf.com http://www.pop4.net\n> 56K Nationwide Dialup from $16.00/mo at Pop4 Networking\n> http://www.camping-usa.com http://www.cloudninegifts.com\n> http://www.meanstreamradio.com http://www.unknown-artists.com\n> ==========================================================================\n> \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> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 20 Aug 2002 17:09:43 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "> Should someone from the core team perhaps get in contact with this guy\n> and ask if he could get in contact with the development team before\n> publicizing any further security holes? AFAIK that is standard\n> operating procedure in most cases...\n>\n> Second, it might be worth pushing a 7.2.2 release containing the fix\n> for this bug, as well as the datetime problem. If that sounds\n> reasonable to the people who have to do the most work on a new release\n> (e.g. Marc), I can volunteer to backport a fix for the datetime\n> problem.\n\nIt'd be better to contact the dude and get all his bugs out of him, fix them\nall and issue a 7.2.2 with all the fixes.\n\nI think this is now essential - people will be using 7.2 series for ages\neven after the 7.3 release...\n\nChris\n\n", "msg_date": "Wed, 21 Aug 2002 09:49:57 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL\n\t(fwd)" }, { "msg_contents": "On Wed, 21 Aug 2002, Christopher Kings-Lynne wrote:\n\n> > Should someone from the core team perhaps get in contact with this guy\n> > and ask if he could get in contact with the development team before\n> > publicizing any further security holes? AFAIK that is standard\n> > operating procedure in most cases...\n> >\n> > Second, it might be worth pushing a 7.2.2 release containing the fix\n> > for this bug, as well as the datetime problem. If that sounds\n> > reasonable to the people who have to do the most work on a new release\n> > (e.g. Marc), I can volunteer to backport a fix for the datetime\n> > problem.\n> \n> It'd be better to contact the dude and get all his bugs out of him, fix them\n> all and issue a 7.2.2 with all the fixes.\n\nThat wouldn't work because it seems he is making advisories at the\ntime he discovers a bug/flaw. That is, he is not directly interested in\nthe robustness of Postgres -- rather, another poster put it, his\nreputation on bugtraq. That's fine but it doesn't mesh well with the\nco-ordinated effort you describe.\n\nI still do not see this as being a serious security problem unless you are\nproviding access to postgres to untrusted users. The advisory's\nrecommendation of killing the postmaster as a solution to these bugs is\nakin to saying 'kill ssh if there's a libc bug'. If you are providing\naccess to untrusted users, that advice is worthwhile. But if your users\nare trusted and could produce the same effect in any other number of\nreasons, the advice is useless.\n\nAs for making a 7.2.2 release for the sake of form and for those users who\ndo provide access to untrusted users (universities, ISPs, say) this would\nbe pointless without the changes to opaque which Tom has put forward and\nmay/may not work on before 7.3 beta. I'm not sure that the core guys would\nbe too happy doing that *and* requiring an initdb for a minor release (as \nI presume Tom's changes will require one).\n\nGavin\n\n", "msg_date": "Wed, 21 Aug 2002 12:15:51 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "Gavin Sherry <swm@linuxworld.com.au> writes:\n> As for making a 7.2.2 release for the sake of form and for those\n> users who do provide access to untrusted users (universities, ISPs,\n> say) this would be pointless without the changes to opaque which Tom\n> has put forward and may/may not work on before 7.3 beta.\n\nI don't think that releasing 7.2.2 without the opaque changes would be\npointless. For one thing, the opaque-related security problems are\ncomparatively minor: the cracker must be able to execute arbitrary SQL\ncommands against the database, and by that stage of the game, a DoS\nattack is already trivial (e.g. disable GEQO and execute a 15 table\njoin query).\n\nAlso, from skimming the discussion on fixing the opaque problems,\nthere will be at least some degree of backwards incompatibility. That\nis definitely undesirable for a stable point release -- as is an\ninitdb, as you point out. This amount of pain to fix a minor security\nhole is *not* worth it, IMHO.\n\nSo I think that fixing the opaque problems in 7.2.x is simply\nimpossible. Given that, the question is whether we should make a 7.2.2\nrelease with fixes for the other security holes (lpad(), rpad(),\nreverse(), and the datetime overruns). IMHO, we should.\n\nThe datetime hole is fairly serious: it's not unreasonable for\ndevelopers to accept datetime input from the user without limiting\nit's length. So it's quite likely that there are 7.2 systems in\nproduction that have a sane security policy (e.g. hidden behind a\nfirewall, validate user input, etc.) that are still vulnerable.\n\nThe alternative seems unattractive: if we require that users wait for\n7.3 to come out, it may be months before the 7.3.0 release. And even\nthen, upgrading to 7.3 is non-trivial: it requires an initdb and\nreload, as well as testing to ensure that the user's applications run\nsmoothly on 7.3. Therefore, it may be several months before many\nproduction sites upgrade to 7.3; leaving them in the cold for that\nperiod isn't something I think we should do, if we can avoid it.\n\nThat said, there's only a limited amount that I can do. I think we\nshould make a 7.2.2 release, and to that end I've posted patches\nagainst REL7_2_STABLE for all four of the security holes. The rest of\nthe work that goes into making a release needs to be done by others\n(Marc, Vince, Bruce, etc.) -- if there's anything I can do to help,\nlet me know.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "20 Aug 2002 23:34:53 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "...\n> So I think that fixing the opaque problems in 7.2.x is simply\n> impossible. Given that, the question is whether we should make a 7.2.2\n> release with fixes for the other security holes (lpad(), rpad(),\n> reverse(), and the datetime overruns). IMHO, we should.\n\nJust a minor point: can someone actually show a symptom with date/time\nproblems in 7.2.x?\n\n - Thomas\n", "msg_date": "Tue, 20 Aug 2002 21:09:58 -0700", "msg_from": "Thomas Lockhart <lockhart@fourpalms.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On Tue, 20 Aug 2002, Thomas Lockhart wrote:\n\n> ...\n> > So I think that fixing the opaque problems in 7.2.x is simply\n> > impossible. Given that, the question is whether we should make a 7.2.2\n> > release with fixes for the other security holes (lpad(), rpad(),\n> > reverse(), and the datetime overruns). IMHO, we should.\n> \n> Just a minor point: can someone actually show a symptom with date/time\n> problems in 7.2.x?\n\ntemplate1=# select version();\n version\n-------------------------------------------------------------\n PostgreSQL 7.2.1 on i686-pc-linux-gnu, compiled by GCC 2.96\n(1 row)\n\ntemplate1=# select\n'1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111'::date\nserver closed the connection unexpectedly\n This probably means the server terminated abnormally\n before or while processing the request.\nThe connection to the server was lost. Attempting reset: Failed.\n!#\n\nParseDateTime() isn't checking that str < MAXDATELEN -- which is the\nproblem you solved in the datetime.c fixes.\n\nGavin\n\n", "msg_date": "Wed, 21 Aug 2002 16:13:22 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On 20 Aug 2002, Neil Conway wrote:\n\n> Vince Vielhaber <vev@michvhf.com> writes:\n> > Here's yet another.\n>\n> Should someone from the core team perhaps get in contact with this guy\n> and ask if he could get in contact with the development team before\n> publicizing any further security holes? AFAIK that is standard\n> operating procedure in most cases...\n>\n> Second, it might be worth pushing a 7.2.2 release containing the fix\n> for this bug, as well as the datetime problem. If that sounds\n> reasonable to the people who have to do the most work on a new release\n> (e.g. Marc), I can volunteer to backport a fix for the datetime\n> problem.\n\nI have no problems doing a v7.2.2 release for something like this ...\n\n\n", "msg_date": "Wed, 21 Aug 2002 03:41:32 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On Tue, 2002-08-20 at 23:34, Neil Conway wrote:\n> Gavin Sherry <swm@linuxworld.com.au> writes:\n> > As for making a 7.2.2 release for the sake of form and for those\n> > users who do provide access to untrusted users (universities, ISPs,\n> > say) this would be pointless without the changes to opaque which Tom\n> > has put forward and may/may not work on before 7.3 beta.\n> \n> I don't think that releasing 7.2.2 without the opaque changes would be\n> pointless. For one thing, the opaque-related security problems are\n> comparatively minor: the cracker must be able to execute arbitrary SQL\n> commands against the database, and by that stage of the game, a DoS\n> attack is already trivial (e.g. disable GEQO and execute a 15 table\n> join query).\n> \n> Also, from skimming the discussion on fixing the opaque problems,\n> there will be at least some degree of backwards incompatibility. That\n> is definitely undesirable for a stable point release -- as is an\n> initdb, as you point out. This amount of pain to fix a minor security\n> hole is *not* worth it, IMHO.\n> \n> So I think that fixing the opaque problems in 7.2.x is simply\n> impossible. Given that, the question is whether we should make a 7.2.2\n> release with fixes for the other security holes (lpad(), rpad(),\n> reverse(), and the datetime overruns). IMHO, we should.\n> \n> The datetime hole is fairly serious: it's not unreasonable for\n> developers to accept datetime input from the user without limiting\n> it's length. So it's quite likely that there are 7.2 systems in\n> production that have a sane security policy (e.g. hidden behind a\n> firewall, validate user input, etc.) that are still vulnerable.\n> \n> The alternative seems unattractive: if we require that users wait for\n> 7.3 to come out, it may be months before the 7.3.0 release. And even\n> then, upgrading to 7.3 is non-trivial: it requires an initdb and\n> reload, as well as testing to ensure that the user's applications run\n> smoothly on 7.3. Therefore, it may be several months before many\n> production sites upgrade to 7.3; leaving them in the cold for that\n> period isn't something I think we should do, if we can avoid it.\n> \n> That said, there's only a limited amount that I can do. I think we\n> should make a 7.2.2 release, and to that end I've posted patches\n> against REL7_2_STABLE for all four of the security holes. The rest of\n> the work that goes into making a release needs to be done by others\n> (Marc, Vince, Bruce, etc.) -- if there's anything I can do to help,\n> let me know.\n> \n> Cheers,\n> \n> Neil\n> \n\nAssuming that we do go ahead with a 7.2.2 release, can we get some kind\nof unofficial statement on pushing back the 7.3 beta? I know Tom was\nhoping to start it by sept 1 but that seems rushed to me. Furthermore,\nbetween schema support and now more backward incompatibility in regards\nto functions/opaque, should we open some discussion on 7.3 really being\n8.0? \n\nRobert Treat\n\n\n\n", "msg_date": "21 Aug 2002 11:43:20 -0400", "msg_from": "Robert Treat <xzilla@users.sourceforge.net>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "Robert Treat wrote:\n<snip>\n> \n> Assuming that we do go ahead with a 7.2.2 release, can we get some kind\n> of unofficial statement on pushing back the 7.3 beta? I know Tom was\n> hoping to start it by sept 1 but that seems rushed to me. Furthermore,\n> between schema support and now more backward incompatibility in regards\n> to functions/opaque, should we open some discussion on 7.3 really being\n> 8.0?\n\nDepending on how far back we push this, we might be able to get Windows\nNative support added. That'd be cool and probably contribute to an 8.0.\n\n:)\n\nRegards and best wishes,\n\nJustin Clift\n\n \n> Robert Treat\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\"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": "Thu, 22 Aug 2002 02:13:08 +1000", "msg_from": "Justin Clift <justin@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "\nWe learned a few lessons from previous releases. First, don't delay\nthe beta by days/weeks that drag on. Delay one month at a time. \nSecond, don't decide on a further delay the day before you are going to\ngo beta. Multiple short-period delays and delays that happen at the\nlast minute cause too many stops/starts for developers to be effective,\nso...\n\nIf we are going to delay beta, we should decide now, not at the end of\nAugust, and the delay should be until the end of September. The big\nquestion is whether we have enough material to warrant a delay.\n\n---------------------------------------------------------------------------\n\nJustin Clift wrote:\n> Robert Treat wrote:\n> <snip>\n> > \n> > Assuming that we do go ahead with a 7.2.2 release, can we get some kind\n> > of unofficial statement on pushing back the 7.3 beta? I know Tom was\n> > hoping to start it by sept 1 but that seems rushed to me. Furthermore,\n> > between schema support and now more backward incompatibility in regards\n> > to functions/opaque, should we open some discussion on 7.3 really being\n> > 8.0?\n> \n> Depending on how far back we push this, we might be able to get Windows\n> Native support added. That'd be cool and probably contribute to an 8.0.\n> \n> :)\n> \n> Regards and best wishes,\n> \n> Justin Clift\n> \n> \n> > Robert Treat\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> \"My grandfather once told me that there are two kinds of people: those\n> who work and those who take the credit. He told me to try to be in the\n> first group; there was less competition there.\"\n> - Indira Gandhi\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 12:25:07 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "Bruce Momjian wrote:\n> \n> We learned a few lessons from previous releases. First, don't delay\n> the beta by days/weeks that drag on. Delay one month at a time.\n> Second, don't decide on a further delay the day before you are going to\n> go beta. Multiple short-period delays and delays that happen at the\n> last minute cause too many stops/starts for developers to be effective,\n> so...\n> \n> If we are going to delay beta, we should decide now, not at the end of\n> August, and the delay should be until the end of September. The big\n> question is whether we have enough material to warrant a delay.\n\nOnly two things which have the potential to be worth waiting for, from\nwhat I'm aware of. There may be others:\n\n - Find out from Sir Mordred if he wants to take a look at the CVS\n version of code and audit in that for a bit, Just In Case he turns\n up something that's serious and requires substantial re-work.\n Although it means he wouldn't have a bunch of \"I found this existing\n exploit\" type releases, we could instead offer him credit on the\n press release along the lines of \"This released has been audited for\n security flaws in its code by Sir Mordred\". Am pretty sure he'd\n do a very thorough job for that, as it means he'd have an official\n \"product reputation\" he'd need to stand by for it.\n\n - Patches to the CVS tree which let us have a truly native windows\n version. This is of huge significance and would *very* much improve\n our growth and adoption by being in this release in comparison to\n being in the release afterwards. Not in an airy fairy way, but\n quite definitely and solidly.\n\nOf the two, Sir Mordred may or may not be willing, so that's kind of\niffy, whereas the Windows Native port which is in beta testing isn't\nin too bad a state at all already. Have been running preliminary\nmulti-user AS3AP tests on it (with OSDB) and getting a significant\nperformance throughput increase in comparison to the cygwin version.\n\n:)\n\nHope I'm not pushing too strongly for this, as, after all, I can't do\nthe coding needed here. :(\n\nRegards and best wishes,\n\nJustin Clift\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": "Thu, 22 Aug 2002 02:42:38 +1000", "msg_from": "Justin Clift <justin@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "Justin Clift wrote:\n> Only two things which have the potential to be worth waiting for, from\n> what I'm aware of. There may be others:\n> \n> - Find out from Sir Mordred if he wants to take a look at the CVS\n> version of code and audit in that for a bit, Just In Case he turns\n> up something that's serious and requires substantial re-work.\n> Although it means he wouldn't have a bunch of \"I found this existing\n> exploit\" type releases, we could instead offer him credit on the\n> press release along the lines of \"This released has been audited for\n> security flaws in its code by Sir Mordred\". Am pretty sure he'd\n> do a very thorough job for that, as it means he'd have an official\n> \"product reputation\" he'd need to stand by for it.\n\nThis is interesting. He would have a month to do it.\n\n> - Patches to the CVS tree which let us have a truly native windows\n> version. This is of huge significance and would *very* much improve\n> our growth and adoption by being in this release in comparison to\n> being in the release afterwards. Not in an airy fairy way, but\n> quite definitely and solidly.\n> \n> Of the two, Sir Mordred may or may not be willing, so that's kind of\n> iffy, whereas the Windows Native port which is in beta testing isn't\n> in too bad a state at all already. Have been running preliminary\n> multi-user AS3AP tests on it (with OSDB) and getting a significant\n> performance throughput increase in comparison to the cygwin version.\n\nOK, now I have to ask, where did this native Windows version come from? \nI don't know anything about it, except that Jan and SRA are both working\non versions.\n\nThe other issue is PITR, which I have been told today will not be ready\nfor a September 1 beta but may be ready for an October 1 beta.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 12:59:50 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "Bruce Momjian wrote:\n> \n> Justin Clift wrote:\n> > Only two things which have the potential to be worth waiting for, from\n> > what I'm aware of. There may be others:\n> >\n> > - Find out from Sir Mordred if he wants to take a look at the CVS\n> > version of code and audit in that for a bit, Just In Case he turns\n> > up something that's serious and requires substantial re-work.\n> > Although it means he wouldn't have a bunch of \"I found this existing\n> > exploit\" type releases, we could instead offer him credit on the\n> > press release along the lines of \"This released has been audited for\n> > security flaws in its code by Sir Mordred\". Am pretty sure he'd\n> > do a very thorough job for that, as it means he'd have an official\n> > \"product reputation\" he'd need to stand by for it.\n> \n> This is interesting. He would have a month to do it.\n\nReckon it's worth asking him, to find out if he'd be interested in this?\n \n> > - Patches to the CVS tree which let us have a truly native windows\n> > version. This is of huge significance and would *very* much improve\n> > our growth and adoption by being in this release in comparison to\n> > being in the release afterwards. Not in an airy fairy way, but\n> > quite definitely and solidly.\n> >\n> > Of the two, Sir Mordred may or may not be willing, so that's kind of\n> > iffy, whereas the Windows Native port which is in beta testing isn't\n> > in too bad a state at all already. Have been running preliminary\n> > multi-user AS3AP tests on it (with OSDB) and getting a significant\n> > performance throughput increase in comparison to the cygwin version.\n> \n> OK, now I have to ask, where did this native Windows version come from?\n> I don't know anything about it, except that Jan and SRA are both working\n> on versions.\n\nIt was kind of quietly let slip out:\n\nhttp://archives.postgresql.org/pgsql-cygwin/2002-08/msg00012.php\n\nBut, it's definitely up and running and functional and pretty decent.\n\n:-)\n\n> The other issue is PITR, which I have been told today will not be ready\n> for a September 1 beta but may be ready for an October 1 beta.\n\nUseful, but not sure it's worth delaying even *further* for.\n\n+ Justin\n\n> --\n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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": "Thu, 22 Aug 2002 03:04:56 +1000", "msg_from": "Justin Clift <justin@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "Justin Clift wrote:\n> Bruce Momjian wrote:\n> > \n> > Justin Clift wrote:\n> > > Only two things which have the potential to be worth waiting for, from\n> > > what I'm aware of. There may be others:\n> > >\n> > > - Find out from Sir Mordred if he wants to take a look at the CVS\n> > > version of code and audit in that for a bit, Just In Case he turns\n> > > up something that's serious and requires substantial re-work.\n> > > Although it means he wouldn't have a bunch of \"I found this existing\n> > > exploit\" type releases, we could instead offer him credit on the\n> > > press release along the lines of \"This released has been audited for\n> > > security flaws in its code by Sir Mordred\". Am pretty sure he'd\n> > > do a very thorough job for that, as it means he'd have an official\n> > > \"product reputation\" he'd need to stand by for it.\n> > \n> > This is interesting. He would have a month to do it.\n> \n> Reckon it's worth asking him, to find out if he'd be interested in this?\n\n\nI wouldn't do it yet until we know if we are going to delay.\n\n> > > - Patches to the CVS tree which let us have a truly native windows\n> > > version. This is of huge significance and would *very* much improve\n> > > our growth and adoption by being in this release in comparison to\n> > > being in the release afterwards. Not in an airy fairy way, but\n> > > quite definitely and solidly.\n> > >\n> > > Of the two, Sir Mordred may or may not be willing, so that's kind of\n> > > iffy, whereas the Windows Native port which is in beta testing isn't\n> > > in too bad a state at all already. Have been running preliminary\n> > > multi-user AS3AP tests on it (with OSDB) and getting a significant\n> > > performance throughput increase in comparison to the cygwin version.\n> > \n> > OK, now I have to ask, where did this native Windows version come from?\n> > I don't know anything about it, except that Jan and SRA are both working\n> > on versions.\n> \n> It was kind of quietly let slip out:\n> \n> http://archives.postgresql.org/pgsql-cygwin/2002-08/msg00012.php\n> \n> But, it's definitely up and running and functional and pretty decent.\n\nOh, so it is Jan's group. Great news; wish someone would have told me\nsooner. I removed the Win32 off the open items list because, with no\ninfo and no one commenting on the item, it seemed dead for 7.3.\n\n> > The other issue is PITR, which I have been told today will not be ready\n> > for a September 1 beta but may be ready for an October 1 beta.\n> \n> Useful, but not sure it's worth delaying even *further* for.\n\nWell, PITR is a much more desired feature even than Win32; the big\nquestion is how long PITR will actually take, seeing as we haven't see\nany patches yet.\n\nHowever, we haven't seen any Win32 patches yet either, so they are in\nthe same boat as far as I am concerned.\n\nWe have an open items list that is pretty much ready for 7.3. The only\nopen items of significance left is whether schema/DROP COLUMN stuff is\nready in all the interfaces/apps.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n\n P O S T G R E S Q L\n\n 7 . 3 O P E N I T E M S\n\n\nCurrent at ftp://candle.pha.pa.us/pub/postgresql/open_items.\n\nSource Code Changes\n-------------------\nSchema handling - ready? interfaces? client apps?\nDrop column handling - ready for all clients, apps?\nhave pg_dumpall dump out db privilege and per-user/db settings\nfix implicit type coercions that are worse\nPrepared statements - to be reviewed (Tom)\nimprove macros in new tuple header code (Tom)\nintegrate or move to gborg libpqxx, Pg:DBD\nAllow PL/PgSQL functions to return sets (Neil)\nAllow easy display of usernames in a group (pg_hba.conf uses groups now)\nfix BeOS and QNX4 ports\n\nOn Hold\n-------\nPoint-in-time recovery - status? (J.R., Richard)\nWin32 port\nSecurity audit\n\nDocumentation Changes\n---------------------\nMention foreign keys and SERIAL dependencies will not be in 7.2 loaded tables\nDocument need to add permissions to loaded functions and languages", "msg_date": "Wed, 21 Aug 2002 13:13:27 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://candle.pha.pa.us/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nNeil Conway wrote:\n> Tom Lane <tgl@sss.pgh.pa.us> writes:\n> > Neil Conway <neilc@samurai.com> writes:\n> > > + \t/* Check for integer overflow */\n> > > + \tif (tlen / slen != count)\n> > > + \t\telog(ERROR, \"Requested buffer is too large.\");\n> > \n> > What about slen == 0?\n> \n> Good point -- that wouldn't cause incorrect results or a security\n> problem, but it would reject input that we should really accept.\n> \n> Revised patch is attached.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 13:22:13 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL" }, { "msg_contents": "On Wed, 2002-08-21 at 13:13, Bruce Momjian wrote:\n> Justin Clift wrote:\n> > Bruce Momjian wrote:\n> > > \n> > > Justin Clift wrote:\n> > > > Only two things which have the potential to be worth waiting for, from\n> > > > what I'm aware of. There may be others:\n> > > >\n> > > > - Find out from Sir Mordred if he wants to take a look at the CVS\n> > > > version of code and audit in that for a bit, Just In Case he turns\n> > > > up something that's serious and requires substantial re-work.\n> > > > Although it means he wouldn't have a bunch of \"I found this existing\n> > > > exploit\" type releases, we could instead offer him credit on the\n> > > > press release along the lines of \"This released has been audited for\n> > > > security flaws in its code by Sir Mordred\". Am pretty sure he'd\n> > > > do a very thorough job for that, as it means he'd have an official\n> > > > \"product reputation\" he'd need to stand by for it.\n> > > \n> > > This is interesting. He would have a month to do it.\n> > \n> > Reckon it's worth asking him, to find out if he'd be interested in this?\n> \n> \n> I wouldn't do it yet until we know if we are going to delay.\n\nI'd ask anyway. 99% of the issues he finds will be fairly localized. \nAnything truly new (not on TODO already) will probably require a fair\nbit of time to track down, then fix time on top (2 months delay?).\n\nAnyway, these types of discoveries are better in beta than after the\nrelease and would still warrent a mention if there is a fair amount of\nground covered.\n\n\nPersonally, I'd be more interested in whats safe (covered) than whats\nbroken. Posting the successful test cases as some proof rowards\nstability / security of the new release would realize immediate gains in\nsettling nervous VPs about the new installation.\n\n\n", "msg_date": "21 Aug 2002 13:30:04 -0400", "msg_from": "Rod Taylor <rbt@zort.ca>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "\nGood point; please ask him. We have at least on month in beta.\n\n---------------------------------------------------------------------------\n\nRod Taylor wrote:\n> On Wed, 2002-08-21 at 13:13, Bruce Momjian wrote:\n> > Justin Clift wrote:\n> > > Bruce Momjian wrote:\n> > > > \n> > > > Justin Clift wrote:\n> > > > > Only two things which have the potential to be worth waiting for, from\n> > > > > what I'm aware of. There may be others:\n> > > > >\n> > > > > - Find out from Sir Mordred if he wants to take a look at the CVS\n> > > > > version of code and audit in that for a bit, Just In Case he turns\n> > > > > up something that's serious and requires substantial re-work.\n> > > > > Although it means he wouldn't have a bunch of \"I found this existing\n> > > > > exploit\" type releases, we could instead offer him credit on the\n> > > > > press release along the lines of \"This released has been audited for\n> > > > > security flaws in its code by Sir Mordred\". Am pretty sure he'd\n> > > > > do a very thorough job for that, as it means he'd have an official\n> > > > > \"product reputation\" he'd need to stand by for it.\n> > > > \n> > > > This is interesting. He would have a month to do it.\n> > > \n> > > Reckon it's worth asking him, to find out if he'd be interested in this?\n> > \n> > \n> > I wouldn't do it yet until we know if we are going to delay.\n> \n> I'd ask anyway. 99% of the issues he finds will be fairly localized. \n> Anything truly new (not on TODO already) will probably require a fair\n> bit of time to track down, then fix time on top (2 months delay?).\n> \n> Anyway, these types of discoveries are better in beta than after the\n> release and would still warrent a mention if there is a fair amount of\n> ground covered.\n> \n> \n> Personally, I'd be more interested in whats safe (covered) than whats\n> broken. Posting the successful test cases as some proof rowards\n> stability / security of the new release would realize immediate gains in\n> settling nervous VPs about the new installation.\n> \n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 13:31:26 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On 21 Aug 2002, Robert Treat wrote:\n\n> Assuming that we do go ahead with a 7.2.2 release, can we get some kind\n> of unofficial statement on pushing back the 7.3 beta? I know Tom was\n\nv7.3 goes beta Sept 1st ... v7.2.2 will be purely a security bugfix\nrelease, with no changes in functionality that should (I hope?) require\nanything more then a simple re-install ...\n\n\n", "msg_date": "Wed, 21 Aug 2002 14:47:26 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On Wed, 21 Aug 2002, Bruce Momjian wrote:\n\n>\n> We learned a few lessons from previous releases. First, don't delay\n> the beta by days/weeks that drag on. Delay one month at a time.\n> Second, don't decide on a further delay the day before you are going to\n> go beta. Multiple short-period delays and delays that happen at the\n> last minute cause too many stops/starts for developers to be effective,\n> so...\n>\n> If we are going to delay beta, we should decide now, not at the end of\n> August, and the delay should be until the end of September. The big\n> question is whether we have enough material to warrant a delay.\n\nBeta goes down in 1 week ... if we follow what we had talked about before,\nwithin a short period of time after beta, we should be able to let ppl\ndive into working on v7.4 (or 8.0, whatever we decide to call it) ... but\nlet's try and stick to a timeline for once, else we are going to hit the\nsame as the last *very* extended release ...\n\n", "msg_date": "Wed, 21 Aug 2002 14:50:36 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On Thu, 22 Aug 2002, Justin Clift wrote:\n\n> - Find out from Sir Mordred if he wants to take a look at the CVS\n> version of code and audit in that for a bit, Just In Case he turns\n> up something that's serious and requires substantial re-work.\n> Although it means he wouldn't have a bunch of \"I found this existing\n> exploit\" type releases, we could instead offer him credit on the\n> press release along the lines of \"This released has been audited for\n> security flaws in its code by Sir Mordred\". Am pretty sure he'd\n> do a very thorough job for that, as it means he'd have an official\n> \"product reputation\" he'd need to stand by for it.\n\n\"Security Relatd Fixed\" are applicable for adoption during the beta\nperiod, leading up to release ...\n\n> - Patches to the CVS tree which let us have a truly native windows\n> version. This is of huge significance and would *very* much improve\n> our growth and adoption by being in this release in comparison to\n> being in the release afterwards. Not in an airy fairy way, but\n> quite definitely and solidly.\n\nIf they aren't in by now, they should wait until the next dev cycle ...\nunless they are *small* changes ...\n\n", "msg_date": "Wed, 21 Aug 2002 14:52:27 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On Wed, 21 Aug 2002, Bruce Momjian wrote:\n\n> Justin Clift wrote:\n> > Only two things which have the potential to be worth waiting for, from\n> > what I'm aware of. There may be others:\n> >\n> > - Find out from Sir Mordred if he wants to take a look at the CVS\n> > version of code and audit in that for a bit, Just In Case he turns\n> > up something that's serious and requires substantial re-work.\n> > Although it means he wouldn't have a bunch of \"I found this existing\n> > exploit\" type releases, we could instead offer him credit on the\n> > press release along the lines of \"This released has been audited for\n> > security flaws in its code by Sir Mordred\". Am pretty sure he'd\n> > do a very thorough job for that, as it means he'd have an official\n> > \"product reputation\" he'd need to stand by for it.\n>\n> This is interesting. He would have a month to do it.\n\nA month in beta, ya ... or more, depending on how beta went ... but a\n'Security Audit' shouldn' tlogically be done until the code base is frozen\nfor Beta anyway, since who knows, if it isn't frozen, whether someone is\ngoing to introduce something else in the mean time ...\n\n> The other issue is PITR, which I have been told today will not be ready\n> for a September 1 beta but may be ready for an October 1 beta.\n\nThen it can wait for v7.4 ... period. Have we not learnt from past\n'delays' ... hell, you yourself use \"may be ready for\", so, what, Oct 1st\nrolls around and we delay for, say, another 2 weeks cause it \"may be ready\nfor then\"?\n\nNo ... if PITR and Native Windows aren't ready for inclusion, then they\ncan be the foundation for the v7.4 release ...\n\nSept 1st - 1st Beta ... Oct 1st is the tentative release date ... which\ngives Mordred a month to audit the code and report any security bugs\nbefore th release, where there are no substantive changes going into the\ncode that will invalidate his results ...\n\n", "msg_date": "Wed, 21 Aug 2002 14:57:35 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On Wed, 2002-08-21 at 13:50, Marc G. Fournier wrote:\n> On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> \n> >\n> > We learned a few lessons from previous releases. First, don't delay\n> > the beta by days/weeks that drag on. Delay one month at a time.\n> > Second, don't decide on a further delay the day before you are going to\n> > go beta. Multiple short-period delays and delays that happen at the\n> > last minute cause too many stops/starts for developers to be effective,\n> > so...\n> >\n> > If we are going to delay beta, we should decide now, not at the end of\n> > August, and the delay should be until the end of September. The big\n> > question is whether we have enough material to warrant a delay.\n> \n> Beta goes down in 1 week ... if we follow what we had talked about before,\n> within a short period of time after beta, we should be able to let ppl\n> dive into working on v7.4 (or 8.0, whatever we decide to call it) ... but\n> let's try and stick to a timeline for once, else we are going to hit the\n> same as the last *very* extended release ...\n\nAgreed. If patches are applied to the 7.4 branch as fast as normal,\nthen maybe 7.4 will only be 6 months out with well tested Windows, PIT,\netc. code that gets applied this October.\n\nWhats the intended branchpoint? Beta with less than 5 patches? 3rd\nbeta start period? Less than 100 lines changed between betas?\n\nWhere is the reasonable point where double patching isn't as annoying as\nwaiting to apply new work?\n\n\n", "msg_date": "21 Aug 2002 13:57:37 -0400", "msg_from": "Rod Taylor <rbt@zort.ca>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On Wed, 21 Aug 2002, Bruce Momjian wrote:\n\n> Justin Clift wrote:\n> > Reckon it's worth asking him, to find out if he'd be interested in this?\n>\n>\n> I wouldn't do it yet until we know if we are going to delay.\n\nAny security audit of the code should *not* be done while the code is in\nflux, and if we delay, the code would be in flux since the delay would be\nto throw in a load of other code that would invalidate the audit results\n...\n\n> Oh, so it is Jan's group. Great news; wish someone would have told me\n> sooner. I removed the Win32 off the open items list because, with no\n> info and no one commenting on the item, it seemed dead for 7.3.\n\nAnd it should be ... we can put the Win32 patches up on the ftp site for\nppl to play with if they want, but to include it at this late a date would\nbe irresponsible ...\n\n> Well, PITR is a much more desired feature even than Win32; the big\n> question is how long PITR will actually take, seeing as we haven't see\n> any patches yet.\n>\n> However, we haven't seen any Win32 patches yet either, so they are in\n> the same boat as far as I am concerned.\n>\n> We have an open items list that is pretty much ready for 7.3. The only\n> open items of significance left is whether schema/DROP COLUMN stuff is\n> ready in all the interfaces/apps.\n\nWe set a timeline for beta ... this time, let's stick to it ... its not\nlike we didn't advertise when we were going into beta ... hell, even when\nthe patches are presented for PITR support, who knows whether they will be\naccepted, or what kinda bugs they are going to throw into the mix, or ...\n\n", "msg_date": "Wed, 21 Aug 2002 15:02:59 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On 21 Aug 2002, Rod Taylor wrote:\n\n> Agreed. If patches are applied to the 7.4 branch as fast as normal,\n> then maybe 7.4 will only be 6 months out with well tested Windows, PIT,\n> etc. code that gets applied this October.\n>\n> Whats the intended branchpoint? Beta with less than 5 patches? 3rd\n> beta start period? Less than 100 lines changed between betas?\n\nActually, I believe the agreement on branchpoint for this release was\nrelease date ... with the normal having been a few weeks *after* release\nin previous releases ...\n\n", "msg_date": "Wed, 21 Aug 2002 15:05:07 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "\nOK, beta starts on time, September 1. I agree we should keep the\nagreed-upon date, and that the propsed features aren't ready, but I had\nto let the discussion happen so people felt their opinions where being\nheard.\n\n---------------------------------------------------------------------------\n\nMarc G. Fournier wrote:\n> On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> \n> > Justin Clift wrote:\n> > > Reckon it's worth asking him, to find out if he'd be interested in this?\n> >\n> >\n> > I wouldn't do it yet until we know if we are going to delay.\n> \n> Any security audit of the code should *not* be done while the code is in\n> flux, and if we delay, the code would be in flux since the delay would be\n> to throw in a load of other code that would invalidate the audit results\n> ...\n> \n> > Oh, so it is Jan's group. Great news; wish someone would have told me\n> > sooner. I removed the Win32 off the open items list because, with no\n> > info and no one commenting on the item, it seemed dead for 7.3.\n> \n> And it should be ... we can put the Win32 patches up on the ftp site for\n> ppl to play with if they want, but to include it at this late a date would\n> be irresponsible ...\n> \n> > Well, PITR is a much more desired feature even than Win32; the big\n> > question is how long PITR will actually take, seeing as we haven't see\n> > any patches yet.\n> >\n> > However, we haven't seen any Win32 patches yet either, so they are in\n> > the same boat as far as I am concerned.\n> >\n> > We have an open items list that is pretty much ready for 7.3. The only\n> > open items of significance left is whether schema/DROP COLUMN stuff is\n> > ready in all the interfaces/apps.\n> \n> We set a timeline for beta ... this time, let's stick to it ... its not\n> like we didn't advertise when we were going into beta ... hell, even when\n> the patches are presented for PITR support, who knows whether they will be\n> accepted, or what kinda bugs they are going to throw into the mix, or ...\n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 14:06:29 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "Marc G. Fournier wrote:\n> On 21 Aug 2002, Rod Taylor wrote:\n> \n> > Agreed. If patches are applied to the 7.4 branch as fast as normal,\n> > then maybe 7.4 will only be 6 months out with well tested Windows, PIT,\n> > etc. code that gets applied this October.\n> >\n> > Whats the intended branchpoint? Beta with less than 5 patches? 3rd\n> > beta start period? Less than 100 lines changed between betas?\n> \n> Actually, I believe the agreement on branchpoint for this release was\n> release date ... with the normal having been a few weeks *after* release\n> in previous releases ...\n\nActually, you proposed beta2 as the branch time for this release, and no\none objected. I think we will have to be flexible and see how heavy the\npatching is during beta, but I think the latest would be on release\ndate. Our releases are very solid now, so we have very little patching\nafter release, and now that I and Tom are fulltime, we can handle the\nload of double-patching better.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 14:08:09 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "Let me see if I have my \"release dates\" straight:\n\nA 7.2.2 release in the next week or so that fixes the bugtraq buffer\noverflows and timestamp issues\n\nA 7.3 beta on Sept 1st that has all the new schema jazz and also the\nfixes for opaque (as well as other stuff from todo) during which time we\nget more security auditing\n\nHopefully an official 7.3 release on October 1.\n\n7.4/8.0 development will start and native windows and PITR patches can\nstart being submitted for that?\n\nRobert Treat\n\nOn Wed, 2002-08-21 at 14:06, Bruce Momjian wrote:\n> \n> OK, beta starts on time, September 1. I agree we should keep the\n> agreed-upon date, and that the propsed features aren't ready, but I had\n> to let the discussion happen so people felt their opinions where being\n> heard.\n> \n> ---------------------------------------------------------------------------\n> \n> Marc G. Fournier wrote:\n> > On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> > \n> > > Justin Clift wrote:\n> > > > Reckon it's worth asking him, to find out if he'd be interested in this?\n> > >\n> > >\n> > > I wouldn't do it yet until we know if we are going to delay.\n> > \n> > Any security audit of the code should *not* be done while the code is in\n> > flux, and if we delay, the code would be in flux since the delay would be\n> > to throw in a load of other code that would invalidate the audit results\n> > ...\n> > \n> > > Oh, so it is Jan's group. Great news; wish someone would have told me\n> > > sooner. I removed the Win32 off the open items list because, with no\n> > > info and no one commenting on the item, it seemed dead for 7.3.\n> > \n> > And it should be ... we can put the Win32 patches up on the ftp site for\n> > ppl to play with if they want, but to include it at this late a date would\n> > be irresponsible ...\n> > \n> > > Well, PITR is a much more desired feature even than Win32; the big\n> > > question is how long PITR will actually take, seeing as we haven't see\n> > > any patches yet.\n> > >\n> > > However, we haven't seen any Win32 patches yet either, so they are in\n> > > the same boat as far as I am concerned.\n> > >\n> > > We have an open items list that is pretty much ready for 7.3. The only\n> > > open items of significance left is whether schema/DROP COLUMN stuff is\n> > > ready in all the interfaces/apps.\n> > \n> > We set a timeline for beta ... this time, let's stick to it ... its not\n> > like we didn't advertise when we were going into beta ... hell, even when\n> > the patches are presented for PITR support, who knows whether they will be\n> > accepted, or what kinda bugs they are going to throw into the mix, or ...\n> > \n> > \n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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", "msg_date": "21 Aug 2002 14:19:19 -0400", "msg_from": "Robert Treat <xzilla@users.sourceforge.net>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On Wed, 21 Aug 2002, Bruce Momjian wrote:\n\n> Marc G. Fournier wrote:\n> > On 21 Aug 2002, Rod Taylor wrote:\n> >\n> > > Agreed. If patches are applied to the 7.4 branch as fast as normal,\n> > > then maybe 7.4 will only be 6 months out with well tested Windows, PIT,\n> > > etc. code that gets applied this October.\n> > >\n> > > Whats the intended branchpoint? Beta with less than 5 patches? 3rd\n> > > beta start period? Less than 100 lines changed between betas?\n> >\n> > Actually, I believe the agreement on branchpoint for this release was\n> > release date ... with the normal having been a few weeks *after* release\n> > in previous releases ...\n>\n> Actually, you proposed beta2 as the branch time for this release, and no\n> one objected. I think we will have to be flexible and see how heavy the\n> patching is during beta, but I think the latest would be on release\n> date. Our releases are very solid now, so we have very little patching\n> after release, and now that I and Tom are fulltime, we can handle the\n> load of double-patching better.\n\nOh, even better :) I didn't think I had succeeded in getting it brought\n*that* far forward :)\n\n\n", "msg_date": "Wed, 21 Aug 2002 15:19:22 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Branch Date (Was: Re: @(#)Mordred Labs advisory 0...)" }, { "msg_contents": "On 21 Aug 2002, Robert Treat wrote:\n\n> Let me see if I have my \"release dates\" straight:\n>\n> A 7.2.2 release in the next week or so that fixes the bugtraq buffer\n> overflows and timestamp issues\n>\n> A 7.3 beta on Sept 1st that has all the new schema jazz and also the\n> fixes for opaque (as well as other stuff from todo) during which time we\n> get more security auditing\n>\n> Hopefully an official 7.3 release on October 1.\n>\n> 7.4/8.0 development will start and native windows and PITR patches can\n> start being submitted for that?\n\nCorrect ... and, if all goes well, patches for the v7.4 branch should be\nacceptable sometime mid-Sept ...\n\n\n\n>\n> Robert Treat\n>\n> On Wed, 2002-08-21 at 14:06, Bruce Momjian wrote:\n> >\n> > OK, beta starts on time, September 1. I agree we should keep the\n> > agreed-upon date, and that the propsed features aren't ready, but I had\n> > to let the discussion happen so people felt their opinions where being\n> > heard.\n> >\n> > ---------------------------------------------------------------------------\n> >\n> > Marc G. Fournier wrote:\n> > > On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> > >\n> > > > Justin Clift wrote:\n> > > > > Reckon it's worth asking him, to find out if he'd be interested in this?\n> > > >\n> > > >\n> > > > I wouldn't do it yet until we know if we are going to delay.\n> > >\n> > > Any security audit of the code should *not* be done while the code is in\n> > > flux, and if we delay, the code would be in flux since the delay would be\n> > > to throw in a load of other code that would invalidate the audit results\n> > > ...\n> > >\n> > > > Oh, so it is Jan's group. Great news; wish someone would have told me\n> > > > sooner. I removed the Win32 off the open items list because, with no\n> > > > info and no one commenting on the item, it seemed dead for 7.3.\n> > >\n> > > And it should be ... we can put the Win32 patches up on the ftp site for\n> > > ppl to play with if they want, but to include it at this late a date would\n> > > be irresponsible ...\n> > >\n> > > > Well, PITR is a much more desired feature even than Win32; the big\n> > > > question is how long PITR will actually take, seeing as we haven't see\n> > > > any patches yet.\n> > > >\n> > > > However, we haven't seen any Win32 patches yet either, so they are in\n> > > > the same boat as far as I am concerned.\n> > > >\n> > > > We have an open items list that is pretty much ready for 7.3. The only\n> > > > open items of significance left is whether schema/DROP COLUMN stuff is\n> > > > ready in all the interfaces/apps.\n> > >\n> > > We set a timeline for beta ... this time, let's stick to it ... its not\n> > > like we didn't advertise when we were going into beta ... hell, even when\n> > > the patches are presented for PITR support, who knows whether they will be\n> > > accepted, or what kinda bugs they are going to throw into the mix, or ...\n> > >\n> > >\n> >\n> > --\n> > Bruce Momjian | http://candle.pha.pa.us\n> > pgman@candle.pha.pa.us | (610) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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", "msg_date": "Wed, 21 Aug 2002 15:28:16 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Dev Cycles (Was: Re: @(#)Mordred Labs advisory 0x...)" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> Tom Lane <tgl@sss.pgh.pa.us> writes:\n> \n> > Vince Vielhaber <vev@michvhf.com> writes:\n> > > Here's yet another. He claims malicious code can be run on the server\n> > > with this one.\n> > \n> > regression=# select repeat('xxx',1431655765);\n> > server closed the connection unexpectedly\n> > \n> > This is probably caused by integer overflow in calculating the size of\n> > the repeat's result buffer. It'd take some considerable doing to create\n> > an arbitrary-code exploit, but perhaps could be done. Anyone want to\n> > investigate a patch?\n> \n> This seems to fix the problem:\n\nNo, no it does not :-)\n\nTom pointed out some obvious braindamage in my previous patch. I've\nattached a revised version.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "21 Aug 2002 17:04:22 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL\n\t(fwd)" }, { "msg_contents": "\nYep, that's the plan!\n\n---------------------------------------------------------------------------\n\nRobert Treat wrote:\n> Let me see if I have my \"release dates\" straight:\n> \n> A 7.2.2 release in the next week or so that fixes the bugtraq buffer\n> overflows and timestamp issues\n> \n> A 7.3 beta on Sept 1st that has all the new schema jazz and also the\n> fixes for opaque (as well as other stuff from todo) during which time we\n> get more security auditing\n> \n> Hopefully an official 7.3 release on October 1.\n> \n> 7.4/8.0 development will start and native windows and PITR patches can\n> start being submitted for that?\n> \n> Robert Treat\n> \n> On Wed, 2002-08-21 at 14:06, Bruce Momjian wrote:\n> > \n> > OK, beta starts on time, September 1. I agree we should keep the\n> > agreed-upon date, and that the propsed features aren't ready, but I had\n> > to let the discussion happen so people felt their opinions where being\n> > heard.\n> > \n> > ---------------------------------------------------------------------------\n> > \n> > Marc G. Fournier wrote:\n> > > On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> > > \n> > > > Justin Clift wrote:\n> > > > > Reckon it's worth asking him, to find out if he'd be interested in this?\n> > > >\n> > > >\n> > > > I wouldn't do it yet until we know if we are going to delay.\n> > > \n> > > Any security audit of the code should *not* be done while the code is in\n> > > flux, and if we delay, the code would be in flux since the delay would be\n> > > to throw in a load of other code that would invalidate the audit results\n> > > ...\n> > > \n> > > > Oh, so it is Jan's group. Great news; wish someone would have told me\n> > > > sooner. I removed the Win32 off the open items list because, with no\n> > > > info and no one commenting on the item, it seemed dead for 7.3.\n> > > \n> > > And it should be ... we can put the Win32 patches up on the ftp site for\n> > > ppl to play with if they want, but to include it at this late a date would\n> > > be irresponsible ...\n> > > \n> > > > Well, PITR is a much more desired feature even than Win32; the big\n> > > > question is how long PITR will actually take, seeing as we haven't see\n> > > > any patches yet.\n> > > >\n> > > > However, we haven't seen any Win32 patches yet either, so they are in\n> > > > the same boat as far as I am concerned.\n> > > >\n> > > > We have an open items list that is pretty much ready for 7.3. The only\n> > > > open items of significance left is whether schema/DROP COLUMN stuff is\n> > > > ready in all the interfaces/apps.\n> > > \n> > > We set a timeline for beta ... this time, let's stick to it ... its not\n> > > like we didn't advertise when we were going into beta ... hell, even when\n> > > the patches are presented for PITR support, who knows whether they will be\n> > > accepted, or what kinda bugs they are going to throw into the mix, or ...\n> > > \n> > > \n> > \n> > -- \n> > Bruce Momjian | http://candle.pha.pa.us\n> > pgman@candle.pha.pa.us | (610) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 17:05:01 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "On Wed, 21 Aug 2002, Gavin Sherry wrote:\n\n> On Tue, 20 Aug 2002, Thomas Lockhart wrote:\n> \n> > ...\n> > > So I think that fixing the opaque problems in 7.2.x is simply\n> > > impossible. Given that, the question is whether we should make a 7.2.2\n> > > release with fixes for the other security holes (lpad(), rpad(),\n> > > reverse(), and the datetime overruns). IMHO, we should.\n> > \n> > Just a minor point: can someone actually show a symptom with date/time\n> > problems in 7.2.x?\n> \n\n[snip]\n\n> server closed the connection unexpectedly\n> This probably means the server terminated abnormally\n> before or while processing the request.\n> The connection to the server was lost. Attempting reset: Failed.\n> !#\n> \n> ParseDateTime() isn't checking that str < MAXDATELEN -- which is the\n> problem you solved in the datetime.c fixes.\n\nI had a look at this code on the train. There does not appear to be any\nway on conventional hardware manipulate this bug to smash the stack. This\nis due to the fact that ParseDateTime() returns to the caller if it\nencounters a non-printable character. It would be perhaps one of the most\nimpressive hacks ever if someone could dream machine code to put in the\noverrun which consisted entirely of printable characters.\n\nAs such, it is remarkably unlikely that someone could exploit this bug to\nexecute arbitary code.\n\nGavin\n\n", "msg_date": "Thu, 22 Aug 2002 11:39:58 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "> > If we are going to delay beta, we should decide now, not at the end of\n> > August, and the delay should be until the end of September. The big\n> > question is whether we have enough material to warrant a delay.\n>\n> Beta goes down in 1 week ... if we follow what we had talked about before,\n> within a short period of time after beta, we should be able to let ppl\n> dive into working on v7.4 (or 8.0, whatever we decide to call it) ... but\n> let's try and stick to a timeline for once, else we are going to hit the\n> same as the last *very* extended release ...\n\nI'd much rather see 7.3 as-is then an 8.0 not so later on with win32 native\nand PITR _properly_tested_.\n\nChris\n\n", "msg_date": "Thu, 22 Aug 2002 09:48:04 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" }, { "msg_contents": "\nPatch applied. Thanks.\n\n---------------------------------------------------------------------------\n\n\nNeil Conway wrote:\n> Tom Lane <tgl@sss.pgh.pa.us> writes:\n> > Neil Conway <neilc@samurai.com> writes:\n> > > + \t/* Check for integer overflow */\n> > > + \tif (tlen / slen != count)\n> > > + \t\telog(ERROR, \"Requested buffer is too large.\");\n> > \n> > What about slen == 0?\n> \n> Good point -- that wouldn't cause incorrect results or a security\n> problem, but it would reject input that we should really accept.\n> \n> Revised patch is attached.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 00:54:28 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL" }, { "msg_contents": "Gavin Sherry <swm@linuxworld.com.au> writes:\n\n> It would be perhaps one of the most impressive hacks ever if someone\n> could dream machine code to put in the overrun which consisted\n> entirely of printable characters.\n\nAt least for the x86 architecture, working ASCII-only shell code\nexists (even shell code which consists just of letters!). See for\nexample:\n\nhttp://cert.uni-stuttgart.de/archive/vuln-dev/2000/10/msg00200.html\n\nASCII-only shellcode for RISC platforms is even harder and might be\nimpossible.\n\n-- \nFlorian Weimer \t Weimer@CERT.Uni-Stuttgart.DE\nUniversity of Stuttgart http://CERT.Uni-Stuttgart.DE/people/fw/\nRUS-CERT fax +49-711-685-5898\n", "msg_date": "Thu, 22 Aug 2002 13:30:19 +0200", "msg_from": "Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in" } ]
[ { "msg_contents": "\n> \"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> > Hard to say what is good for those names imho, don't like \n> \"anytype\" :-(\n> \n> How about \"any\"? It's a reserved word per SQL99, I think.\n\nI would actually stick to opaque in that case, already used in other db's.\n\n> > I like \"cstring\", \"void\" and \"internal\". \n> \n> Okay.\n> \n> > Maybe \"anyarray\" instead of \"anyarraytype\".\n> \n> That would match with \"any\".\n\nI thought you wanted it separate ?\n\n> \n> > And I would prefer \"row\" instead of \"tuple\".\n> \n> I'm leaning towards agreeing with Stephan: we should use typename\n> \"trigger\" to declare triggers. \"Tuple\" (or \"row\") is strictly correct\n> only for BEFORE triggers, not AFTER triggers, so it's a bit of a\n> misnomer for triggers anyhow.\n\nConvinced.\n\n> \n> I'm now also toying with inventing a pseudotype just for procedural\n> language handlers, which are currently \"foo() returns opaque\". If we\n> want the type system to catch misuses of trigger functions, we should\n> want it for handlers too. Maybe name this type \"language_handler\"?\n\n\"HANDLER\" would again already be a reserved word, sounds good.\n\nAndreas\n", "msg_date": "Tue, 20 Aug 2002 22:17:28 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n>> \"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n>>> Hard to say what is good for those names imho, don't like \n>>> \"anytype\" :-(\n>> \n>> How about \"any\"? It's a reserved word per SQL99, I think.\n\n> I would actually stick to opaque in that case, already used in other db's.\n\nI want to change the name because (a) we are changing the semantics,\n(b) we can't throw notices for opaque if we keep it as a valid choice.\n\n>>> Maybe \"anyarray\" instead of \"anyarraytype\".\n>> \n>> That would match with \"any\".\n\n> I thought you wanted it separate ?\n\nI meant that if the one name is \"any\", then making the other \"anyarray\"\n(ie, both without \"type\" on the end) is consistent.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 16:21:50 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " } ]
[ { "msg_contents": "> >>> Hard to say what is good for those names imho, don't like \n> >>> \"anytype\" :-(\n> >> \n> >> How about \"any\"? It's a reserved word per SQL99, I think.\n> \n> > I would actually stick to opaque in that case, already used in other db's.\n> \n> I want to change the name because (a) we are changing the semantics,\n> (b) we can't throw notices for opaque if we keep it as a valid choice.\n\nHmm, \"any\" would sound like it is the same as opaque. Would \"any\" really be\nall allowed types ? I think we would want to eliminate that altogether.\nIf it is not all types I would rather use a more restrictive name (nulltype \n/ anynumeric).\n \nImho opaque is missing a runtime type info, like a descriptor, \nand thus only \"pass by value\" could not be allowed anymore.\n\nI guess I must sleep over this, not convinced about depricating opaque yet :-)\n\n> I meant that if the one name is \"any\", then making the other \"anyarray\"\n> (ie, both without \"type\" on the end) is consistent.\n\nAh, good.\n\nAndreas\n", "msg_date": "Tue, 20 Aug 2002 22:46:34 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " }, { "msg_contents": "On Tue, 2002-08-20 at 16:46, Zeugswetter Andreas SB SD wrote:\n> > >>> Hard to say what is good for those names imho, don't like \n> > >>> \"anytype\" :-(\n> > >> \n> > >> How about \"any\"? It's a reserved word per SQL99, I think.\n> > \n> > > I would actually stick to opaque in that case, already used in other db's.\n> > \n> > I want to change the name because (a) we are changing the semantics,\n> > (b) we can't throw notices for opaque if we keep it as a valid choice.\n> \n> Hmm, \"any\" would sound like it is the same as opaque. Would \"any\" really be\n> all allowed types ? I think we would want to eliminate that altogether.\n\nErm.. count(*) <- * is literally anything.\n\n", "msg_date": "20 Aug 2002 16:50:43 -0400", "msg_from": "Rod Taylor <rbt@zort.ca>", "msg_from_op": false, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in" }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> Hmm, \"any\" would sound like it is the same as opaque. Would \"any\" really be\n> all allowed types ? I think we would want to eliminate that altogether.\n\nDo you plan on eliminating the COUNT() aggregate, then?\n \n> Imho opaque is missing a runtime type info, like a descriptor, \n> and thus only \"pass by value\" could not be allowed anymore.\n\nAFAICS it's only useful for functions that only care whether their\nargument is NULL or not, and don't inspect its value. But that just\nhappens to describe COUNT, as well as nullvalue/nonnullvalue.\n\nI don't really think that using ANY instead of OPAQUE for this purpose\nwill affect users, because they will never be declaring any functions\nthat would legitimately take ANY, much less return ANY (the latter\nprobably makes no sense at all). It seems to me that COUNT, nullvalue,\nand nonnullvalue pretty much cover the spectrum of what you can usefully\ndo with only an isnull bit to look at...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 16:56:00 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " } ]
[ { "msg_contents": "\nAnd another one. Sure would be nice if shit-for-brains would mention\nit to us first.\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n---------- Forwarded message ----------\nDate: Tue, 20 Aug 2002 15:01:34 +0000\nFrom: Sir Mordred The Traitor <mordred@s-mail.com>\nTo: bugtraq@securityfocus.com\nSubject: @(#)Mordred Labs advisory 0x0004: Multiple buffer overflows in\n PostgreSQL.\n\n\n//@(#) Mordred Labs advisory 0x0004\n\nRelease data: 20/08/02\nName: Two buffer overflows in PostgreSQL\nVersions affected: all versions\nConditions: multibyte support\nRisk: average\n\n--[ Description:\n\nI guess all of you already hear about the PostgreSQL.\nIf not, try to visit\nhttp://www.postgresql.org/idocs/index.php?preface.html#INTRO-WHATIS.\n\nThere are two buffer overflows in src/backend/utils/adt/oracle_compat.c.\n1) lpad(text, integer, text) function\n2) rpad(text, integer, text) function\n\n--[ Details:\n\nThe code for this functions is\nsrc/backend/utils/adt/oracle_compat.c::lpad() and\nsrc/backend/utils/adt/oracle_compat.c::rpad() respectively.\nThe code suffers from a buffer overflow (of course).\n\n--[ How to reproduce:\n\nshell> pgsql template1 postgres\ntemplate1=# select version();\n version\n-----------------------------------------------------------\n PostgreSQL 7.2 on i686-pc-linux-gnu, compiled by GCC 2.96\n(1 row)\n\ntemplate1=# create database my_db with encoding='UNICODE';\nCREATE DATABASE\ntemplate1# \\c my_db\nYou are now connected to database my_db.\n\nmy_db=# select lpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');\npqReadData() -- backend closed the channel unexpectedly.\n This probably means the backend terminated abnormally\n before or while processing the request.\nThe connection to the server was lost. Attempting reset: Failed.\n!#\n\nThe same for rpad() function.\n\nThe vulnerable encodings are: EUC_JP, EUC_CN, EUC_KR, EUC_TW, UNICODE,\nMULE_INTERNAL.\n\n--[ Solution\n\nSecure coding of web applications, input validation checks...etc...\n\n\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n\n", "msg_date": "Tue, 20 Aug 2002 16:48:03 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": true, "msg_subject": "@(#)Mordred Labs advisory 0x0004: Multiple buffer overflows in\n\tPostgreSQL. (fwd)" }, { "msg_contents": "Vince Vielhaber <vev@michvhf.com> writes:\n> And another one. Sure would be nice if shit-for-brains would mention\n> it to us first.\n\nI don't even mind the \"first\" part, but it would certainly be polite of\nhim to cc: pghackers rather than expecting us to dig it off bugtraq.\n\nBut, as someone else pointed out, he's not doing this for our benefit,\nit's to make himself look good. The only audience he cares about is\nbugtraq, I suspect.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 20 Aug 2002 17:02:18 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer overflows in\n\tPostgreSQL. (fwd)" }, { "msg_contents": "Vince Vielhaber <vev@michvhf.com> writes:\n> And another one.\n\nThis patch should fix the problem. Doesn't include my previous patch\nfor repeat(). Again, somewhat off-the-cuff, so I might have missed\nsomething...\n\ntest=# select lpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');\nERROR: Requested length too large\ntest=# select rpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');\nERROR: Requested length too large\n\n(That's on a Unicode DB, haven't tested other encodings but AFAICT\nthis fix should still work.)\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "20 Aug 2002 17:08:39 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer overflows in\n\tPostgreSQL. (fwd)" }, { "msg_contents": "On 20 Aug 2002, Neil Conway wrote:\n\n> Vince Vielhaber <vev@michvhf.com> writes:\n> > And another one.\n>\n> This patch should fix the problem. Doesn't include my previous patch\n> for repeat(). Again, somewhat off-the-cuff, so I might have missed\n> something...\n>\n> test=# select lpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');\n> ERROR: Requested length too large\n> test=# select rpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');\n> ERROR: Requested length too large\n>\n> (That's on a Unicode DB, haven't tested other encodings but AFAICT\n> this fix should still work.)\n\nIs there any chance that pg_database_encoding_max_length() could return\nzero and give a divide by zero error? Or is that trapped?\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n", "msg_date": "Tue, 20 Aug 2002 17:19:29 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": true, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer" }, { "msg_contents": "Vince Vielhaber <vev@michvhf.com> writes:\n> On 20 Aug 2002, Neil Conway wrote:\n> Is there any chance that pg_database_encoding_max_length() could return\n> zero and give a divide by zero error? Or is that trapped?\n\nI don't think so (the array of encodings that contains the data seems\nto be pre-defined), but I know next to nothing about multibyte, so I\nmay be wrong. In any case, the \"divide by zero\" error would cause an\nelog(ERROR) anyway, not a segfault...\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "20 Aug 2002 17:29:14 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer overflows in\n\tPostgreSQL. (fwd)" }, { "msg_contents": "> > (That's on a Unicode DB, haven't tested other encodings but AFAICT\n> > this fix should still work.)\n> \n> Is there any chance that pg_database_encoding_max_length() could return\n> zero\n\nThat's impossible or at least is the evidence of something badly\nbroken.\n\n> and give a divide by zero error? Or is that trapped?\n--\nTatsuo Ishii\n", "msg_date": "Wed, 21 Aug 2002 10:08:16 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer" }, { "msg_contents": "\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://candle.pha.pa.us/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nNeil Conway wrote:\n> Vince Vielhaber <vev@michvhf.com> writes:\n> > And another one.\n> \n> This patch should fix the problem. Doesn't include my previous patch\n> for repeat(). Again, somewhat off-the-cuff, so I might have missed\n> something...\n> \n> test=# select lpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');\n> ERROR: Requested length too large\n> test=# select rpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');\n> ERROR: Requested length too large\n> \n> (That's on a Unicode DB, haven't tested other encodings but AFAICT\n> this fix should still work.)\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n[ Attachment, skipping... ]\n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 13:29:57 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer overflows" }, { "msg_contents": "\nPatch applied. Thanks.\n\n---------------------------------------------------------------------------\n\n\nNeil Conway wrote:\n> Vince Vielhaber <vev@michvhf.com> writes:\n> > And another one.\n> \n> This patch should fix the problem. Doesn't include my previous patch\n> for repeat(). Again, somewhat off-the-cuff, so I might have missed\n> something...\n> \n> test=# select lpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');\n> ERROR: Requested length too large\n> test=# select rpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');\n> ERROR: Requested length too large\n> \n> (That's on a Unicode DB, haven't tested other encodings but AFAICT\n> this fix should still work.)\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n[ Attachment, skipping... ]\n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 00:55:12 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer overflows" } ]
[ { "msg_contents": "Dann Corbit wrote:\n> > -----Original Message-----\n> > From: Neil Conway [mailto:neilc@samurai.com] \n> > Sent: Tuesday, August 20, 2002 1:44 PM\n> > To: Vince Vielhaber\n> > Cc: pgsql-hackers@postgreSQL.org\n> > Subject: Re: [HACKERS] @(#)Mordred Labs advisory 0x0003: \n> > Buffer overflow in PostgreSQL (fwd)\n> > \n> > \n> > Vince Vielhaber <vev@michvhf.com> writes:\n> > > Here's yet another.\n> > \n> > Should someone from the core team perhaps get in contact with \n> > this guy and ask if he could get in contact with the \n> > development team before publicizing any further security \n> > holes? AFAIK that is standard operating procedure in most cases...\n> \n> As long as we continue to find out about them, I would just let him work\n> away.\n> He is clearly an excellent tester, and if you had to hire him it would\n> be very expensive.\n> \n> As long as he is producing results of such great value, I think it is\n> wonderful.\n\nYea, someone please contact him and tell him to keep going.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 20 Aug 2002 16:52:05 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL" }, { "msg_contents": "> -----Original Message-----\n> From: Neil Conway [mailto:neilc@samurai.com] \n> Sent: Tuesday, August 20, 2002 1:44 PM\n> To: Vince Vielhaber\n> Cc: pgsql-hackers@postgreSQL.org\n> Subject: Re: [HACKERS] @(#)Mordred Labs advisory 0x0003: \n> Buffer overflow in PostgreSQL (fwd)\n> \n> \n> Vince Vielhaber <vev@michvhf.com> writes:\n> > Here's yet another.\n> \n> Should someone from the core team perhaps get in contact with \n> this guy and ask if he could get in contact with the \n> development team before publicizing any further security \n> holes? AFAIK that is standard operating procedure in most cases...\n\nAs long as we continue to find out about them, I would just let him work\naway.\nHe is clearly an excellent tester, and if you had to hire him it would\nbe very expensive.\n\nAs long as he is producing results of such great value, I think it is\nwonderful.\n \n> Second, it might be worth pushing a 7.2.2 release containing \n> the fix for this bug, as well as the datetime problem. If \n> that sounds reasonable to the people who have to do the most \n> work on a new release (e.g. Marc), I can volunteer to \n> backport a fix for the datetime problem.\n\nBugs that cause a catastrophic error (e.g. \"crash\" of the database\nengine, causing loss of data) should have the highest priority. Call\nthem category zero.\nBugs that cause incorrect results should have the next highest priority.\nCall them category one.\nBugs that are minor annoyances (e.g. \"appearance\" such as a misspelled\nword in a help file) should be low priority.\nBugs that are only suggestions for improvements should have the lowest\npriority.\n\nAll known category zero and one bugs should be fixed before each and\nevery new release. IMO-YMMV.\n\n", "msg_date": "Tue, 20 Aug 2002 13:53:39 -0700", "msg_from": "\"Dann Corbit\" <DCorbit@connx.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0003: Buffer overflow in PostgreSQL\n\t(fwd)" } ]
