threads listlengths 1 2.99k |
|---|
[
{
"msg_contents": "\nI have been looking forward to schemas (namespaces) for sometime.\n\nI had not been able to decipher the schema symantics necessary for a\ndefault schema, until I hacked the source a bit.\n\nNow I know that the rules to get a default schema using\n\tdb_user_namespace = true\n\tsearch_path = '$user,public'\nin postgresql.conf\n\n\nthe user logs in as\n\tpsql testdb testuser\n\nbut the userid is really\n\ttestuser@testdb\n\nif a schema named \"testuser@testdb\" exists\nthe user gets it as a default schema\notherwise the default schema is public;\n\nand if the user is a global user\nthe login is\n\tpsql testdb globuser@\nand the required schema is \"globuser\"\n\nthis use of \"@\" in the default schema is a bit counter intuitive\n\n\nso I offer the following patch against CVS\n\nit tries the userid as found in pg_user,\nthen strips off the @db stuff if possible\n\n--- src/backend/catalog/namespace.c\tThu Oct 17 21:23:34 2002\n+++ src/backend/catalog/namespace_2.c\tThu Oct 17 21:21:52 2002\n@@ -1386,8 +1386,10 @@\n \t\t\tif (HeapTupleIsValid(tuple))\n \t\t\t{\n \t\t\t\tchar\t *uname;\n+\t\t\t\tchar\t *uname_local;\n \t\t\t\tuname = NameStr(((Form_pg_shadow) \nGETSTRUCT(tuple))->usename);\n \t\t\t\tnamespaceId = \nGetSysCacheOid(NAMESPACENAME,\n \t\t\t\t\t\t\t\t\t\t\t \nCStringGetDatum(uname),\n \t\t\t\t\t\t\t\t\t\t\t \n0, 0, 0);\n@@ -1396,7 +1398,25 @@\n \t\t\t\t\t!intMember(namespaceId, \noidlist) &&\n \t\t\t\t\t \npg_namespace_aclcheck(namespaceId, userId,\n \t\t\t\t\t\t\t\t\t\t \nACL_USAGE) == ACLCHECK_OK)\n+\t\t\t\t{\n \t\t\t\t\toidlist = lappendi(oidlist, \nnamespaceId);\n+\t\t\t\t}\n+\t\t\t\telse\n+\t\t\t\t{\t \n+\t\t\t\t\tuname_local = (char *) \nmalloc(strlen(uname)); +\t\t\t\t\tstrcpy \n(uname_local,uname);\n+\t\t\t\t\tuname_local = \nstrtok(uname_local,\"@\");\n+\t\t\t\t\tnamespaceId = \nGetSysCacheOid(NAMESPACENAME,\n+\t\t\t\t\t\t\t\t \nCStringGetDatum(uname_local),\n+\t\t\t\t\t\t\t\t\t \n0, 0, 0);\n+\t\t\t\t\tif (OidIsValid(namespaceId) &&\n+\t\t\t\t\t\t!intMember(namespaceId, \noidlist) &&\n+\t\t\t\t\t\t \npg_namespace_aclcheck(namespaceId, userId,\n+\t\t\t\t\t\t\t\t\t\t \nACL_USAGE) == ACLCHECK_OK)\n+\t\t\t\t\t\toidlist = \nlappendi(oidlist, namespaceId);\n+\t\t\t\t\tfree(uname_local);\n+\t\t\t\t}\n+\n \t\t\t}\n \t\t}\n \t\telse\n\n\n\n\n",
"msg_date": "Thu, 17 Oct 2002 21:42:38 -0400",
"msg_from": "Carl Anderson <candrsn@mindspring.com>",
"msg_from_op": true,
"msg_subject": "default namespace (schema) confusion"
},
{
"msg_contents": "Carl Anderson <candrsn@mindspring.com> writes:\n> this use of \"@\" in the default schema is a bit counter intuitive\n> so I offer the following patch against CVS\n\nHmm, this seems like a wart, but then the db_user_namespace feature\nis an acknowledged wart already.\n\nI think I'd be willing to hold still for this if it happens only when\ndb_user_namespace is on. Comments anyone?\n\n> +\t\t\t\t\tuname_local = (char *) \n> malloc(strlen(uname)); +\t\t\t\t\tstrcpy \n> (uname_local,uname);\n\nUse pstrdup, please.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 17 Oct 2002 22:20:01 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: default namespace (schema) confusion "
},
{
"msg_contents": "Tom Lane wrote:\n> Carl Anderson <candrsn@mindspring.com> writes:\n> > this use of \"@\" in the default schema is a bit counter intuitive\n> > so I offer the following patch against CVS\n> \n> Hmm, this seems like a wart, but then the db_user_namespace feature\n> is an acknowledged wart already.\n> \n> I think I'd be willing to hold still for this if it happens only when\n> db_user_namespace is on. Comments anyone?\n\nI dislike double-testing the username in schema areas but not other\nplaces. Seems if we do it, we should do it consistently for all\nusername references, or not at all.\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, 17 Oct 2002 22:26:35 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: default namespace (schema) confusion"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> I dislike double-testing the username in schema areas but not other\n> places. Seems if we do it, we should do it consistently for all\n> username references, or not at all.\n\nWhat other places do we have an explicit dependence on the username?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 17 Oct 2002 22:36:43 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: default namespace (schema) confusion "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > I dislike double-testing the username in schema areas but not other\n> > places. Seems if we do it, we should do it consistently for all\n> > username references, or not at all.\n> \n> What other places do we have an explicit dependence on the username?\n\nUh, what about CREATE/ALTER/DROP USER, and pg_hba.conf. Of course,\nthose are more admin, but isn't the schema also sort of admin too?\n\nI am also concerned about opening cases where a CURRENT_USER test\ndoesn't match the user's schema.\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, 17 Oct 2002 22:39:02 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: default namespace (schema) confusion"
}
] |
[
{
"msg_contents": "Could anyone please tell me if we found a solution? It seems I missed\nany discussion or whatsoever. The only things I know is that\npostgresql.org still has bison 1.35 installed and 7.3 release is\nnearing.\n\nMichael\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Fri, 18 Oct 2002 17:34:17 +0200",
"msg_from": "Michael Meskes <meskes@postgresql.org>",
"msg_from_op": true,
"msg_subject": "ECPG and bison"
},
{
"msg_contents": "\nYes, we are going to upgrade to bison 1.50 on our main server as soon as\nMarc gets it installed. If you want to make your changes and commit\nthose now, you can, because I assume your bison 1.50 output will get\ninto CVS. I have bison 1.50 here too.\n\n---------------------------------------------------------------------------\n\nMichael Meskes wrote:\n> Could anyone please tell me if we found a solution? It seems I missed\n> any discussion or whatsoever. The only things I know is that\n> postgresql.org still has bison 1.35 installed and 7.3 release is\n> nearing.\n> \n> Michael\n> -- \n> Michael Meskes\n> Michael@Fam-Meskes.De\n> Go SF 49ers! Go Rhein Fire!\n> Use Debian GNU/Linux! Use PostgreSQL!\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 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, 18 Oct 2002 11:48:33 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison"
},
{
"msg_contents": "On Fri, Oct 18, 2002 at 11:48:33AM -0400, Bruce Momjian wrote:\n> Yes, we are going to upgrade to bison 1.50 on our main server as soon as\n> Marc gets it installed. If you want to make your changes and commit\n\nOkay.\n\n> those now, you can, because I assume your bison 1.50 output will get\n> into CVS. I have bison 1.50 here too.\n\nThe changes are there already, I just have to fold the ecpg.big branch\ninto HEAD.\n\nMichael\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Fri, 18 Oct 2002 18:56:56 +0200",
"msg_from": "Michael Meskes <meskes@postgresql.org>",
"msg_from_op": true,
"msg_subject": "Re: ECPG and bison"
},
{
"msg_contents": "Michael Meskes <meskes@postgresql.org> writes:\n>> those now, you can, because I assume your bison 1.50 output will get\n>> into CVS. I have bison 1.50 here too.\n\n> The changes are there already, I just have to fold the ecpg.big branch\n> into HEAD.\n\nProbably you should refrain from doing that until Marc installs 1.50\nat postgresql.org, else the nightly snapshots will be broken, to no\none's advantage.\n\nMarc, can you do something about installing bison 1.50 soon?\n\n(Guess I'd better get it loaded on my own machines, too...)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 18 Oct 2002 14:51:12 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison "
},
{
"msg_contents": "On Fri, 2002-10-18 at 13:51, Tom Lane wrote:\n> Michael Meskes <meskes@postgresql.org> writes:\n> >> those now, you can, because I assume your bison 1.50 output will get\n> >> into CVS. I have bison 1.50 here too.\n> \n> > The changes are there already, I just have to fold the ecpg.big branch\n> > into HEAD.\n> \n> Probably you should refrain from doing that until Marc installs 1.50\n> at postgresql.org, else the nightly snapshots will be broken, to no\n> one's advantage.\n> \n> Marc, can you do something about installing bison 1.50 soon?\n> \n> (Guess I'd better get it loaded on my own machines, too...)\nLooks like threre is a 1.75 out (based on a FreeBSD PR I just saw fly\nby). \n\nLER\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-- \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": "18 Oct 2002 15:03:52 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison"
},
{
"msg_contents": "Larry Rosenman wrote:\n> On Fri, 2002-10-18 at 13:51, Tom Lane wrote:\n> > Michael Meskes <meskes@postgresql.org> writes:\n> > >> those now, you can, because I assume your bison 1.50 output will get\n> > >> into CVS. I have bison 1.50 here too.\n> > \n> > > The changes are there already, I just have to fold the ecpg.big branch\n> > > into HEAD.\n> > \n> > Probably you should refrain from doing that until Marc installs 1.50\n> > at postgresql.org, else the nightly snapshots will be broken, to no\n> > one's advantage.\n> > \n> > Marc, can you do something about installing bison 1.50 soon?\n> > \n> > (Guess I'd better get it loaded on my own machines, too...)\n> Looks like threre is a 1.75 out (based on a FreeBSD PR I just saw fly\n> by). \n\nI can confirm that:\n\n\thttp://ftp.gnu.org/gnu/bison/\n\nAttached is the Changelog since 1.50.\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\n2002-10-14 Akim Demaille <akim@epita.fr>\n\n\tVersion 1.75.\n\n2002-10-14 Akim Demaille <akim@epita.fr>\n\n\t* tests/Makefile.am (maintainer-check-posix): New.\n\n2002-10-14 Akim Demaille <akim@epita.fr>\n\n\t* data/glr.c [YYDEBUG] (YYLEFTMOST_STATE): Initialize the yyloc\n\tmember.\n\n2002-10-14 Akim Demaille <akim@epita.fr>\n\n\t* src/tables.c (table_ninf_remap): base -> tab.\n\tReported by Matt Rosing.\n\n2002-10-14 Paul Eggert <eggert@twinsun.com>\n\n\t* tests/action.at, tests/calc.at, tests/conflicts.at,\n\ttests/cxx-type.at, tests/headers.at, tests/input.at,\n\ttests/regression.at, tests/synclines.at, tests/torture.at:\n\tSay \"bison -o foo.c foo.y\", not \"bison foo.y -o foo.c\",\n\tso that the tests still work even if POSIXLY_CORRECT is set.\n\t* doc/bison.texinfo (Rpcalc Compile, Invocation): Likewise.\n\n\t* data/c.m4 (b4_int_type): Use yysigned_char instead of signed char,\n\tfor portability to K&R hosts. Fix typo: signed char is guaranteed\n\tonly to 127, not to 128.\n\t* data/glr.c (yysigned_char): New type.\n\t* data/yacc.c (yysigned_char): Likewise.\n\t* tests/regression.at (Web2c Actions): signed char -> yysigned_char.\n\n2002-10-13 Paul Eggert <eggert@twinsun.com>\n\n\t* data/yacc.c (yyparse): Rewrite to avoid \"comparison is always\n\ttrue due to limited range of data type\" warning from GCC.\n\n\t* data/c.m4 (b4_token_defines): Protect against double-inclusion\n\tby wrapping enum yytokentype's definition inside #ifndef\n\tYYTOKENTYPE. This undoes a bug I introduced on 2002-10-12.\n\n2002-10-13 Akim Demaille <akim@epita.fr>\n\n\t* data/glr.c (yyglrShiftDefer, yyaddDeferredAction, yydoAction):\n\tUn yy- yyrhs to avoid the name clash with the global YYRHS.\n\n2002-10-13 Akim Demaille <akim@epita.fr>\n\n\t* Makefile.maint: Update from Autoconf 2.54.\n\t* m4/strerror_r.m4 (AC_FUNC_STRERROR_R): Remove, shipped with 2.54.\n\n2002-10-13 Akim Demaille <akim@epita.fr>\n\n\t* src/print.c (print_state): Separate the list of solved conflicts\n\tfrom the other items.\n\t* tests/conflicts.at (Resolved SR Conflicts): Adjust.\n\n2002-10-13 Akim Demaille <akim@epita.fr>\n\n\tLet nondeterministic skeletons be usable with deterministic\n\ttables.\n\n\tWith the patch, GAWK compiled by GCC without -O2 passes its test\n\tsuite using a GLR parser driven by LALR tables. It fails with -O2\n\tbecause `struct stat' gives two different answers on my machine:\n\t88 (definition of an auto var) and later 96 (memset on this var).\n\tHence the stack is badly corrumpted. The headers inclusion is to\n\tblame: if I move the awk.h inclusion before GLR's system header\n\tinclusion, the two struct stat have the same size.\n\n\t* src/tables.c (pack_table): Always create conflict_table.\n\t(token_actions): Always create conflict_list.\n\t* data/glr.c (YYFLAG): Remove, unused.\n\n2002-10-13 Akim Demaille <akim@epita.fr>\n\n\t* configure.ac (AC_GNU_SOURCE): Use it instead of hand written code.\n\t(O0FLAGS): New.\n\t(VALGRIND, GXX): New.\n\t* tests/atlocal.in (CFLAGS): Use O0FLAGS.\n\t* tests/bison.in: Run $PREBISON a pre-command.\n\t* tests/Makefile.am (maintainer-check, maintainer-check-valgrind)\n\t(maintainer-check-g++): New.\n\t* Makefile.am (maintainer-check): New.\n\n2002-10-13 Akim Demaille <akim@epita.fr>\n\n\t* data/glr.c: Formatting changes.\n\tTweak some trace messages to match yacc.c's.\n\n2002-10-13 Akim Demaille <akim@epita.fr>\n\n\tGLR parsers sometimes raise parse errors instead of performing the\n\tdefault reduction.\n\tReported by Charles-Henry de Boysson.\n\n\t* tests/calc.at (_AT_CHECK_CALC, _AT_CHECK_CALC_ERROR): Don't\n\tcheck the length of the\ttraces when %glr.\n\t(_AT_CHECK_CALC_ERROR): Also skip `^Stack' lines, coming from\n\tGLR's traces.\n\t(AT_CHECK_CALC_LALR, AT_CHECK_CALC_GLR): New.\n\tTest GLR parsers.\n\t* data/glr.c (YYLEFTMOST_STATE): Fix its value.\n\t(yyltype): Remove the yy prefix from the member names.\n\t(yytable): Complete its comment.\n\t(yygetLRActions): Map error action number from YYTABLE from\n\tYYTABLE_NINF to 0.\n\t(yyisErrorAction): No longer compare YYACTION to YYPACT_NINF\n\t(which was a bug: it should have been YYTABEL_NINF, and yet it was\n\tnot satisfying as we could compare an YYACTION computed from\n\tYYDEFACT to YYTABLE_NINF although they are unrelated): 0 is the\n\tonly value for error actions.\n\t(yyreportParseError): In verbose parse error messages, don't issue\n\t`error' in the list of expected tokens.\n\t* data/yacc.c (yyparse) <yybackup>: Rewrite the decoding of the\n\tnext action to perform to match glr.c's decoding.\n\t(yytable): Complete its comment.\n\n2002-10-13 Paul Eggert <eggert@twinsun.com>\n\n\tFix problem reported by Henrik Grubbstroem in\n\t<http://mail.gnu.org/pipermail/bug-bison/2002-October/001670.html>:\n\t\"nonterm: { $$ = 123; } { $$ = $1; };\" was wrongly rejected,\n\tbecause the Bison parser reads the second action before reducing\n\tthe first one.\n\t* src/scan-gram.l (rule_length): New static var.\n\tUse it to keep track of the rule length in the scanner, since\n\twe can't expect the parser to be in lock-step sync with the scanner.\n\t(handle_action_dollar, handle_action_at): Use this var.\n\t* tests/actions.at (Exotic Dollars): Test for the problem.\n\n2002-10-12 Paul Eggert <eggert@twinsun.com>\n\n\t* lib/timevar.c [! IN_GCC && HAVE_SYS_TIME_H]: Include <sys/time.h>.\n\t* m4/timevar.m4 (BISON_PREREQ_TIMEVAR): Check for <sys/time.h>.\n\tInclude <sys/time.h> when checking for clock_t and struct tms.\n\tUse same include order as source.\n\tThis is for the SunOS 4.1.4 porting bug reported by Peter Klein in\n\t<http://mail.gnu.org/pipermail/bug-bison/2002-October/001674.html>.\n\n\t* lib/timevar.c: Update copyright date and clarify comments.\n\t(get_time) [IN_GCC]: Keep the GCC version for reference.\n\n\t* lib/timevar.c, lib/timevar.h, lib/timevar.def: Import\n\tGCC version as of today, then merge Bison's changes.\n\tChange \"GCC\" to \"Bison\" in copyright notice. timevar.def's\n\tauthor is Akim, so change that too.\n\n\t* src/reader.c (grammar_current_rule_check):\n\tDon't worry about the default action if $$ is untyped.\n\tPrevents bogus warnings reported by Jim Gifford in\n\t<http://mail.gnu.org/pipermail/bug-bison/2002-October/001673.html>.\n\n\t* data/c.m4 (b4_token_enum): Do not define YYTOKENTYPE.\n\t* data/glr.c, data/lalr1.cc, data/yacc.c:\n\tOutput token definitions before the first part of user declarations.\n\tFixes compatibility problem reported by Jim Gifford for kbd in\n\t<http://mail.gnu.org/pipermail/bug-bison/2002-October/001672.html>.\n\n2002-10-11 Paul Eggert <eggert@twinsun.com>\n\n\t* data/yacc.c (yyreport_parse_error): Remove, putting its body into...\n\t(yyparse): here. This undoes some of the 2002-07-25 change.\n\tCompatibility problem reported by Ralf S. Engelschall with\n\tOSSP cfg <http://www.ossp.org/pkg/lib/cfg/>.\n\n2002-10-11 Akim Demaille <akim@epita.fr>\n\n\t* tests/regression.at Characters Escapes): New.\n\t* src/scan-gram.l (SC_ESCAPED_CHARACTER): Accept \\' in strings and\n\tcharacters.\n\tReported by Jan Nieuwenhuizen.\n\n2002-10-11 Akim Demaille <akim@epita.fr>\n\n\t* po/id.po: New.\n\n2002-10-10 Paul Eggert <eggert@twinsun.com>\n\n\tPortability fixes for bitsets; this also avoids several GCC\n\twarnings.\n\n\t* lib/abitset.c: Include <stddef.h>, for offsetof.\n\t* lib/lbitset.c: Likewise.\n\n\t* lib/abitset.c (abitset_bytes): Return a size that is aligned\n\tproperly for vectors of objects. Do not assume that adding a\n\theader size to a multiple of a word size yields a value that is\n\tproperly aligned for the whole union.\n\t* lib/bitsetv.c (bitsetv_alloc): Likewise.\n\n\t* lib/bitset_stats.c (bitset_stats_bytes): Adjust to new,\n\tunique names for structures.\n\t* lib/ebitset.c (ebitset_bytes): Likewise.\n\t* lib/lbitset.c (lbitset_bytes): Likewise.\n\n\t* lib/abitset.c (abitset_ones, abitset_zero, abitset_empty_p,\n\tabitset_copy1, abitset_not, abitset_equal_p, abitset_subset_p,\n\tabitset_disjoint_p, abitset_and, abitset_and_cmp, abitset_andn,\n\tabitset_andn_cmp, abitset_or, abitset_or_cmp, abitset_xor,\n\tabitset_xor_cmp, abitset_and_or, abitset_and_or_cmp,\n\tabitset_andn_or, abitset_andn_or_cmp, abitset_or_and,\n\tabitset_or_and_cmp, abitset_copy): Supply prototype decls,\n\tto improve the type-checking that GCC can do.\n\t* lib/bitset.c (bitset_op4_cmp): Likewise.\n\t* lib/bitset_stats.c (bitset_stats_count,\n\tbitset_stats_empty_p, bitset_stats_ones, bitset_stats_zero,\n\tbitset_stats_copy, bitset_stats_disjoint_p,\n\tbitset_stats_equal_p, bitset_stats_not, bitset_stats_subset_p,\n\tbitset_stats_and, bitset_stats_and_cmp, bitset_stats_andn,\n\tbitset_stats_andn_cmp, bitset_stats_or, bitset_stats_or_cmp,\n\tbitset_stats_xor, bitset_stats_xor_cmp, bitset_stats_and_or,\n\tbitset_stats_and_or_cmp, bitset_stats_andn_or,\n\tbitset_stats_andn_or_cmp, bitset_stats_or_and,\n\tbitset_stats_or_and_cmp): Likewise.\n\t* lib/lbitset.c (lbitset_and, lbitset_and_cmp, lbitset_andn,\n\tlbitset_andn_cmp, lbitset_or, lbitset_or_cmp, lbitset_xor,\n\tlbitset_xor_cmp, lbitset_empty_p, lbitset_ones, lbitset_not,\n\tlbitset_subset_p, lbitset_disjoint_p, debug_lbitset): Likewise.\n\n\t* lib/abitset.h: Include bitset.h, not bbitset.h.\n\t* lib/ebitset.h: Likewise.\n\t* lib/lbitset.h: Likewise.\n\n\t* lib/bbitset.h: (enum_bitset_ops, enum_bitset_type): New types.\n\tAll instances of parameters of type enum bitset_opts are now of\n\ttype enum_bitset_opts, to conform to the C Standard, and similarly\n\tfor enum_bitset_type.\n\t* lib/ebitset.c (enum_ebitset_find_mode): Likewise.\n\t* lib/lbitset.c (enum_lbitset_find_mode): Likewise.\n\n\tDo not use \"struct bitset_struct\" to mean different things in\n\tdifferent modules. Not only is this confusing, it violates\n\tthe C Standard, which requires that structure types in different\n\tmodules must be compatible if one is to be passed to the other.\n\t* lib/bbitset.h (bitset): Now points to a union, not to a struct.\n\tAll instances of \"struct bitset_struct *\" replaced with \"bitset\".\n\t* lib/bitset.h (struct bitset_struct): Remove, replacing with....\n\t(union bitset_union, struct abitset_struct, struct ebitset_struct,\n\tstruct lbitset_struct, struct bitset_stats_struct): New types.\n\tAll uses of struct bitset_struct changed to union bitset_union,\n\tetc.\n\t* lib/abitset.c\t(struct abitset_struct, abitset,\n\tstruct bitset_struct): Remove.\n\t* lib/bitset_stats.c (struct bitset_stats_struct, bitset_stats,\n\tstruct bitset_struct): Remove.\n\t* lib/ebitset.c (struct ebitset_struct, ebitset, struct\n\tbitset_struct): Remove.\n\t* lib/lbitset.c (struct lbitset_struct, lbitset, bitset_struct):\n\tLikewise.\n\n\tDo not call a function of type T using a call that assumes the\n\tfunction is of a different type U. Standard C requires that a\n\tfunction must be called with a type that is compatible with its\n\tdefinition.\n\t* lib/bbitset.h (bitset_and_or_, bitset_andn_or_, bitset_or_and_):\n\tNew decls.\n\t* lib/bitset.c (bitset_and_or_, bitset_andn_or_, bitset_or_and_):\n\tNew functions.\n\t* lib/ebitset.c (PFV): Remove.\n\t* lib/lbitset.c (PFV): Likewise.\n\t* lib/ebitset.c (ebitset_and, ebitset_andn, ebitset_or,\n\tebitset_xor, ebitset_copy, ebitset_ones, ebitset_empty_p): New\n\tdecls.\n\t(ebitset_and, ebitset_andn, ebitset_or, ebitset_xor): New functions.\n\t(ebitset_vtable): Use them.\n\t* lib/lbitset.c (lbitset_and, lbitset_andn, lbitset_or,\n\tlbitset_xor): New functions.\n\t(lbitset_vtable): Use them.\n\n\t* lib/bitset.h (bitset_next, bitset_prev, bitset_only_set_p):\n\tDeclare.\n\n\t* lib/bitsetv.c (bitsetv_alloc): Add a cast to (void *) to avoid a\n\tGCC warning.\n\t* lib/lbitset.c (LBITSET_CURRENT1): Likewise.\n\tUse offsetof, for simplicity.\n\n2002-10-06 Paul Eggert <eggert@twinsun.com>\n\n\t* lib/bitset.h (bitset_reset): Do not assume that bitset_word is\n\tthe same width as int. This reapplies a hunk of the 2002-08-12 patch\n\t<http://mail.gnu.org/pipermail/bison-patches/2002-August/001111.html>,\n\twhich was inadvertently undone by the 2002-09-30 patch.\n\t* lib/lbitset.c (debug_lbitset): Do not assume that bitset_word is\n\tthe same width as int.",
"msg_date": "Fri, 18 Oct 2002 16:16:29 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Larry Rosenman wrote:\n>> Looks like threre is a 1.75 out (based on a FreeBSD PR I just saw fly\n>> by). \n\n> I can confirm that:\n> \thttp://ftp.gnu.org/gnu/bison/\n> Attached is the Changelog since 1.50.\n\nUrgh. Nothing like major releases 2 weeks apart to give one a nice warm\ncomfy feeling about the stability of one's tools.\n\nOffhand it looks like these are nearly all bug fixes, and so we'd better\nstandardize on 1.75 not 1.50. But I wonder what will be out next\nweek...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 18 Oct 2002 16:49:41 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison "
},
{
"msg_contents": "Peter Eisentraut wrote:\n> Michael Meskes writes:\n> \n> > Could anyone please tell me if we found a solution? It seems I missed\n> > any discussion or whatsoever. The only things I know is that\n> > postgresql.org still has bison 1.35 installed and 7.3 release is\n> > nearing.\n> \n> I suggest that you merge your branch as soon as you are ready and force\n> the issue a bit. It's not like the bison installation on\n> postgresql.org can make the world stop.\n\nAgreed. We need to push this.\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, 18 Oct 2002 18:08:25 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison"
},
{
"msg_contents": "Michael Meskes writes:\n\n> Could anyone please tell me if we found a solution? It seems I missed\n> any discussion or whatsoever. The only things I know is that\n> postgresql.org still has bison 1.35 installed and 7.3 release is\n> nearing.\n\nI suggest that you merge your branch as soon as you are ready and force\nthe issue a bit. It's not like the bison installation on\npostgresql.org can make the world stop.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Sat, 19 Oct 2002 00:09:49 +0200 (CEST)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison"
},
{
"msg_contents": "\ngoing to be dealing with this this weekend ... Bruce already asked me as\nwell, just been busy during the week, so had to defer to the weekend ...\nif its not in place by tomorrow around, so, 8pm GMT,please feel free to\nemail-nag me :)\n\n\nOn Fri, 18 Oct 2002, Michael Meskes wrote:\n\n> Could anyone please tell me if we found a solution? It seems I missed\n> any discussion or whatsoever. The only things I know is that\n> postgresql.org still has bison 1.35 installed and 7.3 release is\n> nearing.\n>\n> Michael\n> --\n> Michael Meskes\n> Michael@Fam-Meskes.De\n> Go SF 49ers! Go Rhein Fire!\n> Use Debian GNU/Linux! Use PostgreSQL!\n>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 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": "Sat, 19 Oct 2002 00:01:49 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison"
},
{
"msg_contents": "On Fri, 18 Oct 2002, Tom Lane wrote:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Larry Rosenman wrote:\n> >> Looks like threre is a 1.75 out (based on a FreeBSD PR I just saw fly\n> >> by).\n>\n> > I can confirm that:\n> > \thttp://ftp.gnu.org/gnu/bison/\n> > Attached is the Changelog since 1.50.\n>\n> Urgh. Nothing like major releases 2 weeks apart to give one a nice warm\n> comfy feeling about the stability of one's tools.\n>\n> Offhand it looks like these are nearly all bug fixes, and so we'd better\n> standardize on 1.75 not 1.50. But I wonder what will be out next\n> week...\n\nthat's one of hte reasons I've been holding off (the other being a busy\nweek) ... I *try* to keep stuff like this to what the FreeBSD ports\ncollection considers a stable release, and it hasn't been updated yet :(\n\n\n",
"msg_date": "Sat, 19 Oct 2002 00:03:30 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison "
},
{
"msg_contents": "On Fri, 2002-10-18 at 22:03, Marc G. Fournier wrote:\n> On Fri, 18 Oct 2002, Tom Lane wrote:\n> \n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > Larry Rosenman wrote:\n> > >> Looks like threre is a 1.75 out (based on a FreeBSD PR I just saw fly\n> > >> by).\n> >\n> > > I can confirm that:\n> > > \thttp://ftp.gnu.org/gnu/bison/\n> > > Attached is the Changelog since 1.50.\n> >\n> > Urgh. Nothing like major releases 2 weeks apart to give one a nice warm\n> > comfy feeling about the stability of one's tools.\n> >\n> > Offhand it looks like these are nearly all bug fixes, and so we'd better\n> > standardize on 1.75 not 1.50. But I wonder what will be out next\n> > week...\n> \n> that's one of hte reasons I've been holding off (the other being a busy\n> week) ... I *try* to keep stuff like this to what the FreeBSD ports\n> collection considers a stable release, and it hasn't been updated yet :(\nThe PR I referred to above is the PR to update to 1.75. \n\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": "18 Oct 2002 22:06:40 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison"
},
{
"msg_contents": "On Fri, Oct 18, 2002 at 04:16:29PM -0400, Bruce Momjian wrote:\n> > Looks like threre is a 1.75 out (based on a FreeBSD PR I just saw fly\n> > by). \n> \n> I can confirm that:\n\nI just tried Marc's installation of 1.75 and compared the result with my\n1.50 compiled version of ecpg grammar and found both .c files to be\nidentical. Seems the changes didn't effect the parser output.\n\nMichael\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Mon, 21 Oct 2002 14:56:26 +0200",
"msg_from": "Michael Meskes <meskes@postgresql.org>",
"msg_from_op": true,
"msg_subject": "Re: ECPG and bison"
},
{
"msg_contents": "Michael Meskes wrote:\n> On Fri, Oct 18, 2002 at 04:16:29PM -0400, Bruce Momjian wrote:\n> > > Looks like threre is a 1.75 out (based on a FreeBSD PR I just saw fly\n> > > by). \n> > \n> > I can confirm that:\n> \n> I just tried Marc's installation of 1.75 and compared the result with my\n> 1.50 compiled version of ecpg grammar and found both .c files to be\n> identical. Seems the changes didn't effect the parser output.\n\nOh, that's very good news. Makes me feel a little more secure.\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, 21 Oct 2002 10:42:20 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: ECPG and bison"
}
] |
[
{
"msg_contents": "Marc,\n\nnot only the global variables are a problem. PostgreSQL doesn't clean\nup\nall the memory before a process terminates and you must deal with\nsignals between threads.\n\nI've modified PostgreSQL that it is completely thread based and nearly\nclean\nwith allocating/deallocating memory and I'm using signals between\nthreads.\nI've written the code that it's easy to combine the changes with new\nreleases of\nPostgreSQL but it has some problems that require a decrease of the\nnumber of\nglobal variables. This is very hard if you don't want to drift away\nfrom the\nmain sourcecode of PostgreSQL.\n\nActually I have a table that holds all the globals that needs to be\naddressed\nand I can tell you that there are a lot of global variables.\n\nHowever. I've tried to give the changes with the memory freeing back to\nthe\nmain source tree but my feeling was that they don't like it because it\ndoesn't\nmatter on UNIX if you're only process based. The same with several\ntools like\nPSQL. After this I wanted to look closer at the global variables.\n\nIf this may change in the future you would be very happy if you don't\nhave to \nsearch for unreleased memory because it is solved.\n\nI'm still VERY interested in giving those changes back to the\ncommunity,\nbut i think that my changes are still not wanted.\n\nAnother problem I have is that the build tools i'm using are not GNU\nand \nthe project file is different in this case. The developers from\nPostgreSQL \ndon't want this too.\n\nCurrently i'm working on a project file that uses GNU make but i'm not\nvery\nhappy with this because it is dramatically slower compared to the\nMetrowerks Tools. With Codewarrior I can build PostgreSQL in around\n2 Minutes with all libraires and so on and GNU make needs much much\nlonger.\n\nBut I still hope that I'll find a way how I can handle all this.\n\nPostgreSQL itslef is running very good with \"only threads\" and the\ndatabase\nitself is a VERY GOOD product.\n\nIn the future I'll make some tests to find out information about the\nperformance\nand how threads affect the performance on Linux.\n\nI'm also working on several additions to PostgreSQL for remote realtime\nMonitoring\nand configuration, LDAP authentication and more.\n\nUlrich \n\n>>> \"Marc G. Fournier\" <scrappy@hub.org> 18.10.2002 05:21:59 >>>\nOn Wed, 16 Oct 2002, Anuradha Ratnaweera wrote:\n\n> On Wed, Oct 16, 2002 at 01:25:23AM -0400, Bruce Momjian wrote:\n> > Anuradha Ratnaweera wrote:\n> >\n> > > ... what I want to know is whether multithreading is likely to\nget\n> > > into in postgresql, say somewhere in 8.x, or even in 9.x?\n> >\n> > It may be optional some day, most likely for Win32 at first, but we\nsee\n> > little value to it on most other platforms; of course, we may be\nwrong.\n>\n> In that case, I wonder if it is worth folking a new project to add\n> threading support to the backend? Of course, keeping in sync with\nthe\n> original would be lot of work.\n\nActually, if you go through the archives, there has been talk about\nwhat\nwould have to be done to the main source tree towards getting\nthreading\nincluded ... as well as a lengthy discussion about the steps involved.\n\nthe first and foremost issue that needs to be addressed is cleaning up\nthe\nglobal variables that are used throughout ...\n\n\n---------------------------(end of\nbroadcast)---------------------------\nTIP 5: Have you checked our extensive FAQ?\n\nhttp://www.postgresql.org/users-lounge/docs/faq.html\n----------------------------------\n This e-mail is virus scanned\n Diese e-mail ist virusgeprueft\n\n",
"msg_date": "Fri, 18 Oct 2002 18:10:44 +0200",
"msg_from": "\"Ulrich Neumann\" <U_Neumann@gne.de>",
"msg_from_op": true,
"msg_subject": "Antw: Re: Postgresql and multithreading"
},
{
"msg_contents": "On 18 Oct 2002 at 18:10, Ulrich Neumann wrote:\n\n> Marc,\n> \n> not only the global variables are a problem. PostgreSQL doesn't clean\n> up\n> all the memory before a process terminates and you must deal with\n> signals between threads.\n\nOK, first of all let me say this. I am interested in seeing postgresql \nmutlithreaded but I don't think it's a good idea to provide a thread per \nconnection. On slick unices like freebsd/linux, threads and processes are \nalmost equally heavy. Besides if connection opening time for a database is an \nissue, application should cache the pool of connections. Databases are not web \nservers which should cater to thousands of connections per minute.\n\nI am interested in seeing postgresql multithreaded so that it performs better \non SMP machine. Right now I am building/designing a *huge* database which is \ngoing to use at least 4 way machines. With number of connections very small, \naround 10-15 and being persistent, I am afraid that postgresql would not use \nall the available horsepower.\n\n \n> I've modified PostgreSQL that it is completely thread based and nearly\n> clean\n> with allocating/deallocating memory and I'm using signals between\n> threads.\n\nI am sure it must have been massive task.\n\n> I'm still VERY interested in giving those changes back to the\n> community,\n> but i think that my changes are still not wanted.\n\nWell, I understand your feeling but I am sure it's not the cases. People here \nhave stated that there are other optimisation areas which needs to be addresses \nfirst. Threading is not ruled out but it's not on current radar either.\n\nI would be more than happy to assist you to make this happen. I have some \nexperience in threads(pthreads on linux) but absolute no experience in \npostgresql code.\n\nCan we make a patch that converts a single module/operation to multithreaded? \nLet's say sequential table scan. \n\nI propose that such a conversion should be done under conditional compilation \nwith default threaded behaviour as off. This would not disturb existing setup \nand behaviour where threads are not good enough. Besides since we are \ndistributing almost exclusive work to threads, there should be little \nsyncronisation issues.\n\nI request postgresql hackers to look up to this proposal favourably. This \nwould be a good and gradual way to get threading in postgresql.\n\nAwaiting your feedback..\n\n\n\nBye\n Shridhar\n\n--\nThe Consultant's Curse:\tWhen the customer has beaten upon you long enough, give \nhim\twhat he asks for, instead of what he needs. This is very strong\tmedicine, \nand is normally only required once.\n\n",
"msg_date": "Sat, 19 Oct 2002 12:27:59 +0530",
"msg_from": "\"Shridhar Daithankar\" <shridhar_daithankar@persistent.co.in>",
"msg_from_op": false,
"msg_subject": "Re: Antw: Re: Postgresql and multithreading"
},
{
"msg_contents": "On Sat, 19 Oct 2002, Shridhar Daithankar wrote:\n\n> On 18 Oct 2002 at 18:10, Ulrich Neumann wrote:\n>\n> > I'm still VERY interested in giving those changes back to the\n> > community, but i think that my changes are still not wanted.\n\nOkay, can you submit such changes in pieces / steps? For instance, I know\nthere was alot of talk about how to fix/improve the global variable\nsituation, but nobody appeared to have picked that up and done anything\nwith it ... if you have, is it something you could submit as a patch for\nreview?\n\nNote that as the source matures, we've become more critical of patches,\nespecially those that affect the *whole* server, so if you can submit\npatches in manageable chunks, it becomes easier to review and get those\nchanges in before the next round of changes ...\n\n\n\n",
"msg_date": "Sun, 20 Oct 2002 23:46:29 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": false,
"msg_subject": "Re: Antw: Re: Postgresql and multithreading"
}
] |
[
{
"msg_contents": "pg_encoding doesn't reject invalid input (like 'pg_encoding foo'). I\nthink it should exit with a non-zero status.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Fri, 18 Oct 2002 18:17:31 +0200 (CEST)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": true,
"msg_subject": "pg_encoding doesn't reject invalid input"
},
{
"msg_contents": "Peter Eisentraut <peter_e@gmx.net> writes:\n> pg_encoding doesn't reject invalid input (like 'pg_encoding foo'). I\n> think it should exit with a non-zero status.\n\nSeems reasonable to me ... shouldn't it complain on stderr as well?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 18 Oct 2002 14:43:47 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: pg_encoding doesn't reject invalid input "
},
{
"msg_contents": "> pg_encoding doesn't reject invalid input (like 'pg_encoding foo'). I\n> think it should exit with a non-zero status.\n\nFixed.\n--\nTatsuo Ishii\n",
"msg_date": "Sun, 20 Oct 2002 20:33:21 +0900 (JST)",
"msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>",
"msg_from_op": false,
"msg_subject": "Re: pg_encoding doesn't reject invalid input"
}
] |
[
{
"msg_contents": "Seems we aren't the only ones concerned about memset-like optimizations:\n\n\thttp://www.bsdforums.org/forums/showthread.php?threadid=3778\n\nThis talks about using MMX assembly instructions for memcpy.\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, 18 Oct 2002 13:01:29 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "memcpy testing"
}
] |
[
{
"msg_contents": "Hi,\n\nI was just approached with an optimizer question I cannot answer. Does\nour optimizer know how expensive different comparisons are? That is can\nit do something like:\n\nIf there are different comparisons in a where clause check the ints\nfirst, then the strings, then everything with regexp, or like clauses\netc. and finally function calls at last, because in most cases a\nfunction call is the most expensive one. \n\nOkay, we may argue whether a regexp is more expensive than a function,\nbut you get the idea. Are we able to get that logic in where clauses\nwhere no index is defined? I just tried a query that given the order in\nthe where clause either takes a few seconds or forever. Well, I killed\nthe query after quite some time.\n\nMichael\n-- \nMichael Meskes\nMichael@Fam-Meskes.De\nGo SF 49ers! Go Rhein Fire!\nUse Debian GNU/Linux! Use PostgreSQL!\n",
"msg_date": "Fri, 18 Oct 2002 19:02:31 +0200",
"msg_from": "Michael Meskes <meskes@postgresql.org>",
"msg_from_op": true,
"msg_subject": "optimizer question"
},
{
"msg_contents": "Michael Meskes <meskes@postgresql.org> writes:\n> I was just approached with an optimizer question I cannot answer. Does\n> our optimizer know how expensive different comparisons are?\n\nIt does not, because there's noplace it could look to find the info.\n\n> If there are different comparisons in a where clause check the ints\n> first, then the strings, then everything with regexp, or like clauses\n> etc. and finally function calls at last, because in most cases a\n> function call is the most expensive one. \n\nUnfortunately, these all look like function calls to the optimizer.\n\nI have played with the notion of forcing subquery-based WHERE clauses\nto the end of the list, but that's about all that we could easily manage\nwithout great leaps of assumptions...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 18 Oct 2002 14:48:58 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: optimizer question "
}
] |
[
{
"msg_contents": "Michael,\n\nPlease post instead of sending private email.\n\nOn Fri, Oct 18, 2002 at 12:08:02PM +0200, Michael Steiner wrote:\n> FAQ says: \"A native port to MS Win NT/2000/XP is currently being worked\n> on.\"\n> \n> => When is it expected to be available as a stable release?\n\nI don't know.\n\n> Will it be based on cygwin dll's?\n\nNo.\n\n> Will the reliability and performance be comparable to the unix-version?\n\nI don't know.\n\n> Will it be free for commercial use?\n\nI presume so.\n\nMaybe others on the CC'd lists can answer your questions better.\n\nJason\n",
"msg_date": "Fri, 18 Oct 2002 15:39:31 -0400",
"msg_from": "Jason Tishler <jason@tishler.net>",
"msg_from_op": true,
"msg_subject": "Re: Postgres Windows Native Port"
},
{
"msg_contents": "what is the command to launch a sql script not in psql\nthanks\n\n",
"msg_date": "Tue, 05 Nov 2002 15:19:44 +0100",
"msg_from": "Florian Litot <flitot@besancon.sema.slb.com>",
"msg_from_op": false,
"msg_subject": "command"
},
{
"msg_contents": "On Tue, Nov 05, 2002 at 03:19:44PM +0100, Florian Litot wrote:\n> what is the command to launch a sql script not in psql\n\nPerl?\n\nOr maybe you just want to read SQL into Postgres using the shell. In\nwhich case, try the -f switch to psql.\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": "Thu, 7 Nov 2002 10:12:27 -0500",
"msg_from": "Andrew Sullivan <andrew@libertyrms.info>",
"msg_from_op": false,
"msg_subject": "Re: command"
},
{
"msg_contents": "On Tue, 5 Nov 2002, Florian Litot wrote:\n\n> what is the command to launch a sql script not in psql\n> thanks\n\nwithout actually being IN psql, you can use it to run one line scripts \nlike this:\n\npsql dbname -c -- 'single query goes here'\n\nor you can run a large file full of sql queries like this:\n\npsql dbname -f filename.sql\n\n",
"msg_date": "Thu, 7 Nov 2002 09:54:18 -0700 (MST)",
"msg_from": "\"scott.marlowe\" <scott.marlowe@ihs.com>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] command"
},
{
"msg_contents": "Be more specific with your question, what exactly are You trying to do ? \nDid you ment PL/PSQL script (stored procedure) , or You are asking about \ninterfaces to PostgreSQL ?\n\nI suppose the right place to ask questions like that would be \npsql-general@postgresql.org, not to hackers.\n\nregards !\n\nOn Tuesday 05 November 2002 14:19, Florian Litot wrote:\n> what is the command to launch a sql script not in psql\n> thanks\n>\n>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n\n",
"msg_date": "Thu, 7 Nov 2002 17:37:38 +0000",
"msg_from": "Darko Prenosil <darko.prenosil@finteh.hr>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] command"
}
] |
[
{
"msg_contents": "$ ./clusterdb\npsql: could not connect to server: No such file or directory\n Is the server running locally and accepting\n connections on Unix domain socket \"/tmp/.s.PGSQL.5432\"?\npsql: could not connect to server: No such file or directory\n Is the server running locally and accepting\n connections on Unix domain socket \"/tmp/.s.PGSQL.5432\"?\nclusterdb: While clustering peter, the following failed:\n$\n\nThis could probably handled a little more gracefully.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Sat, 19 Oct 2002 00:11:32 +0200 (CEST)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": true,
"msg_subject": "clusterdb has poor error recovery"
},
{
"msg_contents": "On Sat, Oct 19, 2002 at 12:11:32AM +0200, Peter Eisentraut wrote:\n> $ ./clusterdb\n> psql: could not connect to server: No such file or directory\n> Is the server running locally and accepting\n> connections on Unix domain socket \"/tmp/.s.PGSQL.5432\"?\n> psql: could not connect to server: No such file or directory\n> Is the server running locally and accepting\n> connections on Unix domain socket \"/tmp/.s.PGSQL.5432\"?\n> clusterdb: While clustering peter, the following failed:\n> $\n> \n> This could probably handled a little more gracefully.\n\nYes, sorry. A patch for this is attached. Please apply.\n\n-- \nAlvaro Herrera (<alvherre[a]dcc.uchile.cl>)\nXML! Exclaimed C++. What are you doing here? You're not a programming\nlanguage.\nTell that to the people who use me, said XML.",
"msg_date": "Sat, 19 Oct 2002 15:14:56 -0300",
"msg_from": "Alvaro Herrera <alvherre@dcc.uchile.cl>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] clusterdb has poor error recovery"
},
{