[ { "msg_contents": "> -----Original Message-----\n> From: Vince Vielhaber [mailto:vev@michvhf.com] \n> Sent: Tuesday, August 20, 2002 1:48 PM\n> To: pgsql-hackers@postgreSQL.org\n> Subject: [HACKERS] @(#)Mordred Labs advisory 0x0004: Multiple \n> buffer overflows inPostgreSQL. (fwd)\n> \n> \n> \n> And another one. Sure would be nice if shit-for-brains would \n> mention it to us first.\n\nIt looks to me like he may be the most valuable tester on the staff.\nAs long as we find out what the problem is, why complain?\n", "msg_date": "Tue, 20 Aug 2002 13:54:53 -0700", "msg_from": "\"Dann Corbit\" <DCorbit@connx.com>", "msg_from_op": true, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer overflows\n\tinPostgreSQL. (fwd)" }, { "msg_contents": " .------[ Dann Corbit wrote (2002/08/20 at 13:54:53) ]------\n | \n | > From: Vince Vielhaber [mailto:vev@michvhf.com] \n | > Sent: Tuesday, August 20, 2002 1:48 PM\n | > To: pgsql-hackers@postgreSQL.org\n | > Subject: [HACKERS] @(#)Mordred Labs advisory 0x0004: Multiple \n | > buffer overflows inPostgreSQL. (fwd)\n | > \n | > \n | > \n | > And another one. Sure would be nice if shit-for-brains would \n | > mention it to us first.\n | \n | It looks to me like he may be the most valuable tester on the staff.\n | As long as we find out what the problem is, why complain?\n | \n `-------------------------------------------------\n\n The reason to complain is that he is not notifying the development\n team before hand. Giving them absolutely no chance to work on a\n fix prior to the whole world freaking out over these bugs. \n\n If I was your neighbor, and I noticed your front door was open I\n would contact you and let you know... not take out a full page\n ad in the local news paper! Same idea applies here. :) \n\n Also, if I'm not mistaken this guy isn't on \"staff\". \n\n ---------------------------------\n Frank Wiles <frank@wiles.org>\n http://frank.wiles.org\n ---------------------------------\n\n", "msg_date": "Tue, 20 Aug 2002 15:57:05 -0500", "msg_from": "Frank Wiles <frank@wiles.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer overflows\n\tinPostgreSQL. (fwd)" } ]
[ { "msg_contents": "> -----Original Message-----\n> From: Frank Wiles [mailto:frank@wiles.org] \n> Sent: Tuesday, August 20, 2002 1:57 PM\n> To: Dann Corbit\n> Cc: pgsql-hackers@postgresql.org\n> Subject: Re: [HACKERS] @(#)Mordred Labs advisory 0x0004: \n> Multiple buffer overflows inPostgreSQL. (fwd)\n> \n> \n> .------[ Dann Corbit wrote (2002/08/20 at 13:54:53) ]------\n> | \n> | > From: Vince Vielhaber [mailto:vev@michvhf.com] \n> | > Sent: Tuesday, August 20, 2002 1:48 PM\n> | > To: pgsql-hackers@postgreSQL.org\n> | > Subject: [HACKERS] @(#)Mordred Labs advisory 0x0004: Multiple \n> | > buffer overflows inPostgreSQL. (fwd)\n> | > \n> | > \n> | > \n> | > And another one. Sure would be nice if shit-for-brains would \n> | > mention it to us first.\n> | \n> | It looks to me like he may be the most valuable tester on \n> the staff. | As long as we find out what the problem is, \n> why complain? | \n> `-------------------------------------------------\n> \n> The reason to complain is that he is not notifying the development\n> team before hand. Giving them absolutely no chance to work on a\n> fix prior to the whole world freaking out over these bugs. \n> \n> If I was your neighbor, and I noticed your front door was open I\n> would contact you and let you know... not take out a full page\n> ad in the local news paper! Same idea applies here. :) \n> \n> Also, if I'm not mistaken this guy isn't on \"staff\". \n\nWell, of course, a well mannered team member would report the bugs\nthrough one of the normal channels.\nOn the other hand, a malicious tester who finds these problems performs\ntwo valuable services:\n1. Through great effort, he has found a problem that needs to be\naddressed or serious consequences will result.\n2. He has raised a large public rancor. The result of which is that\nthe serious problem must be addressed.\n\nThe motivation is suspect. The character is suspect. But the result is\nof great value. In a similar manner, it is a common practice to hire\nhackers to try to break into your site. While their methods will be\nunconventional, and they can be very seedy and immoral characters, they\nwill reveal information of great value to show you exactly where the\nhole needs to be plugged.\n", "msg_date": "Tue, 20 Aug 2002 14:05:37 -0700", "msg_from": "\"Dann Corbit\" <DCorbit@connx.com>", "msg_from_op": true, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer overflows\n\tinPostgreSQL. (fwd)" }, { "msg_contents": " .------[ Dann Corbit wrote (2002/08/20 at 14:05:37) ]------\n | \n | ... [large snip] ... \n |\n | Well, of course, a well mannered team member would report the bugs\n | through one of the normal channels.\n | On the other hand, a malicious tester who finds these problems performs\n | two valuable services:\n | 1. Through great effort, he has found a problem that needs to be\n | addressed or serious consequences will result.\n |\n | ... [small snip] ...\n |\n `-------------------------------------------------\n\n Reading the TODO list is \"great effort\"? What puzzles me most is\n that you speak as if you have personal knowledge of how much effort\n it took. \n\n ---------------------------------\n Frank Wiles <frank@wiles.org>\n http://frank.wiles.org\n ---------------------------------\n\n", "msg_date": "Tue, 20 Aug 2002 16:09:00 -0500", "msg_from": "Frank Wiles <frank@wiles.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0004: Multiple buffer overflows\n\tinPostgreSQL. (fwd)" } ]
[ { "msg_contents": "\nJ.R, Richard, we only have 10 days until beta. Can you give us any\nstatus on PITR?\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 20 Aug 2002 17:34:43 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [PATCHES] Patch for roll forward recovery (PITR)" } ]
[ { "msg_contents": "Hi Mordred,\n\nThanks for doing this testing for vulnerabilities for us.\n\nSome PostgreSQL team members have been looking to organise an \"audit\"\nteam to remove vulnerabilities like this, and your posts to BugTraq have\nassisted in getting that further mobilised.\n\nSomething that is concerning us though, is that whilst one of these bugs\nwas known and on our \"to fix\" list, there are some that were not known\nand you're not notifying us up front so we can fix them before details\nare publicly released.\n\nWould you be able to work in with us from here, notifying us of these\nvulnerabilities with some decent amount of time in advance so we can\ncreate the necessary patches/fixes, etc?\n\n:-)\n\nRegards and best wishes,\n\nJustin Clift\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, 21 Aug 2002 07:54:17 +1000", "msg_from": "Justin Clift <justin@postgresql.org>", "msg_from_op": true, "msg_subject": "Multiple vulnerabilities in PostgreSQL" } ]
[ { "msg_contents": "I've backpatched Thomas's fixes for the potential buffer overruns in\nthe datetime code to the REL7_2_STABLE branch. The required changes\nare pretty minimal, and the code passes the regression tests.\n\nHowever, I haven't yet seen a test-case that demonstrates the buffer\noverrun in 7.2 code (I originally found the problem when playing with\ndatetime on 7.3-dev with integer datetimes) -- I've tested it a little\nbit and it seems to work, but this patch probably requires pretty\nwidespread testing.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "20 Aug 2002 18:24:35 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": true, "msg_subject": "backpatch of datetime fixes" }, { "msg_contents": "Neil Conway wrote:\n> \n> I've backpatched Thomas's fixes for the potential buffer overruns in\n> the datetime code to the REL7_2_STABLE branch. The required changes\n> are pretty minimal, and the code passes the regression tests.\n> \n> However, I haven't yet seen a test-case that demonstrates the buffer\n> overrun in 7.2 code (I originally found the problem when playing with\n> datetime on 7.3-dev with integer datetimes)\n\nI've had trouble provoking an error also, but the circumstances should\nbe related to having a long date/time input, especially if it has more\nfields than one would expect in a valid date/time string.\n\n - Thomas\n", "msg_date": "Tue, 20 Aug 2002 18:18:46 -0700", "msg_from": "Thomas Lockhart <lockhart@fourpalms.org>", "msg_from_op": false, "msg_subject": "Re: backpatch of datetime fixes" } ]
[ { "msg_contents": "I'm going to change contrib/pgstattuple so that it returns a tuple\nrather than a NOTICE using new SRF interface. I believe this way is\nmuch more convenient for users. Any objection?\n--\nTatsuo Ishii\n", "msg_date": "Wed, 21 Aug 2002 12:25:29 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": true, "msg_subject": "pgstattuple change using SRF" }, { "msg_contents": "Tatsuo Ishii <t-ishii@sra.co.jp> writes:\n> I'm going to change contrib/pgstattuple so that it returns a tuple\n> rather than a NOTICE using new SRF interface. I believe this way is\n> much more convenient for users. Any objection?\n\nSure. I like the way Joe did show_all() better than the way Neil\ndid lock_status() --- I'm in the middle of getting rid of zero\ntype OIDs throughout the system, and so I'm not happy with anything\nthat re-introduces them, even if it's only transient during initdb.\nOther than that minor point, go for it...\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 01:54:36 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: pgstattuple change using SRF " }, { "msg_contents": "> Tatsuo Ishii <t-ishii@sra.co.jp> writes:\n> > I'm going to change contrib/pgstattuple so that it returns a tuple\n> > rather than a NOTICE using new SRF interface. I believe this way is\n> > much more convenient for users. Any objection?\n> \n> Sure. I like the way Joe did show_all() better than the way Neil\n> did lock_status() --- I'm in the middle of getting rid of zero\n> type OIDs throughout the system, and so I'm not happy with anything\n> that re-introduces them, even if it's only transient during initdb.\n> Other than that minor point, go for it...\n\nI think pgstattuple() never uses OPAQUE...\n--\nTatsuo Ishii\n", "msg_date": "Wed, 21 Aug 2002 15:11:28 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": true, "msg_subject": "Re: pgstattuple change using SRF " }, { "msg_contents": "> I'm going to change contrib/pgstattuple so that it returns a tuple\n> rather than a NOTICE using new SRF interface. I believe this way is\n> much more convenient for users. Any objection?\n\nI have committed above changes.\n--\nTatsuo Ishii\n", "msg_date": "Fri, 23 Aug 2002 17:20:59 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": true, "msg_subject": "Re: pgstattuple change using SRF" } ]
[ { "msg_contents": "Hi All,\n\nFew quick questions.\n\n1. How do you compile contribs with full debugging symbols. I always get\nheuristic-fencepost-blah probs with gdb even though I've configured postgres\nwith all the debugging stuff. Makes it really hard to find probs in the\ncontribs.\n\n2. Should contribs exclusively use palloc/pfree? I notice that\nfulltextindex is using malloc and free - is there any real reason not to\nchange them?\n\nI've attached the current tarball (extract in contrib) but my regression\ntest is crashing and I've had to fill it full of elogs to see that nargs is\nbeing overwritten by an overflow or something.\n\nFlorian - feel like helping me test/fix it. go 'gmake install' and then\n'gmake installcheck' to test it.\n\nThanks,\n\nChris", "msg_date": "Wed, 21 Aug 2002 11:29:36 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "Trouble debugging contrib/fulltextindex" }, { "msg_contents": "Christopher Kings-Lynne writes:\n\n> 1. How do you compile contribs with full debugging symbols. I always get\n> heuristic-fencepost-blah probs with gdb even though I've configured postgres\n> with all the debugging stuff.\n\nFirst, compile without optimization (-O0). Second, using LOAD to load the\nshared object before the function call may be of advantage. I would allow\nyou to set breakpoints in the loaded object, for example. Just waiting\nfor the code to crash and hoping to get a good backtrace isn't always a\nsuccessful strategy.\n\n> 2. Should contribs exclusively use palloc/pfree?\n\nYes (unless there are some wildly unusual memory allocation requirements).\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Wed, 21 Aug 2002 22:47:59 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: Trouble debugging contrib/fulltextindex" } ]
[ { "msg_contents": "> Hi Tatsuo Ishii :\n> \n> Thank you very much for your reply. Somehow I managed to restore the data by\n> creating by database in SQL_ASCII encoding. Then I changed the encoding of\n> my database into EUC_CN thru \"update pg_database set encoding=2\". But the\n> same problem happened in \"vacuum verbose analyze\". I went thru the changelog\n> and found out that you have add the multibyte validation feature from\n> release 7.2 on. Looks like the only way to get around is to patch the\n> source.\n> \n> But I would like to suggest that this feature should be made configurable.\n> You know encoding is a such a mess in the Chinese language, the popular\n> Chinese input methods actually allow the input of any Chinese character from\n> different encodings(e.g., GB2312/GBK, big5), it is difficult to make sure\n> that the user always input valid characters unless you validate the insert\n> and update clause.\n\nBut the encoding validation is made for such cases. It will prevent\nthe database from INSERT/UPDATE with wrong encoding data. If you\nreally want to allow input any GB2312/GBK, big5 etc. encoded data to\nthe database, why don't you simply use SQL_ASCII or some other single\nbyte encodings? If you mix up GB2312/GBK, big5 in a database you could\nnot perform sort or any other database operation properly anyway.\n--\nTatsuo Ishii\n", "msg_date": "Wed, 21 Aug 2002 15:06:12 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": true, "msg_subject": "Re: How to disable encoding validation during database restore" } ]
[ { "msg_contents": "I had a test schema containing one table.\n\n olly=# drop schema testing;\n NOTICE: table testing.testa depends on schema testing\n\n ERROR: Cannot drop schema testing because other objects depend on it\n \t Use DROP ... CASCADE to drop the dependent objects too\n\nThis seems a little over-restrictive to me. There is no dependency\noutside the schema, so shouldn't that have worked? I should have\nthought CASCADE would be implicit for objects inside the schema to be\ndropped.\n\n-- \nOliver Elphick Oliver.Elphick@lfix.co.uk\nIsle of Wight, UK \nhttp://www.lfix.co.uk/oliver\nGPG: 1024D/3E1D0C1C: CA12 09E0 E8D5 8870 5839 932A 614D 4C34 3E1D 0C1C\n ========================================\n \"For the Lord himself shall descend from heaven with a \n shout, with the voice of the archangel, and with the \n trump of God; and the dead in Christ shall rise first;\n Then we which are alive and remain shall be caught \n up together with them in the clouds, to meet the Lord \n in the air; and so shall we ever be with the Lord.\" \n I Thessalonians 4:16,17 \n\n", "msg_date": "21 Aug 2002 07:27:22 +0100", "msg_from": "Oliver Elphick <olly@lfix.co.uk>", "msg_from_op": true, "msg_subject": "Dropping a schema" }, { "msg_contents": "Oliver Elphick <olly@lfix.co.uk> writes:\n> olly=# drop schema testing;\n> NOTICE: table testing.testa depends on schema testing\n> ERROR: Cannot drop schema testing because other objects depend on it\n> \t Use DROP ... CASCADE to drop the dependent objects too\n\n> This seems a little over-restrictive to me.\n\nIt's per spec: SQL92 saith\n\n 3) If RESTRICT is specified, then S shall not contain any per-\n sistent base tables, global temporary tables, created local\n temporary tables, views, domains, assertions, character sets,\n collations, or translations.\n\n Note: If CASCADE is specified, then such objects will be dropped\n by the effective execution of the SQL schema manipulation state-\n ments specified in the General Rules of this Subclause.\n\nAlso, it seems the safest behavior to me. \"rmdir dir\" won't remove a\nnonempty directory; isn't that a pretty close analogy?\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 10:02:54 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Dropping a schema " }, { "msg_contents": "On Wed, 2002-08-21 at 15:02, Tom Lane wrote:\n> Oliver Elphick <olly@lfix.co.uk> writes:\n> > olly=# drop schema testing;\n> > NOTICE: table testing.testa depends on schema testing\n> > ERROR: Cannot drop schema testing because other objects depend on it\n> > \t Use DROP ... CASCADE to drop the dependent objects too\n> \n> > This seems a little over-restrictive to me.\n> \n> It's per spec: SQL92 saith\n...\n> Also, it seems the safest behavior to me. \"rmdir dir\" won't remove a\n> nonempty directory; isn't that a pretty close analogy?\n\nNot really, seeing that you can't say \"mkdir directory (containing these\nfiles)\". An implicit cascade *inside* the schema seems an appropriate\nparallel to \"CREATE SCHEMA ... (CREATE TABLE ...)\". After all, we don't\nhave to say \"DROP TABLE ... CASCADE\" because the table has rows in it!\n\nBut if that's what the spec says...\n\n-- \nOliver Elphick Oliver.Elphick@lfix.co.uk\nIsle of Wight, UK \nhttp://www.lfix.co.uk/oliver\nGPG: 1024D/3E1D0C1C: CA12 09E0 E8D5 8870 5839 932A 614D 4C34 3E1D 0C1C\n ========================================\n \"For the Lord himself shall descend from heaven with a \n shout, with the voice of the archangel, and with the \n trump of God; and the dead in Christ shall rise first;\n Then we which are alive and remain shall be caught \n up together with them in the clouds, to meet the Lord \n in the air; and so shall we ever be with the Lord.\" \n I Thessalonians 4:16,17 \n\n", "msg_date": "21 Aug 2002 16:07:54 +0100", "msg_from": "Oliver Elphick <olly@lfix.co.uk>", "msg_from_op": true, "msg_subject": "Re: Dropping a schema" }, { "msg_contents": "Oliver Elphick <olly@lfix.co.uk> writes:\n> On Wed, 2002-08-21 at 15:02, Tom Lane wrote:\n>> Also, it seems the safest behavior to me. \"rmdir dir\" won't remove a\n>> nonempty directory; isn't that a pretty close analogy?\n\n> Not really, seeing that you can't say \"mkdir directory (containing these\n> files)\". An implicit cascade *inside* the schema seems an appropriate\n> parallel to \"CREATE SCHEMA ... (CREATE TABLE ...)\". After all, we don't\n> have to say \"DROP TABLE ... CASCADE\" because the table has rows in it!\n\nHm. I could see an argument for being willing to auto-drop stuff that\nhad been made that way (inside CREATE SCHEMA) but not stuff that was\nmade by separate commands. But the spec doesn't seem to make any such\ndistinction: RESTRICT is RESTRICT. In any case, I like the behavior as\nit is, so I'm not gonna go out of my way to change it...\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 12:12:31 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Dropping a schema " } ]
[ { "msg_contents": "\n> > Hmm, \"any\" would sound like it is the same as opaque. Would \"any\" really be\n> > all allowed types ? I think we would want to eliminate that \n> altogether.\n> \n> Do you plan on eliminating the COUNT() aggregate, then?\n\nAh, you want it for aggbasetype in pg_aggregate, I did not \nsee that.\n\nHow could we then disallow it's use in other context ? Seems\nif there is no restriction, \"any\" will be exactly as prone to\n\"wrong use\" as opaque was.\n\nMay be a plan could be to leave opaque, but throw a notice\nwhen it is used in a create stmt, like:\nNOTICE: the use of type OPAQUE should be avoided where possible \n\nAndreas\n", "msg_date": "Wed, 21 Aug 2002 10:02:21 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> How could we then disallow it's use in other context ? Seems\n> if there is no restriction, \"any\" will be exactly as prone to\n> \"wrong use\" as opaque was.\n\nWell, you can always shoot yourself in the foot by creating a C function\nthat misinterprets its input. I'm not here to prevent that. But it\nwon't be easy to make a crashable function without superuser privileges,\nbecause all the PL languages will reject function definitions that use\ntype ANY as an argument or result (ditto the other pseudotypes, except\nmaybe VOID).\n\n> May be a plan could be to leave opaque, but throw a notice\n> when it is used in a create stmt, like:\n> NOTICE: the use of type OPAQUE should be avoided where possible \n\nRight, that's exactly the plan.\n\nActually, I think we can tighten the use of OPAQUE quite a bit, too.\nThe only supported uses will be (a) as an argument or result of a\ndatatype's I/O function, (b) as the result of a trigger function.\nSince I/O functions have to be coded in C anyway, we can disallow\nOPAQUE as an argument type of PL functions without losing any backwards\ncompatibility. Furthermore, *we do not have to treat OPAQUE as meaning\nANY*. It can become a pseudo-type that's not coercion-compatible to\nanything else, thus preventing any direct SQL calls of either I/O\nfunctions or triggers that are declared in the old style. There are\nstill some holes, for example you could do\n\tselect old_input_function(old_trigger_function());\nand the type system wouldn't complain. But it's a lot more nearly\nwatertight than before, even for functions declared with OPAQUE.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 09:46:16 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " } ]
[ { "msg_contents": "Has anyone made a decision about the SET LOCAL needing to be changed to SET\nTRANSATION for SQL compatibiltity?\n\nChris\n\n", "msg_date": "Wed, 21 Aug 2002 16:18:12 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "7.3 TODO" }, { "msg_contents": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au> writes:\n> Has anyone made a decision about the SET LOCAL needing to be changed to SET\n> TRANSATION for SQL compatibiltity?\n\nI think we had decided not to; IIRC the argument that spelling it\nTRANSACTION would be more spec-compatible looked bogus on closer\ninspection.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 10:10:57 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: 7.3 TODO " } ]
[ { "msg_contents": "Hi guys,\n\nSir Mordred seems okay, and is happy to help us out as long as we give\nhim credit where it's due.\n\nCan't see where that would be a problem with anyone here, as it's\ntotally in line with how we work, so am going to say yes to him up\nfront.\n\n:-)\n\nRegards and best wishes,\n\nJustin Clift\n\n\n-------- Original Message --------\nSubject: Re: Multiple vulnerabilities in PostgreSQL\nDate: Wed, 21 Aug 2002 08:12:15 +0000\nFrom: Sir Mordred The Traitor <mordred@s-mail.com>\nTo: justin@postgresql.org\n\nHi Justin.\n\n>Something that is concerning us though, is that whilst one of these bugs\n>was known and on our \"to fix\" list, there are some that were not known\n>and you're not notifying us up front so we can fix them before details\n>are publicly released.\n\nThere is no reason to be concerned really.\nWhile a bastard like me do stupid things sometimes, from now i will be\nworking with you guys.\nI'll be posting to pgsql-hackers@postgresql.org, and believe me, you\nwill\nenough have time for fixing.\nAfter fixing, you will release an advisory and give me a credit. That\nwill\nbe enough for me.\nIf that okay for you, let my know.\n\nBest regards.\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n", "msg_date": "Wed, 21 Aug 2002 18:25:33 +1000", "msg_from": "Justin Clift <justin@postgresql.org>", "msg_from_op": true, "msg_subject": "[Fwd: Re: Multiple vulnerabilities in PostgreSQL]" } ]
[ { "msg_contents": "Hi Sir Mordred,\n\nForwarded your email on to that same \"PostgreSQL Hackers\" mailing list\n(didn't seem to be anything confidential in it), just to let everyone\nknow that things will be ok from here on, etc.\n\nNo-one would generally even think to take credit for your work, as the\npeople in our community are the decent up-front kind of folk, and we\nwelcome your assistance and expertise in helping us find the\nvulnerabilities in PostgreSQL.\n\nSo, yep, it's all cool with us.\n\n:-)\n\nRegards and best wishes,\n\nJustin Clift\n\n\nSir Mordred The Traitor wrote:\n> \n> Hi Justin.\n> \n> >Something that is concerning us though, is that whilst one of these bugs\n> >was known and on our \"to fix\" list, there are some that were not known\n> >and you're not notifying us up front so we can fix them before details\n> >are publicly released.\n> \n> There is no reason to be concerned really.\n> While a bastard like me do stupid things sometimes, from now i will be\n> working with you guys.\n> I'll be posting to pgsql-hackers@postgresql.org, and believe me, you will\n> enough have time for fixing.\n> After fixing, you will release an advisory and give me a credit. That will\n> be enough for me.\n> If that okay for you, let my know.\n> \n> Best regards.\n> \n> ________________________________________________________________________\n> This letter has been delivered unencrypted. We'd like to remind you that\n> the full protection of e-mail correspondence is provided by S-mail\n> encryption mechanisms if only both, Sender and Recipient use S-mail.\n> Register at S-mail.com: http://www.s-mail.com/inf/en\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, 21 Aug 2002 18:30:18 +1000", "msg_from": "Justin Clift <justin@postgresql.org>", "msg_from_op": true, "msg_subject": "Re: Multiple vulnerabilities in PostgreSQL" } ]
[ { "msg_contents": "What's with this?\n\ntemplate1=# select current_timestamp(0);\n timestamptz\n------------------------\n 2002-08-21 16:39:40+08\n(1 row)\n\ntemplate1=# set time zone 'Australia/Sydney';\nSET\ntemplate1=# select current_timestamp(0);\n timestamptz\n------------------------\n 2002-08-21 18:39:49+10\n(1 row)\n\ntemplate1=# select current_timestamp(0) at time zone 'Australia/Sydney';\nERROR: Time zone 'australia/sydney' not recognized\ntemplate1=# select current_timestamp(0) at time zone 'AEST';\n timezone\n---------------------\n 2002-08-21 18:40:07\n(1 row)\n\nShouldn't the textual version of the time zone work using 'at time zone' as\nwell as 'set time zone'?\n\nAnd also, why does the column name change from timestamptz to timezone?\nAnyway, shouldn't it in fact be current_timestamp?\n\nChris\n\n", "msg_date": "Wed, 21 Aug 2002 16:41:50 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "AT TIME ZONE bug in CVS?" }, { "msg_contents": "> template1=# select current_timestamp(0) at time zone 'Australia/Sydney';\n> ERROR: Time zone 'australia/sydney' not recognized\n\nThe input is done using an internal lookup, not your system's time zone\ndatabase. Much faster; setting time zone variables for every input will\nbe substantially slower (though I haven't measured how much, it will\ninvolve opening files etc etc).\n\n> And also, why does the column name change from timestamptz to timezone?\n> Anyway, shouldn't it in fact be current_timestamp?\n\nThe feature is implemented as a function call to timezone(), which\nreturns a string. If it stayed a timestamp or something like that the\ntime zone can not be \"frozen\" through the formatting/output process.\n\n - Thomas\n", "msg_date": "Wed, 21 Aug 2002 08:59:44 -0700", "msg_from": "Thomas Lockhart <lockhart@fourpalms.org>", "msg_from_op": false, "msg_subject": "Re: AT TIME ZONE bug in CVS?" } ]
[ { "msg_contents": "On FreeBSD/Alpha, CVS gives:\n\ngmake[4]: Entering directory `/home/chriskl/pgsql-head/src/backend/parser'\ngmake[4]: Nothing to be done for `all'.\ngmake[4]: Leaving directory `/home/chriskl/pgsql-head/src/backend/parser'\ngcc -pipe -O -g -Wall -Wmissing-prototypes -Wmissing-declarations -I../../..\n/src/interfaces/libpq -I../../../src/include -c -o common.o common.c -MMD\nIn file included from common.c:21:\npg_backup_archiver.h:168: syntax error before `off_t'\ngmake[3]: *** [common.o] Error 1\ngmake[3]: Leaving directory `/home/chriskl/pgsql-head/src/bin/pg_dump'\ngmake[2]: *** [all] Error 2\ngmake[2]: Leaving directory `/home/chriskl/pgsql-head/src/bin'\ngmake[1]: *** [all] Error 2\ngmake[1]: Leaving directory `/home/chriskl/pgsql-head/src'\ngmake: *** [all] Error 2\n\nChris\n\n", "msg_date": "Wed, 21 Aug 2002 16:59:47 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "CVS broken - large file support?" }, { "msg_contents": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au> writes:\n> On FreeBSD/Alpha, CVS gives [trouble]\n\nI'm currently having to use \"configure --disable-largefile\" on HPUX;\nlooks like you'll have to do the same until Peter finishes ironing out\nthe wrinkles with autoconfiguring largefile support. It would be\nhelpful if you'd poke into your system headers and find out (a) can you\ndo large files at all, and if so (b) what is the correct magic\ncombination of #defines for your system.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 10:14:40 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: CVS broken - large file support? " }, { "msg_contents": "Christopher Kings-Lynne writes:\n\n> On FreeBSD/Alpha, CVS gives:\n\n> pg_backup_archiver.h:168: syntax error before `off_t'\n\nI have added sys/types.h, so off_t should now be available.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Wed, 21 Aug 2002 19:25:37 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: CVS broken - large file support?" }, { "msg_contents": "\nPeter, did you go around and remove sys/types.h from all the include\nfiles now that it is in c.h?\n\n---------------------------------------------------------------------------\n\nPeter Eisentraut wrote:\n> Christopher Kings-Lynne writes:\n> \n> > On FreeBSD/Alpha, CVS gives:\n> \n> > pg_backup_archiver.h:168: syntax error before `off_t'\n> \n> I have added sys/types.h, so off_t should now be available.\n> \n> -- \n> Peter Eisentraut peter_e@gmx.net\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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 13:30:45 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: CVS broken - large file support?" }, { "msg_contents": "> > On FreeBSD/Alpha, CVS gives [trouble]\n> \n> I'm currently having to use \"configure --disable-largefile\" on HPUX;\n> looks like you'll have to do the same until Peter finishes ironing out\n> the wrinkles with autoconfiguring largefile support. It would be\n> helpful if you'd poke into your system headers and find out (a) can you\n> do large files at all, and if so (b) what is the correct magic\n> combination of #defines for your system.\n\nFreeBSD 5.0-CURRENT, gcc 3.1 and 3.2 both have this same problem even\nwith largefile disabled. Help? -sc\n\ngmake[4]: Leaving directory `/usr/home/sean/open_source/postgresql/pgsql/src/backend/parser'\ngcc -pipe -Wall -Wmissing-prototypes -Wmissing-declarations -I../../../src/interfaces/libpq -I../../../src/include -c -o pg_dump.o pg_dump.c\ngcc -pipe -Wall -Wmissing-prototypes -Wmissing-declarations -I../../../src/interfaces/libpq -I../../../src/include -c -o common.o common.c\nIn file included from common.c:21:\npg_backup_archiver.h:168: syntax error before \"off_t\"\ngmake[3]: *** [common.o] Error 1\ngmake[3]: Leaving directory `/usr/home/sean/open_source/postgresql/pgsql/src/bin/pg_dump'\ngmake[2]: *** [all] Error 2\ngmake[2]: Leaving directory `/usr/home/sean/open_source/postgresql/pgsql/src/bin'\ngmake[1]: *** [all] Error 2\ngmake[1]: Leaving directory `/usr/home/sean/open_source/postgresql/pgsql/src'\ngmake: *** [all] Error 2\n*** Error code 2\n\nStop in /usr/home/sean/open_source/postgresql/pgsql.\n\n\n-- \nSean Chittenden\n", "msg_date": "Wed, 21 Aug 2002 12:26:26 -0700", "msg_from": "Sean Chittenden <sean@chittenden.org>", "msg_from_op": false, "msg_subject": "Re: CVS broken - large file support?" }, { "msg_contents": "> > > On FreeBSD/Alpha, CVS gives [trouble]\n> > \n> > I'm currently having to use \"configure --disable-largefile\" on HPUX;\n> > looks like you'll have to do the same until Peter finishes ironing out\n> > the wrinkles with autoconfiguring largefile support. It would be\n> > helpful if you'd poke into your system headers and find out (a) can you\n> > do large files at all, and if so (b) what is the correct magic\n> > combination of #defines for your system.\n> \n> FreeBSD 5.0-CURRENT, gcc 3.1 and 3.2 both have this same problem even\n> with largefile disabled. Help? -sc\n\nFor those interested and with commit powers, including sys/types.h in\npg_backup_archiver.h fixes this build problem. -sc\n\n> gcc -pipe -Wall -Wmissing-prototypes -Wmissing-declarations -I../../../src/interfaces/libpq -I../../../src/include -c -o common.o common.c\n> In file included from common.c:21:\n> pg_backup_archiver.h:168: syntax error before \"off_t\"\n> gmake[3]: *** [common.o] Error 1\n\n\n-- \nSean Chittenden", "msg_date": "Wed, 21 Aug 2002 16:06:41 -0700", "msg_from": "Sean Chittenden <sean@chittenden.org>", "msg_from_op": false, "msg_subject": "Re: CVS broken - large file support?" }, { "msg_contents": "> > > > On FreeBSD/Alpha, CVS gives [trouble]\n> > > \n> > > I'm currently having to use \"configure --disable-largefile\" on HPUX;\n> > > looks like you'll have to do the same until Peter finishes ironing out\n> > > the wrinkles with autoconfiguring largefile support. It would be\n> > > helpful if you'd poke into your system headers and find out (a) can you\n> > > do large files at all, and if so (b) what is the correct magic\n> > > combination of #defines for your system.\n> > \n> > FreeBSD 5.0-CURRENT, gcc 3.1 and 3.2 both have this same problem even\n> > with largefile disabled. Help? -sc\n> \n> For those interested and with commit powers, including sys/types.h in\n> pg_backup_archiver.h fixes this build problem. -sc\n> \n> > gcc -pipe -Wall -Wmissing-prototypes -Wmissing-declarations -I../../../src/interfaces/libpq -I../../../src/include -c -o common.o common.c\n> > In file included from common.c:21:\n> > pg_backup_archiver.h:168: syntax error before \"off_t\"\n> > gmake[3]: *** [common.o] Error 1\n\n'nother trivial type-o/patch that gets building working with\n--disable-largefile. -sc\n\n\n-- \nSean Chittenden", "msg_date": "Wed, 21 Aug 2002 19:14:07 -0700", "msg_from": "Sean Chittenden <sean@chittenden.org>", "msg_from_op": false, "msg_subject": "Re: CVS broken - large file support?" }, { "msg_contents": "\nYea, that was my booboo, fixed now.\n\n---------------------------------------------------------------------------\n\nSean Chittenden wrote:\n> > > > > On FreeBSD/Alpha, CVS gives [trouble]\n> > > > \n> > > > I'm currently having to use \"configure --disable-largefile\" on HPUX;\n> > > > looks like you'll have to do the same until Peter finishes ironing out\n> > > > the wrinkles with autoconfiguring largefile support. It would be\n> > > > helpful if you'd poke into your system headers and find out (a) can you\n> > > > do large files at all, and if so (b) what is the correct magic\n> > > > combination of #defines for your system.\n> > > \n> > > FreeBSD 5.0-CURRENT, gcc 3.1 and 3.2 both have this same problem even\n> > > with largefile disabled. Help? -sc\n> > \n> > For those interested and with commit powers, including sys/types.h in\n> > pg_backup_archiver.h fixes this build problem. -sc\n> > \n> > > gcc -pipe -Wall -Wmissing-prototypes -Wmissing-declarations -I../../../src/interfaces/libpq -I../../../src/include -c -o common.o common.c\n> > > In file included from common.c:21:\n> > > pg_backup_archiver.h:168: syntax error before \"off_t\"\n> > > gmake[3]: *** [common.o] Error 1\n> \n> 'nother trivial type-o/patch that gets building working with\n> --disable-largefile. -sc\n> \n> \n> -- \n> Sean Chittenden\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 22:29:42 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: CVS broken - large file support?" }, { "msg_contents": "\nPeter, I have received no reply to this question.\n\n---------------------------------------------------------------------------\n\nBruce Momjian wrote:\n> \n> Peter, did you go around and remove sys/types.h from all the include\n> files now that it is in c.h?\n> \n> ---------------------------------------------------------------------------\n> \n> Peter Eisentraut wrote:\n> > Christopher Kings-Lynne writes:\n> > \n> > > On FreeBSD/Alpha, CVS gives:\n> > \n> > > pg_backup_archiver.h:168: syntax error before `off_t'\n> > \n> > I have added sys/types.h, so off_t should now be available.\n> > \n> > -- \n> > Peter Eisentraut peter_e@gmx.net\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) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 17:55:30 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: CVS broken - large file support?" }, { "msg_contents": "\nDone and committed with clean compile.\n\n---------------------------------------------------------------------------\n\nBruce Momjian wrote:\n> \n> Peter, did you go around and remove sys/types.h from all the include\n> files now that it is in c.h?\n> \n> ---------------------------------------------------------------------------\n> \n> Peter Eisentraut wrote:\n> > Christopher Kings-Lynne writes:\n> > \n> > > On FreeBSD/Alpha, CVS gives:\n> > \n> > > pg_backup_archiver.h:168: syntax error before `off_t'\n> > \n> > I have added sys/types.h, so off_t should now be available.\n> > \n> > -- \n> > Peter Eisentraut peter_e@gmx.net\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) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sun, 1 Sep 2002 22:34:55 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: CVS broken - large file support?" } ]
[ { "msg_contents": "Hi.\nThis post certainly contains no valuable information,\nbut i feel i should clarify some points.\n\n1) I like postgresql and i worked with it for a long time.\n2) Solution like \"killall -9 postmaster\" was just a joke.\n3) ...Hm..i forgot...maybe later ...\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n", "msg_date": "Wed, 21 Aug 2002 14:42:04 +0000", "msg_from": "Sir Mordred The Traitor <mordred@s-mail.com>", "msg_from_op": true, "msg_subject": "i'll promise, i'll be polite :-)" }, { "msg_contents": "On Wednesday 21 August 2002 10:42 am, Sir Mordred The Traitor wrote:\n> Hi.\n> This post certainly contains no valuable information,\n> but i feel i should clarify some points.\n\n> 1) I like postgresql and i worked with it for a long time.\n> 2) Solution like \"killall -9 postmaster\" was just a joke.\n\nFirst, it is good you found the problems.\n\nHowever, you really should have notified the developers first. And giving \njoke advice as a solution isn't the height of responsibility. I have taken \nthe liberty of addressing that issue on BugTraq, if it passes moderation.\n\nHaving said that, I personally look forward to you telling us about problems \nyou find, and helping audit this codebase. After all, if you found it \nsomeone else can too.\n\nBut I hope you don't suffer too much for the repercussions that will \nundoubtedly be voiced here -- you did, afterall, make somewhat of a 'splash' \nwhen diving in. :-)\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Wed, 21 Aug 2002 11:11:24 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: i'll promise, i'll be polite :-)" } ]
[ { "msg_contents": "Seems like this one was lost or was filtered out...\n\n//@(#)Mordred Labs advisory 0x0002\n\nRelease data: 19/08/02\nName: Buffer overflow in PostgreSQL\nVersions affected: all versions\nRisk: high\n\n--[ Description:\nThere exists a buffer overflow in a SET TIME ZONE command, that\nallows an attacker to execute malicious code.\n\n--[ Details:\nUpon executing the SET TIME ZONE 'STRING' command, parse_timezone()\nfunction is invoked,\nwhich will overwrite a static buffer tzbuf with the supplied string.\nLook at the src/backend/commands/variable.c if you need something to laugh\nat.\n\n--[ How to reproduce:\npsql> SET TIMEZONE to 'XXXXXX...very long string...XXXXX'\n...\nNOTICE: Buffer Leak: [27191] (freeNext=0, freePrev=0, rel=0/0, blockNum=0,\nflags=0x0, refcount=0 128)\nNOTICE: Buffer Leak: [27192] (freeNext=0, freePrev=0, rel=0/0, blockNum=0,\nflags=0x0, refcount=0 1249)\nNOTICE: Buffer Leak: [27193] (freeNext=0, freePrev=0, rel=0/0, blockNum=0,\nflags=0x0, refcount=0 1651799137)\nNOTICE: Buffer Leak: [27194] (freeNext=0, freePrev=0, rel=0/0, blockNum=0,\nflags=0x0, refcount=0 1818326649)\n...\npqReadData() -- backend closed the channel unexpectedly.\n This probably means the backend terminated abnormally\n before or while processing the request.\nThe connection to the server was lost. Attempting reset: Failed.\n\n--[ Solution:\nJust wait...\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n", "msg_date": "Wed, 21 Aug 2002 14:45:42 +0000", "msg_from": "Sir Mordred The Traitor <mordred@s-mail.com>", "msg_from_op": true, "msg_subject": "@(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "Sir Mordred The Traitor <mordred@s-mail.com> writes:\n> There exists a buffer overflow in a SET TIME ZONE command, that\n> allows an attacker to execute malicious code.\n\nHere's a patch for the problem. I also fixed some other potential\nbuffer overruns nearby, and added a little paranoia to another routine\nthat uses a statically sized buffer.\n\nThanks for the report.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "21 Aug 2002 12:13:19 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> Sir Mordred The Traitor <mordred@s-mail.com> writes:\n> > There exists a buffer overflow in a SET TIME ZONE command, that\n> > allows an attacker to execute malicious code.\n> \n> Here's a patch for the problem. I also fixed some other potential\n> buffer overruns nearby, and added a little paranoia to another routine\n> that uses a statically sized buffer.\n\nThe handling of the TZ environmental variable is subject to a buffer\noverrun. To see the problem, try:\n\nexport TZ=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\n xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\npostmaster -D /foo/bar&\npsql\n\nYou get:\n\nNOTICE: Buffer Leak: [26914] (freeNext=0, freePrev=0, rel=0/0, blockNum=0, flags=0x0, refcount=0 1)\n[ lots more NOTICEs ]\npsql: server closed the connection unexpectedly\n\tThis probably means the server terminated abnormally\n\tbefore or while processing the request.\n\nA revised patch is attached that fixes the problem.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "21 Aug 2002 13:41:32 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://candle.pha.pa.us/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nNeil Conway wrote:\n> Sir Mordred The Traitor <mordred@s-mail.com> writes:\n> > There exists a buffer overflow in a SET TIME ZONE command, that\n> > allows an attacker to execute malicious code.\n> \n> Here's a patch for the problem. I also fixed some other potential\n> buffer overruns nearby, and added a little paranoia to another routine\n> that uses a statically sized buffer.\n> \n> Thanks for the report.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n[ Attachment, skipping... ]\n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 14:03:15 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://candle.pha.pa.us/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nNeil Conway wrote:\n> Neil Conway <neilc@samurai.com> writes:\n> > Sir Mordred The Traitor <mordred@s-mail.com> writes:\n> > > There exists a buffer overflow in a SET TIME ZONE command, that\n> > > allows an attacker to execute malicious code.\n> > \n> > Here's a patch for the problem. I also fixed some other potential\n> > buffer overruns nearby, and added a little paranoia to another routine\n> > that uses a statically sized buffer.\n> \n> The handling of the TZ environmental variable is subject to a buffer\n> overrun. To see the problem, try:\n> \n> export TZ=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\n> xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\n> postmaster -D /foo/bar&\n> psql\n> \n> You get:\n> \n> NOTICE: Buffer Leak: [26914] (freeNext=0, freePrev=0, rel=0/0, blockNum=0, flags=0x0, refcount=0 1)\n> [ lots more NOTICEs ]\n> psql: server closed the connection unexpectedly\n> \tThis probably means the server terminated abnormally\n> \tbefore or while processing the request.\n> \n> A revised patch is attached that fixes the problem.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n[ Attachment, skipping... ]\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 14:03:20 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> The handling of the TZ environmental variable is subject to a buffer\n> overrun.\n\nThis problem is long gone in current sources, no?\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 00:04:13 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL " }, { "msg_contents": "Tom Lane wrote:\n> Neil Conway <neilc@samurai.com> writes:\n> > The handling of the TZ environmental variable is subject to a buffer\n> > overrun.\n> \n> This problem is long gone in current sources, no?\n> \n\nThe patch looks like it does prevent some problems.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 00:12:59 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Tom Lane wrote:\n> > Neil Conway <neilc@samurai.com> writes:\n> > > The handling of the TZ environmental variable is subject to a buffer\n> > > overrun.\n> > \n> > This problem is long gone in current sources, no?\n\nI quickly tested current sources, and it seems the bug is fixed. I\nonly fixed it to begin with because I saw it while fixing the reported\nproblem.\n\n> The patch looks like it does prevent some problems.\n\nYes: namely, it fixes the bug in REL7_2_STABLE.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "22 Aug 2002 00:56:59 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "\nWhat would you like done with the patch you submitted?\n\n---------------------------------------------------------------------------\n\nNeil Conway wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Tom Lane wrote:\n> > > Neil Conway <neilc@samurai.com> writes:\n> > > > The handling of the TZ environmental variable is subject to a buffer\n> > > > overrun.\n> > > \n> > > This problem is long gone in current sources, no?\n> \n> I quickly tested current sources, and it seems the bug is fixed. I\n> only fixed it to begin with because I saw it while fixing the reported\n> problem.\n> \n> > The patch looks like it does prevent some problems.\n> \n> Yes: namely, it fixes the bug in REL7_2_STABLE.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n> \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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 00:59:27 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> What would you like done with the patch you submitted?\n\nI'd like to see it applied to CVS HEAD and REL7_2_STABLE.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "22 Aug 2002 01:01:04 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > What would you like done with the patch you submitted?\n> \n> I'd like to see it applied to CVS HEAD and REL7_2_STABLE.\n\nUh, sorry -- wrote that without thinking. I'd like to see the patch\napplied to REL7_2_STABLE. I'll prepare a revised patch for CVS HEAD.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "22 Aug 2002 01:05:23 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> Neil Conway <neilc@samurai.com> writes:\n>> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> What would you like done with the patch you submitted?\n>> \n>> I'd like to see it applied to CVS HEAD and REL7_2_STABLE.\n\n> Uh, sorry -- wrote that without thinking. I'd like to see the patch\n> applied to REL7_2_STABLE. I'll prepare a revised patch for CVS HEAD.\n\nI'm pretty certain that no such patch is necessary for HEAD.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 01:08:44 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL " }, { "msg_contents": "\nOK, I have applied this to 7.2.X.\n\nI have applied the lpad/rpad/repeat patch to CVS head. I assume you do\nnot want the others applied to CVS head because the fixes are already present.\n\n---------------------------------------------------------------------------\n\nNeil Conway wrote:\n> Neil Conway <neilc@samurai.com> writes:\n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > What would you like done with the patch you submitted?\n> > \n> > I'd like to see it applied to CVS HEAD and REL7_2_STABLE.\n> \n> Uh, sorry -- wrote that without thinking. I'd like to see the patch\n> applied to REL7_2_STABLE. I'll prepare a revised patch for CVS HEAD.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 01:37:03 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0002: Buffer overflow in PostgreSQL" }, { "msg_contents": "\nAre we all caught up now on the known bugs/fixes? Would it be reasonably\nsafe to do up a quick v7.2.2 Security Fix Release tomorrow afternoon?\n\nOn Thu, 22 Aug 2002, Bruce Momjian wrote:\n\n>\n> OK, I have applied this to 7.2.X.\n>\n> I have applied the lpad/rpad/repeat patch to CVS head. I assume you do\n> not want the others applied to CVS head because the fixes are already present.\n>\n> ---------------------------------------------------------------------------\n>\n> Neil Conway wrote:\n> > Neil Conway <neilc@samurai.com> writes:\n> > > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > > What would you like done with the patch you submitted?\n> > >\n> > > I'd like to see it applied to CVS HEAD and REL7_2_STABLE.\n> >\n> > Uh, sorry -- wrote that without thinking. I'd like to see the patch\n> > applied to REL7_2_STABLE. I'll prepare a revised patch for CVS HEAD.\n> >\n> > Cheers,\n> >\n> > Neil\n> >\n> > --\n> > Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\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) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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", "msg_date": "Thu, 22 Aug 2002 03:36:54 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "\"Marc G. Fournier\" <scrappy@hub.org> writes:\n> Are we all caught up now on the known bugs/fixes? Would it be reasonably\n> safe to do up a quick v7.2.2 Security Fix Release tomorrow afternoon?\n\nStill need to do a release-history entry and version-stamp, of course.\nI will do a diff of the 7.2 branch tree this afternoon and make sure\nthe code looks good.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 09:43:05 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...) " }, { "msg_contents": "\nI was going to ask that too. I need to make up a list of 7.2.2 changes,\nand there are quite a number of them. I will get the list together today.\n\n---------------------------------------------------------------------------\n\nMarc G. Fournier wrote:\n> \n> Are we all caught up now on the known bugs/fixes? Would it be reasonably\n> safe to do up a quick v7.2.2 Security Fix Release tomorrow afternoon?\n> \n> On Thu, 22 Aug 2002, Bruce Momjian wrote:\n> \n> >\n> > OK, I have applied this to 7.2.X.\n> >\n> > I have applied the lpad/rpad/repeat patch to CVS head. I assume you do\n> > not want the others applied to CVS head because the fixes are already present.\n> >\n> > ---------------------------------------------------------------------------\n> >\n> > Neil Conway wrote:\n> > > Neil Conway <neilc@samurai.com> writes:\n> > > > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > > > What would you like done with the patch you submitted?\n> > > >\n> > > > I'd like to see it applied to CVS HEAD and REL7_2_STABLE.\n> > >\n> > > Uh, sorry -- wrote that without thinking. I'd like to see the patch\n> > > applied to REL7_2_STABLE. I'll prepare a revised patch for CVS HEAD.\n> > >\n> > > Cheers,\n> > >\n> > > Neil\n> > >\n> > > --\n> > > Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\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) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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> ---------------------------(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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 10:26:36 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "\nThanks, Tom. That is a big help. I will do the release history and\nversion stamping.\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > Are we all caught up now on the known bugs/fixes? Would it be reasonably\n> > safe to do up a quick v7.2.2 Security Fix Release tomorrow afternoon?\n> \n> Still need to do a release-history entry and version-stamp, of course.\n> I will do a diff of the 7.2 branch tree this afternoon and make sure\n> the code looks good.\n> \n> \t\t\tregards, tom lane\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-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 10:27:16 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "Hi everyone,\n\nIt probably makes sense to wait about a week until releasing 7.2.2, even\nif we get assembled anything else that is needed. Sir Mordred appears\nto be taking a look through our 7.2.x code about now and that probably\nmeans there's a good chance we'll hear of other patches that will need\nto be applied to the 7.2.x branch.\n\n:-)\n\nRegards and best wishes,\n\nJustin Clift\n\n\nOleg Bartunov wrote:\n> \n> Tom,\n> \n> I think it's worth to include patch for query planner which\n> fixes using indices with predicates for join plans. We found it's\n> quite useful.\n> \n> http://fts.postgresql.org/db/mw/msg.html?mid=1018153\n> \n> Oleg\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": "Fri, 23 Aug 2002 00:32:45 +1000", "msg_from": "Justin Clift <justin@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "Justin Clift <justin@postgresql.org> writes:\n> It probably makes sense to wait about a week until releasing 7.2.2, even\n> if we get assembled anything else that is needed.\n\nI think we should go ahead and push it out; by the end of next week\nwe'll be trying to wrap 7.3 beta, and the confusion factor for pushing\nout two releases at the same time will be much too high.\n\nI think it is fairly unlikely that we will find anything else in the\nnext week that is exploitable indirectly through a web-app in the same\nway that the date buffer overrun bug could be. Most of the sorts of\nbugs that I'm expecting to hear about will require being able to issue\nSQL commands --- and if someone can issue arbitrary SQL commands, there\nare plenty of ways to create a DOS situation.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 10:54:26 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...) " }, { "msg_contents": "Tom,\n\nI think it's worth to include patch for query planner which\nfixes using indices with predicates for join plans. We found it's\nquite useful.\n\nhttp://fts.postgresql.org/db/mw/msg.html?mid=1018153\n\n\tOleg\nOn Thu, 22 Aug 2002, Tom Lane wrote:\n\n> \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > Are we all caught up now on the known bugs/fixes? Would it be reasonably\n> > safe to do up a quick v7.2.2 Security Fix Release tomorrow afternoon?\n>\n> Still need to do a release-history entry and version-stamp, of course.\n> I will do a diff of the 7.2 branch tree this afternoon and make sure\n> the code looks good.\n>\n> \t\t\tregards, tom lane\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\tRegards,\n\t\tOleg\n_____________________________________________________________\nOleg Bartunov, sci.researcher, hostmaster of AstroNet,\nSternberg Astronomical Institute, Moscow University (Russia)\nInternet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/\nphone: +007(095)939-16-83, +007(095)939-23-83\n\n", "msg_date": "Thu, 22 Aug 2002 18:14:24 +0300 (GMT)", "msg_from": "Oleg Bartunov <oleg@sai.msu.su>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...) " }, { "msg_contents": "On Thu, 22 Aug 2002, Tom Lane wrote:\n\n> Justin Clift <justin@postgresql.org> writes:\n> > It probably makes sense to wait about a week until releasing 7.2.2, even\n> > if we get assembled anything else that is needed.\n>\n> I think we should go ahead and push it out; by the end of next week\n> we'll be trying to wrap 7.3 beta, and the confusion factor for pushing\n> out two releases at the same time will be much too high.\n>\n> I think it is fairly unlikely that we will find anything else in the\n> next week that is exploitable indirectly through a web-app in the same\n> way that the date buffer overrun bug could be. Most of the sorts of\n> bugs that I'm expecting to hear about will require being able to issue\n> SQL commands --- and if someone can issue arbitrary SQL commands, there\n> are plenty of ways to create a DOS situation.\n\nAnd, worse comes to worse, we *can* issue a v7.2.3 if further security\nissues are found before v7.3 is fully released ...\n\n\n", "msg_date": "Thu, 22 Aug 2002 15:06:22 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...) " }, { "msg_contents": "\nDone, but Tom.\n\n---------------------------------------------------------------------------\n\nOleg Bartunov wrote:\n> Tom,\n> \n> I think it's worth to include patch for query planner which\n> fixes using indices with predicates for join plans. We found it's\n> quite useful.\n> \n> http://fts.postgresql.org/db/mw/msg.html?mid=1018153\n> \n> \tOleg\n> On Thu, 22 Aug 2002, Tom Lane wrote:\n> \n> > \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > > Are we all caught up now on the known bugs/fixes? Would it be reasonably\n> > > safe to do up a quick v7.2.2 Security Fix Release tomorrow afternoon?\n> >\n> > Still need to do a release-history entry and version-stamp, of course.\n> > I will do a diff of the 7.2 branch tree this afternoon and make sure\n> > the code looks good.\n> >\n> > \t\t\tregards, tom lane\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> \tRegards,\n> \t\tOleg\n> _____________________________________________________________\n> Oleg Bartunov, sci.researcher, hostmaster of AstroNet,\n> Sternberg Astronomical Institute, Moscow University (Russia)\n> Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/\n> phone: +007(095)939-16-83, +007(095)939-23-83\n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 15:31:51 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "On Thursday 22 August 2002 10:32 am, Justin Clift wrote:\n> It probably makes sense to wait about a week until releasing 7.2.2, even\n> if we get assembled anything else that is needed. Sir Mordred appears\n> to be taking a look through our 7.2.x code about now and that probably\n> means there's a good chance we'll hear of other patches that will need\n> to be applied to the 7.2.x branch.\n\nFWIW, it will be Monday at the earliest before I can dedicate the time to \nbuild a 7.2.2 release RPMset. If a preliminary tarball was available \nSaturday, I might could have an RPM uploaded Monday morning.\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Thu, 22 Aug 2002 16:36:29 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "I said:\n> Still need to do a release-history entry and version-stamp, of course.\n> I will do a diff of the 7.2 branch tree this afternoon and make sure\n> the code looks good.\n\nCode checked, regression tests run. Bruce, the ball's in your court ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 18:32:44 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...) " }, { "msg_contents": "Tom Lane wrote:\n> I said:\n> > Still need to do a release-history entry and version-stamp, of course.\n> > I will do a diff of the 7.2 branch tree this afternoon and make sure\n> > the code looks good.\n> \n> Code checked, regression tests run. Bruce, the ball's in your court ...\n\nOK, I am getting autoconf warnings with the 7.2.2 configure.in. This is\nwith autoconf 2.53. Do I need a different version for this release?\n\n---------------------------------------------------------------------------\n\n(3) autoconf\nconfigure.in:139: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:146: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:149: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:158: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:168: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:200: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:212: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:223: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:236: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:246: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:253: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:261: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:311: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:320: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:360: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:367: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:378: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:383: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:389: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:397: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:411: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:442: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:465: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:484: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:495: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:517: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:533: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:534: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:535: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:558: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:621: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:972: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\nconfigure.in:292: error: possibly undefined macro: AC_PROG_CC_WORKS\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 18:43:51 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "\nWe're still using 2.53 on the main server:\n\n%autoconf --version\nautoconf (GNU Autoconf) 2.53\nWritten by David J. MacKenzie and Akim Demaille.\n\nCopyright 2002 Free Software Foundation, Inc.\nThis is free software; see the source for copying conditions. There is NO\nwarranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n\nAnd I just autoconf'd after removing the odbc stuff for the HEAD branch ...\n\n\nOn Thu, 22 Aug 2002, Bruce Momjian wrote:\n\n> Tom Lane wrote:\n> > I said:\n> > > Still need to do a release-history entry and version-stamp, of course.\n> > > I will do a diff of the 7.2 branch tree this afternoon and make sure\n> > > the code looks good.\n> >\n> > Code checked, regression tests run. Bruce, the ball's in your court ...\n>\n> OK, I am getting autoconf warnings with the 7.2.2 configure.in. This is\n> with autoconf 2.53. Do I need a different version for this release?\n>\n> ---------------------------------------------------------------------------\n>\n> (3) autoconf\n> configure.in:139: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:146: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:149: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:158: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:168: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:200: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:212: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:223: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:236: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:246: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:253: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:261: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:311: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:320: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:360: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:367: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:378: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:383: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:389: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:397: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:411: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:442: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:465: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:484: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:495: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:517: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:533: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:534: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:535: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:558: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:621: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:972: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> configure.in:292: error: possibly undefined macro: AC_PROG_CC_WORKS\n>\n>\n> --\n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n>\n\n", "msg_date": "Thu, 22 Aug 2002 19:47:20 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "\nNo, the problem is with the 7.2.2 branch, REL7_2_STABLE.\n\n---------------------------------------------------------------------------\n\nMarc G. Fournier wrote:\n> \n> We're still using 2.53 on the main server:\n> \n> %autoconf --version\n> autoconf (GNU Autoconf) 2.53\n> Written by David J. MacKenzie and Akim Demaille.\n> \n> Copyright 2002 Free Software Foundation, Inc.\n> This is free software; see the source for copying conditions. There is NO\n> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n> \n> And I just autoconf'd after removing the odbc stuff for the HEAD branch ...\n> \n> \n> On Thu, 22 Aug 2002, Bruce Momjian wrote:\n> \n> > Tom Lane wrote:\n> > > I said:\n> > > > Still need to do a release-history entry and version-stamp, of course.\n> > > > I will do a diff of the 7.2 branch tree this afternoon and make sure\n> > > > the code looks good.\n> > >\n> > > Code checked, regression tests run. Bruce, the ball's in your court ...\n> >\n> > OK, I am getting autoconf warnings with the 7.2.2 configure.in. This is\n> > with autoconf 2.53. Do I need a different version for this release?\n> >\n> > ---------------------------------------------------------------------------\n> >\n> > (3) autoconf\n> > configure.in:139: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:146: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:149: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:158: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:168: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:200: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:212: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:223: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:236: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:246: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:253: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:261: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:311: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:320: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:360: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:367: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:378: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:383: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:389: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:397: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:411: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:442: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:465: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:484: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:495: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:517: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:533: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:534: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:535: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:558: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:621: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:972: /usr/contrib/bin/gm4: Non-numeric argument to built-in `divert'\n> > configure.in:292: error: possibly undefined macro: AC_PROG_CC_WORKS\n> >\n> >\n> > --\n> > Bruce Momjian | http://candle.pha.pa.us\n> > pgman@candle.pha.pa.us | (610) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n> >\n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 18:48:45 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> OK, I am getting autoconf warnings with the 7.2.2 configure.in. This is\n> with autoconf 2.53. Do I need a different version for this release?\n\n7.2 was built with autoconf 2.13. Simple answer: don't re-autoconf.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 18:50:01 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...) " }, { "msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > OK, I am getting autoconf warnings with the 7.2.2 configure.in. This is\n> > with autoconf 2.53. Do I need a different version for this release?\n> \n> 7.2 was built with autoconf 2.13. Simple answer: don't re-autoconf.\n\nGood idea.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 18:53:51 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "\nDone.\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> I said:\n> > Still need to do a release-history entry and version-stamp, of course.\n> > I will do a diff of the 7.2 branch tree this afternoon and make sure\n> > the code looks good.\n> \n> Code checked, regression tests run. Bruce, the ball's in your court ...\n> \n> \t\t\tregards, tom lane\n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 19:27:46 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Done.\n\nI've attached some improvements to HISTORY: fixed some obvious typos,\nadded info on Tom's partial index backpatch, and elaborated on the\nnature of the fixed security problems.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "22 Aug 2002 20:23:26 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Done.\n\nLooks good here, except I think you should accept Neil's edits of the\nhistory entries. Also, don't forget to forward-patch the 7.2.2 release\ninfo into the CVS tip version of release.sgml.\n\nIs it appropriate to back-patch the current state of the FAQ into 7.2.2?\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 21:22:21 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...) " }, { "msg_contents": "On Thu, 22 Aug 2002, Tom Lane wrote:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Done.\n>\n> Looks good here, except I think you should accept Neil's edits of the\n> history entries.\n\nAgreed, that's what I was waiting on :)\n\n\n", "msg_date": "Thu, 22 Aug 2002 22:27:44 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...) " }, { "msg_contents": "> On Thu, 22 Aug 2002, Tom Lane wrote:\n> \n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > Done.\n> >\n> > Looks good here, except I think you should accept Neil's edits of the\n> > history entries.\n> \n> Agreed, that's what I was waiting on :)\n\nDoing a test build of 7.2.2 on FreeBSD/Alpha atm...\n\nChris\n\n", "msg_date": "Fri, 23 Aug 2002 10:22:24 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...) " }, { "msg_contents": "\nPatch applied. Thanks.\n\n---------------------------------------------------------------------------\n\n\nNeil Conway wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Done.\n> \n> I've attached some improvements to HISTORY: fixed some obvious typos,\n> added info on Tom's partial index backpatch, and elaborated on the\n> nature of the fixed security problems.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n[ Attachment, skipping... ]\n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 22:32:28 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Done.\n> \n> Looks good here, except I think you should accept Neil's edits of the\n> history entries. Also, don't forget to forward-patch the 7.2.2 release\n> info into the CVS tip version of release.sgml.\n\nDone. I actually took a break from the computer for a few hours to\nclear my head.\n\n\n> Is it appropriate to back-patch the current state of the FAQ into 7.2.2?\n\nAlready done.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 22:33:03 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" }, { "msg_contents": "All 79 tests passed on FreeBSD/Alpha. Good to go :)\n\nChris\n\n> > On Thu, 22 Aug 2002, Tom Lane wrote:\n> > \n> > > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > > Done.\n> > >\n> > > Looks good here, except I think you should accept Neil's edits of the\n> > > history entries.\n> > \n> > Agreed, that's what I was waiting on :)\n> \n> Doing a test build of 7.2.2 on FreeBSD/Alpha atm...\n\n", "msg_date": "Fri, 23 Aug 2002 10:52:25 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...) " }, { "msg_contents": "Builds and runs fine under HP/Compaq Tru64 aka Digital Unix aka OSF/1\n(this is getting difficult...) version 4.0f/g using standard cc:\n\ntemplate1=# SELECT version();\n version\n----------------------------------------------------------------\n PostgreSQL 7.2.2 on alphaev56-dec-osf4.0g, compiled by cc -std\n\n$ gmake check\n\n======================\n All 79 tests passed.\n======================\n\nSame using GCC:\n\n======================\n All 79 tests passed.\n======================\n\n-- \nAlessio F. Bragadini\t\talessio@albourne.com\nAPL Financial Services\t\thttp://village.albourne.com\nNicosia, Cyprus\t\t \tphone: +357-22-755750\n\n\"It is more complicated than you think\"\n\t\t-- The Eighth Networking Truth from RFC 1925\n\n", "msg_date": "26 Aug 2002 17:21:04 +0300", "msg_from": "Alessio Bragadini <alessio@albourne.com>", "msg_from_op": false, "msg_subject": "Re: Release of v7.2.2 (Was: Re: @(#)Mordred Labs ad...)" } ]
[ { "msg_contents": "\n> > May be a plan could be to leave opaque, but throw a notice\n> > when it is used in a create stmt, like:\n> > NOTICE: the use of type OPAQUE should be avoided where possible \n> \n> Right, that's exactly the plan.\n> \n> Actually, I think we can tighten the use of OPAQUE quite a bit, too.\n> The only supported uses will be (a) as an argument or result of a\n> datatype's I/O function, (b) as the result of a trigger function.\n\nIn my paper I use C functions that take \"any tuple\".\nI do not yet see how I can do that without opaque if we don't \nhave a \"row\" but only a \"trigger\" type.\n\nAndreas\n", "msg_date": "Wed, 21 Aug 2002 18:10:48 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> In my paper I use C functions that take \"any tuple\".\n> I do not yet see how I can do that without opaque if we don't \n> have a \"row\" but only a \"trigger\" type.\n\nFor that you would have to use \"any\", at the moment. This would give\nyou the same amount of type safety you have with \"opaque\", ie, none.\n\nWe could talk about adding more pseudotypes that are geared to\nrequirements not existing in the standard set of functions ...\nbut I'm going to concentrate on those first.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 12:18:41 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " } ]
[ { "msg_contents": "\n> > In my paper I use C functions that take \"any tuple\".\n> > I do not yet see how I can do that without opaque if we don't \n> > have a \"row\" but only a \"trigger\" type.\n> \n> For that you would have to use \"any\", at the moment. This would give\n> you the same amount of type safety you have with \"opaque\", ie, none.\n\nI would have to use some pg_proc magic to make \"any\" appear there,\nsince the plan was to not make it visible at the sql level, no ?\nIf you do, you would also have to throw the notice for \"any\".\n\nAgain, I think leaving \"any\" be \"opaque\" and throwing the warning NOTICE\nwould be better. I do not think there is enough time to really see through\nthe implications of restricting opaque already in 7.3, (at least for me :-)\n\n> We could talk about adding more pseudotypes that are geared to\n> requirements not existing in the standard set of functions ...\n> but I'm going to concentrate on those first.\n\nYes, certainly a lot of work.\n\nAndreas\n", "msg_date": "Wed, 21 Aug 2002 18:32:48 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n>> For that you would have to use \"any\", at the moment. This would give\n>> you the same amount of type safety you have with \"opaque\", ie, none.\n\n> I would have to use some pg_proc magic to make \"any\" appear there,\n> since the plan was to not make it visible at the sql level, no ?\n\nHuh? It'll be perfectly visible.\n\nThere is one little problem with calling it ANY, it turns out: that word\nis fully reserved in our parser (and trying to make it less reserved\ncreates reduce/reduce conflicts). So unless we go back to \"anytype\"\nyou'd have to quote the name, eg\n\tcreate function foo(\"any\") returns ...\nI do prefer using \"any\" because that's what we have historically used\nin CREATE AGGREGATE, but maybe the keyword conflict will be too\nannoying.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 12:56:23 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " } ]
[ { "msg_contents": "\n> If we are going to delay beta, we should decide now, not at the end of\n> August, and the delay should be until the end of September. The big\n> question is whether we have enough material to warrant a delay.\n\nI think the \"implicit casts\" todo should be adressed soon, \nnot sure there is enough time for discussing this topic in 10 days ?\n\nAm I the only relic, that wants all implicit casts that don't cause\nloss of precision ?\n\nNice ones would imho be such involving numeric constants, like \n\tselect * from atab where int2col=2; -- use index on int2col\n\nAndreas\n", "msg_date": "Wed, 21 Aug 2002 18:41:54 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "delay beta ? (was: RE: @(#)Mordred Labs advisory 0x0003: Buffer\n\toverflow in)" } ]
[ { "msg_contents": "\n> >> For that you would have to use \"any\", at the moment. This would give\n> >> you the same amount of type safety you have with \"opaque\", \n> ie, none.\n> \n> > I would have to use some pg_proc magic to make \"any\" appear there,\n> > since the plan was to not make it visible at the sql level, no ?\n> \n> Huh? It'll be perfectly visible.\n\nI did not mean visible, I meant useable, like in\n\tcreate function xx(any) returns text ...;\n\nIf that is possible, what is the difference to opaque ?\n\nAndreas\n", "msg_date": "Wed, 21 Aug 2002 19:18:46 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> I did not mean visible, I meant useable, like in\n> \tcreate function xx(any) returns text ...;\n> If that is possible, what is the difference to opaque ?\n\n\"any\" will have the same behavior that \"opaque\" used to have, yes.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 21 Aug 2002 13:31:01 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " } ]
[ { "msg_contents": "\n> > I did not mean visible, I meant useable, like in\n> > \tcreate function xx(any) returns text ...;\n> > If that is possible, what is the difference to opaque ?\n> \n> \"any\" will have the same behavior that \"opaque\" used to have, yes.\n\nOk, now I vote, that you don't implement \"any\" and use \"opaque\".\nI don't think we want two types that do the same thing.\nIs it that you like the name \"any\" more than \"opaque\" ?\nI am confused.\n\nAndreas\n", "msg_date": "Wed, 21 Aug 2002 19:39:12 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> Ok, now I vote, that you don't implement \"any\" and use \"opaque\".\n> I don't think we want two types that do the same thing.\n> Is it that you like the name \"any\" more than \"opaque\" ?\n\nNo, it's that I want to deprecate \"opaque\" so that we can catch old\nuses that should not be there anymore. If you look at your code and\nyou decide that \"any\" is the correct semantics, then fine: change\n\"opaque\" to \"any\" and the warnings will go away. But relatively few\nexisting uses of \"opaque\" really mean \"any\", and I don't want the\npeople who are using \"opaque\" to mean \"cstring\", \"trigger\", etc\nto keep using \"opaque\" for those other purposes. The idea here is\nto force a security review.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 00:03:09 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " } ]
[ { "msg_contents": "As you probably know, PostgreSQL is quickly approaching the beta period\nfor its next release (7.3).\n\nIt would be heavily appreciated if you could attack the new source base\nstarting now and throughout the beta period in an attempt to make the\nnext release extra secure and stable from the outset.\n\nMore to the point, I'd love material I could show the local VPs as to a\nreason they should upgrade to 7.3 from either other products or previous\nversions of PostgreSQL. The necessary material could be as simple as\nthe test cases run against the new version and it's status (pass, fail\nwith immediate fix, fail with major rework required (7.4 fix?)). I've\nseen four cases which failed, but how many succeeded and what kind of\ntest were they?\n\nIf possible (based on group decision) it would be great if this material\ncould be included or linked to from the release notes.\n\nKeep up the excellent work.\n\nThanks,\n\tRod\n\n", "msg_date": "21 Aug 2002 13:48:00 -0400", "msg_from": "Rod Taylor <rbt@zort.ca>", "msg_from_op": true, "msg_subject": "Upcoming Beta" } ]
[ { "msg_contents": "\nMorning all ...\n\n\tThis afternoon, Bruce Momjiam created a new project on GBorg for\nlibpq++, and Jeroen T. Vermeulen created one for libpqxx ... Both projects\nsource directory from the central CVS repository have been copied over,\nincluding full history logs, and can be viewed at:\n\n\tlibpqxx:\n\thttp://gborg.postgresql.org/project/libpqxx/projdisplay.php\n\n\tAnd\n\n\tlibpq++:\n\thttp://gborg.postgresql.org/project/libpqpp/projdisplay.php\n\n\tLet us know of any problems with either project ...\n\n\n", "msg_date": "Wed, 21 Aug 2002 20:16:27 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "libpq++ and libpqxx moved to GBorg ..." } ]
[ { "msg_contents": "\nthey are no longer on the central repository, but are on GBorg ... I've\nmade the appropriate chagnes to configure and Makefiles to reflect the\nfact that libpq++ is no longer part of the central distribution, *but* I\nused 'cvs remove' to remove the files themselves, so that the old branches\n(ie. if we do a v7.2.2, we need it) still have access to the code ...\n\nlet me know if there are any problems ...\n\n", "msg_date": "Wed, 21 Aug 2002 21:21:10 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "libpq++ and libpqxx removed " }, { "msg_contents": "Can we perhaps have very prominent linking to gborg from the postgresql.org\nhome page? Especially now that there's libpq++ and stuff on there?\n\nChris\n\n> -----Original Message-----\n> From: pgsql-hackers-owner@postgresql.org\n> [mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Marc G. Fournier\n> Sent: Thursday, 22 August 2002 8:21 AM\n> To: pgsql-interfaces@postgresql.org\n> Cc: pgsql-hackers@postgresql.org\n> Subject: [HACKERS] libpq++ and libpqxx removed\n>\n>\n>\n> they are no longer on the central repository, but are on GBorg ... I've\n> made the appropriate chagnes to configure and Makefiles to reflect the\n> fact that libpq++ is no longer part of the central distribution, *but* I\n> used 'cvs remove' to remove the files themselves, so that the old branches\n> (ie. if we do a v7.2.2, we need it) still have access to the code ...\n>\n> let me know if there are any problems ...\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", "msg_date": "Thu, 22 Aug 2002 09:58:24 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: libpq++ and libpqxx removed " }, { "msg_contents": "\nThere will be on the new portal, but you are right, we should have such on\nthe current one until the new portal is in place ...\n\nVince, can you add a link to it off the menus themselves on the left side?\n\nOn Thu, 22 Aug 2002, Christopher Kings-Lynne wrote:\n\n> Can we perhaps have very prominent linking to gborg from the postgresql.org\n> home page? Especially now that there's libpq++ and stuff on there?\n>\n> Chris\n>\n> > -----Original Message-----\n> > From: pgsql-hackers-owner@postgresql.org\n> > [mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Marc G. Fournier\n> > Sent: Thursday, 22 August 2002 8:21 AM\n> > To: pgsql-interfaces@postgresql.org\n> > Cc: pgsql-hackers@postgresql.org\n> > Subject: [HACKERS] libpq++ and libpqxx removed\n> >\n> >\n> >\n> > they are no longer on the central repository, but are on GBorg ... I've\n> > made the appropriate chagnes to configure and Makefiles to reflect the\n> > fact that libpq++ is no longer part of the central distribution, *but* I\n> > used 'cvs remove' to remove the files themselves, so that the old branches\n> > (ie. if we do a v7.2.2, we need it) still have access to the code ...\n> >\n> > let me know if there are any problems ...\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\n", "msg_date": "Wed, 21 Aug 2002 22:59:24 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] libpq++ and libpqxx removed " }, { "msg_contents": "On Wed, 21 Aug 2002, Marc G. Fournier wrote:\n\n>\n> There will be on the new portal, but you are right, we should have such on\n> the current one until the new portal is in place ...\n>\n> Vince, can you add a link to it off the menus themselves on the left side?\n\nSure. Make me a button that matches the rest. That's what the holdup\nhas been for any changes. That's why I've been soliciting new designs.\nThe new portal isn't going to help in this regard either, I've been\ntrying to tell you that for months.\n\n\n>\n> On Thu, 22 Aug 2002, Christopher Kings-Lynne wrote:\n>\n> > Can we perhaps have very prominent linking to gborg from the postgresql.org\n> > home page? Especially now that there's libpq++ and stuff on there?\n> >\n> > Chris\n> >\n> > > -----Original Message-----\n> > > From: pgsql-hackers-owner@postgresql.org\n> > > [mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Marc G. Fournier\n> > > Sent: Thursday, 22 August 2002 8:21 AM\n> > > To: pgsql-interfaces@postgresql.org\n> > > Cc: pgsql-hackers@postgresql.org\n> > > Subject: [HACKERS] libpq++ and libpqxx removed\n> > >\n> > >\n> > >\n> > > they are no longer on the central repository, but are on GBorg ... I've\n> > > made the appropriate chagnes to configure and Makefiles to reflect the\n> > > fact that libpq++ is no longer part of the central distribution, *but* I\n> > > used 'cvs remove' to remove the files themselves, so that the old branches\n> > > (ie. if we do a v7.2.2, we need it) still have access to the code ...\n> > >\n> > > let me know if there are any problems ...\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>\n>\n\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n", "msg_date": "Wed, 21 Aug 2002 22:24:47 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": false, "msg_subject": "Re: libpq++ and libpqxx removed " }, { "msg_contents": "Where can we see the new portal? Maybe it should be designed in such a way\nas to not use image links at all. From all my experience in doing\nwebsites - that'd be a _really_ good idea.\n\nChris\n\n> > There will be on the new portal, but you are right, we should\n> have such on\n> > the current one until the new portal is in place ...\n> >\n> > Vince, can you add a link to it off the menus themselves on the\n> left side?\n>\n> Sure. Make me a button that matches the rest. That's what the holdup\n> has been for any changes. That's why I've been soliciting new designs.\n> The new portal isn't going to help in this regard either, I've been\n> trying to tell you that for months.\n\n", "msg_date": "Thu, 22 Aug 2002 10:28:03 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: libpq++ and libpqxx removed " }, { "msg_contents": "On Thu, 22 Aug 2002, Christopher Kings-Lynne wrote:\n\n> Where can we see the new portal? Maybe it should be designed in such a way\n> as to not use image links at all. From all my experience in doing\n> websites - that'd be a _really_ good idea.\n\nI don't believe we are using image links on the new portal ...\n>\n> Chris\n>\n> > > There will be on the new portal, but you are right, we should\n> > have such on\n> > > the current one until the new portal is in place ...\n> > >\n> > > Vince, can you add a link to it off the menus themselves on the\n> > left side?\n> >\n> > Sure. Make me a button that matches the rest. That's what the holdup\n> > has been for any changes. That's why I've been soliciting new designs.\n> > The new portal isn't going to help in this regard either, I've been\n> > trying to tell you that for months.\n>\n>\n\n", "msg_date": "Wed, 21 Aug 2002 23:36:14 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] libpq++ and libpqxx removed " } ]
[ { "msg_contents": "\nThat's one thing that I can't touch, since I have no idea about SGML :(\n\nBruce, can you extract that and add it to the CVS repository for libpq++?\n\n", "msg_date": "Wed, 21 Aug 2002 21:23:28 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "libpq++ documentation ..." }, { "msg_contents": "\nI can remove it cleanly, but I have no idea how to add the SGML in a way\nthat will allow it to build.\n\n---------------------------------------------------------------------------\n\nMarc G. Fournier wrote:\n> \n> That's one thing that I can't touch, since I have no idea about SGML :(\n> \n> Bruce, can you extract that and add it to the CVS repository for libpq++?\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 22:26:41 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: libpq++ documentation ..." }, { "msg_contents": "On Wed, 21 Aug 2002, Bruce Momjian wrote:\n\n>\n> I can remove it cleanly, but I have no idea how to add the SGML in a way\n> that will allow it to build.\n\nAnyone with some time on their hands that knows SGML enough to do this? :)\n\nFor now, can you extract what is required and commit it to the project\n*but* don't remove from our central source(s)? Maybe add a prominent\ncomment in the docs for now that the source has moved to GBorg, so that\nppl aren't confused that the docs are there but there is no code, until we\ncan get a 'clean build' of the docs in the project itself?\n\n >\n> ---------------------------------------------------------------------------\n>\n> Marc G. Fournier wrote:\n> >\n> > That's one thing that I can't touch, since I have no idea about SGML :(\n> >\n> > Bruce, can you extract that and add it to the CVS repository for libpq++?\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> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n>\n\n", "msg_date": "Wed, 21 Aug 2002 23:38:14 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: libpq++ documentation ..." }, { "msg_contents": "\nI will remove it from our cvs and move it over, and clean up our SGML to\ncompile properly. I can do that much.\n\n---------------------------------------------------------------------------\n\nMarc G. Fournier wrote:\n> On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> \n> >\n> > I can remove it cleanly, but I have no idea how to add the SGML in a way\n> > that will allow it to build.\n> \n> Anyone with some time on their hands that knows SGML enough to do this? :)\n> \n> For now, can you extract what is required and commit it to the project\n> *but* don't remove from our central source(s)? Maybe add a prominent\n> comment in the docs for now that the source has moved to GBorg, so that\n> ppl aren't confused that the docs are there but there is no code, until we\n> can get a 'clean build' of the docs in the project itself?\n> \n> >\n> > ---------------------------------------------------------------------------\n> >\n> > Marc G. Fournier wrote:\n> > >\n> > > That's one thing that I can't touch, since I have no idea about SGML :(\n> > >\n> > > Bruce, can you extract that and add it to the CVS repository for libpq++?\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> > Bruce Momjian | http://candle.pha.pa.us\n> > pgman@candle.pha.pa.us | (610) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n> >\n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 23:14:17 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: libpq++ documentation ..." }, { "msg_contents": "\nDone.\n\n---------------------------------------------------------------------------\n\nBruce Momjian wrote:\n> \n> I will remove it from our cvs and move it over, and clean up our SGML to\n> compile properly. I can do that much.\n> \n> ---------------------------------------------------------------------------\n> \n> Marc G. Fournier wrote:\n> > On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> > \n> > >\n> > > I can remove it cleanly, but I have no idea how to add the SGML in a way\n> > > that will allow it to build.\n> > \n> > Anyone with some time on their hands that knows SGML enough to do this? :)\n> > \n> > For now, can you extract what is required and commit it to the project\n> > *but* don't remove from our central source(s)? Maybe add a prominent\n> > comment in the docs for now that the source has moved to GBorg, so that\n> > ppl aren't confused that the docs are there but there is no code, until we\n> > can get a 'clean build' of the docs in the project itself?\n> > \n> > >\n> > > ---------------------------------------------------------------------------\n> > >\n> > > Marc G. Fournier wrote:\n> > > >\n> > > > That's one thing that I can't touch, since I have no idea about SGML :(\n> > > >\n> > > > Bruce, can you extract that and add it to the CVS repository for libpq++?\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> > > Bruce Momjian | http://candle.pha.pa.us\n> > > pgman@candle.pha.pa.us | (610) 359-1001\n> > > + If your life is a hard drive, | 13 Roberts Road\n> > > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n> > >\n> > \n> > \n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 23:17:58 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: libpq++ documentation ..." }, { "msg_contents": "\n'K, wanna do DBD::Pg next? :)\n\nOn Wed, 21 Aug 2002, Bruce Momjian wrote:\n\n>\n> Done.\n>\n> ---------------------------------------------------------------------------\n>\n> Bruce Momjian wrote:\n> >\n> > I will remove it from our cvs and move it over, and clean up our SGML to\n> > compile properly. I can do that much.\n> >\n> > ---------------------------------------------------------------------------\n> >\n> > Marc G. Fournier wrote:\n> > > On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> > >\n> > > >\n> > > > I can remove it cleanly, but I have no idea how to add the SGML in a way\n> > > > that will allow it to build.\n> > >\n> > > Anyone with some time on their hands that knows SGML enough to do this? :)\n> > >\n> > > For now, can you extract what is required and commit it to the project\n> > > *but* don't remove from our central source(s)? Maybe add a prominent\n> > > comment in the docs for now that the source has moved to GBorg, so that\n> > > ppl aren't confused that the docs are there but there is no code, until we\n> > > can get a 'clean build' of the docs in the project itself?\n> > >\n> > > >\n> > > > ---------------------------------------------------------------------------\n> > > >\n> > > > Marc G. Fournier wrote:\n> > > > >\n> > > > > That's one thing that I can't touch, since I have no idea about SGML :(\n> > > > >\n> > > > > Bruce, can you extract that and add it to the CVS repository for libpq++?\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> > > > Bruce Momjian | http://candle.pha.pa.us\n> > > > pgman@candle.pha.pa.us | (610) 359-1001\n> > > > + If your life is a hard drive, | 13 Roberts Road\n> > > > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n> > > >\n> > >\n> > >\n> >\n> > --\n> > Bruce Momjian | http://candle.pha.pa.us\n> > pgman@candle.pha.pa.us | (610) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n>\n\n", "msg_date": "Thu, 22 Aug 2002 00:21:39 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: libpq++ documentation ..." }, { "msg_contents": "\nI knew that was coming. Some have been concerned that though Edmund\nsaid OK, there is some new person who is the maintainer, though you\nwould think Edmund would be the final word on that.\n\nAnyway, I will do it now, or as soon as I empty the patch queue.\n\n\n---------------------------------------------------------------------------\n\nMarc G. Fournier wrote:\n> \n> 'K, wanna do DBD::Pg next? :)\n> \n> On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> \n> >\n> > Done.\n> >\n> > ---------------------------------------------------------------------------\n> >\n> > Bruce Momjian wrote:\n> > >\n> > > I will remove it from our cvs and move it over, and clean up our SGML to\n> > > compile properly. I can do that much.\n> > >\n> > > ---------------------------------------------------------------------------\n> > >\n> > > Marc G. Fournier wrote:\n> > > > On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> > > >\n> > > > >\n> > > > > I can remove it cleanly, but I have no idea how to add the SGML in a way\n> > > > > that will allow it to build.\n> > > >\n> > > > Anyone with some time on their hands that knows SGML enough to do this? :)\n> > > >\n> > > > For now, can you extract what is required and commit it to the project\n> > > > *but* don't remove from our central source(s)? Maybe add a prominent\n> > > > comment in the docs for now that the source has moved to GBorg, so that\n> > > > ppl aren't confused that the docs are there but there is no code, until we\n> > > > can get a 'clean build' of the docs in the project itself?\n> > > >\n> > > > >\n> > > > > ---------------------------------------------------------------------------\n> > > > >\n> > > > > Marc G. Fournier wrote:\n> > > > > >\n> > > > > > That's one thing that I can't touch, since I have no idea about SGML :(\n> > > > > >\n> > > > > > Bruce, can you extract that and add it to the CVS repository for libpq++?\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> > > > > Bruce Momjian | http://candle.pha.pa.us\n> > > > > pgman@candle.pha.pa.us | (610) 359-1001\n> > > > > + If your life is a hard drive, | 13 Roberts Road\n> > > > > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n> > > > >\n> > > >\n> > > >\n> > >\n> > > --\n> > > Bruce Momjian | http://candle.pha.pa.us\n> > > pgman@candle.pha.pa.us | (610) 359-1001\n> > > + If your life is a hard drive, | 13 Roberts Road\n> > > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n> >\n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 21 Aug 2002 23:25:14 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: libpq++ documentation ..." }, { "msg_contents": "On Wed, 21 Aug 2002, Bruce Momjian wrote:\n\n>\n> I knew that was coming. Some have been concerned that though Edmund\n> said OK, there is some new person who is the maintainer, though you\n> would think Edmund would be the final word on that.\n>\n> Anyway, I will do it now, or as soon as I empty the patch queue.\n\nIMHO, if there is someone else, they should be step'ng forward to let us\nknow, no?\n\n\n", "msg_date": "Thu, 22 Aug 2002 00:33:28 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: libpq++ documentation ..." }, { "msg_contents": "\nYep, let's create the project.\n\n---------------------------------------------------------------------------\n\nMarc G. Fournier wrote:\n> On Wed, 21 Aug 2002, Bruce Momjian wrote:\n> \n> >\n> > I knew that was coming. Some have been concerned that though Edmund\n> > said OK, there is some new person who is the maintainer, though you\n> > would think Edmund would be the final word on that.\n> >\n> > Anyway, I will do it now, or as soon as I empty the patch queue.\n> \n> IMHO, if there is someone else, they should be step'ng forward to let us\n> know, no?\n> \n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 00:13:38 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: libpq++ documentation ..." } ]
[ { "msg_contents": "Hi,\n\nltree is good and fast at indexing hierarchies, right? XML is a standard\nway of storing hierarchies, right? Would there be any way of allowing ltree\nto index XML text fields to allow really fast XPath searches?\n\nChris\n\n", "msg_date": "Thu, 22 Aug 2002 10:52:59 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "Theoretical XML & ltree Question" }, { "msg_contents": "On Thu, 22 Aug 2002, Christopher Kings-Lynne wrote:\n\n> Hi,\n>\n> ltree is good and fast at indexing hierarchies, right? XML is a standard\n> way of storing hierarchies, right? Would there be any way of allowing ltree\n> to index XML text fields to allow really fast XPath searches?\n\nI was waiting this question :) I was aware of xml, it's in our TODO.\n\n\n>\n> Chris\n>\n>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n>\n\n\tRegards,\n\t\tOleg\n_____________________________________________________________\nOleg Bartunov, sci.researcher, hostmaster of AstroNet,\nSternberg Astronomical Institute, Moscow University (Russia)\nInternet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/\nphone: +007(095)939-16-83, +007(095)939-23-83\n\n", "msg_date": "Thu, 22 Aug 2002 11:16:47 +0300 (GMT)", "msg_from": "Oleg Bartunov <oleg@sai.msu.su>", "msg_from_op": false, "msg_subject": "Re: Theoretical XML & ltree Question" } ]
[ { "msg_contents": "\n\n> -----Original Message-----\n> From: Vince Vielhaber [mailto:vev@michvhf.com] \n> Sent: 22 August 2002 03:25\n> To: Marc G. Fournier\n> Cc: Christopher Kings-Lynne; pgsql-interfaces@postgresql.org; \n> pgsql-hackers@postgresql.org\n> Subject: Re: [HACKERS] libpq++ and libpqxx removed \n> \n> \nThe new portal isn't going to help in this regard either, I've been\ntrying to tell you that for months.\n\nCare to let me in on that one seeing as I'm the one who's been putting\nblood and sweat into it?\n\nRegards, Dave.\n", "msg_date": "Thu, 22 Aug 2002 08:40:59 +0100", "msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>", "msg_from_op": true, "msg_subject": "Re: libpq++ and libpqxx removed " } ]
[ { "msg_contents": "\n\n> -----Original Message-----\n> From: Christopher Kings-Lynne [mailto:chriskl@familyhealth.com.au] \n> Sent: 22 August 2002 03:28\n> To: Vince Vielhaber; Marc G. Fournier\n> Cc: pgsql-interfaces@postgresql.org; pgsql-hackers@postgresql.org\n> Subject: Re: [HACKERS] libpq++ and libpqxx removed \n> \n> \n> Where can we see the new portal? Maybe it should be designed \n> in such a way as to not use image links at all. From all my \n> experience in doing websites - that'd be a _really_ good idea.\n\nIt doesn't use images for links. All the main links & menu items are\nheld in a global php file so if a single change is made, it affects all\npages.\n\nRegards, Dave.\n", "msg_date": "Thu, 22 Aug 2002 08:42:57 +0100", "msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>", "msg_from_op": true, "msg_subject": "Re: libpq++ and libpqxx removed " } ]
[ { "msg_contents": "\n> > Ok, now I vote, that you don't implement \"any\" and use \"opaque\".\n> > I don't think we want two types that do the same thing.\n> > Is it that you like the name \"any\" more than \"opaque\" ?\n> \n> No, it's that I want to deprecate \"opaque\" so that we can catch old\n> uses that should not be there anymore. If you look at your code and\n> you decide that \"any\" is the correct semantics, then fine: change\n> \"opaque\" to \"any\" and the warnings will go away. But relatively few\n> existing uses of \"opaque\" really mean \"any\", and I don't want the\n> people who are using \"opaque\" to mean \"cstring\", \"trigger\", etc\n> to keep using \"opaque\" for those other purposes. The idea here is\n> to force a security review.\n\nThat is what I have been trying to say, imho \"any\" should have the same \nNOTICE as opaque has, since it is potentially dangerous.\nI would suggest a warning NOTICE for opaque and not depricate it.\n\nImho the NOTICE should *not* go away.\n\nIf we want \"any\" in the future, it should imho always be passed a \"safe\" \nDatum that includes type info. This will allow us to create a type \"any\"\nthat does not have the pitfalls of opaque. \n\nAndreas\n", "msg_date": "Thu, 22 Aug 2002 11:21:22 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: @(#) Mordred Labs advisory 0x0001: Buffer overflow in " } ]
[ { "msg_contents": "Hi.\nThere is no need to ask me to attack a postgresql source code,\nas long as i remember myself, i was always in studing someone's source code\nor disasm output..\nBy the way, the code i write being a plain web programmer, even for small\nprojects is fairly buggy:-))).\nOne little thing saves me a bit - i am using a perl, and the code itself\ndont suffer from the buffer/integer overflows much,\nnot talking about DBD::Pg.\n\nSo i can repeat the above, don't ask me to do the things which i am doing\n24 hours in a day...\n\nBest regards.\n\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n", "msg_date": "Thu, 22 Aug 2002 10:20:12 +0000", "msg_from": "Sir Mordred The Traitor <mordred@s-mail.com>", "msg_from_op": true, "msg_subject": "Re: Upcoming Beta" } ]
[ { "msg_contents": ">\"Marc G. Fournier\" <scrappy@hub.org> writes:\n\n> Are we all caught up now on the known bugs/fixes? Would it be reasonably\n> safe to do up a quick v7.2.2 Security Fix Release tomorrow afternoon?\n\nMaybe it makes sense to wait about a week.\n\n\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n", "msg_date": "Thu, 22 Aug 2002 13:50:32 +0000", "msg_from": "Sir Mordred The Traitor <mordred@s-mail.com>", "msg_from_op": true, "msg_subject": "Re: Release of v7.2.2" } ]
[ { "msg_contents": "The original problem with \"opaque\" is gone in CVS tip:\n\nregression=# select cash_out(2);\nERROR: Function cash_out(integer) does not exist\n Unable to identify a function that satisfies the given argument types\n You may need to add explicit typecasts\n\nHowever, I'm not entirely satisfied with it. In particular, the type\nsystem will still allow you to pass the result of an output function to\nanother input function:\n\nregression=# select textin(inet_out('123.123.123.123'));\n textin\n--------\n\n(1 row)\n\nI should have gotten '123.123.123.123' here, but didn't, and in other\ncases I get garbage. (I'm not sure that a crash is possible, but I'm\nnot sure one isn't, either.) This seems like a reasonable operation\nin many cases, so I don't really want to put in a kluge to disallow it.\n\nThe problem comes down to the fact that the system can't really cope\nwith Datums of type cstring: it doesn't know how to copy them. We have\ncstring declared as a pass-by-value type, which it isn't, and so the\npointer gets pushed around without concern for copying the pointed-to\nstring.\n\nI am thinking about declaring type cstring in pg_type as\npass-by-reference with typlen set to -2, and altering the places that\ninterpret typlen to understand -2 as meaning null-terminated string.\n(Currently, typlen > 0 means a fixed-length type, typlen = -1 is used\nfor varlena values which have a length word, and there's no meaning\nassigned to other negative typlen values.)\n\nI haven't really looked to see how many places would need to change,\nbut if it's not too many then this seems reasonable to do --- it would\nmake the cstring feature a lot more robust than it is now.\n\nComments?\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 11:52:08 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Making cstring type less pseudo and more real" }, { "msg_contents": "En Thu, 22 Aug 2002 11:52:08 -0400\nTom Lane <tgl@sss.pgh.pa.us> escribi�:\n\n> The original problem with \"opaque\" is gone in CVS tip:\n> \n> regression=# select cash_out(2);\n> ERROR: Function cash_out(integer) does not exist\n> Unable to identify a function that satisfies the given argument types\n> You may need to add explicit typecasts\n\nYes, but how does it work?\n\nalvh=> CREATE TABLE money_test (a money);\nCREATE TABLE\nalvh=> INSERT INTO money_test VALUES (2);\nERROR: column \"a\" is of type 'money' but expression is of type 'integer'\n\tYou will need to rewrite or cast the expression\nalvh=> INSERT INTO money_test VALUES ('2');\nERROR: Bad money external representation 2\nalvh=> INSERT INTO money_test VALUES ('2.00');\nERROR: Bad money external representation 2.00\nalvh=> INSERT INTO money_test VALUES ('$2.00');\nERROR: Bad money external representation $2.00\nalvh=> INSERT INTO money_test VALUES ('2'::money);\nERROR: Bad money external representation 2\nalvh=> INSERT INTO money_test VALUES (money(2));\nERROR: Function money(integer) does not exist\n\tUnable to identify a function that satisfies the given argument types\n\tYou may need to add explicit typecasts\n\n\n> I am thinking about declaring type cstring in pg_type as\n> pass-by-reference with typlen set to -2, and altering the places that\n> interpret typlen to understand -2 as meaning null-terminated string.\n> (Currently, typlen > 0 means a fixed-length type, typlen = -1 is used\n> for varlena values which have a length word, and there's no meaning\n> assigned to other negative typlen values.)\n\nSounds good.\n\n> I haven't really looked to see how many places would need to change,\n> but if it's not too many then this seems reasonable to do --- it would\n> make the cstring feature a lot more robust than it is now.\n\nAt least all places that call get_typlenbyval() and get_typlen() or get\nthe typlen directly, it seems. They are not too much, AFAICS...\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\nSi no sabes adonde vas, es muy probable que acabes en otra parte.\n", "msg_date": "Thu, 22 Aug 2002 16:58:19 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": false, "msg_subject": "Re: Making cstring type less pseudo and more real" }, { "msg_contents": "Alvaro Herrera <alvherre@atentus.com> writes:\n> alvh=> CREATE TABLE money_test (a money);\n> CREATE TABLE\n> alvh=> INSERT INTO money_test VALUES (2);\n> ERROR: column \"a\" is of type 'money' but expression is of type 'integer'\n> \tYou will need to rewrite or cast the expression\n> alvh=> INSERT INTO money_test VALUES ('2');\n> ERROR: Bad money external representation 2\n> alvh=> INSERT INTO money_test VALUES ('2.00');\n> ERROR: Bad money external representation 2.00\n> alvh=> INSERT INTO money_test VALUES ('$2.00');\n> ERROR: Bad money external representation $2.00\n> alvh=> INSERT INTO money_test VALUES ('2'::money);\n> ERROR: Bad money external representation 2\n\nWhat does \"show lc_monetary\" say? I think you may be in a non-US locale.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 18:05:00 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Re: Making cstring type less pseudo and more real " } ]
[ { "msg_contents": "does the recent security activity, including several reported exploits \nand patches, as well as the mention of creation of an audit team merit \nthe creation of a new pgsql-security list?\n\nas someone working with a paranoid sysadmin, i'd find it to be of use...\n\nany thoughts? would there be sufficient traffic? maybe the list would \nactually _help_ generate traffic?\n\n-tfo\n", "msg_date": "Thu, 22 Aug 2002 11:05:49 -0500", "msg_from": "Thomas O'Connell <tfo@monsterlabs.com>", "msg_from_op": true, "msg_subject": "recent security activity" }, { "msg_contents": "I think that's an excellent idea. It would allow people to subscribe to\nwhat would seemingly be a low volume mailing list and still be alerted\nto possible issues they should be aware of.\n\nSign,\n\n\tGreg Copeland\n\n\n\nOn Thu, 2002-08-22 at 11:05, Thomas O'Connell wrote:\n> does the recent security activity, including several reported exploits \n> and patches, as well as the mention of creation of an audit team merit \n> the creation of a new pgsql-security list?\n> \n> as someone working with a paranoid sysadmin, i'd find it to be of use...\n> \n> any thoughts? would there be sufficient traffic? maybe the list would \n> actually _help_ generate traffic?\n> \n> -tfo\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org", "msg_date": "22 Aug 2002 17:41:03 -0500", "msg_from": "Greg Copeland <greg@CopelandConsulting.Net>", "msg_from_op": false, "msg_subject": "Re: recent security activity" }, { "msg_contents": "Greg Copeland <greg@CopelandConsulting.Net> writes:\n> I think that's an excellent idea. It would allow people to subscribe to\n> what would seemingly be a low volume mailing list and still be alerted\n> to possible issues they should be aware of.\n\nWould the purpose of the list be for publicizing vulnerabilities and\npatches, or for the discussion of potential security problems, code\nauditing, and related development activity?\n\nIf the former, I think pgsql-announce is adequate for that purpose. If\nthe latter, I'd rather see that kind of discussion on -hackers, so\nthat other developers are aware of what's going on.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "22 Aug 2002 18:48:40 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: recent security activity" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> Would the purpose of the list be for publicizing vulnerabilities and\n> patches, or for the discussion of potential security problems, code\n> auditing, and related development activity?\n\n> If the former, I think pgsql-announce is adequate for that purpose. If\n> the latter, I'd rather see that kind of discussion on -hackers, so\n> that other developers are aware of what's going on.\n\nAlso worth noting in this connection: if someone wants to report a\nsecurity issue to the developers *without* publicizing it (as used to\nbe considered good form), you can send to the pgsql-core mailing list.\nThis goes to just the core committee members and is not archived anywhere\npublic.\n\nI tend to agree with Neil that a separate -security list isn't needed,\nbut will not stand in the way if there's sufficient interest.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 19:07:05 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: recent security activity " }, { "msg_contents": "I assumed it would be for patches and security alerts with followups as\nneeded.\n\nI can see where use of announce can serve this purpose, however, if\nsomeone is solely interested in the security advisory aspects, they may\nnot care about the announcement-of-the-day.\n\nJust food for thought. I can see why you wouldn't want another\nlist..otoh, I can see where someone may not want to monitor announce for\nthe sole purpose of watching for security advisories and patches.\n\nPerhaps the use of \"[SECURITY]\" in the subject, or some such item, would\nbetter address the issue and simply continue to use announce? That way,\nMUA filters can easily be used to find and highlight items of interest.\n\n\nGreg\n\n\n\nOn Thu, 2002-08-22 at 17:48, Neil Conway wrote:\n> Greg Copeland <greg@CopelandConsulting.Net> writes:\n> > I think that's an excellent idea. It would allow people to subscribe to\n> > what would seemingly be a low volume mailing list and still be alerted\n> > to possible issues they should be aware of.\n> \n> Would the purpose of the list be for publicizing vulnerabilities and\n> patches, or for the discussion of potential security problems, code\n> auditing, and related development activity?\n> \n> If the former, I think pgsql-announce is adequate for that purpose. If\n> the latter, I'd rather see that kind of discussion on -hackers, so\n> that other developers are aware of what's going on.\n>", "msg_date": "23 Aug 2002 07:37:06 -0500", "msg_from": "Greg Copeland <greg@CopelandConsulting.Net>", "msg_from_op": false, "msg_subject": "Re: recent security activity" } ]
[ { "msg_contents": "Hi,\n\nI've been using postgres for a little over a year now, and overall I've\nbeen very pleased with it. Thanks!\n\nThe psql client is really nice, but I was frustrated by the lack of a\n\"no-autocommit\" mode. So, I made a few small modifications to the client\nto wrap everything in a transaction block automatically. I've been using\nit locally for a while, and it seems to work alright. The patch is\nattached (it's a patch to the 7.2.1 branch).\n\nDo other people think this is a useful feature?\n\nThe one thing that I don't like about the patch, but that I'm not sure how\nto handle, is telling the user that one transaction has ended and another\nhas begun. Currently, I am just printing NOTICE messages (\"NOTICE:\naborting current transaction and beginning a new one\", for example) but\nthis seems wrong, since the messages aren't coming from the backend.\nSuggestions on how to handle this?\n\n...dave", "msg_date": "Thu, 22 Aug 2002 09:38:01 -0700 (PDT)", "msg_from": "David Van Wie <vanwie@cryptio.net>", "msg_from_op": true, "msg_subject": "turning off autocommit behavior in psql" }, { "msg_contents": "David Van Wie <vanwie@cryptio.net> writes:\n> The psql client is really nice, but I was frustrated by the lack of a\n> \"no-autocommit\" mode. So, I made a few small modifications to the client\n> to wrap everything in a transaction block automatically.\n\nI believe our intention has been to modify the backend to offer that\nbehavior (with a GUC variable to turn it on or off). Supporting it\nonly in psql doesn't seem the way to go.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 22 Aug 2002 12:59:16 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: turning off autocommit behavior in psql " }, { "msg_contents": "On Thu, 22 Aug 2002, Tom Lane wrote:\n\n> I believe our intention has been to modify the backend to offer that\n> behavior (with a GUC variable to turn it on or off). Supporting it\n> only in psql doesn't seem the way to go.\n\nAfter playing around with the transaction code for a while, I think I have\na reasonable idea of how to do this in the backend. This is my first time\nworking with the backend code, so it's entirely possible that I'm going\nabout this completely wrong, but the change seems to work and pass\nregression tests.\n\nI added a GUC variable \"autocommit\" which defaults to true. (Current\nbehavior). When it is set to false, the backend should begin a transaction\nblock if it is not inside one when executing any command. I did this by\nchanging StartTransactionCommand() in backend/access/transam/xact.c to\ncall BeginTransactionBlock() after StartTransaction() if it is not\ncurrently in a transaction block.\n\nDoes this sound like a reasonable approach?\n\nMy first attempt at a patch (to current cvs source) is attached.\n\n...dave", "msg_date": "Thu, 22 Aug 2002 18:36:20 -0700 (PDT)", "msg_from": "David Van Wie <vanwie@cryptio.net>", "msg_from_op": true, "msg_subject": "Re: turning off autocommit behavior in psql " }, { "msg_contents": "\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://candle.pha.pa.us/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nDavid Van Wie wrote:\n> On Thu, 22 Aug 2002, Tom Lane wrote:\n> \n> > I believe our intention has been to modify the backend to offer that\n> > behavior (with a GUC variable to turn it on or off). Supporting it\n> > only in psql doesn't seem the way to go.\n> \n> After playing around with the transaction code for a while, I think I have\n> a reasonable idea of how to do this in the backend. This is my first time\n> working with the backend code, so it's entirely possible that I'm going\n> about this completely wrong, but the change seems to work and pass\n> regression tests.\n> \n> I added a GUC variable \"autocommit\" which defaults to true. (Current\n> behavior). When it is set to false, the backend should begin a transaction\n> block if it is not inside one when executing any command. I did this by\n> changing StartTransactionCommand() in backend/access/transam/xact.c to\n> call BeginTransactionBlock() after StartTransaction() if it is not\n> currently in a transaction block.\n> \n> Does this sound like a reasonable approach?\n> \n> My first attempt at a patch (to current cvs source) is attached.\n> \n> ...dave\n\nContent-Description: \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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 16:41:31 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: turning off autocommit behavior in psql" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Your patch has been added to the PostgreSQL unapplied patches list at:\n\nPlease wait for this to be reviewed.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 18:04:36 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: turning off autocommit behavior in psql " }, { "msg_contents": "\nTom Lane has applied this patch and merged it into our code. Thanks.\n\n---------------------------------------------------------------------------\n\nDavid Van Wie wrote:\n> On Thu, 22 Aug 2002, Tom Lane wrote:\n> \n> > I believe our intention has been to modify the backend to offer that\n> > behavior (with a GUC variable to turn it on or off). Supporting it\n> > only in psql doesn't seem the way to go.\n> \n> After playing around with the transaction code for a while, I think I have\n> a reasonable idea of how to do this in the backend. This is my first time\n> working with the backend code, so it's entirely possible that I'm going\n> about this completely wrong, but the change seems to work and pass\n> regression tests.\n> \n> I added a GUC variable \"autocommit\" which defaults to true. (Current\n> behavior). When it is set to false, the backend should begin a transaction\n> block if it is not inside one when executing any command. I did this by\n> changing StartTransactionCommand() in backend/access/transam/xact.c to\n> call BeginTransactionBlock() after StartTransaction() if it is not\n> currently in a transaction block.\n> \n> Does this sound like a reasonable approach?\n> \n> My first attempt at a patch (to current cvs source) is attached.\n> \n> ...dave\n\nContent-Description: \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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sat, 31 Aug 2002 01:42:11 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: turning off autocommit behavior in psql" } ]
[ { "msg_contents": "\nJust got a note from Jeffrey on this, and he acknowledges that he is\ncontinuing to maintain DBD::Pg through CPAN, so we won't be moving it to\nGBorg ...\n\nBruce, he mentions having tried to contact you via email, but being\nunsuccessful due to your anti-spam blacklist filters :(\n\n", "msg_date": "Thu, 22 Aug 2002 15:55:18 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "DBD::Pg ..." }, { "msg_contents": "\"Marc G. Fournier\" <scrappy@hub.org> writes:\n> Just got a note from Jeffrey on this, and he acknowledges that he is\n> continuing to maintain DBD::Pg through CPAN, so we won't be moving it to\n> GBorg ...\n\nOk. Are there plans to remove interfaces/perl5 from the CVS tree?\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "22 Aug 2002 15:05:05 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: DBD::Pg ..." }, { "msg_contents": "On 22 Aug 2002, Neil Conway wrote:\n\n> \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > Just got a note from Jeffrey on this, and he acknowledges that he is\n> > continuing to maintain DBD::Pg through CPAN, so we won't be moving it to\n> > GBorg ...\n>\n> Ok. Are there plans to remove interfaces/perl5 from the CVS tree?\n\nYes ... that will be moved over to GBorg similar to libpq++ ... Bruce, can\nyou create a projet for that as well? maybe call it pg-perl?\n\n\n", "msg_date": "Thu, 22 Aug 2002 16:10:17 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: DBD::Pg ..." }, { "msg_contents": "Marc G. Fournier wrote:\n> \n> Just got a note from Jeffrey on this, and he acknowledges that he is\n> continuing to maintain DBD::Pg through CPAN, so we won't be moving it to\n> GBorg ...\n\nWould you please ask him if he would create a gborg project for it so he\nand I can make changes, and we can have a mailing list and bug list for\nit? I think that is a good idea. I have the code here if he wants me\nto do it.\n\n> Bruce, he mentions having tried to contact you via email, but being\n> unsuccessful due to your anti-spam blacklist filters :(\n\nThat is strange. That is the first report I have gotten that my spam\nblocks are blocking anything but spam. Do you have is address so I can\nwhitelist him. I see his address in the Pg:DBD README as Jeffrey W.\nBaker <jwbaker@acm.org>. I can't believe I am blocking acm.org.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 17:54:45 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: DBD::Pg ..." }, { "msg_contents": "\nCC'd in ...\n\n\nOn Thu, 22 Aug 2002, Bruce Momjian wrote:\n\n> Marc G. Fournier wrote:\n> >\n> > Just got a note from Jeffrey on this, and he acknowledges that he is\n> > continuing to maintain DBD::Pg through CPAN, so we won't be moving it to\n> > GBorg ...\n>\n> Would you please ask him if he would create a gborg project for it so he\n> and I can make changes, and we can have a mailing list and bug list for\n> it? I think that is a good idea. I have the code here if he wants me\n> to do it.\n>\n> > Bruce, he mentions having tried to contact you via email, but being\n> > unsuccessful due to your anti-spam blacklist filters :(\n>\n> That is strange. That is the first report I have gotten that my spam\n> blocks are blocking anything but spam. Do you have is address so I can\n> whitelist him. I see his address in the Pg:DBD README as Jeffrey W.\n> Baker <jwbaker@acm.org>. I can't believe I am blocking acm.org.\n>\n> --\n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n>\n\n", "msg_date": "Thu, 22 Aug 2002 19:11:01 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: DBD::Pg ..." }, { "msg_contents": "\nCreated on gborg as pgperl.\n\n---------------------------------------------------------------------------\n\nMarc G. Fournier wrote:\n> On 22 Aug 2002, Neil Conway wrote:\n> \n> > \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > > Just got a note from Jeffrey on this, and he acknowledges that he is\n> > > continuing to maintain DBD::Pg through CPAN, so we won't be moving it to\n> > > GBorg ...\n> >\n> > Ok. Are there plans to remove interfaces/perl5 from the CVS tree?\n> \n> Yes ... that will be moved over to GBorg similar to libpq++ ... Bruce, can\n> you create a projet for that as well? maybe call it pg-perl?\n> \n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sun, 25 Aug 2002 15:46:32 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: DBD::Pg ..." }, { "msg_contents": "\nJeffrey, I have several Perl:DBD patches my mailbox that I have\naccumulated over the years that never made it over to Edmund or\nyourself. Would you like them? \n\nAlso, would you consider creating a Pg:DBD project on\nhttp://gborg.postgresql.org. We are trying to get all the PostgreSQL\ninterfaces on that site. It doesn't prevent you from distributing via\nCPAN as well, it just gives Pg:DBD more visibility, plus a mailing list\nand bug database, if that would be helpful.\n\nIf it would be helpful, I can create the project on there and make you\nan administrator.\n\n---------------------------------------------------------------------------\n\nBruce Momjian wrote:\n> Marc G. Fournier wrote:\n> > \n> > Just got a note from Jeffrey on this, and he acknowledges that he is\n> > continuing to maintain DBD::Pg through CPAN, so we won't be moving it to\n> > GBorg ...\n> \n> Would you please ask him if he would create a gborg project for it so he\n> and I can make changes, and we can have a mailing list and bug list for\n> it? I think that is a good idea. I have the code here if he wants me\n> to do it.\n> \n> > Bruce, he mentions having tried to contact you via email, but being\n> > unsuccessful due to your anti-spam blacklist filters :(\n> \n> That is strange. That is the first report I have gotten that my spam\n> blocks are blocking anything but spam. Do you have is address so I can\n> whitelist him. I see his address in the Pg:DBD README as Jeffrey W.\n> Baker <jwbaker@acm.org>. I can't believe I am blocking acm.org.\n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sun, 25 Aug 2002 15:56:29 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: DBD::Pg ..." } ]
[ { "msg_contents": "\n> > > Just got a note from Jeffrey on this, and he acknowledges that he is\n> > > continuing to maintain DBD::Pg through CPAN, so we won't be moving it to\n> > > GBorg ...\n> >\n> > Ok. Are there plans to remove interfaces/perl5 from the CVS tree?\n> \n> Yes ... that will be moved over to GBorg similar to libpq++ ... Bruce, can\n> you create a projet for that as well? maybe call it pg-perl?\n\nMaybe it could also be left to die ? Is someone still maintaining it ?\nPeople should really use DBD::Pg (good that it stays on CPAN). \n\nAndreas\n", "msg_date": "Thu, 22 Aug 2002 21:18:02 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: DBD::Pg ..." }, { "msg_contents": "On Thu, Aug 22, 2002 at 21:18:02 +0200,\n Zeugswetter Andreas SB SD <ZeugswetterA@spardat.at> wrote:\n> \n> People should really use DBD::Pg (good that it stays on CPAN). \n\nWhy do you say that?\n\nPg's interface seems a bit closer to libpg's interface than DBD::Pg.\nI would think that would be an advantage.\n", "msg_date": "Thu, 22 Aug 2002 15:34:04 -0500", "msg_from": "Bruno Wolff III <bruno@wolff.to>", "msg_from_op": false, "msg_subject": "Re: DBD::Pg ..." } ]
[ { "msg_contents": "\n> > People should really use DBD::Pg (good that it stays on CPAN). \n> \n> Why do you say that?\n> \n> Pg's interface seems a bit closer to libpg's interface than DBD::Pg.\n> I would think that would be an advantage.\n\nWell libpq is very nonstandard same as interfaces/perl5. Since I like\nsomewhat portable code, I would allways suggest people use the more standard \ninterfaces, like DBD::Pg ODBC ECPG.\n\nAndreas\n", "msg_date": "Thu, 22 Aug 2002 22:44:14 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: DBD::Pg ..." }, { "msg_contents": "\nWell, can't say much to ECPG/ODBC, but I find that using DBI, period, is\neasier then the perl5 interface, as it allows for creating reasonably\nportable perl code ...\n\nOn Thu, 22 Aug 2002, Zeugswetter Andreas SB SD wrote:\n\n>\n> > > People should really use DBD::Pg (good that it stays on CPAN).\n> >\n> > Why do you say that?\n> >\n> > Pg's interface seems a bit closer to libpg's interface than DBD::Pg.\n> > I would think that would be an advantage.\n>\n> Well libpq is very nonstandard same as interfaces/perl5. Since I like\n> somewhat portable code, I would allways suggest people use the more standard\n> interfaces, like DBD::Pg ODBC ECPG.\n>\n> Andreas\n>\n\n", "msg_date": "Thu, 22 Aug 2002 18:26:48 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: DBD::Pg ..." } ]
[ { "msg_contents": "\n\nI've been try to check the anoncvs out since Tom posted about the cstring stuff\ntoday and I keep getting this:\n\n$ cvs -d :pserver:anoncvs@anoncvs.postgresql.org:/projects/cvsroot co -P pgsql\ncvs server: existing repository /projects/cvsroot/CVSROOT/Emptydir does not match /projects/cvsroot/pgsql-server\ncvs server: ignoring module server\ncvs server: Updating pgsql/contrib/earthdistance\n$ \n\nand that's it.\n\nThe cvs command I'm using is exactly the same one I was running a couple of\ndays ago to make sure I had upto date code before having the quick poke around\nin this area myself.\n\nIs it me or is it widespread?\n\n\n-- \nNigel J. Andrews\nDirector\n\n---\nLogictree Systems Limited\nComputer Consultants\n\n", "msg_date": "Thu, 22 Aug 2002 22:13:38 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": true, "msg_subject": "anoncvs failure?" }, { "msg_contents": "-*- Nigel J. Andrews <nandrews@investsystems.co.uk> [ 2002-08-22 21:13 ]:\n> \n> \n> I've been try to check the anoncvs out since Tom posted about the cstring stuff\n> today and I keep getting this:\n> \n> $ cvs -d :pserver:anoncvs@anoncvs.postgresql.org:/projects/cvsroot co -P pgsql\n> cvs server: existing repository /projects/cvsroot/CVSROOT/Emptydir does not match /projects/cvsroot/pgsql-server\n> cvs server: ignoring module server\n> cvs server: Updating pgsql/contrib/earthdistance\n> $ \n> \n> and that's it.\n> \n> The cvs command I'm using is exactly the same one I was running a couple of\n> days ago to make sure I had upto date code before having the quick poke around\n> in this area myself.\n> \n> Is it me or is it widespread?\n\nWorks fine for me, I think (never done this before... :)\n\n0 tolli:tolli@hyperion pgsql $ cvs -d :pserver:anoncvs@anoncvs.postgresql.org:/projects/cvsroot co -P pgsql\n...shitloadoffiles...\ncvs server: Updating pgsql/src/win32\ncvs server: Updating pgsql/contrib/earthdistance\nU pgsql/contrib/earthdistance/Makefile\nU pgsql/contrib/earthdistance/README.earthdistance\nU pgsql/contrib/earthdistance/earthdistance.c\nU pgsql/contrib/earthdistance/earthdistance.sql.in\n0 tolli:tolli@hyperion pgsql $ du -sh pgsql/\n54M pgsql\n0 tolli:tolli@hyperion pgsql $\n\n\n-- \nRegards,\nTolli\ntolli@tol.li\n", "msg_date": "Thu, 22 Aug 2002 21:35:36 +0000", "msg_from": "=?iso-8859-1?Q?=DE=F3rhallur_H=E1lfd=E1narson?= <tolli@tol.li>", "msg_from_op": false, "msg_subject": "Re: anoncvs failure?" }, { "msg_contents": "\nOn Thu, 22 Aug 2002, Þórhallur Hálfdánarson wrote:\n\n> -*- Nigel J. Andrews <nandrews@investsystems.co.uk> [ 2002-08-22 21:13 ]:\n> > \n> > \n> > I've been try to check the anoncvs out since Tom posted about the cstring stuff\n> > today and I keep getting this:\n> > \n> > $ cvs -d :pserver:anoncvs@anoncvs.postgresql.org:/projects/cvsroot co -P pgsql\n> > cvs server: existing repository /projects/cvsroot/CVSROOT/Emptydir does not match /projects/cvsroot/pgsql-server\n> > cvs server: ignoring module server\n> > cvs server: Updating pgsql/contrib/earthdistance\n> > $ \n> > \n> > and that's it.\n>\n> Works fine for me, I think (never done this before... :)\n> \n> 0 tolli:tolli@hyperion pgsql $ cvs -d :pserver:anoncvs@anoncvs.postgresql.org:/projects/cvsroot co -P pgsql\n> ...shitloadoffiles...\n> cvs server: Updating pgsql/src/win32\n> cvs server: Updating pgsql/contrib/earthdistance\n> U pgsql/contrib/earthdistance/Makefile\n> U pgsql/contrib/earthdistance/README.earthdistance\n> U pgsql/contrib/earthdistance/earthdistance.c\n> U pgsql/contrib/earthdistance/earthdistance.sql.in\n> 0 tolli:tolli@hyperion pgsql $ du -sh pgsql/\n> 54M pgsql\n> 0 tolli:tolli@hyperion pgsql $\n\n\nThanks, I've deleted the existing tree and it seems to be checking it all out\nnow.\n\nVery odd.\n\n-- \nNigel J. Andrews\nDirector\n\n---\nLogictree Systems Limited\nComputer Consultants\n\n", "msg_date": "Thu, 22 Aug 2002 22:45:36 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": true, "msg_subject": "Re: anoncvs failure?" } ]
[ { "msg_contents": "There is so much going on, I am having trouble keeping things straight.\nJust thought I would let everyone know. ;-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 18:07:05 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "My head is spinning" }, { "msg_contents": "On Thursday 22 August 2002 06:07 pm, Bruce Momjian wrote:\n> There is so much going on, I am having trouble keeping things straight.\n> Just thought I would let everyone know. ;-)\n\nQuite busy, eh? :-)\n\nHey, Bruce, gyroscopes spin to keep them stable.....\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Thu, 22 Aug 2002 21:57:59 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: My head is spinning" }, { "msg_contents": "Lamar Owen wrote:\n> On Thursday 22 August 2002 06:07 pm, Bruce Momjian wrote:\n> > There is so much going on, I am having trouble keeping things straight.\n> > Just thought I would let everyone know. ;-)\n> \n> Quite busy, eh? :-)\n\nYep. I don't know where to start.\n\n> Hey, Bruce, gyroscopes spin to keep them stable.....\n\nThat doesn't seem to work for me.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 22:40:50 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: My head is spinning" }, { "msg_contents": "On Thursday 22 August 2002 10:40 pm, Bruce Momjian wrote:\n> Lamar Owen wrote:\n> > On Thursday 22 August 2002 06:07 pm, Bruce Momjian wrote:\n> > > There is so much going on, I am having trouble keeping things straight.\n> > > Just thought I would let everyone know. ;-)\n\n> > Quite busy, eh? :-)\n\n> Yep. I don't know where to start.\n\nWell, Ok, I think we're going in the right direction. The one thing I would \nlike to see is a release target for 7.2.2. There is a substantial interest \nin our userbase for this, judging from a thread among OpenACS developers. \nThus far it's not hit bigger news sites -- but I can tell you right now, with \nmy media experience, that people will assume the worst -- and if/when this \nhits slashdot, we really need a response to 'whence 7.2.2'. BugTraq's one \nthing -- slashdot is another thing entirely. I'm quite surprized, in fact, \nit hasn't hit Linux Today, slashdot, or Linux Weekly News. When/if it does, \n*SHA-ZAM*. I'd personally like the news item to read 'PostgreSQL developers \nrelease security update' rather than 'Security hole found in PostgreSQL'.\n\nIf push comes to shove I can push RPM's out Saturday, if the tarball is ready. \nIf not, Monday morning at the earliest.\n\n> > Hey, Bruce, gyroscopes spin to keep them stable.....\n\n> That doesn't seem to work for me.\n\nWell, we've already seen kudos go out to Marc. I think a round of applause \nshould also go to some other people -- but in particular Bruce has always \ndone a fine job of getting the release ready for packaging. Not to take away \nfrom other's efforts, just maybe to help encourage someone else who is \nfeeling overwhelmed (a couple hundred thousand lines of Fortran, a complete \nwebsite redesign, a major renovation, and some major engineering/surveying \nwork have my attention here... and that's just one job).\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Thu, 22 Aug 2002 23:01:01 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: My head is spinning" }, { "msg_contents": "Lamar Owen wrote:\n> On Thursday 22 August 2002 10:40 pm, Bruce Momjian wrote:\n> > Lamar Owen wrote:\n> > > On Thursday 22 August 2002 06:07 pm, Bruce Momjian wrote:\n> > > > There is so much going on, I am having trouble keeping things straight.\n> > > > Just thought I would let everyone know. ;-)\n> \n> > > Quite busy, eh? :-)\n> \n> > Yep. I don't know where to start.\n> \n> Well, Ok, I think we're going in the right direction. The one thing I would \n> like to see is a release target for 7.2.2. There is a substantial interest \n> in our userbase for this, judging from a thread among OpenACS developers. \n> Thus far it's not hit bigger news sites -- but I can tell you right now, with \n> my media experience, that people will assume the worst -- and if/when this \n> hits slashdot, we really need a response to 'whence 7.2.2'. BugTraq's one \n> thing -- slashdot is another thing entirely. I'm quite surprized, in fact, \n> it hasn't hit Linux Today, slashdot, or Linux Weekly News. When/if it does, \n> *SHA-ZAM*. I'd personally like the news item to read 'PostgreSQL developers \n> release security update' rather than 'Security hole found in PostgreSQL'.\n> \n> If push comes to shove I can push RPM's out Saturday, if the tarball is ready. \n> If not, Monday morning at the earliest.\n\nYou bring up a good point. We don't want to appear reactive on this, we\nwant to be proactive. The CVS is all ready for release, so there isn't\nanything holding us up except our quality control.\n\n> > > Hey, Bruce, gyroscopes spin to keep them stable.....\n> \n> > That doesn't seem to work for me.\n> \n> Well, we've already seen kudos go out to Marc. I think a round of applause \n> should also go to some other people -- but in particular Bruce has always \n> done a fine job of getting the release ready for packaging. Not to take away \n> from other's efforts, just maybe to help encourage someone else who is \n> feeling overwhelmed (a couple hundred thousand lines of Fortran, a complete \n> website redesign, a major renovation, and some major engineering/surveying \n> work have my attention here... and that's just one job).\n\nThanks. It is just that I usually have control over all the open threads\nso I can know they are all tied up. Right now, I have a mailbox full of\n1/2 discussed items that haven't been resolved. I will need to start\nreading all of them tomorrow and try to get closure on them.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 23:07:13 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: My head is spinning" }, { "msg_contents": "On Thursday 22 August 2002 11:07 pm, Bruce Momjian wrote:\n> Lamar Owen wrote:\n> > Weekly News. When/if it does, *SHA-ZAM*. I'd personally like the news\n> > item to read 'PostgreSQL developers release security update' rather than\n> > 'Security hole found in PostgreSQL'.\n\n> You bring up a good point. We don't want to appear reactive on this, we\n> want to be proactive. The CVS is all ready for release, so there isn't\n> anything holding us up except our quality control.\n\nAs to QA, I have REL7_2_STABLE coming down my dialup now. I'll try a test \nbuild (given the CVS versus dist tarball quirks -- I really need to duplicate \nthe distribution scripts Marc has so that I can preroll tarballs from CVS \nhere...) tonight or tomorrow and see where it leads. My biggest difficulty \nis merging any necessary RPM-specific patches -- and I have a couple of new \nones to put in, particularly a set of ones to contrib from Peter that I've \nnot yet merged in.\n\n> Thanks. It is just that I usually have control over all the open threads\n> so I can know they are all tied up. Right now, I have a mailbox full of\n> 1/2 discussed items that haven't been resolved. I will need to start\n> reading all of them tomorrow and try to get closure on them.\n\nKeeping 7.3 and 7.2.2 issues separate is going to be a challenge -- but we \nalways knew there'd be this possibility. I _know_ my build environment isn't \nset up to parallel QA releases. And with Sep 1 nearing, I really need to get \nmy act together on building pre-7.3... \n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Thu, 22 Aug 2002 23:18:31 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: My head is spinning" }, { "msg_contents": "On Thursday 22 August 2002 11:07 pm, Bruce Momjian wrote:\n> Lamar Owen wrote:\n> > Weekly News. When/if it does, *SHA-ZAM*. I'd personally like the news\n> > item to read 'PostgreSQL developers release security update' rather than\n> > 'Security hole found in PostgreSQL'.\n\n> You bring up a good point. We don't want to appear reactive on this, we\n> want to be proactive. The CVS is all ready for release, so there isn't\n> anything holding us up except our quality control.\n\nWell, it _has_ been placed on LWN under the moniker 'Multiple buffer overflows \nin PostgreSQL'. Not good, as everyone who claims to be a security expert \n*knows* that all buffer overflows are the bane of a secure system.... It \nisn't a headline item, though -- which is why it took me a minute to find it. \n\nBut even LWN isn't slashdot. It would be nice if we've dodged that bullet.\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Thu, 22 Aug 2002 23:32:34 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: My head is spinning" }, { "msg_contents": "Lamar Owen <lamar.owen@wgcr.org> writes:\n> BugTraq's one thing -- slashdot is another thing entirely. I'm\n> quite surprized, in fact, it hasn't hit Linux Today, slashdot, or\n> Linux Weekly News.\n\nI'd be very surprised if it is mentioned on Slashdot (LWN + Linuxtoday\ncarry routine security advisories, so they'll probably have it at some\npoint). The security problem is *not* that serious.\n\n> If push comes to shove I can push RPM's out Saturday, if the tarball\n> is ready. If not, Monday morning at the earliest.\n\nI'd say release the RPMs when they are ready, a couple days won't make\na big difference either way. Given that the only remotely serious hole\n(the datetime bug) has been public knowledge for a matter of weeks,\nthere's not a lot of point to panicking at this point.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "22 Aug 2002 23:42:01 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: My head is spinning" }, { "msg_contents": "On Thursday 22 August 2002 11:42 pm, Neil Conway wrote:\n> Lamar Owen <lamar.owen@wgcr.org> writes:\n> > If push comes to shove I can push RPM's out Saturday, if the tarball\n> > is ready. If not, Monday morning at the earliest.\n\n> I'd say release the RPMs when they are ready, a couple days won't make\n> a big difference either way. Given that the only remotely serious hole\n> (the datetime bug) has been public knowledge for a matter of weeks,\n> there's not a lot of point to panicking at this point.\n\nOh, I'm not panicking. I just try my best to release RPM's as close to \ncoincident to the tarball release as possible. Otherwise I get complaints.\n\nAnd you're right -- the security issue itself isn't that serious (as I have \nposted to both BugTraq and the OpenACS forum) -- but remember the mindset of \nthe typical slashdot reader.\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Thu, 22 Aug 2002 23:49:11 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: My head is spinning" }, { "msg_contents": "On Thu, 22 Aug 2002, Bruce Momjian wrote:\n\n> Lamar Owen wrote:\n> > Hey, Bruce, gyroscopes spin to keep them stable.....\n>\n> That doesn't seem to work for me.\n\nNot spinning fast enough? Hey, I know. Let's stand him on his head!\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n", "msg_date": "Fri, 23 Aug 2002 05:45:52 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": false, "msg_subject": "Re: My head is spinning" } ]