"msg_contents": "\nYour patch has been added to the PostgreSQL unapplied patches list at:\n\n\thttp://momjian.postgresql.org/cgi-bin/pgpatches\n\nI will try to apply it within the next 48 hours.\n\n---------------------------------------------------------------------------\n\n\nAlvaro Herrera wrote:\n> On Sat, Oct 19, 2002 at 12:11:32AM +0200, Peter Eisentraut wrote:\n> > $ ./clusterdb\n> > psql: could not connect to server: No such file or directory\n> > Is the server running locally and accepting\n> > connections on Unix domain socket \"/tmp/.s.PGSQL.5432\"?\n> > psql: could not connect to server: No such file or directory\n> > Is the server running locally and accepting\n> > connections on Unix domain socket \"/tmp/.s.PGSQL.5432\"?\n> > clusterdb: While clustering peter, the following failed:\n> > $\n> > \n> > This could probably handled a little more gracefully.\n> \n> Yes, sorry. A patch for this is attached. Please apply.\n> \n> -- \n> Alvaro Herrera (<alvherre[a]dcc.uchile.cl>)\n> XML! Exclaimed C++. What are you doing here? You're not a programming\n> language.\n> Tell that to the people who use me, said XML.\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": "Sun, 20 Oct 2002 21:18:36 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] clusterdb has poor error recovery"
},
{
"msg_contents": "\nPatch applied. Thanks.\n\n---------------------------------------------------------------------------\n\n\nAlvaro Herrera wrote:\n> On Sat, Oct 19, 2002 at 12:11:32AM +0200, Peter Eisentraut wrote:\n> > $ ./clusterdb\n> > psql: could not connect to server: No such file or directory\n> > Is the server running locally and accepting\n> > connections on Unix domain socket \"/tmp/.s.PGSQL.5432\"?\n> > psql: could not connect to server: No such file or directory\n> > Is the server running locally and accepting\n> > connections on Unix domain socket \"/tmp/.s.PGSQL.5432\"?\n> > clusterdb: While clustering peter, the following failed:\n> > $\n> > \n> > This could probably handled a little more gracefully.\n> \n> Yes, sorry. A patch for this is attached. Please apply.\n> \n> -- \n> Alvaro Herrera (<alvherre[a]dcc.uchile.cl>)\n> XML! Exclaimed C++. What are you doing here? You're not a programming\n> language.\n> Tell that to the people who use me, said XML.\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": "Mon, 21 Oct 2002 16:32:31 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] clusterdb has poor error recovery"
}
] |
[
{
"msg_contents": "I see the following error with the current CVS code:\n\nmake[3]: Entering directory `/home/nconway/pgsql/src/backend/commands'\n[ ... ]\ngcc -O2 -DLINUX_PROFILE -g -pg -Wall -Wmissing-prototypes -Wmissing-declarations -I../../../src/include -c -o copy.o copy.c\ncopy.c: In function `DoCopy':\ncopy.c:425: `errno' undeclared (first use in this function)\ncopy.c:425: (Each undeclared identifier is reported only once\ncopy.c:425: for each function it appears in.)\nmake[3]: *** [copy.o] Error 1\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n",
"msg_date": "18 Oct 2002 19:22:12 -0400",
"msg_from": "Neil Conway <neilc@samurai.com>",
"msg_from_op": true,
"msg_subject": "compile error in CVS HEAD"
},
{
"msg_contents": "Neil Conway <neilc@samurai.com> writes:\n> I see the following error with the current CVS code:\n\n> copy.c:425: `errno' undeclared (first use in this function)\n\nHmm. Needs a \"#include <errno.h>\" seemingly; but surely that error\nhas been there awhile?\n\n(light dawns...) I'll bet it was masked by all the bogus \"extern int\nerrno\" declarations that we had till this afternoon. I didn't see the\nfailure here, because HPUX's header set seems to include <errno.h> in\nsome pretty basic place. Anywhere else failing for you?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 18 Oct 2002 20:27:11 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: compile error in CVS HEAD "
}
] |
[
{
"msg_contents": "There are a number of statements, such as TRUNCATE TABLE, that refuse to\nrun in a transaction block because they perform actions that can't be\nrolled back later.\n\nThese statements currently do not look at autocommit, which means that\nif autocommit is off, their tests will succeed ... but then a\ntransaction block is started anyway, defeating the point of the test.\n\nWe could fix these statements to fail if autocommit is off, which means\nthat you could not use them at all except by setting autocommit on.\nUgh.\n\nOr we could fix them to force an autocommit. Which would mean that\nthese \"dangerous\" statements would become even more dangerous, since\nthat's exactly the behavior a person using autocommit-off would not\nexpect. Also ugh.\n\nAnyone see a way out of this catch-22? If not, which is the least\nbad alternative?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 18 Oct 2002 21:04:37 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Tom Lane wrote:\n> There are a number of statements, such as TRUNCATE TABLE, that refuse to\n> run in a transaction block because they perform actions that can't be\n> rolled back later.\n> \n> These statements currently do not look at autocommit, which means that\n> if autocommit is off, their tests will succeed ... but then a\n> transaction block is started anyway, defeating the point of the test.\n> \n> We could fix these statements to fail if autocommit is off, which means\n> that you could not use them at all except by setting autocommit on.\n> Ugh.\n> \n> Or we could fix them to force an autocommit. Which would mean that\n> these \"dangerous\" statements would become even more dangerous, since\n> that's exactly the behavior a person using autocommit-off would not\n> expect. Also ugh.\n> \n> Anyone see a way out of this catch-22? If not, which is the least\n> bad alternative?\n> \n\nI think the \"least bad\" is the first option -- disallow TRUNCATE unless \nautocommit is on. With the second option, people would be caught by surprise \nat precisely the worst possible moment. Better to make them take the extra step.\n\nJoe\n\n",
"msg_date": "Fri, 18 Oct 2002 18:14:27 -0700",
"msg_from": "Joe Conway <mail@joeconway.com>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Tom Lane wrote:\n> There are a number of statements, such as TRUNCATE TABLE, that refuse to\n> run in a transaction block because they perform actions that can't be\n> rolled back later.\n> \n> These statements currently do not look at autocommit, which means that\n> if autocommit is off, their tests will succeed ... but then a\n> transaction block is started anyway, defeating the point of the test.\n> \n> We could fix these statements to fail if autocommit is off, which means\n> that you could not use them at all except by setting autocommit on.\n> Ugh.\n> \n> Or we could fix them to force an autocommit. Which would mean that\n> these \"dangerous\" statements would become even more dangerous, since\n> that's exactly the behavior a person using autocommit-off would not\n> expect. Also ugh.\n> \n> Anyone see a way out of this catch-22? If not, which is the least\n> bad alternative?\n\nUgh, I see what you mean. With the old code, you had to do a BEGIN to\nstart a multi-statement transaction, while with autocommit off, you are\nalways in one. I don't think forcing them to turn on autocommit makes\nany sense; it isn't obvious, and it doesn't buy anything. I think we\nshould just do an automatic COMMIT if it is the first statement of a\ntransaction, and if not, throw the same error we used to throw. We are\nperforming autocommit for SET at the start of a transaction now anyway,\nso it isn't totally strange to do it for TRUNCATE, etc. too. In fact,\nyou can just put the xact commit check in the same place SET is handled\nin postgres.c. It isn't great, but it is clean. ;-)\n\nYou could also throw a NOTICE mentioning it was committed as a separate\ntransaction.\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, 18 Oct 2002 21:18:19 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> ... I think we\n> should just do an automatic COMMIT if it is the first statement of a\n> transaction, and if not, throw the same error we used to throw. We are\n> performing autocommit for SET at the start of a transaction now anyway,\n> so it isn't totally strange to do it for TRUNCATE, etc. too. In fact,\n> you can just put the xact commit check in the same place SET is handled\n> in postgres.c. It isn't great, but it is clean. ;-)\n\nWell, \"clean\" isn't the adjective I would use ;-), but this might be the\nmost useful approach. The analogy to SET hadn't occurred to me.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 18 Oct 2002 21:24:26 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > ... I think we\n> > should just do an automatic COMMIT if it is the first statement of a\n> > transaction, and if not, throw the same error we used to throw. We are\n> > performing autocommit for SET at the start of a transaction now anyway,\n> > so it isn't totally strange to do it for TRUNCATE, etc. too. In fact,\n> > you can just put the xact commit check in the same place SET is handled\n> > in postgres.c. It isn't great, but it is clean. ;-)\n> \n> Well, \"clean\" isn't the adjective I would use ;-), but this might be the\n\nClean in coding terms, _only_.\n\n> most useful approach. The analogy to SET hadn't occurred to me.\n\nYea, the SET behavior appeared pretty queer to me, but now that I have\nused it, I am getting used to 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": "Fri, 18 Oct 2002 21:30:34 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Bruce Momjian wrote:\n>>most useful approach. The analogy to SET hadn't occurred to me.\n> \n> \n> Yea, the SET behavior appeared pretty queer to me, but now that I have\n> used it, I am getting used to it.\n> \n\nSo does that mean:\n\n set autocommit to off;\n begin;\n insert into foo values('a');\n insert into bar values('b');\n truncate table foobar;\n\nwill automatically commit the two inserts?\n\nJoe\n\n\n",
"msg_date": "Fri, 18 Oct 2002 18:34:16 -0700",
"msg_from": "Joe Conway <mail@joeconway.com>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Joe Conway wrote:\n> Bruce Momjian wrote:\n> >>most useful approach. The analogy to SET hadn't occurred to me.\n> > \n> > \n> > Yea, the SET behavior appeared pretty queer to me, but now that I have\n> > used it, I am getting used to it.\n> > \n> \n> So does that mean:\n> \n> set autocommit to off;\n> begin;\n> insert into foo values('a');\n> insert into bar values('b');\n> truncate table foobar;\n> \n> will automatically commit the two inserts?\n\nNo, the entire transaction will aborted because TRUNCATE has to be at\nthe start of a multi-statement transaction. This will also fail:\n\n> set autocommit to off;\n> begin;\n> truncate table foobar;\n> insert into foo values('a');\n> insert into bar values('b');\n\nbut this will work:\n\n> set autocommit to off;\n> truncate table foobar;\n> insert into foo values('a');\n> insert into bar values('b');\n\nIn the last case, the TRUNCATE will happen, and the INSERTs will be in\ntheir own multi-statement transaction. A SET in place of TRUNCATE will\nbehave the same way.\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, 18 Oct 2002 21:39:01 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Bruce Momjian wrote:\n> Joe Conway wrote:\n> \n>>Bruce Momjian wrote:\n>>\n>>>>most useful approach. The analogy to SET hadn't occurred to me.\n>>>\n>>>\n>>>Yea, the SET behavior appeared pretty queer to me, but now that I have\n>>>used it, I am getting used to it.\n\n<snip examples>\n\n> \n> In the last case, the TRUNCATE will happen, and the INSERTs will be in\n> their own multi-statement transaction. A SET in place of TRUNCATE will\n> behave the same way.\n> \n\nHmmm. It does look strange. We ought to make this prominent in the release \nnotes and docs.\n\nJoe\n\n",
"msg_date": "Fri, 18 Oct 2002 18:45:50 -0700",
"msg_from": "Joe Conway <mail@joeconway.com>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "On Fri, 18 Oct 2002, Tom Lane wrote:\n\n> Anyone see a way out of this catch-22? If not, which is the least\n> bad alternative?\n\nUltimately, fix TRUNCATE to be transaction safe. This is non-trivial,\nI know :-).\n\nRegardless, the first option seems the less of the two evils.\n\nGavin\n\n",
"msg_date": "Sat, 19 Oct 2002 12:02:11 +1000 (EST)",
"msg_from": "Gavin Sherry <swm@linuxworld.com.au>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Gavin Sherry wrote:\n> On Fri, 18 Oct 2002, Tom Lane wrote:\n> \n> \n>>Anyone see a way out of this catch-22? If not, which is the least\n>>bad alternative?\n> \n> \n> Ultimately, fix TRUNCATE to be transaction safe. This is non-trivial,\n> I know :-).\n> \n> Regardless, the first option seems the less of the two evils.\n\nEven though TRUNCATE was modeled after Oracle's TRUNCATE and \nOracle's TRUNCATE commits the running tx, truncates the \nrelation, and starts a new tx, regardless of whether or not \nTRUNCATE is the first statement of the tx?\n\nMike Mascari\nmascarm@mascari.com\n\n",
"msg_date": "Fri, 18 Oct 2002 22:07:39 -0400",
"msg_from": "Mike Mascari <mascarm@mascari.com>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Mike Mascari wrote:\n> Gavin Sherry wrote:\n> > On Fri, 18 Oct 2002, Tom Lane wrote:\n> > \n> > \n> >>Anyone see a way out of this catch-22? If not, which is the least\n> >>bad alternative?\n> > \n> > \n> > Ultimately, fix TRUNCATE to be transaction safe. This is non-trivial,\n> > I know :-).\n> > \n> > Regardless, the first option seems the less of the two evils.\n> \n> Even though TRUNCATE was modeled after Oracle's TRUNCATE and \n> Oracle's TRUNCATE commits the running tx, truncates the \n> relation, and starts a new tx, regardless of whether or not \n> TRUNCATE is the first statement of the tx?\n\nThat seems just too harsh to me. I think we should impose some\nstructure to it, though we will have compatibility issues.\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, 18 Oct 2002 22:15:32 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "On Fri, 18 Oct 2002, Mike Mascari wrote:\n\n> Gavin Sherry wrote:\n> > On Fri, 18 Oct 2002, Tom Lane wrote:\n> > \n> > \n> >>Anyone see a way out of this catch-22? If not, which is the least\n> >>bad alternative?\n> > \n> > \n> > Ultimately, fix TRUNCATE to be transaction safe. This is non-trivial,\n> > I know :-).\n> > \n> > Regardless, the first option seems the less of the two evils.\n> \n> Even though TRUNCATE was modeled after Oracle's TRUNCATE and \n> Oracle's TRUNCATE commits the running tx, truncates the \n> relation, and starts a new tx, regardless of whether or not \n> TRUNCATE is the first statement of the tx?\n\nWhy should we be *only* as good as Oracle? :-)\n\nGavin\n\n",
"msg_date": "Sat, 19 Oct 2002 12:41:19 +1000 (EST)",
"msg_from": "Gavin Sherry <swm@linuxworld.com.au>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Gavin Sherry <swm@linuxworld.com.au> writes:\n> On Fri, 18 Oct 2002, Tom Lane wrote:\n>> Anyone see a way out of this catch-22? If not, which is the least\n>> bad alternative?\n\n> Ultimately, fix TRUNCATE to be transaction safe. This is non-trivial,\n> I know :-).\n\nI was about to say that the entire *point* of TRUNCATE is to be\ntransaction-unsafe ;-)\n\nBut on the other hand ... now that we have relation versioning (like\nCLUSTER) it seems like TRUNCATE could generate new versions of the\nrelation and its indexes, without touching the originals. This would\nmake it transaction-safe, at the cost of not releasing the original\nversion's disk space till you commit. Seems like a good tradeoff to me.\n\nIt's not happening for 7.3, in any case, so we need a stopgap answer.\nThere are other examples --- CREATE/DROP DATABASE, for example ---\nwhere we'd probably need an answer anyway; I doubt we'll ever make\nthose completely transaction-safe.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 18 Oct 2002 23:25:53 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al "
},
{
"msg_contents": "On Friday 18 October 2002 11:25 pm, Tom Lane wrote:\n> Gavin Sherry <swm@linuxworld.com.au> writes:\n> > On Fri, 18 Oct 2002, Tom Lane wrote:\n> >> Anyone see a way out of this catch-22? If not, which is the least\n> >> bad alternative?\n\n> > Ultimately, fix TRUNCATE to be transaction safe. This is non-trivial,\n> > I know :-).\n\n> I was about to say that the entire *point* of TRUNCATE is to be\n> transaction-unsafe ;-)\n\nI actually was considering using a transaction-safe TRUNCATE in an application \ninvolving daily imports of 170MB of data into a set of linked tables. Since \nthe import takes a finite amount of time, it would be nice to have the \nprevious data available while the new is being imported. And TRUNCATE is \nsignificantly faster than DELETE over 170MB of data.\n-- \nLamar Owen\nWGCR Internet Radio\n1 Peter 4:11\n",
"msg_date": "Sat, 19 Oct 2002 22:43:16 -0400",
"msg_from": "Lamar Owen <lamar.owen@wgcr.org>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "\nAdded to TODO:\n\n\t* Make a transaction-safe TRUNCATE\n\n---------------------------------------------------------------------------\n\nLamar Owen wrote:\n> On Friday 18 October 2002 11:25 pm, Tom Lane wrote:\n> > Gavin Sherry <swm@linuxworld.com.au> writes:\n> > > On Fri, 18 Oct 2002, Tom Lane wrote:\n> > >> Anyone see a way out of this catch-22? If not, which is the least\n> > >> bad alternative?\n> \n> > > Ultimately, fix TRUNCATE to be transaction safe. This is non-trivial,\n> > > I know :-).\n> \n> > I was about to say that the entire *point* of TRUNCATE is to be\n> > transaction-unsafe ;-)\n> \n> I actually was considering using a transaction-safe TRUNCATE in an application \n> involving daily imports of 170MB of data into a set of linked tables. Since \n> the import takes a finite amount of time, it would be nice to have the \n> previous data available while the new is being imported. And TRUNCATE is \n> significantly faster than DELETE over 170MB of data.\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\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, 19 Oct 2002 22:51:12 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Tom Lane wrote:\n>> There are a number of statements, such as TRUNCATE TABLE, that refuse to\n>> run in a transaction block because they perform actions that can't be\n>> rolled back later.\n>> \n>> These statements currently do not look at autocommit, which means that\n>> if autocommit is off, their tests will succeed ... but then a\n>> transaction block is started anyway, defeating the point of the test.\n\n> ... I think we\n> should just do an automatic COMMIT if it is the first statement of a\n> transaction, and if not, throw the same error we used to throw. We are\n> performing autocommit for SET at the start of a transaction now anyway,\n> so it isn't totally strange to do it for TRUNCATE, etc. too.\n\nThere is another aspect of this, which is: what if one of these\nstatements is called by a user-defined function, say a plpgsql function\nthat does a TRUNCATE and then other stuff? If the function is called\nby a SELECT that's not inside a transaction block, then its\nIsTransactionBlock() test will succeed --- but the possibility remains\nthat the later actions of the function could cause an elog and\ntransaction rollback. Which is what we wanted to prevent.\n\nWe can go with the auto-COMMIT idea for statements that are invoked at\nthe outer interactive level, but that doesn't work for stuff invoked\ninside a function. I think we need to forbid these statements inside\nfunctions, too. We already have that for VACUUM, because of its\ninternal commits causing problems for functions, but we'll need to\nextend it to all of them.\n\nJust FYI, the statements involved are\nVACUUM\nTRUNCATE TABLE\nCREATE/DROP DATABASE\nREINDEX (all forms)\nALTER USER changing password\nDROP USER\n\nALTER and DROP USER just issue NOTICEs rather than failing, which seems\npretty bogus in itself. The reason they are worried is that there's\nno mechanism for rolling back updates of the flat password file.\nI think we could fix that by arranging not to write the flat password\nfile at all until we are ready to commit the current transaction;\nwill take a look at it.\n\nREINDEX perhaps could be treated as transaction-safe in the forms that\nbuild a new index file rather than truncating. Will look at that, too.\n\nAnother place that is calling IsTransactionBlock is parser/analyze.c\nwhile processing DECLARE CURSOR. I think this is pretty bogus for\nseveral reasons:\n1. There's no good reason to make DECLARE CURSOR outside a transaction\n block be an ERROR; at most it should be a NOTICE.\n2. Parse analysis is the wrong place anyway; it should be tested\n at execution time, methinks.\n3. If the cursor is being declared and used inside a function, then\n it could be used successfully without being inside a transaction\n block at all.\n\nPoint 3 makes me think we should just get rid of the test entirely.\nComments?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 12:40:54 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Tom Lane wrote:\n> >> There are a number of statements, such as TRUNCATE TABLE, that refuse to\n> >> run in a transaction block because they perform actions that can't be\n> >> rolled back later.\n> >> \n> >> These statements currently do not look at autocommit, which means that\n> >> if autocommit is off, their tests will succeed ... but then a\n> >> transaction block is started anyway, defeating the point of the test.\n> \n> > ... I think we\n> > should just do an automatic COMMIT if it is the first statement of a\n> > transaction, and if not, throw the same error we used to throw. We are\n> > performing autocommit for SET at the start of a transaction now anyway,\n> > so it isn't totally strange to do it for TRUNCATE, etc. too.\n> \n> There is another aspect of this, which is: what if one of these\n> statements is called by a user-defined function, say a plpgsql function\n> that does a TRUNCATE and then other stuff? If the function is called\n> by a SELECT that's not inside a transaction block, then its\n> IsTransactionBlock() test will succeed --- but the possibility remains\n> that the later actions of the function could cause an elog and\n> transaction rollback. Which is what we wanted to prevent.\n> \n> We can go with the auto-COMMIT idea for statements that are invoked at\n> the outer interactive level, but that doesn't work for stuff invoked\n> inside a function. I think we need to forbid these statements inside\n> functions, too. We already have that for VACUUM, because of its\n> internal commits causing problems for functions, but we'll need to\n> extend it to all of them.\n> \n> Just FYI, the statements involved are\n> VACUUM\n> TRUNCATE TABLE\n> CREATE/DROP DATABASE\n> REINDEX (all forms)\n> ALTER USER changing password\n> DROP USER\n> \n> ALTER and DROP USER just issue NOTICEs rather than failing, which seems\n> pretty bogus in itself. The reason they are worried is that there's\n> no mechanism for rolling back updates of the flat password file.\n> I think we could fix that by arranging not to write the flat password\n> file at all until we are ready to commit the current transaction;\n> will take a look at it.\n\nYes, I thought we had those secure, but I see now we don't. We only\nhave interlocking so the file is reread at the proper time.\n\n> REINDEX perhaps could be treated as transaction-safe in the forms that\n> build a new index file rather than truncating. Will look at that, too.\n> \n> Another place that is calling IsTransactionBlock is parser/analyze.c\n> while processing DECLARE CURSOR. I think this is pretty bogus for\n> several reasons:\n> 1. There's no good reason to make DECLARE CURSOR outside a transaction\n> block be an ERROR; at most it should be a NOTICE.\n> 2. Parse analysis is the wrong place anyway; it should be tested\n> at execution time, methinks.\n> 3. If the cursor is being declared and used inside a function, then\n> it could be used successfully without being inside a transaction\n> block at all.\n> \n> Point 3 makes me think we should just get rid of the test entirely.\n> Comments?\n\nI thought the transaction test was in DECLARE so people didn't create\ncursors outside of transactions and then wonder why they didn't work.\nIf it is going to fail, an ERROR seems more appropriate than a NOTICE.\nI can see it happening inside a function, yes.\n\nAnother question related to this is the handling of SET/SHOW/RESET in\nfunctions. People should realize it isn't really the first command in\nthe transaction so will be part of the transaction. The big issue is\nthat SET has a fallback when it is not first in a transaciton, namely to\nbe part of the transaction, while TRUNCATE doesn't have that fallback\nbecause it can't be rolled back.\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, 21 Oct 2002 14:54:57 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": ">>> ... I think we\n>>> should just do an automatic COMMIT if it is the first statement of a\n>>> transaction, and if not, throw the same error we used to throw. We are\n>>> performing autocommit for SET at the start of a transaction now anyway,\n>>> so it isn't totally strange to do it for TRUNCATE, etc. too.\n>> \n>> We can go with the auto-COMMIT idea for statements that are invoked at\n>> the outer interactive level,\n\nWhat I just committed uses your idea of auto-committing TRUNCATE et al,\nbut now that I review the thread I think that everyone else thought that\nthat was a dangerous idea. How do you feel about simply throwing an error\nin autocommit-off mode, instead? (At least it's a localized change now)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 18:18:35 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al "
},
{
"msg_contents": "Tom Lane wrote:\n> >>> ... I think we\n> >>> should just do an automatic COMMIT if it is the first statement of a\n> >>> transaction, and if not, throw the same error we used to throw. We are\n> >>> performing autocommit for SET at the start of a transaction now anyway,\n> >>> so it isn't totally strange to do it for TRUNCATE, etc. too.\n> >> \n> >> We can go with the auto-COMMIT idea for statements that are invoked at\n> >> the outer interactive level,\n> \n> What I just committed uses your idea of auto-committing TRUNCATE et al,\n> but now that I review the thread I think that everyone else thought that\n> that was a dangerous idea. How do you feel about simply throwing an error\n> in autocommit-off mode, instead? (At least it's a localized change now)\n\nYes, I saw more votes to not allow it, as you said, but turning off\nautocommit just seemed really strange to me, because then they have to\nturn it on again to continue. It just seemed strange to tell them to\nset something to execute a command.\n\nMaybe we can throw a WARNING when autocommit is on. Would that make\neveryone happy?\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, 21 Oct 2002 18:29:28 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Maybe we can throw a WARNING when autocommit is on. Would that make\n> everyone happy?\n\nI doubt it, because by the time you read the WARNING it's too late:\nthe statement's already committed.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 18:37:51 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Maybe we can throw a WARNING when autocommit is on. Would that make\n> > everyone happy?\n> \n> I doubt it, because by the time you read the WARNING it's too late:\n> the statement's already committed.\n\nI assume the same limitation would hold for VACUUM, right, that you have\nto turn on autocommit mode to use 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": "Mon, 21 Oct 2002 18:39:51 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Tom Lane wrote:\n>> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n>>> Maybe we can throw a WARNING when autocommit is on. Would that make\n>>> everyone happy?\n>> \n>> I doubt it, because by the time you read the WARNING it's too late:\n>> the statement's already committed.\n\n> I assume the same limitation would hold for VACUUM, right, that you have\n> to turn on autocommit mode to use it?\n\nYeah, it would, unless we wanted to throw in some additional hack to\ndistinguish VACUUM from the \"more dangerous\" cases.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 18:46:37 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Tom Lane wrote:\n> >> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> >>> Maybe we can throw a WARNING when autocommit is on. Would that make\n> >>> everyone happy?\n> >> \n> >> I doubt it, because by the time you read the WARNING it's too late:\n> >> the statement's already committed.\n> \n> > I assume the same limitation would hold for VACUUM, right, that you have\n> > to turn on autocommit mode to use it?\n> \n> Yeah, it would, unless we wanted to throw in some additional hack to\n> distinguish VACUUM from the \"more dangerous\" cases.\n\n From my perspective, I think it would be consistent to disallow all\ntransaction-unsafe commands and tell people they have to turn autocommit\non to execute them, so it would be:\n\n\tSET autocommit TO 'on';\n\tVACUUM;\n\tSET autocommit TO 'off';\n\nThat is a pain, but it is probably the safest, as you explained.\n\nOne particularly nasty problem is issuing a VACUUM or TRUNCATE in cases\nwhere you don't know the autocommit mode. You could set autocommit to\n'on', and issue the command, but how do you know if you need to turn\nautocommit back off again? I suppose you have to conditionally test the\nautocommit value and reset it after the command if needed.\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, 21 Oct 2002 19:08:10 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> One particularly nasty problem is issuing a VACUUM or TRUNCATE in cases\n> where you don't know the autocommit mode. You could set autocommit to\n> 'on', and issue the command, but how do you know if you need to turn\n> autocommit back off again?\n\nPerhaps RESET AUTOCOMMIT is a good enough answer?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 19:10:14 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > One particularly nasty problem is issuing a VACUUM or TRUNCATE in cases\n> > where you don't know the autocommit mode. You could set autocommit to\n> > 'on', and issue the command, but how do you know if you need to turn\n> > autocommit back off again?\n> \n> Perhaps RESET AUTOCOMMIT is a good enough answer?\n\nI was unclear on that. RESET sets it back to the postgresql.conf value,\nright? Do we know that the session didn't change it earlier in the\nscript? That's where it gets tricky.\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, 21 Oct 2002 19:11:52 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Tom Lane wrote:\n>> Perhaps RESET AUTOCOMMIT is a good enough answer?\n\n> I was unclear on that. RESET sets it back to the postgresql.conf value,\n> right? Do we know that the session didn't change it earlier in the\n> script? That's where it gets tricky.\n\nYou're postulating a scenario in which some snippet of code doesn't know\nwhat the surrounding script/application likes for AUTOCOMMIT, but does\nknow enough about the context to know that it's not inside a transaction\nblock already. That combination seems moderately implausible to me.\nAnyone have an example where it'd really be useful?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 19:21:40 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Tom Lane wrote:\n> >> Perhaps RESET AUTOCOMMIT is a good enough answer?\n> \n> > I was unclear on that. RESET sets it back to the postgresql.conf value,\n> > right? Do we know that the session didn't change it earlier in the\n> > script? That's where it gets tricky.\n> \n> You're postulating a scenario in which some snippet of code doesn't know\n> what the surrounding script/application likes for AUTOCOMMIT, but does\n> know enough about the context to know that it's not inside a transaction\n> block already. That combination seems moderately implausible to me.\n> Anyone have an example where it'd really be useful?\n\nWell, in most cases, if you don't know, you do BEGIN ... COMMIT, but in\nthe case of VACUUM/TRUNCATE, you can't do that, so you need some smarts.\n\nIt is a contrived example. I am just throwing it out for illumination.\nI often throw out scenarios at the edges to see if it strikes anyone as\na problem. When it doesn't, we can move ahead.\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, 21 Oct 2002 19:25:25 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
}
] |
[
{
"msg_contents": "I think we should move /contrib/retep to gborg. It is 400k and isn't\nanywhere near our core functionality. I will add to TODO:\n\n\tMove /contrib/retep to gborg\n\nCan I move this during beta? I don't think so.\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, 18 Oct 2002 22:30:40 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "/contrib/retep to gborg"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> I think we should move /contrib/retep to gborg. It is 400k and isn't\n> anywhere near our core functionality. I will add to TODO:\n> \tMove /contrib/retep to gborg\n> Can I move this during beta? I don't think so.\n\nWhy not?\n\nFeature addition during beta is frowned on, but I never heard anyone say\nyou couldn't remove features during beta. (In fact that's a pretty\nstandard response to serious bugs, no?)\n\nWhether retep *should* be moved is not something I feel qualified to\ncomment on, but I don't put any stock in an argument that it can't\nbe done simply because we're in beta.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 18 Oct 2002 23:33:48 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: /contrib/retep to gborg "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > I think we should move /contrib/retep to gborg. It is 400k and isn't\n> > anywhere near our core functionality. I will add to TODO:\n> > \tMove /contrib/retep to gborg\n> > Can I move this during beta? I don't think so.\n> \n> Why not?\n> \n> Feature addition during beta is frowned on, but I never heard anyone say\n> you couldn't remove features during beta. (In fact that's a pretty\n> standard response to serious bugs, no?)\n> \n> Whether retep *should* be moved is not something I feel qualified to\n> comment on, but I don't put any stock in an argument that it can't\n> be done simply because we're in beta.\n\nretep are specialized java tools. They are quite far removed from our\ncore system and have had almost zero activity or mention since being\nadded a few years ago.\n\nI am concerned that removing them may cause instability, particularly in\nsome of the java build scripts. That's why I am suggesting not doing it\nduring 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": "Sat, 19 Oct 2002 01:05:34 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: /contrib/retep to gborg"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> I am concerned that removing them may cause instability, particularly in\n> some of the java build scripts. That's why I am suggesting not doing it\n> during beta.\n\nActually, it looks to me like they belong with the JDBC driver. As long\nas JDBC is in the main distro, I think we should leave contrib/retep\nthere too.\n\nThere was some talk of moving JDBC to gborg, but I'm not sure what the\nplan is, if any.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 19 Oct 2002 12:16:53 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: /contrib/retep to gborg "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > I am concerned that removing them may cause instability, particularly in\n> > some of the java build scripts. That's why I am suggesting not doing it\n> > during beta.\n> \n> Actually, it looks to me like they belong with the JDBC driver. As long\n> as JDBC is in the main distro, I think we should leave contrib/retep\n> there too.\n\nWell, the top of the /contrib/retep README has:\n\n\tBefore you ask what retepTools are, they are my personal suite of\n\tutilities. About 90% of them are JDBC related (either they use JDBC, or\n\tI use them in developing the JDBC driver).\n\nIt is java, but only tangentially related to the driver. It is very far\nfrom our core functionality and seems better in gborg. Just because it\nis java doesn't mean it has to be near jdbc.\n\n> There was some talk of moving JDBC to gborg, but I'm not sure what the\n> plan is, if any.\n\nJDBC folks aren't excited about removing from the main cvs until the\nvitality of gborg is more proven.\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, 19 Oct 2002 22:17:17 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: /contrib/retep to gborg"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Tom Lane wrote:\n>> Actually, it looks to me like they belong with the JDBC driver.\n\n> It is java, but only tangentially related to the driver.\n\nI doubt that either of us are qualified to opine on that point.\nPerhaps you should be asking in pgsql-jdbc whether it's okay to remove\ncontrib/retep. If no one there objects, then I won't complain.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 19 Oct 2002 22:30:17 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: /contrib/retep to gborg "
},
{
"msg_contents": "\nJDBC folks, can someone comment on whether /contrib/retep should be in\nthe /contrib directory, or on gborg? Is it something of general\nusefulness/interest?\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Tom Lane wrote:\n> >> Actually, it looks to me like they belong with the JDBC driver.\n> \n> > It is java, but only tangentially related to the driver.\n> \n> I doubt that either of us are qualified to opine on that point.\n> Perhaps you should be asking in pgsql-jdbc whether it's okay to remove\n> contrib/retep. If no one there objects, then I won't complain.\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": "Sat, 19 Oct 2002 22:32:06 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] /contrib/retep to gborg"
},
{
"msg_contents": "\nI just shot an email over to jdbc. Let's see what they say.\n\n---------------------------------------------------------------------------\n\nBarry Lind wrote:\n> I don't object. In fact I have never even looked at this code before \n> now. I doubt anyone else on the jdbc list would mind either, but it \n> wouldn't hurt to ask.\n> \n> --Barry\n> \n> Tom Lane wrote:\n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > \n> >>Tom Lane wrote:\n> >>\n> >>>Actually, it looks to me like they belong with the JDBC driver.\n> >>\n> > \n> >>It is java, but only tangentially related to the driver.\n> > \n> > \n> > I doubt that either of us are qualified to opine on that point.\n> > Perhaps you should be asking in pgsql-jdbc whether it's okay to remove\n> > contrib/retep. If no one there objects, then I won't complain.\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> ---------------------------(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": "Sat, 19 Oct 2002 22:40:33 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: /contrib/retep to gborg"
},
{
"msg_contents": "I don't object. In fact I have never even looked at this code before \nnow. I doubt anyone else on the jdbc list would mind either, but it \nwouldn't hurt to ask.\n\n--Barry\n\nTom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> \n>>Tom Lane wrote:\n>>\n>>>Actually, it looks to me like they belong with the JDBC driver.\n>>\n> \n>>It is java, but only tangentially related to the driver.\n> \n> \n> I doubt that either of us are qualified to opine on that point.\n> Perhaps you should be asking in pgsql-jdbc whether it's okay to remove\n> contrib/retep. If no one there objects, then I won't complain.\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",
"msg_date": "Sat, 19 Oct 2002 19:41:29 -0700",
"msg_from": "Barry Lind <blind@xythos.com>",
"msg_from_op": false,
"msg_subject": "Re: /contrib/retep to gborg"
}
] |
[
{
"msg_contents": "\nOK, after a week of concentrated work, we now have fewer open items. To\nwrap this up:\n\n\tMichael, please use bison 1.75 to update ecpg\n\tMarc, please install bison 1.75 on postgresql.org\n\tPhilip, please submit your pg_dump patch\n\tTom, please try to fix the vacuum bug\n\nThis will complete almost all our items and we can start to focus on\ndocumentation and final release.\n\n---------------------------------------------------------------------------\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://momjian.postgresql.org/pub/postgresql/open_items.\n\nRequired Changes\n-------------------\nSchema handling - ready? interfaces? client apps?\nDrop column handling - ready for all clients, apps?\nGet bison upgrade on postgresql.org for ecpg only (Marc)\nFix vacuum btree bug (Tom)\nFix pg_dump to handle 64-bit off_t offsets for custom format (Philip)\n\nOptional Changes\n----------------\nAdd schema dump option to pg_dump\nMove /contrib/earthdistance back into main CVS\n\nDocumentation Changes\n---------------------\nMove documation to gborg for moved projects\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, 18 Oct 2002 22:33:42 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Open items"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> \tMichael, please use bison 1.75 to update ecpg\n\nWhy is this necessary? AFAIK we don't keep bison-derived files in\nCVS...\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n",
"msg_date": "19 Oct 2002 04:15:09 -0400",
"msg_from": "Neil Conway <neilc@samurai.com>",
"msg_from_op": false,
"msg_subject": "Re: Open items"
},
{
"msg_contents": "Neil Conway wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > \tMichael, please use bison 1.75 to update ecpg\n> \n> Why is this necessary? AFAIK we don't keep bison-derived files in\n> CVS...\n\nRight, they aren't in CVS, but they are shipped in the tarball so we\ndon't require everyone to have bison. Developers and people building\nfrom CVS do need bison, and will need 1.75 for ecpg.\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, 19 Oct 2002 09:39:47 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Open items"
}
] |
[
{
"msg_contents": "I am curious as to why it is necessary to add PL/pgSQL as a language\ninstead of including it by default.\nIs it to save space on small systems or???\n\nSince the code necessary PL/pgSQL is very small, it does not seem like\nthere would be much savings. I imagine that there is occasionally some\nconfusion with manuals describing the use of PL/pgSQL and yet the\nlanguage does not exist until installation.\n",
"msg_date": "Fri, 18 Oct 2002 19:54:54 -0700",
"msg_from": "\"Dann Corbit\" <DCorbit@connx.com>",
"msg_from_op": true,
"msg_subject": "Design decision curiosity"
},
{
"msg_contents": "\"Dann Corbit\" <DCorbit@connx.com> writes:\n> I am curious as to why it is necessary to add PL/pgSQL as a language\n> instead of including it by default.\n\nI believe security was one of the reasons.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 18 Oct 2002 23:42:33 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Design decision curiosity "
}
] |
[
{