[ { "msg_contents": "\nJust so that everyone knows, Dave Page created a project on GBorg for the\nODBC driver, and I've just finished moving it over and removing it from\nthe main tree ...\n\n2 down ... so many more to go :)\n\n", "msg_date": "Thu, 22 Aug 2002 19:46:03 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "ODBC Driver moved to GBorg ..." }, { "msg_contents": "I must be blind. I don't see links to gborg anywhere on the developer\nor main site web pages. Perhaps more obvious a sister site link would\nbe of value. \n\nThe only link I found was under \"User's Lounge\" and then \"PostgreSQL\nRelated Projects\". Perhaps extra link emphasis is needed since a\nseemingly new level of demand is being placed on gborg's significance?\n\nSign,\n\n\tGreg Copeland\n\n\nOn Thu, 2002-08-22 at 17:46, Marc G. Fournier wrote:\n> \n> Just so that everyone knows, Dave Page created a project on GBorg for the\n> ODBC driver, and I've just finished moving it over and removing it from\n> the main tree ...\n> \n> 2 down ... so many more to go :)\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster", "msg_date": "23 Aug 2002 07:47:56 -0500", "msg_from": "Greg Copeland <greg@CopelandConsulting.Net>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] ODBC Driver moved to GBorg ..." }, { "msg_contents": "On Thu, 22 Aug 2002 19:46:03 -0300 (ADT), Marc G. Fournier wrote:\n\n>Just so that everyone knows, Dave Page created a project on GBorg for the\n>ODBC driver, and I've just finished moving it over and removing it from\n>the main tree ...\n>\n>2 down ... so many more to go :)\n>\n\nSo now I don't have anymore access to cvs. Just registering at GBorg\nisn't sufficient. If IIRC I have to join the team to get access at GBorg.\nBut I'm not a developer...\n:-(\n\nregards\n\n\nJohann Zuschlag\nzuschlag2@online.de\n\n\n", "msg_date": "Fri, 23 Aug 2002 22:05:10 +0200", "msg_from": "\"Johann Zuschlag\" <zuschlag2@online.de>", "msg_from_op": false, "msg_subject": "Re: ODBC Driver moved to GBorg ..." }, { "msg_contents": "\nThis is all in Vince's area ...\n\n\nOn 23 Aug 2002, Greg Copeland wrote:\n\n> I must be blind. I don't see links to gborg anywhere on the developer\n> or main site web pages. Perhaps more obvious a sister site link would\n> be of value.\n>\n> The only link I found was under \"User's Lounge\" and then \"PostgreSQL\n> Related Projects\". Perhaps extra link emphasis is needed since a\n> seemingly new level of demand is being placed on gborg's significance?\n>\n> Sign,\n>\n> \tGreg Copeland\n>\n>\n> On Thu, 2002-08-22 at 17:46, Marc G. Fournier wrote:\n> >\n> > Just so that everyone knows, Dave Page created a project on GBorg for the\n> > ODBC driver, and I've just finished moving it over and removing it from\n> > the main tree ...\n> >\n> > 2 down ... so many more to go :)\n> >\n> >\n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 4: Don't 'kill -9' the postmaster\n>\n>\n\n", "msg_date": "Sat, 31 Aug 2002 21:50:46 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: [ODBC] [HACKERS] ODBC Driver moved to GBorg ..." }, { "msg_contents": "On Sat, 31 Aug 2002, Marc G. Fournier wrote:\n\n>\n> This is all in Vince's area ...\n\nYou're right. You don't see 'em.\n\n>\n>\n> On 23 Aug 2002, Greg Copeland wrote:\n>\n> > I must be blind. I don't see links to gborg anywhere on the developer\n> > or main site web pages. Perhaps more obvious a sister site link would\n> > be of value.\n> >\n> > The only link I found was under \"User's Lounge\" and then \"PostgreSQL\n> > Related Projects\". Perhaps extra link emphasis is needed since a\n> > seemingly new level of demand is being placed on gborg's significance?\n> >\n> > Sign,\n> >\n> > \tGreg Copeland\n> >\n> >\n> > On Thu, 2002-08-22 at 17:46, Marc G. Fournier wrote:\n> > >\n> > > Just so that everyone knows, Dave Page created a project on GBorg for the\n> > > ODBC driver, and I've just finished moving it over and removing it from\n> > > the main tree ...\n> > >\n> > > 2 down ... so many more to go :)\n> > >\n> > >\n> > > ---------------------------(end of broadcast)---------------------------\n> > > TIP 4: Don't 'kill -9' the postmaster\n> >\n> >\n>\n>\n\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n", "msg_date": "Mon, 2 Sep 2002 20:17:21 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": false, "msg_subject": "Re: [ODBC] ODBC Driver moved to GBorg ..." }, { "msg_contents": "On Sat, 31 Aug 2002, Marc G. Fournier wrote:\n\n>\n> This is all in Vince's area ...\n\nCorrection. You're not looking, it's in the users lounge. We've\ncovered the button thing already.\n\n>\n>\n> On 23 Aug 2002, Greg Copeland wrote:\n>\n> > I must be blind. I don't see links to gborg anywhere on the developer\n> > or main site web pages. Perhaps more obvious a sister site link would\n> > be of value.\n> >\n> > The only link I found was under \"User's Lounge\" and then \"PostgreSQL\n> > Related Projects\". Perhaps extra link emphasis is needed since a\n> > seemingly new level of demand is being placed on gborg's significance?\n> >\n> > Sign,\n> >\n> > \tGreg Copeland\n> >\n> >\n> > On Thu, 2002-08-22 at 17:46, Marc G. Fournier wrote:\n> > >\n> > > Just so that everyone knows, Dave Page created a project on GBorg for the\n> > > ODBC driver, and I've just finished moving it over and removing it from\n> > > the main tree ...\n> > >\n> > > 2 down ... so many more to go :)\n> > >\n> > >\n> > > ---------------------------(end of broadcast)---------------------------\n> > > TIP 4: Don't 'kill -9' the postmaster\n> >\n> >\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\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n", "msg_date": "Mon, 2 Sep 2002 20:20:14 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": false, "msg_subject": "Re: [ODBC] ODBC Driver moved to GBorg ..." }, { "msg_contents": "Hot sure if this is the right list, but. . .\n\nOn Mon, Sep 02, 2002 at 08:20:14PM -0400, Vince Vielhaber wrote:\n> On Sat, 31 Aug 2002, Marc G. Fournier wrote:\n> \n> >\n> > This is all in Vince's area ...\n> \n> Correction. You're not looking, it's in the users lounge. We've\n> covered the button thing already.\n> \n> >\n> >\n> > On 23 Aug 2002, Greg Copeland wrote:\n> >\n> > > I must be blind. I don't see links to gborg anywhere on the developer\n> > > or main site web pages. Perhaps more obvious a sister site link would\n> > > be of value.\n\n. . .given that so many projects are being moved out of the tree, and\ngiven that everyone keeps talking about \"gborg\" on the list, would it\nbe possible to change the link name from \"PostgreSQL Related\nProjects\" to \"PostgreSQL Related Projects ('gborg')\". Users are\ngoing to go looking for \"gborg\", because that's what everyone calls\nit. It'd be nice to make it real easy to find.\n\nA\n\n-- \n----\nAndrew Sullivan 204-4141 Yonge Street\nLiberty RMS Toronto, Ontario Canada\n<andrew@libertyrms.info> M2P 2A8\n +1 416 646 3304 x110\n\n", "msg_date": "Wed, 4 Sep 2002 13:56:46 -0400", "msg_from": "Andrew Sullivan <andrew@libertyrms.info>", "msg_from_op": false, "msg_subject": "Re: [ODBC] ODBC Driver moved to GBorg ..." } ]
[ { "msg_contents": "Hi,\n\nMy primary use of Postgres is as the backend database for a busy web site.\nWe have a cron job that just emails us the tail of our database, php, apache\nlogs every night. That way we can see some problems.\n\nThese logs almost always contain some errors. For instance, this is what I\nsee at the moment:\n\n2002-08-22 19:21:57 ERROR: pg_atoi: error in \"334 - 18k\": can't parse \" -\n18k\"\n\nNow there's plenty of places that accept numeric input in the site and\nobviously there's a bug in some script somewhere that's not filtering the\ninput properly or something. However - the error message above is useless\nto me!!!\n\nSo, what I'd like to propose is a new GUC variable called\n'debug_print_query_on_error' or something. Instead of turning on\ndebug_print_query and having my logs totally spammed up with sql, this GUC\nvariable would only print the query if an actual ERROR occurred. This way I\ncould nail the error very quickly by simply finding the query in my\ncodebase.\n\nIs this possible? At the stage of processing where the elog(ERROR) occurs,\ndo we still have access to the original query string?\n\nComments?\n\nChris\n\n", "msg_date": "Fri, 23 Aug 2002 10:04:50 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "Proposed GUC Variable" }, { "msg_contents": "\nSomeone asked for that recently, and the email is in my mailbox for\nconsideration. I think it is a great idea, and we have\ndebug_query_string that holds the current query. You could grab that\nfrom elog.c. Added to TODO:\n\n\t* Add GUC parameter to print queries that generate errors \n\n---------------------------------------------------------------------------\n\nChristopher Kings-Lynne wrote:\n> Hi,\n> \n> My primary use of Postgres is as the backend database for a busy web site.\n> We have a cron job that just emails us the tail of our database, php, apache\n> logs every night. That way we can see some problems.\n> \n> These logs almost always contain some errors. For instance, this is what I\n> see at the moment:\n> \n> 2002-08-22 19:21:57 ERROR: pg_atoi: error in \"334 - 18k\": can't parse \" -\n> 18k\"\n> \n> Now there's plenty of places that accept numeric input in the site and\n> obviously there's a bug in some script somewhere that's not filtering the\n> input properly or something. However - the error message above is useless\n> to me!!!\n> \n> So, what I'd like to propose is a new GUC variable called\n> 'debug_print_query_on_error' or something. Instead of turning on\n> debug_print_query and having my logs totally spammed up with sql, this GUC\n> variable would only print the query if an actual ERROR occurred. This way I\n> could nail the error very quickly by simply finding the query in my\n> codebase.\n> \n> Is this possible? At the stage of processing where the elog(ERROR) occurs,\n> do we still have access to the original query string?\n> \n> Comments?\n> \n> Chris\n> \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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 22 Aug 2002 22:46:31 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "Hi all,\n\nQuick hack while eating a sandwich.\n\ntemplate1=# select * frum;\nERROR: parser: parse error at or near \"frum\" at character 10\nERROR QUERY: select * frum;\n\nNow, I did say quick hack. 'ERROR QUERY' isn't a new error level I just\nstrcat() it to buf_msg in elog() if debug_print_error_query is\ntrue. Question: from Chris's request it doesn't sound like there is much\nuse writing this to the client. Does everyone else feel the same way?\n\nIf so, I'll patch it up and send off.\n\nGavin\n\nOn Thu, 22 Aug 2002, Bruce Momjian wrote:\n\n> \n> Someone asked for that recently, and the email is in my mailbox for\n> consideration. I think it is a great idea, and we have\n> debug_query_string that holds the current query. You could grab that\n> from elog.c. Added to TODO:\n> \n> \t* Add GUC parameter to print queries that generate errors \n> \n> ---------------------------------------------------------------------------\n> \n> Christopher Kings-Lynne wrote:\n> > Hi,\n> > \n> > My primary use of Postgres is as the backend database for a busy web site.\n> > We have a cron job that just emails us the tail of our database, php, apache\n> > logs every night. That way we can see some problems.\n> > \n> > These logs almost always contain some errors. For instance, this is what I\n> > see at the moment:\n> > \n> > 2002-08-22 19:21:57 ERROR: pg_atoi: error in \"334 - 18k\": can't parse \" -\n> > 18k\"\n> > \n> > Now there's plenty of places that accept numeric input in the site and\n> > obviously there's a bug in some script somewhere that's not filtering the\n> > input properly or something. However - the error message above is useless\n> > to me!!!\n> > \n> > So, what I'd like to propose is a new GUC variable called\n> > 'debug_print_query_on_error' or something. Instead of turning on\n> > debug_print_query and having my logs totally spammed up with sql, this GUC\n> > variable would only print the query if an actual ERROR occurred. This way I\n> > could nail the error very quickly by simply finding the query in my\n> > codebase.\n> > \n> > Is this possible? At the stage of processing where the elog(ERROR) occurs,\n> > do we still have access to the original query string?\n> > \n> > Comments?\n> > \n> > Chris\n> > \n> > \n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 4: Don't 'kill -9' the postmaster\n> > \n> \n> \n\n", "msg_date": "Fri, 23 Aug 2002 13:39:35 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Someone asked for that recently, and the email is in my mailbox for\n> consideration. I think it is a great idea, and we have\n> debug_query_string that holds the current query.\n\ndebug_query_string doesn't necessarily have a lot to do with the\nproximate cause of the error. Consider queries issued by rules,\ntriggers, plpgsql functions, etc.\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 23 Aug 2002 00:20:19 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable " }, { "msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Someone asked for that recently, and the email is in my mailbox for\n> > consideration. I think it is a great idea, and we have\n> > debug_query_string that holds the current query.\n> \n> debug_query_string doesn't necessarily have a lot to do with the\n> proximate cause of the error. Consider queries issued by rules,\n> triggers, plpgsql functions, etc.\n\nMaybe. I think giving the user the string that caused the error is\nprobably what they want, rather than a rule or trigger that they can't\ntie back to an actual query.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Fri, 23 Aug 2002 00:24:59 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "> > debug_query_string doesn't necessarily have a lot to do with the\n> > proximate cause of the error. Consider queries issued by rules,\n> > triggers, plpgsql functions, etc.\n> \n> Maybe. I think giving the user the string that caused the error is\n> probably what they want, rather than a rule or trigger that they can't\n> tie back to an actual query.\n\nYeah, I'd agree with that.\n\nChris\n\n", "msg_date": "Fri, 23 Aug 2002 12:26:52 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "> Quick hack while eating a sandwich.\n>\n> template1=# select * frum;\n> ERROR: parser: parse error at or near \"frum\" at character 10\n> ERROR QUERY: select * frum;\n>\n> Now, I did say quick hack. 'ERROR QUERY' isn't a new error level I just\n> strcat() it to buf_msg in elog() if debug_print_error_query is\n> true. Question: from Chris's request it doesn't sound like there is much\n> use writing this to the client. Does everyone else feel the same way?\n>\n> If so, I'll patch it up and send off.\n\nHow about the ERROR occurs and is sent to client and logged. Then you do\nanother elog(LOG, querystring) sort of thing. That way you won't confuse\nclients that are parsing the messages but those interested in the log text\ncan read it quite happily.\n\nThat'd be my preferred solution...\n\nThanks for working on this BTW gavin.\n\nChris\n\n", "msg_date": "Fri, 23 Aug 2002 14:46:50 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "On Fri, 2002-08-23 at 04:46, Bruce Momjian wrote:\n> \n> \t* Add GUC parameter to print queries that generate errors \n\nMaybe don't make that an option, but rather do it always. I don't see\nwhere that would hurt. And killing configuration options that are\nunneeded is always a Good Thing.\n\n-- \nMarkus Bertheau.\nBerlin, Berlin.\nGermany.\n\n", "msg_date": "23 Aug 2002 09:54:57 +0200", "msg_from": "Markus Bertheau <twanger@bluetwanger.de>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "So long as the change is only evident in the log, I agree.\n\nChris\n\n> -----Original Message-----\n> From: pgsql-hackers-owner@postgresql.org\n> [mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Markus Bertheau\n> Sent: Friday, 23 August 2002 3:55 PM\n> To: Bruce Momjian\n> Cc: Christopher Kings-Lynne; Hackers\n> Subject: Re: [HACKERS] Proposed GUC Variable\n> \n> \n> On Fri, 2002-08-23 at 04:46, Bruce Momjian wrote:\n> > \n> > \t* Add GUC parameter to print queries that generate errors \n> \n> Maybe don't make that an option, but rather do it always. I don't see\n> where that would hurt. And killing configuration options that are\n> unneeded is always a Good Thing.\n> \n> -- \n> Markus Bertheau.\n> Berlin, Berlin.\n> Germany.\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n> \n\n", "msg_date": "Fri, 23 Aug 2002 17:13:08 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "\nThis is an intersting mix of features. First, use LOG so it goes only\nto the log file by default _but_ can be turned on to be seen by the\nclient, and remove the GUC completely.\n\nThis does make 100% sense because an ERROR is going to go the client and\nthe server logs by default, and LOG is going to go to the server logs by\ndefault _but_ can optionally be seen by the client by modifying\nmin_client_messagses.\n\n---------------------------------------------------------------------------\n\nChristopher Kings-Lynne wrote:\n> So long as the change is only evident in the log, I agree.\n> \n> Chris\n> \n> > -----Original Message-----\n> > From: pgsql-hackers-owner@postgresql.org\n> > [mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Markus Bertheau\n> > Sent: Friday, 23 August 2002 3:55 PM\n> > To: Bruce Momjian\n> > Cc: Christopher Kings-Lynne; Hackers\n> > Subject: Re: [HACKERS] Proposed GUC Variable\n> > \n> > \n> > On Fri, 2002-08-23 at 04:46, Bruce Momjian wrote:\n> > > \n> > > \t* Add GUC parameter to print queries that generate errors \n> > \n> > Maybe don't make that an option, but rather do it always. I don't see\n> > where that would hurt. And killing configuration options that are\n> > unneeded is always a Good Thing.\n> > \n> > -- \n> > Markus Bertheau.\n> > Berlin, Berlin.\n> > Germany.\n> > \n> > \n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 6: Have you searched our list archives?\n> > \n> > http://archives.postgresql.org\n> > \n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Fri, 23 Aug 2002 11:24:19 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "Markus Bertheau <twanger@bluetwanger.de> writes:\n> On Fri, 2002-08-23 at 04:46, Bruce Momjian wrote:\n> > \n> > \t* Add GUC parameter to print queries that generate errors\n> \n> Maybe don't make that an option, but rather do it always. I don't\n> see where that would hurt.\n\nIt would hurt in situations in which an error is expected to occur\n(e.g. INSERT into table, if constraint violation then do XYZ). It\nwould also make the logs a lot larger + less readable if long (say,\n100 line) queries are being executed.\n\nI don't think either situation is particularly common, but I do think\nit's worth keeping a GUC variable for it. The GUC var should probably\ndefault to being enabled though.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "23 Aug 2002 13:56:44 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "\nI had an idea on this. It seems pretty pointless to show a query error\nwithout a query, but some queries are very large.\n\nHow about if we print only the first 80 characters of the query, with\nnewlines, tabs, and spaces reduced to a single space, and send that as\nLOG to the server logs. That would give people enough context, and\nprevent us from having another GUC variable.\n\n---------------------------------------------------------------------------\n\nGavin Sherry wrote:\n> Hi all,\n> \n> Quick hack while eating a sandwich.\n> \n> template1=# select * frum;\n> ERROR: parser: parse error at or near \"frum\" at character 10\n> ERROR QUERY: select * frum;\n> \n> Now, I did say quick hack. 'ERROR QUERY' isn't a new error level I just\n> strcat() it to buf_msg in elog() if debug_print_error_query is\n> true. Question: from Chris's request it doesn't sound like there is much\n> use writing this to the client. Does everyone else feel the same way?\n> \n> If so, I'll patch it up and send off.\n> \n> Gavin\n> \n> On Thu, 22 Aug 2002, Bruce Momjian wrote:\n> \n> > \n> > Someone asked for that recently, and the email is in my mailbox for\n> > consideration. I think it is a great idea, and we have\n> > debug_query_string that holds the current query. You could grab that\n> > from elog.c. Added to TODO:\n> > \n> > \t* Add GUC parameter to print queries that generate errors \n> > \n> > ---------------------------------------------------------------------------\n> > \n> > Christopher Kings-Lynne wrote:\n> > > Hi,\n> > > \n> > > My primary use of Postgres is as the backend database for a busy web site.\n> > > We have a cron job that just emails us the tail of our database, php, apache\n> > > logs every night. That way we can see some problems.\n> > > \n> > > These logs almost always contain some errors. For instance, this is what I\n> > > see at the moment:\n> > > \n> > > 2002-08-22 19:21:57 ERROR: pg_atoi: error in \"334 - 18k\": can't parse \" -\n> > > 18k\"\n> > > \n> > > Now there's plenty of places that accept numeric input in the site and\n> > > obviously there's a bug in some script somewhere that's not filtering the\n> > > input properly or something. However - the error message above is useless\n> > > to me!!!\n> > > \n> > > So, what I'd like to propose is a new GUC variable called\n> > > 'debug_print_query_on_error' or something. Instead of turning on\n> > > debug_print_query and having my logs totally spammed up with sql, this GUC\n> > > variable would only print the query if an actual ERROR occurred. This way I\n> > > could nail the error very quickly by simply finding the query in my\n> > > codebase.\n> > > \n> > > Is this possible? At the stage of processing where the elog(ERROR) occurs,\n> > > do we still have access to the original query string?\n> > > \n> > > Comments?\n> > > \n> > > Chris\n> > > \n> > > \n> > > ---------------------------(end of broadcast)---------------------------\n> > > TIP 4: Don't 'kill -9' the postmaster\n> > > \n> > \n> > \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> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 16:54:24 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "On Tue, 2002-08-27 at 15:54, Bruce Momjian wrote:\n> \n> I had an idea on this. It seems pretty pointless to show a query error\n> without a query, but some queries are very large.\n> \n> How about if we print only the first 80 characters of the query, with\n> newlines, tabs, and spaces reduced to a single space, and send that as\n> LOG to the server logs. That would give people enough context, and\n> prevent us from having another GUC variable.\nNot necessarily giving enough context. I know I've had program\ngenerated query's that were syntactically invalid WAY after the 80th\ncharacter. \n\nIf you print ANY of the query, you should print all of it. Look at the\ncode in elog.c that does the syslog splitting. \n\nLER\n> \n> ---------------------------------------------------------------------------\n> \n> Gavin Sherry wrote:\n> > Hi all,\n> > \n> > Quick hack while eating a sandwich.\n> > \n> > template1=# select * frum;\n> > ERROR: parser: parse error at or near \"frum\" at character 10\n> > ERROR QUERY: select * frum;\n> > \n> > Now, I did say quick hack. 'ERROR QUERY' isn't a new error level I just\n> > strcat() it to buf_msg in elog() if debug_print_error_query is\n> > true. Question: from Chris's request it doesn't sound like there is much\n> > use writing this to the client. Does everyone else feel the same way?\n> > \n> > If so, I'll patch it up and send off.\n> > \n> > Gavin\n> > \n> > On Thu, 22 Aug 2002, Bruce Momjian wrote:\n> > \n> > > \n> > > Someone asked for that recently, and the email is in my mailbox for\n> > > consideration. I think it is a great idea, and we have\n> > > debug_query_string that holds the current query. You could grab that\n> > > from elog.c. Added to TODO:\n> > > \n> > > \t* Add GUC parameter to print queries that generate errors \n> > > \n> > > ---------------------------------------------------------------------------\n> > > \n> > > Christopher Kings-Lynne wrote:\n> > > > Hi,\n> > > > \n> > > > My primary use of Postgres is as the backend database for a busy web site.\n> > > > We have a cron job that just emails us the tail of our database, php, apache\n> > > > logs every night. That way we can see some problems.\n> > > > \n> > > > These logs almost always contain some errors. For instance, this is what I\n> > > > see at the moment:\n> > > > \n> > > > 2002-08-22 19:21:57 ERROR: pg_atoi: error in \"334 - 18k\": can't parse \" -\n> > > > 18k\"\n> > > > \n> > > > Now there's plenty of places that accept numeric input in the site and\n> > > > obviously there's a bug in some script somewhere that's not filtering the\n> > > > input properly or something. However - the error message above is useless\n> > > > to me!!!\n> > > > \n> > > > So, what I'd like to propose is a new GUC variable called\n> > > > 'debug_print_query_on_error' or something. Instead of turning on\n> > > > debug_print_query and having my logs totally spammed up with sql, this GUC\n> > > > variable would only print the query if an actual ERROR occurred. This way I\n> > > > could nail the error very quickly by simply finding the query in my\n> > > > codebase.\n> > > > \n> > > > Is this possible? At the stage of processing where the elog(ERROR) occurs,\n> > > > do we still have access to the original query string?\n> > > > \n> > > > Comments?\n> > > > \n> > > > Chris\n> > > > \n> > > > \n> > > > ---------------------------(end of broadcast)---------------------------\n> > > > TIP 4: Don't 'kill -9' the postmaster\n> > > > \n> > > \n> > > \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> > \n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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-- \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\n\n", "msg_date": "27 Aug 2002 15:57:24 -0500", "msg_from": "Larry Rosenman <ler@lerctr.org>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "On Tue, 2002-08-27 at 16:54, Bruce Momjian wrote:\n> \n> I had an idea on this. It seems pretty pointless to show a query error\n> without a query, but some queries are very large.\n> \n> How about if we print only the first 80 characters of the query, with\n> newlines, tabs, and spaces reduced to a single space, and send that as\n> LOG to the server logs. That would give people enough context, and\n> prevent us from having another GUC variable.\n\nI could go for the first 1000 characters, but 80 is almost useless for\nmost of our stuff. 80 wouldn't get through the select list a good chunk\nof the time.\n\nIf an application in the product environment is throwing an error, we'd\nwant the full thing. Most of our internal systems are completely hands\noff unless it's been scripted and tested elsewhere, so it's not like\nuser queries would be getting into it.\n\nPerhaps a GUC for the length? But I'd still opt for storing the whole\nthing. Yes, someone could fill up the disk but a rotating log would\nhelp that.\n\n> ---------------------------------------------------------------------------\n> \n> Gavin Sherry wrote:\n> > Hi all,\n> > \n> > Quick hack while eating a sandwich.\n> > \n> > template1=# select * frum;\n> > ERROR: parser: parse error at or near \"frum\" at character 10\n> > ERROR QUERY: select * frum;\n> > \n> > Now, I did say quick hack. 'ERROR QUERY' isn't a new error level I just\n> > strcat() it to buf_msg in elog() if debug_print_error_query is\n> > true. Question: from Chris's request it doesn't sound like there is much\n> > use writing this to the client. Does everyone else feel the same way?\n> > \n> > If so, I'll patch it up and send off.\n> > \n> > Gavin\n> > \n> > On Thu, 22 Aug 2002, Bruce Momjian wrote:\n> > \n> > > \n> > > Someone asked for that recently, and the email is in my mailbox for\n> > > consideration. I think it is a great idea, and we have\n> > > debug_query_string that holds the current query. You could grab that\n> > > from elog.c. Added to TODO:\n> > > \n> > > \t* Add GUC parameter to print queries that generate errors \n> > > \n> > > ---------------------------------------------------------------------------\n> > > \n> > > Christopher Kings-Lynne wrote:\n> > > > Hi,\n> > > > \n> > > > My primary use of Postgres is as the backend database for a busy web site.\n> > > > We have a cron job that just emails us the tail of our database, php, apache\n> > > > logs every night. That way we can see some problems.\n> > > > \n> > > > These logs almost always contain some errors. For instance, this is what I\n> > > > see at the moment:\n> > > > \n> > > > 2002-08-22 19:21:57 ERROR: pg_atoi: error in \"334 - 18k\": can't parse \" -\n> > > > 18k\"\n> > > > \n> > > > Now there's plenty of places that accept numeric input in the site and\n> > > > obviously there's a bug in some script somewhere that's not filtering the\n> > > > input properly or something. However - the error message above is useless\n> > > > to me!!!\n> > > > \n> > > > So, what I'd like to propose is a new GUC variable called\n> > > > 'debug_print_query_on_error' or something. Instead of turning on\n> > > > debug_print_query and having my logs totally spammed up with sql, this GUC\n> > > > variable would only print the query if an actual ERROR occurred. This way I\n> > > > could nail the error very quickly by simply finding the query in my\n> > > > codebase.\n> > > > \n> > > > Is this possible? At the stage of processing where the elog(ERROR) occurs,\n> > > > do we still have access to the original query string?\n> > > > \n> > > > Comments?\n> > > > \n> > > > Chris\n> > > > \n> > > > \n> > > > ---------------------------(end of broadcast)---------------------------\n> > > > TIP 4: Don't 'kill -9' the postmaster\n> > > > \n> > > \n> > > \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> > \n> \n> -- \n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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", "msg_date": "27 Aug 2002 17:03:17 -0400", "msg_from": "Rod Taylor <rbt@zort.ca>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "Larry Rosenman wrote:\n> On Tue, 2002-08-27 at 15:54, Bruce Momjian wrote:\n> > \n> > I had an idea on this. It seems pretty pointless to show a query error\n> > without a query, but some queries are very large.\n> > \n> > How about if we print only the first 80 characters of the query, with\n> > newlines, tabs, and spaces reduced to a single space, and send that as\n> > LOG to the server logs. That would give people enough context, and\n> > prevent us from having another GUC variable.\n> Not necessarily giving enough context. I know I've had program\n> generated query's that were syntactically invalid WAY after the 80th\n> character. \n> \n> If you print ANY of the query, you should print all of it. Look at the\n> code in elog.c that does the syslog splitting. \n\nBut we should have some default to print some of the query, because\nright now we print none of it. I am not saying it is perfect, but it is\nbetter than what we have, and is a reasonable default.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 17:05:33 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "Rod Taylor wrote:\n> On Tue, 2002-08-27 at 16:54, Bruce Momjian wrote:\n> > \n> > I had an idea on this. It seems pretty pointless to show a query error\n> > without a query, but some queries are very large.\n> > \n> > How about if we print only the first 80 characters of the query, with\n> > newlines, tabs, and spaces reduced to a single space, and send that as\n> > LOG to the server logs. That would give people enough context, and\n> > prevent us from having another GUC variable.\n> \n> I could go for the first 1000 characters, but 80 is almost useless for\n> most of our stuff. 80 wouldn't get through the select list a good chunk\n> of the time.\n> \n> If an application in the product environment is throwing an error, we'd\n> want the full thing. Most of our internal systems are completely hands\n> off unless it's been scripted and tested elsewhere, so it's not like\n> user queries would be getting into it.\n> \n> Perhaps a GUC for the length? But I'd still opt for storing the whole\n> thing. Yes, someone could fill up the disk but a rotating log would\n> help that.\n\nA length value would work, default to 80 and let someone turn it off\nwith zero and unlimited with 9999 or -1.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 17:06:48 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "On Tue, 2002-08-27 at 16:05, Bruce Momjian wrote:\n> Larry Rosenman wrote:\n> > On Tue, 2002-08-27 at 15:54, Bruce Momjian wrote:\n> > > \n> > > I had an idea on this. It seems pretty pointless to show a query error\n> > > without a query, but some queries are very large.\n> > > \n> > > How about if we print only the first 80 characters of the query, with\n> > > newlines, tabs, and spaces reduced to a single space, and send that as\n> > > LOG to the server logs. That would give people enough context, and\n> > > prevent us from having another GUC variable.\n> > Not necessarily giving enough context. I know I've had program\n> > generated query's that were syntactically invalid WAY after the 80th\n> > character. \n> > \n> > If you print ANY of the query, you should print all of it. Look at the\n> > code in elog.c that does the syslog splitting. \n> \n> But we should have some default to print some of the query, because\n> right now we print none of it. I am not saying it is perfect, but it is\n> better than what we have, and is a reasonable default.\nOn an error, you may not be able to reproduce it. Why not print the\nwhole query to the log?\n\nI don't see a reason for truncating it at 80 chars. \n\nIMHO, of course. \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\n\n", "msg_date": "27 Aug 2002 16:08:16 -0500", "msg_from": "Larry Rosenman <ler@lerctr.org>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "Larry Rosenman wrote:\n> > But we should have some default to print some of the query, because\n> > right now we print none of it. I am not saying it is perfect, but it is\n> > better than what we have, and is a reasonable default.\n> On an error, you may not be able to reproduce it. Why not print the\n> whole query to the log?\n> \n> I don't see a reason for truncating it at 80 chars. \n> \n> IMHO, of course. \n\nBecause every typo query, every syntax error of a user in psql would\nappear in your logs. That seems excessive. Already the ERROR line\nappears in the logs. Do we want to see their bad query too?\n\nMy concern is that long queries could easily bulk up the logs to the\npoint where the actual important log messages would be lost in the fog.\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 17:14:43 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "On Tue, 2002-08-27 at 16:14, Bruce Momjian wrote:\n> Larry Rosenman wrote:\n> > > But we should have some default to print some of the query, because\n> > > right now we print none of it. I am not saying it is perfect, but it is\n> > > better than what we have, and is a reasonable default.\n> > On an error, you may not be able to reproduce it. Why not print the\n> > whole query to the log?\n> > \n> > I don't see a reason for truncating it at 80 chars. \n> > \n> > IMHO, of course. \n> \n> Because every typo query, every syntax error of a user in psql would\n> appear in your logs. That seems excessive. Already the ERROR line\n> appears in the logs. Do we want to see their bad query too?\n> \n> My concern is that long queries could easily bulk up the logs to the\n> point where the actual important log messages would be lost in the fog.\nHmm. I think the 80 should be a GUC variable (and also settable from\nSQL SET as well), and the 80 should probably be higher. \n\nAnd, maybe send the full query at a different Syslog(3) level.\n\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\n\n", "msg_date": "27 Aug 2002 16:17:07 -0500", "msg_from": "Larry Rosenman <ler@lerctr.org>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> But we should have some default to print some of the query,\n\nWhy? So far you've been told by two different people (make that three\nnow) that such a behavior is useless, and no one's weighed in in its\nfavor ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 18:08:40 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable " }, { "msg_contents": "On Tue, Aug 27, 2002 at 06:08:40PM -0400, Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > But we should have some default to print some of the query,\n> \n> Why? So far you've been told by two different people (make that three\n> now) that such a behavior is useless, and no one's weighed in in its\n> favor ...\n\nI agree that a 'trimmed' query is likely to be useless, but the idea of\nprinting the query on ERROR is a big win for me: right now I'm logging\n_all_ queries on our development machine (and sometimes on our production\nmachine. when there's trouble) so my logs would get considerably smaller.\n\nA settable trim length would probably be a good idea, I suppose, for\nthose slinging 'bytea' and toasted texts around.\n\nRoss\n", "msg_date": "Tue, 27 Aug 2002 17:30:04 -0500", "msg_from": "\"Ross J. Reedstrom\" <reedstrm@rice.edu>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "On Tue, 2002-08-27 at 17:30, Ross J. Reedstrom wrote:\n> On Tue, Aug 27, 2002 at 06:08:40PM -0400, Tom Lane wrote:\n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > But we should have some default to print some of the query,\n> > \n> > Why? So far you've been told by two different people (make that three\n> > now) that such a behavior is useless, and no one's weighed in in its\n> > favor ...\n> \n> I agree that a 'trimmed' query is likely to be useless, but the idea of\n> printing the query on ERROR is a big win for me: right now I'm logging\n> _all_ queries on our development machine (and sometimes on our production\n> machine. when there's trouble) so my logs would get considerably smaller.\nI agree with printing the query on error... just not limiting it to 80\ncharacters by default. \n> \n> A settable trim length would probably be a good idea, I suppose, for\n> those slinging 'bytea' and toasted texts around.\nYes, but the default should be NO TRIM or in 1K-4K range. IMHO\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\n\n", "msg_date": "27 Aug 2002 17:31:53 -0500", "msg_from": "Larry Rosenman <ler@lerctr.org>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "\"Ross J. Reedstrom\" <reedstrm@rice.edu> writes:\n> I agree that a 'trimmed' query is likely to be useless, but the idea of\n> printing the query on ERROR is a big win for me:\n\nCertainly. I think though that an on-or-off GUC option is sufficient.\nWe don't need a length, and we definitely don't need code to strip out\nwhitespace as Bruce was suggesting ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 18:34:40 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable " }, { "msg_contents": "> > A settable trim length would probably be a good idea, I suppose, for\n> > those slinging 'bytea' and toasted texts around.\n> Yes, but the default should be NO TRIM or in 1K-4K range. IMHO\n\nDitto.\n\n\n", "msg_date": "27 Aug 2002 18:39:48 -0400", "msg_from": "Rod Taylor <rbt@zort.ca>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "\nOK, just go with a boolean and admins can decide if they want it.\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> \"Ross J. Reedstrom\" <reedstrm@rice.edu> writes:\n> > I agree that a 'trimmed' query is likely to be useless, but the idea of\n> > printing the query on ERROR is a big win for me:\n> \n> Certainly. I think though that an on-or-off GUC option is sufficient.\n> We don't need a length, and we definitely don't need code to strip out\n> whitespace as Bruce was suggesting ...\n> \n> \t\t\tregards, tom lane\n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 19:17:36 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "On Tue, 2002-08-27 at 17:17, Larry Rosenman wrote:\n> On Tue, 2002-08-27 at 16:14, Bruce Momjian wrote:\n> > Larry Rosenman wrote:\n> > > > But we should have some default to print some of the query, because\n> > > > right now we print none of it. I am not saying it is perfect, but it is\n> > > > better than what we have, and is a reasonable default.\n> > > On an error, you may not be able to reproduce it. Why not print the\n> > > whole query to the log?\n> > > \n> > > I don't see a reason for truncating it at 80 chars. \n> > > \n> > > IMHO, of course. \n> > \n> > Because every typo query, every syntax error of a user in psql would\n> > appear in your logs. That seems excessive. Already the ERROR line\n> > appears in the logs. Do we want to see their bad query too?\n> > \n> > My concern is that long queries could easily bulk up the logs to the\n> > point where the actual important log messages would be lost in the fog.\n> Hmm. I think the 80 should be a GUC variable (and also settable from\n> SQL SET as well), and the 80 should probably be higher. \n> \n> And, maybe send the full query at a different Syslog(3) level.\n\nNot sure whether it can be done easily, but I think I'd prefer to log it\nto a different file/facility. That lets me keep the full query for the\nusers, without swamping the sysadmin's logs. Just my $0.02.\n\n--\nKarl\n\n", "msg_date": "27 Aug 2002 19:18:38 -0400", "msg_from": "Karl DeBisschop <kdebisschop@alert.infoplease.com>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "tgl@sss.pgh.pa.us (Tom Lane) wrote\n\n> \"Ross J. Reedstrom\" <reedstrm@rice.edu> writes:\n>> I agree that a 'trimmed' query is likely to be useless, but the idea of\n>> printing the query on ERROR is a big win for me:\n> \n> Certainly. I think though that an on-or-off GUC option is sufficient.\n> We don't need a length, and we definitely don't need code to strip out\n> whitespace as Bruce was suggesting ...\n\nJust out of curiosity... how much harder would it be to have the GUC \nvariable represent the truncation length? so setting it to zero would be \nequivalent to turning the feature off... I personally would have no use \nfor this feature, but I am just curious.\n", "msg_date": "Wed, 28 Aug 2002 00:37:15 +0000 (UTC)", "msg_from": "ngpg@grymmjack.com", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "On Tue, 27 Aug 2002, Tom Lane wrote:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > But we should have some default to print some of the query,\n> \n> Why? So far you've been told by two different people (make that three\n> now) that such a behavior is useless, and no one's weighed in in its\n> favor ...\n\nI completely agree. Nothing wrong with adding another guc variable and\nsince it is a debug variable people expect lots of verbosity.\n\nOnce I check out some other suggestions by Christopher I'll send a patch\nin -- its only a 10 liner.\n\nGavin\n\n", "msg_date": "Wed, 28 Aug 2002 11:59:40 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable " }, { "msg_contents": "On Wed, 28 Aug 2002, Gavin Sherry wrote:\n\n> On Tue, 27 Aug 2002, Tom Lane wrote:\n> \n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > But we should have some default to print some of the query,\n> > \n> > Why? So far you've been told by two different people (make that three\n> > now) that such a behavior is useless, and no one's weighed in in its\n> > favor ...\n> \n> I completely agree. Nothing wrong with adding another guc variable and\n> since it is a debug variable people expect lots of verbosity.\n\nAttached is the patch. debug_print_error_query is set to false by default.\n\nFor want of a better phrase, I've prepended 'original query: ' to the\nerror message to highlight why it is in the log.\n\nGavin", "msg_date": "Wed, 28 Aug 2002 16:26:47 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable " }, { "msg_contents": "I tested this patch and I can confirm that it works very well on\nFreeBSD/Alpha! This is such a cool feature - it's going to make my life so\nmuch easier!\n\nDo you think that 'original query: ..' looks a bit like bad english? Should\nit be properly capitalised? ie. 'Original query: ...'? Just nitpicking...\n\nChris\n\n> -----Original Message-----\n> From: pgsql-patches-owner@postgresql.org\n> [mailto:pgsql-patches-owner@postgresql.org]On Behalf Of Gavin Sherry\n> Sent: Wednesday, 28 August 2002 2:27 PM\n> To: Tom Lane\n> Cc: Hackers; pgsql-patches@postgresql.org\n> Subject: Re: [PATCHES] [HACKERS] Proposed GUC Variable\n>\n>\n> On Wed, 28 Aug 2002, Gavin Sherry wrote:\n>\n> > On Tue, 27 Aug 2002, Tom Lane wrote:\n> >\n> > > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > > But we should have some default to print some of the query,\n> > >\n> > > Why? So far you've been told by two different people (make that three\n> > > now) that such a behavior is useless, and no one's weighed in in its\n> > > favor ...\n> >\n> > I completely agree. Nothing wrong with adding another guc variable and\n> > since it is a debug variable people expect lots of verbosity.\n>\n> Attached is the patch. debug_print_error_query is set to false by default.\n>\n> For want of a better phrase, I've prepended 'original query: ' to the\n> error message to highlight why it is in the log.\n>\n> Gavin\n>\n\n", "msg_date": "Wed, 28 Aug 2002 15:32:49 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] Proposed GUC Variable " }, { "msg_contents": "Gavin Sherry writes:\n\n> Attached is the patch. debug_print_error_query is set to false by default.\n>\n> For want of a better phrase, I've prepended 'original query: ' to the\n> error message to highlight why it is in the log.\n\n From your resident How-To-Name-Stuff Nitpicker:\n\n1. The names of the debug_* GUC variables are leftovers from the pre-GUC\nera and the names where left to include \"debug\" in them because at the\ntime it wasn't clear whether the implementation had more than server-code\ndebugging quality. New variables should be named log_*.\n\n2. Unless you are only logging queries, the correct term is \"statement\" or\n\"commmand\". Statements are defined in the SQL standard to end at the\nsemicolon, but if you're logging whatever the client passed in (which may\ncontain multiple statements) then \"command\" might be best. (consequently:\nlog_command_on_error or something like that)\n\n3. Not sure what the \"original\" is for -- you're not transforming\nanything.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Wed, 28 Aug 2002 20:28:25 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable " }, { "msg_contents": "Christopher Kings-Lynne writes:\n\n> Do you think that 'original query: ..' looks a bit like bad english? Should\n> it be properly capitalised? ie. 'Original query: ...'? Just nitpicking...\n\nI find it's generally better to not capitalize anything in program\nmessages, unless the sentence/paragraph nature is very evident. Otherwise\nyou get drawn into a big deal about which messages are sentences or\nqualifying fragments, it creates inconsistencies if your messages get\nembedded into other messages, and the next day you start thinking about\nputting periods at the end, which is a really bad idea.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Wed, 28 Aug 2002 20:29:26 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable " }, { "msg_contents": "Peter Eisentraut wrote:\n> Gavin Sherry writes:\n> \n> > Attached is the patch. debug_print_error_query is set to false by default.\n> >\n> > For want of a better phrase, I've prepended 'original query: ' to the\n> > error message to highlight why it is in the log.\n> \n> >From your resident How-To-Name-Stuff Nitpicker:\n> \n> 1. The names of the debug_* GUC variables are leftovers from the pre-GUC\n> era and the names where left to include \"debug\" in them because at the\n> time it wasn't clear whether the implementation had more than server-code\n> debugging quality. New variables should be named log_*.\n\nAgreed. They are not really _debug_ for the server, but debug for user\napps; should be \"log\".\n\n\n> 2. Unless you are only logging queries, the correct term is \"statement\" or\n> \"commmand\". Statements are defined in the SQL standard to end at the\n> semicolon, but if you're logging whatever the client passed in (which may\n> contain multiple statements) then \"command\" might be best. (consequently:\n> log_command_on_error or something like that)\n\nOr log_statement_on_error. I think statement is better because we are\nusing that now for statement_timeout.\n\n> 3. Not sure what the \"original\" is for -- you're not transforming\n> anything.\n\nAgreed. Just call it \"Error query\". Seems clear to me.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 28 Aug 2002 15:05:05 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable" }, { "msg_contents": "On Wed, 2002-08-28 at 14:05, Bruce Momjian wrote:\n> Peter Eisentraut wrote:\n> > Gavin Sherry writes:\n> > \n> > > Attached is the patch. debug_print_error_query is set to false by default.\n> > >\n> > > For want of a better phrase, I've prepended 'original query: ' to the\n> > > error message to highlight why it is in the log.\n> > \n> > >From your resident How-To-Name-Stuff Nitpicker:\n> > \n> > 1. The names of the debug_* GUC variables are leftovers from the pre-GUC\n> > era and the names where left to include \"debug\" in them because at the\n> > time it wasn't clear whether the implementation had more than server-code\n> > debugging quality. New variables should be named log_*.\n> \n> Agreed. They are not really _debug_ for the server, but debug for user\n> apps; should be \"log\".\n> \n> \n> > 2. Unless you are only logging queries, the correct term is \"statement\" or\n> > \"commmand\". Statements are defined in the SQL standard to end at the\n> > semicolon, but if you're logging whatever the client passed in (which may\n> > contain multiple statements) then \"command\" might be best. (consequently:\n> > log_command_on_error or something like that)\n> \n> Or log_statement_on_error. I think statement is better because we are\n> using that now for statement_timeout.\n> \n> > 3. Not sure what the \"original\" is for -- you're not transforming\n> > anything.\n> \n> Agreed. Just call it \"Error query\". Seems clear to me.\nWhat about rule(s) transformation(s)? Will we see the real query or the\ntransformed query?\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\n\n", "msg_date": "28 Aug 2002 14:07:57 -0500", "msg_from": "Larry Rosenman <ler@lerctr.org>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable" }, { "msg_contents": "Larry Rosenman wrote:\n> > > 3. Not sure what the \"original\" is for -- you're not transforming\n> > > anything.\n> > \n> > Agreed. Just call it \"Error query\". Seems clear to me.\n> What about rule(s) transformation(s)? Will we see the real query or the\n> transformed query?\n\nWell, looking at Gavin's patch, I see:\n\n+ if(lev == ERROR && Debug_print_error_query)\n+ elog(LOG,\"original query: %s\",debug_query_string);\n\nThat will be the query supplied by the user. To give them anything else\nwould be even more confusing --- \"How did that query get executed. I\ndon't even see that query in my code\". At least when they see that the\nquery and the error don't match, they can think, rules/triggers, etc. \nWe should mention the possible mismatch in the docs.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 28 Aug 2002 15:11:09 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable" }, { "msg_contents": "One of my users is generating a notice message --> NOTICE: Adding\nmissing FROM-clause entry for table \"msg202\" It might be helpful to\ndump out the query on notice messages like this, and it looks like a\nsimple change as far as elog.c and guc.c are concerned, but would this\nbe overkill?\n\nRobert Treat\n\nOn Wed, 2002-08-28 at 02:26, Gavin Sherry wrote:\n> On Wed, 28 Aug 2002, Gavin Sherry wrote:\n> \n> > On Tue, 27 Aug 2002, Tom Lane wrote:\n> > \n> > > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > > But we should have some default to print some of the query,\n> > > \n> > > Why? So far you've been told by two different people (make that three\n> > > now) that such a behavior is useless, and no one's weighed in in its\n> > > favor ...\n> > \n> > I completely agree. Nothing wrong with adding another guc variable and\n> > since it is a debug variable people expect lots of verbosity.\n> \n> Attached is the patch. debug_print_error_query is set to false by default.\n> \n> For want of a better phrase, I've prepended 'original query: ' to the\n> error message to highlight why it is in the log.\n> \n> Gavin\n> \n> ----\n> \n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n\n\n\n", "msg_date": "29 Aug 2002 12:09:58 -0400", "msg_from": "Robert Treat <xzilla@users.sourceforge.net>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable" }, { "msg_contents": "Robert Treat <xzilla@users.sourceforge.net> writes:\n> One of my users is generating a notice message --> NOTICE: Adding\n> missing FROM-clause entry for table \"msg202\" It might be helpful to\n> dump out the query on notice messages like this, and it looks like a\n> simple change as far as elog.c and guc.c are concerned, but would this\n> be overkill?\n\nHm. Maybe instead of a boolean, what we want is a message level\nvariable: log original query if it triggers a message >= severity X.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 29 Aug 2002 12:14:24 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable " }, { "msg_contents": "On August 29, 2002 12:09 pm, Robert Treat wrote:\n> One of my users is generating a notice message --> NOTICE: Adding\n> missing FROM-clause entry for table \"msg202\" It might be helpful to\n> dump out the query on notice messages like this, and it looks like a\n> simple change as far as elog.c and guc.c are concerned, but would this\n> be overkill?\n\nCould be useful. In the meantime you still have an option. Turn on query \nlogging and then you can go back to your logs and find this notice and the \nquery that generated it. Note that this could create lots of logs so be \nprepared. I think that Tom's suggestion is better in the long run but if you \nneed something now then that works.\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": "Thu, 29 Aug 2002 13:17:35 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" }, { "msg_contents": "On Thu, 2002-08-29 at 19:04, Gavin Sherry wrote:\n> On Thu, 29 Aug 2002, Tom Lane wrote:\n> \n> > Robert Treat <xzilla@users.sourceforge.net> writes:\n> > > One of my users is generating a notice message --> NOTICE: Adding\n> > > missing FROM-clause entry for table \"msg202\" It might be helpful to\n> > > dump out the query on notice messages like this, and it looks like a\n> > > simple change as far as elog.c and guc.c are concerned, but would this\n> > > be overkill?\n> > \n> > Hm. Maybe instead of a boolean, what we want is a message level\n> > variable: log original query if it triggers a message >= severity X.\n> \n> That's a pretty good idea. Now, what format will the argument take: text\n> (NOTICE, ERROR, DEBUG, etc) or integer? The increasing severity is clear\n> with numbers but the correlation to NOTICE, ERROR etc is undocumented\n> IIRC. On the other hand, the textual form is clear but INFO < NOTICE <\n> WARNING < ERROR < FATAL, etc, is note necessarily obvious. (Also, with the\n> textual option the word will need to be converted to the corresponding\n> number by the GUC code).\n> \n> Naturally, the problem with each option can be cleared up with\n> documentation.\nmy gut feeling is use the words. \n\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\n\n", "msg_date": "29 Aug 2002 19:02:44 -0500", "msg_from": "Larry Rosenman <ler@lerctr.org>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable" }, { "msg_contents": "On Thu, 29 Aug 2002, Tom Lane wrote:\n\n> Robert Treat <xzilla@users.sourceforge.net> writes:\n> > One of my users is generating a notice message --> NOTICE: Adding\n> > missing FROM-clause entry for table \"msg202\" It might be helpful to\n> > dump out the query on notice messages like this, and it looks like a\n> > simple change as far as elog.c and guc.c are concerned, but would this\n> > be overkill?\n> \n> Hm. Maybe instead of a boolean, what we want is a message level\n> variable: log original query if it triggers a message >= severity X.\n\nThat's a pretty good idea. Now, what format will the argument take: text\n(NOTICE, ERROR, DEBUG, etc) or integer? The increasing severity is clear\nwith numbers but the correlation to NOTICE, ERROR etc is undocumented\nIIRC. On the other hand, the textual form is clear but INFO < NOTICE <\nWARNING < ERROR < FATAL, etc, is note necessarily obvious. (Also, with the\ntextual option the word will need to be converted to the corresponding\nnumber by the GUC code).\n\nNaturally, the problem with each option can be cleared up with\ndocumentation.\n\nDoes anyone have a preference here?\n\nGavin\n\n\n", "msg_date": "Fri, 30 Aug 2002 10:04:25 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable " }, { "msg_contents": "Gavin Sherry wrote:\n> On Thu, 29 Aug 2002, Tom Lane wrote:\n> \n> > Robert Treat <xzilla@users.sourceforge.net> writes:\n> > > One of my users is generating a notice message --> NOTICE: Adding\n> > > missing FROM-clause entry for table \"msg202\" It might be helpful to\n> > > dump out the query on notice messages like this, and it looks like a\n> > > simple change as far as elog.c and guc.c are concerned, but would this\n> > > be overkill?\n> > \n> > Hm. Maybe instead of a boolean, what we want is a message level\n> > variable: log original query if it triggers a message >= severity X.\n> \n> That's a pretty good idea. Now, what format will the argument take: text\n> (NOTICE, ERROR, DEBUG, etc) or integer? The increasing severity is clear\n> with numbers but the correlation to NOTICE, ERROR etc is undocumented\n> IIRC. On the other hand, the textual form is clear but INFO < NOTICE <\n> WARNING < ERROR < FATAL, etc, is note necessarily obvious. (Also, with the\n> textual option the word will need to be converted to the corresponding\n> number by the GUC code).\n> \n> Naturally, the problem with each option can be cleared up with\n> documentation.\n\nI think the arg has to be text. See server_min_messages GUC for an\nexample.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 29 Aug 2002 20:10:32 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable" }, { "msg_contents": "Gavin Sherry <swm@linuxworld.com.au> writes:\n> That's a pretty good idea. Now, what format will the argument take: text\n> (NOTICE, ERROR, DEBUG, etc) or integer? The increasing severity is clear\n> with numbers but the correlation to NOTICE, ERROR etc is undocumented\n> IIRC. On the other hand, the textual form is clear but INFO < NOTICE <\n> WARNING < ERROR < FATAL, etc, is note necessarily obvious.\n\nThe variable should take the same values as SERVER_MIN_MESSAGES and\nimpose the same priority order as it does. I would assume you could\nshare code, or at worst copy-and-paste a few dozen lines.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 29 Aug 2002 20:42:28 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable " }, { "msg_contents": "On Thu, 29 Aug 2002, Tom Lane wrote:\n\n> Gavin Sherry <swm@linuxworld.com.au> writes:\n> > That's a pretty good idea. Now, what format will the argument take: text\n> > (NOTICE, ERROR, DEBUG, etc) or integer? The increasing severity is clear\n> > with numbers but the correlation to NOTICE, ERROR etc is undocumented\n> > IIRC. On the other hand, the textual form is clear but INFO < NOTICE <\n> > WARNING < ERROR < FATAL, etc, is note necessarily obvious.\n> \n> The variable should take the same values as SERVER_MIN_MESSAGES and\n> impose the same priority order as it does. I would assume you could\n> share code, or at worst copy-and-paste a few dozen lines.\n\nA patch implementing this is attached. Instead of copy-and-pasting the\ncode, I abstracted out of the lookup and converted the existing functions\nto use it.\n\nI was careful in elog() to ignore elog(LOG) calls when\nlog_min_error_query >= LOG.\n\nGavin", "msg_date": "Sun, 1 Sep 2002 02:28:21 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Proposed GUC Variable " }, { "msg_contents": "OK, attached patch applied. I made a few changes. I added a mention\nthey may want to enable LOG_PID because there is no guarantee that the\nstatement and error will appear next to each other in the log file. I\nalso renamed 'query' to 'statement' to be more precise.\n\nAlso, is there any way to disable this feature? I can't see how.\n\nAlso, you added this line:\n\n\t+ extern bool Debug_print_error_query;\n\nI assume it was a mistake.\n\n---------------------------------------------------------------------------\n\nGavin Sherry wrote:\n> On Thu, 29 Aug 2002, Tom Lane wrote:\n> \n> > Gavin Sherry <swm@linuxworld.com.au> writes:\n> > > That's a pretty good idea. Now, what format will the argument take: text\n> > > (NOTICE, ERROR, DEBUG, etc) or integer? The increasing severity is clear\n> > > with numbers but the correlation to NOTICE, ERROR etc is undocumented\n> > > IIRC. On the other hand, the textual form is clear but INFO < NOTICE <\n> > > WARNING < ERROR < FATAL, etc, is note necessarily obvious.\n> > \n> > The variable should take the same values as SERVER_MIN_MESSAGES and\n> > impose the same priority order as it does. I would assume you could\n> > share code, or at worst copy-and-paste a few dozen lines.\n> \n> A patch implementing this is attached. Instead of copy-and-pasting the\n> code, I abstracted out of the lookup and converted the existing functions\n> to use it.\n> \n> I was careful in elog() to ignore elog(LOG) calls when\n> log_min_error_query >= LOG.\n> \n> Gavin\n\nContent-Description: \n\n[ Attachment, skipping... ]\n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n\nIndex: doc/src/sgml/runtime.sgml\n===================================================================\nRCS file: /cvsroot/pgsql-server/doc/src/sgml/runtime.sgml,v\nretrieving revision 1.132\ndiff -c -c -r1.132 runtime.sgml\n*** doc/src/sgml/runtime.sgml\t1 Sep 2002 23:26:06 -0000\t1.132\n--- doc/src/sgml/runtime.sgml\t2 Sep 2002 05:35:45 -0000\n***************\n*** 942,948 ****\n </para>\n </listitem>\n </varlistentry>\n- \n <varlistentry>\n <term><varname>EXPLAIN_PRETTY_PRINT</varname> (<type>boolean</type>)</term>\n <listitem>\n--- 942,947 ----\n***************\n*** 979,984 ****\n--- 978,1005 ----\n </varlistentry>\n \n <varlistentry>\n+ <term><varname>LOG_MIN_ERROR_STATEMENT</varname> (<type>string</type>)</term>\n+ <listitem>\n+ <para>\n+ This controls which log messages are accompanied by the original\n+ query which generated the message. All queries matching the setting\n+ or which are of a higher severity than the setting are logged. The\n+ default is <literal>ERROR</literal>. Valid values are\n+ <literal>DEBUG5</literal>, <literal>DEBUG4</literal>, \n+ <literal>DEBUG3</literal>, <literal>DEBUG2</literal>, \n+ <literal>DEBUG1</literal>, <literal>INFO</literal>,\n+ <literal>NOTICE</literal>, <literal>WARNING</literal>\n+ and <literal>ERROR</literal>.\n+ </para>\n+ <para>\n+ It is recommended you enable <literal>LOG_PID</literal> as well\n+ so you can more easily match the error statement with the error\n+ message.\n+ </para>\n+ </listitem>\n+ </varlistentry>\n+ \n+ <varlistentry>\n <term><varname>LOG_PID</varname> (<type>boolean</type>)</term>\n <listitem>\n <para>\n***************\n*** 1005,1012 ****\n <listitem>\n <para>\n Prints the duration of every completed query. To use this option, \n! enable LOG_STATEMENT and LOG_PID so you can link the original query\n! to the duration using the process id.\n </para>\n </listitem>\n </varlistentry>\n--- 1026,1033 ----\n <listitem>\n <para>\n Prints the duration of every completed query. To use this option, \n! enable <literal>LOG_STATEMENT</> and <literal>LOG_PID</> so you \n! can link the original query to the duration using the process id.\n </para>\n </listitem>\n </varlistentry>\nIndex: src/backend/utils/error/elog.c\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/backend/utils/error/elog.c,v\nretrieving revision 1.101\ndiff -c -c -r1.101 elog.c\n*** src/backend/utils/error/elog.c\t2 Sep 2002 02:47:05 -0000\t1.101\n--- src/backend/utils/error/elog.c\t2 Sep 2002 05:35:51 -0000\n***************\n*** 33,49 ****\n #include \"storage/proc.h\"\n #include \"tcop/tcopprot.h\"\n #include \"utils/memutils.h\"\n \n #include \"mb/pg_wchar.h\"\n \n- int\t\t\tserver_min_messages;\n- char\t *server_min_messages_str = NULL;\n- const char\tserver_min_messages_str_default[] = \"notice\";\n- \n- int\t\t\tclient_min_messages;\n- char\t *client_min_messages_str = NULL;\n- const char\tclient_min_messages_str_default[] = \"notice\";\n- \n #ifdef HAVE_SYSLOG\n /*\n * 0 = only stdout/stderr\n--- 33,42 ----\n #include \"storage/proc.h\"\n #include \"tcop/tcopprot.h\"\n #include \"utils/memutils.h\"\n+ #include \"utils/guc.h\"\n \n #include \"mb/pg_wchar.h\"\n \n #ifdef HAVE_SYSLOG\n /*\n * 0 = only stdout/stderr\n***************\n*** 345,350 ****\n--- 338,344 ----\n \t\t}\n \t}\n \n+ \n \t/*\n \t * Message prepared; send it where it should go\n \t */\n***************\n*** 433,438 ****\n--- 427,440 ----\n \tif (msg_buf != msg_fixedbuf)\n \t\tfree(msg_buf);\n \n+ \t/* If the user wants this elog() generating query logged,\n+ \t * do so. We only want to log if the query has been\n+ \t * written to debug_query_string. Also, avoid infinite loops.\n+ \t */\n+ \n+ \tif(lev != LOG && lev >= log_min_error_statement && debug_query_string)\n+ \t\telog(LOG,\"statement: %s\",debug_query_string);\n+ \n \t/*\n \t * Perform error recovery action as specified by lev.\n \t */\n***************\n*** 835,905 ****\n }\n \n \n- /*\n- * GUC support routines\n- */\n- const char *\n- assign_server_min_messages(const char *newval,\n- \t\t\t\t\t\t bool doit, bool interactive)\n- {\n- \tif (strcasecmp(newval, \"debug\") == 0)\n- \t\t{ if (doit) server_min_messages = DEBUG1; }\n- \telse if (strcasecmp(newval, \"debug5\") == 0)\n- \t\t{ if (doit) server_min_messages = DEBUG5; }\n- \telse if (strcasecmp(newval, \"debug4\") == 0)\n- \t\t{ if (doit) server_min_messages = DEBUG4; }\n- \telse if (strcasecmp(newval, \"debug3\") == 0)\n- \t\t{ if (doit) server_min_messages = DEBUG3; }\n- \telse if (strcasecmp(newval, \"debug2\") == 0)\n- \t\t{ if (doit) server_min_messages = DEBUG2; }\n- \telse if (strcasecmp(newval, \"debug1\") == 0)\n- \t\t{ if (doit) server_min_messages = DEBUG1; }\n- \telse if (strcasecmp(newval, \"info\") == 0)\n- \t\t{ if (doit) server_min_messages = INFO; }\n- \telse if (strcasecmp(newval, \"notice\") == 0)\n- \t\t{ if (doit) server_min_messages = NOTICE; }\n- \telse if (strcasecmp(newval, \"warning\") == 0)\n- \t\t{ if (doit) server_min_messages = WARNING; }\n- \telse if (strcasecmp(newval, \"error\") == 0)\n- \t\t{ if (doit) server_min_messages = ERROR; }\n- \telse if (strcasecmp(newval, \"log\") == 0)\n- \t\t{ if (doit) server_min_messages = LOG; }\n- \telse if (strcasecmp(newval, \"fatal\") == 0)\n- \t\t{ if (doit) server_min_messages = FATAL; }\n- \telse if (strcasecmp(newval, \"panic\") == 0)\n- \t\t{ if (doit) server_min_messages = PANIC; }\n- \telse\n- \t\treturn NULL;\t\t\t/* fail */\n- \treturn newval;\t\t\t\t/* OK */\n- }\n \n- const char *\n- assign_client_min_messages(const char *newval,\n- \t\t\t\t\t\t bool doit, bool interactive)\n- {\n- \tif (strcasecmp(newval, \"debug\") == 0)\n- \t\t{ if (doit) client_min_messages = DEBUG1; }\n- \telse if (strcasecmp(newval, \"debug5\") == 0)\n- \t\t{ if (doit) client_min_messages = DEBUG5; }\n- \telse if (strcasecmp(newval, \"debug4\") == 0)\n- \t\t{ if (doit) client_min_messages = DEBUG4; }\n- \telse if (strcasecmp(newval, \"debug3\") == 0)\n- \t\t{ if (doit) client_min_messages = DEBUG3; }\n- \telse if (strcasecmp(newval, \"debug2\") == 0)\n- \t\t{ if (doit) client_min_messages = DEBUG2; }\n- \telse if (strcasecmp(newval, \"debug1\") == 0)\n- \t\t{ if (doit) client_min_messages = DEBUG1; }\n- \telse if (strcasecmp(newval, \"log\") == 0)\n- \t\t{ if (doit) client_min_messages = LOG; }\n- \telse if (strcasecmp(newval, \"info\") == 0)\n- \t\t{ if (doit) client_min_messages = INFO; }\n- \telse if (strcasecmp(newval, \"notice\") == 0)\n- \t\t{ if (doit) client_min_messages = NOTICE; }\n- \telse if (strcasecmp(newval, \"warning\") == 0)\n- \t\t{ if (doit) client_min_messages = WARNING; }\n- \telse if (strcasecmp(newval, \"error\") == 0)\n- \t\t{ if (doit) client_min_messages = ERROR; }\n- \telse\n- \t\treturn NULL;\t\t\t/* fail */\n- \treturn newval;\t\t\t\t/* OK */\n- }\n--- 837,840 ----\nIndex: src/backend/utils/misc/guc.c\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/backend/utils/misc/guc.c,v\nretrieving revision 1.91\ndiff -c -c -r1.91 guc.c\n*** src/backend/utils/misc/guc.c\t2 Sep 2002 01:05:06 -0000\t1.91\n--- src/backend/utils/misc/guc.c\t2 Sep 2002 05:35:59 -0000\n***************\n*** 71,76 ****\n--- 71,79 ----\n \t\t\t\t\t\t\t\t bool doit, bool interactive);\n #endif\n \n+ static const char *assign_msglvl(int *var, const char *newval,\n+ \t\t\t\t\t\t\t\t\tbool doit, bool interactive);\n+ \n /*\n * Debugging options\n */\n***************\n*** 99,104 ****\n--- 102,120 ----\n \n bool\t\tPassword_encryption = true;\n \n+ int\t\t\tlog_min_error_statement;\n+ char\t\t*log_min_error_statement_str = NULL;\n+ const char\tlog_min_error_statement_str_default[] = \"error\";\n+ \n+ int server_min_messages;\n+ char *server_min_messages_str = NULL;\n+ const char server_min_messages_str_default[] = \"notice\";\n+ \n+ int client_min_messages;\n+ char *client_min_messages_str = NULL;\n+ const char client_min_messages_str_default[] = \"notice\";\n+ \n+ \n #ifndef PG_KRB_SRVTAB\n #define PG_KRB_SRVTAB \"\"\n #endif\n***************\n*** 727,732 ****\n--- 743,753 ----\n \t},\n \n \t{\n+ \t\t{ \"log_min_error_statement\", PGC_USERSET }, &log_min_error_statement_str,\n+ \t\tlog_min_error_statement_str_default, assign_min_error_statement, NULL\n+ \t},\n+ \n+ \t{\n \t\t{ \"DateStyle\", PGC_USERSET, GUC_LIST_INPUT }, &datestyle_string,\n \t\t\"ISO, US\", assign_datestyle, show_datestyle\n \t},\n***************\n*** 2877,2879 ****\n--- 2898,2951 ----\n \n \treturn newarray;\n }\n+ \n+ const char *\n+ assign_server_min_messages(const char *newval,\n+ \t\t\t\t\t\t bool doit, bool interactive)\n+ {\n+ \treturn(assign_msglvl(&server_min_messages,newval,doit,interactive));\n+ }\n+ \n+ const char *\n+ assign_client_min_messages(const char *newval,\n+ \t\t\t\t\t\t bool doit, bool interactive)\n+ {\n+ \treturn(assign_msglvl(&client_min_messages,newval,doit,interactive));\n+ }\n+ \n+ const char *\n+ assign_min_error_statement(const char *newval, bool doit, bool interactive)\n+ {\n+ \treturn(assign_msglvl(&log_min_error_statement,newval,doit,interactive));\n+ }\n+ \n+ static const char *\n+ assign_msglvl(int *var, const char *newval, bool doit, bool interactive)\n+ {\n+ \tif (strcasecmp(newval, \"debug\") == 0)\n+ \t\t{ if (doit) (*var) = DEBUG1; }\n+ \telse if (strcasecmp(newval, \"debug5\") == 0)\n+ \t\t{ if (doit) (*var) = DEBUG5; }\n+ \telse if (strcasecmp(newval, \"debug4\") == 0)\n+ \t\t{ if (doit) (*var) = DEBUG4; }\n+ \telse if (strcasecmp(newval, \"debug3\") == 0)\n+ \t\t{ if (doit) (*var) = DEBUG3; }\n+ \telse if (strcasecmp(newval, \"debug2\") == 0)\n+ \t\t{ if (doit) (*var) = DEBUG2; }\n+ \telse if (strcasecmp(newval, \"debug1\") == 0)\n+ \t\t{ if (doit) (*var) = DEBUG1; }\n+ \telse if (strcasecmp(newval, \"log\") == 0)\n+ \t\t{ if (doit) (*var) = LOG; }\n+ \telse if (strcasecmp(newval, \"info\") == 0)\n+ \t\t{ if (doit) (*var) = INFO; }\n+ \telse if (strcasecmp(newval, \"notice\") == 0)\n+ \t\t{ if (doit) (*var) = NOTICE; }\n+ \telse if (strcasecmp(newval, \"warning\") == 0)\n+ \t\t{ if (doit) (*var) = WARNING; }\n+ \telse if (strcasecmp(newval, \"error\") == 0)\n+ \t\t{ if (doit) (*var) = ERROR; }\n+ \telse\n+ \t\treturn NULL;\t\t\t/* fail */\n+ \treturn newval;\t\t\t\t/* OK */\n+ }\n+ \nIndex: src/backend/utils/misc/postgresql.conf.sample\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/backend/utils/misc/postgresql.conf.sample,v\nretrieving revision 1.51\ndiff -c -c -r1.51 postgresql.conf.sample\n*** src/backend/utils/misc/postgresql.conf.sample\t1 Sep 2002 23:26:06 -0000\t1.51\n--- src/backend/utils/misc/postgresql.conf.sample\t2 Sep 2002 05:36:00 -0000\n***************\n*** 127,132 ****\n--- 127,135 ----\n #log_duration = false\n #log_timestamp = false\n \n+ #log_min_error_statement = error\t\t# Values in order of increasing severity:\n+ \t\t\t\t\t\t\t\t\t# debug5, debug4, debug3, debug2, debug1,\n+ \t\t\t\t\t # info, notice, warning, error\n #debug_print_parse = false\n #debug_print_rewritten = false\n #debug_print_plan = false\nIndex: src/bin/psql/tab-complete.c\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/bin/psql/tab-complete.c,v\nretrieving revision 1.61\ndiff -c -c -r1.61 tab-complete.c\n*** src/bin/psql/tab-complete.c\t1 Sep 2002 23:26:06 -0000\t1.61\n--- src/bin/psql/tab-complete.c\t2 Sep 2002 05:36:05 -0000\n***************\n*** 271,277 ****\n \t\t\"default_transaction_isolation\",\n \t\t\"search_path\",\n \t\t\"statement_timeout\",\n! \n \t\tNULL\n \t};\n \n--- 271,277 ----\n \t\t\"default_transaction_isolation\",\n \t\t\"search_path\",\n \t\t\"statement_timeout\",\n! \t\t\"log_min_error_statement\",\n \t\tNULL\n \t};\n \nIndex: src/include/utils/elog.h\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/include/utils/elog.h,v\nretrieving revision 1.38\ndiff -c -c -r1.38 elog.h\n*** src/include/utils/elog.h\t20 Jun 2002 20:29:52 -0000\t1.38\n--- src/include/utils/elog.h\t2 Sep 2002 05:36:06 -0000\n***************\n*** 47,68 ****\n extern bool Log_timestamp;\n extern bool Log_pid;\n \n- extern char\t *server_min_messages_str;\n- extern char\t *client_min_messages_str;\n- extern const char server_min_messages_str_default[];\n- extern const char client_min_messages_str_default[];\n- \n extern void\n elog(int lev, const char *fmt,...)\n /* This extension allows gcc to check the format string for consistency with\n the supplied arguments. */\n __attribute__((format(printf, 2, 3)));\n \n! extern int\tDebugFileOpen(void);\n! \n! extern const char *assign_server_min_messages(const char *newval,\n! \t\t\t\t\t\t\t\t\t\t\t bool doit, bool interactive);\n! extern const char *assign_client_min_messages(const char *newval,\n! \t\t\t\t\t\t\t\t\t\t\t bool doit, bool interactive);\n \n #endif /* ELOG_H */\n--- 47,58 ----\n extern bool Log_timestamp;\n extern bool Log_pid;\n \n extern void\n elog(int lev, const char *fmt,...)\n /* This extension allows gcc to check the format string for consistency with\n the supplied arguments. */\n __attribute__((format(printf, 2, 3)));\n \n! extern int DebugFileOpen(void);\n \n #endif /* ELOG_H */\nIndex: src/include/utils/guc.h\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/include/utils/guc.h,v\nretrieving revision 1.21\ndiff -c -c -r1.21 guc.h\n*** src/include/utils/guc.h\t1 Sep 2002 23:26:06 -0000\t1.21\n--- src/include/utils/guc.h\t2 Sep 2002 05:36:07 -0000\n***************\n*** 100,105 ****\n--- 100,112 ----\n extern ArrayType *GUCArrayAdd(ArrayType *array, const char *name, const char *value);\n extern ArrayType *GUCArrayDelete(ArrayType *array, const char *name);\n \n+ extern const char *assign_min_error_statement(const char *newval, bool doit,\n+ \t\t\tbool interactive);\n+ \n+ extern const char *assign_server_min_messages(const char *newval,\n+ bool doit, bool interactive);\n+ extern const char *assign_client_min_messages(const char *newval,\n+ bool doit, bool interactive);\n extern bool Log_statement;\n extern bool Log_duration;\n extern bool Debug_print_plan;\n***************\n*** 117,121 ****\n--- 124,143 ----\n \n extern bool SQL_inheritance;\n extern bool Australian_timezones;\n+ \n+ extern char *debug_query_string;\n+ \n+ extern int\tlog_min_error_statement;\n+ extern char *log_min_error_statement_str;\n+ extern const char log_min_error_statement_str_default[];\n+ \n+ extern int\tserver_min_messages;\n+ extern char *server_min_messages_str;\n+ extern const char server_min_messages_str_default[];\n+ \n+ extern int client_min_messages;\n+ extern char *client_min_messages_str;\n+ \n+ extern const char client_min_messages_str_default[];\n \n #endif /* GUC_H */", "msg_date": "Mon, 2 Sep 2002 01:41:27 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Proposed GUC Variable" } ]