"msg_contents": "\n\nI notice there's a leak of memory in SPI_prepare().\n\nThe full fix is nontrival and I don't want to submit a half solution so I\nthought I'd check whether people think it's worth worrying about.\n\nThe leak is that memory is grabbed in SPI_prepare() for a plan within whatever\ncontext is current when it does the palloc(). It may be the caller's or it may\nbe the relevent SPI one. The plan is then copied out of this memory [and\ncontext] into a child of the procedure's context and forgotten about, or just\nplain forgotten. Obviously the intention is that this memory is freed when the\ncontext is deleted and is probably not a problem unless someone does something\nlike:\n\ni = 100000;\nwhile (i--)\n{\n plan = SPI_prepare(\"SELECT 1\", 0, (Oid *)NULL);\n SPI_freeplan(plan); /* SPI_freeplan() is not just for SPI_saveplan() */\n}\n\nIs this worth worrying about?\n\nAny busy person can stop reading now as the above defines the problem while the\nbelow only shows an easily reproducable example.\n\nFWIW, I found it while testing something like, which is a little less daft\nthan the above example:\n\ncreate function atest1 ( ) returns int as '\n a = 0\n while a < 10000:\n plan = plpy.prepare(\"SELECT \" + repr(a))\n a = a + 1\n' language 'plpython';\n\nHere the plpython code uses SPI_freeplan to release the context holding the\nplan memory when each plan object returned by plpy.prepare() is garbage\ncollected. This seems sensibly to happen when the plan variable is\nreassigned. However I was baffled why the process still had an obvious memory\nleak so looked a little closer at SPI.\n\n\n-- \nNigel J. Andrews\n\n",
"msg_date": "Sat, 19 Oct 2002 17:27:05 +0100 (BST)",
"msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>",
"msg_from_op": true,
"msg_subject": "Freeing plan memory"
},
{
"msg_contents": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> The leak is that memory is grabbed in SPI_prepare() for a plan within\n> whatever context is current when it does the palloc(). It may be the\n> caller's or it may be the relevent SPI one. The plan is then copied\n> out of this memory [and context] into a child of the procedure's\n> context and forgotten about, or just plain forgotten.\n\nAu contraire: SPI_prepare builds the plan in its \"execCxt\", which is\nreset before returning (look at _SPI_begin_call and _SPI_end_call).\nSo I see no leak there.\n\nI'm not sure where the leak is in your plpython example, but I'd be\ninclined to look to plpython itself, perhaps even just the string\nconcatenation expression in\n\tplan = plpy.prepare(\"SELECT \" + repr(a))\n\nplpgsql used to have terrible intra-function memory leaks, and only by\ndint of much hard work has it been brought to the point where you can\nexpect a long loop in a plpgsql function not to chew up memory. AFAIK,\nno one has yet done similar work for the other PL languages.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 19 Oct 2002 12:59:14 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Freeing plan memory "
},
{
"msg_contents": "On Sat, 19 Oct 2002, Tom Lane wrote:\n\n> \"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> > The leak is that memory is grabbed in SPI_prepare() for a plan within\n> > whatever context is current when it does the palloc(). It may be the\n> > caller's or it may be the relevent SPI one. The plan is then copied\n> > out of this memory [and context] into a child of the procedure's\n> > context and forgotten about, or just plain forgotten.\n> \n> Au contraire: SPI_prepare builds the plan in its \"execCxt\", which is\n> reset before returning (look at _SPI_begin_call and _SPI_end_call).\n> So I see no leak there.\n\nAh, yes, I see that now.\n\n> I'm not sure where the leak is in your plpython example, but I'd be\n> inclined to look to plpython itself, perhaps even just the string\n> concatenation expression in\n> \tplan = plpy.prepare(\"SELECT \" + repr(a))\n\nWell it's not that string operation.\n\n> plpgsql used to have terrible intra-function memory leaks, and only by\n> dint of much hard work has it been brought to the point where you can\n> expect a long loop in a plpgsql function not to chew up memory. AFAIK,\n> no one has yet done similar work for the other PL languages.\n\nHmmm...my test case should boil down to a fairly small number of other calls in\nthe SPI_prepare wrapper and a quick looks doesn't show anything\ninteresting. Not sure I've got the time to dedicate to investigating this but\nI'll look at it as and when I can.\n\nI'm sending a patch for plpython.c to -patches which fixes a mistake I made in\nthe previous patch.\n\n\n-- \nNigel J. Andrews\n\n",
"msg_date": "Sat, 19 Oct 2002 18:27:23 +0100 (BST)",
"msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: Freeing plan memory"
},
{
"msg_contents": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> On Sat, 19 Oct 2002, Tom Lane wrote:\n>> I'm not sure where the leak is in your plpython example, but I'd be\n>> inclined to look to plpython itself, perhaps even just the string\n>> concatenation expression in\n>> plan = plpy.prepare(\"SELECT \" + repr(a))\n\n> Well it's not that string operation.\n\nActually, I'll bet it's this code in PLy_spi_prepare:\n\n\tplan->plan = SPI_prepare(query, plan->nargs, plan->types);\n\t// error check\n\n\tplan->plan = SPI_saveplan(plan->plan);\n\t// error check\n\nThe copy of the plan that's returned by SPI_prepare is being blithely\nlost --- and since it's in the procCxt, it won't go away until the\nplpython function is exited. Need a SPI_freeplan() here, I think.\nCan you check it out and send a patch?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 19 Oct 2002 13:38:39 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Freeing plan memory "
}
] |
[
{
"msg_contents": "CREATE FUNCTION \"x\" (integer) RETURNS integer AS '\nBEGIN\n\tRETURN x(0);\nEND' LANGUAGE 'plpgsql' WITH ( isstrict );\nSELECT x(0);\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\nThe log says:\n DEBUG: server process (pid 3550) was terminated by signal 11\n DEBUG: terminating any other active server processes\n\nThe thing that worries me most is the last message: postmaster terminates all\nother backends, therefore overflowing the stack via a plpgsql function can be\nused to DoS-attack the server.\n\nIs there anything one (as a postgresql installation admin) can do? Disabling\nprocedural languages in databases of the users I don't trust comes to mind, but\ndoes it really make me safe? I mean: won't such attack still be possible with\npure SQL?\n\nregards,\n\nMarcin\n-- \nMarcin Owsiany <marcin@owsiany.pl> http://marcin.owsiany.pl/\nGnuPG: 1024D/60F41216 FE67 DA2D 0ACA FC5E 3F75 D6F6 3A0D 8AA0 60F4 1216\n \n\"Every program in development at MIT expands until it can read mail.\"\n -- Unknown\n",
"msg_date": "Sat, 19 Oct 2002 19:17:13 +0200",
"msg_from": "Marcin Owsiany <marcin@owsiany.pl>",
"msg_from_op": true,
"msg_subject": "Security implications of (plpgsql) functions"
},
{
"msg_contents": "\n[ Thread moved to hackers.]\n\nCrash reproduced here.\n\nMy guess is that you are recursing and crashing the backend, which then\nforces the other backends to reset. I think we need to fix this by\neither setting a limit in the amount of function recursion, or allowing\nonly the offending backend to crash without forcing all the other\nbackends to crash.\n\n---------------------------------------------------------------------------\n\nMarcin Owsiany wrote:\n> CREATE FUNCTION \"x\" (integer) RETURNS integer AS '\n> BEGIN\n> \tRETURN x(0);\n> END' LANGUAGE 'plpgsql' WITH ( isstrict );\n> SELECT x(0);\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> The log says:\n> DEBUG: server process (pid 3550) was terminated by signal 11\n> DEBUG: terminating any other active server processes\n> \n> The thing that worries me most is the last message: postmaster terminates all\n> other backends, therefore overflowing the stack via a plpgsql function can be\n> used to DoS-attack the server.\n> \n> Is there anything one (as a postgresql installation admin) can do? Disabling\n> procedural languages in databases of the users I don't trust comes to mind, but\n> does it really make me safe? I mean: won't such attack still be possible with\n> pure SQL?\n> \n> regards,\n> \n> Marcin\n> -- \n> Marcin Owsiany <marcin@owsiany.pl> http://marcin.owsiany.pl/\n> GnuPG: 1024D/60F41216 FE67 DA2D 0ACA FC5E 3F75 D6F6 3A0D 8AA0 60F4 1216\n> \n> \"Every program in development at MIT expands until it can read mail.\"\n> -- Unknown\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, 21 Oct 2002 11:51:05 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Security implications of (plpgsql) functions"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Crash reproduced here.\n\nFWIW, I got a regular \"out of memory\" elog. But I can see that this\nwould depend on the relative sizes of data limit and stack limit on\na particular platform.\n\n> I think we need to fix this by\n> either setting a limit in the amount of function recursion, or allowing\n> only the offending backend to crash without forcing all the other\n> backends to crash.\n\nUnless you can think of a way to distinguish stack overflow from other\nkinds of SIGSEGV, I don't think we can avoid a system-wide restart.\nReducing stack overflow to a plain elog(ERROR) would be really nice,\nbut how?\n\nA depth limit for PL-function recursion is perhaps feasible, but I can't\nsay that I care for it a whole lot ... anyone have better ideas?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 12:02:42 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Security implications of (plpgsql) functions "
},
{
"msg_contents": "Tom Lane wrote:\n> A depth limit for PL-function recursion is perhaps feasible, but I can't\n> say that I care for it a whole lot ... anyone have better ideas?\n> \n\nIs there any way to recognize infinite recursion by analyzing the saved \nexecution tree -- i.e. can we assume that a function that calls itself, with \nthe same arguments with which it was called, constitutes infinite recursion?\n\nJoe\n\n",
"msg_date": "Mon, 21 Oct 2002 09:20:36 -0700",
"msg_from": "Joe Conway <mail@joeconway.com>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Security implications of (plpgsql) functions"
},
{
"msg_contents": "Joe Conway <mail@joeconway.com> writes:\n\n> Tom Lane wrote:\n> > A depth limit for PL-function recursion is perhaps feasible, but I can't\n> > say that I care for it a whole lot ... anyone have better ideas?\n> >\n> \n> Is there any way to recognize infinite recursion by analyzing the\n> saved execution tree -- i.e. can we assume that a function that calls\n> itself, with the same arguments with which it was called, constitutes\n> infinite recursion?\n\nSolved the halting problem lately? ;)\n\nSomeone determined to DoS could probably get around any practical\nimplementation of your idea, using dummy argument, mutual recursion or\nwhatever. \n\n-Doug\n",
"msg_date": "21 Oct 2002 12:27:20 -0400",
"msg_from": "Doug McNaught <doug@wireboard.com>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Security implications of (plpgsql) functions"
},
{
"msg_contents": "Doug McNaught wrote:\n> Solved the halting problem lately? ;)\n\nnah -- I'll leave that as an exercise for the reader ;-)\n\n> Someone determined to DoS could probably get around any practical\n> implementation of your idea, using dummy argument, mutual recursion or\n> whatever. \n> \n\nI see your point. I guess a max recursion limit would be the way to go. \nProbably should be a configurable setting.\n\nJoe\n\n",
"msg_date": "Mon, 21 Oct 2002 09:39:51 -0700",
"msg_from": "Joe Conway <mail@joeconway.com>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Security implications of (plpgsql) functions"
},
{
"msg_contents": "Joe Conway <mail@joeconway.com> writes:\n> Is there any way to recognize infinite recursion by analyzing the saved \n> execution tree -- i.e. can we assume that a function that calls itself, with \n> the same arguments with which it was called, constitutes infinite recursion?\n\nA bulletproof solution would be equivalent to solving the halting\nproblem, I believe. The test you mentioned is easily defeated by\nrecursing between two functions. Also, a would-be instigator of\nDOS doesn't need *infinite* recursion; it could be quite finite and\nstill blow out your stack. For example ask for factorial(10million)\nwhere factorial is defined in the traditional recursive way...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 12:44:41 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Security implications of (plpgsql) functions "
},
{
"msg_contents": "On Saturday 19 Oct 2002 5:17 pm, Marcin Owsiany wrote:\n[snipped infinite recursion function]\n> The log says:\n> DEBUG: server process (pid 3550) was terminated by signal 11\n> DEBUG: terminating any other active server processes\n>\n> The thing that worries me most is the last message: postmaster terminates\n> all other backends, therefore overflowing the stack via a plpgsql function\n> can be used to DoS-attack the server.\n\nYou need to be careful allowing create access for *any* procedural language \n(and paranoid about 'C'). If nothing else, someone could craft a procedure \nthat allocates a lot of memory and just wait. I'm note sure a sig11 is what \nwe'd be looking for here though - I'd expect \"unable to allocate memory\". If \none of the developers don't spot this, I'd report it as a bug.\n\n> Is there anything one (as a postgresql installation admin) can do? \n> Disabling procedural languages in databases of the users I don't trust\n> comes to mind, but does it really make me safe? I mean: won't such attack\n> still be possible with pure SQL?\n\nYou can run a DoS fairly simply if you can connect multiple times - just \nselect * from bigtable as a, bigtable as b will soon exhaust memory. It's not \ndifficult to craft queries to drain CPU/disk IO either. You can control the \namount of shared memory, sort memory and number of connections allowed though \n(see postgresql.conf) which gives an upper limit on the drain PG places on \nyour machine.\n\nYou're along the right lines, restricting access to languages - only grant \npermissions where needed (this applies to tables etc too).\n-- \n Richard Huxton\n",
"msg_date": "Mon, 21 Oct 2002 16:56:40 +0000",
"msg_from": "Richard Huxton <dev@archonet.com>",
"msg_from_op": false,
"msg_subject": "Re: Security implications of (plpgsql) functions"
},
{
"msg_contents": "I've seen the idea of \"user resource limits\" bandied about before as a\nway to address these problems; depending on implementation that might be\nthe way to go.\n\nRobert Treat\n\nOn Mon, 2002-10-21 at 12:44, Tom Lane wrote:\n> Joe Conway <mail@joeconway.com> writes:\n> > Is there any way to recognize infinite recursion by analyzing the saved \n> > execution tree -- i.e. can we assume that a function that calls itself, with \n> > the same arguments with which it was called, constitutes infinite recursion?\n> \n> A bulletproof solution would be equivalent to solving the halting\n> problem, I believe. The test you mentioned is easily defeated by\n> recursing between two functions. Also, a would-be instigator of\n> DOS doesn't need *infinite* recursion; it could be quite finite and\n> still blow out your stack. For example ask for factorial(10million)\n> where factorial is defined in the traditional recursive way...\n> \n> \t\t\tregards, tom lane\n> \n\n\n\n",
"msg_date": "21 Oct 2002 14:17:47 -0400",
"msg_from": "Robert Treat <xzilla@users.sourceforge.net>",
"msg_from_op": false,
"msg_subject": "Re: [GENERAL] Security implications of (plpgsql) functions"
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Tom Lane [mailto:tgl@sss.pgh.pa.us] \n> Sent: 19 October 2002 17:17\n> To: Bruce Momjian\n> Cc: PostgreSQL-development\n> Subject: Re: [HACKERS] /contrib/retep to gborg \n> \n> \n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > I am concerned that removing them may cause instability, \n> particularly \n> > in some of the java build scripts. That's why I am suggesting not \n> > doing it during beta.\n> \n> Actually, it looks to me like they belong with the JDBC \n> driver. As long as JDBC is in the main distro, I think we \n> should leave contrib/retep there too.\n> \n> There was some talk of moving JDBC to gborg, but I'm not sure \n> what the plan is, if any.\n\nMarc asked me to sort it along with some other things, but when I spoke\nto Barry he didn't want to move. I passed it back to Marc.\n\nRegards, Dave.\n",
"msg_date": "Sat, 19 Oct 2002 20:36:35 +0100",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: /contrib/retep to gborg "
},
{
"msg_contents": "On Sat, 19 Oct 2002, Dave Page wrote:\n\n> \n> \n> > -----Original Message-----\n> > From: Tom Lane [mailto:tgl@sss.pgh.pa.us] \n> > Sent: 19 October 2002 17:17\n> > To: Bruce Momjian\n> > Cc: PostgreSQL-development\n> > Subject: Re: [HACKERS] /contrib/retep to gborg \n> > \n> > \n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > I am concerned that removing them may cause instability, \n> > particularly \n> > > in some of the java build scripts. That's why I am suggesting not \n> > > doing it during beta.\n> > \n> > Actually, it looks to me like they belong with the JDBC \n> > driver. As long as JDBC is in the main distro, I think we \n> > should leave contrib/retep there too.\n> > \n> > There was some talk of moving JDBC to gborg, but I'm not sure \n> > what the plan is, if any.\n> \n> Marc asked me to sort it along with some other things, but when I spoke\n> to Barry he didn't want to move. I passed it back to Marc.\n\nI did say to Marc about removing contrib/retep several months ago, as more \nup to date versions have been on sourceforge for almost a year now.\n\nI agree with Bruce about the build scripts, but suggest once everything is \ndone, to remove them.\n\nPeter\n\n-- \nPeter Mount\npeter@retep.org.uk\nhttp://www.retep.org.uk/\n Tel/Fax: +44 (0) 1622 749439\n Mobile: +44 (0) 7903 155887\n US Fax: 1 435 304 5165\nUS Voice: 1 435 304 5165\n\n\n",
"msg_date": "Sun, 20 Oct 2002 10:12:07 +0000 (UTC)",
"msg_from": "Peter Mount <peter@retep.org.uk>",
"msg_from_op": false,
"msg_subject": "Re: /contrib/retep to gborg "
},
{
"msg_contents": "\nOK, /contrib/retep removed from CVS. Thanks.\n\n---------------------------------------------------------------------------\n\nPeter Mount wrote:\n> On Sat, 19 Oct 2002, Dave Page wrote:\n> \n> > \n> > \n> > > -----Original Message-----\n> > > From: Tom Lane [mailto:tgl@sss.pgh.pa.us] \n> > > Sent: 19 October 2002 17:17\n> > > To: Bruce Momjian\n> > > Cc: PostgreSQL-development\n> > > Subject: Re: [HACKERS] /contrib/retep to gborg \n> > > \n> > > \n> > > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > > I am concerned that removing them may cause instability, \n> > > particularly \n> > > > in some of the java build scripts. That's why I am suggesting not \n> > > > doing it during beta.\n> > > \n> > > Actually, it looks to me like they belong with the JDBC \n> > > driver. As long as JDBC is in the main distro, I think we \n> > > should leave contrib/retep there too.\n> > > \n> > > There was some talk of moving JDBC to gborg, but I'm not sure \n> > > what the plan is, if any.\n> > \n> > Marc asked me to sort it along with some other things, but when I spoke\n> > to Barry he didn't want to move. I passed it back to Marc.\n> \n> I did say to Marc about removing contrib/retep several months ago, as more \n> up to date versions have been on sourceforge for almost a year now.\n> \n> I agree with Bruce about the build scripts, but suggest once everything is \n> done, to remove them.\n> \n> Peter\n> \n> -- \n> Peter Mount\n> peter@retep.org.uk\n> http://www.retep.org.uk/\n> Tel/Fax: +44 (0) 1622 749439\n> Mobile: +44 (0) 7903 155887\n> US Fax: 1 435 304 5165\n> US Voice: 1 435 304 5165\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": "Sun, 20 Oct 2002 20:36:06 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: /contrib/retep to gborg"
}
] |
[
{
"msg_contents": "Remember how we make SET/SHOW/RESET _not_ be part of a multi-statement\ntransaction when it is at the start of a transaction and autocommit is\noff?\n\nWell, look at this:\n\t\n\ttest=> SET random_page_cost = 2;\n\tSET\n\ttest=> COMMIT;\n\tWARNING: COMMIT: no transaction in progress\n\tCOMMIT\n\nThe WARNING happens with SHOW and RESET too. I wonder if we should\nsuppress the WARNING of a COMMIT with no statements when autocommit is\noff. This will probably be better for portability, though again, it is\nconfusing.\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, 19 Oct 2002 22:56:49 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "New SET/autocommit problem"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> \tWARNING: COMMIT: no transaction in progress\n> \tCOMMIT\n\n> The WARNING happens with SHOW and RESET too. I wonder if we should\n> suppress the WARNING of a COMMIT with no statements when autocommit is\n> off.\n\nI don't think so; that will make it quite hard to check whether a\ntransaction is open.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 19 Oct 2002 23:49:46 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: New SET/autocommit problem "
},
{
"msg_contents": "> > \tWARNING: COMMIT: no transaction in progress\n> > \tCOMMIT\n> \n> > The WARNING happens with SHOW and RESET too. I wonder if we should\n> > suppress the WARNING of a COMMIT with no statements when autocommit is\n> > off.\n> \n> I don't think so; that will make it quite hard to check whether a\n> transaction is open.\n\nI've got tons of these warnings in my logs... is there a programmatic\nway of determining if the current session is in a transaction? Unless\nI misunderstood the fix and the commit message, I'm pretty sure that\nmost of my problem has been fixed in CVS with SET's auto-committing if\nit's not in a transaction, but now that there are some intricate rules\nwith regards to starting transactions, I'd love to provide a DBI\ninterface into a call that returns whether or not we're in a\ntransaction to prevent millions of these:\n\nNOTICE: ROLLBACK: no transaction in progress\n\n-sc\n\n-- \nSean Chittenden\n",
"msg_date": "Tue, 22 Oct 2002 00:12:41 -0700",
"msg_from": "Sean Chittenden <sean@chittenden.org>",
"msg_from_op": false,
"msg_subject": "Re: New SET/autocommit problem"
},
{
"msg_contents": "Sean Chittenden <sean@chittenden.org> writes:\n> WARNING: COMMIT: no transaction in progress\n\n> I've got tons of these warnings in my logs... is there a programmatic\n> way of determining if the current session is in a transaction?\n\nNot at present: you have to track it for yourself. One of the\nsuggestions on the list for the next frontend/backend protocol revision\n(probably in 7.4) is to add a way for the backend to signal its\ntransaction state: no transaction, in transaction, or in failed\ntransaction seems like the set of possible states.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 22 Oct 2002 09:24:30 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: New SET/autocommit problem "
},
{
"msg_contents": "> > WARNING: COMMIT: no transaction in progress\n> \n> > I've got tons of these warnings in my logs... is there a programmatic\n> > way of determining if the current session is in a transaction?\n> \n> Not at present: you have to track it for yourself. One of the\n> suggestions on the list for the next frontend/backend protocol revision\n> (probably in 7.4) is to add a way for the backend to signal its\n> transaction state: no transaction, in transaction, or in failed\n> transaction seems like the set of possible states.\n\nThat would be fabulous because with autocommit set to off, the\ncomplexity for tracking that in application is getting pretty\ngnarly. -sc\n\n-- \nSean Chittenden\n",
"msg_date": "Tue, 22 Oct 2002 10:10:46 -0700",
"msg_from": "Sean Chittenden <sean@chittenden.org>",
"msg_from_op": false,
"msg_subject": "Re: New SET/autocommit problem"
}
] |
[
{
"msg_contents": "> Remember how we make SET/SHOW/RESET _not_ be part of a multi-statement\n> transaction when it is at the start of a transaction and autocommit is\n> off?\n> \n> Well, look at this:\n> \n> test=> SET random_page_cost = 2;\n> SET\n> test=> COMMIT;\n> WARNING: COMMIT: no transaction in progress\n> COMMIT\n> \n> The WARNING happens with SHOW and RESET too. I wonder if we should\n> suppress the WARNING of a COMMIT with no statements when autocommit is\n> off. This will probably be better for portability, though again, it is\n> confusing.\n\n[ Sorry, I deleted Tom's reply.]\n\nTom, you mentioned suppressing the WARNING on COMMIT of an empty\ntransaction would make it hard to know when you are in a transaction,\nbut I was suggesting suppressing the warning only when autocommit was\noff, so by definition you are always in a transaction, sort of. You are\nin a transaction, but perhaps an empty one. Should it be OK to issue a\nCOMMIT of an empty transaction when autocommit is off?\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, 20 Oct 2002 00:40:20 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "New problem with SET/autocommit"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Tom, you mentioned suppressing the WARNING on COMMIT of an empty\n> transaction would make it hard to know when you are in a transaction,\n> but I was suggesting suppressing the warning only when autocommit was\n> off, so by definition you are always in a transaction, sort of. You are\n> in a transaction, but perhaps an empty one.\n\nI don't understand why you're labeling this behavior as a problem.\nTo me, it's the expected behavior, it's useful in debugging, and it\ndoes not actually break anything. (A WARNING is not an ERROR. Though\nI'd not object if you'd like to downgrade the begin/commit/rollback\nwrong-state WARNINGs to NOTICEs, like they were before.)\n\n> Should it be OK to issue a\n> COMMIT of an empty transaction when autocommit is off?\n\nWe need to be careful about adding more and more special cases to\nthe transactional rules. The more there are, the more confusing\nand hard-to-maintain the system will be. I don't think this proposed\nspecial case is justified: it has no value except to suppress a notice.\nMoreover, it's suppressing a notice in a context where the user\ndemonstrably has a misunderstanding of the transactional behavior.\nDon't we usually throw notices to try to teach people what they\nmay be doing wrong?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 20 Oct 2002 01:26:48 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: New problem with SET/autocommit "
},
{
"msg_contents": "\nOK, I guess it isn't a problem.\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Tom, you mentioned suppressing the WARNING on COMMIT of an empty\n> > transaction would make it hard to know when you are in a transaction,\n> > but I was suggesting suppressing the warning only when autocommit was\n> > off, so by definition you are always in a transaction, sort of. You are\n> > in a transaction, but perhaps an empty one.\n> \n> I don't understand why you're labeling this behavior as a problem.\n> To me, it's the expected behavior, it's useful in debugging, and it\n> does not actually break anything. (A WARNING is not an ERROR. Though\n> I'd not object if you'd like to downgrade the begin/commit/rollback\n> wrong-state WARNINGs to NOTICEs, like they were before.)\n> \n> > Should it be OK to issue a\n> > COMMIT of an empty transaction when autocommit is off?\n> \n> We need to be careful about adding more and more special cases to\n> the transactional rules. The more there are, the more confusing\n> and hard-to-maintain the system will be. I don't think this proposed\n> special case is justified: it has no value except to suppress a notice.\n> Moreover, it's suppressing a notice in a context where the user\n> demonstrably has a misunderstanding of the transactional behavior.\n> Don't we usually throw notices to try to teach people what they\n> may be doing wrong?\n> \n> \t\t\tregards, tom lane\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, 20 Oct 2002 15:57:52 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: New problem with SET/autocommit"
}
] |
[
{
"msg_contents": "Hello.\r\nI wonder when postgresql 7.3 will release.\r\nthank you very much.\r\nJinqiang Han\r\n\n\n\n\n\nHello.\nI wonder when postgresql 7.3 will release.\nthank you very much.\nJinqiang Han",
"msg_date": "Sun, 20 Oct 2002 21:39:53 +0800",
"msg_from": "Jinqiang Han <postgresql@db.pku.edu.cn>",
"msg_from_op": true,
"msg_subject": "a question about postgresql 7.3beta2"
}
] |
[
{
"msg_contents": "When I try to grant usage on language c in 7.3b1 I get an error message\nsaying I can't because the language isn't trusted. Is this different in b2?\nI can see using whether or not a language is trusted to set defaults,\nbut now that access can be controlled it would be useful to grant\naccess to load c functions to users that are trusted.\n\nHere is the example:\ngrant usage on language c to bruno;\nERROR: language \"c\" is not trusted\n",
"msg_date": "Sun, 20 Oct 2002 10:50:59 -0500",
"msg_from": "Bruno Wolff III <bruno@wolff.to>",
"msg_from_op": true,
"msg_subject": "Problem granting usage on language c"
},
{
"msg_contents": "On Sun, Oct 20, 2002 at 10:50:59 -0500,\n Bruno Wolff III <bruno@wolff.to> wrote:\n> When I try to grant usage on language c in 7.3b1 I get an error message\n> saying I can't because the language isn't trusted. Is this different in b2?\n> I can see using whether or not a language is trusted to set defaults,\n> but now that access can be controlled it would be useful to grant\n> access to load c functions to users that are trusted.\n\nLooking at yesterday's CVS suggests that it is intended for untrusted\nlanguages to only be used by superusers.\n\nI would like to run up a trial balloon that the follwoing changes be made:\nWhen creating new languages, untrusted ones default to not granting usage\nto public and trusted ones default to granting usage to public. And that\naccess to languages be controlled by just the usage permission and not\nwhether a language is trusted or not.\n\nThe argument for this is that trusted users may want to work in an environment\nwhere some things are blocked off to prevent shooting one's self in the foot.\nSo that you might want to be able to create c functions without being able\nto do anything.\n\nThe risk is that the admins have to realize that granting access to untrusted\nlanguages is giving away the keys to the castle.\n",
"msg_date": "Sun, 20 Oct 2002 11:25:34 -0500",
"msg_from": "Bruno Wolff III <bruno@wolff.to>",
"msg_from_op": true,
"msg_subject": "Re: Problem granting usage on language c"
},
{
"msg_contents": "Bruno Wolff III <bruno@wolff.to> writes:\n> When I try to grant usage on language c in 7.3b1 I get an error message\n> saying I can't because the language isn't trusted. Is this different in b2?\n\nNo.\n\n> I can see using whether or not a language is trusted to set defaults,\n> but now that access can be controlled it would be useful to grant\n> access to load c functions to users that are trusted.\n\nThere is no point at all in providing shades of access to untrusted\nlanguages. A user who can define his own C functions is God: he can\nuse the functions to bypass any security restrictions you may think\nyou have on him. So it's useless to say that any privilege state\nless than superuser-dom is appropriate for creating C functions.\n\nlater ...\n> The risk is that the admins have to realize that granting access to\n> untrusted languages is giving away the keys to the castle.\n\nEquating it to superuser status seems an effective way of reminding\nthem of that.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 20 Oct 2002 13:28:31 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Problem granting usage on language c "
}
] |
[
{
"msg_contents": "We are installing postgres on a MacOS X Server...\n\nOn our Linux box, we had to muddle with some parameters to be able to\nincrease the shared memory and buffers in postgres.\n\nDoes anyone know how to do this in OS X? Or is it required?\n\nThanks,\nHunter\n\n",
"msg_date": "Sun, 20 Oct 2002 12:05:04 -0700",
"msg_from": "Hunter Hillegas <lists@lastonepicked.com>",
"msg_from_op": true,
"msg_subject": "Buffers and MacOS X "
},
{
"msg_contents": "Hunter Hillegas <lists@lastonepicked.com> writes:\n> We are installing postgres on a MacOS X Server...\n> On our Linux box, we had to muddle with some parameters to be able to\n> increase the shared memory and buffers in postgres.\n> Does anyone know how to do this in OS X? Or is it required?\n\nIn OS X 10.1, the max shared memory was a hardwired number that couldn't\nbe changed without recompiling the kernel --- and the standard setting\nwas none too darn large, either :-(. I dunno if Apple has improved\nmatters in 10.2.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 20 Oct 2002 20:48:31 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Buffers and MacOS X "
},
{
"msg_contents": "Tom Lane wrote:\n> Hunter Hillegas <lists@lastonepicked.com> writes:\n> > We are installing postgres on a MacOS X Server...\n> > On our Linux box, we had to muddle with some parameters to be able to\n> > increase the shared memory and buffers in postgres.\n> > Does anyone know how to do this in OS X? Or is it required?\n> \n> In OS X 10.1, the max shared memory was a hardwired number that couldn't\n> be changed without recompiling the kernel --- and the standard setting\n> was none too darn large, either :-(. I dunno if Apple has improved\n> matters in 10.2.\n\nI was going to point him to the development docs, but I see no mention\nof OSX shared kernel limit tuning in there at all. Can someone submit\ninfo to beef it up for OSX?\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, 20 Oct 2002 21:13:29 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Buffers and MacOS X"
},
{
"msg_contents": ">Tom Lane wrote:\n>> Hunter Hillegas <lists@lastonepicked.com> writes:\n>> > We are installing postgres on a MacOS X Server...\n>> > On our Linux box, we had to muddle with some parameters to be able to\n>> > increase the shared memory and buffers in postgres.\n>> > Does anyone know how to do this in OS X? Or is it required?\n>>\n>> In OS X 10.1, the max shared memory was a hardwired number that couldn't\n>> be changed without recompiling the kernel --- and the standard setting\n>> was none too darn large, either :-(. I dunno if Apple has improved\n>> matters in 10.2.\n>\n>I was going to point him to the development docs, but I see no mention\n>of OSX shared kernel limit tuning in there at all. Can someone submit\n>info to beef it up for OSX?\n\n\nHere are instructions for setting shared memory limits on Mac OS X 10.2.\n\nEdit the file:\n\n\t/System/Library/StartupItems/SystemTuning/SystemTuning\n\nNear the bottom, you will see these lines:\n\n sysctl -w kern.sysv.shmmax=4194304\n sysctl -w kern.sysv.shmmin=1\n sysctl -w kern.sysv.shmmni=32\n sysctl -w kern.sysv.shmseg=8\n sysctl -w kern.sysv.shmall=1024\n\nThese parameters are described in this document:\n\n\thttp://www.postgresql.org/idocs/index.php?kernel-resources.html\n\nOS X is simialar to BSD/OS in the values for each parameter. For my system, I want PostgreSQL to use 128MB worth of buffers. I used the calculations provided in the \"Reasonable Values\" column of the parameters table in the document above.\n\nSo here are my settings (allowing a little extra room):\n\n sysctl -w kern.sysv.shmmax=167772160\t# bytes\n sysctl -w kern.sysv.shmmin=1\n sysctl -w kern.sysv.shmmni=32\n sysctl -w kern.sysv.shmseg=8\n sysctl -w kern.sysv.shmall=65536\t\t# 4K pages\n\nNow, I can set shared_buffers in postgresql.conf to:\n\n\tshared_buffers = 16384\t# number of buffers, 8K each\n\n- Jeff\n-- \n\nJeff Bohmer\nVisionLink, Inc.\n_________________________________\n303.402.0170\nwww.visionlink.org\n_________________________________\nPeople. Tools. Change. Community.\n",
"msg_date": "Mon, 21 Oct 2002 10:08:52 -0600",
"msg_from": "Jeffrey Bohmer <bohmer@visionlink.org>",
"msg_from_op": false,
"msg_subject": "Re: Buffers and MacOS X"
},
{
"msg_contents": "\nAdded to 7.3 documentation. Thanks.\n\n---------------------------------------------------------------------------\n\nJeffrey Bohmer wrote:\n> >Tom Lane wrote:\n> >> Hunter Hillegas <lists@lastonepicked.com> writes:\n> >> > We are installing postgres on a MacOS X Server...\n> >> > On our Linux box, we had to muddle with some parameters to be able to\n> >> > increase the shared memory and buffers in postgres.\n> >> > Does anyone know how to do this in OS X? Or is it required?\n> >>\n> >> In OS X 10.1, the max shared memory was a hardwired number that couldn't\n> >> be changed without recompiling the kernel --- and the standard setting\n> >> was none too darn large, either :-(. I dunno if Apple has improved\n> >> matters in 10.2.\n> >\n> >I was going to point him to the development docs, but I see no mention\n> >of OSX shared kernel limit tuning in there at all. Can someone submit\n> >info to beef it up for OSX?\n> \n> \n> Here are instructions for setting shared memory limits on Mac OS X 10.2.\n> \n> Edit the file:\n> \n> \t/System/Library/StartupItems/SystemTuning/SystemTuning\n> \n> Near the bottom, you will see these lines:\n> \n> sysctl -w kern.sysv.shmmax=4194304\n> sysctl -w kern.sysv.shmmin=1\n> sysctl -w kern.sysv.shmmni=32\n> sysctl -w kern.sysv.shmseg=8\n> sysctl -w kern.sysv.shmall=1024\n> \n> These parameters are described in this document:\n> \n> \thttp://www.postgresql.org/idocs/index.php?kernel-resources.html\n> \n> OS X is simialar to BSD/OS in the values for each parameter. For my system, I want PostgreSQL to use 128MB worth of buffers. I used the calculations provided in the \"Reasonable Values\" column of the parameters table in the document above.\n> \n> So here are my settings (allowing a little extra room):\n> \n> sysctl -w kern.sysv.shmmax=167772160\t# bytes\n> sysctl -w kern.sysv.shmmin=1\n> sysctl -w kern.sysv.shmmni=32\n> sysctl -w kern.sysv.shmseg=8\n> sysctl -w kern.sysv.shmall=65536\t\t# 4K pages\n> \n> Now, I can set shared_buffers in postgresql.conf to:\n> \n> \tshared_buffers = 16384\t# number of buffers, 8K each\n> \n> - Jeff\n> -- \n> \n> Jeff Bohmer\n> VisionLink, Inc.\n> _________________________________\n> 303.402.0170\n> www.visionlink.org\n> _________________________________\n> People. Tools. Change. Community.\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": "Tue, 5 Nov 2002 18:16:37 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Buffers and MacOS X"
}
] |
[
{
"msg_contents": "Does anyone know why this returns 0 and not 3600?\n\t\n\ttest=> select '1970-01-01 01:00:00'::timestamp::abstime::int4;\n\t int4 \n\t------\n\t 0\n\t(1 row)\n\nMy time(2) manual page says:\n\n The time() function returns the value of time in seconds since 0\n hours, 0 minutes, 0 seconds, January 1, 1970, Coordinated Universal Time.\n\nAlso, why does my TimeZone show as Poland?\n\t\n\ttest=> SHOW TimeZone;\n\t TimeZone \n\t----------\n\t Poland\n\t(1 row)\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, 20 Oct 2002 21:46:13 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "timestamp cast to abstime problem"
},
{
"msg_contents": "\nI figured this all out; turns out they were related.\n\nI had set my timezone in the postmaster to Poland long ago and forgot,\nand that is why both the abstime and show timezone produced strange\nresults. Sorry.\n\n---------------------------------------------------------------------------\n\nBruce Momjian wrote:\n> Does anyone know why this returns 0 and not 3600?\n> \t\n> \ttest=> select '1970-01-01 01:00:00'::timestamp::abstime::int4;\n> \t int4 \n> \t------\n> \t 0\n> \t(1 row)\n> \n> My time(2) manual page says:\n> \n> The time() function returns the value of time in seconds since 0\n> hours, 0 minutes, 0 seconds, January 1, 1970, Coordinated Universal Time.\n> \n> Also, why does my TimeZone show as Poland?\n> \t\n> \ttest=> SHOW TimeZone;\n> \t TimeZone \n> \t----------\n> \t Poland\n> \t(1 row)\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-- \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, 20 Oct 2002 22:01:04 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: timestamp cast to abstime problem"
}
] |
[
{
"msg_contents": "\nOK, I can now see the 7.3 final release. I have Philip's patch, which\nI will apply sometime tomorrow, with the hope it gets some big-endian\ntesting. If someone can test that, please do. We can't make Tom do\neverything. :-)\n\nMichael, please update ecpg. Marc, please update bison. Other CVS\nbuilders will need to upgrade to Bison 1.75 to build ecpg. \n\n(Marc, also please move /contrib/earthdistance back into the /contrib\nCVS. We are still getting complaints about that.)\n\nWith those items completed in the next day or two, we then have little\nleft on the open list. The remaining items can be marked as done or\nmoved to TODO, and then we can start looking for a release target date.\n\n---------------------------------------------------------------------------\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://momjian.postgresql.org/pub/postgresql/open_items.\n\nRequired Changes\n-------------------\nSchema handling - ready? interfaces? client apps?\nDrop column handling - ready for all clients, apps?\nGet bison upgrade on postgresql.org for ecpg only (Michael, Marc)\nFix pg_dump to handle 64-bit off_t offsets for custom format (Philip)\n\nOptional Changes\n----------------\nAdd schema dump option to pg_dump\nMove /contrib/earthdistance back into main CVS (Marc)\n\nDocumentation Changes\n---------------------\nMove documation to gborg for moved projects\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, 20 Oct 2002 22:18:08 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "I see the end!"
}
] |
[
{
"msg_contents": "\nlet me know if there are any problems with it ....\n\n\n",
"msg_date": "Sun, 20 Oct 2002 23:26:59 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "bison 1.75 installed ..."
},
{
"msg_contents": "\"Marc G. Fournier\" <scrappy@hub.org> writes:\n> let me know if there are any problems with it ....\n\n<grouse>\nOther than the fact that it's about a factor of 16 slower than bison\n1.28, while not offering any substantial gain in functionality? If I\nwere a Bison maintainer, I'd hang my head in shame.\n</grouse>\n\nAll PG regression tests pass here with 1.75-built parsers.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 00:19:10 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: bison 1.75 installed ... "
},
{
"msg_contents": "Tom Lane wrote:\n> \n> All PG regression tests pass here with 1.75-built parsers.\n> \n\nSame here on a Red Hat 7.3 system.\n\nIn case anyone is interested, I made a bison 1.75 rpm based on the 1.35 source \nrpm from RH7.3. It can be found here along with an updated src RPM:\n\nhttp://www.joeconway.com/bison-1.75-custom.1.i686.rpm\nhttp://www.joeconway.com/bison-1.75-custom.1.src.rpm\n\nJoe\n\n",
"msg_date": "Sun, 20 Oct 2002 22:41:08 -0700",
"msg_from": "Joe Conway <mail@joeconway.com>",
"msg_from_op": false,
"msg_subject": "Re: bison 1.75 installed ..."
},
{
"msg_contents": "Tom Lane wrote:\n> \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > let me know if there are any problems with it ....\n> \n> <grouse>\n> Other than the fact that it's about a factor of 16 slower than bison\n> 1.28, while not offering any substantial gain in functionality? If I\n> were a Bison maintainer, I'd hang my head in shame.\n> </grouse>\n> \n> All PG regression tests pass here with 1.75-built parsers.\n\n16x is the grammer output generation, not the actual parsing of the SQL,\nright? There are cases where slow output generation can lead to faster\nparsing, right? Let's hope that happened.\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, 21 Oct 2002 15:01:15 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: bison 1.75 installed ..."