[ { "msg_contents": "\nOkay, just packaged it up (too fast for you Lamar? *grin*) ... I put the\nfiles in /pub/source/v7.2.2-prerelease for now, but will rename the\ndirectory to purely v7.2.2 tomorrow afternoon and put out an announcement,\njust to give 12 or so hours for ppl to confirm that the packaging doesn't\nhave any bugs in it ...\n\n", "msg_date": "Fri, 23 Aug 2002 00:28:00 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "v7.2.2 packaged ..." }, { "msg_contents": "On Thursday 22 August 2002 11:28 pm, Marc G. Fournier wrote:\n> Okay, just packaged it up (too fast for you Lamar? *grin*) ...\n\nJust got the cvs checkout of REL7_2_STABLE downloaded... :-) Oh well. I can \npossibly get an RPM pushed out tomorrow evening (I'm out of the office and \nout of town most of the day tomorrow fixing an errant transmitter). Time to \nload the line up another half hour.... :-)\n\n> I put the\n> files in /pub/source/v7.2.2-prerelease for now, but will rename the\n> directory to purely v7.2.2 tomorrow afternoon and put out an announcement,\n> just to give 12 or so hours for ppl to confirm that the packaging doesn't\n> have any bugs in it ...\n\nTime for me to play catchup. I wasn't expecting it quite _that_ fast, no.\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Thu, 22 Aug 2002 23:43:33 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: v7.2.2 packaged ..." }, { "msg_contents": "On Thursday 22 August 2002 11:43 pm, Lamar Owen wrote:\n> Just got the cvs checkout of REL7_2_STABLE downloaded... :-) Oh well. I\n> can possibly get an RPM pushed out tomorrow evening (I'm out of the office\n> and out of town most of the day tomorrow fixing an errant transmitter). \n> Time to load the line up another half hour.... :-)\n\nWell, the build is successful for RPM purposes for 7.2.2. Can't regression \ntest this one yet -- the machine I built it on is running a live OpenACS \nwebsite backed by 7.2.1. Still have 18 minutes to go on my dialup download \nof the 7.2.2 pre tarball.\n\nSooooo..... look in /pub/binary/v7.2.2-prelease/RPMS/SRPMS for a \nsuper-prerelease SOURCE RPM. No binaries. (Yes, I have them. No, I'm not \nuploading them at this point -- rpm --rebuild is your friend at this point in \ntime. I'll do a proper RPM release when I can integrate a couple of patches \nproperly -- Saturday morning may work, as I'll be on the other end of a T1 \nmost of that day.) For the purposes of release announcements, Marc, a simple \n'RPMS should be available within a day or two' should be sufficient.\n\nBut if you are of that bent, rpm --rebuild that source RPM all you want.... \n:-) Fast enough, Marc? :-)\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Fri, 23 Aug 2002 00:23:11 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: v7.2.2 packaged ..." }, { "msg_contents": "\"Marc G. Fournier\" <scrappy@hub.org> writes:\n> Okay, just packaged it up (too fast for you Lamar? *grin*) ... I put the\n> files in /pub/source/v7.2.2-prerelease for now, but will rename the\n> directory to purely v7.2.2 tomorrow afternoon and put out an announcement,\n> just to give 12 or so hours for ppl to confirm that the packaging doesn't\n> have any bugs in it ...\n\nThe main tarball looks okay from here.\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 23 Aug 2002 01:21:46 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: v7.2.2 packaged ... " }, { "msg_contents": "downloading, and will test on openbsd -current in the morning,\nwhen wget is finished. slow dl speed atm, sure everyone is \ngrabbing it. sure others are probably testing as well. just\nwanted to say so for those who care :)\n\n-chris\n\nMarc G. Fournier writes:\n > \n > Okay, just packaged it up (too fast for you Lamar? *grin*) ... I put the\n > files in /pub/source/v7.2.2-prerelease for now, but will rename the\n > directory to purely v7.2.2 tomorrow afternoon and put out an announcement,\n > just to give 12 or so hours for ppl to confirm that the packaging doesn't\n > have any bugs in it ...\n > \n > \n > ---------------------------(end of broadcast)---------------------------\n > TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n\n-- \n\"Anyone who considers arithmetical methods of producing random digits is, \n of course, in a state of sin.\"\n", "msg_date": "Fri, 23 Aug 2002 01:05:15 -0500", "msg_from": "Chris Humphries <chumphries@devis.com>", "msg_from_op": false, "msg_subject": "v7.2.2 packaged ..." }, { "msg_contents": "\ncompiles clean on openbsd -current.\n\n-chris\n\nTom Lane writes:\n > \"Marc G. Fournier\" <scrappy@hub.org> writes:\n > > Okay, just packaged it up (too fast for you Lamar? *grin*) ... I put the\n > > files in /pub/source/v7.2.2-prerelease for now, but will rename the\n > > directory to purely v7.2.2 tomorrow afternoon and put out an announcement,\n > > just to give 12 or so hours for ppl to confirm that the packaging doesn't\n > > have any bugs in it ...\n > \n > The main tarball looks okay from here.\n > \n > \t\t\tregards, tom lane\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\"Anyone who considers arithmetical methods of producing random digits is, \n of course, in a state of sin.\"\n", "msg_date": "Fri, 23 Aug 2002 08:24:22 -0500", "msg_from": "Chris Humphries <chumphries@devis.com>", "msg_from_op": false, "msg_subject": "Re: v7.2.2 packaged ... " }, { "msg_contents": "Marc G. Fournier wrote:\n> Okay, just packaged it up (too fast for you Lamar? *grin*) ... I put the\n> files in /pub/source/v7.2.2-prerelease for now, but will rename the\n> directory to purely v7.2.2 tomorrow afternoon and put out an announcement,\n> just to give 12 or so hours for ppl to confirm that the packaging doesn't\n> have any bugs in it ...\n\nFWIW: I was able to build my own RPMs and upgrade a 7.2.1 system \nsuccessfully to 7.2.2 -- no problems encountered at all.\n\nSystem is Red Hat 7.3/i686 smp\n\nJoe\n\n\n\n\n", "msg_date": "Fri, 23 Aug 2002 08:47:31 -0700", "msg_from": "Joe Conway <mail@joeconway.com>", "msg_from_op": false, "msg_subject": "Re: v7.2.2 packaged ..." }, { "msg_contents": "Lamar,\n\nI haven't had/needed to upgrade since 7.0 and 7.2BETA so I'm not sure \nhow this is being done currently but I'd like to see a variable or two in \nthe /etc/init.d/postgresql script to allow a change to tje location of \ndebugging/logging information.\n Currently it is hard-coded as /dev/null. I'd like a variable or two\nsay, PGLOGDIR, PGLOGFILE at the top of the script that could be changed\neasily to where ever the admin wants it to go.\n\n\nThanks for your consideration,\nRod\n-- \n \"Open Source Software - Sometimes you get more than you paid for...\"\n\n", "msg_date": "Fri, 23 Aug 2002 09:57:27 -0700 (PDT)", "msg_from": "\"Roderick A. Anderson\" <raanders@acm.org>", "msg_from_op": false, "msg_subject": "RPM Feature Request" } ]
[ { "msg_contents": "This is the same patch as 20020822_01_pitr.patch.gz, but in the correct\nformat (context diff), and the 'wal_arch_dir' parameter of\npostgresql.conf.sample is changed to 'wal_archive_dest', which is what\nit should have been.\n\n\n-- \nJ. R. Nield\njrnield@usol.com", "msg_date": "22 Aug 2002 23:58:47 -0400", "msg_from": "\"J. R. Nield\" <jrnield@usol.com>", "msg_from_op": true, "msg_subject": "20020822_02_pitr.patch.gz" }, { "msg_contents": "\nJ.R, we have split off CVS for 7.4 so if you can supply a PITR patch I\nwill get it applied after the typical review.\n\n---------------------------------------------------------------------------\n\nJ. R. Nield wrote:\n> This is the same patch as 20020822_01_pitr.patch.gz, but in the correct\n> format (context diff), and the 'wal_arch_dir' parameter of\n> postgresql.conf.sample is changed to 'wal_archive_dest', which is what\n> it should have been.\n> \n> \n> -- \n> J. R. Nield\n> jrnield@usol.com\n> \n> \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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sat, 28 Sep 2002 22:04:49 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: 20020822_02_pitr.patch.gz" }, { "msg_contents": "\nJ.R, do you have an updated version of this patch. We are ready for\n7.4 development.\n\n---------------------------------------------------------------------------\n\nJ. R. Nield wrote:\n> This is the same patch as 20020822_01_pitr.patch.gz, but in the correct\n> format (context diff), and the 'wal_arch_dir' parameter of\n> postgresql.conf.sample is changed to 'wal_archive_dest', which is what\n> it should have been.\n> \n> \n> -- \n> J. R. Nield\n> jrnield@usol.com\n> \n> \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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 6 Nov 2002 10:42:00 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: 20020822_02_pitr.patch.gz" }, { "msg_contents": "\nJ.R. are you still around? Do you want your patch reviewed/applied?\n\n---------------------------------------------------------------------------\n\nJ. R. Nield wrote:\n> This is the same patch as 20020822_01_pitr.patch.gz, but in the correct\n> format (context diff), and the 'wal_arch_dir' parameter of\n> postgresql.conf.sample is changed to 'wal_archive_dest', which is what\n> it should have been.\n> \n> \n> -- \n> J. R. Nield\n> jrnield@usol.com\n> \n> \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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Thu, 14 Nov 2002 23:07:19 -0500 (EST)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [PATCHES] 20020822_02_pitr.patch.gz" } ]
[ { "msg_contents": "Hi,\n\nThe current prompt in psql for password is merely\n\nPassword:\n\nwhich is Ok when one is using psql interactively as one know exactly\nwhich username and database server is specified. However, when using it\nfor programs like vacuumdb, createuser, createdb, clusterdb etc is not\nimmediately obvious which (whose?) password it is asking.\n\nIs there a reason not to change it to something like\nusername@host:port Password:\n?\n\nIt's a trivial change and I can send a patch if people agree.\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\nA male gynecologist is like an auto mechanic who never owned a car.\n- Carrie Snow\n", "msg_date": "Fri, 23 Aug 2002 01:01:00 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "psql password prompt" }, { "msg_contents": "Alvaro Herrera <alvherre@atentus.com> writes:\n> Is there a reason not to change it to something like\n> username@host:port Password:\n\nNot sure. I can't immediately spot any security risk in this, but that\ndoesn't mean there isn't any. It bothers me that I cannot think of\n*any* other password-prompting program that gives you such feedback.\nSeems like if this were really a good idea, we'd not be the first to\ndo it ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 23 Aug 2002 01:12:06 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: psql password prompt " }, { "msg_contents": "On Fri, 23 Aug 2002, Tom Lane wrote:\n\n> Alvaro Herrera <alvherre@atentus.com> writes:\n> > Is there a reason not to change it to something like\n> > username@host:port Password:\n> \n> Not sure. I can't immediately spot any security risk in this, but that\n> doesn't mean there isn't any. It bothers me that I cannot think of\n> *any* other password-prompting program that gives you such feedback.\n\n[swm@laptop swm]$ ssh zipperii.zip.com.au\nswm@zipperii.zip.com.au's password:\n\n:-)\n\nGavin\n\n", "msg_date": "Fri, 23 Aug 2002 15:19:05 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": false, "msg_subject": "Re: psql password prompt " }, { "msg_contents": "Gavin Sherry <swm@linuxworld.com.au> writes:\n> On Fri, 23 Aug 2002, Tom Lane wrote:\n>> It bothers me that I cannot think of\n>> *any* other password-prompting program that gives you such feedback.\n\n> [swm@laptop swm]$ ssh zipperii.zip.com.au\n> swm@zipperii.zip.com.au's password:\n\nDuh. Okay, complaint withdrawn ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 23 Aug 2002 01:33:29 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: psql password prompt " }, { "msg_contents": "En Fri, 23 Aug 2002 01:12:06 -0400\nTom Lane <tgl@sss.pgh.pa.us> escribi�:\n\n> Alvaro Herrera <alvherre@atentus.com> writes:\n> > Is there a reason not to change it to something like\n> > username@host:port Password:\n> \n> Not sure. I can't immediately spot any security risk in this, but that\n> doesn't mean there isn't any. It bothers me that I cannot think of\n> *any* other password-prompting program that gives you such feedback.\n> Seems like if this were really a good idea, we'd not be the first to\n> do it ...\n\nIt's not that easy anyway. Generally psql does not have a lot of\ninformation about connection options: those are deduced from the\nenvironment by libpq and psql does not have access to what libpq\nguesses.\n\nIt can be resolved using the same logic as libpq in psql. It seems a\nbad idea to just copy the code; another way would be separating libpq's\nlogic in an exportable function so psql can call it.\n\nHowever, as this turns out to be more difficult than I had originally\nthought, I think it's not THAT useful so I rather leave it alone, unless\nsomebody thinks it's useful.\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"Para tener mas hay que desear menos\"\n", "msg_date": "Fri, 23 Aug 2002 15:20:48 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "Re: psql password prompt" } ]
[ { "msg_contents": "Anyone seen this:\n\nhttp://www.palslib.com/sitemap.html\n\nLoads of fantastic DBMS articles - many from techdocs. The XML & Database\nsection is really interesting and is already giving me heaps of ideas about\nimproving the XML stuff for later releases.\n\neg.\n\nhttp://www.rpbourret.com/xml/XMLAndDatabases.htm\n\nJustin - might it be worth linking to this site from TechDocs?\n\nChris\n\n\n", "msg_date": "Fri, 23 Aug 2002 14:24:37 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "Excellent DBMS Resource Site" }, { "msg_contents": "Hi Chris,\n\nChristopher Kings-Lynne wrote:\n> \n> Anyone seen this:\n> \n> http://www.palslib.com/sitemap.html\n> \n> Loads of fantastic DBMS articles - many from techdocs. The XML & Database\n> section is really interesting and is already giving me heaps of ideas about\n> improving the XML stuff for later releases.\n> \n> eg.\n> \n> http://www.rpbourret.com/xml/XMLAndDatabases.htm\n> \n> Justin - might it be worth linking to this site from TechDocs?\n\nCool. Adding it now.\n\n:)\n\nRegards and best wishes,\n\nJustin Clift\n\n \n> Chris\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\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\n", "msg_date": "Fri, 23 Aug 2002 16:34:47 +1000", "msg_from": "Justin Clift <justin@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] Excellent DBMS Resource Site" } ]
[ { "msg_contents": "Greeting Hackers! Had some questions about integer array type...\n\nIf I have a integer[] field in a table, which already has the values\n{1,2,3,6,17} and I additionally want to add the value '45' to this\narray, is there any easier way than SELECTing the current values and\nthen updating the entire record?\n\nI see that I could do:\nUPDATE table SET array_field[6] = 45 WHERE ...\n\nBut how do I know that 6 is the correct index without first SELECTing\nall the data?\n\nIs there an array push/pop function? Is there a method of counting the\nnumber of integers in an array?\n\nThanks for any help you can provide - indexable integer arrays are an\nincredible feature, thanks for the hard work!\n\nRyan Mahoney\n", "msg_date": "23 Aug 2002 10:39:42 -0400", "msg_from": "Ryan Mahoney <ryan@paymentalliance.net>", "msg_from_op": true, "msg_subject": "integer[] Update Questions" }, { "msg_contents": "Well, CVS version (7.3) of contrib/intarray has:\nFUNCTIONS:\n\n int icount(int[]) - the number of elements in intarray\n int[] sort(int[], 'asc' | 'desc') - sort intarray\n int[] sort(int[]) - sort in ascending order\n int[] sort_asc(int[]),sort_desc(int[]) - shortcuts for sort\n int[] uniq(int[]) - returns unique elements\n int idx(int[], int item) - returns index of first intarray matching element to item, or\n '0' if matching failed.\n int[] subarray(int[],int START [, int LEN]) - returns part of intarray starting from\n element number START (from 1) and length LEN.\n\nOPERATIONS:\n\n\n int[] && int[] - overlap - returns TRUE if arrays has at least one common elements.\n int[] @ int[] - contains - returns TRUE if left array contains right array\n int[] ~ int[] - contained - returns TRUE if left array is contained in right array\n # int[] - return the number of elements in array\n int[] + int - push element to array ( add to end of array)\n int[] + int[] - merge of arrays (right array added to the end of left one)\n int[] - int - remove entries matched by right argument from array\n int[] - int[] - remove left array from right\n int[] | int - returns intarray - union of arguments\n int[] | int[] - returns intarray as a union of two arrays\n int[] & int[] - returns intersection of arrays\n\n\nI don't remember if I have a patch for 7.2\n\n\tOleg\nOn 23 Aug 2002, Ryan Mahoney wrote:\n\n> Greeting Hackers! Had some questions about integer array type...\n>\n> If I have a integer[] field in a table, which already has the values\n> {1,2,3,6,17} and I additionally want to add the value '45' to this\n> array, is there any easier way than SELECTing the current values and\n> then updating the entire record?\n>\n> I see that I could do:\n> UPDATE table SET array_field[6] = 45 WHERE ...\n>\n> But how do I know that 6 is the correct index without first SELECTing\n> all the data?\n>\n> Is there an array push/pop function? Is there a method of counting the\n> number of integers in an array?\n>\n> Thanks for any help you can provide - indexable integer arrays are an\n> incredible feature, thanks for the hard work!\n>\n> Ryan Mahoney\n>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n>\n\n\tRegards,\n\t\tOleg\n_____________________________________________________________\nOleg Bartunov, sci.researcher, hostmaster of AstroNet,\nSternberg Astronomical Institute, Moscow University (Russia)\nInternet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/\nphone: +007(095)939-16-83, +007(095)939-23-83\n\n", "msg_date": "Fri, 23 Aug 2002 19:40:43 +0300 (GMT)", "msg_from": "Oleg Bartunov <oleg@sai.msu.su>", "msg_from_op": false, "msg_subject": "Re: integer[] Update Questions" }, { "msg_contents": "Wow Oleg, that's great to hear! I am compiling 7.3 from CVS right now. \nIf you do have a patch for 7.2 could you send that - I would like to use\nthese functions on some production machines running 7.2.1 currently.\n\nThanks for the prompt response and keep up the great work!\n\nRyan Mahoney\n\nOn Fri, 2002-08-23 at 12:40, Oleg Bartunov wrote:\n> Well, CVS version (7.3) of contrib/intarray has:\n> FUNCTIONS:\n> \n> int icount(int[]) - the number of elements in intarray\n> int[] sort(int[], 'asc' | 'desc') - sort intarray\n> int[] sort(int[]) - sort in ascending order\n> int[] sort_asc(int[]),sort_desc(int[]) - shortcuts for sort\n> int[] uniq(int[]) - returns unique elements\n> int idx(int[], int item) - returns index of first intarray matching element to item, or\n> '0' if matching failed.\n> int[] subarray(int[],int START [, int LEN]) - returns part of intarray starting from\n> element number START (from 1) and length LEN.\n> \n> OPERATIONS:\n> \n> \n> int[] && int[] - overlap - returns TRUE if arrays has at least one common elements.\n> int[] @ int[] - contains - returns TRUE if left array contains right array\n> int[] ~ int[] - contained - returns TRUE if left array is contained in right array\n> # int[] - return the number of elements in array\n> int[] + int - push element to array ( add to end of array)\n> int[] + int[] - merge of arrays (right array added to the end of left one)\n> int[] - int - remove entries matched by right argument from array\n> int[] - int[] - remove left array from right\n> int[] | int - returns intarray - union of arguments\n> int[] | int[] - returns intarray as a union of two arrays\n> int[] & int[] - returns intersection of arrays\n> \n> \n> I don't remember if I have a patch for 7.2\n> \n> \tOleg\n> On 23 Aug 2002, Ryan Mahoney wrote:\n> \n> > Greeting Hackers! Had some questions about integer array type...\n> >\n> > If I have a integer[] field in a table, which already has the values\n> > {1,2,3,6,17} and I additionally want to add the value '45' to this\n> > array, is there any easier way than SELECTing the current values and\n> > then updating the entire record?\n> >\n> > I see that I could do:\n> > UPDATE table SET array_field[6] = 45 WHERE ...\n> >\n> > But how do I know that 6 is the correct index without first SELECTing\n> > all the data?\n> >\n> > Is there an array push/pop function? Is there a method of counting the\n> > number of integers in an array?\n> >\n> > Thanks for any help you can provide - indexable integer arrays are an\n> > incredible feature, thanks for the hard work!\n> >\n> > Ryan Mahoney\n> >\n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 4: Don't 'kill -9' the postmaster\n> >\n> \n> \tRegards,\n> \t\tOleg\n> _____________________________________________________________\n> Oleg Bartunov, sci.researcher, hostmaster of AstroNet,\n> Sternberg Astronomical Institute, Moscow University (Russia)\n> Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/\n> phone: +007(095)939-16-83, +007(095)939-23-83\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n> \n\n", "msg_date": "23 Aug 2002 13:41:49 -0400", "msg_from": "Ryan Mahoney <ryan@paymentalliance.net>", "msg_from_op": true, "msg_subject": "Re: [GENERAL] integer[] Update Questions" }, { "msg_contents": "\nWell, we think beta is starting in 8 days, so you could be an _early_\nbeta tester.\n\n---------------------------------------------------------------------------\n\nRyan Mahoney wrote:\n> Wow Oleg, that's great to hear! I am compiling 7.3 from CVS right now. \n> If you do have a patch for 7.2 could you send that - I would like to use\n> these functions on some production machines running 7.2.1 currently.\n> \n> Thanks for the prompt response and keep up the great work!\n> \n> Ryan Mahoney\n> \n> On Fri, 2002-08-23 at 12:40, Oleg Bartunov wrote:\n> > Well, CVS version (7.3) of contrib/intarray has:\n> > FUNCTIONS:\n> > \n> > int icount(int[]) - the number of elements in intarray\n> > int[] sort(int[], 'asc' | 'desc') - sort intarray\n> > int[] sort(int[]) - sort in ascending order\n> > int[] sort_asc(int[]),sort_desc(int[]) - shortcuts for sort\n> > int[] uniq(int[]) - returns unique elements\n> > int idx(int[], int item) - returns index of first intarray matching element to item, or\n> > '0' if matching failed.\n> > int[] subarray(int[],int START [, int LEN]) - returns part of intarray starting from\n> > element number START (from 1) and length LEN.\n> > \n> > OPERATIONS:\n> > \n> > \n> > int[] && int[] - overlap - returns TRUE if arrays has at least one common elements.\n> > int[] @ int[] - contains - returns TRUE if left array contains right array\n> > int[] ~ int[] - contained - returns TRUE if left array is contained in right array\n> > # int[] - return the number of elements in array\n> > int[] + int - push element to array ( add to end of array)\n> > int[] + int[] - merge of arrays (right array added to the end of left one)\n> > int[] - int - remove entries matched by right argument from array\n> > int[] - int[] - remove left array from right\n> > int[] | int - returns intarray - union of arguments\n> > int[] | int[] - returns intarray as a union of two arrays\n> > int[] & int[] - returns intersection of arrays\n> > \n> > \n> > I don't remember if I have a patch for 7.2\n> > \n> > \tOleg\n> > On 23 Aug 2002, Ryan Mahoney wrote:\n> > \n> > > Greeting Hackers! Had some questions about integer array type...\n> > >\n> > > If I have a integer[] field in a table, which already has the values\n> > > {1,2,3,6,17} and I additionally want to add the value '45' to this\n> > > array, is there any easier way than SELECTing the current values and\n> > > then updating the entire record?\n> > >\n> > > I see that I could do:\n> > > UPDATE table SET array_field[6] = 45 WHERE ...\n> > >\n> > > But how do I know that 6 is the correct index without first SELECTing\n> > > all the data?\n> > >\n> > > Is there an array push/pop function? Is there a method of counting the\n> > > number of integers in an array?\n> > >\n> > > Thanks for any help you can provide - indexable integer arrays are an\n> > > incredible feature, thanks for the hard work!\n> > >\n> > > Ryan Mahoney\n> > >\n> > > ---------------------------(end of broadcast)---------------------------\n> > > TIP 4: Don't 'kill -9' the postmaster\n> > >\n> > \n> > \tRegards,\n> > \t\tOleg\n> > _____________________________________________________________\n> > Oleg Bartunov, sci.researcher, hostmaster of AstroNet,\n> > Sternberg Astronomical Institute, Moscow University (Russia)\n> > Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/\n> > phone: +007(095)939-16-83, +007(095)939-23-83\n> > \n> > \n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n> > \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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Fri, 23 Aug 2002 13:46:03 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] integer[] Update Questions" }, { "msg_contents": "On 23 Aug 2002, Ryan Mahoney wrote:\n\n> Wow Oleg, that's great to hear! I am compiling 7.3 from CVS right now.\n> If you do have a patch for 7.2 could you send that - I would like to use\n> these functions on some production machines running 7.2.1 currently.\n\nunfortuneately, we don't have a *reliable* patch for 7.2 :-(\nbut Beta is started (should) very soon, so you'd be able to test your\nreal life application :0\n\n\tOleg\n>\n> Thanks for the prompt response and keep up the great work!\n>\n> Ryan Mahoney\n>\n> On Fri, 2002-08-23 at 12:40, Oleg Bartunov wrote:\n> > Well, CVS version (7.3) of contrib/intarray has:\n> > FUNCTIONS:\n> >\n> > int icount(int[]) - the number of elements in intarray\n> > int[] sort(int[], 'asc' | 'desc') - sort intarray\n> > int[] sort(int[]) - sort in ascending order\n> > int[] sort_asc(int[]),sort_desc(int[]) - shortcuts for sort\n> > int[] uniq(int[]) - returns unique elements\n> > int idx(int[], int item) - returns index of first intarray matching element to item, or\n> > '0' if matching failed.\n> > int[] subarray(int[],int START [, int LEN]) - returns part of intarray starting from\n> > element number START (from 1) and length LEN.\n> >\n> > OPERATIONS:\n> >\n> >\n> > int[] && int[] - overlap - returns TRUE if arrays has at least one common elements.\n> > int[] @ int[] - contains - returns TRUE if left array contains right array\n> > int[] ~ int[] - contained - returns TRUE if left array is contained in right array\n> > # int[] - return the number of elements in array\n> > int[] + int - push element to array ( add to end of array)\n> > int[] + int[] - merge of arrays (right array added to the end of left one)\n> > int[] - int - remove entries matched by right argument from array\n> > int[] - int[] - remove left array from right\n> > int[] | int - returns intarray - union of arguments\n> > int[] | int[] - returns intarray as a union of two arrays\n> > int[] & int[] - returns intersection of arrays\n> >\n> >\n> > I don't remember if I have a patch for 7.2\n> >\n> > \tOleg\n> > On 23 Aug 2002, Ryan Mahoney wrote:\n> >\n> > > Greeting Hackers! Had some questions about integer array type...\n> > >\n> > > If I have a integer[] field in a table, which already has the values\n> > > {1,2,3,6,17} and I additionally want to add the value '45' to this\n> > > array, is there any easier way than SELECTing the current values and\n> > > then updating the entire record?\n> > >\n> > > I see that I could do:\n> > > UPDATE table SET array_field[6] = 45 WHERE ...\n> > >\n> > > But how do I know that 6 is the correct index without first SELECTing\n> > > all the data?\n> > >\n> > > Is there an array push/pop function? Is there a method of counting the\n> > > number of integers in an array?\n> > >\n> > > Thanks for any help you can provide - indexable integer arrays are an\n> > > incredible feature, thanks for the hard work!\n> > >\n> > > Ryan Mahoney\n> > >\n> > > ---------------------------(end of broadcast)---------------------------\n> > > TIP 4: Don't 'kill -9' the postmaster\n> > >\n> >\n> > \tRegards,\n> > \t\tOleg\n> > _____________________________________________________________\n> > Oleg Bartunov, sci.researcher, hostmaster of AstroNet,\n> > Sternberg Astronomical Institute, Moscow University (Russia)\n> > Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/\n> > phone: +007(095)939-16-83, +007(095)939-23-83\n> >\n> >\n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n> >\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\tRegards,\n\t\tOleg\n_____________________________________________________________\nOleg Bartunov, sci.researcher, hostmaster of AstroNet,\nSternberg Astronomical Institute, Moscow University (Russia)\nInternet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/\nphone: +007(095)939-16-83, +007(095)939-23-83\n\n", "msg_date": "Fri, 23 Aug 2002 23:06:39 +0300 (GMT)", "msg_from": "Oleg Bartunov <oleg@sai.msu.su>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] integer[] Update Questions" } ]
[ { "msg_contents": "Sorry all,\n\nI got in a hurry and posted before clearing out the Cc stuff. I only\nmeant to bother Lamar. :-)\n\n\nRod\n-- \n \"Open Source Software - Sometimes you get more than you paid for...\"\n\n", "msg_date": "Fri, 23 Aug 2002 10:05:53 -0700 (PDT)", "msg_from": "\"Roderick A. Anderson\" <raanders@acm.org>", "msg_from_op": true, "msg_subject": "Sorry about previous post." } ]
[ { "msg_contents": "\nJust to give time for the mirrors to pick up the directory change, and\nVince to change his pointers, I'm going to announce the release later this\nevening ... both the binary and source directories have been renamed\naccordingly, as well as appropriate links changed from the top of hte ftp\ntree itself ...\n\n\n\n", "msg_date": "Fri, 23 Aug 2002 14:22:06 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "v7.2.2 Released ... but not announced ..." }, { "msg_contents": "On Friday 23 August 2002 01:22 pm, Marc G. Fournier wrote:\n> Just to give time for the mirrors to pick up the directory change, and\n> Vince to change his pointers, I'm going to announce the release later this\n> evening ... both the binary and source directories have been renamed\n> accordingly, as well as appropriate links changed from the top of hte ftp\n> tree itself ...\n\nOh, Ok. I was going to do that for the binary tree, but since you already did \nit.... :-)\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Fri, 23 Aug 2002 14:16:02 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: v7.2.2 Released ... but not announced ..." }, { "msg_contents": "On Fri, 23 Aug 2002, Marc G. Fournier wrote:\n\n>\n> Just to give time for the mirrors to pick up the directory change, and\n> Vince to change his pointers, I'm going to announce the release later this\n> evening ... both the binary and source directories have been renamed\n> accordingly, as well as appropriate links changed from the top of hte ftp\n> tree itself ...\n\nYeah, I've already been hearing about the directory changes :)\n\nI've got the software webpage changed and am ready for the announcement\nfor the announcement (home) and news pages. An advance copy of that\nannouncement would be appreciated Marc. Or at least CC me on the real\nthing. Either will work.\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n", "msg_date": "Fri, 23 Aug 2002 18:38:41 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": false, "msg_subject": "Re: v7.2.2 Released ... but not announced ..." } ]
[ { "msg_contents": "The new CREATE OPERATOR CLASS command will presently let you create an\nindex opclass if you own the datatype the class is for. With the\nrecent emphasis on security I'm thinking that this is not an adequate\npermission check. We don't have any reasonable way of checking that\nthe provided set of operators and support functions meet the\nexpectations of the index AM and are mutually consistent. This means\nit's not at all difficult to make an index opclass that will crash\nthe backend when used.\n\nI'm inclined to require superuser permissions to do CREATE OPERATOR\nCLASS. This would not be a loss of functionality compared to prior\nreleases, since the old way of creating an opclass involved manual\ninsertions into system catalogs, also a superuser-only thing.\n\nComments?\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 23 Aug 2002 15:17:25 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Permissions for CREATE OPERATOR CLASS" }, { "msg_contents": "En Fri, 23 Aug 2002 15:17:25 -0400\nTom Lane <tgl@sss.pgh.pa.us> escribi�:\n\n> The new CREATE OPERATOR CLASS command will presently let you create an\n> index opclass if you own the datatype the class is for. With the\n> recent emphasis on security I'm thinking that this is not an adequate\n> permission check. We don't have any reasonable way of checking that\n> the provided set of operators and support functions meet the\n> expectations of the index AM and are mutually consistent. This means\n> it's not at all difficult to make an index opclass that will crash\n> the backend when used.\n\nWell, maybe this reduces the usefulness of having a CREATE TYPE as\ncompared to being able to create operator classes as well, but I wonder\nhow much people create their own types and want to create operator\nclasses without being superusers.\n\nIs it too difficult to check whether the functions are \"good for AM\"?\nDoes not sound like an easy task... (halting problem maybe?)\n\n> I'm inclined to require superuser permissions to do CREATE OPERATOR\n> CLASS. This would not be a loss of functionality compared to prior\n> releases, since the old way of creating an opclass involved manual\n> insertions into system catalogs, also a superuser-only thing.\n\nIf it's unsafe, users should not be able to mess with it.\n\n\nI was playing around and got this:\n\nalvh=> create type my_cash (input = cash_in, output = cash_out, internallength = variable);\nERROR: TypeCreate: function cash_out(opaque) does not exist\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"La fuerza no est� en los medios f�sicos\nsino que reside en una voluntad indomable\" (Gandhi)\n", "msg_date": "Fri, 23 Aug 2002 15:42:14 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": false, "msg_subject": "Re: Permissions for CREATE OPERATOR CLASS" }, { "msg_contents": "Alvaro Herrera <alvherre@atentus.com> writes:\n> I was playing around and got this:\n\n> alvh=> create type my_cash (input = cash_in, output = cash_out, internallength = variable);\n> ERROR: TypeCreate: function cash_out(opaque) does not exist\n\nYeah, the CREATE TYPE code is now much stricter about the allowed\nsignatures of the I/O functions. The input function for a type foo\nmust be one of\n\tfoo_in(cstring) returns foo\n\tfoo_in(opaque) returns foo\n\tfoo_in(cstring) returns opaque\n\tfoo_in(opaque) returns opaque\nwhile the output must be one of\n\tfoo_out(foo) returns cstring\n\tfoo_out(opaque) returns cstring\n\tfoo_out(foo) returns opaque\n\tfoo_out(opaque) returns opaque\nso unless you use \"opaque\" you cannot use the same I/O function for\ntwo different types. (Possibly the error message should complain about\nfoo_out(foo) not foo_out(opaque).)\n\nIf we hear a lot of squawks about that, we can discuss how to weaken\nthe rules ... but IMHO the entire point here is to introduce some type\nsafety into the use of I/O functions ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 23 Aug 2002 18:01:38 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "Re: Permissions for CREATE OPERATOR CLASS " } ]
[ { "msg_contents": "This should not happen, I guess:\n\nalvh=> CREATE TABLE test_inh (a int);\nCREATE TABLE\nalvh=> CREATE TABLE test_inh_child (b int) INHERITS (test_inh);\nCREATE TABLE\nalvh=> ALTER TABLE test_inh_child RENAME a TO c;\nALTER TABLE\nalvh=> SELECT * FROM test_inh;\nERROR: Relation \"test_inh_child\" has no column \"a\"\n\nalvh=> ALTER TABLE test_inh_child RENAME c TO a;\nALTER TABLE\nalvh=> ALTER TABLE test_inh_child DROP COLUMN a;\nALTER TABLE\nalvh=> SELECT * FROM test_inh;\nERROR: Relation \"test_inh_child\" has no column \"a\"\n\n\nI remember Tom suggested adding something like attisinherited and\npreventing this kind of operations on such attributes, because one can\ndo things such as\n\nalvh=> ALTER TABLE test_inh_child ADD COLUMN a TEXT;\nALTER TABLE\nalvh=> INSERT INTO test_inh_child VALUES (1, 'hello world');\nINSERT 33449 1\nalvh=> SELECT * FROM test_inh;\nserver closed the connection unexpectedly\n This probably means the server terminated abnormally\n before or while processing the request.\nThe connection to the server was lost. Attempting reset: Failed.\n!> \n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"Entristecido, Wutra\necha a Freyr a rodar\ny a nosotros al mar\" (cancion de Las Barreras, Heliconia)\n\n", "msg_date": "Fri, 23 Aug 2002 18:14:28 -0400 (CLT)", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "pg_attribute.attisinherited ?" }, { "msg_contents": "On Fri, 23 Aug 2002 18:14:28 -0400 (CLT)\nI said:\n\n> I remember Tom suggested adding something like attisinherited and\n> preventing this kind of operations on such attributes, because one can\n> do things such as [...]\n\nWell, maybe nobody cares or are just too busy (maybe it's weekend)... \nanyway I made a patch that creates the attisinherited attribute and I\nthink I got it working (that is, inherited attributes have it set and\nfor non-inherited it is false). I haven't yet written checks for\ndisallowing the unwanted operations, though.\n\nI will test it some more and post a patch later.\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"Como puedes confiar en algo que pagas y que no ves,\ny no confiar en algo que te dan y te lo muestran?\" (German Poo)\n", "msg_date": "Sat, 24 Aug 2002 17:09:55 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "Re: pg_attribute.attisinherited ?" }, { "msg_contents": "En Sat, 24 Aug 2002 17:09:55 -0400\nI said:\n\n> > I remember Tom suggested adding something like attisinherited and\n> > preventing this kind of operations on such attributes, because one can\n> > do things such as [...]\n\nOk, I attach a patch that does this. It doesn't include regression\ntests, docs nor the checks against unwanted operations; these will come\nlater if people think this is a good approach.\n\nIt passes 86 of 88 tests. The 2 failures are ordering issues (diff\nbelow) I don't know what causes it.\n\nPlease review.\n\n\n*** ./expected/select_having.out\tWed Jun 26 17:58:56 2002\n--- ./results/select_having.out\tSat Aug 24 18:32:16 2002\n***************\n*** 26,33 ****\n \tGROUP BY b, c HAVING b = 3; \n b | c \n ---+----------\n- 3 | BBBB \n 3 | bbbb \n (2 rows)\n \n SELECT lower(c), count(c) FROM test_having\n--- 26,33 ----\n \tGROUP BY b, c HAVING b = 3; \n b | c \n ---+----------\n 3 | bbbb \n+ 3 | BBBB \n (2 rows)\n \n SELECT lower(c), count(c) FROM test_having\n***************\n*** 43,50 ****\n \tGROUP BY c HAVING count(*) > 2 OR min(a) = max(a);\n c | max \n ----------+-----\n- XXXX | 0\n bbbb | 5\n (2 rows)\n \n DROP TABLE test_having;\n--- 43,50 ----\n \tGROUP BY c HAVING count(*) > 2 OR min(a) = max(a);\n c | max \n ----------+-----\n bbbb | 5\n+ XXXX | 0\n (2 rows)\n \n DROP TABLE test_having;\n\n======================================================================\n\n*** ./expected/rules.out\tMon Aug 19 01:08:30 2002\n--- ./results/rules.out\tSat Aug 24 18:32:46 2002\n***************\n*** 404,412 ****\n ----------------------+--------------+------------+------------+------------\n gates | t | fired | $0.00 | $80,000.00\n gates | t | hired | $80,000.00 | $0.00\n- wiech | t | hired | $5,000.00 | $0.00\n wieck | t | honored | $6,000.00 | $5,000.00\n wieck | t | honored | $7,000.00 | $6,000.00\n (5 rows)\n \n insert into rtest_empmass values ('meyer', '4000.00');\n--- 404,412 ----\n ----------------------+--------------+------------+------------+------------\n gates | t | fired | $0.00 | $80,000.00\n gates | t | hired | $80,000.00 | $0.00\n wieck | t | honored | $6,000.00 | $5,000.00\n wieck | t | honored | $7,000.00 | $6,000.00\n+ wiech | t | hired | $5,000.00 | $0.00\n (5 rows)\n \n insert into rtest_empmass values ('meyer', '4000.00');\n***************\n*** 421,429 ****\n maier | t | hired | $5,000.00 | $0.00\n mayr | t | hired | $6,000.00 | $0.00\n meyer | t | hired | $4,000.00 | $0.00\n- wiech | t | hired | $5,000.00 | $0.00\n wieck | t | honored | $6,000.00 | $5,000.00\n wieck | t | honored | $7,000.00 | $6,000.00\n (8 rows)\n \n update rtest_empmass set salary = salary + '1000.00';\n--- 421,429 ----\n maier | t | hired | $5,000.00 | $0.00\n mayr | t | hired | $6,000.00 | $0.00\n meyer | t | hired | $4,000.00 | $0.00\n wieck | t | honored | $6,000.00 | $5,000.00\n wieck | t | honored | $7,000.00 | $6,000.00\n+ wiech | t | hired | $5,000.00 | $0.00\n (8 rows)\n \n update rtest_empmass set salary = salary + '1000.00';\n***************\n*** 439,447 ****\n mayr | t | honored | $7,000.00 | $6,000.00\n meyer | t | hired | $4,000.00 | $0.00\n meyer | t | honored | $5,000.00 | $4,000.00\n- wiech | t | hired | $5,000.00 | $0.00\n wieck | t | honored | $6,000.00 | $5,000.00\n wieck | t | honored | $7,000.00 | $6,000.00\n (11 rows)\n \n delete from rtest_emp where ename = rtest_empmass.ename;\n--- 439,447 ----\n mayr | t | honored | $7,000.00 | $6,000.00\n meyer | t | hired | $4,000.00 | $0.00\n meyer | t | honored | $5,000.00 | $4,000.00\n wieck | t | honored | $6,000.00 | $5,000.00\n wieck | t | honored | $7,000.00 | $6,000.00\n+ wiech | t | hired | $5,000.00 | $0.00\n (11 rows)\n \n delete from rtest_emp where ename = rtest_empmass.ename;\n***************\n*** 459,467 ****\n meyer | t | fired | $0.00 | $5,000.00\n meyer | t | hired | $4,000.00 | $0.00\n meyer | t | honored | $5,000.00 | $4,000.00\n- wiech | t | hired | $5,000.00 | $0.00\n wieck | t | honored | $6,000.00 | $5,000.00\n wieck | t | honored | $7,000.00 | $6,000.00\n (14 rows)\n \n --\n--- 459,467 ----\n meyer | t | fired | $0.00 | $5,000.00\n meyer | t | hired | $4,000.00 | $0.00\n meyer | t | honored | $5,000.00 | $4,000.00\n wieck | t | honored | $6,000.00 | $5,000.00\n wieck | t | honored | $7,000.00 | $6,000.00\n+ wiech | t | hired | $5,000.00 | $0.00\n (14 rows)\n \n --\n\n======================================================================\n\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"Un poeta es un mundo encerrado en un hombre\" (Victor Hugo)", "msg_date": "Sat, 24 Aug 2002 18:49:12 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" }, { "msg_contents": "Hi Alvaro,\n\nYeah it is an issue that needs to be fixed. I'll have a look at your\npatch once someone comments on those ordering issue?\n\nAlthough I note that I've had ordering issues in the RULES test for ages\nnow on FreeBSD/Alpha...\n\nChris\n\nOn Sat, 24 Aug 2002, Alvaro Herrera wrote:\n\n> En Sat, 24 Aug 2002 17:09:55 -0400\n> I said:\n>\n> > > I remember Tom suggested adding something like attisinherited and\n> > > preventing this kind of operations on such attributes, because one can\n> > > do things such as [...]\n>\n> Ok, I attach a patch that does this. It doesn't include regression\n> tests, docs nor the checks against unwanted operations; these will come\n> later if people think this is a good approach.\n>\n> It passes 86 of 88 tests. The 2 failures are ordering issues (diff\n> below) I don't know what causes it.\n>\n> Please review.\n>\n>\n> *** ./expected/select_having.out\tWed Jun 26 17:58:56 2002\n> --- ./results/select_having.out\tSat Aug 24 18:32:16 2002\n> ***************\n> *** 26,33 ****\n> \tGROUP BY b, c HAVING b = 3;\n> b | c\n> ---+----------\n> - 3 | BBBB\n> 3 | bbbb\n> (2 rows)\n>\n> SELECT lower(c), count(c) FROM test_having\n> --- 26,33 ----\n> \tGROUP BY b, c HAVING b = 3;\n> b | c\n> ---+----------\n> 3 | bbbb\n> + 3 | BBBB\n> (2 rows)\n>\n> SELECT lower(c), count(c) FROM test_having\n> ***************\n> *** 43,50 ****\n> \tGROUP BY c HAVING count(*) > 2 OR min(a) = max(a);\n> c | max\n> ----------+-----\n> - XXXX | 0\n> bbbb | 5\n> (2 rows)\n>\n> DROP TABLE test_having;\n> --- 43,50 ----\n> \tGROUP BY c HAVING count(*) > 2 OR min(a) = max(a);\n> c | max\n> ----------+-----\n> bbbb | 5\n> + XXXX | 0\n> (2 rows)\n>\n> DROP TABLE test_having;\n>\n> ======================================================================\n>\n> *** ./expected/rules.out\tMon Aug 19 01:08:30 2002\n> --- ./results/rules.out\tSat Aug 24 18:32:46 2002\n> ***************\n> *** 404,412 ****\n> ----------------------+--------------+------------+------------+------------\n> gates | t | fired | $0.00 | $80,000.00\n> gates | t | hired | $80,000.00 | $0.00\n> - wiech | t | hired | $5,000.00 | $0.00\n> wieck | t | honored | $6,000.00 | $5,000.00\n> wieck | t | honored | $7,000.00 | $6,000.00\n> (5 rows)\n>\n> insert into rtest_empmass values ('meyer', '4000.00');\n> --- 404,412 ----\n> ----------------------+--------------+------------+------------+------------\n> gates | t | fired | $0.00 | $80,000.00\n> gates | t | hired | $80,000.00 | $0.00\n> wieck | t | honored | $6,000.00 | $5,000.00\n> wieck | t | honored | $7,000.00 | $6,000.00\n> + wiech | t | hired | $5,000.00 | $0.00\n> (5 rows)\n>\n> insert into rtest_empmass values ('meyer', '4000.00');\n> ***************\n> *** 421,429 ****\n> maier | t | hired | $5,000.00 | $0.00\n> mayr | t | hired | $6,000.00 | $0.00\n> meyer | t | hired | $4,000.00 | $0.00\n> - wiech | t | hired | $5,000.00 | $0.00\n> wieck | t | honored | $6,000.00 | $5,000.00\n> wieck | t | honored | $7,000.00 | $6,000.00\n> (8 rows)\n>\n> update rtest_empmass set salary = salary + '1000.00';\n> --- 421,429 ----\n> maier | t | hired | $5,000.00 | $0.00\n> mayr | t | hired | $6,000.00 | $0.00\n> meyer | t | hired | $4,000.00 | $0.00\n> wieck | t | honored | $6,000.00 | $5,000.00\n> wieck | t | honored | $7,000.00 | $6,000.00\n> + wiech | t | hired | $5,000.00 | $0.00\n> (8 rows)\n>\n> update rtest_empmass set salary = salary + '1000.00';\n> ***************\n> *** 439,447 ****\n> mayr | t | honored | $7,000.00 | $6,000.00\n> meyer | t | hired | $4,000.00 | $0.00\n> meyer | t | honored | $5,000.00 | $4,000.00\n> - wiech | t | hired | $5,000.00 | $0.00\n> wieck | t | honored | $6,000.00 | $5,000.00\n> wieck | t | honored | $7,000.00 | $6,000.00\n> (11 rows)\n>\n> delete from rtest_emp where ename = rtest_empmass.ename;\n> --- 439,447 ----\n> mayr | t | honored | $7,000.00 | $6,000.00\n> meyer | t | hired | $4,000.00 | $0.00\n> meyer | t | honored | $5,000.00 | $4,000.00\n> wieck | t | honored | $6,000.00 | $5,000.00\n> wieck | t | honored | $7,000.00 | $6,000.00\n> + wiech | t | hired | $5,000.00 | $0.00\n> (11 rows)\n>\n> delete from rtest_emp where ename = rtest_empmass.ename;\n> ***************\n> *** 459,467 ****\n> meyer | t | fired | $0.00 | $5,000.00\n> meyer | t | hired | $4,000.00 | $0.00\n> meyer | t | honored | $5,000.00 | $4,000.00\n> - wiech | t | hired | $5,000.00 | $0.00\n> wieck | t | honored | $6,000.00 | $5,000.00\n> wieck | t | honored | $7,000.00 | $6,000.00\n> (14 rows)\n>\n> --\n> --- 459,467 ----\n> meyer | t | fired | $0.00 | $5,000.00\n> meyer | t | hired | $4,000.00 | $0.00\n> meyer | t | honored | $5,000.00 | $4,000.00\n> wieck | t | honored | $6,000.00 | $5,000.00\n> wieck | t | honored | $7,000.00 | $6,000.00\n> + wiech | t | hired | $5,000.00 | $0.00\n> (14 rows)\n>\n> --\n>\n> ======================================================================\n>\n>\n> --\n> Alvaro Herrera (<alvherre[a]atentus.com>)\n> \"Un poeta es un mundo encerrado en un hombre\" (Victor Hugo)\n>\n\n", "msg_date": "Sun, 25 Aug 2002 17:34:21 +0800 (WST)", "msg_from": "Christopher Kings-Lynne <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" }, { "msg_contents": "En Sun, 25 Aug 2002 17:34:21 +0800 (WST)\nChristopher Kings-Lynne <chriskl@familyhealth.com.au> escribi�:\n\nHi again,\n\n> Yeah it is an issue that needs to be fixed.\n\nI'm thinking about the ONLY part in the grammar in ALTER TABLE... DROP\nCOLUMN and RENAME COLUMN. I think they should not be there: they only\ncreate noise and chances of ill behavior. If I modify only the parent\ntable, then I'm able to create a column on the child table with\ndifferent datatype and same name as new column on parent, causing\nsubsequent backend crash.\n\nConsider\n\nCREATE TABLE foo (a int);\nCREATE TABLE bar () INHERITS (foo);\nALTER TABLE ONLY foo RENAME a TO b;\nALTER TABLE bar ADD COLUMN b TEXT;\n\nregression=# INSERT INTO bar values (1, 'hello world');\nINSERT 205625 1\nregression=# SELECT * FROM foo;\nserver closed the connection unexpectedly\n\tThis probably means the server terminated abnormally\n\tbefore or while processing the request.\nThe connection to the server was lost. Attempting reset: Failed.\n!# \n\nWhat does people think about removing the support for ONLY in these\ndirectives?\n\nBut this is a different problem and requires a different patch.\n\n\nHere I post a new version of attisinherited; this one includes the tests\nin AlterTableDropColumn and renameatt so inherited columns can not be\ndropped nor renamed. Please review this new version. Regression tests\nare also included, as is the modification of catalog.sgml.\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"El Maquinismo fue proscrito so pena de cosquilleo hasta la muerte\"\n(Ijon Tichy en Viajes, Stanislaw Lem)", "msg_date": "Mon, 26 Aug 2002 05:28:21 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" }, { "msg_contents": "Alvaro Herrera <alvherre@atentus.com> writes:\n> I'm thinking about the ONLY part in the grammar in ALTER TABLE... DROP\n> COLUMN and RENAME COLUMN. I think they should not be there:\n\nLocal DROP COLUMN is fine: it just causes the column to become\nnon-inherited in any children. (Your patch for attisinherited will\nneed to cover this case.)\n\nLocal RENAME COLUMN does need to be prohibited, as does local ADD\nCOLUMN, as does local ALTER COLUMN if we ever allow changing column\ntype. Basically we need to prohibit the column from becoming\nincompatible with its children.\n\nI don't agree with the notion of changing the grammar to achieve that,\nbtw. Simpler and more friendly to add a specific error check in\n(most likely place) utility/tcop.c. Then if you try to say ONLY you'll\nget a more useful complaint than \"parse error\".\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 26 Aug 2002 10:26:01 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ? " }, { "msg_contents": "Tom Lane dijo: \n\n> Alvaro Herrera <alvherre@atentus.com> writes:\n> > I'm thinking about the ONLY part in the grammar in ALTER TABLE... DROP\n> > COLUMN and RENAME COLUMN. I think they should not be there:\n> \n> Local DROP COLUMN is fine: it just causes the column to become\n> non-inherited in any children. (Your patch for attisinherited will\n> need to cover this case.)\n\nOh, I see.\n\n> Local RENAME COLUMN does need to be prohibited, as does local ADD\n> COLUMN, as does local ALTER COLUMN if we ever allow changing column\n> type. Basically we need to prohibit the column from becoming\n> incompatible with its children.\n\n> I don't agree with the notion of changing the grammar to achieve that,\n> btw. Simpler and more friendly to add a specific error check in\n> (most likely place) utility/tcop.c. Then if you try to say ONLY you'll\n> get a more useful complaint than \"parse error\".\n\nUh, I added checks in the command itself (command/tablecmds.c), just\nbecause I had already done so and to not make tcop/utility.c messier\nthan it already is; I can probably move the check if people thinks it's\nbetter. Also implemented is the change from inherited to non-inherited\nwhen local-dropping a column.\n\nI also changed the text of some error messages from \"renameatt: cannot\nfoo\" to \"ALTER TABLE: cannot foo\". But my choose in wording of new\nerror messages probably needs improvement (suggestions welcome).\n\nPlease review; I have not received comments on whether this\nimplementation is a good approach: note the signature change of\nTupleDescInitEntry().\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"La espina, desde que nace, ya pincha\" (Proverbio africano)", "msg_date": "Mon, 26 Aug 2002 16:49:07 -0400 (CLT)", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ? " }, { "msg_contents": "\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://candle.pha.pa.us/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nAlvaro Herrera wrote:\n> Tom Lane dijo: \n> \n> > Alvaro Herrera <alvherre@atentus.com> writes:\n> > > I'm thinking about the ONLY part in the grammar in ALTER TABLE... DROP\n> > > COLUMN and RENAME COLUMN. I think they should not be there:\n> > \n> > Local DROP COLUMN is fine: it just causes the column to become\n> > non-inherited in any children. (Your patch for attisinherited will\n> > need to cover this case.)\n> \n> Oh, I see.\n> \n> > Local RENAME COLUMN does need to be prohibited, as does local ADD\n> > COLUMN, as does local ALTER COLUMN if we ever allow changing column\n> > type. Basically we need to prohibit the column from becoming\n> > incompatible with its children.\n> \n> > I don't agree with the notion of changing the grammar to achieve that,\n> > btw. Simpler and more friendly to add a specific error check in\n> > (most likely place) utility/tcop.c. Then if you try to say ONLY you'll\n> > get a more useful complaint than \"parse error\".\n> \n> Uh, I added checks in the command itself (command/tablecmds.c), just\n> because I had already done so and to not make tcop/utility.c messier\n> than it already is; I can probably move the check if people thinks it's\n> better. Also implemented is the change from inherited to non-inherited\n> when local-dropping a column.\n> \n> I also changed the text of some error messages from \"renameatt: cannot\n> foo\" to \"ALTER TABLE: cannot foo\". But my choose in wording of new\n> error messages probably needs improvement (suggestions welcome).\n> \n> Please review; I have not received comments on whether this\n> implementation is a good approach: note the signature change of\n> TupleDescInitEntry().\n> \n> -- \n> Alvaro Herrera (<alvherre[a]atentus.com>)\n> \"La espina, desde que nace, ya pincha\" (Proverbio africano)\n\nContent-Description: \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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 17:19:42 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" }, { "msg_contents": "Bruce Momjian dijo: \n\n> Your patch has been added to the PostgreSQL unapplied patches list at:\n> \n> \thttp://candle.pha.pa.us/cgi-bin/pgpatches\n> \n> I will try to apply it within the next 48 hours.\n\nDon't. I need to resync. Will post a new version later, hopefully\ntoday.\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"Aprender sin pensar es inutil; pensar sin aprender, peligroso\" (Confucio)\n\n", "msg_date": "Tue, 27 Aug 2002 17:25:35 -0400 (CLT)", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" }, { "msg_contents": "\nPatch withdrawn by author.\n\n---------------------------------------------------------------------------\n\nAlvaro Herrera wrote:\n> Tom Lane dijo: \n> \n> > Alvaro Herrera <alvherre@atentus.com> writes:\n> > > I'm thinking about the ONLY part in the grammar in ALTER TABLE... DROP\n> > > COLUMN and RENAME COLUMN. I think they should not be there:\n> > \n> > Local DROP COLUMN is fine: it just causes the column to become\n> > non-inherited in any children. (Your patch for attisinherited will\n> > need to cover this case.)\n> \n> Oh, I see.\n> \n> > Local RENAME COLUMN does need to be prohibited, as does local ADD\n> > COLUMN, as does local ALTER COLUMN if we ever allow changing column\n> > type. Basically we need to prohibit the column from becoming\n> > incompatible with its children.\n> \n> > I don't agree with the notion of changing the grammar to achieve that,\n> > btw. Simpler and more friendly to add a specific error check in\n> > (most likely place) utility/tcop.c. Then if you try to say ONLY you'll\n> > get a more useful complaint than \"parse error\".\n> \n> Uh, I added checks in the command itself (command/tablecmds.c), just\n> because I had already done so and to not make tcop/utility.c messier\n> than it already is; I can probably move the check if people thinks it's\n> better. Also implemented is the change from inherited to non-inherited\n> when local-dropping a column.\n> \n> I also changed the text of some error messages from \"renameatt: cannot\n> foo\" to \"ALTER TABLE: cannot foo\". But my choose in wording of new\n> error messages probably needs improvement (suggestions welcome).\n> \n> Please review; I have not received comments on whether this\n> implementation is a good approach: note the signature change of\n> TupleDescInitEntry().\n> \n> -- \n> Alvaro Herrera (<alvherre[a]atentus.com>)\n> \"La espina, desde que nace, ya pincha\" (Proverbio africano)\n\nContent-Description: \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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 17:26:56 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Your patch has been added to the PostgreSQL unapplied patches list at:\n\nI'd like to review this before it's applied ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 17:28:18 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ? " }, { "msg_contents": "En Tue, 27 Aug 2002 17:26:56 -0400 (EDT)\nBruce Momjian <pgman@candle.pha.pa.us> escribi�:\n\n> \n> Patch withdrawn by author.\n\nOk, new version. Please remember to change catversion.\n\nDescription of this patch:\n\n- Adds a new attribute in pg_attribute named attisinherited.\n- Creation of tables marks it true for attributes that are inherited\n- Addition of new attribute to existing inherited table marks the\n attribute as inherited for child tables.\n- Checked when trying to rename inherited attributes: if table has\n inheritors, only allow renaming if asked to recurse. Disallow\n renaming for child tables only.\n- Checked when trying to drop inherited attributes: if table has\n inheritors, mark attribute as non-inherited for direct inheritors.\n Disallow dropping for child tables only.\n\nAs an added bonus\n- Check inheritance when adding new attributes (if table has inheritors,\n only allow new attribute if it's inherited also).\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"Acepta los honores y aplausos y perderas tu libertad\"\n", "msg_date": "Wed, 28 Aug 2002 03:59:48 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" }, { "msg_contents": "En Wed, 28 Aug 2002 03:59:48 -0400\nAlvaro Herrera <alvherre@atentus.com> escribi�:\n\n> En Tue, 27 Aug 2002 17:26:56 -0400 (EDT)\n> Bruce Momjian <pgman@candle.pha.pa.us> escribi�:\n> \n> > \n> > Patch withdrawn by author.\n> \n> Ok, new version. Please remember to change catversion.\n\nDoh, patch really attached this time.\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"Cuando miro a alguien, mas me atrae como cambia que quien es\" (J. Binoche)", "msg_date": "Wed, 28 Aug 2002 04:41:27 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" }, { "msg_contents": "\nHere is a tip. After sending too many emails with no attachment, I\ndecided that as soon as I mention an attachment, I attach it, so I don't\nforget to do it at the end.\n\nI still send emails lacking attachments, but it happens less frequently.\n\n---------------------------------------------------------------------------\n\nAlvaro Herrera wrote:\n> En Wed, 28 Aug 2002 03:59:48 -0400\n> Alvaro Herrera <alvherre@atentus.com> escribi?:\n> \n> > En Tue, 27 Aug 2002 17:26:56 -0400 (EDT)\n> > Bruce Momjian <pgman@candle.pha.pa.us> escribi?:\n> > \n> > > \n> > > Patch withdrawn by author.\n> > \n> > Ok, new version. Please remember to change catversion.\n> \n> Doh, patch really attached this time.\n> \n> -- \n> Alvaro Herrera (<alvherre[a]atentus.com>)\n> \"Cuando miro a alguien, mas me atrae como cambia que quien es\" (J. Binoche)\n\n[ Attachment, skipping... ]\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 28 Aug 2002 10:50:59 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" }, { "msg_contents": "\n[ Tom Lane will be reviewing this patch.]\n\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://candle.pha.pa.us/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nAlvaro Herrera wrote:\n> En Tue, 27 Aug 2002 17:26:56 -0400 (EDT)\n> Bruce Momjian <pgman@candle.pha.pa.us> escribi?:\n> \n> > \n> > Patch withdrawn by author.\n> \n> Ok, new version. Please remember to change catversion.\n> \n> Description of this patch:\n> \n> - Adds a new attribute in pg_attribute named attisinherited.\n> - Creation of tables marks it true for attributes that are inherited\n> - Addition of new attribute to existing inherited table marks the\n> attribute as inherited for child tables.\n> - Checked when trying to rename inherited attributes: if table has\n> inheritors, only allow renaming if asked to recurse. Disallow\n> renaming for child tables only.\n> - Checked when trying to drop inherited attributes: if table has\n> inheritors, mark attribute as non-inherited for direct inheritors.\n> Disallow dropping for child tables only.\n> \n> As an added bonus\n> - Check inheritance when adding new attributes (if table has inheritors,\n> only allow new attribute if it's inherited also).\n> \n> -- \n> Alvaro Herrera (<alvherre[a]atentus.com>)\n> \"Acepta los honores y aplausos y perderas tu libertad\"\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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 28 Aug 2002 17:13:01 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" }, { "msg_contents": "Alvaro Herrera <alvherre@atentus.com> writes:\n> - Adds a new attribute in pg_attribute named attisinherited.\n> - Creation of tables marks it true for attributes that are inherited\n> - Addition of new attribute to existing inherited table marks the\n> attribute as inherited for child tables.\n> - Checked when trying to rename inherited attributes: if table has\n> inheritors, only allow renaming if asked to recurse. Disallow\n> renaming for child tables only.\n> - Checked when trying to drop inherited attributes: if table has\n> inheritors, mark attribute as non-inherited for direct inheritors.\n> Disallow dropping for child tables only.\n\nI've applied this patch after a little editorializing. FYI ---\n\n* copyfuncs.c,equalfuncs.c,outfuncs.c,readfuncs.c needed to be updated\n for the field added to ColumnDef. In general, any time you alter the\n definition of a Node structure, you gotta update these files.\n\n* I didn't like having to touch all the callers of TupleDescInitEntry,\n so I just made it initialize attisinherited to false. In the one\n place where attisinherited might be set true, just update after return\n from TupleDescInitEntry.\n\n* Moved the checks for rename/drop ONLY with child tables into\n tablecmds.c instead of utility.c, so that they'd be applied after\n grabbing an exclusive lock on the table, not before. Otherwise a\n child could be added after you look.\n\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 30 Aug 2002 15:29:45 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ? " }, { "msg_contents": "En Fri, 30 Aug 2002 15:29:45 -0400\nTom Lane <tgl@sss.pgh.pa.us> escribi�:\n\n> Alvaro Herrera <alvherre@atentus.com> writes:\n> > - Adds a new attribute in pg_attribute named attisinherited.\n> \n> I've applied this patch after a little editorializing. FYI ---\n> \n> * copyfuncs.c,equalfuncs.c,outfuncs.c,readfuncs.c needed to be updated\n> for the field added to ColumnDef. In general, any time you alter the\n> definition of a Node structure, you gotta update these files.\n\nOk, will make a note on that.\n\n> * I didn't like having to touch all the callers of TupleDescInitEntry,\n> so I just made it initialize attisinherited to false. In the one\n> place where attisinherited might be set true, just update after return\n> from TupleDescInitEntry.\n\nYes, I had thought of doing that. It's much simpler and cleaner.\n\n\n> * Moved the checks for rename/drop ONLY with child tables into\n> tablecmds.c instead of utility.c, so that they'd be applied after\n> grabbing an exclusive lock on the table, not before. Otherwise a\n> child could be added after you look.\n\nHuh, that's where I had put them in the first place. I moved them to\ntcop without thinking about the locking issues. I'll be more careful on\nthis also.\n\nThank you,\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\nVoy a acabar con todos los humanos / con los humanos yo acabar�\nvoy a acabar con todos / con todos los humanos acabar� (Bender)\n", "msg_date": "Sat, 31 Aug 2002 01:52:59 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] pg_attribute.attisinherited ?" } ]