}
] |
[
{
"msg_contents": "Hello all,\n\nNewbie time... where do I browse for CVS membership?\nWhat tools are hackers using for IDE?\n\nAny other advice for a new interest in MIN() MAX()\nand indexes?\n\n\nKeith Gray\nkapgray@ocean.com.au\n\n",
"msg_date": "Mon, 21 Oct 2002 12:55:14 +1000",
"msg_from": "Keith Gray <kapgray@ocean.com.au>",
"msg_from_op": true,
"msg_subject": "CVS - projects"
}
] |
[
{
"msg_contents": "Hi all,\n\nWithout modifying anything, postgresql (since today) has a strange\nbehavior:\n\nAll connections are rejected with No space left on device.\n\nThere's plenty of space in shm, disk...\n\nIt occurs on UNIX as well as on INET socket.\nI can't even shut the postmaster down (it doesn't respond) and have a few\ndefunct process.\n\n\nSHMAX is at 40000000 here, version is 7.2.3 \nHeres postgresql.conf:\n#\n# PostgreSQL configuration file\n# -----------------------------\n#\n# This file consists of lines of the form\n#\n# name = value\n#\n# (The `=' is optional.) White space is collapsed, comments are\n# introduced by `#' anywhere on a line. The complete list of option\n# names and allowed values can be found in the PostgreSQL\n# documentation. The commented-out settings shown in this file\n# represent the default values.\n\n# Any option can also be given as a command line switch to the\n# postmaster, e.g., 'postmaster -c log_connections=on'. Some options\n# can be changed at run-time with the 'SET' SQL command.\n\n\n#========================================================================\n\n\n#\n#\tConnection Parameters\n#\ntcpip_socket = true\n#ssl = false\n\nmax_connections = 64\n\nport = 5432 \nhostname_lookup = true\n#show_source_port = false\n\n#unix_socket_directory = ''\n#unix_socket_group = ''\n#unix_socket_permissions = 0777\n\n#virtual_host = ''\n\n#krb_server_keyfile = ''\n\n\n#\n#\tShared Memory Size\n#\nshared_buffers = 512 # 2*max_connections, min 16\n#max_fsm_relations = 100 # min 10, fsm is free space map\n#max_fsm_pages = 10000 # min 1000, fsm is free space map\n#max_locks_per_transaction = 64 # min 10\n#wal_buffers = 8 # min 4\n\n#\n#\tNon-shared Memory Sizes\n#\nsort_mem = 10240 # min 32\n#vacuum_mem = 8192 # min 1024\n\n\n#\n#\tWrite-ahead log (WAL)\n#\n#wal_files = 4 # range 0-64\n#wal_sync_method = fsync # the default varies across platforms:\n#\t\t\t # fsync, fdatasync, open_sync, or open_datasync\n#wal_debug = 0 # range 0-16\n#commit_delay = 0 # range 0-100000\n#commit_siblings = 5 # range 1-1000\n#checkpoint_segments = 3 # in logfile segments (16MB each), min 1\n#checkpoint_timeout = 300 # in seconds, range 30-3600\n#fsync = true\n\n\n#\n#\tOptimizer Parameters\n#\n#enable_seqscan = true\n#enable_indexscan = true\n#enable_tidscan = true\n#enable_sort = true\n#enable_nestloop = true\n#enable_mergejoin = true\n#enable_hashjoin = true\n\n#ksqo = false\n\n#effective_cache_size = 1000 # default in 8k pages\n#random_page_cost = 4\n#cpu_tuple_cost = 0.01\n#cpu_index_tuple_cost = 0.001\n#cpu_operator_cost = 0.0025\n\n\n#\n#\tGEQO Optimizer Parameters\n#\n#geqo = true\n#geqo_selection_bias = 2.0 # range 1.5-2.0\n#geqo_threshold = 11\n#geqo_pool_size = 0 # default based on #tables in query, range 128-1024\n#geqo_effort = 1\n#geqo_generations = 0\n#geqo_random_seed = -1 # auto-compute seed\n\n\n#\n#\tDebug display\n#\n#silent_mode = false\n\nlog_connections = true\n#log_timestamp = true\nlog_pid = true\n\n#debug_level = 0 # range 0-16\n\ndebug_print_query = true\n#debug_print_parse = false\n#debug_print_rewritten = false\n#debug_print_plan = false\ndebug_pretty_print = true\n\n# requires USE_ASSERT_CHECKING\n#debug_assertions = true\n\n\n#\n#\tSyslog\n#\n# requires ENABLE_SYSLOG\nsyslog = 2 # range 0-2\n#syslog_facility = 'LOCAL0'\n#syslog_ident = 'postgres'\n\n\n#\n#\tStatistics\n#\nshow_parser_stats = false\nshow_planner_stats = false\nshow_executor_stats = false\nshow_query_stats = false\n\n# requires BTREE_BUILD_STATS\n#show_btree_build_stats = false\n\n\n#\n#\tAccess statistics collection\n#\nstats_start_collector = true\nstats_reset_on_server_start = true\nstats_command_string = true\nstats_row_level = true\nstats_block_level = true\n\n\n#\n#\tLock Tracing\n#\n#trace_notify = false\n\n# requires LOCK_DEBUG\n#trace_locks = false\n#trace_userlocks = false\n#trace_lwlocks = false\n#debug_deadlocks = false\n#trace_lock_oidmin = 16384\n#trace_lock_table = 0\n\n\n#\n#\tMisc\n#\n#dynamic_library_path = '$libdir'\n#australian_timezones = false\n#authentication_timeout = 60 # min 1, max 600\n#deadlock_timeout = 1000\n#default_transaction_isolation = 'read committed'\n#max_expr_depth = 10000 # min 10\n#max_files_per_process = 1000 # min 25\n#password_encryption = false\n#sql_inheritance = true\n#transform_null_equals = false\n\nAnyone can come with an idea?\n\nIt NEVER happend before...\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": "Mon, 21 Oct 2002 15:42:39 +0200 (MET DST)",
"msg_from": "Olivier PRENANT <ohp@pyrenet.fr>",
"msg_from_op": true,
"msg_subject": "Please help"
},
{
"msg_contents": "Olivier PRENANT <ohp@pyrenet.fr> writes:\n> Without modifying anything, postgresql (since today) has a strange\n> behavior:\n\n> All connections are rejected with No space left on device.\n\nCould you be out of swap space?\n\nI'd like to see the *exact* context in which you see this error message,\nthough.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 11:02:43 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Please help "
},
{
"msg_contents": "On Monday 21 October 2002 15:42, Olivier PRENANT wrote:\n> Hi all,\n>\n> Without modifying anything, postgresql (since today) has a strange\n> behavior:\n>\n> All connections are rejected with No space left on device.\n>\n> There's plenty of space in shm, disk...\n\nI have no idea whether it's relevant, but maybe you have a problem with\nsemaphores? See:\n\nhttp://www.ca.postgresql.org/docs/faq-english.html#3.4\n\n(A lack of available semaphores can also produce the message \n\"No space left on device.\")\n\nSorry I can't help any further.\n\nIan Barwick\nbarwick@gmx.net\n",
"msg_date": "Mon, 21 Oct 2002 17:16:37 +0200",
"msg_from": "Ian Barwick <barwick@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: Please help"
},
{
"msg_contents": "Thanks for your reply.\n\nActually, I just found xhat happened 2 mn ago!\n\nLast week I changed my pg_hba.conf to require an md5 password for one\nspecific database and updated all my script but ONE.\n\nIt seems that connection from php pg_connect not supplying a password\nlives the process for a \"certain ammount of time\" running, then postmaster\njust hangs.\n\nDon't really know what happens here, but supplying a password on this\nscript made the problem go away... (at least for now)\n\nRegards\n On Mon, 21 Oct 2002, Tom Lane wrote:\n\n> Date: Mon, 21 Oct 2002 11:02:43 -0400\n> From: Tom Lane <tgl@sss.pgh.pa.us>\n> To: ohp@pyrenet.fr\n> Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>\n> Subject: Re: [HACKERS] Please help \n> \n> Olivier PRENANT <ohp@pyrenet.fr> writes:\n> > Without modifying anything, postgresql (since today) has a strange\n> > behavior:\n> \n> > All connections are rejected with No space left on device.\n> \n> Could you be out of swap space?\n> \n> I'd like to see the *exact* context in which you see this error message,\n> though.\n> \n> \t\t\tregards, tom lane\n> \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": "Mon, 21 Oct 2002 17:22:59 +0200 (MET DST)",
"msg_from": "Olivier PRENANT <ohp@pyrenet.fr>",
"msg_from_op": true,
"msg_subject": "Re: Please help "
},
{
"msg_contents": "Olivier PRENANT <ohp@pyrenet.fr> writes:\n> It seems that connection from php pg_connect not supplying a password\n> lives the process for a \"certain ammount of time\" running, then postmaster\n> just hangs.\n\nThat's hard to believe. In 7.2 or later, the backend should give up and\nclose the connection and exit if the client doesn't finish the\nauthentication handshake within 60 seconds.\n\nCan anyone else reproduce a problem with lack of a password on a PHP\nconnection?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 11:35:33 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Please help "
},
{
"msg_contents": "Hi Tom,\n\nI can assure you that there are no problem anymore now that I have put the\npassword.\n\nHowever the modification pg_hba.conf has been done a few days ago and\nnever noticed the error until today where I had a *LOT* of visits to my\nsite.\n\nSo I still think it's a matter of bad connection delay.\n\nConnexion where comming very fast and postmaster did'nt release them fast\nenough.\n\nI now have this problem when I run vacuum analyze:\n\nScript started on Mon Oct 21 18:20:35 2002\n~ 18:20:35: psql pyrenet\nPassword: \nWelcome to psql, the PostgreSQL interactive terminal.\n\nType: \\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\npyrenet=# VACUUM ANALYZE ;\nFATAL 2: could not open transaction-commit log directory (/usr/local/pgsql/data/pg_clog): Too many open files\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: NOTICE: Message from PostgreSQL backend:\n\tThe Postmaster has informed me that some other backend\n\tdied abnormally and possibly corrupted shared memory.\n\tI have rolled back the current transaction and am\n\tgoing to terminate your database system connection and exit.\n\tPlease reconnect to the database system and repeat your query.\nFailed.\n!# \\q\n~ 18:21:21: psql pyrenet\nPassword: \nWelcome to psql, the PostgreSQL interactive terminal.\n\nType: \\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\npyrenet=# VACUUM ANALYZE ;\nFATAL 2: could not open transaction-commit log directory (/usr/local/pgsql/data/pg_clog): Too many open files\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!# \\q\n\nscript done on Mon Oct 21 18:21:51 2002\n\nWhat causes (FILES=64).\n\nAgain, this is the very first time I have such problems on postgresql!!\nIt works so well, it's the central point of my system!!\n On Mon, 21 Oct 2002, Tom Lane wrote:\n\n> Date: Mon, 21 Oct 2002 11:35:33 -0400\n> From: Tom Lane <tgl@sss.pgh.pa.us>\n> To: ohp@pyrenet.fr\n> Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>\n> Subject: Re: [HACKERS] Please help \n> \n> Olivier PRENANT <ohp@pyrenet.fr> writes:\n> > It seems that connection from php pg_connect not supplying a password\n> > lives the process for a \"certain ammount of time\" running, then postmaster\n> > just hangs.\n> \n> That's hard to believe. In 7.2 or later, the backend should give up and\n> close the connection and exit if the client doesn't finish the\n> authentication handshake within 60 seconds.\n> \n> Can anyone else reproduce a problem with lack of a password on a PHP\n> connection?\n> \n> \t\t\tregards, tom lane\n> \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": "Mon, 21 Oct 2002 18:23:53 +0200 (MET DST)",
"msg_from": "Olivier PRENANT <ohp@pyrenet.fr>",
"msg_from_op": true,
"msg_subject": "Re: Please help "
},
{
"msg_contents": "Olivier PRENANT <ohp@pyrenet.fr> writes:\n> pyrenet=# VACUUM ANALYZE ;\n> FATAL 2: could not open transaction-commit log directory (/usr/local/pgsql/data/pg_clog): Too many open files\n\nHmm. Do you have any idea what sysconf(_SC_OPEN_MAX) returns on your\nplatform? You could try reducing the max_files_per_process parameter.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 21 Oct 2002 12:52:10 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Please help "
},
{
"msg_contents": "On Mon, 21 Oct 2002, Tom Lane wrote:\n\n> Date: Mon, 21 Oct 2002 12:52:10 -0400\n> From: Tom Lane <tgl@sss.pgh.pa.us>\n> To: ohp@pyrenet.fr\n> Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>\n> Subject: Re: [HACKERS] Please help \n> \n> Olivier PRENANT <ohp@pyrenet.fr> writes:\n> > pyrenet=# VACUUM ANALYZE ;\n> > FATAL 2: could not open transaction-commit log directory (/usr/local/pgsql/data/pg_clog): Too many open files\n> \n> Hmm. Do you have any idea what sysconf(_SC_OPEN_MAX) returns on your\n> platform? You could try reducing the max_files_per_process parameter.\nThe point is, it occurs today for the very first time!\nQuestion: does (with 7.2) augmenting max_connection suffice, or do I have\nto recompile?\n\nThat's the only thing that comes to my mind! I changed max_coneections\n(and related parameters) in postgresql.conf only...\n\nI say that, because I tried to change socket_directory in postgresql.conf \nand clients didn't work anymore\n> \n> \t\t\tregards, tom lane\n> \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": "Mon, 21 Oct 2002 19:26:18 +0200 (MET DST)",
"msg_from": "Olivier PRENANT <ohp@pyrenet.fr>",
"msg_from_op": true,
"msg_subject": "Re: Please help "
},
{
"msg_contents": "On Mon, 2002-10-21 at 12:26, Olivier PRENANT wrote:\n> On Mon, 21 Oct 2002, Tom Lane wrote:\n> \n> > Date: Mon, 21 Oct 2002 12:52:10 -0400\n> > From: Tom Lane <tgl@sss.pgh.pa.us>\n> > To: ohp@pyrenet.fr\n> > Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>\n> > Subject: Re: [HACKERS] Please help \n> > \n> > Olivier PRENANT <ohp@pyrenet.fr> writes:\n> > > pyrenet=# VACUUM ANALYZE ;\n> > > FATAL 2: could not open transaction-commit log directory (/usr/local/pgsql/data/pg_clog): Too many open files\n> > \n> > Hmm. Do you have any idea what sysconf(_SC_OPEN_MAX) returns on your\n> > platform? You could try reducing the max_files_per_process parameter.\n> The point is, it occurs today for the very first time!\n> Question: does (with 7.2) augmenting max_connection suffice, or do I have\n> to recompile?\nYou might need to up the Shared Memory parameters and the Semaphore\nParameters in your OS (UnixWare IIRC). \n> \n> That's the only thing that comes to my mind! I changed max_coneections\n> (and related parameters) in postgresql.conf only...\n> \n> I say that, because I tried to change socket_directory in postgresql.conf \n> and clients didn't work anymore\nSee above. \n\n\n> > \n> > \t\t\tregards, tom lane\n> > \n> \n> -- \n> Olivier PRENANT \tTel:\t+33-5-61-50-97-00 (Work)\n> Quartier d'Harraud Turrou +33-5-61-50-97-01 (Fax)\n> 31190 AUTERIVE +33-6-07-63-80-64 (GSM)\n> FRANCE Email: ohp@pyrenet.fr\n> ------------------------------------------------------------------------------\n> Make your life a dream, make your dream a reality. (St Exupery)\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\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": "21 Oct 2002 12:34:48 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": false,
"msg_subject": "Re: Please help"
},
{
"msg_contents": "Hi larry,\n\nGlad to see you around...\nOn 21 Oct 2002, Larry Rosenman wrote:\n\n> Date: 21 Oct 2002 12:34:48 -0500\n> From: Larry Rosenman <ler@lerctr.org>\n> To: ohp@pyrenet.fr\n> Cc: Tom Lane <tgl@sss.pgh.pa.us>,\n> pgsql-hackers list <pgsql-hackers@postgresql.org>\n> Subject: Re: [HACKERS] Please help\n> > The point is, it occurs today for the very first time!\n> > Question: does (with 7.2) augmenting max_connection suffice, or do I have\n> > to recompile?\n> You might need to up the Shared Memory parameters and the Semaphore\n> Parameters in your OS (UnixWare IIRC). \nI did!\n> > \n> > That's the only thing that comes to my mind! I changed max_coneections\n> > (and related parameters) in postgresql.conf only...\n> > \n> > I say that, because I tried to change socket_directory in postgresql.conf \n> > and clients didn't work anymore\nSorry, I mis-explain!\nI mean changing socket_directory in postgresql.conf and restart server did\ncreate .s.PGSQL.5432 in the new dir, however clients (like psql) still\nwant it in /tmp!!\n\n > See above. \n> \n> \n> > > \n> > > \t\t\tregards, tom lane\n> > > \n> > \n> > -- \n> > Olivier PRENANT \tTel:\t+33-5-61-50-97-00 (Work)\n> > Quartier d'Harraud Turrou +33-5-61-50-97-01 (Fax)\n> > 31190 AUTERIVE +33-6-07-63-80-64 (GSM)\n> > FRANCE Email: ohp@pyrenet.fr\n> > ------------------------------------------------------------------------------\n> > Make your life a dream, make your dream a reality. (St Exupery)\n> > \n> > \n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 4: Don't 'kill -9' the postmaster\n> > \n> \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": "Mon, 21 Oct 2002 19:39:53 +0200 (MET DST)",
"msg_from": "Olivier PRENANT <ohp@pyrenet.fr>",
"msg_from_op": true,
"msg_subject": "Re: Please help"
},
{
"msg_contents": "On Mon, 2002-10-21 at 12:39, Olivier PRENANT wrote:\n> Hi larry,\n> \n> Glad to see you around...\n> On 21 Oct 2002, Larry Rosenman wrote:\n> \n> > Date: 21 Oct 2002 12:34:48 -0500\n> > From: Larry Rosenman <ler@lerctr.org>\n> > To: ohp@pyrenet.fr\n> > Cc: Tom Lane <tgl@sss.pgh.pa.us>,\n> > pgsql-hackers list <pgsql-hackers@postgresql.org>\n> > Subject: Re: [HACKERS] Please help\n> > > The point is, it occurs today for the very first time!\n> > > Question: does (with 7.2) augmenting max_connection suffice, or do I have\n> > > to recompile?\n> > You might need to up the Shared Memory parameters and the Semaphore\n> > Parameters in your OS (UnixWare IIRC). \n> I did!\nOk.\n> > > \n> > > That's the only thing that comes to my mind! I changed max_coneections\n> > > (and related parameters) in postgresql.conf only...\n> > > \n> > > I say that, because I tried to change socket_directory in postgresql.conf \n> > > and clients didn't work anymore\n> Sorry, I mis-explain!\n> I mean changing socket_directory in postgresql.conf and restart server did\n> create .s.PGSQL.5432 in the new dir, however clients (like psql) still\n> want it in /tmp!!\nThat **WOULD** take a recompile. \n\nLER\n> \n> > See above. \n> > \n> > \n> > > > \n> > > > \t\t\tregards, tom lane\n> > > > \n> > > \n> > > -- \n> > > Olivier PRENANT \tTel:\t+33-5-61-50-97-00 (Work)\n> > > Quartier d'Harraud Turrou +33-5-61-50-97-01 (Fax)\n> > > 31190 AUTERIVE +33-6-07-63-80-64 (GSM)\n> > > FRANCE Email: ohp@pyrenet.fr\n> > > ------------------------------------------------------------------------------\n> > > Make your life a dream, make your dream a reality. (St Exupery)\n> > > \n> > > \n> > > ---------------------------(end of broadcast)---------------------------\n> > > TIP 4: Don't 'kill -9' the postmaster\n> > > \n> > \n> \n> -- \n> Olivier PRENANT \tTel:\t+33-5-61-50-97-00 (Work)\n> Quartier d'Harraud Turrou +33-5-61-50-97-01 (Fax)\n> 31190 AUTERIVE +33-6-07-63-80-64 (GSM)\n> FRANCE Email: ohp@pyrenet.fr\n> ------------------------------------------------------------------------------\n> Make your life a dream, make your dream a reality. (St Exupery)\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": "21 Oct 2002 12:57:19 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": false,
"msg_subject": "Re: Please help"
},
{
"msg_contents": "On Mon, 2002-10-21 at 12:57, Larry Rosenman wrote:\n> On Mon, 2002-10-21 at 12:39, Olivier PRENANT wrote:\n> > Hi larry,\n> > \n> > Glad to see you around...\n> > On 21 Oct 2002, Larry Rosenman wrote:\n> > \n> > > Date: 21 Oct 2002 12:34:48 -0500\n> > > From: Larry Rosenman <ler@lerctr.org>\n> > > To: ohp@pyrenet.fr\n> > > Cc: Tom Lane <tgl@sss.pgh.pa.us>,\n> > > pgsql-hackers list <pgsql-hackers@postgresql.org>\n> > > Subject: Re: [HACKERS] Please help\n> > > > The point is, it occurs today for the very first time!\n> > > > Question: does (with 7.2) augmenting max_connection suffice, or do I have\n> > > > to recompile?\n> > > You might need to up the Shared Memory parameters and the Semaphore\n> > > Parameters in your OS (UnixWare IIRC). \n> > I did!\n> Ok.\n> > > > \n> > > > That's the only thing that comes to my mind! I changed max_coneections\n> > > > (and related parameters) in postgresql.conf only...\n> > > > \n> > > > I say that, because I tried to change socket_directory in postgresql.conf \n> > > > and clients didn't work anymore\n> > Sorry, I mis-explain!\n> > I mean changing socket_directory in postgresql.conf and restart server did\n> > create .s.PGSQL.5432 in the new dir, however clients (like psql) still\n> > want it in /tmp!!\n> That **WOULD** take a recompile. \nOr (IIRC), changing the connect string passed from PHP to PostgreSQL.\n\n> \n> LER\n> > \n> > > See above. \n> > > \n> > > \n> > > > > \n> > > > > \t\t\tregards, tom lane\n> > > > > \n> > > > \n> > > > -- \n> > > > Olivier PRENANT \tTel:\t+33-5-61-50-97-00 (Work)\n> > > > Quartier d'Harraud Turrou +33-5-61-50-97-01 (Fax)\n> > > > 31190 AUTERIVE +33-6-07-63-80-64 (GSM)\n> > > > FRANCE Email: ohp@pyrenet.fr\n> > > > ------------------------------------------------------------------------------\n> > > > Make your life a dream, make your dream a reality. (St Exupery)\n> > > > \n> > > > \n> > > > ---------------------------(end of broadcast)---------------------------\n> > > > TIP 4: Don't 'kill -9' the postmaster\n> > > > \n> > > \n> > \n> > -- \n> > Olivier PRENANT \tTel:\t+33-5-61-50-97-00 (Work)\n> > Quartier d'Harraud Turrou +33-5-61-50-97-01 (Fax)\n> > 31190 AUTERIVE +33-6-07-63-80-64 (GSM)\n> > FRANCE Email: ohp@pyrenet.fr\n> > ------------------------------------------------------------------------------\n> > Make your life a dream, make your dream a reality. (St Exupery)\n> -- \n> Larry Rosenman http://www.lerctr.org/~ler\n> Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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-- \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": "21 Oct 2002 12:59:38 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": false,
"msg_subject": "Re: Please help"
}
] |
[
{
"msg_contents": "Hi, \n\nI just merged the big_bison branch back into HEAD. Hopefully I did this\nin the correct way. Keep in mind that I'm not really fit in using cvs.\n\nPlease test 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": "Mon, 21 Oct 2002 15:59:12 +0200",
"msg_from": "Michael Meskes <meskes@postgresql.org>",
"msg_from_op": true,
"msg_subject": "ECPG"
}
] |
[
{
"msg_contents": "Hi everyone,\n\nThanks to Cornelia Boenigk <poppcorn@cornelia-boenigk.de> and Michael\nPaesold <mpaesold@gmx.at>, the German translation of the PostgreSQL\n\"Advocacy and Marketing\" site, is now completed and ready for public\nuse:\n\nhttp://advocacy.postgresql.org/?lang=de\n\nThis makes 5 languages done, with Brazilian Portuguese and Spanish\nnearly ready too.\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": "Tue, 22 Oct 2002 09:05:19 +1000",
"msg_from": "Justin Clift <justin@postgresql.org>",
"msg_from_op": true,
"msg_subject": "German version of the PostgreSQL \"Advocacy and Marketing\" site is\n\tready"
},
{
"msg_contents": "Folks,\n\nFLYER: I can start work on the flyer next week sometime. Should I just use \nthe text from the website? Does anyone have any ideas for graphics? Are we \nlooking for a 1-page, 2-page, 4-page flyer or 3-fold brochure?\n\n7.3 PRESS RELEASE: Postgres 7.3 will be coming out soon. We should do a \npress release; I even have a press list of tech press we can e-mail it to. \nBut I think we should start on wording now, so that we can have it nailed \ndown and translated the day that 7.3 is ready.\n\tI'm sure that Justin has already thought of this, but one of the reasons that \nMySQL is so much higher profile than us is press releases. They get press \nmilage out of announcing \"enhancements\" which we have had for years. Time to \nturn that around ...\n\n-- \n-Josh Berkus\n Aglio Database Solutions\n San Francisco\n\n",
"msg_date": "Mon, 21 Oct 2002 17:04:36 -0700",
"msg_from": "Josh Berkus <josh@agliodbs.com>",
"msg_from_op": false,
"msg_subject": "Flyer, Press Release"
},
{
"msg_contents": "On 21 Oct 2002 at 17:04, Josh Berkus wrote:\n\n> Folks,\n> \n> FLYER: I can start work on the flyer next week sometime. Should I just use \n> the text from the website? Does anyone have any ideas for graphics? Are we \n> looking for a 1-page, 2-page, 4-page flyer or 3-fold brochure?\n> \n> 7.3 PRESS RELEASE: Postgres 7.3 will be coming out soon. We should do a \n> press release; I even have a press list of tech press we can e-mail it to. \n> But I think we should start on wording now, so that we can have it nailed \n> down and translated the day that 7.3 is ready.\n> \tI'm sure that Justin has already thought of this, but one of the reasons that \n> MySQL is so much higher profile than us is press releases. They get press \n> milage out of announcing \"enhancements\" which we have had for years. Time to \n> turn that around ...\n\nApart from that, I think we need to post HOWTO's which can enable one to start \nusing database in less than 5 minutes. Getting this HOWTO to as many people as \npossible, would let them know how easy it is to start using postgresql..\n\nBye\n Shridhar\n\n--\nASCII:\tThe control code for all beginning programmers and those who would\t\nbecome computer literate. Etymologically, the term has come down as\ta \ncontraction of the often-repeated phrase \"ascii and you shall\treceive.\"\t\t-- \nRobb Russon\n\n",
"msg_date": "Tue, 22 Oct 2002 12:12:08 +0530",
"msg_from": "\"Shridhar Daithankar\" <shridhar_daithankar@persistent.co.in>",
"msg_from_op": false,
"msg_subject": "Re: Flyer, Press Release"
}
] |
[
{
"msg_contents": "\nOK, we are down to one open item, related to pg_dumping on 64-bit off_t.\nWe had discussion today on this so it should be completed shortly.\n\n---------------------------------------------------------------------------\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://momjian.postgresql.org/pub/postgresql/open_items.\n\nRequired Changes\n-------------------\nSchema handling - ready? interfaces? client apps?\nDrop column handling - ready for all clients, apps?\nFix pg_dump to handle 64-bit off_t offsets for custom format (Philip)\n\nOptional Changes\n----------------\n\nDocumentation Changes\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, 21 Oct 2002 19:18:23 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "One 7.3 item left"
},
{
"msg_contents": "> Schema handling - ready? interfaces? client apps?\n\nWhat is the state of the Perl interface?\n\nWill it work when 7.3 is released \nWill it work, but no schema support\nWill it pop up later on CPAN\n\n-- \nKaare Rasmussen --Linux, spil,-- Tlf: 3816 2582\nKaki Data tshirts, merchandize Fax: 3816 2501\nHowitzvej 75 �ben 12.00-18.00 Email: kar@kakidata.dk\n2000 Frederiksberg L�rdag 12.00-16.00 Web: www.suse.dk\n",
"msg_date": "Tue, 22 Oct 2002 19:01:20 +0200",
"msg_from": "Kaare Rasmussen <kar@kakidata.dk>",
"msg_from_op": false,
"msg_subject": "Re: One 7.3 item left"
},
{
"msg_contents": "Kaare Rasmussen wrote:\n> > Schema handling - ready? interfaces? client apps?\n> \n> What is the state of the Perl interface?\n> \n> Will it work when 7.3 is released \n> Will it work, but no schema support\n> Will it pop up later on CPAN\n\nWe have a separate gborg project for the old perl5 in interface and\ndbd-pg. The DBD group is making improvements right now. Not sure how\nit works with 7.3 but I am sure they will get to testing it soon. David\nWheeler is working on it, and he is involved in 7.3.\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, 22 Oct 2002 13:10:21 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: One 7.3 item left"
},
{
"msg_contents": "Bruce Momjian writes:\n\n> OK, we are down to one open item, related to pg_dumping on 64-bit off_t.\n> We had discussion today on this so it should be completed shortly.\n\nI hate to spoil the fun, but we have at least the Linux + Perl\n5.8.1 + _GNU_SOURCE and the AIX + Large File + Flex problems to fix.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Tue, 22 Oct 2002 22:11:18 +0200 (CEST)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: One 7.3 item left"
},
{
"msg_contents": "Peter Eisentraut <peter_e@gmx.net> writes:\n> Bruce Momjian writes:\n>> OK, we are down to one open item, related to pg_dumping on 64-bit off_t.\n>> We had discussion today on this so it should be completed shortly.\n\n> I hate to spoil the fun, but we have at least the Linux + Perl\n> 5.8.1 + _GNU_SOURCE and the AIX + Large File + Flex problems to fix.\n\nWe should not, however, wait longer before pushing out a beta3 release.\nPortability problems on individual platforms may hold up RC1, but we're\noverdue to put out a final beta...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 22 Oct 2002 16:15:42 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: One 7.3 item left "
},
{
"msg_contents": "On Tue, 22 Oct 2002, Tom Lane wrote:\n\n> Peter Eisentraut <peter_e@gmx.net> writes:\n> > Bruce Momjian writes:\n> >> OK, we are down to one open item, related to pg_dumping on 64-bit off_t.\n> >> We had discussion today on this so it should be completed shortly.\n>\n> > I hate to spoil the fun, but we have at least the Linux + Perl\n> > 5.8.1 + _GNU_SOURCE and the AIX + Large File + Flex problems to fix.\n>\n> We should not, however, wait longer before pushing out a beta3 release.\n> Portability problems on individual platforms may hold up RC1, but we're\n> overdue to put out a final beta...\n\nWas just about to ask that ... Friday sound reasonable for beta3 then?\nBruce, can you have all your files updated by then?\n\n\n",
"msg_date": "Tue, 22 Oct 2002 22:14:19 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": false,
"msg_subject": "Re: One 7.3 item left "
},
{
"msg_contents": "Marc G. Fournier wrote:\n> On Tue, 22 Oct 2002, Tom Lane wrote:\n> \n> > Peter Eisentraut <peter_e@gmx.net> writes:\n> > > Bruce Momjian writes:\n> > >> OK, we are down to one open item, related to pg_dumping on 64-bit off_t.\n> > >> We had discussion today on this so it should be completed shortly.\n> >\n> > > I hate to spoil the fun, but we have at least the Linux + Perl\n> > > 5.8.1 + _GNU_SOURCE and the AIX + Large File + Flex problems to fix.\n> >\n> > We should not, however, wait longer before pushing out a beta3 release.\n> > Portability problems on individual platforms may hold up RC1, but we're\n> > overdue to put out a final beta...\n> \n> Was just about to ask that ... Friday sound reasonable for beta3 then?\n> Bruce, can you have all your files updated by then?\n\nI can, sure.\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, 22 Oct 2002 21:16:17 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: One 7.3 item left"
},
{
"msg_contents": "On Tue, Oct 22, 2002 at 19:01:20 +0200,\n Kaare Rasmussen <kar@kakidata.dk> wrote:\n> > Schema handling - ready? interfaces? client apps?\n> \n> What is the state of the Perl interface?\n> \n> Will it work when 7.3 is released \n> Will it work, but no schema support\n> Will it pop up later on CPAN\n\nI am using Pg with 7.3b1 and it works OK for what I am doing. I am not\nexplicitly naming schemas when referencing objects though.\n",
"msg_date": "Tue, 22 Oct 2002 22:24:15 -0500",
"msg_from": "Bruno Wolff III <bruno@wolff.to>",
"msg_from_op": false,
"msg_subject": "Re: One 7.3 item left"
},
{
"msg_contents": "Peter Eisentraut wrote:\n> Bruce Momjian writes:\n> \n> > OK, we are down to one open item, related to pg_dumping on 64-bit off_t.\n> > We had discussion today on this so it should be completed shortly.\n> \n> I hate to spoil the fun, but we have at least the Linux + Perl\n> 5.8.1 + _GNU_SOURCE and the AIX + Large File + Flex problems to fix.\n\nI was hoping those had gone away. :-(\n\nOpen items updated.\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://momjian.postgresql.org/pub/postgresql/open_items.\n\nRequired Changes\n-------------------\nSchema handling - ready? interfaces? client apps?\nDrop column handling - ready for all clients, apps?\nAdd configure check for sizeof(off_t) > sizeof(long) and no fseek()\nFix Linux + Perl 5.8.1 + _GNU_SOURCE problem\nFix AIX + Large File + Flex problem\n\nOptional Changes\n----------------\n\nDocumentation Changes\n---------------------",
"msg_date": "Wed, 23 Oct 2002 21:02:46 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: One 7.3 item left"
},
{
"msg_contents": "Marc G. Fournier wrote:\n> On Tue, 22 Oct 2002, Tom Lane wrote:\n> \n> > Peter Eisentraut <peter_e@gmx.net> writes:\n> > > Bruce Momjian writes:\n> > >> OK, we are down to one open item, related to pg_dumping on 64-bit off_t.\n> > >> We had discussion today on this so it should be completed shortly.\n> >\n> > > I hate to spoil the fun, but we have at least the Linux + Perl\n> > > 5.8.1 + _GNU_SOURCE and the AIX + Large File + Flex problems to fix.\n> >\n> > We should not, however, wait longer before pushing out a beta3 release.\n> > Portability problems on individual platforms may hold up RC1, but we're\n> > overdue to put out a final beta...\n> \n> Was just about to ask that ... Friday sound reasonable for beta3 then?\n> Bruce, can you have all your files updated by then?\n\nDone.\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, 23 Oct 2002 23:02:45 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: One 7.3 item left"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Marc G. Fournier wrote:\n>> Was just about to ask that ... Friday sound reasonable for beta3 then?\n>> Bruce, can you have all your files updated by then?\n\n> Done.\n\nAre we going to back-merge CVS tip into the REL7_3_STABLE branch now?\nWhat about opening CVS tip for 7.4 development?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 23 Oct 2002 23:17:03 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: One 7.3 item left "
},
{
"msg_contents": "On Wed, 23 Oct 2002, Tom Lane wrote:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Marc G. Fournier wrote:\n> >> Was just about to ask that ... Friday sound reasonable for beta3 then?\n> >> Bruce, can you have all your files updated by then?\n>\n> > Done.\n>\n> Are we going to back-merge CVS tip into the REL7_3_STABLE branch now?\n> What about opening CVS tip for 7.4 development?\n\nconsidering the number of changes that have lead up to beta3, I'd saw wait\n... I'm almost thinking that rc1 makes more sense to do it, since once rc1\ngoes out, then we're at the point of \"only critical changes\", which means\nalot less commits then what we see through beta ...\n\n\n",
"msg_date": "Thu, 24 Oct 2002 01:18:04 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": false,
"msg_subject": "Re: One 7.3 item left "
},
{
"msg_contents": "I came across a quite interesting issue I don't really understand but \nmaybe Tom will know.\nThis happened rather accidentally.\n\nI have a rather complex query which executes efficiently.\n\nThere is one interesting thing - let's have a look at the query:\n\n\nSELECT t_struktur.id, t_text.code, COUNT(t_wert.wert) AS x\n FROM t_struktur JOIN t_sportstruktur\n ON (t_struktur.id = t_sportstruktur.strukturid),\n t_text, t_master, t_strukturtyp,\n t_master AS a JOIN t_struktur AS b\n ON (a.slave_id = b.id) JOIN t_strukturtyp AS c\n ON (b.typid = c.id),\n t_wert JOIN t_werttyp\n ON (t_werttyp.id = t_wert.werttypid)\n WHERE t_struktur.id = t_text.suchid\n AND t_text.sprache = 1\n AND t_text.texttyp IS NULL\n AND t_text.icode = 'struktur'\n\n AND t_master.master_id IN (11, 6, 10, 9, 5, 3, 7, 8, 13)\n AND t_master.slave_id = t_struktur.id\n AND t_struktur.typid = t_strukturtyp.id\n AND t_strukturtyp.kommentar = 'geoort'\n\n AND a.master_id = t_struktur.id\n AND c.sortierung = '60005'\n\n AND t_sportstruktur.sportid IN (1, 2, 3, 4, 5)\n AND t_struktur.id = t_wert.strukturid\n AND t_werttyp.id = t_wert.werttypid\n AND t_werttyp.anzeige IN (40550, 40555, 40525,\n 41070, 41073, 41075, 41077, 41080,\n 40745, 40750, 40775, 40735, 40780,\n 40785, 40760, 40710, 41110, 41115,\n 41090, 41120, 40265, 41085, 41030,\n 41570, 41550)\n AND (t_wert.wert > '0' OR t_wert.wert = 't')\n GROUP BY t_struktur.id, t_text.code\n ORDER BY x DESC;\n\n\nOn my good old P166 it takes\n\nroot@actionscouts:/tmp# time psql action < c.sql\n id | code | x\n-----+------------+----\n 301 | Schladming | 16\n 204 | Kitzb�hel | 8\n(2 rows)\n\n\nreal 0m1.475s\nuser 0m0.050s\nsys 0m0.010s\n\nIt takes around 5 seconds to execute the query without explicit joins \n(brief comment to the discussion we had yesterday).\n\nAs you can see the query is redundant:\n\nt_wert JOIN t_werttyp\n ON (t_werttyp.id = t_wert.werttypid)\n\nI also use:\n\n AND t_werttyp.id = t_wert.werttypid\n\nI have done with join twice since I have forgotten to remove the line \nbelow when tweaking the stuff.\n\nHowever, when I remove this AND the query is logically the same but ...\n\n\nroot@actionscouts:/tmp# time psql action < c.sql\n id | code | x\n-----+------------+----\n 301 | Schladming | 16\n 204 | Kitzb�hel | 8\n(2 rows)\n\n\nreal 0m2.280s\nuser 0m0.060s\nsys 0m0.010s\n\nIt is 50% slower ...\nDoes anybody have an idea why?\n\nHere are the execution plans - the first one uses the redundant query; \nthe second one does not use the AND in the WHERE clause.\n\n\n\nroot@actionscouts:/tmp# time psql action < c.sql\nNOTICE: QUERY PLAN:\n\nSort (cost=425.34..425.34 rows=1 width=132)\n -> Aggregate (cost=425.32..425.33 rows=1 width=132)\n -> Group (cost=425.32..425.33 rows=1 width=132)\n -> Sort (cost=425.32..425.32 rows=1 width=132)\n -> Nested Loop (cost=240.47..425.31 rows=1 width=132)\n -> Nested Loop (cost=240.47..415.76 rows=1 \nwidth=124)\n -> Hash Join (cost=240.47..399.06 \nrows=1 width=101)\n -> Nested Loop \n(cost=0.00..154.76 rows=765 width=29)\n -> Seq Scan on t_werttyp \n(cost=0.00..14.69 rows=23 width=8)\n -> Index Scan using \nidx_wert_werttypid on t_wert (cost=0.00..5.98 rows=1 width=21)\n -> Hash (cost=240.47..240.47 \nrows=1 width=72)\n -> Hash Join \n(cost=114.57..240.47 rows=1 width=72)\n -> Hash Join \n(cost=22.45..148.23 rows=24 width=40)\n -> Hash Join \n(cost=18.82..128.85 rows=3091 width=32)\n -> Seq \nScan on t_master a (cost=0.00..55.59 rows=3159 width=16)\n -> Hash \n(cost=16.66..16.66 rows=866 width=16)\n -> \nSeq Scan on t_struktur b (cost=0.00..16.66 rows=866 width=16)\n -> Hash \n(cost=3.62..3.62 rows=1 width=8)\n -> Seq \nScan on t_strukturtyp c (cost=0.00..3.62 rows=1 width=8)\n -> Hash \n(cost=92.11..92.11 rows=3 width=32)\n -> Hash Join \n(cost=41.12..92.11 rows=3 width=32)\n -> Hash \nJoin (cost=37.49..86.40 rows=273 width=24)\n -> \nSeq Scan on t_sportstruktur (cost=0.00..44.13 rows=273 width=8)\n -> \nHash (cost=16.66..16.66 rows=866 width=16)\n \n-> Seq Scan on t_struktur (cost=0.00..16.66 rows=866 width=16)\n -> Hash \n(cost=3.62..3.62 rows=1 width=8)\n -> \nSeq Scan on t_strukturtyp (cost=0.00..3.62 rows=1 width=8)\n -> Index Scan using idx_text_suchid on \nt_text (cost=0.00..16.68 rows=1 width=23)\n -> Index Scan using idx_master_slaveid on \nt_master (cost=0.00..9.54 rows=1 width=8)\n\nEXPLAIN\n\nreal 0m0.616s\nuser 0m0.050s\nsys 0m0.010s\n\n\n\n\n\noot@actionscouts:/tmp# time psql action < c.sql\nNOTICE: QUERY PLAN:\n\nSort (cost=824.56..824.56 rows=1 width=132)\n -> Aggregate (cost=824.55..824.55 rows=1 width=132)\n -> Group (cost=824.55..824.55 rows=1 width=132)\n -> Sort (cost=824.55..824.55 rows=1 width=132)\n -> Nested Loop (cost=255.22..824.54 rows=1 width=132)\n -> Nested Loop (cost=255.22..814.98 rows=1 \nwidth=124)\n -> Hash Join (cost=255.22..798.28 \nrows=1 width=101)\n -> Hash Join (cost=14.75..553.98 \nrows=765 width=29)\n -> Seq Scan on t_wert \n(cost=0.00..501.03 rows=5729 width=21)\n -> Hash (cost=14.69..14.69 \nrows=23 width=8)\n -> Seq Scan on \nt_werttyp (cost=0.00..14.69 rows=23 width=8)\n -> Hash (cost=240.47..240.47 \nrows=1 width=72)\n -> Hash Join \n(cost=114.57..240.47 rows=1 width=72)\n -> Hash Join \n(cost=22.45..148.23 rows=24 width=40)\n -> Hash Join \n(cost=18.82..128.85 rows=3091 width=32)\n -> Seq \nScan on t_master a (cost=0.00..55.59 rows=3159 width=16)\n -> Hash \n(cost=16.66..16.66 rows=866 width=16)\n -> \nSeq Scan on t_struktur b (cost=0.00..16.66 rows=866 width=16)\n -> Hash \n(cost=3.62..3.62 rows=1 width=8)\n -> Seq \nScan on t_strukturtyp c (cost=0.00..3.62 rows=1 width=8)\n -> Hash \n(cost=92.11..92.11 rows=3 width=32)\n -> Hash Join \n(cost=41.12..92.11 rows=3 width=32)\n -> Hash \nJoin (cost=37.49..86.40 rows=273 width=24)\n -> \nSeq Scan on t_sportstruktur (cost=0.00..44.13 rows=273 width=8)\n -> \nHash (cost=16.66..16.66 rows=866 width=16)\n \n-> Seq Scan on t_struktur (cost=0.00..16.66 rows=866 width=16)\n -> Hash \n(cost=3.62..3.62 rows=1 width=8)\n -> \nSeq Scan on t_strukturtyp (cost=0.00..3.62 rows=1 width=8)\n -> Index Scan using idx_text_suchid on \nt_text (cost=0.00..16.68 rows=1 width=23)\n -> Index Scan using idx_master_slaveid on \nt_master (cost=0.00..9.54 rows=1 width=8)\n\nEXPLAIN\n\nreal 0m0.659s\nuser 0m0.040s\nsys 0m0.030s\n\n\nThe execution plans are slightly different.\nIs it \"normal\"?\n\nAlso: My third PostgreSQL book is ready. It is in German - does anybody \nof those PostgreSQL hackers out there want a free issue?\n\n Hans\n\n-- \n*Cybertec Geschwinde u Schoenig*\nLudo-Hartmannplatz 1/14, A-1160 Vienna, Austria\nTel: +43/1/913 68 09; +43/664/233 90 75\nwww.postgresql.at <http://www.postgresql.at>, cluster.postgresql.at \n<http://cluster.postgresql.at>, www.cybertec.at \n<http://www.cybertec.at>, kernel.cybertec.at <http://kernel.cybertec.at>\n\n\n",
"msg_date": "Thu, 24 Oct 2002 10:18:57 +0200",
"msg_from": "=?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <postgres@cybertec.at>",
"msg_from_op": false,
"msg_subject": "Using the same condition twice"
},
{
"msg_contents": "=?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <postgres@cybertec.at> writes:\n> I came across a quite interesting issue I don't really understand but \n> maybe Tom will know.\n\nInteresting. We seem to recognize the fact that the extra clause is\nredundant in nearly all places ... but not in indexscan plan generation.\n\nI tried this simplified test case:\n\ncreate table t_wert(werttypid int);\ncreate table t_werttyp(id int);\ncreate index idx_wert_werttypid on t_wert(werttypid);\n\nexplain select * from\nt_wert JOIN t_werttyp ON (t_werttyp.id = t_wert.werttypid)\nwhere t_werttyp.id = t_wert.werttypid;\n\nexplain select * from\nt_wert JOIN t_werttyp ON (t_werttyp.id = t_wert.werttypid);\n\nI got identical merge-join plans and row count estimates both ways.\nI then turned off enable_mergejoin, and got identical hash-join plans\nand row counts. But with enable_hashjoin also off:\n\nregression=# explain select * from\nregression-# t_wert JOIN t_werttyp ON (t_werttyp.id = t_wert.werttypid)\nregression-# where t_werttyp.id = t_wert.werttypid;\n QUERY PLAN\n-------------------------------------------------------------------------------------------\n Nested Loop (cost=0.00..4858.02 rows=5000 width=8)\n -> Seq Scan on t_werttyp (cost=0.00..20.00 rows=1000 width=4)\n -> Index Scan using idx_wert_werttypid on t_wert (cost=0.00..4.83 rows=1 width=4)\n Index Cond: ((\"outer\".id = t_wert.werttypid) AND (\"outer\".id = t_wert.werttypid))\n(4 rows)\n\nregression=# explain select * from\nregression-# t_wert JOIN t_werttyp ON (t_werttyp.id = t_wert.werttypid);\n QUERY PLAN\n----------------------------------------------------------------------------------------\n Nested Loop (cost=0.00..17150.00 rows=5000 width=8)\n -> Seq Scan on t_werttyp (cost=0.00..20.00 rows=1000 width=4)\n -> Index Scan using idx_wert_werttypid on t_wert (cost=0.00..17.07 rows=5 width=4)\n Index Cond: (\"outer\".id = t_wert.werttypid)\n(4 rows)\n\nLooks like a bug is lurking someplace ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 24 Oct 2002 10:23:29 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Using the same condition twice "
},
{
"msg_contents": "I wrote:\n> Interesting. We seem to recognize the fact that the extra clause is\n> redundant in nearly all places ... but not in indexscan plan generation.\n\n> I tried this simplified test case:\n\n> create table t_wert(werttypid int);\n> create table t_werttyp(id int);\n> create index idx_wert_werttypid on t_wert(werttypid);\n\n> explain select * from\n> t_wert JOIN t_werttyp ON (t_werttyp.id = t_wert.werttypid)\n> where t_werttyp.id = t_wert.werttypid;\n\n> explain select * from\n> t_wert JOIN t_werttyp ON (t_werttyp.id = t_wert.werttypid);\n\n\nFYI, I have committed changes that seem to fix this problem in CVS tip.\n\nregression=# set enable_mergejoin to 0;\nSET\nregression=# set enable_hashjoin to 0;\nSET\nregression=# explain select * from\nregression-# t_wert JOIN t_werttyp ON (t_werttyp.id = t_wert.werttypid)\nregression-# where t_werttyp.id = t_wert.werttypid;\n QUERY PLAN\n----------------------------------------------------------------------------------------\n Nested Loop (cost=0.00..17150.00 rows=5000 width=8)\n -> Seq Scan on t_werttyp (cost=0.00..20.00 rows=1000 width=4)\n -> Index Scan using idx_wert_werttypid on t_wert (cost=0.00..17.07 rows=5 width=4)\n Index Cond: (\"outer\".id = t_wert.werttypid)\n(4 rows)\n\nregression=# explain select * from\nregression-# t_wert JOIN t_werttyp ON (t_werttyp.id = t_wert.werttypid);\n QUERY PLAN\n----------------------------------------------------------------------------------------\n Nested Loop (cost=0.00..17150.00 rows=5000 width=8)\n -> Seq Scan on t_werttyp (cost=0.00..20.00 rows=1000 width=4)\n -> Index Scan using idx_wert_werttypid on t_wert (cost=0.00..17.07 rows=5 width=4)\n Index Cond: (\"outer\".id = t_wert.werttypid)\n(4 rows)\n\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 24 Nov 2002 16:56:09 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Using the same condition twice "
}
] |
[
{