[ { "msg_contents": "\nDue to recent security vulnerabilities reported on BugTraq, concerning\nseveral buffer overruns found in PostgreSQL, the PostgreSQL Global\nDevelopment Team today released v7.2.2 of PostgreSQL that fixes these\nvulnerabilities.\n\nThe following buffer overruns have been identified and addressed:\n\n\t\t... in handling long datetime input\n\t\t... in repeat()\n\t\t... in lpad() and rpad() with multibyte\n\t\t... in SET TIME ZONE and TZ env var\n\nAlthough v7.2.2 is a purely plug-n-play upgrade from v7.2.1, requiring no\ndump-n-reload of the database, it should be noted that these\nvulnerabilities are only critical on \"open\" or \"shared\" systems, as they\nrequire the ability to be able to connect to the database before they can\nbe exploited.\n\nThe latest release is available at:\n\n\tftp://ftp.postgresql.org/pub/sources/v7.2.2\n\nAs well as at appropriate mirror sites.\n\nPlease report any bugs/problems with this release to:\n\n\t\tpgsql-bugs@postgresql.org\n\nMarc G. Fournier\nCo-ordinator\nPostgreSQL Global Development Group\n\n", "msg_date": "Sat, 24 Aug 2002 00:22:17 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "PostgreSQL 7.2.2: Security Release " }, { "msg_contents": "Marc G. Fournier wrote:\n> Although v7.2.2 is a purely plug-n-play upgrade from v7.2.1, requiring no\n> dump-n-reload of the database, it should be noted that these\n> vulnerabilities are only critical on \"open\" or \"shared\" systems, as they\n> require the ability to be able to connect to the database before they can\n> be exploited.\n\nExcellent idea you pointed this out.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Fri, 23 Aug 2002 23:36:57 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Marc G. Fournier wrote:\n> > Although v7.2.2 is a purely plug-n-play upgrade from v7.2.1, requiring no\n> > dump-n-reload of the database, it should be noted that these\n> > vulnerabilities are only critical on \"open\" or \"shared\" systems, as they\n> > require the ability to be able to connect to the database before they can\n> > be exploited.\n> \n> Excellent idea you pointed this out.\n\n... except that it's not correct. The datetime overrun does not\nrequire the ability to connect to the database.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "23 Aug 2002 23:58:02 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "On 23 Aug 2002, Neil Conway wrote:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Marc G. Fournier wrote:\n> > > Although v7.2.2 is a purely plug-n-play upgrade from v7.2.1, requiring no\n> > > dump-n-reload of the database, it should be noted that these\n> > > vulnerabilities are only critical on \"open\" or \"shared\" systems, as they\n> > > require the ability to be able to connect to the database before they can\n> > > be exploited.\n> >\n> > Excellent idea you pointed this out.\n>\n> ... except that it's not correct. The datetime overrun does not\n> require the ability to connect to the database.\n\nAck ... obviously I missed something, but, if you can't get a connection\nto the database, how exactly is this one triggered? :(\n\n\n", "msg_date": "Sat, 24 Aug 2002 01:02:36 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "\"Marc G. Fournier\" <scrappy@hub.org> writes:\n\n> On 23 Aug 2002, Neil Conway wrote:\n> > The datetime overrun does not require the ability to connect to\n> > the database.\n> \n> Ack ... obviously I missed something, but, if you can't get a\n> connection to the database, how exactly is this one triggered? :(\n\nIf the application is accepting datetime input from the user ('what's\nyour birthday?', for example), and isn't doing some non-obvious input\nvalidation on it (namely, checking that the input string isn't too\nlong), you can crash the backend. Gavin says executing arbitrary code\nusing the hole would be extremely difficult, but it's at least\nconceivable.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "24 Aug 2002 00:11:03 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "On 24 Aug 2002, Neil Conway wrote:\n\n> \"Marc G. Fournier\" <scrappy@hub.org> writes:\n>\n> > On 23 Aug 2002, Neil Conway wrote:\n> > > The datetime overrun does not require the ability to connect to\n> > > the database.\n> >\n> > Ack ... obviously I missed something, but, if you can't get a\n> > connection to the database, how exactly is this one triggered? :(\n>\n> If the application is accepting datetime input from the user ('what's\n> your birthday?', for example), and isn't doing some non-obvious input\n> validation on it (namely, checking that the input string isn't too\n> long), you can crash the backend. Gavin says executing arbitrary code\n> using the hole would be extremely difficult, but it's at least\n> conceivable.\n\nRight, but you have to get a connection to the backend in order to crash\nit ... no?\n\n\n", "msg_date": "Sat, 24 Aug 2002 01:13:11 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "\"Marc G. Fournier\" <scrappy@hub.org> writes:\n> On 24 Aug 2002, Neil Conway wrote:\n> > If the application is accepting datetime input from the user ('what's\n> > your birthday?', for example), and isn't doing some non-obvious input\n> > validation on it (namely, checking that the input string isn't too\n> > long), you can crash the backend. Gavin says executing arbitrary code\n> > using the hole would be extremely difficult, but it's at least\n> > conceivable.\n> \n> Right, but you have to get a connection to the backend in order to crash\n> it ... no?\n\nYou need to be using an application accepts datetime input from the\nuser, and at some point inserts it into the database. For example, if\nyou wrote a webapp that accepted datetime input of some kind (to use\nmy previous example, the user's birthday), any user of the webapp\ncould enter bogus data that would crash the backend.\n\nIn this case, the user does not make a connection to the backend (the\nweb app does), and does not have the ability to execute arbitrary SQL\n(i.e. it's not a \"shared\" or \"open\" system) -- but a security problem\nstill exists.\n\nThis is in contrast to the other security holes (repeat(), lpad(),\nrpad(), SET TIME ZONE, and TZ env var), in which the probability of\nsomeone without SQL access being able to exercise the bug is\nnegligible.\n\nCheers,\n\nNeil\n\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "24 Aug 2002 00:23:13 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "\"Marc G. Fournier\" <scrappy@hub.org> writes:\n> Right, but you have to get a connection to the backend in order to crash\n> it ... no?\n\nThe point was that it might be possible to exploit this with only\nindirect access to the database, such as entering \"date\" information\ninto a webform that would hand off the value to the database with\nlittle or no checking. Most of the risks we've been discussing require\nthe ability to issue chosen SQL commands, but this one only requires\nthe ability to determine a data value that's used in a SQL command.\nBig difference.\n\n\t\t\tregards, tom lane\n", "msg_date": "Sat, 24 Aug 2002 00:37:44 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release " }, { "msg_contents": "\nThe issue is data-provoked crashes vs. query-invoked crashes. Marc's\npoint, and I think it was clear enough, is that you can't just poke at\nthe TCP port and hope to do anything bad, which was the thrust of the\nargument, I think.\n\n---------------------------------------------------------------------------\n\nNeil Conway wrote:\n> \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > On 24 Aug 2002, Neil Conway wrote:\n> > > If the application is accepting datetime input from the user ('what's\n> > > your birthday?', for example), and isn't doing some non-obvious input\n> > > validation on it (namely, checking that the input string isn't too\n> > > long), you can crash the backend. Gavin says executing arbitrary code\n> > > using the hole would be extremely difficult, but it's at least\n> > > conceivable.\n> > \n> > Right, but you have to get a connection to the backend in order to crash\n> > it ... no?\n> \n> You need to be using an application accepts datetime input from the\n> user, and at some point inserts it into the database. For example, if\n> you wrote a webapp that accepted datetime input of some kind (to use\n> my previous example, the user's birthday), any user of the webapp\n> could enter bogus data that would crash the backend.\n> \n> In this case, the user does not make a connection to the backend (the\n> web app does), and does not have the ability to execute arbitrary SQL\n> (i.e. it's not a \"shared\" or \"open\" system) -- but a security problem\n> still exists.\n> \n> This is in contrast to the other security holes (repeat(), lpad(),\n> rpad(), SET TIME ZONE, and TZ env var), in which the probability of\n> someone without SQL access being able to exercise the bug is\n> negligible.\n> \n> Cheers,\n> \n> Neil\n> \n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sat, 24 Aug 2002 00:38:07 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "On Saturday 24 August 2002 12:37 am, Tom Lane wrote:\n> \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > Right, but you have to get a connection to the backend in order to crash\n> > it ... no?\n\n> The point was that it might be possible to exploit this with only\n> indirect access to the database, such as entering \"date\" information\n\nAnd this is the reason I was in favor of 7.2.2.\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Sat, 24 Aug 2002 00:44:55 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> The issue is data-provoked crashes vs. query-invoked crashes. Marc's\n> point, and I think it was clear enough, is that you can't just poke at\n> the TCP port and hope to do anything bad, which was the thrust of the\n> argument, I think.\n\nThe point I objected to is the suggestion that only those running\n\"shared\" or \"open\" systems are vulnerable to the security\nproblem. That is simply incorrect.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "24 Aug 2002 00:47:16 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "Neil Conway wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > The issue is data-provoked crashes vs. query-invoked crashes. Marc's\n> > point, and I think it was clear enough, is that you can't just poke at\n> > the TCP port and hope to do anything bad, which was the thrust of the\n> > argument, I think.\n> \n> The point I objected to is the suggestion that only those running\n> \"shared\" or \"open\" systems are vulnerable to the security\n> problem. That is simply incorrect.\n\nYes, I remember now. It is a bad data vunerability vs. a bad query\nvulnerability.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sat, 24 Aug 2002 07:23:48 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "On Sat, 24 Aug 2002, Marc G. Fournier wrote:\n\n> On 24 Aug 2002, Neil Conway wrote:\n>\n> > \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> >\n> > > On 23 Aug 2002, Neil Conway wrote:\n> > > > The datetime overrun does not require the ability to connect to\n> > > > the database.\n> > >\n> > > Ack ... obviously I missed something, but, if you can't get a\n> > > connection to the database, how exactly is this one triggered? :(\n> >\n> > If the application is accepting datetime input from the user ('what's\n> > your birthday?', for example), and isn't doing some non-obvious input\n> > validation on it (namely, checking that the input string isn't too\n> > long), you can crash the backend. Gavin says executing arbitrary code\n> > using the hole would be extremely difficult, but it's at least\n> > conceivable.\n>\n> Right, but you have to get a connection to the backend in order to crash\n> it ... no?\n\nAnd what are the odds your application is going to bomb due to a buffer\noverflow before it even gets to the database. I can see maybe with php,\nbut a web form should always be length limited.\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n", "msg_date": "Sat, 24 Aug 2002 10:51:30 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "On Saturday 24 August 2002 10:51 am, Vince Vielhaber wrote:\n> And what are the odds your application is going to bomb due to a buffer\n> overflow before it even gets to the database. I can see maybe with php,\n> but a web form should always be length limited.\n\nIIRC, web forms are limited to 32k in an input field of any kind. TEXTAREA \nis, I believe, the limit.\n\nThere are other limits -- but a web browser may not be the source of the POST \nformdata.\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Sat, 24 Aug 2002 13:53:22 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "On Sat, 24 Aug 2002, Bruce Momjian wrote:\n\n>\n> The issue is data-provoked crashes vs. query-invoked crashes. Marc's\n> point, and I think it was clear enough, is that you can't just poke at\n> the TCP port and hope to do anything bad, which was the thrust of the\n> argument, I think.\n\nCorrect\n\n\n", "msg_date": "Sat, 24 Aug 2002 16:29:50 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "On Sat, 24 Aug 2002, Bruce Momjian wrote:\n\n> \n> The issue is data-provoked crashes vs. query-invoked crashes. Marc's\n> point, and I think it was clear enough, is that you can't just poke at\n> the TCP port and hope to do anything bad, which was the thrust of the\n> argument, I think.\n\nBruce,\n\nI am convinced that someone with enough time on their hands and some code\npointed to by Florian Weimer could exploit the datetime overrun issue by\ncrafting a datetime string in such a way as to overrun the buffer and\nsmash the stack.\n\nIn applications which pass date/time data directly to the database without\nany validation (is this datetime string greater than 52 bytes? does it\nlook like a date/time string?) then a malicious user without direct\ndatabase access could crash the database by taking advantage of the short\ncomings in Postgres and the application.\n\nAs such, I would recommend all people who offer direct access to the\ndatabase and/or have applications which user date/time data\ntypes/functionality to upgrade to 7.2.2.\n\nGavin\n\n\n", "msg_date": "Sun, 25 Aug 2002 23:44:03 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "\nOK, I understand your point. What do we need to do now that the\nannouncement has already been made?\n\n---------------------------------------------------------------------------\n\nGavin Sherry wrote:\n> On Sat, 24 Aug 2002, Bruce Momjian wrote:\n> \n> > \n> > The issue is data-provoked crashes vs. query-invoked crashes. Marc's\n> > point, and I think it was clear enough, is that you can't just poke at\n> > the TCP port and hope to do anything bad, which was the thrust of the\n> > argument, I think.\n> \n> Bruce,\n> \n> I am convinced that someone with enough time on their hands and some code\n> pointed to by Florian Weimer could exploit the datetime overrun issue by\n> crafting a datetime string in such a way as to overrun the buffer and\n> smash the stack.\n> \n> In applications which pass date/time data directly to the database without\n> any validation (is this datetime string greater than 52 bytes? does it\n> look like a date/time string?) then a malicious user without direct\n> database access could crash the database by taking advantage of the short\n> comings in Postgres and the application.\n> \n> As such, I would recommend all people who offer direct access to the\n> database and/or have applications which user date/time data\n> types/functionality to upgrade to 7.2.2.\n> \n> Gavin\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> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sun, 25 Aug 2002 10:34:29 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "On Sun, 25 Aug 2002, Bruce Momjian wrote:\n\n>\n> OK, I understand your point. What do we need to do now that the\n> announcement has already been made?\n\nI'm still slightly confused here ... from what Neil/Gavin have stated so\nfar, all it sounds like is that if I pass a wrong date/time string, it\nwill crash the backend ... or is this what I'm missing?\n\n\n >\n> ---------------------------------------------------------------------------\n>\n> Gavin Sherry wrote:\n> > On Sat, 24 Aug 2002, Bruce Momjian wrote:\n> >\n> > >\n> > > The issue is data-provoked crashes vs. query-invoked crashes. Marc's\n> > > point, and I think it was clear enough, is that you can't just poke at\n> > > the TCP port and hope to do anything bad, which was the thrust of the\n> > > argument, I think.\n> >\n> > Bruce,\n> >\n> > I am convinced that someone with enough time on their hands and some code\n> > pointed to by Florian Weimer could exploit the datetime overrun issue by\n> > crafting a datetime string in such a way as to overrun the buffer and\n> > smash the stack.\n> >\n> > In applications which pass date/time data directly to the database without\n> > any validation (is this datetime string greater than 52 bytes? does it\n> > look like a date/time string?) then a malicious user without direct\n> > database access could crash the database by taking advantage of the short\n> > comings in Postgres and the application.\n> >\n> > As such, I would recommend all people who offer direct access to the\n> > database and/or have applications which user date/time data\n> > types/functionality to upgrade to 7.2.2.\n> >\n> > Gavin\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> >\n>\n> --\n> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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", "msg_date": "Sun, 25 Aug 2002 23:16:41 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "*sigh* Someone's marked postgres 7.2.1 as forbidden in FreeBSD ports:\n\nFORBIDDEN= \"buffer overruns acknowledged by authors--see\n<URL:http://www3.us.postgresql.org/news.html>\"\n\nSomewhat of an overreaction...I'm hassling the maintainer at the moment...\n\nChris\n\n> -----Original Message-----\n> From: pgsql-hackers-owner@postgresql.org\n> [mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Marc G. Fournier\n> Sent: Monday, 26 August 2002 10:17 AM\n> To: Bruce Momjian\n> Cc: Gavin Sherry; Neil Conway; PostgreSQL Hackers\n> Subject: Re: [HACKERS] [GENERAL] PostgreSQL 7.2.2: Security Release\n>\n>\n> On Sun, 25 Aug 2002, Bruce Momjian wrote:\n>\n> >\n> > OK, I understand your point. What do we need to do now that the\n> > announcement has already been made?\n>\n> I'm still slightly confused here ... from what Neil/Gavin have stated so\n> far, all it sounds like is that if I pass a wrong date/time string, it\n> will crash the backend ... or is this what I'm missing?\n>\n>\n> >\n> >\n> ------------------------------------------------------------------\n> ---------\n> >\n> > Gavin Sherry wrote:\n> > > On Sat, 24 Aug 2002, Bruce Momjian wrote:\n> > >\n> > > >\n> > > > The issue is data-provoked crashes vs. query-invoked\n> crashes. Marc's\n> > > > point, and I think it was clear enough, is that you can't\n> just poke at\n> > > > the TCP port and hope to do anything bad, which was the\n> thrust of the\n> > > > argument, I think.\n> > >\n> > > Bruce,\n> > >\n> > > I am convinced that someone with enough time on their hands\n> and some code\n> > > pointed to by Florian Weimer could exploit the datetime\n> overrun issue by\n> > > crafting a datetime string in such a way as to overrun the buffer and\n> > > smash the stack.\n> > >\n> > > In applications which pass date/time data directly to the\n> database without\n> > > any validation (is this datetime string greater than 52 bytes? does it\n> > > look like a date/time string?) then a malicious user without direct\n> > > database access could crash the database by taking advantage\n> of the short\n> > > comings in Postgres and the application.\n> > >\n> > > As such, I would recommend all people who offer direct access to the\n> > > database and/or have applications which user date/time data\n> > > types/functionality to upgrade to 7.2.2.\n> > >\n> > > Gavin\n> > >\n> > >\n> > >\n> > > ---------------------------(end of\n> 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) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square,\n> Pennsylvania 19073\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>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n>\n\n", "msg_date": "Mon, 26 Aug 2002 13:06:48 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "On Mon, 26 Aug 2002, Christopher Kings-Lynne wrote:\n\n> *sigh* Someone's marked postgres 7.2.1 as forbidden in FreeBSD ports:\n>\n> FORBIDDEN= \"buffer overruns acknowledged by authors--see\n> <URL:http://www3.us.postgresql.org/news.html>\"\n>\n> Somewhat of an overreaction...I'm hassling the maintainer at the moment...\n\nalready fixed ...\n\n> > -----Original Message-----\n> > From: pgsql-hackers-owner@postgresql.org\n> > [mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Marc G. Fournier\n> > Sent: Monday, 26 August 2002 10:17 AM\n> > To: Bruce Momjian\n> > Cc: Gavin Sherry; Neil Conway; PostgreSQL Hackers\n> > Subject: Re: [HACKERS] [GENERAL] PostgreSQL 7.2.2: Security Release\n> >\n> >\n> > On Sun, 25 Aug 2002, Bruce Momjian wrote:\n> >\n> > >\n> > > OK, I understand your point. What do we need to do now that the\n> > > announcement has already been made?\n> >\n> > I'm still slightly confused here ... from what Neil/Gavin have stated so\n> > far, all it sounds like is that if I pass a wrong date/time string, it\n> > will crash the backend ... or is this what I'm missing?\n> >\n> >\n> > >\n> > >\n> > ------------------------------------------------------------------\n> > ---------\n> > >\n> > > Gavin Sherry wrote:\n> > > > On Sat, 24 Aug 2002, Bruce Momjian wrote:\n> > > >\n> > > > >\n> > > > > The issue is data-provoked crashes vs. query-invoked\n> > crashes. Marc's\n> > > > > point, and I think it was clear enough, is that you can't\n> > just poke at\n> > > > > the TCP port and hope to do anything bad, which was the\n> > thrust of the\n> > > > > argument, I think.\n> > > >\n> > > > Bruce,\n> > > >\n> > > > I am convinced that someone with enough time on their hands\n> > and some code\n> > > > pointed to by Florian Weimer could exploit the datetime\n> > overrun issue by\n> > > > crafting a datetime string in such a way as to overrun the buffer and\n> > > > smash the stack.\n> > > >\n> > > > In applications which pass date/time data directly to the\n> > database without\n> > > > any validation (is this datetime string greater than 52 bytes? does it\n> > > > look like a date/time string?) then a malicious user without direct\n> > > > database access could crash the database by taking advantage\n> > of the short\n> > > > comings in Postgres and the application.\n> > > >\n> > > > As such, I would recommend all people who offer direct access to the\n> > > > database and/or have applications which user date/time data\n> > > > types/functionality to upgrade to 7.2.2.\n> > > >\n> > > > Gavin\n> > > >\n> > > >\n> > > >\n> > > > ---------------------------(end of\n> > 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) 359-1001\n> > > + If your life is a hard drive, | 13 Roberts Road\n> > > + Christ can be your backup. | Newtown Square,\n> > Pennsylvania 19073\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> >\n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 4: Don't 'kill -9' the postmaster\n> >\n>\n>\n\n", "msg_date": "Mon, 26 Aug 2002 10:43:53 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": true, "msg_subject": "Re: [GENERAL] PostgreSQL 7.2.2: Security Release" }, { "msg_contents": "In article <20020824000845.Q1769-100000@mail1.hub.org>,\n scrappy@hub.org (\"Marc G. Fournier\") wrote:\n\n> Due to recent security vulnerabilities reported on BugTraq, concerning\n> several buffer overruns found in PostgreSQL, the PostgreSQL Global\n> Development Team today released v7.2.2 of PostgreSQL that fixes these\n> vulnerabilities.\n\nmight i recommend posting this to announce, as well? i never saw it show \nup there, and i certainly know some people who only read the announce \nlist.\n\nif it was there, then the subsidiary problem is that announce is not \nworking on the news server...\n\nFYI.\n\n-tfo\n", "msg_date": "Mon, 26 Aug 2002 13:52:28 -0500", "msg_from": "Thomas O'Connell <tfo@monsterlabs.com>", "msg_from_op": false, "msg_subject": "Re: PostgreSQL 7.2.2: Security Release" } ]
[ { "msg_contents": "I'm out of town and my car broke down. I probably won't be back at my\nmachine until Tuesday. Don't expect too many replies from me until then. Sorry\nthat I can't fix the large file stuff right away.\n\nBtw., the 7.2.2 release contains the wrong documentation, as usual.\n\n-- \nGMX - Die Kommunikationsplattform im Internet.\nhttp://www.gmx.net\n\n", "msg_date": "Sat, 24 Aug 2002 09:22:38 +0200 (MEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": true, "msg_subject": "Delayed..." }, { "msg_contents": "Peter Eisentraut wrote:\n> I'm out of town and my car broke down. I probably won't be back at my\n> machine until Tuesday. Don't expect too many replies from me until then. Sorry\n> that I can't fix the large file stuff right away.\n\nSorry to hear that.\n\n> Btw., the 7.2.2 release contains the wrong documentation, as usual.\n\nOh. I wonder if that is because we didn't rebuild it.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sat, 24 Aug 2002 07:27:56 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: Delayed..." } ]
[ { "msg_contents": "I want to make a function that converts text to cube and the \"obvious\"\nthing to do is to call cube_in with the pointer from the text data.\nHowever this depends on the text string being 0 terminated. I am very\nsure that this is the case, but since I want to submit this change back\nas a patch to contrib/cube I was hoping someone could make me 100%\nsure this is correct. Thanks.\n", "msg_date": "Sat, 24 Aug 2002 10:34:22 -0500", "msg_from": "Bruno Wolff III <bruno@wolff.to>", "msg_from_op": true, "msg_subject": "Are \"text\" strings 0 terminated internally?" }, { "msg_contents": "No, they aren't.\n\nYou should call textout if you want to convert a TEXT object into a C\nstring.\n\n\t\t\tregards, tom lane\n", "msg_date": "Sun, 25 Aug 2002 10:57:12 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Are \"text\" strings 0 terminated internally? " } ]
[ { "msg_contents": "\n\nWould someone apply the attached patch to the development source and let me\nknow if the autoconf step fails or works. I've only got autoconf 2.13 available\nand the file needs 2.53 apparently. If it works could I also have a copy of the\nresulting configure script, or patch, please.\n\nFor the record, this is related to reserving the last few backend slots for the\nsuperuser and I just need to test what I've done.\n\n\nTIA\n\n\n-- \nNigel J. Andrews\nDirector\n\n---\nLogictree Systems Limited\nComputer Consultants\n\n", "msg_date": "Sun, 25 Aug 2002 14:36:19 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": true, "msg_subject": "A configure.in patch check" } ]
[ { "msg_contents": "Helps if I attach the patch...\n\n\n---------- Forwarded message ----------\nDate: Sun, 25 Aug 2002 14:36:19 +0100 (BST)\nFrom: Nigel J. Andrews <nandrews@investsystems.co.uk>\nTo: pgsql-hackers@postgresql.org\nSubject: A configure.in patch check\n\n\n\nWould someone apply the attached patch to the development source and let me\nknow if the autoconf step fails or works. I've only got autoconf 2.13 available\nand the file needs 2.53 apparently. If it works could I also have a copy of the\nresulting configure script, or patch, please.\n\nFor the record, this is related to reserving the last few backend slots for the\nsuperuser and I just need to test what I've done.\n\n\nTIA\n\n\n-- \nNigel J. Andrews\nDirector\n\n---\nLogictree Systems Limited\nComputer Consultants", "msg_date": "Sun, 25 Aug 2002 14:37:31 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": true, "msg_subject": "A configure.in patch check (fwd)" }, { "msg_contents": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> + AC_MSG_CHECKING([for default superuser reserved number of connections])\n> + PGAC_ARG_REQ(with, reservedbackends, [ --with-reservedbackends=N set default superuser reserved number of connections [2]],\n> + [],\n> + [with_reservedbackends=2])\n\nThis will be rejected anyway; what you want is to set up\nreserved_backends as a GUC parameter, not as something that has to be\nhard-wired at configure time. I can't see any reason to make it\nhard-wired...\n\n\t\t\tregards, tom lane\n", "msg_date": "Sun, 25 Aug 2002 10:55:02 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: A configure.in patch check (fwd) " }, { "msg_contents": "On Sun, 25 Aug 2002, Tom Lane wrote:\n\n> \"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> > + AC_MSG_CHECKING([for default superuser reserved number of connections])\n> > + PGAC_ARG_REQ(with, reservedbackends, [ --with-reservedbackends=N set default superuser reserved number of connections [2]],\n> > + [],\n> > + [with_reservedbackends=2])\n> \n> This will be rejected anyway; what you want is to set up\n> reserved_backends as a GUC parameter, not as something that has to be\n> hard-wired at configure time. I can't see any reason to make it\n> hard-wired...\n\n\nIt is a GUC. It's exactly like max_backends. I took the easy route out and\njust followed where DEF_MAXBACKENDS was being set rather than hard wiring\nthe value any where.\n\nRather distressingly in order to get this new value into where it's needed\nI had to hit quite a few files, more than I would have expected. Again I\njust followed how MaxBackends was being sent to where it was needed but is\nthere any particular reason why storage/ipc/sinvaladt.c:SIBackendInit()\ncan't access MaxBackends and my new ReservedBackends directly? The are\nglobal variables afterall, I think #include \"miscadmin.h\" would need to be\nadded but is that bad?\n\n\n-- \nNigel J. Andrews\nDirector\n\n---\nLogictree Systems Limited\nComputer Consultants\n\n", "msg_date": "Sun, 25 Aug 2002 17:09:04 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": true, "msg_subject": "Re: A configure.in patch check (fwd)" }, { "msg_contents": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> It is a GUC. It's exactly like max_backends. I took the easy route out and\n> just followed where DEF_MAXBACKENDS was being set rather than hard wiring\n> the value any where.\n\nOh. Well, skip the configure part: the only reason there's still a\nconfigure parameter for maxbackends is backwards compatibility with\nancient configure scripts (from days when it was in fact frozen at\nconfigure time). I don't see a need to provide one for reserved_slots.\n\n> Rather distressingly in order to get this new value into where it's needed\n> I had to hit quite a few files, more than I would have expected. Again I\n> just followed how MaxBackends was being sent to where it was needed but is\n> there any particular reason why storage/ipc/sinvaladt.c:SIBackendInit()\n> can't access MaxBackends and my new ReservedBackends directly?\n\nProbably not. Again, the way that MaxBackends is handled is largely\nlegacy code. You'd have been better off looking at almost any other\nGUC parameter as a template ;-)\n\n\t\t\tregards, tom lane\n", "msg_date": "Sun, 25 Aug 2002 12:43:11 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: A configure.in patch check (fwd) " } ]
[ { "msg_contents": "Hi all,\n\nI've just been thinking that the documentation doesn't cover the maximum\ninput string lengths for various data types well. Case in point, the\ndate/time code: there was a fair amount of discussion about validation\ninput, including checking for 'unreasonable' string lengths. Unless you\nwent and poked around the code, you couldn't know that the maximum string\nlength for a date/time string is 52 bytes.\n\nSo, the question is this: should there be documentation of the maximum\nstring length of a data structure so that application programmers can\nprovide string length validation? Perhaps MAXDATELEN and other static\nvalues in adt/ should be put in pg_config.h so that the user does not need\nto hard code them? Perhaps it is a bad idea, since programmers should also\ndo other validation, such as ensuring that a submitted date/time value\nis what the program(mer) is expecting?\n\nGavin\n\n\n", "msg_date": "Mon, 26 Aug 2002 01:16:59 +1000 (EST)", "msg_from": "Gavin Sherry <swm@linuxworld.com.au>", "msg_from_op": true, "msg_subject": "Documentation of maximum input string lengths" }, { "msg_contents": "Gavin Sherry <swm@linuxworld.com.au> writes:\n> So, the question is this: should there be documentation of the maximum\n> string length of a data structure so that application programmers can\n> provide string length validation?\n\nI don't think so; that's just going to make it harder to fix things if,\nsay, one day we need to support longer timezone names than we do today.\nThe more places that know about these limits the worse it will be.\n\nThe date buffer overrun bug was a backend bug, nothing more nor less,\nand it was *not* the frontends' responsibility to guard against.\n\n\t\t\tregards, tom lane\n", "msg_date": "Sun, 25 Aug 2002 12:55:54 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Documentation of maximum input string lengths " } ]
[ { "msg_contents": ";;; From: sugita@sra.co.jp \n;;; Date: Tue, 16 Jul 2002 10:22:14 +0900 (JST) \n;;; \n;;; ;;; However, that doesn't explain our OS X problem. I added some debug \n;;; ;;; printouts, and can now report positively that (a) the fork() call \n;;; ;;; returns normally in the parent process, providing an apparently-correct \n;;; ;;; child PID value; but (b) the fork never returns in the child. It \n;;; ;;; doesn't ever get as far as trying to enable SIGTERM. \n;;; & \n;;; ;;; Is it possible that something in the child's fork() processing will wait \n;;; ;;; around for a response from a service that's already died? Why is fork() \n;;; ;;; dependent on any outside service whatever --- isn't that a certain \n;;; ;;; recipe for system failures? \n;;; \n;;; I asked Apple this issue. This is a bug of Mac OS X. The problem is registered \n;;; to their bug database for the appropriate eingineers for investigation. \n\nThis problem is fixed for Mac OS X 10.2.\n\n\nKenji Sugita\n\n", "msg_date": "Mon, 26 Aug 2002 00:35:08 +0900 (JST)", "msg_from": "sugita@sra.co.jp", "msg_from_op": true, "msg_subject": "Re: Mac OS X: system shutdown prevents checkpoint" } ]
[ { "msg_contents": "I am going through my mailbox trying to tie up any loose ends before\nbeta. This is SOP --- standard operating procedure. ;-)\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sun, 25 Aug 2002 16:31:55 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Going through my mailbox" } ]
[ { "msg_contents": "\nI have been corresponding with Bob Devine for a few years. He was at\nBerkeley during the Postgres days and knows quite a bit about\noptimizers and storage systems. I will put his name at the bottom of\nthe optimizer README and if people have questions, he is willing to\nanswer them as best he can.\n\n---------------------------------------------------------------------------\n\nBob Devine wrote:\n> Bruce Momjian wrote:\n> > \n> > Bob, any status on when you may want to get involved with PostgreSQL?\n> \n> Umm, hem, haw...\n> \n> I'm maxed out with current stuff now. It would be very hard\n> for me to realistically promise any substantial committment\n> to doing any major work.\n> \n> However, I'm always available for discussion on designs.\n> I've built two optimizers so I can help there. Otherwise\n> I can help with general query execution or storage subsystems.\n> \n> Bob Devine\n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sun, 25 Aug 2002 18:41:07 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: Fwd: Joins and links" } ]
[ { "msg_contents": "TODO item:\n\nAdministration -\n\tReserve last few process slots for super-user if max_connections\n\treached\n\n\nNotes:\n\nAdded GUC superuser_reserved_connections such that non-superuser connections\nare only acceptable in the first\n(max_connections - superuser_reserved_connections) backend slots.\n\nSuperuser connections within these first n slots count towards this\nnon-superuser connection limit. Therefore there can be at most this number\nof non-superuser connections but may be less.\n\nIn addition, this limit is only checked on initialisation of a backend\nprocess. So reserved slots can be taken by connections that subsequently\nlose superuser priviledges thus evading the lower limit on backends.\n\nPassed regression tests, not that it was likely not to.\nBehaved as expected in a manual test.\n\n\n-- \nNigel J. Andrews\nDirector\n\n---\nLogictree Systems Limited\nComputer Consultants", "msg_date": "Sun, 25 Aug 2002 23:56:31 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": true, "msg_subject": "TODO Done. Superuser backend slot reservations" }, { "msg_contents": "[only replying to -patches, this doesn't belong on -hackers AFAICS ]\n\n\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> Added GUC superuser_reserved_connections such that non-superuser connections\n> are only acceptable in the first\n> (max_connections - superuser_reserved_connections) backend slots.\n\nI'd prefer that we not change the meaning of max_connections. I had in\nmind the following:\n\n - max_connections denotes the # of \"regular\" connections to\n the database (i.e. admin & non-admin)\n - max_admin_connections denotes an additional # of backend\n slots that are reserved for connections from superusers. Not\n sure if this should default to zero or not\n - Therefore, the # of backend slots created is\n (max_connections + max_admin_connections)\n\n> In addition, this limit is only checked on initialisation of a backend\n> process. So reserved slots can be taken by connections that subsequently\n> lose superuser priviledges thus evading the lower limit on backends.\n\nHow can that happen?\n\n+ \t/*\n+ \t * Force ReservedBackends is less than MaxBackends if need be.\n+ \t * A cluster only allowing superuser connections seems silly whereas\n+ \t * a cluster reserving none for superusers doesn't.\n+ \t */\n+ \tif (ReservedBackends >= MaxBackends)\n+ \t\tReservedBackends = MaxBackends - 1;\n\nIMHO, we should elog(FATAL) here, or at least emit a warning.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "25 Aug 2002 19:26:06 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations" }, { "msg_contents": "Neil Conway wrote:\n> [only replying to -patches, this doesn't belong on -hackers AFAICS ]\n> \n> \"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> > Added GUC superuser_reserved_connections such that non-superuser connections\n> > are only acceptable in the first\n> > (max_connections - superuser_reserved_connections) backend slots.\n> \n> I'd prefer that we not change the meaning of max_connections. I had in\n> mind the following:\n> \n> - max_connections denotes the # of \"regular\" connections to\n> the database (i.e. admin & non-admin)\n\nWell, if you do that, then max_connections is really not max\nconnections, it is maximum connections minus admin_connections. That\nseems confusing.\n\nI think I prefer just reserving 1-2 of the last slots. The kernel's\nproc code is the same; if you specify a proc table of 150, it is 150,\nwith 149 and 150 reserved for root.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sun, 25 Aug 2002 19:31:13 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations" }, { "msg_contents": "On Sun, 25 Aug 2002, Bruce Momjian wrote:\n\n> Neil Conway wrote:\n> > [only replying to -patches, this doesn't belong on -hackers AFAICS ]\n> > \n> > \"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> > > Added GUC superuser_reserved_connections such that non-superuser connections\n> > > are only acceptable in the first\n> > > (max_connections - superuser_reserved_connections) backend slots.\n> > \n> > I'd prefer that we not change the meaning of max_connections. I had in\n> > mind the following:\n> > \n> > - max_connections denotes the # of \"regular\" connections to\n> > the database (i.e. admin & non-admin)\n> \n> Well, if you do that, then max_connections is really not max\n> connections, it is maximum connections minus admin_connections. That\n> seems confusing.\n\nThat was my thinking. Plus, MaxBackends gets used in quite a few places, to\navoid having to go to extremes and look at and possibly hit each of those\nplaces to ensure things are correct seems silly. An alternative would be to\nchange the variable max_connections is tied to and make MaxBackends the\nsum of two GUCs. I couldn't see the point in doing either of those when\nmax_connections should define the maximum number of connections possible\nand not something less.\n\nHmmm...rereading that it strikes me that we're all reading from the same\npage of the same book. All three of us are saying max_connections gives\nthe maximum number of connections possible from admin and non-admin users\ntogether.\n\n> I think I prefer just reserving 1-2 of the last slots. The kernel's\n> proc code is the same; if you specify a proc table of 150, it is 150,\n> with 149 and 150 reserved for root.\n\nYes, only the exact number reserved at the top of the slot array is\nconfigurable via the GUC variable.\n\n\n-- \nNigel J. Andrews\n\n", "msg_date": "Mon, 26 Aug 2002 00:59:01 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations" }, { "msg_contents": "On 25 Aug 2002, Neil Conway wrote:\n\n> [only replying to -patches, this doesn't belong on -hackers AFAICS ]\n\nIs -patches for discussions? I thought it was only for the patches\nthemselves, I'd better go subscribe...\n\n> \n> \"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> > In addition, this limit is only checked on initialisation of a backend\n> > process. So reserved slots can be taken by connections that subsequently\n> > lose superuser priviledges thus evading the lower limit on backends.\n> \n> How can that happen?\n\nWell, the test is located somewhere that is only called once, when the backend\nprocess is forked. At least that's what I think but as I say I'm not 100%\ncertain, mostly becuase I haven't checked only looked to see what the\nroutine is doing and it looks like a one shot routine to me. Therefore,\nassuming doing a SET SESSION AUTH... doesn't drop and then reconnect to the\nserver, a change from a superuser to a normaluser is not going to result in a\ndropped connection. Nor should it do I believe.\n\n> \n> + \t/*\n> + \t * Force ReservedBackends is less than MaxBackends if need be.\n> + \t * A cluster only allowing superuser connections seems silly whereas\n> + \t * a cluster reserving none for superusers doesn't.\n> + \t */\n> + \tif (ReservedBackends >= MaxBackends)\n> + \t\tReservedBackends = MaxBackends - 1;\n> \n> IMHO, we should elog(FATAL) here, or at least emit a warning.\n\nThe warning sounds reasonable to me. I'll add one and resubmit in a day or two\nafter I've seen what else gets said.\n\n\n-- \nNigel J. Andrews\n\n\n", "msg_date": "Mon, 26 Aug 2002 01:15:10 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations" }, { "msg_contents": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> \tReserve last few process slots for super-user if max_connections\n> \treached\n\nAnother minor coding suggestion:\n\n+ \t/*\n+ \t * Check a normal user hasn't connected to a superuser reserved slot.\n+ \t * Do this here since we need the user information and that only happens\n+ \t * after we've started bringing the shared memory online. So we wait\n+ \t * until we've registered exit handlers and potentially shut an open\n+ \t * transaction down for an as safety conscious rejection as possible.\n+ \t */\n+ \tif (!superuser() && MyBackendId > MaxBackends - ReservedBackends)\n+ \t\telog(ERROR, \"Normal user limit exceeded\");\n\nThis would be more efficient if you placed the superuser() test after\nthe BackendId test, as it is both more expensive to evaluate and more\nlikely to be true.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "25 Aug 2002 21:41:34 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> - Therefore, the # of backend slots created is\n> (max_connections + max_admin_connections)\n\nI tend to agree with Bruce on this: max_connections means\nmax_connections. Therefore, the number of backend slots is\nmax_connections, of which max_connections - max_admin_connections\nare available to non-superusers.\n\n(There is provision in the existing code for one extra child process\nfor checkpoints, but it's not a \"real\" backend and so it's reasonable\nnot to count it against max_connections.)\n\n\t\t\tregards, tom lane\n", "msg_date": "Sun, 25 Aug 2002 23:56:04 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations " }, { "msg_contents": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> + \tif (!superuser() && MyBackendId > MaxBackends - ReservedBackends)\n> + \t\telog(ERROR, \"Normal user limit exceeded\");\n\nThis coding is wrong on its face: the slot number you happen to find has\nno relationship to the number of slots remaining free, except as an\nexistence proof that the number of slots free was > 0 before you took\none.\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 26 Aug 2002 00:04:24 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations " }, { "msg_contents": "\nOn Mon, 26 Aug 2002, Tom Lane wrote:\n\n> \"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> > + \tif (!superuser() && MyBackendId > MaxBackends - ReservedBackends)\n> > + \t\telog(ERROR, \"Normal user limit exceeded\");\n> \n> This coding is wrong on its face: the slot number you happen to find has\n> no relationship to the number of slots remaining free, except as an\n> existence proof that the number of slots free was > 0 before you took\n> one.\n\nYes.\n\nI was taking the line that the last slots in the array are reserved. Those are\nnot going to be taken by non su connections. Therefore, if MyBackendId is\nunder the lower limit it doesn't matter if it's the only slot free since the\n'safety' measure has already been used in restricting access to the last free\nslots and it just so happens that those sessions are still active.\n\nI take Neil's point about the order of the tests. That's my stupidity when\nrearranging stuff after noticing in tests that the user information wasn't\navailable where I was [also stupidly] expecting it to be first time around.\n\n\n-- \nNigel J. Andrews\n\n", "msg_date": "Mon, 26 Aug 2002 11:36:10 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations" }, { "msg_contents": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> I was taking the line that the last slots in the array are\n> reserved. Those are not going to be taken by non su connections.\n\nBut that doesn't do the job, does it? My view of the feature is that\nwhen there are at least MaxBackends - ReservedBackends slots in use (by\neither su or non-su connections) then no new non-su jobs should be let\nin. For example, if the system is full (with a mix of su and non-su\njobs) and one non-su job quits, don't we want to hold that slot for a\npossible su connection?\n\nYour approach does have the advantage of being very cheap to test\n(I think my semantics would require counting the active backends),\nbut I'm not sure that it really does what we want.\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 26 Aug 2002 10:00:48 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations " }, { "msg_contents": "Tom Lane wrote:\n> \"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> > I was taking the line that the last slots in the array are\n> > reserved. Those are not going to be taken by non su connections.\n> \n> But that doesn't do the job, does it? My view of the feature is that\n> when there are at least MaxBackends - ReservedBackends slots in use (by\n> either su or non-su connections) then no new non-su jobs should be let\n> in. For example, if the system is full (with a mix of su and non-su\n> jobs) and one non-su job quits, don't we want to hold that slot for a\n> possible su connection?\n> \n> Your approach does have the advantage of being very cheap to test\n> (I think my semantics would require counting the active backends),\n> but I'm not sure that it really does what we want.\n\nTom is right. If the last two slots are held by two long-running\nsuper-user backends, and the slots fill, there will be no reserved\nslots. The trick is that when the maximum number of backends is almost\nexceeded, only let the supuer-user in.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Mon, 26 Aug 2002 12:16:32 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations" }, { "msg_contents": "On Mon, 26 Aug 2002, Bruce Momjian wrote:\n\n> Tom Lane wrote:\n> > \"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> > > I was taking the line that the last slots in the array are\n> > > reserved. Those are not going to be taken by non su connections.\n> > \n> > But that doesn't do the job, does it? My view of the feature is that\n> > when there are at least MaxBackends - ReservedBackends slots in use (by\n> > either su or non-su connections) then no new non-su jobs should be let\n> > in. For example, if the system is full (with a mix of su and non-su\n> > jobs) and one non-su job quits, don't we want to hold that slot for a\n> > possible su connection?\n> > \n> > Your approach does have the advantage of being very cheap to test\n> > (I think my semantics would require counting the active backends),\n> > but I'm not sure that it really does what we want.\n> \n> Tom is right. If the last two slots are held by two long-running\n> super-user backends, and the slots fill, there will be no reserved\n> slots. The trick is that when the maximum number of backends is almost\n> exceeded, only let the supuer-user in.\n\nOkay, it's not how I was thinking as you know but I've got nothing against it\nother than the backend slot scan time. I don't think that would be a\nsignificant drain of cpu time so I'll implement that scheme and resubmit.\n\nGot some other stuff to do first so it won't be done immediately but will in\nthe next day or so; in time for beta assuming it doesn't fall foul of any patch\nreview interval required.\n\n\n-- \nNigel J. Andrews\n\n", "msg_date": "Mon, 26 Aug 2002 23:58:00 +0100 (BST)", "msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>", "msg_from_op": true, "msg_subject": "Re: [HACKERS] TODO Done. Superuser backend slot reservations" } ]
[ { "msg_contents": "Does anybody know 7.2.2 if xtar ball has been repacked? Peter pointed\nout the previous tar ball included wrong docs.\n--\nTatsuo Ishii\n", "msg_date": "Mon, 26 Aug 2002 11:08:50 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": true, "msg_subject": "PostgreSQL 7.2.2 and docs" }, { "msg_contents": "> Does anybody know 7.2.2 if xtar ball has been repacked? Peter pointed\n> out the previous tar ball included wrong docs.\n\nI have confirmed\n\ncvs.postgres.org://var/spool/ftp/pub/source/v7.2.2/postgresql-7.2.2.tar.gz\n\n(-rw-r--r-- 1 pgsql pgsql 9239158 Aug 22 23:25 postgresql-7.2.2.tar.gz)\n\nincludes old documents. Also I notice an announce for release 7.2.3\nappeaing on http://www.postgresql.org. Maybe we need to release\n7.2.3 for just a documentation update sake. Sigh.\n--\nTatsuo Ishii\n", "msg_date": "Mon, 26 Aug 2002 12:58:32 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": true, "msg_subject": "Re: PostgreSQL 7.2.2 and docs" }, { "msg_contents": "I say a re-release of 7.2.2 is sufficient, as it's just docs...\n\nWhat docs were in there? 7.3? or 7.2.1?\n\nChris\n\n> -----Original Message-----\n> From: pgsql-hackers-owner@postgresql.org\n> [mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Tatsuo Ishii\n> Sent: Monday, 26 August 2002 11:59 AM\n> To: pgsql-hackers@postgresql.org\n> Subject: Re: [HACKERS] PostgreSQL 7.2.2 and docs\n>\n>\n> > Does anybody know 7.2.2 if xtar ball has been repacked? Peter pointed\n> > out the previous tar ball included wrong docs.\n>\n> I have confirmed\n>\n> cvs.postgres.org://var/spool/ftp/pub/source/v7.2.2/postgresql-7.2.2.tar.gz\n>\n> (-rw-r--r-- 1 pgsql pgsql 9239158 Aug 22 23:25 postgresql-7.2.2.tar.gz)\n>\n> includes old documents. Also I notice an announce for release 7.2.3\n> appeaing on http://www.postgresql.org. Maybe we need to release\n> 7.2.3 for just a documentation update sake. Sigh.\n> --\n> Tatsuo Ishii\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", "msg_date": "Mon, 26 Aug 2002 12:01:42 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: PostgreSQL 7.2.2 and docs" }, { "msg_contents": "\nthe release docs are pulled from petere's account:\n\ncp ~petere/man.tar.gz ~petere/postgres.tar.gz doc\n\n\n\nOn Mon, 26 Aug 2002, Christopher Kings-Lynne wrote:\n\n> I say a re-release of 7.2.2 is sufficient, as it's just docs...\n>\n> What docs were in there? 7.3? or 7.2.1?\n>\n> Chris\n>\n> > -----Original Message-----\n> > From: pgsql-hackers-owner@postgresql.org\n> > [mailto:pgsql-hackers-owner@postgresql.org]On Behalf Of Tatsuo Ishii\n> > Sent: Monday, 26 August 2002 11:59 AM\n> > To: pgsql-hackers@postgresql.org\n> > Subject: Re: [HACKERS] PostgreSQL 7.2.2 and docs\n> >\n> >\n> > > Does anybody know 7.2.2 if xtar ball has been repacked? Peter pointed\n> > > out the previous tar ball included wrong docs.\n> >\n> > I have confirmed\n> >\n> > cvs.postgres.org://var/spool/ftp/pub/source/v7.2.2/postgresql-7.2.2.tar.gz\n> >\n> > (-rw-r--r-- 1 pgsql pgsql 9239158 Aug 22 23:25 postgresql-7.2.2.tar.gz)\n> >\n> > includes old documents. Also I notice an announce for release 7.2.3\n> > appeaing on http://www.postgresql.org. Maybe we need to release\n> > 7.2.3 for just a documentation update sake. Sigh.\n> > --\n> > Tatsuo Ishii\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>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n>\n\n", "msg_date": "Mon, 26 Aug 2002 01:23:54 -0300 (ADT)", "msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>", "msg_from_op": false, "msg_subject": "Re: PostgreSQL 7.2.2 and docs" }, { "msg_contents": "> the release docs are pulled from petere's account:\n> \n> cp ~petere/man.tar.gz ~petere/postgres.tar.gz doc\n\nThey are too old.\n\n$ ls -l man.tar.gz postgres.tar.gz \n-rw-r--r-- 1 petere wheel 125177 Mar 27 11:51 man.tar.gz\n-rw-r--r-- 1 petere wheel 843714 Mar 26 23:39 postgres.tar.gz\n--\nTatsuo Ishii\n", "msg_date": "Mon, 26 Aug 2002 14:18:49 +0900 (JST)", "msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>", "msg_from_op": true, "msg_subject": "Re: PostgreSQL 7.2.2 and docs" }, { "msg_contents": "On Mon, 26 Aug 2002, Tatsuo Ishii wrote:\n\n> Also I notice an announce for release 7.2.3\n> appeaing on http://www.postgresql.org.\n\nYou do? Where?\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 http://www.camping-usa.com http://www.cloudninegifts.com\n http://www.meanstreamradio.com http://www.unknown-artists.com\n==========================================================================\n\n\n\n", "msg_date": "Mon, 26 Aug 2002 06:15:53 -0400 (EDT)", "msg_from": "Vince Vielhaber <vev@michvhf.com>", "msg_from_op": false, "msg_subject": "Re: PostgreSQL 7.2.2 and docs" }, { "msg_contents": "I think he means 7.2.2...\n\nChris\n\nOn Mon, 26 Aug 2002, Vince Vielhaber wrote:\n\n> On Mon, 26 Aug 2002, Tatsuo Ishii wrote:\n>\n> > Also I notice an announce for release 7.2.3\n> > appeaing on http://www.postgresql.org.\n>\n> You do? Where?\n>\n> Vince.\n> --\n> ==========================================================================\n> Vince Vielhaber -- KA8CSH email: vev@michvhf.com http://www.pop4.net\n> 56K Nationwide Dialup from $16.00/mo at Pop4 Networking\n> http://www.camping-usa.com http://www.cloudninegifts.com\n> http://www.meanstreamradio.com http://www.unknown-artists.com\n> ==========================================================================\n>\n>\n>\n>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n>\n\n", "msg_date": "Mon, 26 Aug 2002 18:21:04 +0800 (WST)", "msg_from": "Christopher Kings-Lynne <chriskl@familyhealth.com.au>", "msg_from_op": false, "msg_subject": "Re: PostgreSQL 7.2.2 and docs" } ]
[ { "msg_contents": "Hi all,\n\nI've been having a lot of trouble with deadlocks in 7.2.1 because of\nforeign keys. I dug out a couple of messages from the list archives\nwhich cover this topic.\n\nOne particular message indicates a fix was being worked on.\n--------\nDate: Thu, 11 Apr 2002 09:03:06 -0700 (PDT)\nFrom: Stephan Szabo\n\nPlease see past disussions on the fact that the lock grabbed is too\nstrong. I'm going to (when I get time to work on it) try out a lower\nstrength lock that Alex Hayward made a patch for that should\nlimit/prevent these cases. Thanks for sending a nice simple test case\nto try against :)\n----------\n\nMy question is if this is now resolved for 7.3? I'm trying to figure out\nwhat I can do with my 7.2.1 problems... Has anyone got a back-ported\npatch, perhaps against 7.2.2?\n\nAny help appreciated, thanks!\n\nTom.\n-- \nThomas O'Dowd. - Nooping - http://nooper.com\ntom@nooper.com - Testing - http://nooper.co.jp/labs\n\n", "msg_date": "26 Aug 2002 11:27:12 +0900", "msg_from": "Thomas O'Dowd <tom@nooper.com>", "msg_from_op": true, "msg_subject": "Re: Deadlock situation using foreign keys (reproduceable)" }, { "msg_contents": "\nOn 26 Aug 2002, Thomas O'Dowd wrote:\n\n> Hi all,\n>\n> I've been having a lot of trouble with deadlocks in 7.2.1 because of\n> foreign keys. I dug out a couple of messages from the list archives\n> which cover this topic.\n>\n> One particular message indicates a fix was being worked on.\n> --------\n> Date: Thu, 11 Apr 2002 09:03:06 -0700 (PDT)\n> From: Stephan Szabo\n>\n> Please see past disussions on the fact that the lock grabbed is too\n> strong. I'm going to (when I get time to work on it) try out a lower\n> strength lock that Alex Hayward made a patch for that should\n> limit/prevent these cases. Thanks for sending a nice simple test case\n> to try against :)\n> ----------\n>\n> My question is if this is now resolved for 7.3? I'm trying to figure out\n> what I can do with my 7.2.1 problems... Has anyone got a back-ported\n> patch, perhaps against 7.2.2?\n\nIt's actually not resolved because my attempts at the lower strength lock\nopened up holes and fixing those opened up new deadlock situations, and\nI ran out of time in the 7.3 cycle to try to make it work.\n\n", "msg_date": "Sun, 25 Aug 2002 20:46:35 -0700 (PDT)", "msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>", "msg_from_op": false, "msg_subject": "Re: Deadlock situation using foreign keys (reproduceable)" }, { "msg_contents": "Thanks for your feedback Stephan. Seems like a tough fix. Pitty it won't\nmake it into 7.3. I presume there are other folk out there suffering\nfrom the same problems that I'm having. What approaches if any have\npeople taken to work around this problem? I read in the list that one\nuser patched his postmaster to explictly ignore the RI \"FOR UPDATE\" on\ntables he knew weren't changing. Can't find the detailed message in the\narchives right now, but I read it earlier. Any other work-arounds?\n\nTom. \n\nOn Mon, 2002-08-26 at 12:46, Stephan Szabo wrote:\n> > My question is if this is now resolved for 7.3? I'm trying to figure out\n> > what I can do with my 7.2.1 problems... Has anyone got a back-ported\n> > patch, perhaps against 7.2.2?\n> \n> It's actually not resolved because my attempts at the lower strength lock\n> opened up holes and fixing those opened up new deadlock situations, and\n> I ran out of time in the 7.3 cycle to try to make it work.\n-- \nThomas O'Dowd. - Nooping - http://nooper.com\ntom@nooper.com - Testing - http://nooper.co.jp/labs\n\n", "msg_date": "26 Aug 2002 13:33:39 +0900", "msg_from": "Thomas O'Dowd <tom@nooper.com>", "msg_from_op": true, "msg_subject": "Re: Deadlock situation using foreign keys (reproduceable)" }, { "msg_contents": "On 26 Aug 2002, Thomas O'Dowd wrote:\n\n> Thanks for your feedback Stephan. Seems like a tough fix. Pitty it won't\n> make it into 7.3. I presume there are other folk out there suffering\n> from the same problems that I'm having. What approaches if any have\n> people taken to work around this problem? I read in the list that one\n> user patched his postmaster to explictly ignore the RI \"FOR UPDATE\" on\n> tables he knew weren't changing. Can't find the detailed message in the\n> archives right now, but I read it earlier. Any other work-arounds?\n\nFor many schemas setting the constraints to deferrable makes the problem\ngo away in practice (due to the fact the constraint locks are held for\nless time). Removing for update really only does work if you know that\none side of the constraint never changes concurrently with the other side.\n\n\n", "msg_date": "Sun, 25 Aug 2002 21:58:23 -0700 (PDT)", "msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>", "msg_from_op": false, "msg_subject": "Re: Deadlock situation using foreign keys (reproduceable)" }, { "msg_contents": "I wrote this patch for my system, and it works fine. However, it's a really ugly workaround. I can publish the source\nif anybody is interested.\n\nAm Montag, 26. August 2002 06:33 schrieb Thomas O'Dowd:\n> Thanks for your feedback Stephan. Seems like a tough fix. Pitty it won't\n> make it into 7.3. I presume there are other folk out there suffering\n> from the same problems that I'm having. What approaches if any have\n> people taken to work around this problem? I read in the list that one\n> user patched his postmaster to explictly ignore the RI \"FOR UPDATE\" on\n> tables he knew weren't changing. Can't find the detailed message in the\n> archives right now, but I read it earlier. Any other work-arounds?\n>\n> Tom.\n>\n> On Mon, 2002-08-26 at 12:46, Stephan Szabo wrote:\n> > > My question is if this is now resolved for 7.3? I'm trying to figure\n> > > out what I can do with my 7.2.1 problems... Has anyone got a\n> > > back-ported patch, perhaps against 7.2.2?\n> >\n> > It's actually not resolved because my attempts at the lower strength lock\n> > opened up holes and fixing those opened up new deadlock situations, and\n> > I ran out of time in the 7.3 cycle to try to make it work.\n\n", "msg_date": "Mon, 26 Aug 2002 08:50:42 +0200", "msg_from": "Mario Weilguni <mweilguni@sime.com>", "msg_from_op": false, "msg_subject": "Re: Deadlock situation using foreign keys (reproduceable)" }, { "msg_contents": "\nOn Mon, 26 Aug 2002, Mario Weilguni wrote:\n\n> I wrote this patch for my system, and it works fine. However, it's a\n> really ugly workaround. I can publish the source if anybody is\n> interested.\n\nWhich, the for update one or a lower strength lock? In either case,\ncertainly.\n\n\n", "msg_date": "Mon, 26 Aug 2002 08:29:09 -0700 (PDT)", "msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>", "msg_from_op": false, "msg_subject": "Re: Deadlock situation using foreign keys (reproduceable)" } ]
[ { "msg_contents": "\n\"..if someone has direct SQL access to your database, they can\nalready do more damage than what this vulnerability addresses. Specifically\nDROP TABLE is available to users with direct SQL command line access...\"\n\nThat's true of course, but i really dont want to do any damage, i might\neven don't want\nto get noticed...\n\n\".. the cracker must be able to execute arbitrary SQL commands against the\ndatabase, \nand by that stage of the game, a DoS\nattack is already trivial (e.g. disable GEQO and execute a 15 table join\nquery)...\"\n\nHmm, sounds complex, why bother?. Thanks for the advice anyway :-)\n\n//@(#) Mordred Labs advisory 0x0006\n\nRelease data: 26/08/02\nName: Two dumb DoS conditions in PostgreSQL\nVersions affected: all versions\nRisk: very low\n\n---[ Description:\n\n1) Upon invoking a substring(text, integer, integer) function, a\nsrc/backend/utils/adt/varlena.c:text_substr()\nfunction will gets called, which fails to detect a simple dos condition,\ntriggered by a very high third argument.\nMultibyte support must be enabled for this attack to work.\nThe vulnerable encodings are: SQL_ASCII, LATIN1.\nWith others, you just will get SIGSEGV (not checked).\n\n2) Upon invoking a bpchar(char, integer) function, a\nsrc/backend/utils/adt/varchar.c:bpchar() function\nwill gets called, which suffers from a dos condition.\n\n--[ How to reproduce:\n\ntemplate1=# select substring('xxxxxxxx',2,2147483647); \ntemplate1=# select bpchar('x',100000000); \n\n--[ Solution\n\nNo one is available.\n\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n", "msg_date": "Mon, 26 Aug 2002 12:35:44 +0000", "msg_from": "Sir Mordred The Traitor <mordred@s-mail.com>", "msg_from_op": true, "msg_subject": "@(#)Mordred Labs advisory 0x0006: Two minor DoS conditions in\n\tPostgreSQL" }, { "msg_contents": "Sir Mordred The Traitor <mordred@s-mail.com> writes:\n> template1=# select substring('xxxxxxxx',2,2147483647); \n\nWith CVS HEAD (with database encoding = SQL_ASCII and UNICODE), I get:\n\nnconway=# select substring('xxxxxxxx',2,2147483647);\nERROR: negative substring length not allowed\n\nWith REL7_2_STABLE, I get:\n\ntemplate1=# select substring('xxxxxxxx',2,2147483647);\nERROR: MemoryContextAlloc: invalid request size 2147483651\n\n> template1=# select bpchar('x',100000000); \n\nWith both CVS HEAD and REL7_2_STABLE, I get:\n\ntemplate1=# select bpchar('x',100000000);\nzsh: 7312 segmentation fault (core dumped) ./psql template1\n\n(note that it's the client, and not the backend, that crashes)\n\nAlthough the backend does allocate a couple hundred megs of memory\nwhile processing the query.\n\nAlthough I haven't looked at the code yet, it's probably worth noting\nthat the two test cases posted above are not cut-and-dry DoS\nopportunities, AFAICT -- however, the code may still be vulnerable.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "26 Aug 2002 10:11:54 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0006: Two minor DoS conditions in\n\tPostgreSQL" } ]
[ { "msg_contents": "\n//@(#) Mordred Labs advisory 0x0007\n\nRelease data: 26/08/02\nName: Remote DoS condition in PostgreSQL\nVersions affected: all versions\nConditions: entry in a pg_hba.conf file that matches attacker's host.\nRisk: average\n\n---[ Description:\n\nUpon connecting to a database, postmaster will fork a new process. \nAfter that, a child process will call a\nsrc/backend/postmaster/postmaster.c:DoBackend() routine, \nwhich after processing a startup packet (see src/include/libpq/pqcomm.h), \nwill invoke a src/backend/libpq/auth.c:ClientAuthentication() routine to\nperform client authentication.\nIf there is an entry in pg_hba.conf file, that matches an attacker's host, \nan attacker could trigger\ninvocation of src/backend/libpq/auth.c:recv_and_check_password0(), which\nfails to detect a DoS condition.\n\n---[ Details:\nConsider this snip of code from src/backend/libpq/auth.c:\n\n[snip]\nstatic int recv_and_check_password0(Port *port) {\n\tint32 len;\n\tchar *buf;\n \n\tif (pq_getint(&len, 4) == EOF)\n\t\treturn STATUS_EOF;\n\tlen -= 4;\n\tbuf = palloc(len); /* len is taken from a packet */\n[snip]\n\nNote, that the size of palloced memory is taken from the user's input,\nwhich is stupid if you ask me.\n\n--[ How to reproduce:\n\nI dont want to provide any tools to illustrate this vulnerability.\n\n--[ Solution\n\nDisable network access for untrusted users.\n\n\n\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n", "msg_date": "Mon, 26 Aug 2002 14:46:19 +0000", "msg_from": "Sir Mordred The Traitor <mordred@s-mail.com>", "msg_from_op": true, "msg_subject": "@(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "On 26 Aug 2002 at 14:46, Sir Mordred The Traitor wrote:\n> [snip]\n> static int recv_and_check_password0(Port *port) {\n> \tint32 len;\n> \tchar *buf;\n> \n> \tif (pq_getint(&len, 4) == EOF)\n> \t\treturn STATUS_EOF;\n> \tlen -= 4;\n> \tbuf = palloc(len); /* len is taken from a packet */\n> [snip]\n\nSo that should read,\n\n buf=palloc((len>LENMAX?SAFELEN:len));\n\nis what you want to say? \n\nsounds good to me.. But if it is taken from the packet, won't that be tripped \nto MTA size? Just a na�ve question. Never saw much of postgres code myself..\n\n\nBye\n Shridhar\n\n--\nWilcox's Law:\tA pat on the back is only a few centimeters from a kick in the \npants.\n\n", "msg_date": "Mon, 26 Aug 2002 20:25:15 +0530", "msg_from": "\"Shridhar Daithankar\" <shridhar_daithankar@persistent.co.in>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "Sir Mordred The Traitor <mordred@s-mail.com> writes:\n> Note, that the size of palloced memory is taken from the user's input,\n> which is stupid if you ask me.\n\nBeyond causing an \"out of memory\" error during the handshake, I fail to\nsee how there can be any problem. palloc is considerably more robust\nthan malloc.\n\n> I dont want to provide any tools to illustrate this vulnerability.\n\nPerhaps you haven't tried.\n\nIt may indeed make sense to put a range check here, but I'm getting\ntired of hearing the words \"dos attack\" applied to conditions that\ncannot be exploited to cause any real problem. All you are\naccomplishing is to spread FUD among people who aren't sufficiently\nfamiliar with the code to evaluate the seriousness of problems...\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 26 Aug 2002 11:02:56 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL " }, { "msg_contents": "On Monday 26 August 2002 10:46 am, Sir Mordred The Traitor wrote:\n> Conditions: entry in a pg_hba.conf file that matches attacker's host.\n> Risk: average\n\n> --[ Solution\n>\n> Disable network access for untrusted users.\n\nTCP/IP access must be enabled as well. TCP/IP accessibility is OFF by \ndefault.\n\nI for one thought that it was normal operating procedure to only allow access \nto trusted machines; maybe I'm odd in that regard.\n\nHey, if I can connect to postmaster I can DoS it quite easily, but flooding it \nwith connection requests.....\n\nBut, if we can thwart this, all the better.\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Mon, 26 Aug 2002 11:18:48 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "-*- Lamar Owen <lamar.owen@wgcr.org> [ 2002-08-26 15:19 ]:\n> TCP/IP access must be enabled as well. TCP/IP accessibility is OFF by \n> default.\n> \n> I for one thought that it was normal operating procedure to only allow access \n> to trusted machines; maybe I'm odd in that regard.\n> \n> Hey, if I can connect to postmaster I can DoS it quite easily, but flooding it \n> with connection requests.....\n> \n> But, if we can thwart this, all the better.\n\nWell, ISP's that offer webhosting and database connectivity might also be running a PostgreSQL server that only allows connections from that specific webserver (TCP port 5432 access not blocked as well as an pg_hba.conf entry). Now, if a user with access to the webserver has privileges to open a socket connection, he could exploit this.\n\n\n-- \nRegards,\nTolli\ntolli@tol.li\n", "msg_date": "Mon, 26 Aug 2002 15:27:57 +0000", "msg_from": "=?iso-8859-1?Q?=DE=F3rhallur_H=E1lfd=E1narson?= <tolli@tol.li>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "so basically if you are an idiot admin, and leave the postgresql box\nopen (explicitly opening stuff), and under certian conditions, you can\nget DoS'd? hrm, this may not be your biggest problem.\n\nmaybe if the dba has a clue and only explicitly allows certian ips\nto even route to the box, and then certian users (1 or 2 or so) that\nis not available to the public (ie, internet), they would be better off.\ni would be that with the lazy/ignorant setup of the dba/admin, that a \nDoS of postgresql is not the biggest problem, sure one of their redhat\nboxes has gotten rooted already...\n\nthere is nothing that is more important for security and databases than\nsetting them up correctly, and their place on the network. the database\nis the crown jewel that should never been seen or touched except for when\n_absolutely_ needed, and that must be under heavy control. \n\nthere is a bigger problem here than postgresql, it is the dumbass factor\nof people that try to run a db, and are vuln to anything... and then complain\nabout it... i find this very annoying. \n\nknow what you are doing, or stfu is my opinion\n\n-chris\n\nps -> note this was not directed at any one person, but to the mass of \n people that never should run a db, and go back to eating paint chips.\n-----\ndisclaimer: i do not speak on behalf of devis (devis.com). i speak\n on my own behalf. \n-----\n\n</rant-mode>\n\n\nLamar Owen writes:\n > On Monday 26 August 2002 10:46 am, Sir Mordred The Traitor wrote:\n > > Conditions: entry in a pg_hba.conf file that matches attacker's host.\n > > Risk: average\n > \n > > --[ Solution\n > >\n > > Disable network access for untrusted users.\n > \n > TCP/IP access must be enabled as well. TCP/IP accessibility is OFF by \n > default.\n > \n > I for one thought that it was normal operating procedure to only allow access \n > to trusted machines; maybe I'm odd in that regard.\n > \n > Hey, if I can connect to postmaster I can DoS it quite easily, but flooding it \n > with connection requests.....\n > \n > But, if we can thwart this, all the better.\n > -- \n > Lamar Owen\n > WGCR Internet Radio\n > 1 Peter 4:11\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-- \nChris Humphries\nDevelopment InfoStructure\n540.366.9809 \n", "msg_date": "Mon, 26 Aug 2002 11:25:35 -0500", "msg_from": "Chris Humphries <chumphries@devis.com>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "Tom Lane wrote:\n> Sir Mordred The Traitor <mordred@s-mail.com> writes:\n> > Note, that the size of palloced memory is taken from the user's input,\n> > which is stupid if you ask me.\n> \n> Beyond causing an \"out of memory\" error during the handshake, I fail to\n> see how there can be any problem. palloc is considerably more robust\n> than malloc.\n> \n> > I dont want to provide any tools to illustrate this vulnerability.\n> \n> Perhaps you haven't tried.\n> \n> It may indeed make sense to put a range check here, but I'm getting\n> tired of hearing the words \"dos attack\" applied to conditions that\n> cannot be exploited to cause any real problem. All you are\n> accomplishing is to spread FUD among people who aren't sufficiently\n> familiar with the code to evaluate the seriousness of problems...\n\nSir-* does have a point. A valid host in pg_hba.conf can cause DOS by\njust connecting over and over, but allocating almost all of the memory\non the machine would affect other applications running on the machine,\neven non-networked applications, as well as PostgreSQL, while a\nconnection DOS effects only PostgreSQL.\n\nIt isn't fun to have our code nit-picked apart, and Sir-* is over-hyping\nthe vulnerability, but it is a valid concern. The length should\nprobably be clipped to a reasonable length and a comment put in the code\ndescribing why.\n\nThere are a whole host of vulnerability scenarios that we have not\nexplored. In fact, with 7.2.2, we got into the data-based crash vs.\nquery-based crash issue that we have not discussed before in detail. \n\nWith this report, we have a non-authorized user causing possible memory\nexhaustion for all applications on the server. This is a greater\nvulnerability than the random query argument because in such cases the\nuser is authorized to issue queries.\n\nSo, Sir-* and others, please understand that databases have their own\nvulnerability envelope that is much more complex than a standard network\napplication like ftp. There is almost an entirely new vocabulary for\ndescribing such vulnerabilities which we have been crafting as we\ndiscuss it. This vocabulary includes:\n\n\tconnection-enabled hosts vs. any host\n\tauthenticated vs. non-authenticated users\n\tusers with data insertion access\n\tusers with query access\n\nI think the last group can not be protected from malevolent queries, but\nthe former can be tightened, and Sir-* is working on that. Sir-*, when\ndescribing vulnerabilities, especially in public forums, please try to\nuse this language so people are not overly agitated by your warnings.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Mon, 26 Aug 2002 12:59:08 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "On Monday 26 August 2002 12:59 pm, Bruce Momjian wrote:\n> Tom Lane wrote:\n> > It may indeed make sense to put a range check here, but I'm getting\n> > tired of hearing the words \"dos attack\" applied to conditions that\n> > cannot be exploited to cause any real problem. All you are\n> > accomplishing is to spread FUD among people who aren't sufficiently\n> > familiar with the code to evaluate the seriousness of problems...\n\n> It isn't fun to have our code nit-picked apart, and Sir-* is over-hyping\n> the vulnerability, but it is a valid concern. The length should\n> probably be clipped to a reasonable length and a comment put in the code\n> describing why.\n\nThe pseudo-security-alert format used isn't terribly palatable here, IMHO. On \nBugTraq it might fly -- but not here. A simple 'Hey guys, I found a possible \nproblem when.....' without the big-sounding fluff would sit better with me, \nat least. The substance of the message is perhaps valuable -- but the \nwrapper distracts from the substance.\n\nAnd dealing with a real name would be nice, IMHO. Otherwise we may end up \nwith 'SMtT' as the nickname -- Hmmm, 'SMitTy' perhaps? :-) Reminds me of \n'Uncle George' who did quite a bit for the Alpha port and then disappeared.\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Mon, 26 Aug 2002 13:50:09 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "Lamar Owen wrote:\n> And dealing with a real name would be nice, IMHO. Otherwise we may end up \n> with 'SMtT' as the nickname -- Hmmm, 'SMitTy' perhaps? :-) Reminds me of \n> 'Uncle George' who did quite a bit for the Alpha port and then disappeared.\n\nFunny you mention that. Now knowing someone's name is troubling, but I\nam not sure why.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Mon, 26 Aug 2002 14:14:58 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "pgman@candle.pha.pa.us (Bruce Momjian) wrote \n\n> Sir-* does have a point. A valid host in pg_hba.conf can cause DOS by\n> just connecting over and over, but allocating almost all of the memory\n> on the machine would affect other applications running on the machine,\n> even non-networked applications, as well as PostgreSQL, while a\n> connection DOS effects only PostgreSQL.\n\n\noh woe is he the man who does not use rlimits, or perhaps, concurrency \nlimits?\n\nit seems this is nothing new, all network available services are subject to \ndos or ddos attacks... and if you dont setup limits on your machine, then \nother things can be affected... inetd, bind, sendmail, finger, <insert \nfavorite network accessible program here>, etc...\n\nI do agree that pgsql should not just arbitrarily allocate memory like \nthis, as defensive programming, but I cannot agree that this is a bug or \nproblem in pgsql per se.\n\nAs a side note, if someone wanted to shift the discussion to allowing \nconcurrency limits in pgsql, how would/could this fit into the context of \nanother thread where it was discussed to be able to always allow certain \nusers to login...\n", "msg_date": "Mon, 26 Aug 2002 18:28:55 +0000 (UTC)", "msg_from": "ngpg@grymmjack.com", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" } ]
[ { "msg_contents": "The point is really simple.\nAllocate a huge chunk of memory (no sense to cause out of memory error,\nas palloc will bail is a requested size > 1 gb). The postgres will be ready\nto suck your input,\nvia pg_getbytes(), now in a loop send junk to postgresql.\nOf course you can fork a number of processes to improve your effect.\nThe issues is that postgres allocate a chunk of memory and reads data,\nusing an\nuser's input, which has not completed authentication.\nThis is badly anyway.\nOf course i tried, and wrote proggy for that, \nbut i can repeat, i dont want to provide it here.\n\n\n>Sir Mordred The Traitor <mordred@s-mail.com> writes:\n>> Note, that the size of palloced memory is taken from the user's input,\n>> which is stupid if you ask me.\n>\n>Beyond causing an \"out of memory\" error during the handshake, I fail to\n>see how there can be any problem. palloc is considerably more robust\n>than malloc.\n>\n>> I dont want to provide any tools to illustrate this vulnerability.\n>\n>Perhaps you haven't tried.\n>\n>It may indeed make sense to put a range check here, but I'm getting\n>tired of hearing the words \"dos attack\" applied to conditions that\n>cannot be exploited to cause any real problem. All you are\n>accomplishing is to spread FUD among people who aren't sufficiently\n>familiar with the code to evaluate the seriousness of problems...\n>\n>\t\t\tregards, tom lane\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\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n", "msg_date": "Mon, 26 Aug 2002 15:25:18 +0000", "msg_from": "Sir Mordred The Traitor <mordred@s-mail.com>", "msg_from_op": true, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" } ]
[ { "msg_contents": "Not a lower strength lock, I would, but I'm not so familiar with the postgres\ninternals. I modified ri-triggers.c to exclude certain tables from the\nlocking itself (because I know the tables are not updated).\n\n-----Ursprüngliche Nachricht-----\nVon: Stephan Szabo [mailto:sszabo@megazone23.bigpanda.com]\nGesendet: Montag, 26. August 2002 17:29\nAn: Mario Weilguni\nCc: pgsql-hackers@postgresql.org\nBetreff: Re: [HACKERS] Deadlock situation using foreign keys\n(reproduceable)\n\n\n\nOn Mon, 26 Aug 2002, Mario Weilguni wrote:\n\n> I wrote this patch for my system, and it works fine. However, it's a\n> really ugly workaround. I can publish the source if anybody is\n> interested.\n\nWhich, the for update one or a lower strength lock? In either case,\ncertainly.\n\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", "msg_date": "Mon, 26 Aug 2002 17:30:28 +0200", "msg_from": "\"Mario Weilguni\" <mario.weilguni@icomedias.com>", "msg_from_op": true, "msg_subject": "Re: Deadlock situation using foreign keys (reproduceable)" }, { "msg_contents": "On Mon, 26 Aug 2002, Mario Weilguni wrote:\n\n> Not a lower strength lock, I would, but I'm not so familiar with the postgres\n> internals. I modified ri-triggers.c to exclude certain tables from the\n> locking itself (because I know the tables are not updated).\n\nIt might help the op of this thread, so it's probably worth\nposting. :)\n\n\n", "msg_date": "Mon, 26 Aug 2002 08:33:25 -0700 (PDT)", "msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>", "msg_from_op": false, "msg_subject": "Re: Deadlock situation using foreign keys (reproduceable)" } ]
[ { "msg_contents": ">Hey, if I can connect to postmaster I can DoS it quite easily, but\nflooding it\n>with connection requests.....\n\nHm, that's true of course, but now i will do this with a couple of\nconnections.\nLets say, bot on a owned machine, connects to a database, \nsend a crafted packet,\npostgresql will allocate a huge amount of memory, and will be \nhappy to read anything it recvs from my bot.\n\n\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n", "msg_date": "Mon, 26 Aug 2002 15:31:26 +0000", "msg_from": "Sir Mordred The Traitor <mordred@s-mail.com>", "msg_from_op": true, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "-*- Sir Mordred The Traitor <mordred@s-mail.com> [ 2002-08-26 15:32 ]:\n> >Hey, if I can connect to postmaster I can DoS it quite easily, but\n> flooding it\n> >with connection requests.....\n> \n> Hm, that's true of course, but now i will do this with a couple of\n> connections.\n> Lets say, bot on a owned machine, connects to a database, \n> send a crafted packet,\n> postgresql will allocate a huge amount of memory, and will be \n> happy to read anything it recvs from my bot.\n\nSpeaking of which.\n\nIf I understand correctly, a new backend is forked and the connection dispatched to that specific backend, once access has been granted (with means of user/pass authentication, ident or whatever).\n\nIs there any check for connection to the postmaster that have not been dispatched to a new backend after X bytes (or seconds?), to free resources (would that make any sense? :)\n\nAnd another (perhaps silly) thought: Currently, if the authentication process is exploited, it would kill the postmaster, resulting in a total crash of the whole database system. Would it be beneficial to split the connection handling/authorization process to a seperate process, and if that process dies, the postmaster would simply start a new one, there for not affecting any other backends that are running (for authorized users) ? Or am I way of track? :)\n\n\n-- \nRegards,\nTolli\ntolli@tol.li\n", "msg_date": "Mon, 26 Aug 2002 15:42:12 +0000", "msg_from": "=?iso-8859-1?Q?=DE=F3rhallur_H=E1lfd=E1narson?= <tolli@tol.li>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "=?iso-8859-1?Q?=DE=F3rhallur_H=E1lfd=E1narson?= <tolli@tol.li> writes:\n> And another (perhaps silly) thought: Currently, if the authentication\n> process is exploited, it would kill the postmaster, resulting in a\n> total crash of the whole database system. Would it be beneficial to\n> split the connection handling/authorization process to a seperate\n> process, and if that process dies, the postmaster would simply start a\n> new one, there for not affecting any other backends that are running\n> (for authorized users) ? Or am I way of track? :) \n\nNo, just behind the times ;-). We did that in 7.2.\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 30 Aug 2002 00:42:00 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL " } ]
[ { "msg_contents": "By the way, seems like a beer DOSsed me a little:-).\nI've made i mistake in email subject. Remove->remote\n\n\n________________________________________________________________________\nThis letter has been delivered unencrypted. We'd like to remind you that\nthe full protection of e-mail correspondence is provided by S-mail\nencryption mechanisms if only both, Sender and Recipient use S-mail.\nRegister at S-mail.com: http://www.s-mail.com/inf/en\n", "msg_date": "Mon, 26 Aug 2002 15:34:39 +0000", "msg_from": "Sir Mordred The Traitor <mordred@s-mail.com>", "msg_from_op": true, "msg_subject": "Re: btw" } ]
[ { "msg_contents": "I'm looking at modifications to dblink to allow it to make use of table \nfunctions. One item which is very much needed is the ability to define \ndblink as returning setof record, so that it can return a tuple as \ndescribed at runtime in the query string.\n\nI'm trying to come up with the best method to pass the query string \ncolumndef, or better yet the tuple description, to the function. Any \nsuggestions on an approach?\n\nThanks,\n\nJoe\n\n", "msg_date": "Mon, 26 Aug 2002 11:21:49 -0700", "msg_from": "Joe Conway <mail@joeconway.com>", "msg_from_op": true, "msg_subject": "anonymous composite types - how to pass tupdesc to the function" }, { "msg_contents": "Joe Conway <mail@joeconway.com> writes:\n> I'm trying to come up with the best method to pass the query string \n> columndef, or better yet the tuple description, to the function. Any \n> suggestions on an approach?\n\nCan't it get it for itself from the results of the query, ie, look at\nPQftype() and so on to build a tupledesc?\n\nI guess there are some gotchas with inconsistent type OIDs between\nremote and local databases, but that still seems much less of a risk\nthan manual errors in giving the columnset definition. You could at\nleast check that PQfsize matches the local type's typlen as a way of\ndetecting chance collisions of user-defined type OIDs.\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 26 Aug 2002 15:11:04 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: anonymous composite types - how to pass tupdesc to the function " }, { "msg_contents": "Tom Lane wrote:\n> Joe Conway <mail@joeconway.com> writes:\n> \n>>I'm trying to come up with the best method to pass the query string \n>>columndef, or better yet the tuple description, to the function. Any \n>>suggestions on an approach?\n> \n> \n> Can't it get it for itself from the results of the query, ie, look at\n> PQftype() and so on to build a tupledesc?\n\nHmm. Good point. That certainly works for dblink.\n\nI guess most functions with need for anonymous composite types would be \nable to derive a tupdesc from libpq (dblink), SPI \n(tablefunc.c:crosstab), function arguments (tablefunc.c:crosstab), or it \nwould be known in advance (guc.c:show_all_settings).\n\nCan anyone think of a use case where the *only* source of tuple \ndescription would come from the query column def?\n\n\n> I guess there are some gotchas with inconsistent type OIDs between\n> remote and local databases, but that still seems much less of a risk\n> than manual errors in giving the columnset definition. You could at\n> least check that PQfsize matches the local type's typlen as a way of\n> detecting chance collisions of user-defined type OIDs.\n\nAnother good point.\n\nThanks!\n\nJoe\n\n\n", "msg_date": "Mon, 26 Aug 2002 12:52:13 -0700", "msg_from": "Joe Conway <mail@joeconway.com>", "msg_from_op": true, "msg_subject": "Re: anonymous composite types - how to pass tupdesc to" } ]
[ { "msg_contents": "RPMs for the security bugfix release 7.2.2 are available now for your happy \ndownloading in two forms and versions on ftp.postgresql.org. You may have to \nwait on mirror propagation, which could take a few hours.\n\nFor those who want it as close to the previous release as possible, I have \nuploaded the binaries I built the other day for the 0.1PGDG prerelease -- \nthey are identical in packaging to the last 7.2.1 PGDG RPM's I released. \n\nFor those who want a more serviceable contrib packaging, I have finally \napplied Peter Eisentraut's patch for the contrib tree building. Also, due to \nthe difficulty usually found when upgrading postgresql from a previous MAJOR \nversion, this version's server subpackage Conflicts with server subpackages \nof release 7.1.3 and earlier. This is a prelude to the complete removal of \nthe broken semifunctional upgrade capability -- you have been warned, and you \nwill have to either override the upgrade with rpm --nodeps or rpm -e the \nprevious postgresql-server FIRST, after dumping your data.\n\nDownload the 7.2.2-1PGDG release for the latter, 7.2.2-0.1PGDG release for the \nformer.\n\nBoth are available at \nftp://ftp.postgresql.org/pub/binary/v7.2.2/RPMS/redhat-7.3\nSource RPMS at\nftp://ftp.postgresql.org/pub/binary/v7.2.2/RPMS/SRPMS\n\nNo SPARC binaries for Red Hat 6.2 yet... :-)\n\nCHANGELOG:\n* Mon Aug 26 2002 Lamar Owen <lamar.owen@ramifordistat.net>\n- 7.2.2-1PGDG\n- Applied PeterE's contrib patch -- contrib is now completely restructured.\n- server conflicts with less than 7.2, as a warning to the upcoming\n-- yank of the broken partial upgrade.\n\n* Thu Aug 22 2002 Lamar Owen <lamar.owen@ramifordistat.net>\n- 7.2.2 security fix.\n- deprecate postgresql-dump system.\n- 7.2.2-0.1PGDG PRERELEASE!\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Mon, 26 Aug 2002 14:25:27 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": true, "msg_subject": "RPMs for release 7.2.2" } ]
[ { "msg_contents": "Joe Conway and I have just committed some changes in the internal\nrepresentation of Postgres arrays: an element-type-OID field is added to\nthe array header, and alignment calculations are now done the same way\nas in ordinary tuple storage, instead of taking shortcuts. I believe\nthat these changes need to be reflected into the intarray, ltree, and\nintagg contrib modules.\n\nintarray and ltree both seem to be mapping their own declarations onto\narrays using largely-similar code. But while intarray fails its\nregression test, I find ltree still passes. So I'm confused about what\nthat code is really doing and don't want to touch it.\n\nI tried to fix intagg, but since there is no regression test for it\nI'm unsure whether it's okay.\n\nCould you folks take a look at CVS tip and see what changes are needed,\nif any?\n\nIn the longer run, it might be possible to improve these routines to be\narray-type-polymorphic using the new features. But with the 7.3 beta\ndate nearly upon us, I'd counsel first making the existing functionality\nwork again...\n\n\t\t\tregards, tom lane\n", "msg_date": "Mon, 26 Aug 2002 14:37:47 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": true, "msg_subject": "contrib/ intarray, ltree, intagg broken(?) by array changes" }, { "msg_contents": "\nCan someone address the intagg issue here, or is the code OK?\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> Joe Conway and I have just committed some changes in the internal\n> representation of Postgres arrays: an element-type-OID field is added to\n> the array header, and alignment calculations are now done the same way\n> as in ordinary tuple storage, instead of taking shortcuts. I believe\n> that these changes need to be reflected into the intarray, ltree, and\n> intagg contrib modules.\n> \n> intarray and ltree both seem to be mapping their own declarations onto\n> arrays using largely-similar code. But while intarray fails its\n> regression test, I find ltree still passes. So I'm confused about what\n> that code is really doing and don't want to touch it.\n> \n> I tried to fix intagg, but since there is no regression test for it\n> I'm unsure whether it's okay.\n> \n> Could you folks take a look at CVS tip and see what changes are needed,\n> if any?\n> \n> In the longer run, it might be possible to improve these routines to be\n> array-type-polymorphic using the new features. But with the 7.3 beta\n> date nearly upon us, I'd counsel first making the existing functionality\n> work again...\n> \n> \t\t\tregards, tom lane\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 Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sun, 1 Sep 2002 22:51:09 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: contrib/ intarray, ltree, intagg broken(?) by array changes" }, { "msg_contents": "This built and worked on my system.\nfamous last words, huh?\n\n\nBruce Momjian wrote:\n> \n> Can someone address the intagg issue here, or is the code OK?\n> \n> ---------------------------------------------------------------------------\n> \n> Tom Lane wrote:\n> > Joe Conway and I have just committed some changes in the internal\n> > representation of Postgres arrays: an element-type-OID field is added to\n> > the array header, and alignment calculations are now done the same way\n> > as in ordinary tuple storage, instead of taking shortcuts. I believe\n> > that these changes need to be reflected into the intarray, ltree, and\n> > intagg contrib modules.\n> >\n> > intarray and ltree both seem to be mapping their own declarations onto\n> > arrays using largely-similar code. But while intarray fails its\n> > regression test, I find ltree still passes. So I'm confused about what\n> > that code is really doing and don't want to touch it.\n> >\n> > I tried to fix intagg, but since there is no regression test for it\n> > I'm unsure whether it's okay.\n> >\n> > Could you folks take a look at CVS tip and see what changes are needed,\n> > if any?\n> >\n> > In the longer run, it might be possible to improve these routines to be\n> > array-type-polymorphic using the new features. But with the 7.3 beta\n> > date nearly upon us, I'd counsel first making the existing functionality\n> > work again...\n> >\n> > regards, tom lane\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> Bruce Momjian | http://candle.pha.pa.us\n> pgman@candle.pha.pa.us | (610) 359-1001\n> + If your life is a hard drive, | 13 Roberts Road\n> + Christ can be your backup. | Newtown Square, Pennsylvania 19073", "msg_date": "Sun, 01 Sep 2002 22:54:44 -0400", "msg_from": "mlw <developer@mohawksoft.com>", "msg_from_op": false, "msg_subject": "Re: contrib/ intarray, ltree, intagg broken(?) by array " }, { "msg_contents": "\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://207.106.42.251/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nmlw wrote:\n> This built and worked on my system.\n> famous last words, huh?\n> \n> \n> Bruce Momjian wrote:\n> > \n> > Can someone address the intagg issue here, or is the code OK?\n> > \n> > ---------------------------------------------------------------------------\n> > \n> > Tom Lane wrote:\n> > > Joe Conway and I have just committed some changes in the internal\n> > > representation of Postgres arrays: an element-type-OID field is added to\n> > > the array header, and alignment calculations are now done the same way\n> > > as in ordinary tuple storage, instead of taking shortcuts. I believe\n> > > that these changes need to be reflected into the intarray, ltree, and\n> > > intagg contrib modules.\n> > >\n> > > intarray and ltree both seem to be mapping their own declarations onto\n> > > arrays using largely-similar code. But while intarray fails its\n> > > regression test, I find ltree still passes. So I'm confused about what\n> > > that code is really doing and don't want to touch it.\n> > >\n> > > I tried to fix intagg, but since there is no regression test for it\n> > > I'm unsure whether it's okay.\n> > >\n> > > Could you folks take a look at CVS tip and see what changes are needed,\n> > > if any?\n> > >\n> > > In the longer run, it might be possible to improve these routines to be\n> > > array-type-polymorphic using the new features. But with the 7.3 beta\n> > > date nearly upon us, I'd counsel first making the existing functionality\n> > > work again...\n> > >\n> > > regards, tom lane\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> > Bruce Momjian | http://candle.pha.pa.us\n> > pgman@candle.pha.pa.us | (610) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n\n> ? int_aggregate.sql\n> ? intagg.patch\n> ? intagg_test.sql\n> Index: int_aggregate.c\n> ===================================================================\n> RCS file: /projects/cvsroot/pgsql-server/contrib/intagg/int_aggregate.c,v\n> retrieving revision 1.4\n> diff -u -r1.4 int_aggregate.c\n> --- int_aggregate.c\t2002/08/30 00:28:40\t1.4\n> +++ int_aggregate.c\t2002/08/30 15:22:03\n> @@ -11,7 +11,8 @@\n> * This file is the property of the Digital Music Network (DMN).\n> * It is being made available to users of the PostgreSQL system\n> * under the BSD license.\n> - *\n> + * \n> + * NOTE: This module requires sizeof(void *) to be the same as sizeof(int)\n> */\n> #include \"postgres.h\"\n> \n> @@ -37,6 +38,9 @@\n> #include \"utils/lsyscache.h\"\n> \n> \n> +/* Uncomment this define if you are compiling for postgres 7.2.x */\n> +/* #define PG_7_2 */\n> +\n> /* This is actually a postgres version of a one dimensional array */\n> \n> typedef struct\n> @@ -96,7 +100,9 @@\n> \t\tp->a.size = cb;\n> \t\tp->a.ndim = 0;\n> \t\tp->a.flags = 0;\n> +#ifndef PG_7_2\n> \t\tp->a.elemtype = INT4OID;\n> +#endif\n> \t\tp->items = 0;\n> \t\tp->lower= START_NUM;\n> \t}\n> @@ -149,7 +155,9 @@\n> \t\t\tpnew->a.size = cb;\n> \t\t\tpnew->a.ndim=1;\n> \t\t\tpnew->a.flags = 0;\n> +#ifndef PG_7_2\n> \t\t\tpnew->a.elemtype = INT4OID;\n> +#endif\n> \t\t\tpnew->lower = 0;\n> \t\t}\n> \t\telse\n> Index: int_aggregate.sql.in\n> ===================================================================\n> RCS file: /projects/cvsroot/pgsql-server/contrib/intagg/int_aggregate.sql.in,v\n> retrieving revision 1.1\n> diff -u -r1.1 int_aggregate.sql.in\n> --- int_aggregate.sql.in\t2002/02/25 03:45:27\t1.1\n> +++ int_aggregate.sql.in\t2002/08/30 15:22:03\n> @@ -1,7 +1,7 @@\n> -- Drop functions\n> +drop aggregate int_array_aggregate(int4);\n> drop function int_agg_state (int4, int4);\n> drop function int_agg_final_array (int4);\n> -drop aggregate int_array_aggregate(int4);\n> drop function int_array_enum (int4[]);\n> \n> \n> @@ -9,14 +9,14 @@\n> -- Is called for each item in an aggregation\n> create function int_agg_state (int4, int4)\n> \treturns int4\n> -\tas 'MODULE_FILENAME','int_agg_state'\n> +\tas 'MODULE_PATHNAME','int_agg_state'\n> \tlanguage 'c';\n> \n> -- Internal function for the aggregate\n> -- Is called at the end of the aggregation, and returns an array.\n> create function int_agg_final_array (int4)\n> \treturns int4[]\n> -\tas 'MODULE_FILENAME','int_agg_final_array'\n> +\tas 'MODULE_PATHNAME','int_agg_final_array'\n> \tlanguage 'c';\n> \n> -- The aggration funcion.\n> @@ -35,6 +35,6 @@\n> -- as a row.\n> create function int_array_enum(int4[])\n> \treturns setof integer\n> -\tas 'MODULE_FILENAME','int_enum'\n> +\tas 'MODULE_PATHNAME','int_enum'\n> \tlanguage 'c';\n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Sun, 1 Sep 2002 23:37:32 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: contrib/ intarray, ltree, intagg broken(?) by array changes" }, { "msg_contents": "\nPatch applied. Thanks.\n\n---------------------------------------------------------------------------\n\n\nmlw wrote:\n> This built and worked on my system.\n> famous last words, huh?\n> \n> \n> Bruce Momjian wrote:\n> > \n> > Can someone address the intagg issue here, or is the code OK?\n> > \n> > ---------------------------------------------------------------------------\n> > \n> > Tom Lane wrote:\n> > > Joe Conway and I have just committed some changes in the internal\n> > > representation of Postgres arrays: an element-type-OID field is added to\n> > > the array header, and alignment calculations are now done the same way\n> > > as in ordinary tuple storage, instead of taking shortcuts. I believe\n> > > that these changes need to be reflected into the intarray, ltree, and\n> > > intagg contrib modules.\n> > >\n> > > intarray and ltree both seem to be mapping their own declarations onto\n> > > arrays using largely-similar code. But while intarray fails its\n> > > regression test, I find ltree still passes. So I'm confused about what\n> > > that code is really doing and don't want to touch it.\n> > >\n> > > I tried to fix intagg, but since there is no regression test for it\n> > > I'm unsure whether it's okay.\n> > >\n> > > Could you folks take a look at CVS tip and see what changes are needed,\n> > > if any?\n> > >\n> > > In the longer run, it might be possible to improve these routines to be\n> > > array-type-polymorphic using the new features. But with the 7.3 beta\n> > > date nearly upon us, I'd counsel first making the existing functionality\n> > > work again...\n> > >\n> > > regards, tom lane\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> > Bruce Momjian | http://candle.pha.pa.us\n> > pgman@candle.pha.pa.us | (610) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n\n> ? int_aggregate.sql\n> ? intagg.patch\n> ? intagg_test.sql\n> Index: int_aggregate.c\n> ===================================================================\n> RCS file: /projects/cvsroot/pgsql-server/contrib/intagg/int_aggregate.c,v\n> retrieving revision 1.4\n> diff -u -r1.4 int_aggregate.c\n> --- int_aggregate.c\t2002/08/30 00:28:40\t1.4\n> +++ int_aggregate.c\t2002/08/30 15:22:03\n> @@ -11,7 +11,8 @@\n> * This file is the property of the Digital Music Network (DMN).\n> * It is being made available to users of the PostgreSQL system\n> * under the BSD license.\n> - *\n> + * \n> + * NOTE: This module requires sizeof(void *) to be the same as sizeof(int)\n> */\n> #include \"postgres.h\"\n> \n> @@ -37,6 +38,9 @@\n> #include \"utils/lsyscache.h\"\n> \n> \n> +/* Uncomment this define if you are compiling for postgres 7.2.x */\n> +/* #define PG_7_2 */\n> +\n> /* This is actually a postgres version of a one dimensional array */\n> \n> typedef struct\n> @@ -96,7 +100,9 @@\n> \t\tp->a.size = cb;\n> \t\tp->a.ndim = 0;\n> \t\tp->a.flags = 0;\n> +#ifndef PG_7_2\n> \t\tp->a.elemtype = INT4OID;\n> +#endif\n> \t\tp->items = 0;\n> \t\tp->lower= START_NUM;\n> \t}\n> @@ -149,7 +155,9 @@\n> \t\t\tpnew->a.size = cb;\n> \t\t\tpnew->a.ndim=1;\n> \t\t\tpnew->a.flags = 0;\n> +#ifndef PG_7_2\n> \t\t\tpnew->a.elemtype = INT4OID;\n> +#endif\n> \t\t\tpnew->lower = 0;\n> \t\t}\n> \t\telse\n> Index: int_aggregate.sql.in\n> ===================================================================\n> RCS file: /projects/cvsroot/pgsql-server/contrib/intagg/int_aggregate.sql.in,v\n> retrieving revision 1.1\n> diff -u -r1.1 int_aggregate.sql.in\n> --- int_aggregate.sql.in\t2002/02/25 03:45:27\t1.1\n> +++ int_aggregate.sql.in\t2002/08/30 15:22:03\n> @@ -1,7 +1,7 @@\n> -- Drop functions\n> +drop aggregate int_array_aggregate(int4);\n> drop function int_agg_state (int4, int4);\n> drop function int_agg_final_array (int4);\n> -drop aggregate int_array_aggregate(int4);\n> drop function int_array_enum (int4[]);\n> \n> \n> @@ -9,14 +9,14 @@\n> -- Is called for each item in an aggregation\n> create function int_agg_state (int4, int4)\n> \treturns int4\n> -\tas 'MODULE_FILENAME','int_agg_state'\n> +\tas 'MODULE_PATHNAME','int_agg_state'\n> \tlanguage 'c';\n> \n> -- Internal function for the aggregate\n> -- Is called at the end of the aggregation, and returns an array.\n> create function int_agg_final_array (int4)\n> \treturns int4[]\n> -\tas 'MODULE_FILENAME','int_agg_final_array'\n> +\tas 'MODULE_PATHNAME','int_agg_final_array'\n> \tlanguage 'c';\n> \n> -- The aggration funcion.\n> @@ -35,6 +35,6 @@\n> -- as a row.\n> create function int_array_enum(int4[])\n> \treturns setof integer\n> -\tas 'MODULE_FILENAME','int_enum'\n> +\tas 'MODULE_PATHNAME','int_enum'\n> \tlanguage 'c';\n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Mon, 2 Sep 2002 02:27:49 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: contrib/ intarray, ltree, intagg broken(?) by array changes" }, { "msg_contents": "We'll be back to work next week and look into code.\n\n\tOleg\nOn Sun, 1 Sep 2002, Bruce Momjian wrote:\n\n>\n> Can someone address the intagg issue here, or is the code OK?\n>\n> ---------------------------------------------------------------------------\n>\n> Tom Lane wrote:\n> > Joe Conway and I have just committed some changes in the internal\n> > representation of Postgres arrays: an element-type-OID field is added to\n> > the array header, and alignment calculations are now done the same way\n> > as in ordinary tuple storage, instead of taking shortcuts. I believe\n> > that these changes need to be reflected into the intarray, ltree, and\n> > intagg contrib modules.\n> >\n> > intarray and ltree both seem to be mapping their own declarations onto\n> > arrays using largely-similar code. But while intarray fails its\n> > regression test, I find ltree still passes. So I'm confused about what\n> > that code is really doing and don't want to touch it.\n> >\n> > I tried to fix intagg, but since there is no regression test for it\n> > I'm unsure whether it's okay.\n> >\n> > Could you folks take a look at CVS tip and see what changes are needed,\n> > if any?\n> >\n> > In the longer run, it might be possible to improve these routines to be\n> > array-type-polymorphic using the new features. But with the 7.3 beta\n> > date nearly upon us, I'd counsel first making the existing functionality\n> > work again...\n> >\n> > \t\t\tregards, tom lane\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\tRegards,\n\t\tOleg\n_____________________________________________________________\nOleg Bartunov, sci.researcher, hostmaster of AstroNet,\nSternberg Astronomical Institute, Moscow University (Russia)\nInternet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/\nphone: +007(095)939-16-83, +007(095)939-23-83\n\n", "msg_date": "Mon, 2 Sep 2002 12:26:02 +0300 (GMT)", "msg_from": "Oleg Bartunov <oleg@sai.msu.su>", "msg_from_op": false, "msg_subject": "Re: contrib/ intarray, ltree, intagg broken(?) by array" }, { "msg_contents": "\nI have already received a patch that modifies the regression tests and\nit seems things are working. However, it would be good for you to take\na look. Fortunately, that can be done anytime during beta. Thanks.\n\n---------------------------------------------------------------------------\n\nOleg Bartunov wrote:\n> We'll be back to work next week and look into code.\n> \n> \tOleg\n> On Sun, 1 Sep 2002, Bruce Momjian wrote:\n> \n> >\n> > Can someone address the intagg issue here, or is the code OK?\n> >\n> > ---------------------------------------------------------------------------\n> >\n> > Tom Lane wrote:\n> > > Joe Conway and I have just committed some changes in the internal\n> > > representation of Postgres arrays: an element-type-OID field is added to\n> > > the array header, and alignment calculations are now done the same way\n> > > as in ordinary tuple storage, instead of taking shortcuts. I believe\n> > > that these changes need to be reflected into the intarray, ltree, and\n> > > intagg contrib modules.\n> > >\n> > > intarray and ltree both seem to be mapping their own declarations onto\n> > > arrays using largely-similar code. But while intarray fails its\n> > > regression test, I find ltree still passes. So I'm confused about what\n> > > that code is really doing and don't want to touch it.\n> > >\n> > > I tried to fix intagg, but since there is no regression test for it\n> > > I'm unsure whether it's okay.\n> > >\n> > > Could you folks take a look at CVS tip and see what changes are needed,\n> > > if any?\n> > >\n> > > In the longer run, it might be possible to improve these routines to be\n> > > array-type-polymorphic using the new features. But with the 7.3 beta\n> > > date nearly upon us, I'd counsel first making the existing functionality\n> > > work again...\n> > >\n> > > \t\t\tregards, tom lane\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> \tRegards,\n> \t\tOleg\n> _____________________________________________________________\n> Oleg Bartunov, sci.researcher, hostmaster of AstroNet,\n> Sternberg Astronomical Institute, Moscow University (Russia)\n> Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/\n> phone: +007(095)939-16-83, +007(095)939-23-83\n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Mon, 2 Sep 2002 09:37:26 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: contrib/ intarray, ltree, intagg broken(?) by array changes" }, { "msg_contents": "> intarray and ltree both seem to be mapping their own declarations onto\n> arrays using largely-similar code. But while intarray fails its\n> regression test, I find ltree still passes. So I'm confused about what\n> that code is really doing and don't want to touch it.\n\nPlease, apply attached patch, it solves the problem.\n\n\n-- \nTeodor Sigaev\nteodor@stack.net", "msg_date": "Mon, 09 Sep 2002 15:27:38 +0400", "msg_from": "Teodor Sigaev <teodor@stack.net>", "msg_from_op": false, "msg_subject": "Re: contrib/ intarray, ltree, intagg broken(?) by array" }, { "msg_contents": "\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://candle.pha.pa.us/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nTeodor Sigaev wrote:\n> \n> \n> > intarray and ltree both seem to be mapping their own declarations onto\n> > arrays using largely-similar code. But while intarray fails its\n> > regression test, I find ltree still passes. So I'm confused about what\n> > that code is really doing and don't want to touch it.\n> \n> Please, apply attached patch, it solves the problem.\n> \n> \n> -- \n> Teodor Sigaev\n> teodor@stack.net\n> \n\n[ application/gzip is not supported, skipping... ]\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 11 Sep 2002 00:13:08 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: contrib/ intarray, ltree, intagg broken(?) by array changes" }, { "msg_contents": "\nPatch applied. Thanks.\n\n---------------------------------------------------------------------------\n\n\nTeodor Sigaev wrote:\n> \n> \n> > intarray and ltree both seem to be mapping their own declarations onto\n> > arrays using largely-similar code. But while intarray fails its\n> > regression test, I find ltree still passes. So I'm confused about what\n> > that code is really doing and don't want to touch it.\n> \n> Please, apply attached patch, it solves the problem.\n> \n> \n> -- \n> Teodor Sigaev\n> teodor@stack.net\n> \n\n[ application/gzip is not supported, skipping... ]\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 11 Sep 2002 20:24:50 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: contrib/ intarray, ltree, intagg broken(?) by array changes" } ]
[ { "msg_contents": "> -----Original Message-----\n> From: Lamar Owen [mailto:lamar.owen@wgcr.org] \n> Sent: Monday, August 26, 2002 11:41 AM\n> To: Dann Corbit; Bruce Momjian; Tom Lane\n> Cc: Sir Mordred The Traitor; pgsql-hackers@postgresql.org\n> Subject: Re: [HACKERS] @(#)Mordred Labs advisory 0x0007: \n> Remove DoS in PostgreSQL\n[snip]\n> And, yes, I agree he IS providing a valuable service -- with \n> that I have no \n> complaints. But there is a distinct civility and culture to \n> this list, and \n> I'd like to see it stay that way.\n\nI want to make it clear that I think the best way to report a problem is\nwith formal, rigorous, complete structure. That structure should be\nknown to the receiving body. If there is a procedure or standard form\nfor producing the needed information, it is better to follow the\nstandard procedure.\n\nOn the other hand (and the point I had hoped to make) some people like\nto take lined paper and write on it sideways [Bradbury's \"Fahrenheight\n451\" quote]. Instead of complaining that they are writing sideways on\nthe page, extract the information of value. If the can be convinced to\n\"stop writing in a perpendicular manner\" so that the majority will be\nmore comfortable with it, so much the better. But standardized or not,\nit can still contain information of great value.\n", "msg_date": "Mon, 26 Aug 2002 11:51:43 -0700", "msg_from": "\"Dann Corbit\" <DCorbit@connx.com>", "msg_from_op": true, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" }, { "msg_contents": "On Monday 26 August 2002 02:51 pm, Dann Corbit wrote:\n> I want to make it clear that I think the best way to report a problem is\n> with formal, rigorous, complete structure. That structure should be\n> known to the receiving body. If there is a procedure or standard form\n> for producing the needed information, it is better to follow the\n> standard procedure.\n\nOk, let's do that then. SOP is to use the bug reporting form contained on the \nwebsite, which posts a bug report to pgsql-bugs@postgresql.org, which is the \ncanonical bugreport list. That is if we want to get that formal.\n\n> But standardized or not,\n> it can still contain information of great value.\n\nYes, it can. I can use steganography to issue an RPM release announcement, \ntoo. But will it be effective? :-)\n\nIf it isn't read, it won't be acted upon. His announcements have been a \ndifficult read, that's all. The substance is OK; the presentation is lacking, \nIMHO.\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n", "msg_date": "Mon, 26 Aug 2002 15:01:23 -0400", "msg_from": "Lamar Owen <lamar.owen@wgcr.org>", "msg_from_op": false, "msg_subject": "Re: @(#)Mordred Labs advisory 0x0007: Remove DoS in PostgreSQL" } ]
[ { "msg_contents": "I have been getting the subject message ever since upgrading to 7.2.1. I \ntried 7.2.2 with the same thing. It seems to be related to my chkpass type \n(see contrib) as it only happens on tables with that type. I tried it on a \nnew database with a very simple table and still see it. After compiling \nchkpass.c and running the SQL to create the type create a table with one \nfield with chkpass type. Add a number of rows, I did 24, then vacuum it. \nYou get something similar to the above. Sometimes you get \"Memory exhausted \nin AllocSetAlloc(929654141)\" instead and once in a while there is no error.\n\nGiven table x with field c as chkpass run \"UPDATE x SET c = ':a';\" on it. \nThis never fails. Now try UPDATE x SET c = 'a';\" and let chkpass crypt the \nvalue. This usually fails with one of the above messages. The number is \nconstant until you run the UPDATE again.\n\nSomehow the value of the password string is polluting the size storage. I \nknow this because every time this happens, the first 2nd, 3rd and 4th bytes \n(after adjusting for endianness) of the integer are the 6th, 7th and 8th \ncharacters of the encrypted password.\n\nI have another type which is built like this except that it is an indexable \ntype and that doesn't seem to have any problem. It is constructed the same \nway otherwise. The palloc calls appear to be correct. Can anyone see why \nthis would suddenly be a problem?\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": "Mon, 26 Aug 2002 18:04:04 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "MemoryContextAlloc: invalid request size 1934906735" }, { "msg_contents": "\"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> I have been getting the subject message ever since upgrading to 7.2.1. I \n> tried 7.2.2 with the same thing. It seems to be related to my chkpass type \n> (see contrib) as it only happens on tables with that type.\n\nFWIW, I couldn't see any problem in CVS tip. Could you provide an exact\nsequence-to-reproduce?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 02:01:01 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735 " }, { "msg_contents": "On August 27, 2002 02:01 am, Tom Lane wrote:\n> \"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> > I have been getting the subject message ever since upgrading to 7.2.1. I\n> > tried 7.2.2 with the same thing. It seems to be related to my chkpass\n> > type (see contrib) as it only happens on tables with that type.\n>\n> FWIW, I couldn't see any problem in CVS tip. Could you provide an exact\n> sequence-to-reproduce?\n\nSurely. Create a database (chkpass_test) and, after loading the chkpass \ntype, follow this bouncing ball.\n\n1. CREATE TABLE x (c chkpass);\n2. INSERT INTO x VALUES ('a'); [Repeat 20 times]\n3. VACUUM ANALYZE x;\n4. UPDATE x SET c = ':a';\n5. VACUUM ANALYZE x;\n6. UPDATE x SET c = 'a';\n7. VACUUM ANALYZE x;\n8. GOTO 4\n\nNote that 3 and 7 should fail most (95%?) of the time. When it does, convert \nthe invalid size to hex and compare the bytes of the integer with the \nencrypted value in the table.\n\nNote that the fact that 5 does not fail has nothing to do with the path \nthrough chkpass.c. If you take a failing password and insert it raw with \n\"UPDATE x SET c = ':2w3dhratDt7xo';\" then you get the same failure (even the \nsame number - 1952543340 in my case) as you did when chkpass encrypted with \nthat salt.\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, 27 Aug 2002 07:31:31 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735" }, { "msg_contents": "\"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> On August 27, 2002 02:01 am, Tom Lane wrote:\n>> FWIW, I couldn't see any problem in CVS tip. Could you provide an exact\n>> sequence-to-reproduce?\n\n> Surely. Create a database (chkpass_test) and, after loading the chkpass \n> type, follow this bouncing ball.\n> ...\n> Note that 3 and 7 should fail most (95%?) of the time.\n\nZero failures in a dozen iterations here. Anyone else see it?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 09:13:49 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735 " }, { "msg_contents": "On August 27, 2002 09:13 am, Tom Lane wrote:\n> \"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> > On August 27, 2002 02:01 am, Tom Lane wrote:\n> >> FWIW, I couldn't see any problem in CVS tip. Could you provide an exact\n> >> sequence-to-reproduce?\n> >\n> > Surely. Create a database (chkpass_test) and, after loading the chkpass\n> > type, follow this bouncing ball.\n> > ...\n> > Note that 3 and 7 should fail most (95%?) of the time.\n>\n> Zero failures in a dozen iterations here. Anyone else see it?\n\nNetBSD issue? It does it on all the NetBSD systems I tried it on. I will be \nputting 7.2.2 on AIX shortly. I can try again. Can you think of any OS \nissue here?\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, 27 Aug 2002 09:59:37 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735" }, { "msg_contents": "\"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> On August 27, 2002 09:13 am, Tom Lane wrote:\n>> Zero failures in a dozen iterations here. Anyone else see it?\n\n> NetBSD issue? It does it on all the NetBSD systems I tried it on.\n\nHm. The first thing I thought was \"portability problem\" --- I had been\ntesting on HPUX. But I just tried it on a Linux Intel box and see no\nfailure there either.\n\n> I will be putting 7.2.2 on AIX shortly.\n\nNote I'm testing CVS tip, not 7.2. Could we have fixed whatever the bug\nis? Seems unlikely, if it just appeared between 7.2 and 7.2.1 for you.\n\nAnother possible reason for differences is configuration. I used a\npretty plain-vanilla configure:\n\n./configure --with-CXX --with-tcl --enable-cassert --enable-debug\n\nHow about you?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 15:39:00 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735 " }, { "msg_contents": "On August 27, 2002 03:39 pm, Tom Lane wrote:\n> \"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> > NetBSD issue? It does it on all the NetBSD systems I tried it on.\n>\n> Hm. The first thing I thought was \"portability problem\" --- I had been\n> testing on HPUX. But I just tried it on a Linux Intel box and see no\n> failure there either.\n\nWell, Linux != NetBSD except for processor.\n\n> Note I'm testing CVS tip, not 7.2. Could we have fixed whatever the bug\n> is? Seems unlikely, if it just appeared between 7.2 and 7.2.1 for you.\n\nActually it was between 7.1.2 and 7.2.1. I upgraded to 7.2.2 to see if it \nwould go away. I can try tip on another box.\n\n> Another possible reason for differences is configuration. I used a\n> pretty plain-vanilla configure:\n>\n> ./configure --with-CXX --with-tcl --enable-cassert --enable-debug\n>\n> How about you?\n\nI am using the NetBSD pkgsrc system. It generates this configuration.\nHow about that --enable-multibyte?\n\n--enable-multibyte --disable-odbc --without-java --without-perl \n--without-python --without-tcl --without-tk \n--includedir=/usr/pkg/include/pgsql \n--with-htmldir=/usr/pkg/share/doc/html/postgresql --disable-readline \n--enable-locale --enable-syslog --with-CXX --with-template=netbsd \n--with-openssl=/usr --host=i386--netbsdelf --prefix=/usr/pkg\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, 27 Aug 2002 18:05:29 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735" }, { "msg_contents": "\"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> How about that --enable-multibyte?\n\n--enable-multibyte is default (indeed only) option on CVS tip,\nso that's not it. Ditto locale.\n\nCould you try CVS tip on one of the boxes where you see the failure?\nThat'd help to narrow down the issue.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 18:20:29 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735 " }, { "msg_contents": "On August 27, 2002 06:20 pm, Tom Lane wrote:\n> \"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> > How about that --enable-multibyte?\n>\n> --enable-multibyte is default (indeed only) option on CVS tip,\n> so that's not it. Ditto locale.\n>\n> Could you try CVS tip on one of the boxes where you see the failure?\n> That'd help to narrow down the issue.\n\nSame issue. It must be a NetBSD issue but I can't think what. Could there \nbe some data size issues where NetBSD is different than Linux, etc? And why \nonly on the chkpass type? I have other user defined types and they work fine.\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 425 1212 (DoD#0082) (eNTP) | what's for dinner.\n", "msg_date": "Wed, 28 Aug 2002 04:05:21 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735" }, { "msg_contents": "\"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n>> Could you try CVS tip on one of the boxes where you see the failure?\n>> That'd help to narrow down the issue.\n\n> Same issue. It must be a NetBSD issue but I can't think what.\n\nThe behavior looks a lot like a memory clobber, so perhaps the key\nvariable is some difference in malloc's allocation strategy, causing\ntwo items to be adjacent in NetBSD where they are not on the other\nplatforms we've tried.\n\nI eyeballed the chkpass code and didn't see any sign of buffer overruns,\nbut maybe it needs a harder look.\n\nHm --- I guess another possible variable is behavior of the local\ncrypt() function. Is NetBSD's crypt perhaps willing to return strings\nlonger than 13 chars?\n\nDid you try CVS tip both with and without --enable-cassert? That turns\non memory context checking which might alter the failure in interesting\nways.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Aug 2002 09:23:51 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735 " }, { "msg_contents": "On August 28, 2002 09:23 am, Tom Lane wrote:\n\n> The behavior looks a lot like a memory clobber, so perhaps the key\n> variable is some difference in malloc's allocation strategy, causing\n> two items to be adjacent in NetBSD where they are not on the other\n> platforms we've tried.\n\nHmm. I might try adding some buffer in MemoryContextAlloc() and see if that \nchanges anything. One thing that may be different is that NetBSD is 64 bit \nclean. I don't think the other i386 systems are.\n\n> I eyeballed the chkpass code and didn't see any sign of buffer overruns,\n> but maybe it needs a harder look.\n\nIt's pretty simple. Not even indexing. In fact, I wondered if I should add \nsome just like my other type that does do indexing. That seemed to be the \nonly real difference between the two and the other works.\n\n> Hm --- I guess another possible variable is behavior of the local\n> crypt() function. Is NetBSD's crypt perhaps willing to return strings\n> longer than 13 chars?\n\nWell, the value that it stores is the correct 13 character DES string.\n\n> Did you try CVS tip both with and without --enable-cassert? That turns\n> on memory context checking which might alter the failure in interesting\n> ways.\n\nNo difference. It seems that PostgreSQL is too good at catching the problem \nbefore the assert macros see it.\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 425 1212 (DoD#0082) (eNTP) | what's for dinner.\n", "msg_date": "Wed, 28 Aug 2002 12:48:19 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735" }, { "msg_contents": "On August 28, 2002 12:48 pm, D'Arcy J.M. Cain wrote:\n> On August 28, 2002 09:23 am, Tom Lane wrote:\n> > The behavior looks a lot like a memory clobber, so perhaps the key\n> > variable is some difference in malloc's allocation strategy, causing\n> > two items to be adjacent in NetBSD where they are not on the other\n> > platforms we've tried.\n>\n> Hmm. I might try adding some buffer in MemoryContextAlloc() and see if\n\nNope. Tried adding \"size += 64;\" into MemoryContextAlloc() and it made no \ndifference. \n\nI had also tried changing palloc.h and mcxt.c to turn MemoryContextAlloc() \ninto a macro that called a modified version of the real one to try to narrow \ndown where it was being called from but that wouldn't even run. Is there \nanother file I have to modify as well if I try that? Here are the changes I \ntried. See any reason that it shouldn't have worked if I rebuilt everything?\n\n*** postgresql-server/work.i386/postgresql-7.2.2/src/backend/utils/mmgr/mcxt.c\nWed Aug 28 14:01:31 2002\n--- mcxt.c Mon Aug 26 21:51:31 2002\n***************\n*** 409,416 ****\n * nodes/memnodes.h into postgres.h which seems a bad idea.\n */\n void *\n! MemoryContextAlloc(MemoryContext context, Size size)\n {\n AssertArg(MemoryContextIsValid(context));\n\n if (!AllocSizeIsValid(size))\n--- 409,417 ----\n * nodes/memnodes.h into postgres.h which seems a bad idea.\n */\n void *\n! _MemoryContextAlloc(MemoryContext context, Size size, const char *f, int l)\n {\n+ elog(NOTICE, \"_MemoryContextAlloc called from %s line %d\", f, l);\n AssertArg(MemoryContextIsValid(context));\n\n if (!AllocSizeIsValid(size))\n\n*** postgresql-server/work.i386/postgresql-7.2.2/src/include/utils/palloc.h \nMon\nNov 5 12:46:36 2001\n--- palloc.h Mon Aug 26 21:51:46 2002\n***************\n*** 45,51 ****\n /*\n * Fundamental memory-allocation operations (more are in utils/memutils.h)\n */\n! extern void *MemoryContextAlloc(MemoryContext context, Size size);\n\n #define palloc(sz) MemoryContextAlloc(CurrentMemoryContext, (sz))\n\n--- 45,52 ----\n /*\n * Fundamental memory-allocation operations (more are in utils/memutils.h)\n */\n! extern void *_MemoryContextAlloc(MemoryContext context, Size size, const \nchar\n*s, int l);\n! #define MemoryContextAlloc(context, size) _MemoryContextAlloc(context, \nsize, __FILE__, __LINE__)\n\n #define palloc(sz) MemoryContextAlloc(CurrentMemoryContext, (sz))\n\n\n> that changes anything. One thing that may be different is that NetBSD is\n> 64 bit clean. I don't think the other i386 systems are.\n>\n> > I eyeballed the chkpass code and didn't see any sign of buffer overruns,\n> > but maybe it needs a harder look.\n>\n> It's pretty simple. Not even indexing. In fact, I wondered if I should\n> add some just like my other type that does do indexing. That seemed to be\n> the only real difference between the two and the other works.\n>\n> > Hm --- I guess another possible variable is behavior of the local\n> > crypt() function. Is NetBSD's crypt perhaps willing to return strings\n> > longer than 13 chars?\n>\n> Well, the value that it stores is the correct 13 character DES string.\n>\n> > Did you try CVS tip both with and without --enable-cassert? That turns\n> > on memory context checking which might alter the failure in interesting\n> > ways.\n>\n> No difference. It seems that PostgreSQL is too good at catching the\n> problem before the assert macros see it.\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 425 1212 (DoD#0082) (eNTP) | what's for dinner.\n", "msg_date": "Wed, 28 Aug 2002 14:07:34 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735" }, { "msg_contents": "On August 28, 2002 02:07 pm, D'Arcy J.M. Cain wrote:\n> I had also tried changing palloc.h and mcxt.c to turn MemoryContextAlloc()\n> into a macro that called a modified version of the real one to try to\n> narrow down where it was being called from but that wouldn't even run. Is\n> there another file I have to modify as well if I try that? Here are the\n> changes I tried. See any reason that it shouldn't have worked if I rebuilt\n> everything?\n\nI forgot to mention the error when I do this. The server runs but when I try \nto connect to a database I get this:\n\npsql: could not receive server response to SSL negotiation packet: \nInappropriate ioctl for device\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 425 1212 (DoD#0082) (eNTP) | what's for dinner.\n", "msg_date": "Wed, 28 Aug 2002 14:48:28 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735" }, { "msg_contents": "On August 28, 2002 09:23 am, Tom Lane wrote:\n> The behavior looks a lot like a memory clobber, so perhaps the key\n> variable is some difference in malloc's allocation strategy, causing\n> two items to be adjacent in NetBSD where they are not on the other\n> platforms we've tried.\n\nHere's some other wackiness. The following is various encrypted passwords \nsplit for convenience of analyzing them along with the hex version of the big \nnumber in the error. The OK ones were ones that didn't trigger the error. \nThe first line following that is the last byte in the number in hex and \nbinary. The next is the first character of the second chunk of the password. \n Note how the integer basically is the 4 bytes of the second chunk except for \nthe first byte which differs in a somewhat regular way. Looks like some sort \nof bit mask operation somewhere.\n\nwvx8 42kQ 34jyY (OK)\n0x34 00110100\n\nQ/Jz mdRb HSwE. 0x62526471\n0x71 01110001\n0x6d 01101101\n\nccIx mriB VsviU 0x42697271\n0x71 01110001\n0x6d 01101101\n\nwsnr TAub uIelw 0x62754158\n0x58 01011000\n0x54 01010100\n\ntGep W3d5 EX5pU 0x3564335b\n0x5b 01011011\n0x57 01010111\n\ngJTk uYzh fb3LM 0x687a5979\n0x79 01111001\n0x75 01110101\n\nEfFt qWDL RgVjY 0x4c445775\n0x75 01110101\n0x71 01110001\n\nMy2J GCTv 8A3GI 0x7654434b\n0x4b 01001011\n0x47 01000111\n\nuWPk 7xcQ ZpTi. 0x5163783b\n0x3b 00111011\n0x37 00110111\n\nAlD5 naNP oDKdc (OK)\n0x6e 01101110\n\n>\n> I eyeballed the chkpass code and didn't see any sign of buffer overruns,\n> but maybe it needs a harder look.\n\nHmm. I did give it a harder look and look what jumped out. Both chkpass_out \nand chkpass_rout return PG_RETURN_CSTRING but chkpass_out builds a standard c \nstring while chkpass_rout builds a variable text structure. That can't be \nright. It's odd that this always worked before. It seems to me that \nchkpass_rout should be changed to build a c string like chkpass_out given the \nname of the return macro. I tried that and it made no difference. I'm not \nentirely surprised since I never used the chkpass_rout function in any of the \ntests.\n\nIs it possible that my thinking is wrong and I should be creating a text type \nfor both?\n\nStill doesn't explain why no one else sees this though.\n\nOh, one more datapoint - the error only happens on vacuum analyze, not just \nvacuum. Not sure what that means exactly.\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 425 1212 (DoD#0082) (eNTP) | what's for dinner.\n", "msg_date": "Wed, 28 Aug 2002 22:40:04 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735" }, { "msg_contents": "\"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> Hmm. I did give it a harder look and look what jumped out. Both\n> chkpass_out and chkpass_rout return PG_RETURN_CSTRING but chkpass_out\n> builds a standard c string while chkpass_rout builds a variable text\n> structure. That can't be right.\n\nIt's not, but chkpass_rout is declared to return text, so it should be\nsaying PG_RETURN_TEXT_P. It turns out both macros do the same thing,\nso this is just a cosmetic issue.\n\n> Oh, one more datapoint - the error only happens on vacuum analyze, not just \n> vacuum. Not sure what that means exactly.\n\nThat is odd. You only have the chkpass operators shown in the contrib\nmodule, right? No \"chkpass = chkpass\" operator? Without one, vacuum\nanalyze should pretty much ignore the chkpass column ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Aug 2002 23:07:05 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735 " }, { "msg_contents": "On August 28, 2002 11:07 pm, Tom Lane wrote:\n> \"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> > Hmm. I did give it a harder look and look what jumped out. Both\n> > chkpass_out and chkpass_rout return PG_RETURN_CSTRING but chkpass_out\n> > builds a standard c string while chkpass_rout builds a variable text\n> > structure. That can't be right.\n>\n> It's not, but chkpass_rout is declared to return text, so it should be\n> saying PG_RETURN_TEXT_P. It turns out both macros do the same thing,\n> so this is just a cosmetic issue.\n\nOK, I will make that cosmetic change.\n\n> > Oh, one more datapoint - the error only happens on vacuum analyze, not\n> > just vacuum. Not sure what that means exactly.\n>\n> That is odd. You only have the chkpass operators shown in the contrib\n> module, right? No \"chkpass = chkpass\" operator? Without one, vacuum\n> analyze should pretty much ignore the chkpass column ...\n\nYES! Well, sort of. I didn't have any other operators but while I thought \nthat both were the same (after all, I contributed it) someone must have fixed \nthe one in CVS before adding it. The one I was working with had the \noperators working with chkpass on both sides. As soon as I fixed that it \nworked again.\n\nIn 7.2 the cstring and chkpass types fail in the function definitions because \nthey have not been defined so I had to stay with opaque. In fact, how will \nthat work in 7.3 anyway? We declare the functions to take or return a \nchkpass before we define it.\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": "Thu, 29 Aug 2002 08:16:35 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: MemoryContextAlloc: invalid request size 1934906735" }, { "msg_contents": "\"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> YES! Well, sort of. I didn't have any other operators but while I thought \n> that both were the same (after all, I contributed it) someone must have fixed\n> the one in CVS before adding it. The one I was working with had the \n> operators working with chkpass on both sides. As soon as I fixed that it \n> worked again.\n\nAh-hah, so vacuum was trying to use the \"chkpass = text\" operator to\ncompare two chkpass values. That explains the whole problem --- the\ntext code of course would take the first four bytes of the chkpass\nstring as a length word.\n\n> In 7.2 the cstring and chkpass types fail in the function definitions because\n> they have not been defined so I had to stay with opaque. In fact, how will \n> that work in 7.3 anyway? We declare the functions to take or return a \n> chkpass before we define it.\n\nYeah, you'll get warnings about the type not being defined yet, but it\nwill take them anyway. There's a fundamental circularity involved in\ndefining these things with any sort of accuracy, so we're going to have\nto live with either warnings or kluges :-(.\n\nI suppose that if the warnings really irritate people, we could think\nabout exposing the shell-type-entry mechanism more explicitly. For\nexample, if you did something like\n\n\t-- make a shell pg_type entry\n\tCREATE TYPE chkpass;\n\n\t-- make the I/O functions\n\tCREATE FUNCTION chkpass_in(cstring) RETURNS chkpass ...;\n\n\tCREATE FUNCTION chkpass_out(chkpass) RETURNS cstring ...;\n\n\t-- replace shell entry with real one\n\tCREATE TYPE chkpass(input = chkpass_in, output = ...);\n\nThis looks rather ugly to me but it would be pretty easy to make it\nwork and not give any warnings. Comments?\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 29 Aug 2002 09:45:45 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Type definition process (was Re: MemoryContextAlloc: invalid request\n\tsize 1934906735)" }, { "msg_contents": "On August 29, 2002 09:45 am, Tom Lane wrote:\n> \"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> > YES! Well, sort of. I didn't have any other operators but while I\n> > thought that both were the same (after all, I contributed it) someone\n> > must have fixed the one in CVS before adding it. The one I was working\n> > with had the operators working with chkpass on both sides. As soon as I\n> > fixed that it worked again.\n>\n> Ah-hah, so vacuum was trying to use the \"chkpass = text\" operator to\n> compare two chkpass values. That explains the whole problem --- the\n> text code of course would take the first four bytes of the chkpass\n> string as a length word.\n\nExactly.\n\n> > In 7.2 the cstring and chkpass types fail in the function definitions\n> > because they have not been defined so I had to stay with opaque. In\n> > fact, how will that work in 7.3 anyway? We declare the functions to take\n> > or return a chkpass before we define it.\n>\n> Yeah, you'll get warnings about the type not being defined yet, but it\n> will take them anyway. There's a fundamental circularity involved in\n> defining these things with any sort of accuracy, so we're going to have\n> to live with either warnings or kluges :-(.\n>\n> I suppose that if the warnings really irritate people, we could think\n> about exposing the shell-type-entry mechanism more explicitly. For\n> example, if you did something like\n>\n> \t-- make a shell pg_type entry\n> \tCREATE TYPE chkpass;\n>\n> \t-- make the I/O functions\n> \tCREATE FUNCTION chkpass_in(cstring) RETURNS chkpass ...;\n>\n> \tCREATE FUNCTION chkpass_out(chkpass) RETURNS cstring ...;\n>\n> \t-- replace shell entry with real one\n> \tCREATE TYPE chkpass(input = chkpass_in, output = ...);\n>\n> This looks rather ugly to me but it would be pretty easy to make it\n> work and not give any warnings. Comments?\n\nWell, magic (DWIM) parsing would be nice but this doesn't seem all that ugly \nto me. One thing I do see though is that there is a completion issue. Maybe \nwe should specify that this can only happen within a transaction and add some \ncode to the transaction handling. Some simple rules (not to suggest that \nthey are necessarily simple to implement of course) I see are;\n\n 1. An incomplete CREATE TYPE raises an error if not inside a transaction \nblock.\n 2. CREATE TYPE and CREATE FUNCTION will be backed out on an abort.\n 3. Closing a transaction aborts if an incomplete type has not been completed.\n\nI think that this closes the loop without leaving functions defined on \nincomplete types. With enough clever programming perhaps we can even make \nthe incomplete declarion automatic when it is used in the CREATE FUNCTION. \nWe just don't raise an error until the COMMIT.\n\n BEGIN TRANSACTION\n\n -- an incomplete type \"chkpass\" is conditionally created here\n CREATE FUNCTION chkpass_in(cstring) RETURNS chkpass;\n\n -- the existing conditional type is used here\n CREATE FUNCTION chkpass_out(chkpass) RETURNS cstring;\n\n -- define the actual type\n CREATE TYPE chkpass(input = chkpass_in, output = chkpass_out);\n\n END TRANSACTION\n\nDoes this make sense?\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": "Thu, 29 Aug 2002 15:18:11 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: Type definition process (was Re: MemoryContextAlloc: invalid\n\trequest size 1934906735)" }, { "msg_contents": "\"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> One thing I do see though is that there is a completion issue.\n\nWell, (a) the shell type can't be used for anything till you turn it\ninto a real type, and (b) the completion issue already exists, and has\nfor a long time; you've always been able to create a shell type by using\na not-yet-known type name as the return type of a function. It's just\nnot well documented.\n\n> 1. An incomplete CREATE TYPE raises an error if not inside a transaction \n> block.\n\nI have no intention of implementing this. (1) It wouldn't really\nsimplify life anyway, since we'd still need all the same guard code to\nprevent you from using the shell type within the creating transaction.\n(2) It would break existing pg_dump scripts, which don't know they'd\nneed to do this.\n\nWrapping the sequence inside a transaction is a good practice, but\nI don't feel that we have to try to force good practice on people.\n\n\t\t\tregards, tom lane\n", "msg_date": "Thu, 29 Aug 2002 15:37:29 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Type definition process (was Re: MemoryContextAlloc: invalid\n\trequest size 1934906735)" }, { "msg_contents": "On August 29, 2002 03:37 pm, Tom Lane wrote:\n> \"D'Arcy J.M. Cain\" <darcy@druid.net> writes:\n> > One thing I do see though is that there is a completion issue.\n>\n> Well, (a) the shell type can't be used for anything till you turn it\n> into a real type, and (b) the completion issue already exists, and has\n> for a long time; you've always been able to create a shell type by using\n> a not-yet-known type name as the return type of a function. It's just\n> not well documented.\n\nAnd gives a warning, right?\n\n> > 1. An incomplete CREATE TYPE raises an error if not inside a transaction\n> > block.\n>\n> I have no intention of implementing this. (1) It wouldn't really\n> simplify life anyway, since we'd still need all the same guard code to\n> prevent you from using the shell type within the creating transaction.\n> (2) It would break existing pg_dump scripts, which don't know they'd\n> need to do this.\n\nYes, I see your point.\n\n> Wrapping the sequence inside a transaction is a good practice, but\n> I don't feel that we have to try to force good practice on people.\n\nOK but how about a little reward if they do. Do everything as we do now \nexcept that if they wrap it in a transaction then they don't get the warning \nunless they exit the transaction without completing the type? Some people \n(e.g. me) like to code as if warnings were as bad as errors.\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": "Thu, 29 Aug 2002 21:28:08 -0400", "msg_from": "\"D'Arcy J.M. Cain\" <darcy@druid.net>", "msg_from_op": true, "msg_subject": "Re: Type definition process (was Re: MemoryContextAlloc: invalid\n\trequest size 1934906735)" } ]
[ { "msg_contents": "How do I do an insert and retreival for bytea field using embedded SQL? \nAnybody have a code example? I'm trying to port from informix to \npostgresQL.\n\nThanks\n\nMike\n", "msg_date": "Tue, 27 Aug 2002 05:59:36 GMT", "msg_from": "Mike Sacauskis <mikejsNOSPAM@mbay.net>", "msg_from_op": true, "msg_subject": "BYTEA with ecpg" }, { "msg_contents": "On Tue, Aug 27, 2002 at 05:59:36AM +0000, Mike Sacauskis wrote:\n> How do I do an insert and retreival for bytea field using embedded SQL? \n> Anybody have a code example? I'm trying to port from informix to \n> postgresQL.\n\nActually I never tried. Doesn't it work with using an array of char as C\nvariable?\n\nMichael\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n", "msg_date": "Tue, 3 Sep 2002 17:29:45 +0200", "msg_from": "Michael Meskes <meskes@postgresql.org>", "msg_from_op": false, "msg_subject": "Re: BYTEA with ecpg" } ]
[ { "msg_contents": "\n> > Could the index scan be made to\n> > handle cases where the index tuple it was stopped on is gone?\n> \n> Don't see how. With no equal keys, you could test each tuple you scan\n> over to see if it's > the expected key; but that would slow things down\n> tremendously I fear. In any case it fails completely when there are\n> equal keys, since you could not tell where in a run of equal keys to\n> resume scanning. You really have to find the exact index tuple you\n> stopped on, AFAICS.\n\nWon't it still point to the same heap page and slot ? That additional info \nshould be sufficient to find the exact index tuple. \nAnd it usually won't be that far away, no ?\n\nAndreas\n", "msg_date": "Tue, 27 Aug 2002 09:14:41 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: Think I see a btree vacuuming bug " }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n>>> Could the index scan be made to\n>>> handle cases where the index tuple it was stopped on is gone?\n\n> Won't it still point to the same heap page and slot ?\n\nNot if it's gone ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 09:06:52 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Think I see a btree vacuuming bug " } ]