"msg_contents": "Hello hackers,\n\nI've been toying around with freelist in btrees, getting lots of\ndeadlocks and bootstrap problems. I think I've learned something about\nthe code, though. Just for the record: I already read Lehman and Yao's\npaper and I think I grok it.\n\nThere's few ideas I'd like to present and get some feedback about the\nimplementation, hoping that something that I miss may be catched by\nsomeone with more experience in these things.\n\n\nFirst is about freelist structure. I had originally proposed that the\nfreelist should keep an array of BlockNumbers in the metapage. Tom\nargued that the number that fit in there was too small, and proposed\nthat the first 2000 or so freepages be recorded in the metapage, and\nthe rest should have a link to the next freepage each.\n\nI propose instead an hybrid approach: the metapage has an array of\nBlockNumbers, _and_ a pointer to a \"next freelist page\". That page has\nanother array of BlockNumbers and another link to next freelist page.\nThis allows for easier \"compaction\" of the freelist, an operation which\nshould be done on a regular basis (with each VACUUM FULL, for example).\nThe list of freelist-pages should actually be double linked; that way,\nthe compaction process can take the BlockNumbers from the last page and\nput it on the first to fill it up, etc. (Remember that each newpage\noperation has to sequentially scan the freelist, and put a zero when it\ntakes one).\n\n\nThe second idea is about how to do the free page detection. Each time a\ntuple is deleted the page is checked to see if it's empty. If it is,\nit's added to a \"candidate-empty\" list. At the end of the\nbtbulkdelete() operation, we have a list of pages that are candidates\nfor adding into the freelist. For each one, the page is\nexclusive-locked, and rechecked if it's empty. If it is, the parent is\nalso exclusive-locked (beware of deadlock here!) and also its left and\nright siblings. In the parent, the item pointing to this page is\ndeleted; in the siblings, the side pointers are updated (high key\non the left sibling also?). Then this page is added to the freelist.\n\nFor the btree_xlog_delete() operation the recovery should be similar,\nbut the state should be checked with every operation, not with a\ncandidate-empty list.\n\nOn the \"add-to-free-list\" operation, the page is checked to see if it's\nthe last page of the relation. If it is, the page just before is\nchecked for emptyness (using the BTP_FREE flag) iteratively until a\nnonfree page is found. All those pages are deleted from the freelist.\nThen the relation is shrinked in that many pages.\n\n\nThird and last idea: how the _bt_getbuf() operation is modified. I\nthink the best way is to create a new _bt_newbuf() operation that grabs\none from the freelist, and use that whenever _bt_getbuf() is called with\nP_NEW as BlockNumber in the current code (current _bt_getbuf() will have\nand Assert(blkno != P_NEW). To prevent deadlocks, the newroot operation\ndoes not get a page from the freelist; it always extends the relation.\n\n\nComments? I think I've put too many things in one email. Sorry for\nthis.\n\n-- \nAlvaro Herrera (<alvherre[a]atentus.com>)\n\"The eagle never lost so much time as\nwhen he submitted to learn from the crow.\" (William Blake, citado por Nobody)\n",
"msg_date": "Tue, 22 Oct 2002 00:12:30 -0300",
"msg_from": "Alvaro Herrera <alvherre@dcc.uchile.cl>",
"msg_from_op": true,
"msg_subject": "BTree free pages again"
},
{
"msg_contents": "Dear hackers,\n\n\nI have recently been playing with DB2 8.1 Beta. It has introduced a \nfeature to enable index clustering on more than one key. This reminded \nme of a previous thread on HACKERS about index access anding/bitmaps in \nFirebird. So anyway, here is a little snip from the 8.1 manual as a FYI. \n\n-- snip\n\nAs the name implies, MDC tables cluster data on more than one dimension. \nEach dimension is determined by a column or set of columns that you \nspecify in the ORGANIZE BY DIMENSIONS clause of the CREATE TABLE \nstatement. When you create an MDC table, the following two kinds of \nindexes are created automatically:\n\n * A dimension-block index, which contains pointers to each occupied\n block for a single dimension.\n * A composite block index, which contains all dimension key columns.\n The composite block index is used to maintain clustering during\n insert and update activity.\n\nThe optimizer considers dimension-block index scan plans when it \ndetermines the most efficient access plan for a particular query. When \nqueries have predicates on dimension values, the optimizer can use the \ndimension block index to identify, and fetch from, only extents that \ncontain these values. In addition, because extents are physically \ncontiguous pages on disk, this results in more efficient performance and \nminimizes I/O.\n\n-- snipped\n\n\nregards\n\nMark\n\n\n",
"msg_date": "Tue, 22 Oct 2002 20:09:48 +1300",
"msg_from": "Mark Kirkwood <markir@paradise.net.nz>",
"msg_from_op": false,
"msg_subject": "Multiple Key Clustering In Db2 8.1 - Interesting FYI"
},
{
"msg_contents": "Alvaro,\n\nsome time ago I started to collect ideas for btree reorganization but\nnever got beyond brainstorming. Maybe it helps if I add them to your\nideas ...\n\nOn Tue, 22 Oct 2002 00:12:30 -0300, Alvaro Herrera\n<alvherre@dcc.uchile.cl> wrote:\n>I propose instead an hybrid approach: the metapage has an array of\n>BlockNumbers, _and_ a pointer to a \"next freelist page\". That page has\n>another array of BlockNumbers and another link to next freelist page.\n>This allows for easier \"compaction\" of the freelist, an operation which\n>should be done on a regular basis (with each VACUUM FULL, for example).\n>The list of freelist-pages should actually be double linked; that way,\n>the compaction process can take the BlockNumbers from the last page and\n>put it on the first to fill it up, etc.\n\n>(Remember that each newpage\n>operation has to sequentially scan the freelist, and put a zero when it\n>takes one).\n\nWhat do you mean by \"sequentially scan the freelist\"? Scan the array\nof page numbers on the meta page or scan the whole list of freelist\npages? I'd take a page number from the meta page. When the list of\nfree pages on the meta page is empty, read the first freelist page,\nmove its freelist to the metapage, let the meta page point to its\nsuccessor, and return the (formerly) first freelist page as the new\npage.\n\nPutting a page on the free list: If there is room for one more page\nnumber on the meta page, put the page number there and we are\nfinished. Otherwise move the freelist from the meta page to the new\npage and insert this page at the start of the freelist chain.\n\nTo avoid freelist ping-pong you might want to move only, say, 90% or\n95% of the freelist from the metapage to the new free page ...\n\nI don't know if we need to doubly link the list. When do we have to\nscan it backwards?\n\n\n>Each time a\n>tuple is deleted the page is checked to see if it's empty. If it is,\n>it's added to a \"candidate-empty\" list.\n\nAt this point the page is marked as \"dead\" (by setting a new flag in\nBTPageOpaqueData.btpo_flags) and the lock is released (not sure about\nreleasing or keeping the pin). A dead page is not yet available for\nreuse. It is always empty, so it is immediately skipped by index\nscans. The insertion code is modified to never put an index tuple\nonto a dead page (you can put it onto the right sibling without\nbreaking consistency).\n\n>At the end of the\n>btbulkdelete() operation, we have a list of pages that are candidates\n>for adding into the freelist.\n\nThere are several possible approaches:\n(a) Maintain a list of dead pages in memory\n(b) Rescan the whole index for dead pages\n(c) Handle each dead page immediately\n\nEach of these has its drawbacks: (a) leaves behind dead pages when the\nreorganization crashes, (b) might have to read lots of non-dead pages,\n(c) is inefficient if there are adjacent dead pages.\n\n>For each one, the page is\n>exclusive-locked, and rechecked if it's empty.\n\nThe \"dead\"-flag guarantees emptiness, just assert it's still dead.\n(Can VACUUMs run concurrently? If yes, this algorithm has to be\nadjusted.) Also no need to lock the page now.\n\n>If it is, the parent is\n>also exclusive-locked (beware of deadlock here!) and also its left and\n>right siblings. In the parent, the item pointing to this page is\n>deleted; in the siblings, the side pointers are updated (high key\n>on the left sibling also?).\n\nDo one page at a time: Exclusively lock the parent page, remove the\nindex tuple pointing to the dead page, unlock parent page. Lock one\nsibling, update pointer, unlock. Lock the other sibling, update\npointer, unlock.\n\nAttention! Maybe there's a problem lurking regarding recent splits of\nthe left sibling! Have to think more about this ...\n\nNow we have reached a state where the dead page will not be visited by\nany new index scan. There might still be danger from scans that were\npaused just before they were about to touch the dead page. Have to\nlook at the code whether pins overlap when a scan is sent down from a\nparent page or left/right from a sibling ...\n\n>Then this page is added to the freelist.\n>On the \"add-to-free-list\" operation, the page is checked to see if it's\n>the last page of the relation. If it is, the page just before is\n>checked for emptyness (using the BTP_FREE flag) iteratively until a\n>nonfree page is found.\n\nSpecial handling for freelist pages required here.\n\n>All those pages are deleted from the freelist.\n\nYou don't have to delete them explicitly. Just store the maximum free\npage number on the meta page and let the free page search ignore (and\nset to 0) all page numbers greater than this number. Making this work\nwith newroot always extending the relation needs some thought ...\n\n>Then the relation is shrinked in that many pages.\n>[...]\n>To prevent deadlocks, the newroot operation\n>does not get a page from the freelist; it always extends the relation.\n\n>I think I've put too many things in one email. Sorry for this.\n\nMe too :-)\n\nServus\n Manfred\n",
"msg_date": "Wed, 23 Oct 2002 16:19:45 +0200",
"msg_from": "Manfred Koizar <mkoi-pg@aon.at>",
"msg_from_op": false,
"msg_subject": "Re: BTree free pages again"
}
] |
[
{
"msg_contents": "Tom Lane wrote:\n > We can go with the auto-COMMIT idea for statements that are invoked at\n > the outer interactive level, but that doesn't work for stuff invoked\n > inside a function. I think we need to forbid these statements inside\n > functions, too. We already have that for VACUUM, because of its\n > internal commits causing problems for functions, but we'll need to\n > extend it to all of them.\n >\n > Just FYI, the statements involved are\n > VACUUM\n > TRUNCATE TABLE\n > CREATE/DROP DATABASE\n\nI just noticed that this afternoon's changes cause dblink's regression test to \nfail due to:\n\nCREATE OR REPLACE FUNCTION conditional_drop()\n[...]\n IF FOUND THEN\n DROP DATABASE regression_slave;\n END IF;\n[...]\n' LANGUAGE 'plpgsql';\nSELECT conditional_drop();\n\n\nI'm wondering what is the best alternative? Should we simply do a \"DROP \nDATABASE regression_slave;\" and have the expected results include the \n'database \"regression_slave\" does not exist'? In this case there would be an \nexpected failure whenever the regression test was run more than once.\n\nJoe\n\n",
"msg_date": "Mon, 21 Oct 2002 22:25:42 -0700",
"msg_from": "Joe Conway <mail@joeconway.com>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Joe Conway <mail@joeconway.com> writes:\n> I just noticed that this afternoon's changes cause dblink's regression\n> test to fail due to:\n\n> CREATE OR REPLACE FUNCTION conditional_drop()\n> [...]\n> IF FOUND THEN\n> DROP DATABASE regression_slave;\n> END IF;\n> [...]\n> ' LANGUAGE 'plpgsql';\n> SELECT conditional_drop();\n\n> I'm wondering what is the best alternative?\n\nWell, the *best* alternative would be to make CREATE/DROP DATABASE\ntransaction-safe ;-). I was speculating to myself earlier today about\nhow we might do that. It seems like it's not that far out of reach:\nwe could make smgr's list of files-to-remove-at-xact-commit-or-abort\ninclude whole database subdirectories. But I'm not sure how that would\ninteract with upcoming features like tablespaces, so I don't want to\ngo off and implement it right now.\n\nIn the meantime, to tell you the truth, the cleanest way to handle the\ndblink regression test would be to make it circularly connect to\ndatabase \"regression\". I know this seems cheesy, but as long as the\nsoftware under test doesn't know that it's a connection-to-self, seems\nlike the test is perfectly good. And it's surely easier to manage that\nway.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 22 Oct 2002 01:38:01 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al "
},
{
"msg_contents": "Tom Lane wrote:\n> In the meantime, to tell you the truth, the cleanest way to handle the\n> dblink regression test would be to make it circularly connect to\n> database \"regression\". I know this seems cheesy, but as long as the\n> software under test doesn't know that it's a connection-to-self, seems\n> like the test is perfectly good. And it's surely easier to manage that\n> way.\n\nOK, easy enough. Patch attached.\n\nI also added \"SET autocommit TO ''on'';\" to the beginning of each dblink_exec \ninput statement because the \"SET autocommit TO 'on';\" at the top of the script \nwon't help for the connected database.\n\nJoe",
"msg_date": "Mon, 21 Oct 2002 23:02:48 -0700",
"msg_from": "Joe Conway <mail@joeconway.com>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
},
{
"msg_contents": "Tom Lane wrote:\n> Joe Conway <mail@joeconway.com> writes:\n> > I just noticed that this afternoon's changes cause dblink's regression\n> > test to fail due to:\n> \n> > CREATE OR REPLACE FUNCTION conditional_drop()\n> > [...]\n> > IF FOUND THEN\n> > DROP DATABASE regression_slave;\n> > END IF;\n> > [...]\n> > ' LANGUAGE 'plpgsql';\n> > SELECT conditional_drop();\n> \n> > I'm wondering what is the best alternative?\n> \n> Well, the *best* alternative would be to make CREATE/DROP DATABASE\n> transaction-safe ;-). I was speculating to myself earlier today about\n> how we might do that. It seems like it's not that far out of reach:\n> we could make smgr's list of files-to-remove-at-xact-commit-or-abort\n> include whole database subdirectories. But I'm not sure how that would\n> interact with upcoming features like tablespaces, so I don't want to\n> go off and implement it right now.\n\nFYI, the MSWin port in 7.4 will have C versions of 'cp' and 'rm -r', so\nthose can be used to hook into the smgr layer for all platforms.\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, 22 Oct 2002 12:50:56 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: autocommit vs TRUNCATE et al"
}
] |
[
{
"msg_contents": "%gmake\n....\nbison -y -d preproc.y\npreproc.y:5560: fatal error: maximum table size (32767) exceeded\ngmake[4]: *** [preproc.h] Error 1\ngmake[4]: Leaving directory `/spool/home/teodor/pgsql/src/interfaces/ecpg/preproc'\ngmake[3]: *** [all] Error 2\ngmake[3]: Leaving directory `/spool/home/teodor/pgsql/src/interfaces/ecpg'\ngmake[2]: *** [all] Error 2\ngmake[2]: Leaving directory `/spool/home/teodor/pgsql/src/interfaces'\ngmake[1]: *** [all] Error 2\ngmake[1]: Leaving directory `/spool/home/teodor/pgsql/src'\ngmake: *** [all] Error 2\n% bison -V\nbison (GNU Bison) 1.35\n\nCopyright 1984, 1986, 1989, 1992, 2000, 2001, 2002\nFree 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% uname -a\nFreeBSD xor 4.6-STABLE FreeBSD 4.6-STABLE #2: Tue Jun 18 20:48:48 MSD 2002 \nteodor@xor:/usr/src/sys/compile/XOR i386\n\n-- \nTeodor Sigaev\nteodor@stack.net\n\n\n",
"msg_date": "Tue, 22 Oct 2002 11:57:43 +0400",
"msg_from": "Teodor Sigaev <teodor@stack.net>",
"msg_from_op": true,
"msg_subject": "Current CVS is broken"
},
{
"msg_contents": "Teodor Sigaev <teodor@stack.net> wrote:\n\n> %gmake\n> ....\n> bison -y -d preproc.y\n> preproc.y:5560: fatal error: maximum table size (32767) exceeded\n> gmake[4]: *** [preproc.h] Error 1\n> gmake[4]: Leaving directory\n`/spool/home/teodor/pgsql/src/interfaces/ecpg/preproc'\n> gmake[3]: *** [all] Error 2\n> gmake[3]: Leaving directory `/spool/home/teodor/pgsql/src/interfaces/ecpg'\n> gmake[2]: *** [all] Error 2\n> gmake[2]: Leaving directory `/spool/home/teodor/pgsql/src/interfaces'\n> gmake[1]: *** [all] Error 2\n> gmake[1]: Leaving directory `/spool/home/teodor/pgsql/src'\n> gmake: *** [all] Error 2\n> % bison -V\n> bison (GNU Bison) 1.35\n\nAlthough I am not a hacker, I think you just need to upgrade\nbison to version 1.75 if you want to build from CVS. The ecpg\ninterface is broken with version 1.35 of bison.\n\nBest Regards,\nMichael Paesold\n\n",
"msg_date": "Tue, 22 Oct 2002 10:21:08 +0200",
"msg_from": "\"Michael Paesold\" <mpaesold@gmx.at>",
"msg_from_op": false,
"msg_subject": "Re: Current CVS is broken"
},
{
"msg_contents": "On Tue, Oct 22, 2002 at 10:21:08AM +0200, Michael Paesold wrote:\n> Although I am not a hacker, I think you just need to upgrade\n> bison to version 1.75 if you want to build from CVS. The ecpg\n\n1.50 would do as well. \n\n> interface is broken with version 1.35 of bison.\n\nLet's say ecpg grammar broke bison 1.35. :-)\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, 22 Oct 2002 10:23:20 +0200",
"msg_from": "Michael Meskes <meskes@postgresql.org>",
"msg_from_op": false,
"msg_subject": "Re: Current CVS is broken"
},
{
"msg_contents": "install bison 1.75\n\n\nOn Tue, 22 Oct 2002, Teodor Sigaev wrote:\n\n> %gmake\n> ....\n> bison -y -d preproc.y\n> preproc.y:5560: fatal error: maximum table size (32767) exceeded\n> gmake[4]: *** [preproc.h] Error 1\n> gmake[4]: Leaving directory `/spool/home/teodor/pgsql/src/interfaces/ecpg/preproc'\n> gmake[3]: *** [all] Error 2\n> gmake[3]: Leaving directory `/spool/home/teodor/pgsql/src/interfaces/ecpg'\n> gmake[2]: *** [all] Error 2\n> gmake[2]: Leaving directory `/spool/home/teodor/pgsql/src/interfaces'\n> gmake[1]: *** [all] Error 2\n> gmake[1]: Leaving directory `/spool/home/teodor/pgsql/src'\n> gmake: *** [all] Error 2\n> % bison -V\n> bison (GNU Bison) 1.35\n>\n> Copyright 1984, 1986, 1989, 1992, 2000, 2001, 2002\n> 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> % uname -a\n> FreeBSD xor 4.6-STABLE FreeBSD 4.6-STABLE #2: Tue Jun 18 20:48:48 MSD 2002\n> teodor@xor:/usr/src/sys/compile/XOR i386\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": "Tue, 22 Oct 2002 13:50:21 +0400 (MSD)",
"msg_from": "Oleg Bartunov <oleg@sai.msu.su>",
"msg_from_op": false,
"msg_subject": "Re: Current CVS is broken"
},
{
"msg_contents": "Oleg Bartunov <oleg@sai.msu.su> writes:\n> install bison 1.75\n\nI've fixed the INSTALL and installation.sgml docs to say you need bison\n1.50 or later (it used to say 1.28 or later). Is there anyplace else we\nshould point this out?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 22 Oct 2002 09:48:58 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Current CVS is broken "
}
] |
[
{
"msg_contents": "> edit *pg_hba.conf *\n> # Allow any user on the local system to connect to any\n> # database under any username, but only via an IP connection:\n> host all 127.0.0.1 255.255.255.255 trust \n> # The same, over Unix-socket connections:\n> local all trust\nwhat about reading pg_hba.conf comments?\n local all md5\n\n> or *my nightmare *a cygwin on Win 98 everybody can can access everything \n> :-((((\n/me shrugs\ni don't use cygwin\n\n\n\n",
"msg_date": "Tue, 22 Oct 2002 18:15:29 +0400",
"msg_from": "dima <_pppp@mail.ru>",
"msg_from_op": true,
"msg_subject": "Re: Security question : Database access control"
},
{
"msg_contents": "On Tue, Oct 22, 2002 at 17:05:38 +0200,\n Igor Georgiev <gory@alphasoft-bg.com> wrote:\n> Is there any way to prevent superuser to acces the database ?\n> I mean something like \"GRANT / REVOKE CONNECT\" MECHANISM\n> \n> I have no idea how to prevent root from access data in one of this ways :\n> root @ linux:~#su - postgres\n> postgres @ linux:/usr/local/pgsql/bin$pg_dump ....\n> or\n> edit pg_hba.conf \n> # Allow any user on the local system to connect to any\n> # database under any username, but only via an IP connection:\n> host all 127.0.0.1 255.255.255.255 trust \n> # The same, over Unix-socket connections:\n> local all trust\n> or my nightmare a cygwin on Win 98 everybody can can access everything :-((((\n\nThey can just read the raw database files as well. You have to be able to\ntrust whoever has root access to the system, as well as anyone who has\nphysical access to the system.\n",
"msg_date": "Tue, 22 Oct 2002 09:27:09 -0500",
"msg_from": "Bruno Wolff III <bruno@wolff.to>",
"msg_from_op": false,
"msg_subject": "Re: Security question : Database access control"
},
{
"msg_contents": "Is there any way to prevent superuser to acces the database ?\nI mean something like \"GRANT / REVOKE CONNECT\" MECHANISM\n\nI have no idea how to prevent root from access data in one of this ways :\n root @ linux:~#su - postgres\n postgres @ linux:/usr/local/pgsql/bin$pg_dump ....\nor\n edit pg_hba.conf \n # Allow any user on the local system to connect to any\n # database under any username, but only via an IP connection:\n host all 127.0.0.1 255.255.255.255 trust \n # The same, over Unix-socket connections:\n local all trust\nor my nightmare a cygwin on Win 98 everybody can can access everything :-((((\n\n\n\n\n\n\n\n\n\nIs there any way to prevent superuser to acces the \ndatabase ?\nI mean something like \"GRANT / \nREVOKE CONNECT\" MECHANISM\n \nI have no idea how to prevent root from access data \nin one of this ways :\n root @ linux:~#su \n- postgres\n postgres @ \nlinux:/usr/local/pgsql/bin$pg_dump ....\nor\n edit pg_hba.conf \n\n # Allow any \nuser on the local system to connect to any \n # database under any username, but only via an IP \nconnection: \nhost \nall \n127.0.0.1 255.255.255.255 \ntrust \n# The same, over Unix-socket connections: \n local \nall \ntrustor my nightmare a cygwin on Win 98 everybody can can \naccess everything :-((((",
"msg_date": "Tue, 22 Oct 2002 17:05:38 +0200",
"msg_from": "\"Igor Georgiev\" <gory@alphasoft-bg.com>",
"msg_from_op": false,
"msg_subject": "Security question : Database access control"
},
{
"msg_contents": "On Tue, 22 Oct 2002, Igor Georgiev wrote:\n\n> > > edit *pg_hba.conf *\n> > > # Allow any user on the local system to connect to any\n> > > # database under any username, but only via an IP connection:\n> > > host all 127.0.0.1 255.255.255.255 trust \n> > > # The same, over Unix-socket connections:\n> > > local all trust\n> > what about reading pg_hba.conf comments?\n> > local all md5\n> > \n> \n> Ok, but my question actually isn't about pg_hba.conf comments, i read enough\n> but what will stop root from adding this lines or doing su - postgres ??\n> \n\nNothing, root is GOD in unix. He can do whatever he wants.\n\n",
"msg_date": "Tue, 22 Oct 2002 09:13:43 -0600 (MDT)",
"msg_from": "\"scott.marlowe\" <scott.marlowe@ihs.com>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Security question : Database access control"
},
{
"msg_contents": "On Tue, 2002-10-22 at 12:12, Igor Georgiev wrote:\n> > > edit *pg_hba.conf *\n> > > # Allow any user on the local system to connect to any\n> > > # database under any username, but only via an IP connection:\n> > > host all 127.0.0.1 255.255.255.255 trust \n> > > # The same, over Unix-socket connections:\n> > > local all trust\n> > what about reading pg_hba.conf comments?\n> > local all md5\n> > \n> \n> Ok, but my question actually isn't about pg_hba.conf comments, i read enough\n> but what will stop root from adding this lines or doing su - postgres ??\n\nNext your going to ask what will stop root from stopping your\nPostgreSQL, compiling a second copy with authentication disabled and\nusing your data directory as it's source :)\n\nIf you want to prevent root from accomplishing these things, you're\ngoing to have to look to your kernel for help. The kernel must prevent\nroot from changing users, starting / stopping applications, or touching\ncertain filesystems.\n\n\nPostgreSQL will let you put a password on the data. But that only works\nif they actually try to use PostgreSQL to get at the data.\n\nThere are a couple of tools which were designed to recover database data\nwhile the db is not running.\n\n-- \n Rod Taylor\n\n",
"msg_date": "22 Oct 2002 11:26:30 -0400",
"msg_from": "Rod Taylor <rbt@rbt.ca>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Security question : Database access control"
},
{
"msg_contents": "\nOn Tue, 22 Oct 2002, Igor Georgiev wrote:\n\n> > > edit *pg_hba.conf *\n> > > # Allow any user on the local system to connect to any\n> > > # database under any username, but only via an IP connection:\n> > > host all 127.0.0.1 255.255.255.255 trust\n> > > # The same, over Unix-socket connections:\n> > > local all trust\n> > what about reading pg_hba.conf comments?\n> > local all md5\n> >\n>\n> Ok, but my question actually isn't about pg_hba.conf comments, i read enough\n> but what will stop root from adding this lines or doing su - postgres ??\n\nNot much really. But given that they have access to the raw data\nfiles, preventing them access to the server doesn't gain you that\nmuch if they really want to get the data.\n\n",
"msg_date": "Tue, 22 Oct 2002 08:28:08 -0700 (PDT)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: Security question : Database access control"
},
{
"msg_contents": "\"Igor Georgiev\" <gory@alphasoft-bg.com> writes:\n> Ok, but my question actually isn't about pg_hba.conf comments, i read enough\n> but what will stop root from adding this lines or doing su - postgres ??\n\nAs somebody already pointed out, you *must* trust the people with root\naccess to your machine; there is not anything you can do to defend\nyourself against them. If you can't trust the admins on the machine\nyou're using, better get your own machine.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 22 Oct 2002 11:35:25 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Security question : Database access control "
},
{
"msg_contents": "> > edit *pg_hba.conf *\n> > # Allow any user on the local system to connect to any\n> > # database under any username, but only via an IP connection:\n> > host all 127.0.0.1 255.255.255.255 trust \n> > # The same, over Unix-socket connections:\n> > local all trust\n> what about reading pg_hba.conf comments?\n> local all md5\n> \n\nOk, but my question actually isn't about pg_hba.conf comments, i read enough\nbut what will stop root from adding this lines or doing su - postgres ??\n\n\n\n\n\n\n\n> \n> edit *pg_hba.conf *> \n> # Allow any user on the \nlocal system to connect to any> \n> # database under any \nusername, but only via an IP connection:> \n> \nhost \nall \n127.0.0.1 255.255.255.255 \ntrust > \n> # The same, over \nUnix-socket connections:> \n> \nlocal \nall \ntrust\n> what about reading pg_hba.conf \ncomments?> \nlocal \nall \nmd5> \nOk, but my question actually isn't about pg_hba.conf comments, i read \nenough\nbut what will stop root from adding this lines or \ndoing su - postgres ??",
"msg_date": "Tue, 22 Oct 2002 18:12:16 +0200",
"msg_from": "\"Igor Georgiev\" <gory@alphasoft-bg.com>",
"msg_from_op": false,
"msg_subject": "Re: Security question : Database access control"
},
{
"msg_contents": "> They can just read the raw database files as well. \nwow I'm not sure\nhow about this\n\n edit pg_hba.conf \n # Allow any user on the local system to connect to any\n # database under any username\n local all trust\n\nsu - posgres\npsql test -U dba\nor \npg_dump test\n\n> You have to be able to trust whoever has root access to the system, as well as anyone who has\n> physical access to the system.\n\n\n\n\n\n\n\n\n> They can just read the raw database files as \nwell. \nwow I'm not sure\nhow about this\n \n\n edit pg_hba.conf \n\n # Allow any user on the local system to connect to \nany # database under any \nusername\n \nlocal \nall \ntrust\n \nsu - posgres\npsql test -U dba\nor \npg_dump test\n \n> You have to be able to trust whoever has root \naccess to the system, as well as anyone who has> physical access to the \nsystem.",
"msg_date": "Tue, 22 Oct 2002 18:34:57 +0200",
"msg_from": "\"Igor Georgiev\" <gory@alphasoft-bg.com>",
"msg_from_op": false,
"msg_subject": "Re: Security question : Database access control"
},
{
"msg_contents": "> Next your going to ask what will stop root from stopping your\n> PostgreSQL, compiling a second copy with authentication disabled and\n> using your data directory as it's source :)\n\nHe he i'm enough paranoic :))\n\n> If you want to prevent root from accomplishing these things, you're\n> going to have to look to your kernel for help. The kernel must prevent\n> root from changing users, starting / stopping applications, or touching\n> certain filesystems.\n> \n> PostgreSQL will let you put a password on the data. But that only works\n> if they actually try to use PostgreSQL to get at the data.\n\nuse PostgreSQL to get at the data - Yeah this will be enough \ni want just only REVOKE CONNECT PRIVILEGES on database\n\n\n\n\n\n\n\n\n> Next your going to ask what will stop root \nfrom stopping your> PostgreSQL, compiling a second copy with \nauthentication disabled and> using your data directory as it's source \n:)\n \nHe he i'm enough paranoic :))\n> If you want to prevent root from \naccomplishing these things, you're> going to have to look to your kernel \nfor help. The kernel must prevent> root from changing users, \nstarting / stopping applications, or touching> certain \nfilesystems.> > PostgreSQL will let you put a password on the \ndata. But that only works> if they actually try to use \nPostgreSQL to get at the data.\n \nuse PostgreSQL to get at the data \n- Yeah this will be enough \ni want just only REVOKE CONNECT PRIVILEGES on \ndatabase",
"msg_date": "Tue, 22 Oct 2002 18:45:07 +0200",
"msg_from": "\"Igor Georgiev\" <gory@alphasoft-bg.com>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Security question : Database access control"
},
{
"msg_contents": "On Tue, Oct 22, 2002 at 05:05:38PM +0200, Igor Georgiev wrote:\n> or my nightmare a cygwin on Win 98 everybody can can access everything\n\nOr =my= nightmare: Anything important on any Windows platform.\n\n -crl\n--\nChad R. Larson (CRL22) chad@eldocomp.com\n Eldorado Computing, Inc. 602-604-3100\n 5353 North 16th Street, Suite 400\n Phoenix, Arizona 85016-3228\n",
"msg_date": "Tue, 22 Oct 2002 21:21:23 -0700",
"msg_from": "\"Chad R. Larson\" <chad@eldocomp.com>",
"msg_from_op": false,
"msg_subject": "Re: Security question : Database access control"
},
{
"msg_contents": "\n> Next your going to ask what will stop root from stopping your\n> PostgreSQL, compiling a second copy with authentication disabled and\n> using your data directory as it's source :)\n\nHe he somebody can blow up ur home with C4, but this don't stop you from\nlocking ur door !?!\n\n",
"msg_date": "Fri, 25 Oct 2002 12:57:09 +0200",
"msg_from": "\"Igor Georgiev\" <gory@alphasoft-bg.com>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Security question : Database access control"
},
{
"msg_contents": "On Fri, 25 Oct 2002, Igor Georgiev wrote:\n\n>\n> > Next your going to ask what will stop root from stopping your\n> > PostgreSQL, compiling a second copy with authentication disabled and\n> > using your data directory as it's source :)\n>\n> He he somebody can blow up ur home with C4, but this don't stop you from\n> locking ur door !?!\n\nIt's more equivalent to giving him the keys and then wondering why the\nlock doesn't stop him.\n\n\n",
"msg_date": "Fri, 25 Oct 2002 05:57:22 -0700 (PDT)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Security question : Database access control"
}
] |
[
{
"msg_contents": "\n> What I just committed uses your idea of auto-committing TRUNCATE et al,\n> but now that I review the thread I think that everyone else thought that\n> that was a dangerous idea. How do you feel about simply throwing an error\n> in autocommit-off mode, instead? (At least it's a localized \n> change now)\n\nWell, if I can throw in another opinion, I think what you did is perfect. \nIt will make Oracle users happy too. Only very shrewd applications would\ncommit previous changes with a \"truncate\" statement, and those will learn \nto issue a commit before truncate. I don't like the solutions involving \n\"set autocommit ...\".\n\nAndreas\n",
"msg_date": "Tue, 22 Oct 2002 16:35:46 +0200",
"msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>",
"msg_from_op": true,
"msg_subject": "Re: autocommit vs TRUNCATE et al "
}
] |
[
{
"msg_contents": "Hi there all PostgreSQL Hackers.\n\nWe have just upgraded from 7.1.3 to 7.2.3 over the weekend, and have just\nnoticed something weird with regards 'epoch'::timestamp.\n\nIn 7.1.3, 'epoch'::timestamp always returned the value 1970-01-01\n00:00:00+00, following the upgrade we now get:-\n\n timestamptz \n------------------------\n 1970-01-01 01:00:00+01\n\nAlso, any fields set to 'epoch'::timestamp before the upgrade now return:-\n\n e_app_xfer_date \n------------------------\n 1970-01-01 00:00:00+01\n\nIf we issue a SET TIMEZONE TO 'GMT' the results are the same as for the\nprevious release.\n\nThis is causing us a problem as we use epoch to indicate whether the data\nhas been processed, and to log the date and time of processing.\n\nWe are running on RedHat 7.1 with kernel 2.4.9-21smp, and glibc\nglibc-2.2.4-19.3 with PostgreSQL compiled from the source tarfile.\n\nOutput from version(): PostgreSQL 7.2.3 on i686-pc-linux-gnu, compiled by\nGCC 2.96\n\nWith regards PostgreSQL in general, we have been using it since version\n6.4.x and have found it (even the earier versions) extremely reliable an\npowerful. Keep up the good work guys!\n\nJohn Hosen\nSenior Network Analyst\nCapita RAS\nhttp://www.capitaras.co.uk/\n\n\n***********************************************************************\n\nThis email and any files attached to it are confidential and intended \nsolely for the use of the individual or entity to whom they are\naddressed. If you have received this email in error please notify \nthe system manager.\n\nThe message and any files attached to it have been scanned by \nMIMEsweeper with Sophos Sweep and found to be free from all known\nviruses.\n\nInformation on MIMEsweeper can be found at http://www.mimesweeper.com/\n\n***********************************************************************\n\n",
"msg_date": "Tue, 22 Oct 2002 17:11:35 +0100",
"msg_from": "\"Hosen, John\" <John.Hosen@capita.co.uk>",
"msg_from_op": true,
"msg_subject": "'epoch'::timestamp and Daylight Savings"
},
{
"msg_contents": "> We have just upgraded from 7.1.3 to 7.2.3 over the weekend, and have just\n> noticed something weird with regards 'epoch'::timestamp.\n> In 7.1.3, 'epoch'::timestamp always returned the value 1970-01-01\n> 00:00:00+00,...\n\nI would expect 'epoch'::timestamp to always act as though the value were \nthe same as Unix time zero. But it did not explicitly return that value:\n\nlockhart=# select version();\n-------------------------------------------------------------\n PostgreSQL 7.1.2 on i686-pc-linux-gnu, compiled by GCC 2.96\n\nlockhart=# select 'epoch'::timestamp;\n----------\n epoch\n\nlockhart=# select 'epoch'::timestamp + '0 sec'::interval;\n------------------------\n 1969-12-31 16:00:00-08\n\n\n> following the upgrade we now get:-\n> timestamptz \n> ------------------------\n> 1970-01-01 01:00:00+01\n\nWhich is consistant with the previous result in absolute time.\n\nYou will find that the most recent versions of PostgreSQL convert \n'epoch' on input, and that with- and without time zone data types are \navailable:\n\nlockhart=# select cast('epoch' as timestamp with time zone);\n timestamptz\n------------------------\n 1969-12-31 16:00:00-08\n\nlockhart=# select cast('epoch' as timestamp without time zone);\n timestamp\n---------------------\n 1970-01-01 00:00:00\n\n\n> Also, any fields set to 'epoch'::timestamp before the upgrade now return:-\n> e_app_xfer_date \n> ------------------------\n> 1970-01-01 00:00:00+01\n> If we issue a SET TIMEZONE TO 'GMT' the results are the same as for the\n> previous release.\n\nNot sure about this one. What is the schema? Can you give an example \nwhere the time gets shifted by an hour?\n\n> This is causing us a problem as we use epoch to indicate whether the data\n> has been processed, and to log the date and time of processing.\n\nUsing \"special values\" to indicate status can be troublesome, as you are \nfinding. I'd suggest using NULL to indicate that a field is not known or \nnot yet set.\n\n> This email and any files attached to it are confidential and intended \n> solely for the use of the individual or entity to whom they are\n> addressed. If you have received this email in error please notify \n> the system manager.\n\nOops. Can you please confirm that we were the intended recipients? ;)\n\nhth\n\n - Thomas\n\n",
"msg_date": "Wed, 23 Oct 2002 15:06:09 +0000",
"msg_from": "Thomas Lockhart <lockhart@fourpalms.org>",
"msg_from_op": false,
"msg_subject": "Re: 'epoch'::timestamp and Daylight Savings"
}
] |
[
{
"msg_contents": "I got tired of wondering what timing dependencies might still be lurking\nin the parallel regression tests, and wrote a little Perl script to find\nout by making variant versions of the parallel_schedule file.\n\nEssentially, the script forces each test in a parallel group to run\nbefore all the other ones in its group, and also after all the other\nones. This isn't a completely bulletproof check: you could imagine\nthat test A might be unhappy about some intermediate state created by\ntest B, while not being unhappy with either the starting or ending\nstates. But it's a lot better than guess-and-hope.\n\nThe script successfully exposed the problem reported earlier today by\nRobert Hentosh (create_index test depends on create_misc having run),\nand didn't find any other problems, which I guess is a good sign.\n(I've now committed a fix for that mistake, BTW.)\n\nI'm not sure if the script has any long-term usefulness (anyone feel it\ndeserves to get into CVS in src/tools/?). But I'll attach it anyway\njust so it gets into the pghackers archives.\n\nTo use it you'd do something like\n\nmkdir scheds\n./sched_variants parallel_schedule scheds/sch\nfor f in scheds/sch*\ndo\n\techo $f\n\t/bin/sh ./pg_regress --schedule=$f\n\tff=`basename $f`\n\tmv regression.out scheds/regression.out.$ff\n\tmv regression.diffs scheds/regression.diffs.$ff\ndone\n\n\n\t\t\tregards, tom lane\n\n\n#! /usr/bin/perl\n\n# Generate variants of parallel_schedule file to verify that there are\n# no order dependencies between tests executed in parallel.\n\n# Usage: sched_variants input_file output_prefix\n\n# Output files are named with 'output_prefix' suffixed .1, .2, etc.\n\ndie \"Usage: sched_variants input_file output_prefix\\n\" if ($#ARGV != 1);\n\n$infile = $ARGV[0];\n$outprefix = $ARGV[1];\n\n$outcount = 0;\t\t\t# number of output files created\n\n# We scan the input file repeatedly. On each pass we generate two\n# output files, one where the k'th entry of each parallel test set\n# has been extracted and forced to run first, and one where it's been\n# forced to run last. The number of passes needed is the same as the\n# largest number of tests in a parallel test set.\n\n$k = 1;\t\t\t\t# test index we are currently hacking\n\n$more = 1;\t\t\t# true if we need another pass\n\nwhile ($more) {\n $more = 0;\t\t\t# until proven differently\n\n open(INFILE, $infile) || die \"$infile: $!\\n\";\n\n $outcount++;\n $outbefore = $outprefix . \".\" . $outcount;\n open(OUTBEFORE, \"> $outbefore\") || die \"$outbefore: $!\\n\";\n\n $outcount++;\n $outafter = $outprefix . \".\" . $outcount;\n open(OUTAFTER, \"> $outafter\") || die \"$outafter: $!\\n\";\n\n while (<INFILE>) {\n\tif (! /^test:/) {\n\t # comment line\n\t print OUTBEFORE $_;\n\t print OUTAFTER $_;\n\t next;\n\t}\n\t@tests = split;\n\tshift(@tests);\t\t# remove test:\n\tif ($#tests < $k-1 || $#tests == 0) {\n\t # too few tests in this set, just repeat as-is\n\t print OUTBEFORE $_;\n\t print OUTAFTER $_;\n\t next;\n\t}\n\tif ($#tests >= $k) {\n\t $more = 1;\t\t# need more passes to process this set\n\t}\n\t@thistest = splice(@tests, $k-1, 1);\n\tprint OUTBEFORE \"test: @thistest\\n\";\n\tprint OUTBEFORE \"test: @tests\\n\";\n\n\tprint OUTAFTER \"test: @tests\\n\";\n\tprint OUTAFTER \"test: @thistest\\n\";\n }\n\n close OUTBEFORE;\n close OUTAFTER;\n close INFILE;\n\n $k++;\n}\n\nprint \"$outcount test files generated.\\n\";\n\nexit 0;",
"msg_date": "Tue, 22 Oct 2002 16:31:59 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Silly little tool for making parallel_schedule variants"
}
] |
[
{
"msg_contents": "On Wed, 2002-10-23 at 03:09, Tom Lane wrote:\n> It's fairly difficult to get anywhere with standard leak-tracking tools,\n> since they don't know anything about palloc. What's worse, it is *not*\n> a bug for a routine to palloc space it never pfrees, if it knows that\n> it's palloc'ing in a short-lived memory context. The fact that a\n> context may be released with much still-allocated memory in it is not a\n> bug but a feature; but try teaching that to any standard leak checker...\n\nSeems that Valgrind should have no problems with it, as it tracks actual\nusage of _memory_ (down to single bits :)) , not malloc/free. \n\nSee: http://developer.kde.org/~sewardj/\n\n---------------\nHannu\n\n\n",
"msg_date": "23 Oct 2002 01:36:41 +0500",
"msg_from": "Hannu Krosing <hannu@tm.ee>",
"msg_from_op": true,
"msg_subject": "Re: Memory leaks"
},
{
"msg_contents": "On Tue, Oct 22, 2002 at 04:16:16PM -0500, Greg Copeland wrote:\n> I've started playing a little with Postgres to determine if there were\n> memory leaks running around. After some very brief checking, I'm\n> starting[1] to think that the answer is yes. Has anyone already gone\n> through a significant effort to locate and eradicate memory leaks? Is\n> this done periodically? If so, what tools are others using? I'm\n> currently using dmalloc for my curiosity.\n\nvalgrind is a great tool I used -- didn't get the time to try it out on\nPostgres yet, though. Besides leaks, it also catches uninitialized\nvariable access and stuff like that.\n\n\nPetru\n",
"msg_date": "Tue, 22 Oct 2002 23:43:24 +0300",
"msg_from": "Petru Paler <petru@paler.net>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks"
},
{
"msg_contents": "I've started playing a little with Postgres to determine if there were\nmemory leaks running around. After some very brief checking, I'm\nstarting[1] to think that the answer is yes. Has anyone already gone\nthrough a significant effort to locate and eradicate memory leaks? Is\nthis done periodically? If so, what tools are others using? I'm\ncurrently using dmalloc for my curiosity.\n\n[1] Not sure yet as I'm really wanting to find culumative leaks rather\nthan one shot allocations which are simply never freed prior to process\ntermination.\n\n\nRegards,\n\n\tGreg Copeland\n\n",
"msg_date": "22 Oct 2002 16:16:16 -0500",
"msg_from": "Greg Copeland <greg@CopelandConsulting.Net>",
"msg_from_op": false,
"msg_subject": "Memory leaks"
},
{
"msg_contents": "Greg Copeland <greg@CopelandConsulting.Net> writes:\n> I've started playing a little with Postgres to determine if there were\n> memory leaks running around. After some very brief checking, I'm\n> starting[1] to think that the answer is yes. Has anyone already gone\n> through a significant effort to locate and eradicate memory leaks?\n\nYes, this has been dealt with before. Have you read\nsrc/backend/utils/mmgr/README?\n\nAFAIK the major problems these days are not in the backend as a whole,\nbut in the lesser-used PL language modules (cf recent discussions).\nplpgsql has some issues too, I suspect, but not as bad as pltcl etc.\nPossibly the best answer is to integrate the memory-context notion into\nthose modules; if they did most of their work in a temp context that\ncould be freed once per PL statement or so, the problems would pretty\nmuch go away.\n\nIt's fairly difficult to get anywhere with standard leak-tracking tools,\nsince they don't know anything about palloc. What's worse, it is *not*\na bug for a routine to palloc space it never pfrees, if it knows that\nit's palloc'ing in a short-lived memory context. The fact that a\ncontext may be released with much still-allocated memory in it is not a\nbug but a feature; but try teaching that to any standard leak checker...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 22 Oct 2002 18:09:23 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks "
},
{
"msg_contents": "Petru Paler <petru@paler.net> writes:\n> valgrind is a great tool I used -- didn't get the time to try it out on\n> Postgres yet, though. Besides leaks, it also catches uninitialized\n> variable access and stuff like that.\n\nI've used Valgrind with PostgreSQL a little bit, and it's been fairly\nuseful (I used it to fix some memory leaks in psql and pg_dump and a\ncouple of uninitialized memory accesses in the backend).\n\nIf you want to use it on the backend, you'll need to stop postgres\nfrom clobbering ARGV (as this causes valgrind problems, for some\nreason) -- add '-DPS_USE_NONE -UPS_USE_CLOBBER_ARGV' to CFLAGS. I\nmentioned it to the author of valgrind, but IIRC he didn't mention\nany plans to change this behavior.\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n",
"msg_date": "22 Oct 2002 19:27:20 -0400",
"msg_from": "Neil Conway <neilc@samurai.com>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks"
},
{