[ { "msg_contents": "Would it be worth adding REINDEX ALL and CLUSTER ALL as actual SQL commands?\nThis would be neat. Plus, it means we don't have to worry about having\nunix-only script in the distro once we have Win32 support.\n\nActually, we should just leave the 'ALL' off. That will make them behave\nlike VACUUM without arguments...\n\nChris\n\n", "msg_date": "Tue, 27 Aug 2002 15:33:37 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "REINDEX ALL and CLUSTER ALL" }, { "msg_contents": "Christopher Kings-Lynne wrote:\n> Would it be worth adding REINDEX ALL and CLUSTER ALL as actual SQL commands?\n> This would be neat. Plus, it means we don't have to worry about having\n> unix-only script in the distro once we have Win32 support.\n> \n> Actually, we should just leave the 'ALL' off. That will make them behave\n> like VACUUM without arguments...\n\nWow, now that is a nify idea! Let me add it to TODO and we can get rid\nof the shell scripts entirely:\n\n o Allow CLUSTER to cluster all tables, remove clusterdb\n\to Allow REINDEX to rebuild all indexes, remove /contrib/reindex\n\nIf we ever get the index growth fixed, we will not need the reindex\nchange, I guess, but maybe if they have some index corruption but they\nare not sure where it may be helpful.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 10:17:06 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: REINDEX ALL and CLUSTER ALL" }, { "msg_contents": "On Tue, 27 Aug 2002, Bruce Momjian wrote:\n\n> Christopher Kings-Lynne wrote:\n> > Would it be worth adding REINDEX ALL and CLUSTER ALL as actual SQL commands?\n> > This would be neat. Plus, it means we don't have to worry about having\n> > unix-only script in the distro once we have Win32 support.\n> > \n> > Actually, we should just leave the 'ALL' off. That will make them behave\n> > like VACUUM without arguments...\n> \n> Wow, now that is a nify idea! Let me add it to TODO and we can get rid\n> of the shell scripts entirely:\n> \n> o Allow CLUSTER to cluster all tables, remove clusterdb\n> \to Allow REINDEX to rebuild all indexes, remove /contrib/reindex\n> \n> If we ever get the index growth fixed, we will not need the reindex\n> change, I guess, but maybe if they have some index corruption but they\n> are not sure where it may be helpful.\n\nIsn't it true that reindex's behavior is to simply, quietly delete the \nindex? that was reported by someone when all this was going around \nbefore. I wrote my own reindex script that basically (in a single \ntransaction) grabbed the definition of the index, dropped said index, then \nrecreated it, then committed the transaction, so that if it failed for any \nreason, the old index was still there.\n\nIf reindex does \"lose\" the index on failure then we need to look at \nchanging how it works before we recommend it as a \"daily maintenance \nroutine\".\n\n", "msg_date": "Tue, 27 Aug 2002 09:51:10 -0600 (MDT)", "msg_from": "\"scott.marlowe\" <scott.marlowe@ihs.com>", "msg_from_op": false, "msg_subject": "Re: REINDEX ALL and CLUSTER ALL" }, { "msg_contents": "\nREINDEX just rebuilds the index, not just drop it. In fact, 7.3 will\nhave a reindexdb script.\n\n---------------------------------------------------------------------------\n\nscott.marlowe wrote:\n> On Tue, 27 Aug 2002, Bruce Momjian wrote:\n> \n> > Christopher Kings-Lynne wrote:\n> > > Would it be worth adding REINDEX ALL and CLUSTER ALL as actual SQL commands?\n> > > This would be neat. Plus, it means we don't have to worry about having\n> > > unix-only script in the distro once we have Win32 support.\n> > > \n> > > Actually, we should just leave the 'ALL' off. That will make them behave\n> > > like VACUUM without arguments...\n> > \n> > Wow, now that is a nify idea! Let me add it to TODO and we can get rid\n> > of the shell scripts entirely:\n> > \n> > o Allow CLUSTER to cluster all tables, remove clusterdb\n> > \to Allow REINDEX to rebuild all indexes, remove /contrib/reindex\n> > \n> > If we ever get the index growth fixed, we will not need the reindex\n> > change, I guess, but maybe if they have some index corruption but they\n> > are not sure where it may be helpful.\n> \n> Isn't it true that reindex's behavior is to simply, quietly delete the \n> index? that was reported by someone when all this was going around \n> before. I wrote my own reindex script that basically (in a single \n> transaction) grabbed the definition of the index, dropped said index, then \n> recreated it, then committed the transaction, so that if it failed for any \n> reason, the old index was still there.\n> \n> If reindex does \"lose\" the index on failure then we need to look at \n> changing how it works before we recommend it as a \"daily maintenance \n> routine\".\n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 12:08:25 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: REINDEX ALL and CLUSTER ALL" }, { "msg_contents": "On Tue, 27 Aug 2002, Bruce Momjian wrote:\n\n> Christopher Kings-Lynne wrote:\n> > Would it be worth adding REINDEX ALL and CLUSTER ALL as actual SQL commands?\n> > This would be neat. Plus, it means we don't have to worry about having\n> > unix-only script in the distro once we have Win32 support.\n> > \n> > Actually, we should just leave the 'ALL' off. That will make them behave\n> > like VACUUM without arguments...\n> \n> Wow, now that is a nify idea! Let me add it to TODO and we can get rid\n> of the shell scripts entirely:\n> \n> o Allow CLUSTER to cluster all tables, remove clusterdb\n> \to Allow REINDEX to rebuild all indexes, remove /contrib/reindex\n\nHuh... I asked whether to do the CLUSTER ALL thing, and someone said it\nwas just bloat; no one seemed to think it was useful, so I abandoned the\nidea.\n\n-- \nAlvaro Herrera (<alvherre[@]dcc.uchile.cl>)\n\"Porque Kim no hacia nada, pero, eso si,\ncon extraordinario exito\" (\"Kim\", Kipling)\n\n", "msg_date": "Tue, 27 Aug 2002 12:13:02 -0400 (CLT)", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": false, "msg_subject": "Re: REINDEX ALL and CLUSTER ALL" }, { "msg_contents": "Alvaro Herrera wrote:\n> On Tue, 27 Aug 2002, Bruce Momjian wrote:\n> \n> > Christopher Kings-Lynne wrote:\n> > > Would it be worth adding REINDEX ALL and CLUSTER ALL as actual SQL commands?\n> > > This would be neat. Plus, it means we don't have to worry about having\n> > > unix-only script in the distro once we have Win32 support.\n> > > \n> > > Actually, we should just leave the 'ALL' off. That will make them behave\n> > > like VACUUM without arguments...\n> > \n> > Wow, now that is a nify idea! Let me add it to TODO and we can get rid\n> > of the shell scripts entirely:\n> > \n> > o Allow CLUSTER to cluster all tables, remove clusterdb\n> > \to Allow REINDEX to rebuild all indexes, remove /contrib/reindex\n> \n> Huh... I asked whether to do the CLUSTER ALL thing, and someone said it\n> was just bloat; no one seemed to think it was useful, so I abandoned the\n> idea.\n\nOh, we did? Yes, I remember that.\n\nWell, seeing as we now need clusterdb command, it would be better to get\nthe backend to do it rather than have a separate command floating\naround. A separate script is certainly more bloat than whatever code we\nwould add in cluster.c.\n\nI think this may have been before we got on the idea of marking\npg_attribute with cluster info so we could more easily do cluster of all\ntables.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 12:18:13 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: REINDEX ALL and CLUSTER ALL" }, { "msg_contents": "Sorry, that should have been:\n\nIsn't it true that reindex's behavior ON A FAILURE is to simply, quietly \ndelete the index? that was reported ^^^^^^^^^^^^^\n\n\nOn Tue, 27 Aug 2002, Bruce Momjian wrote:\n\n> \n> REINDEX just rebuilds the index, not just drop it. In fact, 7.3 will\n> have a reindexdb script.\n> \n> ---------------------------------------------------------------------------\n> \n> scott.marlowe wrote:\n> > On Tue, 27 Aug 2002, Bruce Momjian wrote:\n> > \n> > > Christopher Kings-Lynne wrote:\n> > > > Would it be worth adding REINDEX ALL and CLUSTER ALL as actual SQL commands?\n> > > > This would be neat. Plus, it means we don't have to worry about having\n> > > > unix-only script in the distro once we have Win32 support.\n> > > > \n> > > > Actually, we should just leave the 'ALL' off. That will make them behave\n> > > > like VACUUM without arguments...\n> > > \n> > > Wow, now that is a nify idea! Let me add it to TODO and we can get rid\n> > > of the shell scripts entirely:\n> > > \n> > > o Allow CLUSTER to cluster all tables, remove clusterdb\n> > > \to Allow REINDEX to rebuild all indexes, remove /contrib/reindex\n> > > \n> > > If we ever get the index growth fixed, we will not need the reindex\n> > > change, I guess, but maybe if they have some index corruption but they\n> > > are not sure where it may be helpful.\n> > \n> > Isn't it true that reindex's behavior is to simply, quietly delete the \n> > index? that was reported by someone when all this was going around \n> > before. I wrote my own reindex script that basically (in a single \n> > transaction) grabbed the definition of the index, dropped said index, then \n> > recreated it, then committed the transaction, so that if it failed for any \n> > reason, the old index was still there.\n> > \n> > If reindex does \"lose\" the index on failure then we need to look at \n> > changing how it works before we recommend it as a \"daily maintenance \n> > routine\".\n> > \n> > \n> \n> \n\n", "msg_date": "Tue, 27 Aug 2002 11:14:58 -0600 (MDT)", "msg_from": "\"scott.marlowe\" <scott.marlowe@ihs.com>", "msg_from_op": false, "msg_subject": "Re: REINDEX ALL and CLUSTER ALL" }, { "msg_contents": "\nI am not sure, but it certainly makes sense that it would drop the index\non failure. I would never expect it to fail, however.\n\n---------------------------------------------------------------------------\n\nscott.marlowe wrote:\n> Sorry, that should have been:\n> \n> Isn't it true that reindex's behavior ON A FAILURE is to simply, quietly \n> delete the index? that was reported ^^^^^^^^^^^^^\n> \n> \n> On Tue, 27 Aug 2002, Bruce Momjian wrote:\n> \n> > \n> > REINDEX just rebuilds the index, not just drop it. In fact, 7.3 will\n> > have a reindexdb script.\n> > \n> > ---------------------------------------------------------------------------\n> > \n> > scott.marlowe wrote:\n> > > On Tue, 27 Aug 2002, Bruce Momjian wrote:\n> > > \n> > > > Christopher Kings-Lynne wrote:\n> > > > > Would it be worth adding REINDEX ALL and CLUSTER ALL as actual SQL commands?\n> > > > > This would be neat. Plus, it means we don't have to worry about having\n> > > > > unix-only script in the distro once we have Win32 support.\n> > > > > \n> > > > > Actually, we should just leave the 'ALL' off. That will make them behave\n> > > > > like VACUUM without arguments...\n> > > > \n> > > > Wow, now that is a nify idea! Let me add it to TODO and we can get rid\n> > > > of the shell scripts entirely:\n> > > > \n> > > > o Allow CLUSTER to cluster all tables, remove clusterdb\n> > > > \to Allow REINDEX to rebuild all indexes, remove /contrib/reindex\n> > > > \n> > > > If we ever get the index growth fixed, we will not need the reindex\n> > > > change, I guess, but maybe if they have some index corruption but they\n> > > > are not sure where it may be helpful.\n> > > \n> > > Isn't it true that reindex's behavior is to simply, quietly delete the \n> > > index? that was reported by someone when all this was going around \n> > > before. I wrote my own reindex script that basically (in a single \n> > > transaction) grabbed the definition of the index, dropped said index, then \n> > > recreated it, then committed the transaction, so that if it failed for any \n> > > reason, the old index was still there.\n> > > \n> > > If reindex does \"lose\" the index on failure then we need to look at \n> > > changing how it works before we recommend it as a \"daily maintenance \n> > > routine\".\n> > > \n> > > \n> > \n> > \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> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 13:45:20 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: REINDEX ALL and CLUSTER ALL" }, { "msg_contents": "I would guess that if someone mentioned it, then it HAS happened at least \nonce, maybe more. Would doing it in a transaction be a good idea or not? \nI'm not that familiar with the implications of doing a reindex by hand in \na transaction. \n\nSince reindex was designed to fix broken indexes, it's use to reclaim \nspace may awaken bugs no man has dared to dream exist before. Or \nsomething like that. \n\nOn Tue, 27 Aug 2002, Bruce Momjian wrote:\n\n> \n> I am not sure, but it certainly makes sense that it would drop the index\n> on failure. I would never expect it to fail, however.\n> \n> ---------------------------------------------------------------------------\n> \n> scott.marlowe wrote:\n> > Sorry, that should have been:\n> > \n> > Isn't it true that reindex's behavior ON A FAILURE is to simply, quietly \n> > delete the index? that was reported ^^^^^^^^^^^^^\n> > \n> > \n> > On Tue, 27 Aug 2002, Bruce Momjian wrote:\n> > \n> > > \n> > > REINDEX just rebuilds the index, not just drop it. In fact, 7.3 will\n> > > have a reindexdb script.\n> > > \n> > > ---------------------------------------------------------------------------\n> > > \n> > > scott.marlowe wrote:\n> > > > On Tue, 27 Aug 2002, Bruce Momjian wrote:\n> > > > \n> > > > > Christopher Kings-Lynne wrote:\n> > > > > > Would it be worth adding REINDEX ALL and CLUSTER ALL as actual SQL commands?\n> > > > > > This would be neat. Plus, it means we don't have to worry about having\n> > > > > > unix-only script in the distro once we have Win32 support.\n> > > > > > \n> > > > > > Actually, we should just leave the 'ALL' off. That will make them behave\n> > > > > > like VACUUM without arguments...\n> > > > > \n> > > > > Wow, now that is a nify idea! Let me add it to TODO and we can get rid\n> > > > > of the shell scripts entirely:\n> > > > > \n> > > > > o Allow CLUSTER to cluster all tables, remove clusterdb\n> > > > > \to Allow REINDEX to rebuild all indexes, remove /contrib/reindex\n> > > > > \n> > > > > If we ever get the index growth fixed, we will not need the reindex\n> > > > > change, I guess, but maybe if they have some index corruption but they\n> > > > > are not sure where it may be helpful.\n> > > > \n> > > > Isn't it true that reindex's behavior is to simply, quietly delete the \n> > > > index? that was reported by someone when all this was going around \n> > > > before. I wrote my own reindex script that basically (in a single \n> > > > transaction) grabbed the definition of the index, dropped said index, then \n> > > > recreated it, then committed the transaction, so that if it failed for any \n> > > > reason, the old index was still there.\n> > > > \n> > > > If reindex does \"lose\" the index on failure then we need to look at \n> > > > changing how it works before we recommend it as a \"daily maintenance \n> > > > routine\".\n> > > > \n> > > > \n> > > \n> > > \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> > \n> \n> \n\n", "msg_date": "Tue, 27 Aug 2002 13:39:49 -0600 (MDT)", "msg_from": "\"scott.marlowe\" <scott.marlowe@ihs.com>", "msg_from_op": false, "msg_subject": "Is REINDEX ALL safe?" }, { "msg_contents": "\"scott.marlowe\" <scott.marlowe@ihs.com> writes:\n> Sorry, that should have been:\n> Isn't it true that reindex's behavior ON A FAILURE is to simply, quietly \n> delete the index? that was reported ^^^^^^^^^^^^^\n\nNo.\n\nIf you are doing a standalone system index rebuild (with backend -P\nswitch) then REINDEX does a \"TRUNCATE\" of the index relation and\nrebuilds it in place. If that fails partway through, you'd be left\nwith a corrupted index ... which presumably is the same problem you\nstarted with, so I'm not that concerned about it.\n\nThe TRUNCATE approach is also used for rebuilding indexes on shared\nsystem relations (pg_database, pg_shadow, pg_group). This seems\nnecessary since REINDEX has no way to update pg_class.relfilenode in\ndatabases other than the current one.\n\nIn all other cases the rebuild is rollback-able, and a failure should\nleave you exactly where you were before.\n\n\nGiven these facts I think it would be a bad idea to include the shared\nsystem relations in any automatic \"REINDEX ALL\" command. One could\nmake a good argument that any such command should skip *all* system\ntables, actually.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 16:08:53 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: REINDEX ALL and CLUSTER ALL " }, { "msg_contents": "Tom Lane wrote:\n> \"scott.marlowe\" <scott.marlowe@ihs.com> writes:\n> > Sorry, that should have been:\n> > Isn't it true that reindex's behavior ON A FAILURE is to simply, quietly \n> > delete the index? that was reported ^^^^^^^^^^^^^\n> \n> No.\n> \n> If you are doing a standalone system index rebuild (with backend -P\n> switch) then REINDEX does a \"TRUNCATE\" of the index relation and\n> rebuilds it in place. If that fails partway through, you'd be left\n> with a corrupted index ... which presumably is the same problem you\n> started with, so I'm not that concerned about it.\n> \n> The TRUNCATE approach is also used for rebuilding indexes on shared\n> system relations (pg_database, pg_shadow, pg_group). This seems\n> necessary since REINDEX has no way to update pg_class.relfilenode in\n> databases other than the current one.\n> \n> In all other cases the rebuild is rollback-able, and a failure should\n> leave you exactly where you were before.\n> \n> \n> Given these facts I think it would be a bad idea to include the shared\n> system relations in any automatic \"REINDEX ALL\" command. One could\n> make a good argument that any such command should skip *all* system\n> tables, actually.\n\nYes, absolutely. REINDEX is not like vacuum. It needs to skip all\nsystem tables, I think. Those indexes are tied into backend structures.\n\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 16:15:35 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: REINDEX ALL and CLUSTER ALL" } ]
[ { "msg_contents": "\n> What we really need is some mode on template1 that says, \"I am not\n> world-writable, but the admin hasn't made me world-non-writable, so I\n> will create new databases that are world-writable\". Does that make\n> sense?\n\nI think template1 public should be non-world-writeable by default, and pass that\nto newly created databases. What exactly was it, why we don't want that ?\nBackwards compatibility ? Imho doing one (additional) grant after creating a db\ncannot be such a problem, no ? \n\nAndreas\n", "msg_date": "Tue, 27 Aug 2002 09:34:11 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: Default privileges for new databases (was Re: Can't" }, { "msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> ... What exactly was it, why we don't want that ?\n> Backwards compatibility ? Imho doing one (additional) grant after creating a db\n> cannot be such a problem, no ? \n\nBreaking every existing pg_dumpall script doesn't strike you as a\nproblem?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 09:08:23 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: Default privileges for new databases (was Re: Can't " } ]
[ { "msg_contents": "Sorry to bother, but it seems my last to messages (one to the news and\none yesterday to the pghackers list) did'nt go through..\n\nThe problem related to difficulties with plperl and 7.3 on UW 711\n\nRegards\n\n-- \nOlivier PRENANT \tTel:\t+33-5-61-50-97-00 (Work)\nQuartier d'Harraud Turrou +33-5-61-50-97-01 (Fax)\n31190 AUTERIVE +33-6-07-63-80-64 (GSM)\nFRANCE Email: ohp@pyrenet.fr\n------------------------------------------------------------------------------\nMake your life a dream, make your dream a reality. (St Exupery)\n\n", "msg_date": "Tue, 27 Aug 2002 15:31:09 +0200 (MET DST)", "msg_from": "Olivier PRENANT <ohp@pyrenet.fr>", "msg_from_op": true, "msg_subject": "can't post to the list" } ]
[ { "msg_contents": "\nFolks, I discussed with Harald a month ago how to improve the psql \\?\ndisplay. I have implemented his idea of mentioning '+' for the options\nthat can use it, and that is in CVS.\n\nHis other idea is to group the backslash commands into sections, so they\nare not all one big alphabetical list.\n\nHow do people like the idea of grouping them? I would do it in psql \\?\nand in the psql manual page.\n\n---------------------------------------------------------------------------\n\nHarald Armin Massa wrote:\n> But there could really be an improvement of the psql help by grouping the\n> commands after \"Areas of interest\" instead of sorting on alphabet. I have\n> done a first try:\n> \n> Servers, Connections\n> \\c[onnect] [DBNAME|- [USER]]\n> connect to new database (currently \"test\")\n> \\h NAME help on syntax of SQL commands, * for all commands\n> \\g FILENAME send SQL command to server (and write results to file or\n> |pipe)\n> \\copy ... perform SQL COPY with data stream to the client host\n> \\cd [DIRNAME] change the current working directory\n> \\! [COMMAND] execute command in shell or start interactive shell\n> \\encoding ENCODING set client encoding\n> \\set NAME VALUE set internal variable\n> \\unset NAME unset (delete) internal variable\n> \\q quit psql\n> \n> \n> Information Functions\n> \\copyright show PostgreSQL usage and distribution terms\n> \\d TABLE describe table (or view, index, sequence)\n> \\d{t|i|s|v}... list tables/indexes/sequences/views (add + for extended)\n> \\d{p|S|l} list access privileges, system tables, or large objects\n> \\da list aggregate functions\n> \\dd NAME show comment for table, type, function, or operator\n> \\df list functions\n> \\do list operators\n> \\dT list data types\n> \\l list all databases (add + for extended output)\n> \\z list table access privileges\n> \\d+ extended List\n> \\l+\n> \n> Query Buffer\n> \\r reset (clear) the query buffer\n> \\e FILENAME edit the current query buffer or file with external editor\n> \\i FILENAME execute commands from file\n> \\p show the content of the current query buffer\n> \\w FILENAME write current query buffer to file\n> \n> \n> Large Object\n> \\lo_export, \\lo_import, \\lo_list, \\lo_unlink\n> large object operations\n> \n> Output and Formatting\n> \\o FILENAME send all query results to file or |pipe\n> \\a toggle between unaligned and aligned output mode\n> \\f STRING set field separator\n> \\t show only rows (currently off)\n> \\x toggle expanded output (currently off)\n> \\echo TEXT write text to standard output\n> \\qecho TEXT write text to query output stream (see \\o)\n> \\s FILENAME print history or save it to file\n> \\C TITLE set table title\n> \\H toggle HTML output mode (currently off)\n> \\T TEXT set HTML table tag attributes\n> \\pset VAR set table output option (VAR := {format|border|expanded|\n> fieldsep|null|recordsep|tuples_only|title|tableattr|pager})\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 14:36:00 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [BUGS] Bug #718: request for improvement of /? to show /d+ /l+" }, { "msg_contents": "Bruce Momjian dijo: \n\n> Folks, I discussed with Harald a month ago how to improve the psql \\?\n> display. I have implemented his idea of mentioning '+' for the options\n> that can use it, and that is in CVS.\n> \n> His other idea is to group the backslash commands into sections, so they\n> are not all one big alphabetical list.\n> \n> How do people like the idea of grouping them? I would do it in psql \\?\n> and in the psql manual page.\n\nI think the grouping improves usability in general; however, the list\nyou propose below is too long. Maybe having a \"table of contents\" with\nthe groups and allow \\? to take an argument?\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"Aprende a avergonzarte mas ante ti que ante los demas\" (Democrito)\n\n", "msg_date": "Tue, 27 Aug 2002 17:07:27 -0400 (CLT)", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": false, "msg_subject": "Re: [BUGS] Bug #718: request for improvement of /? to" }, { "msg_contents": "Alvaro Herrera wrote:\n> Bruce Momjian dijo: \n> \n> > Folks, I discussed with Harald a month ago how to improve the psql \\?\n> > display. I have implemented his idea of mentioning '+' for the options\n> > that can use it, and that is in CVS.\n> > \n> > His other idea is to group the backslash commands into sections, so they\n> > are not all one big alphabetical list.\n> > \n> > How do people like the idea of grouping them? I would do it in psql \\?\n> > and in the psql manual page.\n> \n> I think the grouping improves usability in general; however, the list\n> you propose below is too long. Maybe having a \"table of contents\" with\n> the groups and allow \\? to take an argument?\n\nI will pair down the actual output to look reasonable, and you can\nscroll through the \\? help anyway, so we should be OK.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 17:12:31 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: [BUGS] Bug #718: request for improvement of /? to show" }, { "msg_contents": "Bruce Momjian writes:\n\n> How do people like the idea of grouping them? I would do it in psql \\?\n> and in the psql manual page.\n\nSeems reasonable, but then we should also have something along the lines\nof\n\n\\? \\r\n\nso I don't have to dig through the whole pile if I'm interested in a\nparticular command. (cf. \\help behavior)\n\nIn the man page the change would be OK since you can easily search the\npage for any string.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Fri, 30 Aug 2002 00:14:20 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": false, "msg_subject": "Re: [BUGS] Bug #718: request for improvement of /? to" }, { "msg_contents": "En Fri, 30 Aug 2002 00:14:20 +0200 (CEST)\nPeter Eisentraut <peter_e@gmx.net> escribi�:\n\n> Bruce Momjian writes:\n> \n> > How do people like the idea of grouping them? I would do it in psql \\?\n> > and in the psql manual page.\n> \n> Seems reasonable, but then we should also have something along the lines\n> of\n> \n> \\? \\r\n> \n> so I don't have to dig through the whole pile if I'm interested in a\n> particular command. (cf. \\help behavior)\n\nThat'd be neat. Also you can include some meta-items that allow the\ndisplay of a single group with more detail.\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"Nunca se desea ardientemente lo que solo se desea por razon\" (F. Alexandre)\n", "msg_date": "Thu, 29 Aug 2002 18:23:32 -0400", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": false, "msg_subject": "Re: [BUGS] Bug #718: request for improvement of /? to" } ]
[ { "msg_contents": "pg_resetxlog uses a non-standard options parsing method: The -l option\nrequires two arguments (-l fileid seg). I propose to change this to -l\nfileid,seg which is the standard way to separate suboptions.\n\nSecondly, the -n option appears to be redundant with pg_controldata. Do\nwe need it?\n\nThirdly, pg_resetxlog uses the term \"guessed\" controldata values if it\ncan't read the real ones. I found this to be confusing, because the code\ndoesn't do a whole lot of guessing. Would it be better to say that the\nvalues are simply defaulted (and to what)?\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n", "msg_date": "Tue, 27 Aug 2002 22:16:30 +0200 (CEST)", "msg_from": "Peter Eisentraut <peter_e@gmx.net>", "msg_from_op": true, "msg_subject": "pg_resetxlog options" }, { "msg_contents": "Peter Eisentraut wrote:\n> pg_resetxlog uses a non-standard options parsing method: The -l option\n> requires two arguments (-l fileid seg). I propose to change this to -l\n> fileid,seg which is the standard way to separate suboptions.\n\nAgreed.\n\n> Secondly, the -n option appears to be redundant with pg_controldata. Do\n> we need it?\n\nYep.\n\n> Thirdly, pg_resetxlog uses the term \"guessed\" controldata values if it\n> can't read the real ones. I found this to be confusing, because the code\n> doesn't do a whole lot of guessing. Would it be better to say that the\n> values are simply defaulted (and to what)?\n\n\"Attempts to determine the proper values\"?\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 16:16:56 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: pg_resetxlog options" }, { "msg_contents": "Peter Eisentraut <peter_e@gmx.net> writes:\n> pg_resetxlog uses a non-standard options parsing method: The -l option\n> requires two arguments (-l fileid seg). I propose to change this to -l\n> fileid,seg which is the standard way to separate suboptions.\n\nNo objection. I think pg_upgrade uses that option, so please adjust it\ntoo.\n\n> Secondly, the -n option appears to be redundant with pg_controldata. Do\n> we need it?\n\nI would like to keep it. It gives some comfort factor that pg_resetxlog\nhas chosen the right things to do, before it does them.\n\n> Thirdly, pg_resetxlog uses the term \"guessed\" controldata values if it\n> can't read the real ones. I found this to be confusing, because the code\n> doesn't do a whole lot of guessing.\n\nThere needs to be more AI in there than there presently is ;-), but I\nthink the term is quite appropriate. Without a readable pg_control,\npg_resetxlog really is guessing at a number of critical data items,\nsuch as the next transaction ID, the locale, etc. I *want* the user\nto be apprehensive if that flow of control is taken, and I think a term\nlike \"guessed\" will induce an appropriately paranoid frame of mind.\nIf you'd like to propose alternate wording, feel free, but \"default\" is\nnot it. If I read \"we used the default values\", I'm going to think\neverything is fine and no thought is required. That's exactly the wrong\nthing for a user of pg_resetxlog to think.\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 18:01:11 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: pg_resetxlog options " } ]
[ { "msg_contents": "\nYes, this is the counter case, where the '@' disappears; so it appears\nmagically for local users, and disappears for global users.\n\n---------------------------------------------------------------------------\n\nRobert Treat wrote:\n> Is the converse to this:\n> \n> $ psql -U postgres@ test\n> \n> Welcome to psql 7.3devel, the PostgreSQL interactive terminal.\n> \n> Type: \\copyright for distribution terms\n> \\h for help with SQL commands\n> \\? for help on internal slash commands\n> \\g or terminate with semicolon to execute query\n> \\q to quit\n> \n> test=> select current_user;\n> current_user \n> --------------\n> postgres\n> (1 row)\n> \n> \n> this seems counterintuitive to me, so I'd like to see the strong\n> practical application that makes it necessary. (This is where mark comes\n> in I suppose)\n> \n> Robert Treat\n> \n> On Tue, 2002-08-27 at 15:43, Bruce Momjian wrote:\n> > Lamar Owen wrote:\n> > > On Tuesday 27 August 2002 03:19 pm, Bruce Momjian wrote:\n> > > > I think we need to resolve this discussion from a week ago. The current\n> > > > code is this:\n> > > \n> > > I thought it WAS resolved, to do:\n> > > \n> > > > \tglobal usernames are stored just like before, e.g. postgres\n> > > > \tlocal users are stored as user@dbname\n> > > > \twhen connecting, global users add '@' to their names\n> > > > \twhen connecting, local users use just their user name, no @dbname\n> > > \n> > > > Tom likes this because it is the fewer global users who have to append\n> > > > the '@'.\n> > > \n> > > At least that was my perception of the uneasy consensus reached.\n> > > \n> > > Basically, this tags the @ as magic saying, during the client connect process, \n> > > 'I'm GLOBAL, treat me differently'. Now that I actually understand how this \n> > > is supposed to work, which your four lines above elucidate nicely, I am in \n> > > more agreement than I was that this is the right answer to this issue.\n> > \n> > OK, you have now split the vote because we have two for the change, and\n> > two against. Why do you prefer to tag the globals? Is it Tom's\n> > argument? I think it is kind of strange to tag the globals when it is\n> > the locals who have @ in their username, and when they do:\n> > \n> > \t$ psql -U dave test\n> > \tWelcome to psql 7.3devel, the PostgreSQL interactive terminal.\n> > \t\n> > \tType: \\copyright for distribution terms\n> > \t \\h for help with SQL commands\n> > \t \\? for help on internal slash commands\n> > \t \\g or terminate with semicolon to execute query\n> > \t \\q to quit\n> > \t\n> > \ttest=> select current_user;\n> > \t current_user \n> > \t--------------\n> > \t dave@test\n> > \t(1 row)\n> > \n> > they will see their full username.\n> > \n> > I can go either way. I am just saying we need to hear from more people\n> > to make sure we are doing this properly.\n> > \n> > -- \n> > Bruce Momjian | http://candle.pha.pa.us\n> > pgman@candle.pha.pa.us | (610) 359-1001\n> > + If your life is a hard drive, | 13 Roberts Road\n> > + Christ can be your backup. | Newtown Square, Pennsylvania 19073\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> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 16:18:32 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": true, "msg_subject": "Re: Open 7.3 items" } ]
[ { "msg_contents": "\nHello,\n\nI tried the chkpass thing on both CVS tip and 7.2 (REL_7_2 or something\nlike that), and didn't see any error.\n\nThis is Linux on Intel.\n\n(sorry, I deleted the thread by accident).\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"El miedo atento y previsor es la madre de la seguridad\" (E. Burke)\n\n", "msg_date": "Tue, 27 Aug 2002 17:28:55 -0400 (CLT)", "msg_from": "Alvaro Herrera <alvherre@atentus.com>", "msg_from_op": true, "msg_subject": "test result: chkpass" } ]
[ { "msg_contents": "This patch fixes the so-called DoS possibility when processing the\npassword packet in recv_and_check_passwordv0(). Nothing fancy, I just\nadded a sanity check to ensure that we bail out if the client enters\nan obviously-bogus length.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "27 Aug 2002 18:12:44 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": true, "msg_subject": "fix for palloc() of user-supplied length" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> This patch fixes the so-called DoS possibility when processing the\n> password packet in recv_and_check_passwordv0().\n\nIf len is signed, then something like \"len < 1\" needs to be in there\nas well.\n\nMore generally, though, I was thinking that the appropriate answer at\nthis point is to rip out support for version-0 authentication\naltogether. I can't believe anyone will be trying to connect to a 7.3\nor beyond server with 6.2 client libraries (v0 went away in 6.3 as best\nI can tell from the CVS logs). And if they try, it's not unreasonable\nto force them to upgrade --- those old client libraries have got to be\npretty buggy themselves. So the utility of the v0 backend code is\ndubious, while its potential for more problems is real.\n\nAnyone want to argue that we should keep the v0 protocol support\nany longer?\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 18:32:48 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length " }, { "msg_contents": "Tom Lane <tgl@sss.pgh.pa.us> writes:\n> More generally, though, I was thinking that the appropriate answer\n> at this point is to rip out support for version-0 authentication\n> altogether. I can't believe anyone will be trying to connect to a\n> 7.3 or beyond server with 6.2 client libraries (v0 went away in 6.3\n> as best I can tell from the CVS logs).\n\nFurther, has this code actually been tested within recent memory? If\nnot, I wouldn't be surprised to learn that it's suffered some\nbitrot...\n\n> Anyone want to argue that we should keep the v0 protocol support any\n> longer?\n\nNope, exactly the same thought crossed my mind while I was reading\nthrough the code...\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "27 Aug 2002 19:00:25 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": true, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> Tom Lane <tgl@sss.pgh.pa.us> writes:\n>> More generally, though, I was thinking that the appropriate answer\n>> at this point is to rip out support for version-0 authentication\n>> altogether.\n\n> Further, has this code actually been tested within recent memory? If\n> not, I wouldn't be surprised to learn that it's suffered some\n> bitrot...\n\nYup, that's another good point. I don't think we *have* a way of\ntesting it any longer, unless someone cares to pull a 6.2 psql from the\narchives ...\n\n\t\t\tregards, tom lane\n", "msg_date": "Tue, 27 Aug 2002 19:34:18 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length " }, { "msg_contents": "Neil Conway wrote:\n> Tom Lane <tgl@sss.pgh.pa.us> writes:\n> > More generally, though, I was thinking that the appropriate answer\n> > at this point is to rip out support for version-0 authentication\n> > altogether. I can't believe anyone will be trying to connect to a\n> > 7.3 or beyond server with 6.2 client libraries (v0 went away in 6.3\n> > as best I can tell from the CVS logs).\n> \n> Further, has this code actually been tested within recent memory? If\n> not, I wouldn't be surprised to learn that it's suffered some\n> bitrot...\n> \n> > Anyone want to argue that we should keep the v0 protocol support any\n> > longer?\n> \n> Nope, exactly the same thought crossed my mind while I was reading\n> through the code...\n\nFeel free to rip it out.\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 23:00:58 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "\nNeil, is this the one Sir-* complained about?\n\n---------------------------------------------------------------------------\n\nNeil Conway wrote:\n> This patch fixes the so-called DoS possibility when processing the\n> password packet in recv_and_check_passwordv0(). Nothing fancy, I just\n> added a sanity check to ensure that we bail out if the client enters\n> an obviously-bogus length.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n[ Attachment, skipping... ]\n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 27 Aug 2002 23:01:38 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Neil, is this the one Sir-* complained about?\n\nYes.\n\nI've attached a revised patch that includes the additional check Tom\nsuggested (len < 1). Unless anyone else steps forward, I'm inclined to\nrip out support for version 0 of the protocol -- but I have more\nurgent things to do for the beta, so it will likely need to wait for\n7.4.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "27 Aug 2002 23:39:15 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": true, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "Quoting Neil Conway <neilc@samurai.com>:\n\n> I've attached a revised patch that includes the additional check Tom\n> suggested (len < 1). Unless anyone else steps forward, I'm inclined to\n\n+ \tif (len < 1 || len > 8192)\n+ \t{\n+ \t\telog(LOG, \"Password packet length too long: %d\", len);\n ^^^^^^^^\nShouldn't it be changed to 'too long || too long' then? ;)\n\nAnd also for the message to be more descriptive for the innocent, I'd included\nthe current boundaries in it (like: \"expected: 1 <= len <= 8192\")\n(a question: isn't hardcoding an evil?)\n\nBut I guess it's not a must-to-do on your list :)\n\n-- \nSerguei A. Mokhov, <mailto: mokhov @ cs.concordia.ca>\nComputer Science, Concordia University\n\n-------------------------------------------------\nThis mail sent through IMP: http://horde.org/imp/\n", "msg_date": "Wed, 28 Aug 2002 00:07:22 -0400", "msg_from": "Serguei Mokhov <mokhov@cs.concordia.ca>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "I wrote:\n\n> > I've attached a revised patch that includes the additional check Tom\n> > suggested (len < 1). Unless anyone else steps forward, I'm inclined to\n> \n> + if (len < 1 || len > 8192)\n> + {\n> + elog(LOG, \"Password packet length too long: %d\", len);\n> ^^^^^^^^\n> Shouldn't it be changed to 'too long || too long' then? ;)\n\nA typo: [too short or too short] :)\n\n-- \nSerguei A. Mokhov, <mailto: mokhov @ cs.concordia.ca>\nComputer Science, Concordia University\n\n-------------------------------------------------\nThis mail sent through IMP: http://horde.org/imp/\n", "msg_date": "Wed, 28 Aug 2002 00:12:26 -0400", "msg_from": "Serguei Mokhov <mokhov@cs.concordia.ca>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "Serguei Mokhov <mokhov@cs.concordia.ca> writes:\n> + \tif (len < 1 || len > 8192)\n> + \t{\n> + \t\telog(LOG, \"Password packet length too long: %d\", len);\n> ^^^^^^^^\n> Shouldn't it be changed to 'too long || too long' then? ;)\n\nWoops, sorry for being careless. Changed the wording to refer to\n'invalid' rather than 'too long' or 'too short'.\n\n> And also for the message to be more descriptive for the innocent, I'd included\n> the current boundaries in it (like: \"expected: 1 <= len <= 8192\")\n\nAlso fixed, although I'm not sure it's worth worrying about.\n\n> (a question: isn't hardcoding an evil?)\n\nYes, probably -- as the comment notes, it is just an arbitrary\nlimitation. But given that (a) it is extremely unlikely to ever be\nencountered in a real-life situation (b) the limits it imposes are\nvery lax (c) it is temporary code that will be ripped out shortly, I'm\nnot too concerned... \n\nThanks for taking a look at the code, BTW.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC", "msg_date": "28 Aug 2002 00:33:26 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": true, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://candle.pha.pa.us/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nNeil Conway wrote:\n> Serguei Mokhov <mokhov@cs.concordia.ca> writes:\n> > + \tif (len < 1 || len > 8192)\n> > + \t{\n> > + \t\telog(LOG, \"Password packet length too long: %d\", len);\n> > ^^^^^^^^\n> > Shouldn't it be changed to 'too long || too long' then? ;)\n> \n> Woops, sorry for being careless. Changed the wording to refer to\n> 'invalid' rather than 'too long' or 'too short'.\n> \n> > And also for the message to be more descriptive for the innocent, I'd included\n> > the current boundaries in it (like: \"expected: 1 <= len <= 8192\")\n> \n> Also fixed, although I'm not sure it's worth worrying about.\n> \n> > (a question: isn't hardcoding an evil?)\n> \n> Yes, probably -- as the comment notes, it is just an arbitrary\n> limitation. But given that (a) it is extremely unlikely to ever be\n> encountered in a real-life situation (b) the limits it imposes are\n> very lax (c) it is temporary code that will be ripped out shortly, I'm\n> not too concerned... \n> \n> Thanks for taking a look at the code, BTW.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n[ Attachment, skipping... ]\n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 28 Aug 2002 17:15:53 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "> > > Anyone want to argue that we should keep the v0 protocol support any\n> > > longer?\n> > \n> > Nope, exactly the same thought crossed my mind while I was reading\n> > through the code...\n> \n> Feel free to rip it out.\n\nShould probably be mentioned in the release notes.\n\n", "msg_date": "28 Aug 2002 20:15:03 -0400", "msg_from": "\"Matthew T. O'Connor\" <matthew@zeut.net>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] fix for palloc() of user-supplied length" }, { "msg_contents": "\nIt will, if a patch is supplied. Anything significant that is mentioned\nin the CVS logs gets shown in the release notes.\n\n---------------------------------------------------------------------------\n\nMatthew T. O'Connor wrote:\n> > > > Anyone want to argue that we should keep the v0 protocol support any\n> > > > longer?\n> > > \n> > > Nope, exactly the same thought crossed my mind while I was reading\n> > > through the code...\n> > \n> > Feel free to rip it out.\n> \n> Should probably be mentioned in the release notes.\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 28 Aug 2002 21:15:10 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: [HACKERS] fix for palloc() of user-supplied length" }, { "msg_contents": "I have applied the following modified version of your patch. The\noriginal version would not apply to CVS.\n\n---------------------------------------------------------------------------\n\nNeil Conway wrote:\n> Serguei Mokhov <mokhov@cs.concordia.ca> writes:\n> > + \tif (len < 1 || len > 8192)\n> > + \t{\n> > + \t\telog(LOG, \"Password packet length too long: %d\", len);\n> > ^^^^^^^^\n> > Shouldn't it be changed to 'too long || too long' then? ;)\n> \n> Woops, sorry for being careless. Changed the wording to refer to\n> 'invalid' rather than 'too long' or 'too short'.\n> \n> > And also for the message to be more descriptive for the innocent, I'd included\n> > the current boundaries in it (like: \"expected: 1 <= len <= 8192\")\n> \n> Also fixed, although I'm not sure it's worth worrying about.\n> \n> > (a question: isn't hardcoding an evil?)\n> \n> Yes, probably -- as the comment notes, it is just an arbitrary\n> limitation. But given that (a) it is extremely unlikely to ever be\n> encountered in a real-life situation (b) the limits it imposes are\n> very lax (c) it is temporary code that will be ripped out shortly, I'm\n> not too concerned... \n> \n> Thanks for taking a look at the code, BTW.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n[ Attachment, skipping... ]\n\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n\nIndex: src/backend/libpq/auth.c\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/backend/libpq/auth.c,v\nretrieving revision 1.86\ndiff -c -c -r1.86 auth.c\n*** src/backend/libpq/auth.c\t29 Aug 2002 03:22:01 -0000\t1.86\n--- src/backend/libpq/auth.c\t29 Aug 2002 21:40:40 -0000\n***************\n*** 709,714 ****\n--- 709,727 ----\n \tif (pq_eof() == EOF || pq_getint(&len, 4) == EOF)\n \t\treturn STATUS_EOF;\t\t/* client didn't want to send password */\n \n+ \t/*\n+ \t * Since the remote client has not yet been authenticated, we need\n+ \t * to be careful when using the data they send us. The 8K limit is\n+ \t * arbitrary, and somewhat bogus: the intent is to ensure we don't\n+ \t * allocate an enormous chunk of memory.\n+ \t */\n+ \tif (len < 1 || len > 8192)\n+ \t{\n+ \t\telog(LOG, \"Invalid password packet length: %d; \"\n+ \t\t\t \"must satisfy 1 <= length <= 8192\", len);\n+ \t\treturn STATUS_EOF;\n+ \t}\n+ \n \tinitStringInfo(&buf);\n \tif (pq_getstr(&buf) == EOF) /* receive password */\n \t{", "msg_date": "Thu, 29 Aug 2002 17:45:56 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> I have applied the following modified version of your patch. The\n> original version would not apply to CVS.\n\nYes, the reason being that Tom removed the entire section of code that\nmy patch modified (and that is the better solution, IMHO).\n\nThe patch you've applied does something rather different, and is\nunrelated to the \"vulnerability\" reported by Mordred and referred to\nin the Subject -- your patch adds some additional sanity checking when\nreading the password packet from v1 protocol clients. This is\nunnecessary for two reasons:\n\n (1) We use a StringInfo to hold the input data, which is\n dynamically allocated as necessary. Since there's no\n palloc() with user-supplied data, you'd need to write x\n bytes to the backend to force it to allocate x bytes of\n memory (i.e. potential for DoS is low).\n\n (2) The length supplied by the user is completely ignored by\n the code, and it simply reads the input until it sees a\n NULL terminator (read the comments in the code about 10\n lines down.) Therefore, any sanity checking on the length\n specified by the user is a waste of time.\n\nYou should probably back out your patch.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n", "msg_date": "30 Aug 2002 01:10:29 -0400", "msg_from": "Neil Conway <neilc@samurai.com>", "msg_from_op": true, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> (2) The length supplied by the user is completely ignored by\n> the code, and it simply reads the input until it sees a\n> NULL terminator (read the comments in the code about 10\n> lines down.) Therefore, any sanity checking on the length\n> specified by the user is a waste of time.\n\nAgreed; the fact that the protocol requires a length word at all is just\na hangover from the past. We can read the length word and forget it.\n\nI wonder though if it'd be worthwhile to limit the length of the string\nthat we are willing to read from the client in the second step. We are\nat this point dealing with an unauthenticated user, so we should be\nuntrusting. And I think Sir Mordred has a point: forcing a backend to\nallocate a lot of memory can be a form of DoS attack.\n\n\t\t\tregards, tom lane\n", "msg_date": "Fri, 30 Aug 2002 01:24:41 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length " }, { "msg_contents": "\nPatch backed out. Thanks.\n\n---------------------------------------------------------------------------\n\nNeil Conway wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > I have applied the following modified version of your patch. The\n> > original version would not apply to CVS.\n> \n> Yes, the reason being that Tom removed the entire section of code that\n> my patch modified (and that is the better solution, IMHO).\n> \n> The patch you've applied does something rather different, and is\n> unrelated to the \"vulnerability\" reported by Mordred and referred to\n> in the Subject -- your patch adds some additional sanity checking when\n> reading the password packet from v1 protocol clients. This is\n> unnecessary for two reasons:\n> \n> (1) We use a StringInfo to hold the input data, which is\n> dynamically allocated as necessary. Since there's no\n> palloc() with user-supplied data, you'd need to write x\n> bytes to the backend to force it to allocate x bytes of\n> memory (i.e. potential for DoS is low).\n> \n> (2) The length supplied by the user is completely ignored by\n> the code, and it simply reads the input until it sees a\n> NULL terminator (read the comments in the code about 10\n> lines down.) Therefore, any sanity checking on the length\n> specified by the user is a waste of time.\n> \n> You should probably back out your patch.\n> \n> Cheers,\n> \n> Neil\n> \n> -- \n> Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n> \n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Fri, 30 Aug 2002 11:24:16 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "\nWould someone submit a patch for this?\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> Neil Conway <neilc@samurai.com> writes:\n> > (2) The length supplied by the user is completely ignored by\n> > the code, and it simply reads the input until it sees a\n> > NULL terminator (read the comments in the code about 10\n> > lines down.) Therefore, any sanity checking on the length\n> > specified by the user is a waste of time.\n> \n> Agreed; the fact that the protocol requires a length word at all is just\n> a hangover from the past. We can read the length word and forget it.\n> \n> I wonder though if it'd be worthwhile to limit the length of the string\n> that we are willing to read from the client in the second step. We are\n> at this point dealing with an unauthenticated user, so we should be\n> untrusting. And I think Sir Mordred has a point: forcing a backend to\n> allocate a lot of memory can be a form of DoS attack.\n> \n> \t\t\tregards, tom lane\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) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Mon, 2 Sep 2002 01:05:39 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "----- Original Message ----- \nFrom: \"Bruce Momjian\" <pgman@candle.pha.pa.us>\nSent: September 02, 2002 1:05 AM\n\n> Would someone submit a patch for this?\n\nWorking on it.\n\n-s\n\n> Tom Lane wrote:\n> > Neil Conway <neilc@samurai.com> writes:\n> > > (2) The length supplied by the user is completely ignored by\n> > > the code, and it simply reads the input until it sees a\n> > > NULL terminator (read the comments in the code about 10\n> > > lines down.) Therefore, any sanity checking on the length\n> > > specified by the user is a waste of time.\n> > \n> > Agreed; the fact that the protocol requires a length word at all is just\n> > a hangover from the past. We can read the length word and forget it.\n> > \n> > I wonder though if it'd be worthwhile to limit the length of the string\n> > that we are willing to read from the client in the second step. We are\n> > at this point dealing with an unauthenticated user, so we should be\n> > untrusting. And I think Sir Mordred has a point: forcing a backend to\n> > allocate a lot of memory can be a form of DoS attack.\n> > \n> > regards, tom lane\n\n", "msg_date": "Mon, 2 Sep 2002 02:15:24 -0400", "msg_from": "\"Serguei Mokhov\" <sa_mokho@alcor.concordia.ca>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "Hello,\n\n----- Original Message ----- \nFrom: \"Bruce Momjian\" <pgman@candle.pha.pa.us>\nSent: September 02, 2002 1:05 AM\n\n> Would someone submit a patch for this?\n\nAttached please find an attempt to fix the volunerability issue below.\n\nAffected files are:\n\n/src/include/libpq/libpq.h\n/src/include/libpq/pqformat.h\n/src/backend/libpq/pqformat.c\n/src/backend/libpq/pqcomm.c\n/src/backend/libpq/auth.c\n\n\"Briefly\" the changes:\n\nMain victims for the change were pq_getstring() and pq_getstr()\n(which calls the former) in pqformat.c and pqcomm.c. pq_getstring() is the one reading\nuntil \\0 and might possibly render the system run out of memory.\n\nChanging pq_getstring() alone would break a lot code, so I \nadded a two more functions: pq_getstring_common() and\npq_getstring_bounded(). The former is a big part of what used to be\npq_getstring() and the latter is a copy of the new pq_getstring()\nwith the string length check. Creating pq_getstring_common()\nwas suggested by its reuse in pq_getstring() and pq_getstring_bounded()\navoiding code duplication.\n\nSimilar changes were done for pq_getstr(). Its common code converting\nto MULTIBYTE was placed in pq_getstr_multibyte() and pq_getstr() and\n(newly added) pq_getstr_bounded() both call it before returning a result.\n\nWRT above, two places in auth.c were changed to call pq_getstr_bounded()\ninstead of pq_getstr() on password read. I'm not sure if\nthere are other places where that might be needed...\n\nMight look ugly for some, but looks like a not-so-bad solution\nto me. If I'm completely wrong, I'd like to have some guidance then :)\nPlease review with care. I'm off to bed.\n\nThanks,\n-s\n\nPS: The patch also fixes a typo in the be-secure.c comment :)\n\n> Tom Lane wrote:\n> > Neil Conway <neilc@samurai.com> writes:\n> > > (2) The length supplied by the user is completely ignored by\n> > > the code, and it simply reads the input until it sees a\n> > > NULL terminator (read the comments in the code about 10\n> > > lines down.) Therefore, any sanity checking on the length\n> > > specified by the user is a waste of time.\n> > \n> > Agreed; the fact that the protocol requires a length word at all is just\n> > a hangover from the past. We can read the length word and forget it.\n> > \n> > I wonder though if it'd be worthwhile to limit the length of the string\n> > that we are willing to read from the client in the second step. We are\n> > at this point dealing with an unauthenticated user, so we should be\n> > untrusting. And I think Sir Mordred has a point: forcing a backend to\n> > allocate a lot of memory can be a form of DoS attack.\n> > \n> > regards, tom lane", "msg_date": "Mon, 2 Sep 2002 04:11:17 -0400", "msg_from": "\"Serguei Mokhov\" <sa_mokho@alcor.concordia.ca>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "Hello again,\n\n*any* comment on this at all?\n\nthanks,\n-s\n\n----- Original Message ----- \nFrom: \"Serguei Mokhov\" <sa_mokho@alcor.concordia.ca>\nSent: September 02, 2002 4:11 AM\n\n> Hello,\n> \n> ----- Original Message ----- \n> From: \"Bruce Momjian\" <pgman@candle.pha.pa.us>\n> Sent: September 02, 2002 1:05 AM\n> \n> > Would someone submit a patch for this?\n> \n> Attached please find an attempt to fix the volunerability issue below.\n> \n> Affected files are:\n> \n> /src/include/libpq/libpq.h\n> /src/include/libpq/pqformat.h\n> /src/backend/libpq/pqformat.c\n> /src/backend/libpq/pqcomm.c\n> /src/backend/libpq/auth.c\n> \n> \"Briefly\" the changes:\n> \n> Main victims for the change were pq_getstring() and pq_getstr()\n> (which calls the former) in pqformat.c and pqcomm.c. pq_getstring() is the one reading\n> until \\0 and might possibly render the system run out of memory.\n> \n> Changing pq_getstring() alone would break a lot code, so I \n> added a two more functions: pq_getstring_common() and\n> pq_getstring_bounded(). The former is a big part of what used to be\n> pq_getstring() and the latter is a copy of the new pq_getstring()\n> with the string length check. Creating pq_getstring_common()\n> was suggested by its reuse in pq_getstring() and pq_getstring_bounded()\n> avoiding code duplication.\n> \n> Similar changes were done for pq_getstr(). Its common code converting\n> to MULTIBYTE was placed in pq_getstr_multibyte() and pq_getstr() and\n> (newly added) pq_getstr_bounded() both call it before returning a result.\n> \n> WRT above, two places in auth.c were changed to call pq_getstr_bounded()\n> instead of pq_getstr() on password read. I'm not sure if\n> there are other places where that might be needed...\n> \n> Might look ugly for some, but looks like a not-so-bad solution\n> to me. If I'm completely wrong, I'd like to have some guidance then :)\n> Please review with care. I'm off to bed.\n> \n> Thanks,\n> -s\n> \n> PS: The patch also fixes a typo in the be-secure.c comment :)\n> \n> > Tom Lane wrote:\n> > > Neil Conway <neilc@samurai.com> writes:\n> > > > (2) The length supplied by the user is completely ignored by\n> > > > the code, and it simply reads the input until it sees a\n> > > > NULL terminator (read the comments in the code about 10\n> > > > lines down.) Therefore, any sanity checking on the length\n> > > > specified by the user is a waste of time.\n> > > \n> > > Agreed; the fact that the protocol requires a length word at all is just\n> > > a hangover from the past. We can read the length word and forget it.\n> > > \n> > > I wonder though if it'd be worthwhile to limit the length of the string\n> > > that we are willing to read from the client in the second step. We are\n> > > at this point dealing with an unauthenticated user, so we should be\n> > > untrusting. And I think Sir Mordred has a point: forcing a backend to\n> > > allocate a lot of memory can be a form of DoS attack.\n> > > \n> > > regards, tom lane\n\n", "msg_date": "Tue, 3 Sep 2002 00:01:36 -0400", "msg_from": "\"Serguei Mokhov\" <mokhov@cs.concordia.ca>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "\nI haven't seen any. If no one comments in a few days, I will apply it\nbecause I need a fix before 7.3 final. I will add it to the patches\nqueue in a day or two.\n\n---------------------------------------------------------------------------\n\nSerguei Mokhov wrote:\n> Hello again,\n> \n> *any* comment on this at all?\n> \n> thanks,\n> -s\n> \n> ----- Original Message ----- \n> From: \"Serguei Mokhov\" <sa_mokho@alcor.concordia.ca>\n> Sent: September 02, 2002 4:11 AM\n> \n> > Hello,\n> > \n> > ----- Original Message ----- \n> > From: \"Bruce Momjian\" <pgman@candle.pha.pa.us>\n> > Sent: September 02, 2002 1:05 AM\n> > \n> > > Would someone submit a patch for this?\n> > \n> > Attached please find an attempt to fix the volunerability issue below.\n> > \n> > Affected files are:\n> > \n> > /src/include/libpq/libpq.h\n> > /src/include/libpq/pqformat.h\n> > /src/backend/libpq/pqformat.c\n> > /src/backend/libpq/pqcomm.c\n> > /src/backend/libpq/auth.c\n> > \n> > \"Briefly\" the changes:\n> > \n> > Main victims for the change were pq_getstring() and pq_getstr()\n> > (which calls the former) in pqformat.c and pqcomm.c. pq_getstring() is the one reading\n> > until \\0 and might possibly render the system run out of memory.\n> > \n> > Changing pq_getstring() alone would break a lot code, so I \n> > added a two more functions: pq_getstring_common() and\n> > pq_getstring_bounded(). The former is a big part of what used to be\n> > pq_getstring() and the latter is a copy of the new pq_getstring()\n> > with the string length check. Creating pq_getstring_common()\n> > was suggested by its reuse in pq_getstring() and pq_getstring_bounded()\n> > avoiding code duplication.\n> > \n> > Similar changes were done for pq_getstr(). Its common code converting\n> > to MULTIBYTE was placed in pq_getstr_multibyte() and pq_getstr() and\n> > (newly added) pq_getstr_bounded() both call it before returning a result.\n> > \n> > WRT above, two places in auth.c were changed to call pq_getstr_bounded()\n> > instead of pq_getstr() on password read. I'm not sure if\n> > there are other places where that might be needed...\n> > \n> > Might look ugly for some, but looks like a not-so-bad solution\n> > to me. If I'm completely wrong, I'd like to have some guidance then :)\n> > Please review with care. I'm off to bed.\n> > \n> > Thanks,\n> > -s\n> > \n> > PS: The patch also fixes a typo in the be-secure.c comment :)\n> > \n> > > Tom Lane wrote:\n> > > > Neil Conway <neilc@samurai.com> writes:\n> > > > > (2) The length supplied by the user is completely ignored by\n> > > > > the code, and it simply reads the input until it sees a\n> > > > > NULL terminator (read the comments in the code about 10\n> > > > > lines down.) Therefore, any sanity checking on the length\n> > > > > specified by the user is a waste of time.\n> > > > \n> > > > Agreed; the fact that the protocol requires a length word at all is just\n> > > > a hangover from the past. We can read the length word and forget it.\n> > > > \n> > > > I wonder though if it'd be worthwhile to limit the length of the string\n> > > > that we are willing to read from the client in the second step. We are\n> > > > at this point dealing with an unauthenticated user, so we should be\n> > > > untrusting. And I think Sir Mordred has a point: forcing a backend to\n> > > > allocate a lot of memory can be a form of DoS attack.\n> > > > \n> > > > regards, tom lane\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\n> \n\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Tue, 3 Sep 2002 00:04:21 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "\nI wish there was an easier way to fix this, but it seems you have done\nthe research and this is what is required.\n\n\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://207.106.42.251/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\nSerguei Mokhov wrote:\n> Hello,\n> \n> ----- Original Message ----- \n> From: \"Bruce Momjian\" <pgman@candle.pha.pa.us>\n> Sent: September 02, 2002 1:05 AM\n> \n> > Would someone submit a patch for this?\n> \n> Attached please find an attempt to fix the volunerability issue below.\n> \n> Affected files are:\n> \n> /src/include/libpq/libpq.h\n> /src/include/libpq/pqformat.h\n> /src/backend/libpq/pqformat.c\n> /src/backend/libpq/pqcomm.c\n> /src/backend/libpq/auth.c\n> \n> \"Briefly\" the changes:\n> \n> Main victims for the change were pq_getstring() and pq_getstr()\n> (which calls the former) in pqformat.c and pqcomm.c. pq_getstring() is the one reading\n> until \\0 and might possibly render the system run out of memory.\n> \n> Changing pq_getstring() alone would break a lot code, so I \n> added a two more functions: pq_getstring_common() and\n> pq_getstring_bounded(). The former is a big part of what used to be\n> pq_getstring() and the latter is a copy of the new pq_getstring()\n> with the string length check. Creating pq_getstring_common()\n> was suggested by its reuse in pq_getstring() and pq_getstring_bounded()\n> avoiding code duplication.\n> \n> Similar changes were done for pq_getstr(). Its common code converting\n> to MULTIBYTE was placed in pq_getstr_multibyte() and pq_getstr() and\n> (newly added) pq_getstr_bounded() both call it before returning a result.\n> \n> WRT above, two places in auth.c were changed to call pq_getstr_bounded()\n> instead of pq_getstr() on password read. I'm not sure if\n> there are other places where that might be needed...\n> \n> Might look ugly for some, but looks like a not-so-bad solution\n> to me. If I'm completely wrong, I'd like to have some guidance then :)\n> Please review with care. I'm off to bed.\n> \n> Thanks,\n> -s\n> \n> PS: The patch also fixes a typo in the be-secure.c comment :)\n> \n> > Tom Lane wrote:\n> > > Neil Conway <neilc@samurai.com> writes:\n> > > > (2) The length supplied by the user is completely ignored by\n> > > > the code, and it simply reads the input until it sees a\n> > > > NULL terminator (read the comments in the code about 10\n> > > > lines down.) Therefore, any sanity checking on the length\n> > > > specified by the user is a waste of time.\n> > > \n> > > Agreed; the fact that the protocol requires a length word at all is just\n> > > a hangover from the past. We can read the length word and forget it.\n> > > \n> > > I wonder though if it'd be worthwhile to limit the length of the string\n> > > that we are willing to read from the client in the second step. We are\n> > > at this point dealing with an unauthenticated user, so we should be\n> > > untrusting. And I think Sir Mordred has a point: forcing a backend to\n> > > allocate a lot of memory can be a form of DoS attack.\n> > > \n> > > regards, tom lane\n\n[ Attachment, skipping... ]\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\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 4 Sep 2002 18:05:40 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> I wish there was an easier way to fix this, but it seems you have done\n> the research and this is what is required.\n\nThis is awfully messy. There's got to be a cleaner way of divvying up\nthis code...\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 04 Sep 2002 18:30:00 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length " }, { "msg_contents": "Quoting Tom Lane <tgl@sss.pgh.pa.us>:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > I wish there was an easier way to fix this, but it seems you have done\n> > the research and this is what is required.\n> \n> This is awfully messy. There's got to be a cleaner way of divvying up\n> this code...\n\nCould you point out, what's exactly unclean? Most importantly,\nwhat would be the way you'd fix it?\n\nThank you,\n\n-- \nSerguei A. Mokhov, <mailto: mokhov @ cs.concordia.ca>\nComputer Science, Concordia University\n\n-------------------------------------------------\nThis mail sent through IMP: http://horde.org/imp/\n", "msg_date": "Wed, 4 Sep 2002 18:37:44 -0400", "msg_from": "Serguei Mokhov <mokhov@cs.concordia.ca>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length " }, { "msg_contents": "Quoting Tom Lane <tgl@sss.pgh.pa.us>:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > I wish there was an easier way to fix this, but it seems you have done\n> > the research and this is what is required.\n> \n> This is awfully messy. There's got to be a cleaner way of divvying up\n> this code...\n\nJust to clarify a bit on my solution in case my English didn't get through \nproperly the first time...\n\nI simply provided two versions of pq_getstr - pq_getstr() with the same \nbehaviour as before (read until input isn't over by \\0) and pq_getstr_bounded() \nthat reads up to a certain limit or till \\0. Functions needed split, IMNSHO,\nbecause grep of the source gave more invocations of pq_getstr, which I was \nafaraid to break, so that's why two functions.\n\nI can justify the rest as well, if you wish. If you are positive, be the change \nin one function only it would not break anything, then the cleaner solution is \njust to change that one function - pg_gestring() invoked directly by\npg_getstr().\n\n-s\n\n-- \nSerguei A. Mokhov, <mailto: mokhov @ cs.concordia.ca>\nComputer Science, Concordia University\n\n-------------------------------------------------\nThis mail sent through IMP: http://horde.org/imp/\n", "msg_date": "Wed, 4 Sep 2002 18:51:03 -0400", "msg_from": "Serguei Mokhov <mokhov@cs.concordia.ca>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length " }, { "msg_contents": "Serguei Mokhov <mokhov@cs.concordia.ca> writes:\n> Could you point out, what's exactly unclean? Most importantly,\n> what would be the way you'd fix it?\n\nWhat's bugging me is that even though the patch goes out of its way to\nshare code, there still seems to be a lot of duplicate code. You're not\ngetting the full benefit of sharing code between both cases, yet you\nstill seem to be paying the price of extra code complexity compared to\njust copy-paste-and-modify.\n\nWhat I'm thinking about is\n\n-- pq_getstr takes a length limit parameter, which is (say) 0 for \"no\nlimit\". Since it's only called in one place, we can just change its\nAPI; there's hardly any point in providing a backward-compatible routine.\n(BTW, I agree with your implementation choice to check the limit only\nonce per bufferload, and thus have a fuzzy limit, but this needs to be\ndocumented.)\n\n-- pq_getstring becomes pq_getstring_bounded, with a limit parameter\nthat it just passes down.\n\n-- We can \"#define pq_getstring(buf) pq_getstring_bounded(buf, 0)\" to\navoid changing the call sites that want unbounded input (not that there\nare that many of 'em, but we may as well provide the macro).\n\nWill adjust your patch in this way and apply.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 04 Sep 2002 19:04:32 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length " }, { "msg_contents": "\nThis was already applied by Tom Lane. I was not sure you were informed.\n\n---------------------------------------------------------------------------\n\n\nrevision 1.91\ndate: 2002/09/04 23:31:34; author: tgl; state: Exp; lines: +4 -5\nGuard against send-lots-and-lots-of-data DoS attack from unauthenticated\nusers, by limiting the length of string we will accept for a password.\nPatch by Serguei Mokhov, some editorializing by Tom Lane.\n\n---------------------------------------------------------------------------\n\nSerguei Mokhov wrote:\n> Hello,\n> \n> ----- Original Message ----- \n> From: \"Bruce Momjian\" <pgman@candle.pha.pa.us>\n> Sent: September 02, 2002 1:05 AM\n> \n> > Would someone submit a patch for this?\n> \n> Attached please find an attempt to fix the volunerability issue below.\n> \n> Affected files are:\n> \n> /src/include/libpq/libpq.h\n> /src/include/libpq/pqformat.h\n> /src/backend/libpq/pqformat.c\n> /src/backend/libpq/pqcomm.c\n> /src/backend/libpq/auth.c\n> \n> \"Briefly\" the changes:\n> \n> Main victims for the change were pq_getstring() and pq_getstr()\n> (which calls the former) in pqformat.c and pqcomm.c. pq_getstring() is the one reading\n> until \\0 and might possibly render the system run out of memory.\n> \n> Changing pq_getstring() alone would break a lot code, so I \n> added a two more functions: pq_getstring_common() and\n> pq_getstring_bounded(). The former is a big part of what used to be\n> pq_getstring() and the latter is a copy of the new pq_getstring()\n> with the string length check. Creating pq_getstring_common()\n> was suggested by its reuse in pq_getstring() and pq_getstring_bounded()\n> avoiding code duplication.\n> \n> Similar changes were done for pq_getstr(). Its common code converting\n> to MULTIBYTE was placed in pq_getstr_multibyte() and pq_getstr() and\n> (newly added) pq_getstr_bounded() both call it before returning a result.\n> \n> WRT above, two places in auth.c were changed to call pq_getstr_bounded()\n> instead of pq_getstr() on password read. I'm not sure if\n> there are other places where that might be needed...\n> \n> Might look ugly for some, but looks like a not-so-bad solution\n> to me. If I'm completely wrong, I'd like to have some guidance then :)\n> Please review with care. I'm off to bed.\n> \n> Thanks,\n> -s\n> \n> PS: The patch also fixes a typo in the be-secure.c comment :)\n> \n> > Tom Lane wrote:\n> > > Neil Conway <neilc@samurai.com> writes:\n> > > > (2) The length supplied by the user is completely ignored by\n> > > > the code, and it simply reads the input until it sees a\n> > > > NULL terminator (read the comments in the code about 10\n> > > > lines down.) Therefore, any sanity checking on the length\n> > > > specified by the user is a waste of time.\n> > > \n> > > Agreed; the fact that the protocol requires a length word at all is just\n> > > a hangover from the past. We can read the length word and forget it.\n> > > \n> > > I wonder though if it'd be worthwhile to limit the length of the string\n> > > that we are willing to read from the client in the second step. We are\n> > > at this point dealing with an unauthenticated user, so we should be\n> > > untrusting. And I think Sir Mordred has a point: forcing a backend to\n> > > allocate a lot of memory can be a form of DoS attack.\n> > > \n> > > regards, tom lane\n\n[ Attachment, skipping... ]\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\n-- \n Bruce Momjian | http://candle.pha.pa.us\n pgman@candle.pha.pa.us | (610) 359-1001\n + If your life is a hard drive, | 13 Roberts Road\n + Christ can be your backup. | Newtown Square, Pennsylvania 19073\n", "msg_date": "Wed, 11 Sep 2002 20:04:18 -0400 (EDT)", "msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" }, { "msg_contents": "Quoting Bruce Momjian <pgman@candle.pha.pa.us>:\n\n> \n> This was already applied by Tom Lane. I was not sure you were informed.\n\nYes, I was. Thank you.\n\n-s\n\n> ---------------------------------------------------------------------------\n> \n> \n> revision 1.91\n> date: 2002/09/04 23:31:34; author: tgl; state: Exp; lines: +4 -5\n> Guard against send-lots-and-lots-of-data DoS attack from unauthenticated\n> users, by limiting the length of string we will accept for a password.\n> Patch by Serguei Mokhov, some editorializing by Tom Lane.\n\n-- \nSerguei A. Mokhov, <mailto: mokhov @ cs.concordia.ca>\nComputer Science, Concordia University\n\n-------------------------------------------------\nThis mail sent through IMP: http://horde.org/imp/\n", "msg_date": "Wed, 11 Sep 2002 22:07:23 -0400", "msg_from": "Serguei Mokhov <mokhov@cs.concordia.ca>", "msg_from_op": false, "msg_subject": "Re: fix for palloc() of user-supplied length" } ]
[ { "msg_contents": "Hi,\n\nNow that we have prepared statements, should the foreign key code be changed\nto use them? I think it's highly likely that they will be reused in a\nconnection. Might be an idea.\n\nAlso, what if the person wants to prepare a view? Should they have to\nPREPARE <view select> or should we automatically prepare them the first time\nthey're used in a connection? Or should we offer a PREPARE VIEW <viewname>\noption?\n\nI personally think the second option would be neat.\n\nChris\n\n", "msg_date": "Wed, 28 Aug 2002 14:53:46 +0800", "msg_from": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au>", "msg_from_op": true, "msg_subject": "PREPARE, FK's and VIEWs" }, { "msg_contents": "On Wed, Aug 28, 2002 at 02:53:46PM +0800, Christopher Kings-Lynne wrote:\n> \n> Now that we have prepared statements, should the foreign key code be changed\n> to use them? I think it's highly likely that they will be reused in a\n> connection. Might be an idea.\n\n No FK, but more common is integrate query plan cache to SPI (saveplan).\n I don't how it's in the current implementation.\n\n Karel\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": "Wed, 28 Aug 2002 09:11:04 +0200", "msg_from": "Karel Zak <zakkr@zf.jcu.cz>", "msg_from_op": false, "msg_subject": "Re: PREPARE, FK's and VIEWs" }, { "msg_contents": "\"Christopher Kings-Lynne\" <chriskl@familyhealth.com.au> writes:\n> Now that we have prepared statements, should the foreign key code be changed\n> to use them? I think it's highly likely that they will be reused in a\n> connection. Might be an idea.\n\nWaste of time; the FK code already does this for itself.\n\n> Also, what if the person wants to prepare a view?\n\nWhat does that mean? A view isn't preparable in isolation from the\nquery that's going to use it.\n\n\t\t\tregards, tom lane\n", "msg_date": "Wed, 28 Aug 2002 09:30:19 -0400", "msg_from": "Tom Lane <tgl@sss.pgh.pa.us>", "msg_from_op": false, "msg_subject": "Re: PREPARE, FK's and VIEWs " }, { "msg_contents": "On Wed, 28 Aug 2002, Christopher Kings-Lynne wrote:\n\n> Now that we have prepared statements, should the foreign key code be changed\n> to use them? I think it's highly likely that they will be reused in a\n> connection. Might be an idea.\n\nIt might be worth doing eventually just to standardize, but it should\nsave query plans internally already.\n\n", "msg_date": "Wed, 28 Aug 2002 07:56:24 -0700 (PDT)", "msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>", "msg_from_op": false, "msg_subject": "Re: PREPARE, FK's and VIEWs" } ]
[ { "msg_contents": "\n> My guess, seeing as very few probably use LIMIT and FOR UPDATE together,\n> is to swap them and document it in the release notes. Was I correct in\n> my guess?\n\nSounds good to me.\n\nAndreas\n", "msg_date": "Wed, 28 Aug 2002 09:29:46 +0200", "msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>", "msg_from_op": true, "msg_subject": "Re: [SQL] LIMIT 1 FOR UPDATE or FOR UPDATE LIMIT 1?" } ]