"msg_contents": "On Tue, 2002-10-22 at 17:09, Tom Lane wrote:\n> Greg Copeland <greg@CopelandConsulting.Net> writes:\n> > I've started playing a little with Postgres to determine if there were\n> > memory leaks running around. After some very brief checking, I'm\n> > starting[1] to think that the answer is yes. Has anyone already gone\n> > through a significant effort to locate and eradicate memory leaks?\n> \n> Yes, this has been dealt with before.\n\nWhat tools, aside from noggin v1.0, did they use? Do we know?\n\n> Have you read\n> src/backend/utils/mmgr/README?\n\nYes...but it's been some time since I last opened it. It was because I\nknew there were some caveats that I wasn't attempting to claim for sure\nthat there were leaks.\n\nI then moved on to psql, again, just for fun. Here, I'm thinking that I\nstarted to find some other leaks...but again, I've not spent any real\ntime on it. So again, I'm not really sure it they are meaningful at\nthis point.\n\n\n> \n> AFAIK the major problems these days are not in the backend as a whole,\n> but in the lesser-used PL language modules (cf recent discussions).\n\nYa, that's what made me wonder about the topic on a larger scale.\n\n> plpgsql has some issues too, I suspect, but not as bad as pltcl etc.\n> Possibly the best answer is to integrate the memory-context notion into\n> those modules; if they did most of their work in a temp context that\n> could be freed once per PL statement or so, the problems would pretty\n> much go away.\n\nInteresting. Having not looked at memory management schemes used in the\npl implementations, can you enlighten me by what you mean by \"integrate\nthe memory-context notion\"? Does that mean they are not using\npalloc/pfree stuff?\n\n> \n> It's fairly difficult to get anywhere with standard leak-tracking tools,\n> since they don't know anything about palloc. What's worse, it is *not*\n> a bug for a routine to palloc space it never pfrees, if it knows that\n> it's palloc'ing in a short-lived memory context. The fact that a\n> context may be released with much still-allocated memory in it is not a\n> bug but a feature; but try teaching that to any standard leak checker...\n> \n> \t\t\tregards, tom lane\n\nWell, the thing that really got my attention is that dmalloc is\nreporting frees on null pointers. While that may be safe on specific\nplatforms, IIRC, it's actually undefined. Again, this is as reported by\ndmalloc so I've yet to actually track it down to determine if it's\npossibly something else going on (magic voodoo of some heap manager,\netc).\n\n\nGreg\n\n\n",
"msg_date": "22 Oct 2002 18:30:13 -0500",
"msg_from": "Greg Copeland <greg@CopelandConsulting.Net>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks"
},
{
"msg_contents": "On 22 Oct 2002, Greg Copeland wrote:\n\n> On Tue, 2002-10-22 at 17:09, Tom Lane wrote:\n> \n> > plpgsql has some issues too, I suspect, but not as bad as pltcl etc.\n> > Possibly the best answer is to integrate the memory-context notion into\n> > those modules; if they did most of their work in a temp context that\n> > could be freed once per PL statement or so, the problems would pretty\n> > much go away.\n> \n> Interesting. Having not looked at memory management schemes used in the\n> pl implementations, can you enlighten me by what you mean by \"integrate\n> the memory-context notion\"? Does that mean they are not using\n> palloc/pfree stuff?\n\nI saw use of a couple of malloc (or Python specific malloc) calls the other day\nbut I also seem to recall that, after consideration, I decided the memory\nneeded to survive for the duration of the backend. Should I have created a new\nchild of the top context and changed these malloc calls?\n\nI was going to ask about thoughts on redirecting malloc etc to palloc etc and\nthereby intercepting memory allocation within the languages and automatically\nbringing them into the memory context realm. However, that would just be making\nlife way too awkward, bearing in mind the above paragraph. Can't we get Sir\nMongle (or whatever the name was) to test these things under the auspices of\nthem being DoS attacks?\n\n\n-- \nNigel J. Andrews\n\n",
"msg_date": "Wed, 23 Oct 2002 00:48:12 +0100 (BST)",
"msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks"
},
{
"msg_contents": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> I saw use of a couple of malloc (or Python specific malloc) calls the\n> other day but I also seem to recall that, after consideration, I\n> decided the memory needed to survive for the duration of the\n> backend. Should I have created a new child of the top context and\n> changed these malloc calls?\n\nIf there is truly no reason to release the memory before the backend\ndies, then I see no reason to avoid malloc(). Basically what the memory\ncontext stuff gives you is the ability to bunch allocations together and\nrelease a whole tree of stuff for little work.\n\nAn example: currently, the plpgsql parser builds its parsed syntax tree\nwith a bunch of malloc's. It has no way to free a syntax tree, so that\ntree lives till the backend exits, whether it's ever used again or not.\nIt would be better to build the tree via palloc's in a context created\nspecially for the specific function: then we could free the whole\ncontext if we discovered that the function had gone away or been\nredefined. (Compare what relcache does for rule syntaxtrees for rules\nassociated with relcache entries.) But right at the moment, there's no\nmechanism to discover that situation, and so there's not now any direct\nvalue in using palloc instead of malloc for plpgsql syntaxtrees. Yet\nthat's surely the direction to go in for the long term.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 22 Oct 2002 23:21:41 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks "
},
{
"msg_contents": "Greg Copeland <greg@copelandconsulting.net> writes:\n> On Tue, 2002-10-22 at 17:09, Tom Lane wrote:\n>> Yes, this has been dealt with before.\n\n> What tools, aside from noggin v1.0, did they use? Do we know?\n\ns/they/me/ ... none. But I don't know of any that I think would be\nuseful.\n\n> I then moved on to psql, again, just for fun. Here, I'm thinking that I\n> started to find some other leaks...but again, I've not spent any real\n> time on it. So again, I'm not really sure it they are meaningful at\n> this point.\n\npsql might well have some internal leaks; the backend memory-context\ndesign doesn't apply to it.\n\n>> Possibly the best answer is to integrate the memory-context notion into\n>> those modules; if they did most of their work in a temp context that\n>> could be freed once per PL statement or so, the problems would pretty\n>> much go away.\n\n> Interesting. Having not looked at memory management schemes used in the\n> pl implementations, can you enlighten me by what you mean by \"integrate\n> the memory-context notion\"? Does that mean they are not using\n> palloc/pfree stuff?\n\nNot everywhere. plpgsql is full of malloc's and I think the other PL\nmodules are too --- and that's not to mention the allocation policies of\nthe perl, tcl, etc, language interpreters. We could use a thorough\nreview of that whole area.\n\n> Well, the thing that really got my attention is that dmalloc is\n> reporting frees on null pointers.\n\nAFAIK that would dump core on many platforms (it sure does here...),\nso I don't think I believe it without seeing chapter and verse. But\nif you can point out where it's really happening, then we must fix it.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 22 Oct 2002 23:28:23 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks "
},
{
"msg_contents": "On Tue, Oct 22, 2002 at 11:28:23PM -0400, Tom Lane wrote:\n> > I then moved on to psql, again, just for fun. Here, I'm thinking that I\n> > started to find some other leaks...but again, I've not spent any real\n> > time on it. So again, I'm not really sure it they are meaningful at\n> > this point.\n> \n> psql might well have some internal leaks; the backend memory-context\n> design doesn't apply to it.\n\n But why? In the Mape project is used mmgr based on PostgreSQL's mmgr and\n it's used for BE and FE. There is not problem with it (BTW backend is\n multithread:-). IMHO use memory-context design for FE is good idea\n if FE a lot works with memory. I already long time think about shared\n lib with PostgreSQL mmgr...\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, 23 Oct 2002 08:13:53 +0200",
"msg_from": "Karel Zak <zakkr@zf.jcu.cz>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks"
},
{
"msg_contents": "On Tue, 22 Oct 2002, Tom Lane wrote:\n\n> Greg Copeland <greg@copelandconsulting.net> writes:\n> \n> > Interesting. Having not looked at memory management schemes used in the\n> > pl implementations, can you enlighten me by what you mean by \"integrate\n> > the memory-context notion\"? Does that mean they are not using\n> > palloc/pfree stuff?\n> \n> Not everywhere. plpgsql is full of malloc's and I think the other PL\n> modules are too --- and that's not to mention the allocation policies of\n> the perl, tcl, etc, language interpreters...\n\nI was going to make the suggestion that malloc et al. could be replaced with\npalloc etc but then that raises too many complications without just shooving\neverything into a long lived context anyway. Also I think we've got to rely on,\ni.e. it is sensible to do so, the underlying language handling memory\ncorrectly.\n\nHmmm...there do seem to be a few mallocs in plpython.c . I haven't looked very\nclosely but nothing jumped out at me as being obviously wrong from the grep\noutput.\n\n\n-- \nNigel J. Andrews\n\n",
"msg_date": "Wed, 23 Oct 2002 12:44:09 +0100 (BST)",
"msg_from": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks"
},
{
"msg_contents": "On Tue, 2002-10-22 at 22:28, Tom Lane wrote:\n> Greg Copeland <greg@copelandconsulting.net> writes:\n> >So again, I'm not really sure it they are meaningful at\n> > this point.\n> \n> psql might well have some internal leaks; the backend memory-context\n> design doesn't apply to it.\n\nOkay. Thanks. I'll probably take another look at it a little later and\nreport back if I find anything of value.\n\n> >Does that mean they are not using\n> > palloc/pfree stuff?\n> \n> Not everywhere. plpgsql is full of malloc's and I think the other PL\n> modules are too --- and that's not to mention the allocation policies of\n> the perl, tcl, etc, language interpreters. We could use a thorough\n> review of that whole area.\n> \n\nOkay. I've started looking at plpython to better understand it's memory\nneeds. I'm seeing a mix of mallocs and PLy_malloc. The PLy version is\nbasically malloc which also checks and reports on memory allocation\nerrors. Anyone know if the cases where malloc was used was purposely\ndone so for performance reasons or simply the flavor or the day?\n\nI thinking for starters, the plpython module could be normalized to use\nthe PLy_malloc stuff across the board. Then again, I still need to\nspend some more time on it. ;)\n\n> > Well, the thing that really got my attention is that dmalloc is\n> > reporting frees on null pointers.\n> \n> AFAIK that would dump core on many platforms (it sure does here...),\n> so I don't think I believe it without seeing chapter and verse. \n\nI actually expected it to do that here on my x86-Linux platform but a\nquick check showed that it was quiet happy with it. What platforms are\nyou using -- just curious?\n\n> But if you can point out where it's really happening, then we must fix it.\n> \n\nI'll trying track this down some more this coming week to see if this is\nreally occurring. After thinking about it, I'm not sure why dmalloc\nwould ever report a free on null if it were not actually occurring. \nAfter all, any call to free should still be showing some memory address\n(valid or otherwise). Off the top of my head, I can't think of an\nartifact that would cause it to falsely report it.\n\n\nGreg\n\n\n",
"msg_date": "23 Oct 2002 08:33:13 -0500",
"msg_from": "Greg Copeland <greg@CopelandConsulting.Net>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks"
},
{
"msg_contents": "\"Nigel J. Andrews\" <nandrews@investsystems.co.uk> writes:\n> On Tue, 22 Oct 2002, Tom Lane wrote:\n>> Not everywhere. plpgsql is full of malloc's and I think the other PL\n>> modules are too --- and that's not to mention the allocation policies of\n>> the perl, tcl, etc, language interpreters...\n\n> I was going to make the suggestion that malloc et al. could be replaced with\n> palloc etc but then that raises too many complications without just shooving\n> everything into a long lived context anyway. Also I think we've got to rely on,\n> i.e. it is sensible to do so, the underlying language handling memory\n> correctly.\n\nIf the perl/tcl/etc interpreters have internal memory leaks, there's\nlittle we can do about that except file bug reports. What I was\nwondering about is whether there isn't deliberate action we need to\ntake to inform those interpreters when data is no longer required.\n\nAn example: when a procedure is updated with CREATE OR REPLACE FUNCTION,\nthe only thing pltcl does about it is a solitary Tcl_DeleteHashEntry();\nit doesn't try to tell Tcl to delete the existing Tcl procedure. That\nmight happen for free (will we always regenerate the same Tcl procedure\nname? not sure), but if the omission causes a leak it's surely not Tcl's\nfault. That's on top of our own data structures about the pltcl\nfunction (pltcl_proc_desc and related stuff), which are definitely\nleaked in this scenario.\n\nSticking all the data about a given function into a memory context\nthat's specific to the function would make it easier to reclaim our own\nmemory in this scenario, but we'd still have to tell Tcl to clean up\nits own memory.\n\nActually pltcl's internal structures about a function look simple enough\nthat it may not be worth using a context for them. It would definitely\nbe useful to do that for plpgsql, though, which builds an extremely\ncomplicated structure for each function (and leaks it all on\nfunction redefinition :-().\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 23 Oct 2002 09:39:47 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks "
},
{
"msg_contents": "Greg Copeland <greg@copelandconsulting.net> writes:\n> Okay. I've started looking at plpython to better understand it's memory\n> needs. I'm seeing a mix of mallocs and PLy_malloc. The PLy version is\n> basically malloc which also checks and reports on memory allocation\n> errors. Anyone know if the cases where malloc was used was purposely\n> done so for performance reasons or simply the flavor or the day?\n\nProbably either oversight or the result of different people's different\ncoding styles.\n\n> I thinking for starters, the plpython module could be normalized to use\n> the PLy_malloc stuff across the board. Then again, I still need to\n> spend some more time on it. ;)\n\nConsistency is good. What I'd wonder about, though, is whether you\nshouldn't be using palloc ;-). malloc, with or without a PLy_ wrapper,\ndoesn't provide any leverage to help you get rid of stuff when you don't\nwant it anymore.\n\n>>> Well, the thing that really got my attention is that dmalloc is\n>>> reporting frees on null pointers.\n>> \n>> AFAIK that would dump core on many platforms (it sure does here...),\n\nI have to take that back: I was thinking about pfree() not free().\nThe ANSI C spec says that free(NULL) is a legal no-op, and there are\njust a few ancient C libraries (perhaps none anymore) where it'll crash.\nI tend to do \"if (ptr) free(ptr)\" from force of habit, but I notice that\npsql (among other places) relies heavily on the ANSI behavior. It's\nprobably pointless to try to convince people to change that coding style.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 23 Oct 2002 09:48:56 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks "
},
{
"msg_contents": "On Wed, 2002-10-23 at 08:48, Tom Lane wrote:\n> Greg Copeland <greg@copelandconsulting.net> writes:\n> > Okay. I've started looking at plpython to better understand it's memory\n> > needs. I'm seeing a mix of mallocs and PLy_malloc. The PLy version is\n> > basically malloc which also checks and reports on memory allocation\n> > errors. Anyone know if the cases where malloc was used was purposely\n> > done so for performance reasons or simply the flavor or the day?\n> \n> Probably either oversight or the result of different people's different\n> coding styles.\n\nMy local copy has this changed to PLy stuff now. Testing shows it's\ngood...then again, I didn't really expect it to change anything. I'll\nsubmit patches later.\n\n> \n> > I thinking for starters, the plpython module could be normalized to use\n> > the PLy_malloc stuff across the board. Then again, I still need to\n> > spend some more time on it. ;)\n> \n> Consistency is good. What I'd wonder about, though, is whether you\n> shouldn't be using palloc ;-). malloc, with or without a PLy_ wrapper,\n> doesn't provide any leverage to help you get rid of stuff when you don't\n> want it anymore.\n\nYa, I'm currently looking to see how the memory is being used and why. \nI'm trying to better understand it's life cycle. You implying that even\nthe short term memory should be using the palloc stuff? What about long\nterm? Blanket statement that pretty much all the PLy stuff should\nreally be using palloc?\n\n> \n> >>> Well, the thing that really got my attention is that dmalloc is\n> >>> reporting frees on null pointers.\n> >> \n> >> AFAIK that would dump core on many platforms (it sure does here...),\n> \n> I have to take that back: I was thinking about pfree() not free().\n> The ANSI C spec says that free(NULL) is a legal no-op, and there are\n\nOh really. I didn't realize that. I've been using the \"if( ptr ) \"\nstuff for so long I didn't realize I didn't need to anymore. Thanks for\nthe update. That was, of course, the cause for alarm.\n\n> It's\n> probably pointless to try to convince people to change that coding style.\n\nWell at this late time, I think it's safe to say that it's not causing\nproblems for anyone on any of the supported platforms. So I'll not\nwaste time looking for it even though I happen think it's a poor\npractice just the same.\n\n\nThanks,\n\n\tGreg\n\n\n",
"msg_date": "23 Oct 2002 09:30:34 -0500",
"msg_from": "Greg Copeland <greg@CopelandConsulting.Net>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks"
},
{
"msg_contents": "Greg Copeland <greg@copelandconsulting.net> writes:\n> Ya, I'm currently looking to see how the memory is being used and why. \n> I'm trying to better understand it's life cycle. You implying that even\n> the short term memory should be using the palloc stuff? What about long\n> term? Blanket statement that pretty much all the PLy stuff should\n> really be using palloc?\n\nShort-term stuff almost certainly should be using palloc, IMHO; anything\nthat is not going to survive the current function invocation should be\npalloc'd in CurrentMemoryContext. The main reason for this is that you\ndon't need to fear leaking such memory if the function is aborted by\nelog(). Depending on what you are doing, you may not have to bother\nwith explicit pfree's at all for such stuff. (In a PL handler you could\nprobably get away with omitting pfree's for stuff allocated once per\ncall, but not for stuff allocated once per statement. Unless you were to\nmake a new context that gets reset for each statement ... which might be\na good idea.)\n\nFor stuff that is going to live a long time and then be explicitly\nfreed, I don't think there's a hard-and-fast rule about which to use.\nIf you are building a complex data structure (parsetree, say) then it's\ngoing to be easier to build it in a memory context and then just free\nthe context rather than tearing down the data structure piece-by-piece.\nBut when you are talking about a single object, there's not a heck of a\nlot of difference between malloc() and palloc in TopMemoryContext.\n\nI'd lean towards using the palloc routines anyway, for consistency of\ncoding style, but that's a judgment call not a must-do thing.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 23 Oct 2002 10:52:33 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks "
},
{
"msg_contents": "Greg Copeland wrote:\n\n>I've started playing a little with Postgres to determine if there were\n>memory leaks running around. After some very brief checking, I'm\n>starting[1] to think that the answer is yes. Has anyone already gone\n>through a significant effort to locate and eradicate memory leaks? Is\n>this done periodically? If so, what tools are others using? I'm\n>currently using dmalloc for my curiosity.\n>\n>[1] Not sure yet as I'm really wanting to find culumative leaks rather\n>than one shot allocations which are simply never freed prior to process\n>termination.\n>\nWhile all leaks should be fixed, obviously, this is one of the \"good\" \nthings in the parennial \"Thread vs process\" argument for processes.\n\n> \n>\n\n\n",
"msg_date": "Tue, 29 Oct 2002 06:11:37 -0500",
"msg_from": "mlw <pgsql@mohawksoft.com>",
"msg_from_op": false,
"msg_subject": "Re: Memory leaks"
}
] |
[
{
"msg_contents": "I've been thinking about how to convert \"x IN (subselect)\" and EXISTS\nconstructs into join-like processing, and I've run into a small problem\nin getting the planner to do it nicely. The issue is that I need to\ntake the subselect and push it into the jointree --- essentially, make\nit look like a subselect-in-FROM --- so that the join planner can deal\nwith it. Basically, I need to rearrange\n\n\tSELECT ... FROM ... WHERE ... AND x IN (SELECT y FROM ...)\n\ninto\n\n\tSELECT ... FROM ..., (SELECT y FROM ...) ss\n WHERE ... AND x =* ss.y\n\nwhere =* represents some specially-marked RestrictInfo node. (NOT IN is the\nsame except that the RestrictInfo node will be marked differently.)\n\nThe difficulty is that there's no good place to do this in\nsubquery_planner(). We should push the subselect into FROM before we\nrun the pull_up_subqueries() and preprocess_jointree() operations;\nif we don't pull up the subselect into the main query then we won't have\naccomplished very much. But the WHERE clause isn't simplified into a\nform that makes it easy to spot top-level IN() expressions until after\nthat. We can't simply switch the order of the subselect and\nWHERE-clause processing, because pulling up subqueries typically adds\nconditions to the WHERE clause.\n\nI haven't been able to think of a solution to this that doesn't involve\nwasting a lot of cycles by repeating some of these processing steps,\nor missing some optimization possibilities. (For example, if we pull up\na subquery that came from a view, it might contain an IN where-clause,\nwhich ideally we'd want to be able to optimize. It almost seems like\nwe need to be able to loop around the whole operation; but most of the\ntime this will just waste cycles.)\n\nAnyone see a nice way to do this?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 22 Oct 2002 19:18:11 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Thinking about IN/EXISTS optimization"
},
{
"msg_contents": "\nThis sounds like one of those classic optimizer problems we have had to\ndeal with in the past. I suggest you go through the optimizer pass and\nset a boolean in Query whenever you do something that may require\nanother loop through, then at the end, you check the boolean and loop if\nrequired.\n\nI think the rules system has to do something similar. I don't see any\nway around that, but because you are setting the boolean you only loop\nwhen you need to.\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> I've been thinking about how to convert \"x IN (subselect)\" and EXISTS\n> constructs into join-like processing, and I've run into a small problem\n> in getting the planner to do it nicely. The issue is that I need to\n> take the subselect and push it into the jointree --- essentially, make\n> it look like a subselect-in-FROM --- so that the join planner can deal\n> with it. Basically, I need to rearrange\n> \n> \tSELECT ... FROM ... WHERE ... AND x IN (SELECT y FROM ...)\n> \n> into\n> \n> \tSELECT ... FROM ..., (SELECT y FROM ...) ss\n> WHERE ... AND x =* ss.y\n> \n> where =* represents some specially-marked RestrictInfo node. (NOT IN is the\n> same except that the RestrictInfo node will be marked differently.)\n> \n> The difficulty is that there's no good place to do this in\n> subquery_planner(). We should push the subselect into FROM before we\n> run the pull_up_subqueries() and preprocess_jointree() operations;\n> if we don't pull up the subselect into the main query then we won't have\n> accomplished very much. But the WHERE clause isn't simplified into a\n> form that makes it easy to spot top-level IN() expressions until after\n> that. We can't simply switch the order of the subselect and\n> WHERE-clause processing, because pulling up subqueries typically adds\n> conditions to the WHERE clause.\n> \n> I haven't been able to think of a solution to this that doesn't involve\n> wasting a lot of cycles by repeating some of these processing steps,\n> or missing some optimization possibilities. (For example, if we pull up\n> a subquery that came from a view, it might contain an IN where-clause,\n> which ideally we'd want to be able to optimize. It almost seems like\n> we need to be able to loop around the whole operation; but most of the\n> time this will just waste cycles.)\n> \n> Anyone see a nice way to do this?\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": "Tue, 22 Oct 2002 22:38:31 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Thinking about IN/EXISTS optimization"
}
] |
[
{
"msg_contents": "Hi everyone,\n\nThanks to Diogo Biazus <diogo@ikono.com.br>, the Brazilian Portuguese\ntranslation of the PostgreSQL \"Advocacy and Marketing\" site is now\ncompleted and ready for public use:\n\nhttp://advocacy.postgresql.org/?lang=br\n\n:-)\n\nWow, that's 6 languages already, and more are coming along.\n\nAm very, very proud of our community members.\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, 23 Oct 2002 13:48:26 +1000",
"msg_from": "Justin Clift <justin@postgresql.org>",
"msg_from_op": true,
"msg_subject": "Brazilian Portuguese version of the PostgreSQL \"Advocacy and\n\tMarketing\" site is ready"
}
] |
[
{
"msg_contents": "First of all PREPARE/EXECUTE is a wonderful thing to speed up things \nsignificantly.\nI wonder if there is a way to store a parsed/rewritten/planned query in \na table so that it can be loaded again.\n\nThis might be useful when it comes to VERY complex queries (> 10 tables).\nI many applications the situation is like that:\n\na. The user connects to the database.\nb. The user sends various different queries to the server (some might be \nthe same)\nc. The user disconnects.\n\nIf there was a way to store execution plans in a table the user could \nload the execution plans of the most time consuming stuff into the \nbackend without parsing and optimizing it every time he authenticates.\n\nDoes it sound useful to anybody? Is it possible to do it or are there \nsome technical problems?\n\nMaybe this is worth thinking about.\n\n Hans\n\n-- \n*Cybertec Geschwinde u Schoenig*\nLudo-Hartmannplatz 1/14, A-1160 Vienna, Austria\nTel: +43/1/913 68 09; +43/664/233 90 75\nwww.postgresql.at <http://www.postgresql.at>, cluster.postgresql.at \n<http://cluster.postgresql.at>, www.cybertec.at \n<http://www.cybertec.at>, kernel.cybertec.at <http://kernel.cybertec.at>\n\n",
"msg_date": "Wed, 23 Oct 2002 16:24:27 +0200",
"msg_from": "=?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <postgres@cybertec.at>",
"msg_from_op": true,
"msg_subject": "PREPARE / EXECUTE"
},
{
"msg_contents": "If you were using them that frequently, couldn't you just keep a\npersistent connection? If it's not used that often, wouldn't the\noverhead of preparing the query following a new connection become noise?\n\nGreg\n\n\nOn Wed, 2002-10-23 at 09:24, Hans-Jürgen Schönig wrote:\n> First of all PREPARE/EXECUTE is a wonderful thing to speed up things \n> significantly.\n> I wonder if there is a way to store a parsed/rewritten/planned query in \n> a table so that it can be loaded again.\n> \n> This might be useful when it comes to VERY complex queries (> 10 tables).\n> I many applications the situation is like that:\n> \n> a. The user connects to the database.\n> b. The user sends various different queries to the server (some might be \n> the same)\n> c. The user disconnects.\n> \n> If there was a way to store execution plans in a table the user could \n> load the execution plans of the most time consuming stuff into the \n> backend without parsing and optimizing it every time he authenticates.\n> \n> Does it sound useful to anybody? Is it possible to do it or are there \n> some technical problems?\n> \n> Maybe this is worth thinking about.\n> \n> Hans\n> \n> -- \n> *Cybertec Geschwinde u Schoenig*\n> Ludo-Hartmannplatz 1/14, A-1160 Vienna, Austria\n> Tel: +43/1/913 68 09; +43/664/233 90 75\n> www.postgresql.at <http://www.postgresql.at>, cluster.postgresql.at \n> <http://cluster.postgresql.at>, www.cybertec.at \n> <http://www.cybertec.at>, kernel.cybertec.at <http://kernel.cybertec.at>\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": "23 Oct 2002 09:39:28 -0500",
"msg_from": "Greg Copeland <greg@CopelandConsulting.Net>",
"msg_from_op": false,
"msg_subject": "Re: PREPARE / EXECUTE"
},
{
"msg_contents": "On Wed, 2002-10-23 at 10:39, Greg Copeland wrote:\n> If you were using them that frequently, couldn't you just keep a\n> persistent connection? If it's not used that often, wouldn't the\n> overhead of preparing the query following a new connection become noise?\n\nEspecially by the time you add in the dependency tracking (drop table,\nquery must go), and modifications to analyze to clear out the stored\nlist.\n\n> On Wed, 2002-10-23 at 09:24, Hans-Jürgen Schönig wrote:\n> > First of all PREPARE/EXECUTE is a wonderful thing to speed up things \n> > significantly.\n> > I wonder if there is a way to store a parsed/rewritten/planned query in \n> > a table so that it can be loaded again.\n> > \n> > This might be useful when it comes to VERY complex queries (> 10 tables).\n> > I many applications the situation is like that:\n> > \n> > a. The user connects to the database.\n> > b. The user sends various different queries to the server (some might be \n> > the same)\n> > c. The user disconnects.\n> > \n> > If there was a way to store execution plans in a table the user could \n> > load the execution plans of the most time consuming stuff into the \n> > backend without parsing and optimizing it every time he authenticates.\n> > \n> > Does it sound useful to anybody? Is it possible to do it or are there \n> > some technical problems?\n> > \n> > Maybe this is worth thinking about.\n> > \n> > Hans\n> > \n> > -- \n> > *Cybertec Geschwinde u Schoenig*\n> > Ludo-Hartmannplatz 1/14, A-1160 Vienna, Austria\n> > Tel: +43/1/913 68 09; +43/664/233 90 75\n> > www.postgresql.at <http://www.postgresql.at>, cluster.postgresql.at \n> > <http://cluster.postgresql.at>, www.cybertec.at \n> > <http://www.cybertec.at>, kernel.cybertec.at <http://kernel.cybertec.at>\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> ---------------------------(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 Rod Taylor\n\n",
"msg_date": "23 Oct 2002 10:52:58 -0400",
"msg_from": "Rod Taylor <rbt@rbt.ca>",
"msg_from_op": false,
"msg_subject": "Re: PREPARE / EXECUTE"
},
{
"msg_contents": "=?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <postgres@cybertec.at> writes:\n> I wonder if there is a way to store a parsed/rewritten/planned query in \n> a table so that it can be loaded again.\n\nThe original version of the PREPARE patch used a shared-across-backends\ncache for PREPAREd statements. We rejected that for a number of\nreasons, one being the increased difficulty of keeping such a cache up\nto date. I think actually storing the plans on disk would have all the\nsame problems, but worse.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 23 Oct 2002 11:02:14 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: PREPARE / EXECUTE "
},
{
"msg_contents": "The idea is not to have it accross multiple backends and having it in \nsync with the tables in the database. This is not the point.\nMy problem is that I have seen many performance critical applications \nsending just a few complex queries to the server. The problem is: If you \nhave many queries where the relation \"time planner\"/\"time executor\" is \nvery high (eg. complex joins with just one value as the result).\nThese applications stay the same for a long time (maybe even years) and \nso there is no need to worry about new tables and so forth - maybe there \nis not even a need to worry about new data. In these cases we could \nspeed up the database significantly just by avoiding the use of the planner:\n\nAn example:\nI have a join across 10 tables + 2 subselects across 4 tables\non the machine I use for testing:\n planner: 12 seconds\n executor: 1 second\n\nThe application will stay the same forever.\nI could be 10 times faster if there was a way to load the execution plan \ninto the backend.\nThere is no way to use a persistent connection (many clients on \ndifferent machines, dynamic IPs, etc. ...)\nThere is no way to have an \"invalid\" execution plan because there are no \nchanges (new tables etc.) in the database.\n\nAlso: If people execute a prepared query and it fails they will know why \n- queries will fail if people drop a table even if these queries are not \nprepared.\nA new feature like the one we are discussing might be used rarely but if \npeople use it they will benefit A LOT.\n\nIf we had a simple ASCII interface to load the stuff into the planner \npeople could save MANY cycles.\nWhen talking about tuning it is nice to gain 10% or even 20% but in many \ncases it does not solve a problem - if a problem can be reduced by 90% \nit is a REAL gain.\nGaining 10% can be done by tweaking the database a little - gaining \n1000% cannot be done so it might be worth thinking about it even it the \nfeature is only used by 20% of those users out there. 20% of all \npostgres users is most likely more than 15.000 people.\n\nAgain; it is not supposed to be a every-day solution. It is a solution \nfor applications staying the same for a very long time.\n\n Hans\n\n\nTom Lane wrote:\n\n>=?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <postgres@cybertec.at> writes:\n> \n>\n>>I wonder if there is a way to store a parsed/rewritten/planned query in \n>>a table so that it can be loaded again.\n>> \n>>\n>\n>The original version of the PREPARE patch used a shared-across-backends\n>cache for PREPAREd statements. We rejected that for a number of\n>reasons, one being the increased difficulty of keeping such a cache up\n>to date. I think actually storing the plans on disk would have all the\n>same problems, but worse.\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-- \n*Cybertec Geschwinde u Schoenig*\nLudo-Hartmannplatz 1/14, A-1160 Vienna, Austria\nTel: +43/1/913 68 09; +43/664/233 90 75\nwww.postgresql.at <http://www.postgresql.at>, cluster.postgresql.at \n<http://cluster.postgresql.at>, www.cybertec.at \n<http://www.cybertec.at>, kernel.cybertec.at <http://kernel.cybertec.at>\n\n\n",
"msg_date": "Wed, 23 Oct 2002 18:04:01 +0200",
"msg_from": "=?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <postgres@cybertec.at>",
"msg_from_op": true,
"msg_subject": "Re: PREPARE / EXECUTE"
},
{
"msg_contents": "Bruno Wolff III <bruno@wolff.to> writes:\n> Hans-J�rgen Sch�nig <postgres@cybertec.at> wrote:\n>> I have a join across 10 tables + 2 subselects across 4 tables\n>> on the machine I use for testing:\n>> planner: 12 seconds\n>> executor: 1 second\n\n> One option you have is to explicitly give the join order.\n\nYes, this is exactly the sort of situation where forcing the join order\nis a big performance win. See\nhttp://www.ca.postgresql.org/users-lounge/docs/7.2/postgres/explicit-joins.html\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 23 Oct 2002 12:15:11 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: PREPARE / EXECUTE "
},
{
"msg_contents": "On Wed, Oct 23, 2002 at 18:04:01 +0200,\n Hans-J�rgen Sch�nig <postgres@cybertec.at> wrote:\n> \n> An example:\n> I have a join across 10 tables + 2 subselects across 4 tables\n> on the machine I use for testing:\n> planner: 12 seconds\n> executor: 1 second\n> \n> The application will stay the same forever.\n> I could be 10 times faster if there was a way to load the execution plan \n> into the backend.\n\nOne option you have is to explicitly give the join order. You can look at\nexplain to see what order the joins are done in and then rewrite the sql\nto force them to be done in that order. This should keep things simple\nfor the planner.\n",
"msg_date": "Wed, 23 Oct 2002 11:20:42 -0500",
"msg_from": "Bruno Wolff III <bruno@wolff.to>",
"msg_from_op": false,
"msg_subject": "Re: PREPARE / EXECUTE"
},
{
"msg_contents": "This is exactly what we do in case of complex stuff. I know that it can \nhelp to reduce the problem for the planner.\nHowever: If you have explicit joins across 10 tables the SQL statement \nis not that readable any more and it is still slower than a prepared \nexecution plan.\n\nI guess it is worth thinking about prepared plans somewhere on disk.\nIs there a way to transform ASCII -> plan?\n\n Hans\n\n\n\nBruno Wolff III wrote:\n\n>On Wed, Oct 23, 2002 at 18:04:01 +0200,\n> Hans-J�rgen Sch�nig <postgres@cybertec.at> wrote:\n> \n>\n>>An example:\n>>I have a join across 10 tables + 2 subselects across 4 tables\n>>on the machine I use for testing:\n>> planner: 12 seconds\n>> executor: 1 second\n>>\n>>The application will stay the same forever.\n>>I could be 10 times faster if there was a way to load the execution plan \n>>into the backend.\n>> \n>>\n\n-- \n*Cybertec Geschwinde u Schoenig*\nLudo-Hartmannplatz 1/14, A-1160 Vienna, Austria\nTel: +43/1/913 68 09; +43/664/233 90 75\nwww.postgresql.at <http://www.postgresql.at>, cluster.postgresql.at \n<http://cluster.postgresql.at>, www.cybertec.at \n<http://www.cybertec.at>, kernel.cybertec.at <http://kernel.cybertec.at>\n\n\n",
"msg_date": "Wed, 23 Oct 2002 18:55:06 +0200",
"msg_from": "=?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <postgres@cybertec.at>",
"msg_from_op": true,
"msg_subject": "Re: PREPARE / EXECUTE"
},
{
"msg_contents": "Could you use some form of connection proxy where the proxy is actually\nkeeping persistent connections but your application is making transient\nconnections to the proxy? I believe this would result in the desired\nperformance boost and behavior.\n\nNow, the next obvious question...anyone know of any proxy apps available\nfor postgresql?\n\nRegards,\n\n\tGreg\n\n\nOn Wed, 2002-10-23 at 11:04, Hans-Jürgen Schönig wrote:\n> The idea is not to have it accross multiple backends and having it in \n> sync with the tables in the database. This is not the point.\n> My problem is that I have seen many performance critical applications \n> sending just a few complex queries to the server. The problem is: If you \n> have many queries where the relation \"time planner\"/\"time executor\" is \n> very high (eg. complex joins with just one value as the result).\n> These applications stay the same for a long time (maybe even years) and \n> so there is no need to worry about new tables and so forth - maybe there \n> is not even a need to worry about new data. In these cases we could \n> speed up the database significantly just by avoiding the use of the planner:\n> \n> An example:\n> I have a join across 10 tables + 2 subselects across 4 tables\n> on the machine I use for testing:\n> planner: 12 seconds\n> executor: 1 second\n> \n> The application will stay the same forever.\n> I could be 10 times faster if there was a way to load the execution plan \n> into the backend.\n> There is no way to use a persistent connection (many clients on \n> different machines, dynamic IPs, etc. ...)\n> There is no way to have an \"invalid\" execution plan because there are no \n> changes (new tables etc.) in the database.\n> \n> Also: If people execute a prepared query and it fails they will know why \n> - queries will fail if people drop a table even if these queries are not \n> prepared.\n> A new feature like the one we are discussing might be used rarely but if \n> people use it they will benefit A LOT.\n> \n> If we had a simple ASCII interface to load the stuff into the planner \n> people could save MANY cycles.\n> When talking about tuning it is nice to gain 10% or even 20% but in many \n> cases it does not solve a problem - if a problem can be reduced by 90% \n> it is a REAL gain.\n> Gaining 10% can be done by tweaking the database a little - gaining \n> 1000% cannot be done so it might be worth thinking about it even it the \n> feature is only used by 20% of those users out there. 20% of all \n> postgres users is most likely more than 15.000 people.\n> \n> Again; it is not supposed to be a every-day solution. It is a solution \n> for applications staying the same for a very long time.\n> \n> Hans\n> \n> \n> Tom Lane wrote:\n> \n> >=?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <postgres@cybertec.at> writes:\n> > \n> >\n> >>I wonder if there is a way to store a parsed/rewritten/planned query in \n> >>a table so that it can be loaded again.\n> >> \n> >>\n> >\n> >The original version of the PREPARE patch used a shared-across-backends\n> >cache for PREPAREd statements. We rejected that for a number of\n> >reasons, one being the increased difficulty of keeping such a cache up\n> >to date. I think actually storing the plans on disk would have all the\n> >same problems, but worse.\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> -- \n> *Cybertec Geschwinde u Schoenig*\n> Ludo-Hartmannplatz 1/14, A-1160 Vienna, Austria\n> Tel: +43/1/913 68 09; +43/664/233 90 75\n> www.postgresql.at <http://www.postgresql.at>, cluster.postgresql.at \n> <http://cluster.postgresql.at>, www.cybertec.at \n> <http://www.cybertec.at>, kernel.cybertec.at <http://kernel.cybertec.at>\n> \n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n\n\n",
"msg_date": "23 Oct 2002 12:52:34 -0500",
"msg_from": "Greg Copeland <greg@CopelandConsulting.Net>",
"msg_from_op": false,
"msg_subject": "Re: PREPARE / EXECUTE"
},
{
"msg_contents": "Greg Copeland wrote:\n\n>Could you use some form of connection proxy where the proxy is actually\n>keeping persistent connections but your application is making transient\n>connections to the proxy? I believe this would result in the desired\n>performance boost and behavior.\n>\n>Now, the next obvious question...anyone know of any proxy apps available\n>for postgresql?\n>\n>Regards,\n>\n>\tGreg\n>\n> \n>\n\nThere is one load balancing software available which is based on the ACE \nlibrary.\nJust have a look at:\n\nhttp://freshmeat.net/projects/dbbalancer/\n\nI haven't tested it up to now.\n\nI am now looking for a workaround - I am sure that there are many \nworkarounds for this issue (explicit joins, persistent connections, etc. \n...).\nI thought it might be useful to have something like a data type (or \nmaybe a binary field) used to store execution plans.\n\nPeople could use this feature as some sort of \"server side\" function or \nso ...\nIt can be seend as some sort of optimized function in the backend which \ncan be loaded/executed more efficiently.\n\nMaybe others would like to see that feature as well.\n\n Hans\n\n-- \n*Cybertec Geschwinde u Schoenig*\nLudo-Hartmannplatz 1/14, A-1160 Vienna, Austria\nTel: +43/1/913 68 09; +43/664/233 90 75\nwww.postgresql.at <http://www.postgresql.at>, cluster.postgresql.at \n<http://cluster.postgresql.at>, www.cybertec.at \n<http://www.cybertec.at>, kernel.cybertec.at <http://kernel.cybertec.at>\n\n\n",
"msg_date": "Wed, 23 Oct 2002 20:09:37 +0200",
"msg_from": "=?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <hs@cybertec.at>",
"msg_from_op": false,
"msg_subject": "Re: PREPARE / EXECUTE"
},
{
"msg_contents": "On Wed, Oct 23, 2002 at 11:02:14AM -0400, Tom Lane wrote:\n> =?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <postgres@cybertec.at> writes:\n> > I wonder if there is a way to store a parsed/rewritten/planned query in \n> > a table so that it can be loaded again.\n> \n> The original version of the PREPARE patch used a shared-across-backends\n> cache for PREPAREd statements. We rejected that for a number of\n> reasons, one being the increased difficulty of keeping such a cache up\n> to date. I think actually storing the plans on disk would have all the\n> same problems, but worse.\n\n Right.\n\n There's solution: persisten backend (for example like classic\n apache). This solve problem with lifetime of all persistent caches.\n It's already in TODO.\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": "Thu, 24 Oct 2002 08:18:20 +0200",
"msg_from": "Karel Zak <zakkr@zf.jcu.cz>",
"msg_from_op": false,
"msg_subject": "Re: PREPARE / EXECUTE"
}
] |
[
{
"msg_contents": "Hello!\n\nI'm tring to move a database and an application from linux to IRIX and I \nhave some problems getting postgresql to work on IRIX.\nMy problem is that postgres crashes is killed by the kernel because of \n(this is from /var/log/SYSLOG):\n\"unix: |$(0x6db)ALERT: Process [postgres] pid 105818 killed: not enough \nmemory to lock stack\"\nThis happens when running the regression tests at the trigger and also \nwhen I try to import my db schema that I've dump with pg_dump on the \nlinux machine.\nUnfortunately I have not much experience with IRIX, so I don't know if \nit is postgres fault or is a bad configuration. What I've tried to do is \nto increase the \"maxkmem\" kernel parameter from 2000 to 20000 but it \nmakes no difference. The machine has 256MB ram and when running postgres \nseems to take no more than 15MB.\n\nSome system information:\nuname -a: \"IRIX cassini 6.5 04101931 IP32 mips\"\ncc -version: \"MIPSpro Compilers: Version 7.2.1\"\nsrc/templates/irix5 file:\nCC=cc\nCFLAGS='-O2'\nLDFLAGS='-O2 -lcurses -n32'\n\nconfigure command:\n./configure --prefix=/disk1/users/robust/famous/local --with-tcl \n--with-includes=/disk1/users/robust/famous/local/include \n--with-libraries=/disk1/users/robust/famous/local/lib --disable-locale \n--without-java --without-CXX\n\n\nThank you for any ideea that can help,\nNicolae Mihalache\n\n",
"msg_date": "Wed, 23 Oct 2002 17:57:57 +0200",
"msg_from": "Nicolae Mihalache <mache@abcpages.com>",
"msg_from_op": true,
"msg_subject": "crashes with postgresql 7.2.1 on IRIX 6.5"
},
{
"msg_contents": "could send the output from:\nlimit\nand\nsystune | grep shm\nalso, what kind of machine is\nregards\n\n\n\n----- Original Message ----- \nFrom: \"Nicolae Mihalache\" <mache@abcpages.com>\nTo: <lamigo@atc.unican.es>; <pgsql-hackers@postgresql.org>\nSent: Wednesday, October 23, 2002 5:57 PM\nSubject: crashes with postgresql 7.2.1 on IRIX 6.5\n\n\n> Hello!\n> \n> I'm tring to move a database and an application from linux to IRIX and I \n> have some problems getting postgresql to work on IRIX.\n> My problem is that postgres crashes is killed by the kernel because of \n> (this is from /var/log/SYSLOG):\n> \"unix: |$(0x6db)ALERT: Process [postgres] pid 105818 killed: not enough \n> memory to lock stack\"\n> This happens when running the regression tests at the trigger and also \n> when I try to import my db schema that I've dump with pg_dump on the \n> linux machine.\n> Unfortunately I have not much experience with IRIX, so I don't know if \n> it is postgres fault or is a bad configuration. What I've tried to do is \n> to increase the \"maxkmem\" kernel parameter from 2000 to 20000 but it \n> makes no difference. The machine has 256MB ram and when running postgres \n> seems to take no more than 15MB.\n> \n> Some system information:\n> uname -a: \"IRIX cassini 6.5 04101931 IP32 mips\"\n> cc -version: \"MIPSpro Compilers: Version 7.2.1\"\n> src/templates/irix5 file:\n> CC=cc\n> CFLAGS='-O2'\n> LDFLAGS='-O2 -lcurses -n32'\n> \n> configure command:\n> ./configure --prefix=/disk1/users/robust/famous/local --with-tcl \n> --with-includes=/disk1/users/robust/famous/local/include \n> --with-libraries=/disk1/users/robust/famous/local/lib --disable-locale \n> --without-java --without-CXX\n> \n> \n> Thank you for any ideea that can help,\n> Nicolae Mihalache\n> \n> \n\n",
"msg_date": "Wed, 23 Oct 2002 18:05:01 +0200",
"msg_from": "\"Luis Alberto Amigo Navarro\" <lamigo@atc.unican.es>",
"msg_from_op": false,
"msg_subject": "Re: crashes with postgresql 7.2.1 on IRIX 6.5"
},
{
"msg_contents": "Luis Alberto Amigo Navarro wrote:\n> could send the output from:\n> limit\n> and\n> systune | grep shm\n> also, what kind of machine is\n> regards\n> \nrobust@cassini> limit\ncputime unlimited\nfilesize unlimited\ndatasize 2097152 kbytes\nstacksize 524288 kbytes\ncoredumpsize unlimited\nmemoryuse 524288 kbytes\nvmemoryuse 2097152 kbytes\ndescriptors 2500\nthreads 1024\n\nrobust@cassini> systune |grep shm\n group: shm (statically changeable)\n sshmseg = 2000 (0x7d0)\n shmmni = 1064 (0x428)\n shmmin = 1 (0x1)\n shmmax = 214748365 (0xccccccd)\n\nrobust@cassini> hinv\nCPU: MIPS R10000 Processor Chip Revision: 2.7\nFPU: MIPS R10010 Floating Point Chip Revision: 0.0\n1 195 MHZ IP32 Processor\nMain memory size: 256 Mbytes\nSecondary unified instruction/data cache size: 1 Mbyte on Processor 0\nInstruction cache size: 32 Kbytes\nData cache size: 32 Kbytes\nFLASH PROM version 4.17\nIntegral SCSI controller 0: Version ADAPTEC 7880\n Disk drive: unit 2 on SCSI controller 0\n CDROM: unit 4 on SCSI controller 0\nIntegral SCSI controller 1: Version ADAPTEC 7880\n Disk drive: unit 5 on SCSI controller 1\nOn-board serial ports: tty1\nOn-board serial ports: tty2\nOn-board EPP/ECP parallel port\nCRM graphics installed\nIntegral Ethernet: ec0, version 1\nIris Audio Processor: version A3 revision 0\nVideo: MVP unit 0 version 1.4\nAV: AV2 Card version 0, Camera not connected.\nVice: TRE\nIEEE 1394 High performance serial bus controller 0: Type: OHCI, Version \n0x104C-1 0\n\n",
"msg_date": "Wed, 23 Oct 2002 18:09:07 +0200",
"msg_from": "Nicolae Mihalache <mache@abcpages.com>",
"msg_from_op": true,
"msg_subject": "Re: crashes with postgresql 7.2.1 on IRIX 6.5"
},
{
"msg_contents": "System configuration looks fine to me, there is no upper limit, the only\nconfiguration problem a can see is that stacksize limit is set to 512Mb when\nu only have 256Mb ram, try \"limit stacksize 65536\" (64Mb) a most serious\nsize, if it does not help I can't be of any more help.\nIf it helps u may need to change hard limit(this is a soft limit) with\nsystune.\nRegards\n\n\n\n----- Original Message -----\nFrom: \"Nicolae Mihalache\" <mache@abcpages.com>\nTo: \"Luis Alberto Amigo Navarro\" <lamigo@atc.unican.es>\nCc: <pgsql-hackers@postgresql.org>\nSent: Wednesday, October 23, 2002 6:09 PM\nSubject: Re: crashes with postgresql 7.2.1 on IRIX 6.5\n\n\n> Luis Alberto Amigo Navarro wrote:\n> > could send the output from:\n> > limit\n> > and\n> > systune | grep shm\n> > also, what kind of machine is\n> > regards\n> >\n> robust@cassini> limit\n> cputime unlimited\n> filesize unlimited\n> datasize 2097152 kbytes\n> stacksize 524288 kbytes\n> coredumpsize unlimited\n> memoryuse 524288 kbytes\n> vmemoryuse 2097152 kbytes\n> descriptors 2500\n> threads 1024\n>\n> robust@cassini> systune |grep shm\n> group: shm (statically changeable)\n> sshmseg = 2000 (0x7d0)\n> shmmni = 1064 (0x428)\n> shmmin = 1 (0x1)\n> shmmax = 214748365 (0xccccccd)\n>\n> robust@cassini> hinv\n> CPU: MIPS R10000 Processor Chip Revision: 2.7\n> FPU: MIPS R10010 Floating Point Chip Revision: 0.0\n> 1 195 MHZ IP32 Processor\n> Main memory size: 256 Mbytes\n> Secondary unified instruction/data cache size: 1 Mbyte on Processor 0\n> Instruction cache size: 32 Kbytes\n> Data cache size: 32 Kbytes\n> FLASH PROM version 4.17\n> Integral SCSI controller 0: Version ADAPTEC 7880\n> Disk drive: unit 2 on SCSI controller 0\n> CDROM: unit 4 on SCSI controller 0\n> Integral SCSI controller 1: Version ADAPTEC 7880\n> Disk drive: unit 5 on SCSI controller 1\n> On-board serial ports: tty1\n> On-board serial ports: tty2\n> On-board EPP/ECP parallel port\n> CRM graphics installed\n> Integral Ethernet: ec0, version 1\n> Iris Audio Processor: version A3 revision 0\n> Video: MVP unit 0 version 1.4\n> AV: AV2 Card version 0, Camera not connected.\n> Vice: TRE\n> IEEE 1394 High performance serial bus controller 0: Type: OHCI, Version\n> 0x104C-1 0\n>\n>\n\n\n",
"msg_date": "Wed, 23 Oct 2002 18:33:53 +0200",
"msg_from": "\"Luis Alberto Amigo Navarro\" <lamigo@atc.unican.es>",
"msg_from_op": false,
"msg_subject": "Re: crashes with postgresql 7.2.1 on IRIX 6.5"
},
{
"msg_contents": "Luis Alberto Amigo Navarro wrote:\n> System configuration looks fine to me, there is no upper limit, the only\n> configuration problem a can see is that stacksize limit is set to 512Mb when\n> u only have 256Mb ram, try \"limit stacksize 65536\" (64Mb) a most serious\n> size, if it does not help I can't be of any more help.\n\nNo it does not help :(\nIs there any place where I can download binaries for postgresql that \ninclude support for pltcl?\n\nThanks,\nmache\n\n",
"msg_date": "Wed, 23 Oct 2002 18:46:41 +0200",
"msg_from": "Nicolae Mihalache <mache@abcpages.com>",
"msg_from_op": true,
"msg_subject": "Re: crashes with postgresql 7.2.1 on IRIX 6.5"
},
{
"msg_contents": "Hello again to everybody!\n\nI've just downloaded and installed a binary version of postgresql 7.1.1 \nfrom ftp.postgresql.org and I have the same bad results: crashes during \nregression tests. Can please someone confirm me that the binary version \nposted there is able to pass all the regression tests? This way I will \nknow that the problem is somwere in my IRIX setup.\n\nThank you,\nmache\n\nOn Wed, 23 Oct 2002, Nicolae Mihalache wrote:\n\n> Hello!\n> \n> I'm tring to move a database and an application from linux to IRIX and I \n> have some problems getting postgresql to work on IRIX.\n> My problem is that postgres crashes is killed by the kernel because of \n> (this is from /var/log/SYSLOG):\n> \"unix: |$(0x6db)ALERT: Process [postgres] pid 105818 killed: not enough \n> memory to lock stack\"\n> This happens when running the regression tests at the trigger and also \n> when I try to import my db schema that I've dump with pg_dump on the \n> linux machine.\n> Unfortunately I have not much experience with IRIX, so I don't know if \n> it is postgres fault or is a bad configuration. What I've tried to do is \n> to increase the \"maxkmem\" kernel parameter from 2000 to 20000 but it \n> makes no difference. The machine has 256MB ram and when running postgres \n> seems to take no more than 15MB.\n> \n> Some system information:\n> uname -a: \"IRIX cassini 6.5 04101931 IP32 mips\"\n> cc -version: \"MIPSpro Compilers: Version 7.2.1\"\n> src/templates/irix5 file:\n> CC=cc\n> CFLAGS='-O2'\n> LDFLAGS='-O2 -lcurses -n32'\n> \n> configure command:\n> ./configure --prefix=/disk1/users/robust/famous/local --with-tcl \n> --with-includes=/disk1/users/robust/famous/local/include \n> --with-libraries=/disk1/users/robust/famous/local/lib --disable-locale \n> --without-java --without-CXX\n> \n> \n> Thank you for any ideea that can help,\n> Nicolae Mihalache\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",
"msg_date": "Wed, 23 Oct 2002 18:01:29 -0400 (EDT)",
"msg_from": "mache@abcpages.com",
"msg_from_op": false,
"msg_subject": "Re: crashes with postgresql 7.2.1 on IRIX 6.5"
},
{
"msg_contents": "regression tests are tested and passed with IRIX 6.5.13 and 6.5.16 and\npostgres from 7.1.3 to 7.2.2\nthere where previously passed tests with IRIX 6.5.5, but I don't know what\npostgres versions\nRegards\n\n\n----- Original Message -----\nFrom: <mache@abcpages.com>\nTo: <lamigo@atc.unican.es>; <pgsql-hackers@postgresql.org>\nSent: Thursday, October 24, 2002 12:01 AM\nSubject: Re: [HACKERS] crashes with postgresql 7.2.1 on IRIX 6.5\n\n\n> Hello again to everybody!\n>\n> I've just downloaded and installed a binary version of postgresql 7.1.1\n> from ftp.postgresql.org and I have the same bad results: crashes during\n> regression tests. Can please someone confirm me that the binary version\n> posted there is able to pass all the regression tests? This way I will\n> know that the problem is somwere in my IRIX setup.\n>\n> Thank you,\n> mache\n>\n> On Wed, 23 Oct 2002, Nicolae Mihalache wrote:\n>\n> > Hello!\n> >\n> > I'm tring to move a database and an application from linux to IRIX and I\n> > have some problems getting postgresql to work on IRIX.\n> > My problem is that postgres crashes is killed by the kernel because of\n> > (this is from /var/log/SYSLOG):\n> > \"unix: |$(0x6db)ALERT: Process [postgres] pid 105818 killed: not enough\n> > memory to lock stack\"\n> > This happens when running the regression tests at the trigger and also\n> > when I try to import my db schema that I've dump with pg_dump on the\n> > linux machine.\n> > Unfortunately I have not much experience with IRIX, so I don't know if\n> > it is postgres fault or is a bad configuration. What I've tried to do is\n> > to increase the \"maxkmem\" kernel parameter from 2000 to 20000 but it\n> > makes no difference. The machine has 256MB ram and when running postgres\n> > seems to take no more than 15MB.\n> >\n> > Some system information:\n> > uname -a: \"IRIX cassini 6.5 04101931 IP32 mips\"\n> > cc -version: \"MIPSpro Compilers: Version 7.2.1\"\n> > src/templates/irix5 file:\n> > CC=cc\n> > CFLAGS='-O2'\n> > LDFLAGS='-O2 -lcurses -n32'\n> >\n> > configure command:\n> > ./configure --prefix=/disk1/users/robust/famous/local --with-tcl\n> > --with-includes=/disk1/users/robust/famous/local/include\n> > --with-libraries=/disk1/users/robust/famous/local/lib --disable-locale\n> > --without-java --without-CXX\n> >\n> >\n> > Thank you for any ideea that can help,\n> > Nicolae Mihalache\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> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n>\n> http://archives.postgresql.org\n>\n>\n\n\n",
"msg_date": "Thu, 24 Oct 2002 10:24:01 +0200",
"msg_from": "\"Luis Alberto Amigo Navarro\" <lamigo@atc.unican.es>",
"msg_from_op": false,
"msg_subject": "Re: crashes with postgresql 7.2.1 on IRIX 6.5"
}
] |
[
{
"msg_contents": "Thomas,\n\nThanks for replying so quickly.\n\nUpon some further investigation, the conversion of the time to BST is OK on\ndata inserted before the upgrade (the display was just a red herring), the\nproblem is just on new data inserted into the table with a default value for\nthe timestamp fields.\n\nThe output from \\d on the table is below:\n\ncapitaras_live=# \\d e_application\n Table \"e_application\"\n Column | Type |\nModifiers \n-------------------+--------------------------+-----------------------------\n---------------------------------------\n e_app_id_num | integer | not null default\nnextval('\"e_application_e_app_id_num_seq\"'::text)\n e_app_cand_id_num | integer | \n e_app_job_id | character varying(30) | not null\n e_app_cid_num | bigint | \n e_app_cts | character varying(30) | \n e_app_mts | character varying(30) | \n e_app_print_date | timestamp with time zone | default\n\"timestamp\"('epoch'::text)\n e_app_xfer_date | timestamp with time zone | default\n\"timestamp\"('epoch'::text)\n e_app_completed | character varying(6) | \n e_app_xml | text | \n e_app_modified | timestamp with time zone | \n e_app_created | timestamp with time zone | \n e_app_wd_feedback | text | \n e_app_submitted | timestamp with time zone | \nIndexes: e_app_completed_idx,\n e_app_job_id_and_completed_idx\nPrimary key: e_application_pkey\nTriggers: RI_ConstraintTrigger_27961\n\nWith a bit more playing, it definitely looks like something wrong with the\nway the default value has been set follwing the export & import for the\nupgrade:-\n\nCreate table arnold (\n\ta\tint8,\n\tb\ttimestamp default 'epoch'::timestamp,\n\tc\ttimestamp default \"timestamp\"('epoch'::text)\n);\n\ncapitaras_test=# \\d arnold\n Table \"arnold\"\n Column | Type | Modifiers\n\n--------+--------------------------+----------------------------------------\n-------\n a | bigint | \n b | timestamp with time zone | default '1970-01-01\n01:00:00+01'::timestamptz\n c | timestamp with time zone | default \"timestamp\"('epoch'::text) \n\ncapitaras_test=# insert into arnold (a) values (1);\nINSERT 182907 1\ncapitaras_test=# insert into arnold (a) values (2);\nINSERT 182907 1\n\ncapitaras_test=# set timezone to 'Europe/London';\nSET VARIABLE\ncapitaras_test=# select * from arnold;\n a | b | c \n---+------------------------+------------------------\n 1 | 1970-01-01 01:00:00+01 | 1970-01-01 00:00:00+01\n 2 | 1970-01-01 01:00:00+01 | 1970-01-01 00:00:00+01\n(2 rows)\n\ncapitaras_test=# set timezone to 'GMT';\nSET VARIABLE\ncapitaras_test=# select * from arnold\ncapitaras_test-# ;\n a | b | c \n---+------------------------+------------------------\n 1 | 1970-01-01 00:00:00+00 | 1969-12-31 23:00:00+00\n 2 | 1970-01-01 00:00:00+00 | 1969-12-31 23:00:00+00\n(2 rows)\n\n\nI think that the best way forward for us (short of re-writing the backend to\nuse NULL) will be to just alter the default value to the one in column b in\nthe test table above.\n\nWith regards the disclaimer, don't you just love lawyers ;-)\n\nOnce again, thanks for replying so quickly.\n\nRegards\n\nJohn.\n\n\n\n-----Original Message-----\nFrom: Thomas Lockhart [mailto:lockhart@fourpalms.org] \nSent: 23 October 2002 16:06\nTo: Hosen, John\nCc: 'pgsql-hackers@postgresql.org'\nSubject: Re: 'epoch'::timestamp and Daylight Savings\n\n\n> We have just upgraded from 7.1.3 to 7.2.3 over the weekend, and have \n> just noticed something weird with regards 'epoch'::timestamp. In \n> 7.1.3, 'epoch'::timestamp always returned the value 1970-01-01 \n> 00:00:00+00,...\n\nI would expect 'epoch'::timestamp to always act as though the value were \nthe same as Unix time zero. But it did not explicitly return that value:\n\nlockhart=# select version();\n-------------------------------------------------------------\n PostgreSQL 7.1.2 on i686-pc-linux-gnu, compiled by GCC 2.96\n\nlockhart=# select 'epoch'::timestamp;\n----------\n epoch\n\nlockhart=# select 'epoch'::timestamp + '0 sec'::interval;\n------------------------\n 1969-12-31 16:00:00-08\n\n\n> following the upgrade we now get:-\n> timestamptz \n> ------------------------\n> 1970-01-01 01:00:00+01\n\nWhich is consistant with the previous result in absolute time.\n\nYou will find that the most recent versions of PostgreSQL convert \n'epoch' on input, and that with- and without time zone data types are \navailable:\n\nlockhart=# select cast('epoch' as timestamp with time zone);\n timestamptz\n------------------------\n 1969-12-31 16:00:00-08\n\nlockhart=# select cast('epoch' as timestamp without time zone);\n timestamp\n---------------------\n 1970-01-01 00:00:00\n\n\n> Also, any fields set to 'epoch'::timestamp before the upgrade now return:-\n> e_app_xfer_date \n> ------------------------\n> 1970-01-01 00:00:00+01\n> If we issue a SET TIMEZONE TO 'GMT' the results are the same as for \n> the previous release.\n\nNot sure about this one. What is the schema? Can you give an example \nwhere the time gets shifted by an hour?\n\n> This is causing us a problem as we use epoch to indicate whether the \n> data has been processed, and to log the date and time of processing.\n\nUsing \"special values\" to indicate status can be troublesome, as you are \nfinding. I'd suggest using NULL to indicate that a field is not known or \nnot yet set.\n\n> This email and any files attached to it are confidential and intended\n> solely for the use of the individual or entity to whom they are\n> addressed. If you have received this email in error please notify \n> the system manager.\n\nOops. Can you please confirm that we were the intended recipients? ;)\n\nhth\n\n - Thomas\n\n\nThis email has been scanned for all viruses by the MessageLabs SkyScan\nservice.\n\n\n***********************************************************************\n\nThis email and any files attached to it are confidential and intended \nsolely for the use of the individual or entity to whom they are\naddressed. If you have received this email in error please notify \nthe system manager.\n\nThe message and any files attached to it have been scanned by \nMIMEsweeper with Sophos Sweep and found to be free from all known\nviruses.\n\nInformation on MIMEsweeper can be found at http://www.mimesweeper.com/\n\n***********************************************************************\n\n",
"msg_date": "Wed, 23 Oct 2002 16:58:40 +0100",
"msg_from": "\"Hosen, John\" <John.Hosen@capita.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: 'epoch'::timestamp and Daylight Savings"
},
{
"msg_contents": "> Create table arnold (\n> \ta\tint8,\n> \tb\ttimestamp default 'epoch'::timestamp,\n> \tc\ttimestamp default \"timestamp\"('epoch'::text)\n> );\n> I think that the best way forward for us (short of re-writing the backend to\n> use NULL) will be to just alter the default value to the one in column b in\n> the test table above.\n\nHmm. The \"timestamp\"() call is forcing TIMESTAMP WITHOUT TIME ZONE, \nwhich then gets converted to TIMESTAMP WITH TIME ZONE (adding in your \none hour offset), which is the default for the unadorned, unquoted \n'timestamp' type.\n\nYou can use something like\n\n cast('epoch'::text as timestamp with time zone)\n\nto get what you want, and can use \"timestamptz\"() if you insist. But \nthat is not recommended for direct use in schema definitions, even if \npgsql chooses to use it for dump/reloads at the moment.\n\n - Thomas\n\n",
"msg_date": "Wed, 23 Oct 2002 18:06:39 +0000",
"msg_from": "Thomas Lockhart <lockhart@fourpalms.org>",
"msg_from_op": false,
"msg_subject": "Re: 'epoch'::timestamp and Daylight Savings"
},
{
"msg_contents": "\"Hosen, John\" <John.Hosen@capita.co.uk> writes:\n\n> e_app_print_date | timestamp with time zone | default\n> \"timestamp\"('epoch'::text)\n\nYeah, there's your problem. You are casting 'epoch' to type timestamp\nwithout time zone, and thence to timestamp with time zone. The first\nstep gives \"midnight\" and the second assumes that that means \"midnight\nlocal time\".\n\nFor example, in US EST zone I get:\n\nregression=# select \"timestamp\"('epoch'::text);\n timestamp\n---------------------\n 1970-01-01 00:00:00\n(1 row)\n\nregression=# select (\"timestamp\"('epoch'::text))::timestamptz;\n timestamp\n------------------------\n 1970-01-01 00:00:00-05\n(1 row)\n\nwhereas what is wanted is\n\nregression=# select \"timestamptz\"('epoch'::text);\n timestamptz\n------------------------\n 1969-12-31 19:00:00-05\n(1 row)\n\nSo you can fix the problem just by setting the default to be\n'epoch'::timestamptz.\n\nThe problem is probably related to the renaming we've been carrying out\nto get closer to the SQL spec: \"timestamp\" now means timestamp without\ntime zone, which is not what it meant in older Postgres releases.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 23 Oct 2002 14:46:33 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: 'epoch'::timestamp and Daylight Savings "
}
] |
[
{
"msg_contents": "\n> The idea is not to have it accross multiple backends and having it in \n> sync with the tables in the database. This is not the point.\n> My problem is that I have seen many performance critical applications \n> sending just a few complex queries to the server. The problem is: If you \n> have many queries where the relation \"time planner\"/\"time executor\" is \n> very high (eg. complex joins with just one value as the result).\n\nThe standard approach to such a scenario would imho be to write stored procedures\nfor the complex queries (e.g. plpgsql) and use that from the client.\nMaybe even eliminate a few ping pongs between client and server.\n\nAndreas\n",
"msg_date": "Wed, 23 Oct 2002 20:58:13 +0200",
"msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>",
"msg_from_op": true,
"msg_subject": "Re: PREPARE / EXECUTE"
},
{
"msg_contents": ">\n>\n>\n>The standard approach to such a scenario would imho be to write stored procedures\n>for the complex queries (e.g. plpgsql) and use that from the client.\n>Maybe even eliminate a few ping pongs between client and server.\n>\n>Andreas\n> \n>\n\nDoes it reduce the time taken by the planner?\nAre server side SQL functions optimized at runtime or at \"create \nfunction\" time?\n\nIf the function is optimized at runtime it is not a gain.\n\n Hans\n\n\n-- \n*Cybertec Geschwinde u Schoenig*\nLudo-Hartmannplatz 1/14, A-1160 Vienna, Austria\nTel: +43/1/913 68 09; +43/664/233 90 75\nwww.postgresql.at <http://www.postgresql.at>, cluster.postgresql.at \n<http://cluster.postgresql.at>, www.cybertec.at \n<http://www.cybertec.at>, kernel.cybertec.at <http://kernel.cybertec.at>\n\n\n",
"msg_date": "Wed, 23 Oct 2002 21:09:45 +0200",
"msg_from": "=?ISO-8859-1?Q?Hans-J=FCrgen_Sch=F6nig?= <hs@cybertec.at>",
"msg_from_op": false,
"msg_subject": "Re: PREPARE / EXECUTE"
},
{
"msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> The standard approach to such a scenario would imho be to write\n> stored procedures for the complex queries (e.g. plpgsql) and use\n> that from the client. Maybe even eliminate a few ping pongs between\n> client and server.\n\nSince PL/PgSQL cached query plans are flushed when the backend exits,\nhow would this help?\n\nRegarding the original suggestion of storing prepared plans on disk, I\nagree with Tom -- it's basically the same idea as storing plans in\nshared memory, which we previously considered (and Karel implemented),\nbut ultimately decided to remove. IMHO, the utility of this feature\ndoesn't justify the problems that would come with implementing it (see\nthe archives for the original implementation discussions).\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n",
"msg_date": "23 Oct 2002 17:35:49 -0400",
"msg_from": "Neil Conway <neilc@samurai.com>",
"msg_from_op": false,
"msg_subject": "Re: PREPARE / EXECUTE"
}
] |
[
{
"msg_contents": "Tree weeks later than I promised, but it is finished (I hope).\n\nIn attachment are files:\n\tdblink.c\n\tdblink.h\n\tdblink.sql.in\n\tpqtest.sql\n\nIn file pqtest.sql is sample queries and results. It seem OK to me.\n\nThere are two reasons why I did not make a diff.\n\n1. The source I started from is 7.3b1, not the latest.\n2. I would like You to check the code, especially the part that touches memory \nmanagement. \nI can say that it works, but I do not know exactly why, and this can be \ndangerous. With my knowledge of postgres internals this is\nas far I can go at the moment. And once more sorry for bad English !\n\nRegards !",
"msg_date": "Wed, 23 Oct 2002 21:51:06 -0100",
"msg_from": "Darko Prenosil <darko.prenosil@finteh.hr>",
"msg_from_op": true,
"msg_subject": "plpq"
},
{
"msg_contents": "\nWould you tell us exactly what this is.\n\n---------------------------------------------------------------------------\n\nDarko Prenosil wrote:\n> \tTree weeks later than I promised, but it is finished (I hope).\n> \n> In attachment are files:\n> \tdblink.c\n> \tdblink.h\n> \tdblink.sql.in\n> \tpqtest.sql\n> \n> In file pqtest.sql is sample queries and results. It seem OK to me.\n> \n> There are two reasons why I did not make a diff.\n> \n> 1. The source I started from is 7.3b1, not the latest.\n> 2. I would like You to check the code, especially the part that touches memory \n> management. \n> I can say that it works, but I do not know exactly why, and this can be \n> dangerous. With my knowledge of postgres internals this is\n> as far I can go at the moment. And once more sorry for bad English !\n> \n> Regards !\n\n[ Attachment, skipping... ]\n\n[ Attachment, skipping... ]\n\n[ Attachment, skipping... ]\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": "Mon, 28 Oct 2002 15:35:03 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: plpq"
},
{
"msg_contents": "\tIt is wrapper about libpq client library functions for use in PL/PSQL.\nI agreed with Joe Conway that it may fit within dblink, because dblink is much\nmore easy to work with than the libpq, but sometimes it is Just not enough.\nSo, the idea is to re-implement all important libpq functions like \nPQConnectdb,\nPQExec ... and make one postgres server become just a client for another (or \nanother's) postgres server. As You know, dblink still does not support \nsimultaneous connections to two different databases, but with this it can !\nI agreed with Joe that when he is back from holiday he will take a look.\n\nRegards !\n\nOn Monday 28 October 2002 20:35, Bruce Momjian wrote:\n> Would you tell us exactly what this is.\n>\n> ---------------------------------------------------------------------------\n>\n> Darko Prenosil wrote:\n> > \tTree weeks later than I promised, but it is finished (I hope).\n> >\n> > In attachment are files:\n> > \tdblink.c\n> > \tdblink.h\n> > \tdblink.sql.in\n> > \tpqtest.sql\n> >\n> > In file pqtest.sql is sample queries and results. It seem OK to me.\n> >\n> > There are two reasons why I did not make a diff.\n> >\n> > 1. The source I started from is 7.3b1, not the latest.\n> > 2. I would like You to check the code, especially the part that touches\n> > memory management.\n> > I can say that it works, but I do not know exactly why, and this can be\n> > dangerous. With my knowledge of postgres internals this is\n> > as far I can go at the moment. And once more sorry for bad English !\n> >\n> > Regards !\n>\n> [ Attachment, skipping... ]\n>\n> [ Attachment, skipping... ]\n>\n> [ Attachment, skipping... ]\n>\n> [ Attachment, skipping... ]\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",
"msg_date": "Tue, 29 Oct 2002 08:27:11 +0000",
"msg_from": "Darko Prenosil <darko.prenosil@finteh.hr>",
"msg_from_op": true,
"msg_subject": "Re: plpq"
},
{
"msg_contents": "Darko Prenosil wrote:\n> \tIt is wrapper about libpq client library functions for use in PL/PSQL.\n> I agreed with Joe Conway that it may fit within dblink, because dblink is much\n> more easy to work with than the libpq, but sometimes it is Just not enough.\n> So, the idea is to re-implement all important libpq functions like \n> PQConnectdb,\n> PQExec ... and make one postgres server become just a client for another (or \n> another's) postgres server. As You know, dblink still does not support \n> simultaneous connections to two different databases, but with this it can !\n> I agreed with Joe that when he is back from holiday he will take a look.\n\nBrilliant idea. I assume you added those functions and rewrote dblink\nto use them. dblink already is great, but by adding libpq inside\nplpgsql, all sorts of other things become possible. \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, 29 Oct 2002 12:39:40 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: plpq"
}
] |
[
{
"msg_contents": "Does anyone remember why a standalone backend doesn't read\npostgresql.conf?\n\nFor example, I just set 'autocommit' to off in postgresql.conf, and\nafter starting the standalone backend, it doesn't have autocommit off.\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, 23 Oct 2002 21:22:29 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Standalone backend doesn't read postgresql.conf"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Does anyone remember why a standalone backend doesn't read\n> postgresql.conf?\n\nHm. I remember why it doesn't pay attention to the per-user and\nper-database settings (if they're screwed up, you might have no\nway to get in and fix 'em). But that argument doesn't apply to\npostgresql.conf, since you can always fix it with a text editor.\n\nThis might just be an oversight. Peter, any comments?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 23 Oct 2002 22:27:53 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Standalone backend doesn't read postgresql.conf "
}
] |
[
{
"msg_contents": "Hi:\n\n Is there a way to emulate a SELECT..FOR UPDATE to\nseries of LOCK/SELECT statement.\n\n I tried the following statements using 2 psql\nterminals.\n\n T1 | T2\n1) BEGIN; |\n2) SELECT x | BEGIN;\n FROM y |\n WHERE y=1 |\n FOR UPDATE; |\n3) | SELECT x\n | FROM y\n | WHERE y=1\n | FOR UPDATE;\n4) COMMIT; |\n5) | COMMIT;\n\n At point #3 T2 will wait, however changing the WHERE\nclause to other clause such as \"WHERE y=2\" will allow\nT2 to proceed.\n\n - I tried changing the SELECT..FOR UPDATE into LOCK\nSHARE MODE followed by a SELECT (but w/o FOR UPDATE)\nbut it T2 is allowed to proceed even for the clause\n\"where y=1\".\n\n I am surprised because according to the docs\n(version 7.2), it says:\n ROW SHARE MODE \n Note: Automatically acquired by SELECT ... FOR\nUPDATE. \n I'm assuming that the SELECT..FOR UPDATE performs\na lock in ROW SHARE MODE before the SELECT.\n\n\n I also tried changing the lock mode into SHARE ROW\nEXCLUSIVE MODE / EXCLUSIVE MODE but still T2 won't be\nallowed to passed even the for the clause \"where y=2\".\n\n Is there any to do it? I'm asking becuase our db\nlibaries (using libpq) always a cursor when generating\na SELECT statement thus I'm encourtering the following\nerror message when I use SELECT..FOR UPDATE:\n Cursor must be READ ONLY.\n DECLARE/UPDATE is not supported.\n\n Another alternative would be studying libpq and\nremoving the cursors in a SELECT statement.\n\n By the way is there any side effect / disadavtages\nwhen I remove the \"DELCARE CURSOR\" statement and\nchange it to plain SELECT statememt in C++ codes?\n\n Thank you in advance,\n\nludwig.\n\n\n\n__________________________________________________\nDo you Yahoo!?\nNew DSL Internet Access from SBC & Yahoo!\nhttp://sbc.yahoo.com\n",
"msg_date": "Wed, 23 Oct 2002 21:10:32 -0700 (PDT)",
"msg_from": "Ludwig Lim <lud_nowhere_man@yahoo.com>",
"msg_from_op": true,
"msg_subject": "Simulating a SELECT..FOR UPDATE to LOCK and SELECT statement"
},
{
"msg_contents": "Ludwig Lim <lud_nowhere_man@yahoo.com> writes:\n> Is there a way to emulate a SELECT..FOR UPDATE to\n> series of LOCK/SELECT statement.\n\nNo. LOCK can only acquire table-level locks; the point of SELECT FOR\nUPDATE is to acquire row-level locks.\n\n> I am surprised because according to the docs\n> (version 7.2), it says:\n> ROW SHARE MODE \n> Note: Automatically acquired by SELECT ... FOR\n> UPDATE. \n\nSELECT FOR UPDATE gets a table-level lock just to ensure that the table\ndoesn't disappear from under it while it's doing its scan. The ROW\nSHARE lock is pretty weak though, and doesn't block anything less\ndrastic than DROP/ALTER TABLE or VACUUM FULL.\n\nThe 7.3 development docs are perhaps clearer about this; try\nhttp://developer.postgresql.org/docs/postgres/explicit-locking.html\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 24 Oct 2002 09:42:46 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Simulating a SELECT..FOR UPDATE to LOCK and SELECT statement "
},
{
"msg_contents": "\nHi i think a hit a major problem on 7.2.1.\nI run 3 systems with postgresql 7.2.1.\nIts a redhat 7.1 for development, a redhat 7.3 for production\nand a FreeBSD 4.6.1RC2 for testing.\n\nAfter long runs (with periodic (daily) vacuum analyze's)\ni noticed that some of the triggers that implement referential integrity\nconstraints just disapeared.\nSome of these triggers were still present on the FreeBSD system\n(which has been idle for a month or so), whereas on the linux\nsystems they were absent.\n\nHas any one have a clue??\nAny comment would be really valuable at this moment of darkness.\n\nThanx.\n\n==================================================================\nAchilleus Mantzios\nS/W Engineer\nIT dept\nDynacom Tankers Mngmt\nNikis 4, Glyfada\nAthens 16610\nGreece\ntel: +30-10-8981112\nfax: +30-10-8981877\nemail: achill@matrix.gatewaynet.com\n mantzios@softlab.ece.ntua.gr\n\n",
"msg_date": "Wed, 6 Nov 2002 14:41:52 +0200 (EET)",
"msg_from": "Achilleus Mantzios <achill@matrix.gatewaynet.com>",
"msg_from_op": false,
"msg_subject": "Problem: Referential Integrity Constraints lost"
},
{
"msg_contents": "Also i must that these lost triggers implement the parent table\nside of the constraint, e.g.\n\nCREATE TABLE VslSections(\nid serial NOT NULL UNIQUE PRIMARY KEY,\nname varchar(20) NOT NULL UNIQUE);\n\nCREATE TABLE MachClasses(\nid serial NOT NULL UNIQUE PRIMARY KEY,\nname varchar(20) NOT NULL UNIQUE,\nvslsecid int4 NOT NULL,\nFOREIGN KEY (vslsecid) REFERENCES VslSections (id));\n\nThen the triggers created are :\n\n1)\nCREATE CONSTRAINT TRIGGER \"<unnamed>\" AFTER INSERT OR UPDATE ON\n\"machclasses\" FROM \"vslsections\" NOT DEFERRABLE INITIALLY IMMEDIATE FOR\nEACH ROW EXECUTE PROCEDURE \"RI_FKey_check_ins\" ('<unnamed>',\n'machclasses', 'vslsections', 'UNSPECIFIED', 'vslsecid', 'id');\n2)\nCREATE CONSTRAINT TRIGGER \"<unnamed>\" AFTER DELETE ON \"vslsections\" FROM\n\"machclasses\" NOT DEFERRABLE INITIALLY IMMEDIATE FOR EACH ROW EXECUTE\nPROCEDURE \"RI_FKey_noaction_del\" ('<unnamed>', 'machclasses',\n'vslsections', 'UNSPECIFIED', 'vslsecid', 'id');\n3)\nCREATE CONSTRAINT TRIGGER \"<unnamed>\" AFTER UPDATE ON \"vslsections\" FROM\n\"machclasses\" NOT DEFERRABLE INITIALLY IMMEDIATE FOR EACH ROW EXECUTE\nPROCEDURE \"RI_FKey_noaction_upd\" ('<unnamed>', 'machclasses',\n'vslsections', 'UNSPECIFIED', 'vslsecid', 'id');\n\nThe *LOST* triggers are 2 and 3.\n\n==================================================================\nAchilleus Mantzios\nS/W Engineer\nIT dept\nDynacom Tankers Mngmt\nNikis 4, Glyfada\nAthens 16610\nGreece\ntel: +30-10-8981112\nfax: +30-10-8981877\nemail: achill@matrix.gatewaynet.com\n mantzios@softlab.ece.ntua.gr\n\n\n",
"msg_date": "Wed, 6 Nov 2002 14:53:21 +0200 (EET)",
"msg_from": "Achilleus Mantzios <achill@matrix.gatewaynet.com>",
"msg_from_op": false,
"msg_subject": "Re: [SQL] Problem: Referential Integrity Constraints lost"
},
{
"msg_contents": "\nI was wrong about parent side triggers only having\ndisappeared.\n\nTriggers of both sides are missing.\n\n==================================================================\nAchilleus Mantzios\nS/W Engineer\nIT dept\nDynacom Tankers Mngmt\nNikis 4, Glyfada\nAthens 16610\nGreece\ntel: +30-10-8981112\nfax: +30-10-8981877\nemail: achill@matrix.gatewaynet.com\n mantzios@softlab.ece.ntua.gr\n\n",
"msg_date": "Wed, 6 Nov 2002 16:10:19 +0200 (EET)",
"msg_from": "Achilleus Mantzios <achill@matrix.gatewaynet.com>",
"msg_from_op": false,
"msg_subject": "Re: [SQL] Problem: Referential Integrity Constraints lost: Correction"
},
{
"msg_contents": "On Wed, 6 Nov 2002, Achilleus Mantzios wrote:\n\n>\n> Hi i think a hit a major problem on 7.2.1.\n> I run 3 systems with postgresql 7.2.1.\n> Its a redhat 7.1 for development, a redhat 7.3 for production\n> and a FreeBSD 4.6.1RC2 for testing.\n>\n> After long runs (with periodic (daily) vacuum analyze's)\n> i noticed that some of the triggers that implement referential integrity\n> constraints just disapeared.\n> Some of these triggers were still present on the FreeBSD system\n> (which has been idle for a month or so), whereas on the linux\n> systems they were absent.\n>\n> Has any one have a clue??\n\nHmm, you haven't done anything like cluster or an incomplete dump and\nreload have you?\n\n\n",
"msg_date": "Wed, 6 Nov 2002 08:25:22 -0800 (PST)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: [SQL] Problem: Referential Integrity Constraints lost"
},
{
"msg_contents": "On Wed, 6 Nov 2002, Stephan Szabo wrote:\n\n> On Wed, 6 Nov 2002, Achilleus Mantzios wrote:\n>\n> >\n> > Hi i think a hit a major problem on 7.2.1.\n> > I run 3 systems with postgresql 7.2.1.\n> > Its a redhat 7.1 for development, a redhat 7.3 for production\n> > and a FreeBSD 4.6.1RC2 for testing.\n> >\n> > After long runs (with periodic (daily) vacuum analyze's)\n> > i noticed that some of the triggers that implement referential integrity\n> > constraints just disapeared.\n> > Some of these triggers were still present on the FreeBSD system\n> > (which has been idle for a month or so), whereas on the linux\n> > systems they were absent.\n> >\n> > Has any one have a clue??\n>\n> Hmm, you haven't done anything like cluster or an incomplete dump and\n> reload have you?\n\nNo,\nAlso the FreeBSD system's database was populated with data from\nthe production on 2002-10-22, so the problem on the FreeBSD\nwas partially inherited from the production databse.\n\n>\n>\n>\n\n==================================================================\nAchilleus Mantzios\nS/W Engineer\nIT dept\nDynacom Tankers Mngmt\nNikis 4, Glyfada\nAthens 16610\nGreece\ntel: +30-10-8981112\nfax: +30-10-8981877\nemail: achill@matrix.gatewaynet.com\n mantzios@softlab.ece.ntua.gr\n\n",
"msg_date": "Thu, 7 Nov 2002 09:33:21 +0200 (EET)",
"msg_from": "Achilleus Mantzios <achill@matrix.gatewaynet.com>",
"msg_from_op": false,
"msg_subject": "Re: [SQL] Problem: Referential Integrity Constraints lost"
},
{
"msg_contents": "On Thu, 7 Nov 2002, Achilleus Mantzios wrote:\n\n> On Wed, 6 Nov 2002, Stephan Szabo wrote:\n>\n> > On Wed, 6 Nov 2002, Achilleus Mantzios wrote:\n> >\n> > >\n> > > Hi i think a hit a major problem on 7.2.1.\n> > > I run 3 systems with postgresql 7.2.1.\n> > > Its a redhat 7.1 for development, a redhat 7.3 for production\n> > > and a FreeBSD 4.6.1RC2 for testing.\n> > >\n> > > After long runs (with periodic (daily) vacuum analyze's)\n> > > i noticed that some of the triggers that implement referential integrity\n> > > constraints just disapeared.\n> > > Some of these triggers were still present on the FreeBSD system\n> > > (which has been idle for a month or so), whereas on the linux\n> > > systems they were absent.\n> > >\n> > > Has any one have a clue??\n> >\n> > Hmm, you haven't done anything like cluster or an incomplete dump and\n> > reload have you?\n>\n> No,\n> Also the FreeBSD system's database was populated with data from\n> the production on 2002-10-22, so the problem on the FreeBSD\n> was partially inherited from the production databse.\n\nAlso i must add, that the database on the production system\nwas never dumped/reloaded since the creation of the system.\n\nThe production 7.2.1 pgsql db was created and loaded on 2002-04-20,\nfrom a 7.1.3 pgsql on our previous Solaris box (which we replaced\nwith a new linux one).\nThe production pgsql is started/stopped only during\nsystem shutdowns/boots.\n\nWe had some unexpected system failures due to some\nLinux/MotherBoard/BIOS problems.\n(I was too enthusiastic about pgsql and its stability\nthat i was overconfident about our database's state\nafter these failures).\nBTW, could that be the cause of the problem??\n\nThe problem is that i didnt realize the problem until yesterday.\nThe only thing i am sure, is that some of the triggers lost\none both linux'es are present on the FreeBSD system,\nwhich was populated on Oct 22, and had NO deletion activity\nat all.\n\nI plan to make a huge map of all my tables, and configure\nall the pairs of tables with inter-referential integrity constraints,\npg_dump --schema-only, see which triggers are missing\nand then create them by hand.\n\nHas anyone got a better idea??\nAfter recreating the missing triggers should i upgrade\nto 7.2.3??\n\nThanx.\n\n>\n> >\n> >\n> >\n>\n> ==================================================================\n> Achilleus Mantzios\n> S/W Engineer\n> IT dept\n> Dynacom Tankers Mngmt\n> Nikis 4, Glyfada\n> Athens 16610\n> Greece\n> tel: +30-10-8981112\n> fax: +30-10-8981877\n> email: achill@matrix.gatewaynet.com\n> mantzios@softlab.ece.ntua.gr\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==================================================================\nAchilleus Mantzios\nS/W Engineer\nIT dept\nDynacom Tankers Mngmt\nNikis 4, Glyfada\nAthens 16610\nGreece\ntel: +30-10-8981112\nfax: +30-10-8981877\nemail: achill@matrix.gatewaynet.com\n mantzios@softlab.ece.ntua.gr\n\n",
"msg_date": "Thu, 7 Nov 2002 09:57:20 +0200 (EET)",
"msg_from": "Achilleus Mantzios <achill@matrix.gatewaynet.com>",
"msg_from_op": false,
"msg_subject": "Re: [SQL] Problem: Referential Integrity Constraints lost"
},
{
"msg_contents": "Achilleus Mantzios <achill@matrix.gatewaynet.com> writes:\n> After recreating the missing triggers should i upgrade\n> to 7.2.3??\n\nMake that \"before\".\n\nI frankly suspect pilot error here. Triggers do not simply disappear.\nIf you did have crash-induced corruption leading to loss of some rows\nin pg_trigger, it would be exceedingly obvious because *no* operations\non the affected tables would work --- relcache would complain about the\nfact that pg_class.reltriggers didn't match the number of rows in\npg_trigger. I think the missing triggers must have been removed or\ndisabled deliberately. (Which is not to say that it couldn't have been\na software bug, but you're barking up the wrong tree to blame it on a\ncrash.)\n\nDid all the triggers of the affected tables disappear, or only some\nof them?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 07 Nov 2002 09:55:29 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: [SQL] Problem: Referential Integrity Constraints lost "
},
{
"msg_contents": "On Thu, 7 Nov 2002, Tom Lane wrote:\n\n> Achilleus Mantzios <achill@matrix.gatewaynet.com> writes:\n> > After recreating the missing triggers should i upgrade\n> > to 7.2.3??\n>\n> Make that \"before\".\n>\n> I frankly suspect pilot error here. Triggers do not simply disappear.\n> If you did have crash-induced corruption leading to loss of some rows\n> in pg_trigger, it would be exceedingly obvious because *no* operations\n> on the affected tables would work --- relcache would complain about the\n> fact that pg_class.reltriggers didn't match the number of rows in\n> pg_trigger. I think the missing triggers must have been removed or\n> disabled deliberately. (Which is not to say that it couldn't have been\n> a software bug, but you're barking up the wrong tree to blame it on a\n> crash.)\n> Did all the triggers of the affected tables disappear, or only some\n> of them?\n\nJust some of them.\nI really dont know what happened.\nLooking back at july backups the problem was already there.\nI never played with system tables in production.\nI hope to be able somehow to reproduce the problem,\nor convince myself its my fault for some reason.\n\nP.S.\nI was surprized when i looked up in my english dictionary the\nword \"deliberately\" :)\n\n>\n> \t\t\tregards, tom lane\n>\n\n==================================================================\nAchilleus Mantzios\nS/W Engineer\nIT dept\nDynacom Tankers Mngmt\nNikis 4, Glyfada\nAthens 16610\nGreece\ntel: +30-10-8981112\nfax: +30-10-8981877\nemail: achill@matrix.gatewaynet.com\n mantzios@softlab.ece.ntua.gr\n\n",
"msg_date": "Thu, 7 Nov 2002 18:38:06 +0200 (EET)",
"msg_from": "Achilleus Mantzios <achill@matrix.gatewaynet.com>",
"msg_from_op": false,
"msg_subject": "Re: [SQL] Problem: Referential Integrity Constraints lost"
}
] |
[
{
"msg_contents": "\n> The question is *which* seek APIs we need to support. Are there any\n> besides fseeko() and fgetpos()?\n\nOn AIX we have \nint fseeko64 (FILE* Stream, off64_t Offset, int Whence);\nwhich is intended for large file access for programs that do NOT\n#define _LARGE_FILES\n\nIt is functionality that is available if _LARGE_FILE_API is defined,\nwhich is the default if _LARGE_FILES is not defined.\n\nThat would have been my preferred way of handling large files on AIX\nin the two/three? places that need it (pg_dump/restore, psql and backend COPY).\nThis would have had the advantage that off_t is not 64 bit in all other places\nwhere it is actually not needed, no ?\n\nAndreas\n",
"msg_date": "Thu, 24 Oct 2002 11:23:13 +0200",
"msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>",
"msg_from_op": true,
"msg_subject": "Re: pg_dump and large files - is this a problem? "
},
{
"msg_contents": "Zeugswetter Andreas SB SD wrote:\n> \n> > The question is *which* seek APIs we need to support. Are there any\n> > besides fseeko() and fgetpos()?\n> \n> On AIX we have \n> int fseeko64 (FILE* Stream, off64_t Offset, int Whence);\n> which is intended for large file access for programs that do NOT\n> #define _LARGE_FILES\n> \n> It is functionality that is available if _LARGE_FILE_API is defined,\n> which is the default if _LARGE_FILES is not defined.\n> \n> That would have been my preferred way of handling large files on AIX\n> in the two/three? places that need it (pg_dump/restore, psql and backend COPY).\n> This would have had the advantage that off_t is not 64 bit in all other places\n> where it is actually not needed, no ?\n\nOK, I am focusing on AIX now. I don't think we can go down the road of\nsaying where large file support is needed or not needed. I think for\neach platform either we support large files or we don't. Is there a way\nto have off_t be 64 bits everywhere, and if it is, why wouldn't we just\nenable that rather than poke around figuring out where it is needed?\n\nAlso, I have the open item:\n\n\tFix AIX + Large File + Flex problem\n\nIs there an AIX problem with Flex?\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, 24 Oct 2002 21:38:00 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and large files - is this a problem?"
},
{
"msg_contents": "At 09:38 PM 24/10/2002 -0400, Bruce Momjian wrote:\n>OK, I am focusing on AIX now. I don't think we can go down the road of\n>saying where large file support is needed or not needed. I think for\n>each platform either we support large files or we don't.\n\nRather than having a different patch file for each platform and refusing to \ncode fseek/tell because we can't do SEEK_CUR, why not check for FSEEKO64 \nand revert to a simple solution:\n\n#ifdef HAVE_FSEEKO64\n#define FSEEK fseeko64\n#define FTELL ftello64\n#define FILE_OFFSET off64_t\n#else\n#ifdef HAVE_FSEEKO\n#define FSEEK fseeko\n#define FTELL ftello\n#define FILE_OFFSET off_t\n#else\n#if HAVE_FSEEK_BETTER_THAN_32_BIT\n#define FSEEK FSEEK_BETTER_THAN_32_BIT\n#define FTELL FTELL_BETTER_THAN_32_BIT\n#define FILE_OFFSET FILE_OFFSET_BETTER_THAN_32_BIT\n#else\n#if sizeof(off_t) > sizeof(long)\n#define IGNORE_FSEEK\n#else\n#define FSEEK fseek\n#define FTELL ftell\n#define FILE_OFFSET long\n#end if...\n\nThen use a correct checkSeek which also checks IGNORE_FSEEK.\n\nAFAICT, this *will* do the job on all systems discussed. And we can \ncertainly skip the HAVE_FSEEK_BETTER_THAN_32_BIT bit, but coding a trivial \nseek/tell pair for fsetpos/fgetpos is easy, even in a macro.\n\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n\n",
"msg_date": "Fri, 25 Oct 2002 12:10:11 +1000",
"msg_from": "Philip Warner <pjw@rhyme.com.au>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and large files - is this a problem?"
},
{
"msg_contents": "Philip Warner wrote:\n> Rather than having a different patch file for each platform and refusing to \n> code fseek/tell because we can't do SEEK_CUR, why not check for FSEEKO64 \n> and revert to a simple solution:\n> \n> #ifdef HAVE_FSEEKO64\n> #define FSEEK fseeko64\n> #define FTELL ftello64\n> #define FILE_OFFSET off64_t\n\nWe can do this, but there is the problem of making the code pretty ugly.\nAlso, it is not immediately clear when off_t is something to be used by\nfseek and when it is being used in file offsets that will never be\nseeked. I am concerned about perhaps making things worse than they are\nnow.\n\n> #else\n> #ifdef HAVE_FSEEKO\n> #define FSEEK fseeko\n> #define FTELL ftello\n> #define FILE_OFFSET off_t\n> #else\n> #if HAVE_FSEEK_BETTER_THAN_32_BIT\n> #define FSEEK FSEEK_BETTER_THAN_32_BIT\n> #define FTELL FTELL_BETTER_THAN_32_BIT\n> #define FILE_OFFSET FILE_OFFSET_BETTER_THAN_32_BIT\n> #else\n> #if sizeof(off_t) > sizeof(long)\n\nCan't do sizeof() tests in cpp, which is where the #if is processed.\n\n> #define IGNORE_FSEEK\n> #else\n> #define FSEEK fseek\n> #define FTELL ftell\n> #define FILE_OFFSET long\n> #end if...\n> \n> Then use a correct checkSeek which also checks IGNORE_FSEEK.\n> \n> AFAICT, this *will* do the job on all systems discussed. And we can \n> certainly skip the HAVE_FSEEK_BETTER_THAN_32_BIT bit, but coding a trivial \n> seek/tell pair for fsetpos/fgetpos is easy, even in a macro.\n\nI don't think we can assume that off_t can be passed to fset/getpos\nunless we know the platform supports it, unless people think fpos_t\nbeing integral and the same size as fpos_t is enough.\n\nAlso, I don't think these can be done a macro, perhaps\nfseeko(...,SEEK_SET), but not the others, and not ftello. See\nport/fseeko.c for the reason.\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, 25 Oct 2002 00:07:37 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and large files - is this a problem?"
},
{
"msg_contents": "At 12:07 AM 25/10/2002 -0400, Bruce Momjian wrote:\n>I don't think we can assume that off_t can be passed to fset/getpos\n>unless we know the platform supports it, unless people think fpos_t\n>being integral and the same size as fpos_t is enough.\n\nWe don't need to. We would #define FILE_OFFSET as fpos_t in that case.\n\n>Also, I don't think these can be done a macro, perhaps\n>fseeko(...,SEEK_SET), but not the others, and not ftello. See\n>port/fseeko.c for the reason.\n\nMy understanding was that you could define a block and declare variables in \na macro; just use a local for the temp storage of the in/out args. If this \nis the only thing stopping you adopting this approach, then I am very happy \nto try to code the macros properly.\n\nHowever, I do get the impression that there is more resistance to the idea \nthan just this.\n\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n\n",
"msg_date": "Fri, 25 Oct 2002 15:39:11 +1000",
"msg_from": "Philip Warner <pjw@rhyme.com.au>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and large files - is this a problem?"
}
] |
[
{
"msg_contents": "Thomas & Tom,\n\nThanks for all your help, I have updated all the defaults to use timestamptz\nand everything is working fine again.\n\nWe can't wait for 7.3, keep up the good work.\n\n\nJohn\n\n\n-----Original Message-----\nFrom: Tom Lane [mailto:tgl@sss.pgh.pa.us] \nSent: 23 October 2002 19:47\nTo: Hosen, John\nCc: 'Thomas Lockhart'; 'pgsql-hackers@postgresql.org'\nSubject: Re: [HACKERS] 'epoch'::timestamp and Daylight Savings \n\n\n\"Hosen, John\" <John.Hosen@capita.co.uk> writes:\n\n> e_app_print_date | timestamp with time zone | default\n> \"timestamp\"('epoch'::text)\n\nYeah, there's your problem. You are casting 'epoch' to type timestamp\nwithout time zone, and thence to timestamp with time zone. The first step\ngives \"midnight\" and the second assumes that that means \"midnight local\ntime\".\n\nFor example, in US EST zone I get:\n\nregression=# select \"timestamp\"('epoch'::text);\n timestamp\n---------------------\n 1970-01-01 00:00:00\n(1 row)\n\nregression=# select (\"timestamp\"('epoch'::text))::timestamptz;\n timestamp\n------------------------\n 1970-01-01 00:00:00-05\n(1 row)\n\nwhereas what is wanted is\n\nregression=# select \"timestamptz\"('epoch'::text);\n timestamptz\n------------------------\n 1969-12-31 19:00:00-05\n(1 row)\n\nSo you can fix the problem just by setting the default to be\n'epoch'::timestamptz.\n\nThe problem is probably related to the renaming we've been carrying out to\nget closer to the SQL spec: \"timestamp\" now means timestamp without time\nzone, which is not what it meant in older Postgres releases.\n\n\t\t\tregards, tom lane\n\nThis email has been scanned for all viruses by the MessageLabs SkyScan\nservice.\n\n\n***********************************************************************\n\nThis email and any files attached to it are confidential and intended \nsolely for the use of the individual or entity to whom they are\naddressed. If you have received this email in error please notify \nthe system manager.\n\nThe message and any files attached to it have been scanned by \nMIMEsweeper with Sophos Sweep and found to be free from all known\nviruses.\n\nInformation on MIMEsweeper can be found at http://www.mimesweeper.com/\n\n***********************************************************************\n\n",
"msg_date": "Thu, 24 Oct 2002 13:19:02 +0100",
"msg_from": "\"Hosen, John\" <John.Hosen@capita.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: 'epoch'::timestamp and Daylight Savings "
}
] |
[
{
"msg_contents": "I'm looking at different ways of optimizing queries with a large number of\njoins. I write the same query in a number of different ways and compare\nthe running times. Now the problem is I do not want the optimizer\nchanging the queries. So I explicit state the order of the joins in the\nFROM clause. I also turn off everything I can except for one type of join\n(say hash join), and I've turned off geqo. But I find that the PLANNER\nstill takes an enormous amount of time for some queries. It doesn't look\nlike the Planner is actually optimizing (changing) anything, but just in\ncase, I was wondering if there was a way to turn off the PLANNER.\n\nNote, when I say an enormous amount of time, I mean at least double the\ntime the EXECUTOR takes to actually answer the query.\n\nThanks for your help,\n\nBen McMahan\n\nps. here is a small example of what my queries look like (so you can see\nif there is something else it might be deciding on):\n\nSELECT DISTINCT c0.x1 , c1.x2 , c0.x3 , c0.x4 , c2.x5\nFROM r1 c4 (x4,x2,x5) JOIN (r0 c3 (x2,x3,x5) JOIN (r2 c2 (x3,x1,x5) JOIN (r1 c1\n(x4,x1,x2) JOIN r1 c0 (x1,x3,x4)\nON ( c0.x4 = c1.x4 AND c0.x1 = c1.x1 ))\nON ( c0.x3 = c2.x3 AND c0.x1 = c2.x1 ))\nON ( c1.x2 = c3.x2 AND c0.x3 = c3.x3 AND c2.x5 = c3.x5 ))\nON ( c0.x4 = c4.x4 AND c1.x2 = c4.x2 AND c2.x5 = c4.x5 );\n\nA quick reminder, FROM r1 c4 (x4,x2,x5) just renames a table r1 into c4\nwhere it also renames the columns to x4, x2, and x5 respectively.\n\n",
"msg_date": "Thu, 24 Oct 2002 15:09:00 -0500 (CDT)",
"msg_from": "Ben McMahan <mcmahanb@cs.rice.edu>",
"msg_from_op": true,
"msg_subject": "Turning the PLANNER off"
},
{
"msg_contents": " \nThat is a good question. The planner does more than just analyse the\nquery. It generates the Plan used by the executor, so that can't be\nremoved.\n\nIt is always a pain when the optimizer/planner takes longer than the\nexecutor. We do have PREPARE/EXECUTE in 7.3beta for you to use.\n\n\n---------------------------------------------------------------------------\n\nBen McMahan wrote:\n> I'm looking at different ways of optimizing queries with a large number of\n> joins. I write the same query in a number of different ways and compare\n> the running times. Now the problem is I do not want the optimizer\n> changing the queries. So I explicit state the order of the joins in the\n> FROM clause. I also turn off everything I can except for one type of join\n> (say hash join), and I've turned off geqo. But I find that the PLANNER\n> still takes an enormous amount of time for some queries. It doesn't look\n> like the Planner is actually optimizing (changing) anything, but just in\n> case, I was wondering if there was a way to turn off the PLANNER.\n> \n> Note, when I say an enormous amount of time, I mean at least double the\n> time the EXECUTOR takes to actually answer the query.\n> \n> Thanks for your help,\n> \n> Ben McMahan\n> \n> ps. here is a small example of what my queries look like (so you can see\n> if there is something else it might be deciding on):\n> \n> SELECT DISTINCT c0.x1 , c1.x2 , c0.x3 , c0.x4 , c2.x5\n> FROM r1 c4 (x4,x2,x5) JOIN (r0 c3 (x2,x3,x5) JOIN (r2 c2 (x3,x1,x5) JOIN (r1 c1\n> (x4,x1,x2) JOIN r1 c0 (x1,x3,x4)\n> ON ( c0.x4 = c1.x4 AND c0.x1 = c1.x1 ))\n> ON ( c0.x3 = c2.x3 AND c0.x1 = c2.x1 ))\n> ON ( c1.x2 = c3.x2 AND c0.x3 = c3.x3 AND c2.x5 = c3.x5 ))\n> ON ( c0.x4 = c4.x4 AND c1.x2 = c4.x2 AND c2.x5 = c4.x5 );\n> \n> A quick reminder, FROM r1 c4 (x4,x2,x5) just renames a table r1 into c4\n> where it also renames the columns to x4, x2, and x5 respectively.\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": "Mon, 28 Oct 2002 19:55:02 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Turning the PLANNER off"
},
{
"msg_contents": "Does PREPARE turn \nselect * from mytable \ninto\nselect mytable.field1,mytable.field2 from mynamespace.mytable\n?\n\nI was looking for this functionality for one of my projects so I'm curious.\n\n\nOn Monday 28 October 2002 06:55 pm, (Via wrote:\n> That is a good question. The planner does more than just analyse the\n> query. It generates the Plan used by the executor, so that can't be\n> removed.\n>\n> It is always a pain when the optimizer/planner takes longer than the\n> executor. We do have PREPARE/EXECUTE in 7.3beta for you to use.\n>\n>\n> ---------------------------------------------------------------------------\n>\n> Ben McMahan wrote:\n> > I'm looking at different ways of optimizing queries with a large number\n> > of joins. I write the same query in a number of different ways and\n> > compare the running times. Now the problem is I do not want the\n> > optimizer changing the queries. So I explicit state the order of the\n> > joins in the FROM clause. I also turn off everything I can except for\n> > one type of join (say hash join), and I've turned off geqo. But I find\n> > that the PLANNER still takes an enormous amount of time for some queries.\n> > It doesn't look like the Planner is actually optimizing (changing)\n> > anything, but just in case, I was wondering if there was a way to turn\n> > off the PLANNER.\n> >\n> > Note, when I say an enormous amount of time, I mean at least double the\n> > time the EXECUTOR takes to actually answer the query.\n> >\n> > Thanks for your help,\n> >\n> > Ben McMahan\n> >\n> > ps. here is a small example of what my queries look like (so you can see\n> > if there is something else it might be deciding on):\n> >\n> > SELECT DISTINCT c0.x1 , c1.x2 , c0.x3 , c0.x4 , c2.x5\n> > FROM r1 c4 (x4,x2,x5) JOIN (r0 c3 (x2,x3,x5) JOIN (r2 c2 (x3,x1,x5) JOIN\n> > (r1 c1 (x4,x1,x2) JOIN r1 c0 (x1,x3,x4)\n> > ON ( c0.x4 = c1.x4 AND c0.x1 = c1.x1 ))\n> > ON ( c0.x3 = c2.x3 AND c0.x1 = c2.x1 ))\n> > ON ( c1.x2 = c3.x2 AND c0.x3 = c3.x3 AND c2.x5 = c3.x5 ))\n> > ON ( c0.x4 = c4.x4 AND c1.x2 = c4.x2 AND c2.x5 = c4.x5 );\n> >\n> > A quick reminder, FROM r1 c4 (x4,x2,x5) just renames a table r1 into c4\n> > where it also renames the columns to x4, x2, and x5 respectively.\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",
"msg_date": "Tue, 29 Oct 2002 08:47:45 -0600",
"msg_from": "David Walker <pgsql@grax.com>",
"msg_from_op": false,
"msg_subject": "Re: Turning the PLANNER off"
},
{
"msg_contents": "David Walker wrote:\n> Does PREPARE turn \n> select * from mytable \n> into\n> select mytable.field1,mytable.field2 from mynamespace.mytable\n> ?\n> \n> I was looking for this functionality for one of my projects so I'm curious.\n\nPREPARE turns the query into an internal format used by the executor.\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, 29 Oct 2002 13:23:10 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Turning the PLANNER off"
},
{
"msg_contents": "Ah, so Ben finally got around to posting here. Ben's a CS Grad student\nhere at Rice. His (current) project involves taking some interesting\nresults from constraint satisfaction and implementing them on a database:\none of the CS faculty has demonstrated that one class of highly joined\nDB queries maps to a solved problem in constraint satisfaction. The end\ngoal would be an optimizer module or setting that recognizes this class\nof query, and spits out a mathematically optimized join order.\n\nSo, in the interim, Ben's trying to do it by hand: preorder the joins\nand demonstrate that the 'best' order is in fact the best. Then move on\nto looking into integrating this, if possible: part of the problem is\nrecognizing the structure of the query, of course. Right now, the planner\nis getting in the way - although he can extract the needed timing info,\nhe's wasting CPU cycles planning things that don't need it, limiting\nthe number of cases he can try.\n\nAs mentioned before, there's currently no interface to feed in a Plan,\nso he's out of luck. Is there, programmatically, a way to do it?\nSerialize a plan tree to a file, and feed it in latter, purely for\ndevelopment purposes. How painful would that be? Should I send him in to\nsee if he can implement one quickly, or are there dragons hiding in there?\n\nRoss\n\nOn Mon, Oct 28, 2002 at 07:55:02PM -0500, Bruce Momjian wrote:\n> \n> That is a good question. The planner does more than just analyse the\n> query. It generates the Plan used by the executor, so that can't be\n> removed.\n> \n> It is always a pain when the optimizer/planner takes longer than the\n> executor. We do have PREPARE/EXECUTE in 7.3beta for you to use.\n> \n> \n> ---------------------------------------------------------------------------\n> \n> Ben McMahan wrote:\n> > I'm looking at different ways of optimizing queries with a large number of\n> > joins. I write the same query in a number of different ways and compare\n> > the running times. Now the problem is I do not want the optimizer\n> > changing the queries. So I explicit state the order of the joins in the\n> > FROM clause. I also turn off everything I can except for one type of join\n> > (say hash join), and I've turned off geqo. But I find that the PLANNER\n> > still takes an enormous amount of time for some queries. It doesn't look\n> > like the Planner is actually optimizing (changing) anything, but just in\n> > case, I was wondering if there was a way to turn off the PLANNER.\n> > \n> > Note, when I say an enormous amount of time, I mean at least double the\n> > time the EXECUTOR takes to actually answer the query.\n> > \n> > Thanks for your help,\n> > \n> > Ben McMahan\n> > \n> > ps. here is a small example of what my queries look like (so you can see\n> > if there is something else it might be deciding on):\n> > \n> > SELECT DISTINCT c0.x1 , c1.x2 , c0.x3 , c0.x4 , c2.x5\n> > FROM r1 c4 (x4,x2,x5) JOIN (r0 c3 (x2,x3,x5) JOIN (r2 c2 (x3,x1,x5) JOIN (r1 c1\n> > (x4,x1,x2) JOIN r1 c0 (x1,x3,x4)\n> > ON ( c0.x4 = c1.x4 AND c0.x1 = c1.x1 ))\n> > ON ( c0.x3 = c2.x3 AND c0.x1 = c2.x1 ))\n> > ON ( c1.x2 = c3.x2 AND c0.x3 = c3.x3 AND c2.x5 = c3.x5 ))\n> > ON ( c0.x4 = c4.x4 AND c1.x2 = c4.x2 AND c2.x5 = c4.x5 );\n> > \n> > A quick reminder, FROM r1 c4 (x4,x2,x5) just renames a table r1 into c4\n> > where it also renames the columns to x4, x2, and x5 respectively.\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> ---------------------------(end of broadcast)---------------------------\n> TIP 5: Have you checked our extensive FAQ?\n> \n> http://www.postgresql.org/users-lounge/docs/faq.html\n",
"msg_date": "Wed, 30 Oct 2002 17:24:32 -0600",
"msg_from": "\"Ross J. Reedstrom\" <reedstrm@rice.edu>",
"msg_from_op": false,
"msg_subject": "Re: Turning the PLANNER off"
},
{
"msg_contents": "\"Ross J. Reedstrom\" <reedstrm@rice.edu> writes:\n> So, in the interim, Ben's trying to do it by hand: preorder the joins\n> and demonstrate that the 'best' order is in fact the best. Then move on\n> to looking into integrating this, if possible: part of the problem is\n> recognizing the structure of the query, of course. Right now, the planner\n> is getting in the way - although he can extract the needed timing info,\n> he's wasting CPU cycles planning things that don't need it, limiting\n> the number of cases he can try.\n\nWhat's the basis for your assertion that it's \"planning things that\ndon't need it\"? Given a JOIN-constrained query I do not believe the\nplanner will look at any cases other than the intended join order.\n\n> As mentioned before, there's currently no interface to feed in a Plan,\n> so he's out of luck. Is there, programmatically, a way to do it?\n> Serialize a plan tree to a file, and feed it in latter, purely for\n> development purposes. How painful would that be? Should I send him in to\n> see if he can implement one quickly, or are there dragons hiding in there?\n\nHe can do whatever he wants, as long as he has no illusions about\ngetting it accepted back into the sources ;-).\n\nWhat would probably be more useful is to do some profiling to understand\nwhy the planner is taking longer than he wants even with a\nJOIN-constrained query. I should think this would be pretty quick.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 30 Oct 2002 22:59:39 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Turning the PLANNER off "
},
{
"msg_contents": "On Wed, Oct 30, 2002 at 10:59:39PM -0500, Tom Lane wrote:\n> \n> What's the basis for your assertion that it's \"planning things that\n> don't need it\"? Given a JOIN-constrained query I do not believe the\n> planner will look at any cases other than the intended join order.\n\nWell, that was a loose choice of words - let's say the planner seems to\nbe taking awfully long to build an execution tree with only one choice\navailable.\n\n> He can do whatever he wants, as long as he has no illusions about\n> getting it accepted back into the sources ;-).\n\nUnderstood - this would be a hacking tool only.\n\n> What would probably be more useful is to do some profiling to understand\n> why the planner is taking longer than he wants even with a\n> JOIN-constrained query. I should think this would be pretty quick.\n\nYup, that targets the same question as above - anything 'uneeded' actually\nhappening in the planner? I'll send him off with this suggestion.\n\nRoss\n",
"msg_date": "Thu, 31 Oct 2002 12:34:54 -0600",
"msg_from": "\"Ross J. Reedstrom\" <reedstrm@rice.edu>",
"msg_from_op": false,
"msg_subject": "Re: Turning the PLANNER off"
}
] |
[
{
"msg_contents": "Hello,\n\nIn the docs it is mentioned for datistemplate that\n\n\"If true then this database can be used in the \"TEMPLATE\" clause of\nCREATE DATABASE to create the new database as a clone of this one.\"\n\nHowever, one can create a database using as template another DB that has\ndatistemplate set to false.\n\nIn this situation, what is the point of having datistemplate?\n\n-- \nAlvaro Herrera (<alvherre[a]dcc.uchile.cl>)\n\"Now I have my system running, not a byte was off the shelf;\nIt rarely breaks and when it does I fix the code myself.\nIt's stable, clean and elegant, and lightning fast as well,\nAnd it doesn't cost a nickel, so Bill Gates can go to hell.\"\n",
"msg_date": "Thu, 24 Oct 2002 17:12:10 -0300",
"msg_from": "Alvaro Herrera <alvherre@dcc.uchile.cl>",
"msg_from_op": true,
"msg_subject": "pg_database datistemplate"
},
{
"msg_contents": "\n>In the docs it is mentioned for datistemplate that\n>\n>\"If true then this database can be used in the \"TEMPLATE\" clause of\n>CREATE DATABASE to create the new database as a clone of this one.\"\n>\n>However, one can create a database using as template another DB that has\n>datistemplate set to false.\n>\n>In this situation, what is the point of having datistemplate?\n\nIf datistemplate is true, any user with CREATEDB privileges can clone the \ndatabase.\nIf datistemplate is false, only a cluster superuser can clone the database.\n\n",
"msg_date": "Thu, 24 Oct 2002 16:29:53 -0400",
"msg_from": "korry <korry@starband.net>",
"msg_from_op": false,
"msg_subject": "Re: pg_database datistemplate"
},
{
"msg_contents": "Alvaro Herrera <alvherre@dcc.uchile.cl> writes:\n> In the docs it is mentioned for datistemplate that\n> \"If true then this database can be used in the \"TEMPLATE\" clause of\n> CREATE DATABASE to create the new database as a clone of this one.\"\n\nRight.\n\n> However, one can create a database using as template another DB that has\n> datistemplate set to false.\n\nOnly if one is owner of the source database (or superuser).\n\n> In this situation, what is the point of having datistemplate?\n\nEffectively, it's a \"grant right to copy\" flag.\n\nNow that we have per-database ACLs, we should probably replace\ndatistemplate with an access right; instead of setting it you'd\ndo something like GRANT COPY ON DATABASE foo TO PUBLIC.\n\n(We'd also talked about replacing datallowconn with an access right,\nalthough that is more likely to break existing apps, since a fair\nnumber of them look at datallowconn.)\n\nToo late for 7.3 though ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 24 Oct 2002 16:39:51 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: pg_database datistemplate "
},
{
"msg_contents": "On Thu, Oct 24, 2002 at 04:39:51PM -0400, Tom Lane wrote:\n> Alvaro Herrera <alvherre@dcc.uchile.cl> writes:\n> > In the docs it is mentioned for datistemplate that\n> \n> > However, one can create a database using as template another DB that has\n> > datistemplate set to false.\n> \n> Only if one is owner of the source database (or superuser).\n\nOh, I see. This is a doc bug, isn't it? I will submit a patch for\nthis. I think I've seen other oversights; will try to keep note of\nthem.\n\n> Now that we have per-database ACLs, we should probably replace\n> datistemplate with an access right; instead of setting it you'd\n> do something like GRANT COPY ON DATABASE foo TO PUBLIC.\n\nSounds good. Altering system catalogs directly is \"a bad thing\", IMHO.\n\n> (We'd also talked about replacing datallowconn with an access right,\n> although that is more likely to break existing apps, since a fair\n> number of them look at datallowconn.)\n\nMaybe keep both for a release, and deprecate datallowconn?\n\nAnyway, there are a number of minor things that could use ALTER <foo>\nsupport. I'll try to make a note of those too, and fix them if I can.\n\n-- \nAlvaro Herrera (<alvherre[a]dcc.uchile.cl>)\n\"La conclusion que podemos sacar de esos estudios es que\nno podemos sacar ninguna conclusion de ellos\" (Tanenbaum)\n",
"msg_date": "Thu, 24 Oct 2002 19:48:46 -0300",
"msg_from": "Alvaro Herrera <alvherre@dcc.uchile.cl>",
"msg_from_op": true,
"msg_subject": "Re: pg_database datistemplate"
},
{
"msg_contents": "Alvaro Herrera <alvherre@dcc.uchile.cl> writes:\n> On Thu, Oct 24, 2002 at 04:39:51PM -0400, Tom Lane wrote:\n>> Alvaro Herrera <alvherre@dcc.uchile.cl> writes:\n>>> However, one can create a database using as template another DB that has\n>>> datistemplate set to false.\n>> \n>> Only if one is owner of the source database (or superuser).\n\n> Oh, I see. This is a doc bug, isn't it?\n\nIt's mentioned somewhere, but perhaps not where you were looking ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 24 Oct 2002 19:15:53 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: pg_database datistemplate "
}
] |
[
{
"msg_contents": "I am wondering whether my recent hack in DetermineLocalTimeZone has\nremoved all need for NO_MKTIME_BEFORE_1970. I see that the AIX and\nIRIX5 port headers define that symbol. Can anyone check regression\ntest results for 7.2.3 or 7.3betaN on these platforms to see whether\nwe could/should remove it?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 24 Oct 2002 16:14:52 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": true,
"msg_subject": "Do we still need NO_MKTIME_BEFORE_1970 ?"
}
] |
[
{
"msg_contents": "\ng'day ...\n\n\tjust went through the new config files for v7.3, to make sure, but\nit doens't look like we have such ... has anyone looked at adding a 'idle\ntimeout' for a postgres process? Or am I missing something in the docs?\n\n\n\n",
"msg_date": "Fri, 25 Oct 2002 00:00:37 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "idle connection timeout ... "
},
{
"msg_contents": "\"Marc G. Fournier\" <scrappy@hub.org> writes:\n> \tjust went through the new config files for v7.3, to make sure, but\n> it doens't look like we have such ... has anyone looked at adding a 'idle\n> timeout' for a postgres process? Or am I missing something in the docs?\n\nAre you looking for the backend to arbitrarily disconnect from a client\nthat hasn't done anything in X amount of time? Seems to me that has\nbeen proposed and rejected, more than once.\n\nWe already have logic that checks for loss of connectivity (see TCP\nkeepalive option). If the client is *still there*, but has just not\nchosen to issue any commands lately, I have a very hard time buying\nany argument that it is the backend's province to abort the connection.\nThat's a recipe for degrading reliability, not improving it.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 24 Oct 2002 23:27:05 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "On Thu, 24 Oct 2002, Tom Lane wrote:\n\n> \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > \tjust went through the new config files for v7.3, to make sure, but\n> > it doens't look like we have such ... has anyone looked at adding a 'idle\n> > timeout' for a postgres process? Or am I missing something in the docs?\n>\n> Are you looking for the backend to arbitrarily disconnect from a client\n> that hasn't done anything in X amount of time? Seems to me that has\n> been proposed and rejected, more than once.\n>\n> We already have logic that checks for loss of connectivity (see TCP\n> keepalive option). If the client is *still there*, but has just not\n> chosen to issue any commands lately, I have a very hard time buying\n> any argument that it is the backend's province to abort the connection.\n> That's a recipe for degrading reliability, not improving it.\n\nYa, I've thought that one through ... I think what I'm more looking at is\nsome way of 'limiting' persistent connections, where a server opens n\nconnections during a spike, which then sit idle indefinitely since it was\none fo those 'slashdot effect' kinda spikes ...\n\nIs there any way of the 'master process' *safely/accurately* knowing,\nthrough the shared memory link, the # of connections currently open to a\nparticular database? So that a limit could be set on a per db basis, say\nas an additional arg to pg_hba.conf?\n\n\n\n",
"msg_date": "Fri, 25 Oct 2002 02:52:53 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "On Fri, 2002-10-25 at 00:52, Marc G. Fournier wrote:\n> Ya, I've thought that one through ... I think what I'm more looking at is\n> some way of 'limiting' persistent connections, where a server opens n\n> connections during a spike, which then sit idle indefinitely since it was\n> one fo those 'slashdot effect' kinda spikes ...\n> \n> Is there any way of the 'master process' *safely/accurately* knowing,\n> through the shared memory link, the # of connections currently open to a\n> particular database? So that a limit could be set on a per db basis, say\n> as an additional arg to pg_hba.conf?\n\nWell, if you're application is smart enough to know it needs to\ndynamically add connections, it should also be smart enough to tear them\ndown after some idle period. I agree with Tom. I think that sounds\nlike application domain.\n\nGreg\n\n\n",
"msg_date": "25 Oct 2002 08:47:42 -0500",
"msg_from": "Greg Copeland <greg@CopelandConsulting.Net>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "\"Marc G. Fournier\" <scrappy@hub.org> writes:\n> Is there any way of the 'master process' *safely/accurately* knowing,\n> through the shared memory link, the # of connections currently open to a\n> particular database? So that a limit could be set on a per db basis, say\n> as an additional arg to pg_hba.conf?\n\nIt would be better/easier to apply the check later on, when a backend is\nadding itself to the PGPROC array. It'd be easy enough to count the\nnumber of other backends showing the same DB OID in their PGPROC\nentries, and reject if too many.\n\n\t\t\tregards, tom lane\n\n\n\n",
"msg_date": "Fri, 25 Oct 2002 10:00:10 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "Greg Copeland <greg@CopelandConsulting.Net> writes:\n> On Fri, 2002-10-25 at 00:52, Marc G. Fournier wrote:\n>> Ya, I've thought that one through ... I think what I'm more looking at is\n>> some way of 'limiting' persistent connections, where a server opens n\n>> connections during a spike, which then sit idle indefinitely since it was\n>> one fo those 'slashdot effect' kinda spikes ...\n>> \n>> Is there any way of the 'master process' *safely/accurately* knowing,\n>> through the shared memory link, the # of connections currently open to a\n>> particular database? So that a limit could be set on a per db basis, say\n>> as an additional arg to pg_hba.conf?\n\n> Well, if you're application is smart enough to know it needs to\n> dynamically add connections, it should also be smart enough to tear them\n> down after some idle period. I agree with Tom. I think that sounds\n> like application domain.\n\nWell, there are two different things here. I agree that if an app\nis going to use persistent connections, it should be the app's\nresponsibility to manage them. But a per-database, as opposed to\ninstallation-wide, limit on number of connections seems like a\nreasonable idea. Note that the limit would result in new connections\nbeing rejected, not old ones being summarily cut.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 25 Oct 2002 10:40:13 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "Tom Lane wrote:\n> Greg Copeland <greg@CopelandConsulting.Net> writes:\n> > On Fri, 2002-10-25 at 00:52, Marc G. Fournier wrote:\n> >> Ya, I've thought that one through ... I think what I'm more looking at is\n> >> some way of 'limiting' persistent connections, where a server opens n\n> >> connections during a spike, which then sit idle indefinitely since it was\n> >> one fo those 'slashdot effect' kinda spikes ...\n> >> \n> >> Is there any way of the 'master process' *safely/accurately* knowing,\n> >> through the shared memory link, the # of connections currently open to a\n> >> particular database? So that a limit could be set on a per db basis, say\n> >> as an additional arg to pg_hba.conf?\n> \n> > Well, if you're application is smart enough to know it needs to\n> > dynamically add connections, it should also be smart enough to tear them\n> > down after some idle period. I agree with Tom. I think that sounds\n> > like application domain.\n> \n> Well, there are two different things here. I agree that if an app\n> is going to use persistent connections, it should be the app's\n> responsibility to manage them. But a per-database, as opposed to\n> installation-wide, limit on number of connections seems like a\n> reasonable idea. Note that the limit would result in new connections\n> being rejected, not old ones being summarily cut.\n\nBut then the app is going to keep trying to connect over and over unless\nit knows something about why it can't connect.\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, 25 Oct 2002 10:47:12 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n>> Well, there are two different things here. I agree that if an app\n>> is going to use persistent connections, it should be the app's\n>> responsibility to manage them. But a per-database, as opposed to\n>> installation-wide, limit on number of connections seems like a\n>> reasonable idea. Note that the limit would result in new connections\n>> being rejected, not old ones being summarily cut.\n\n> But then the app is going to keep trying to connect over and over unless\n> it knows something about why it can't connect.\n\nSo? If it hits the installation-wide limit, you'll have the same\nproblem; and at that point the (presumably runaway) app would have\nsucked up all the connections, denying service to other apps using other\ndatabases. I think Marc's point here is to limit his exposure to\nmisbehavior of any one client app, in a database server that is serving\nmultiple clients using multiple databases.\n\nIt occurs to me that a per-user connection limit is going to be the next\nthing he asks for ;-). We could implement that too, if we wanted.\n(Not sure whether PGPROC stores the user id, but it easily could.)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 25 Oct 2002 11:02:48 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> >> Well, there are two different things here. I agree that if an app\n> >> is going to use persistent connections, it should be the app's\n> >> responsibility to manage them. But a per-database, as opposed to\n> >> installation-wide, limit on number of connections seems like a\n> >> reasonable idea. Note that the limit would result in new connections\n> >> being rejected, not old ones being summarily cut.\n> \n> > But then the app is going to keep trying to connect over and over unless\n> > it knows something about why it can't connect.\n> \n> So? If it hits the installation-wide limit, you'll have the same\n> problem; and at that point the (presumably runaway) app would have\n> sucked up all the connections, denying service to other apps using other\n> databases. I think Marc's point here is to limit his exposure to\n> misbehavior of any one client app, in a database server that is serving\n> multiple clients using multiple databases.\n\nWhat I am saying is that using the backend to throttle per-db\nconnections may not work too well because they will just keep retrying. \nI realize that the total limit can be hit too, but I assume that limit\nis set so it will not be hit (it's a resource tradeoff), while the\nper-db limit is there to try to throttle back the persistent\nconnections.\n\nBasically, total connections is to be set larger than you think you will\never need, while you expect per-db to be hit, and if something keeps\ntrying to connect and failing, we may get very bad connection\nperformance for other backends. This is where doing the limiting on the\npersistent connection end would be a better solution.\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, 25 Oct 2002 11:19:14 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Basically, total connections is to be set larger than you think you will\n> ever need, while you expect per-db to be hit, and if something keeps\n> trying to connect and failing, we may get very bad connection\n> performance for other backends.\n\nHmm, I see your point. A per-db limit *could* be useful even if it's\nset high enough that you don't expect it to be hit ... but most likely\npeople would try to use it in a way that it wouldn't be very efficient\ncompared to a client-side solution.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 25 Oct 2002 11:46:47 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Basically, total connections is to be set larger than you think you will\n> > ever need, while you expect per-db to be hit, and if something keeps\n> > trying to connect and failing, we may get very bad connection\n> > performance for other backends.\n> \n> Hmm, I see your point. A per-db limit *could* be useful even if it's\n> set high enough that you don't expect it to be hit ... but most likely\n> people would try to use it in a way that it wouldn't be very efficient\n> compared to a client-side solution.\n\nThe only way to do it would be, after a few hits of the limit, to start\ndelaying the connection rejections so you don't get hammered. It could\nbe done, but even then, I am not sure if it would be optimal.\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, 25 Oct 2002 11:52:36 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Tom Lane <tgl@sss.pgh.pa.us> wrote:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Basically, total connections is to be set larger than you think you will\n> > ever need, while you expect per-db to be hit, and if something keeps\n> > trying to connect and failing, we may get very bad connection\n> > performance for other backends.\n>\n> Hmm, I see your point. A per-db limit *could* be useful even if it's\n> set high enough that you don't expect it to be hit ... but most likely\n> people would try to use it in a way that it wouldn't be very efficient\n> compared to a client-side solution.\n\nWhat about a shared database server, where you want to have resource\nlimits for each database/user?\nCould be usefull in such a case, even if it is not very efficient, it\nwould be the only way. As dba you need not have control over the\nclient apps.\n\nJust a thought.\n\nRegards,\nMichael\n\n",
"msg_date": "Fri, 25 Oct 2002 18:07:10 +0200",
"msg_from": "\"Michael Paesold\" <mpaesold@gmx.at>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "On Fri, 25 Oct 2002, Tom Lane wrote:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> >> Well, there are two different things here. I agree that if an app\n> >> is going to use persistent connections, it should be the app's\n> >> responsibility to manage them. But a per-database, as opposed to\n> >> installation-wide, limit on number of connections seems like a\n> >> reasonable idea. Note that the limit would result in new connections\n> >> being rejected, not old ones being summarily cut.\n>\n> > But then the app is going to keep trying to connect over and over unless\n> > it knows something about why it can't connect.\n>\n> So? If it hits the installation-wide limit, you'll have the same\n> problem; and at that point the (presumably runaway) app would have\n> sucked up all the connections, denying service to other apps using other\n> databases. I think Marc's point here is to limit his exposure to\n> misbehavior of any one client app, in a database server that is serving\n> multiple clients using multiple databases.\n>\n> It occurs to me that a per-user connection limit is going to be the next\n> thing he asks for ;-)\n\nActually, sounds like a good idea, but have been trying to leave (and\nmove) multiple client auth to be within the database/application itself\n...\n\n",
"msg_date": "Fri, 25 Oct 2002 14:00:27 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "On Fri, 25 Oct 2002, Bruce Momjian wrote:\n\n> Tom Lane wrote:\n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > >> Well, there are two different things here. I agree that if an app\n> > >> is going to use persistent connections, it should be the app's\n> > >> responsibility to manage them. But a per-database, as opposed to\n> > >> installation-wide, limit on number of connections seems like a\n> > >> reasonable idea. Note that the limit would result in new connections\n> > >> being rejected, not old ones being summarily cut.\n> >\n> > > But then the app is going to keep trying to connect over and over unless\n> > > it knows something about why it can't connect.\n> >\n> > So? If it hits the installation-wide limit, you'll have the same\n> > problem; and at that point the (presumably runaway) app would have\n> > sucked up all the connections, denying service to other apps using other\n> > databases. I think Marc's point here is to limit his exposure to\n> > misbehavior of any one client app, in a database server that is serving\n> > multiple clients using multiple databases.\n>\n> What I am saying is that using the backend to throttle per-db\n> connections may not work too well because they will just keep retrying.\n\nOkay, but also bear in mind that alot of the time, when I'm bringign up\nstuff like this, I'm coming from the \"ISP\" perspective ... if I have one\nclient that is using up all 512 connections on the server, none of my\nother clients are getting any connections ...\n\nYes, the client should have tested his code better, but I want to be able\nto put 'limits' to make it so that everyone isn't affected by ones mistake\n...\n\n> I realize that the total limit can be hit too, but I assume that limit\n> is set so it will not be hit (it's a resource tradeoff), while the\n> per-db limit is there to try to throttle back the persistent\n> connections.\n\nNope, the per-db limit is there to try and eliminate the impact of one\nclient/application from essentially creating a DoS for all other\ndatabase/clients ...\n\n> Basically, total connections is to be set larger than you think you will\n> ever need, while you expect per-db to be hit, and if something keeps\n> trying to connect and failing, we may get very bad connection\n> performance for other backends. This is where doing the limiting on the\n> persistent connection end would be a better solution.\n\nAgreed, but unless you have control over both the client and server sides,\nits not possible ...\n\n\n",
"msg_date": "Fri, 25 Oct 2002 14:28:38 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "\nYes, my comments related to useing db/user limits to control the number\nof persistent connections. From an ISP perspective, I can see value in\nuser/db limits.\n\n---------------------------------------------------------------------------\n\nMarc G. Fournier wrote:\n> On Fri, 25 Oct 2002, Bruce Momjian wrote:\n> \n> > Tom Lane wrote:\n> > > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > >> Well, there are two different things here. I agree that if an app\n> > > >> is going to use persistent connections, it should be the app's\n> > > >> responsibility to manage them. But a per-database, as opposed to\n> > > >> installation-wide, limit on number of connections seems like a\n> > > >> reasonable idea. Note that the limit would result in new connections\n> > > >> being rejected, not old ones being summarily cut.\n> > >\n> > > > But then the app is going to keep trying to connect over and over unless\n> > > > it knows something about why it can't connect.\n> > >\n> > > So? If it hits the installation-wide limit, you'll have the same\n> > > problem; and at that point the (presumably runaway) app would have\n> > > sucked up all the connections, denying service to other apps using other\n> > > databases. I think Marc's point here is to limit his exposure to\n> > > misbehavior of any one client app, in a database server that is serving\n> > > multiple clients using multiple databases.\n> >\n> > What I am saying is that using the backend to throttle per-db\n> > connections may not work too well because they will just keep retrying.\n> \n> Okay, but also bear in mind that alot of the time, when I'm bringign up\n> stuff like this, I'm coming from the \"ISP\" perspective ... if I have one\n> client that is using up all 512 connections on the server, none of my\n> other clients are getting any connections ...\n> \n> Yes, the client should have tested his code better, but I want to be able\n> to put 'limits' to make it so that everyone isn't affected by ones mistake\n> ...\n> \n> > I realize that the total limit can be hit too, but I assume that limit\n> > is set so it will not be hit (it's a resource tradeoff), while the\n> > per-db limit is there to try to throttle back the persistent\n> > connections.\n> \n> Nope, the per-db limit is there to try and eliminate the impact of one\n> client/application from essentially creating a DoS for all other\n> database/clients ...\n> \n> > Basically, total connections is to be set larger than you think you will\n> > ever need, while you expect per-db to be hit, and if something keeps\n> > trying to connect and failing, we may get very bad connection\n> > performance for other backends. This is where doing the limiting on the\n> > persistent connection end would be a better solution.\n> \n> Agreed, but unless you have control over both the client and server sides,\n> its not possible ...\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": "Fri, 25 Oct 2002 13:31:24 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "On Fri, 25 Oct 2002, Tom Lane wrote:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Basically, total connections is to be set larger than you think you will\n> > ever need, while you expect per-db to be hit, and if something keeps\n> > trying to connect and failing, we may get very bad connection\n> > performance for other backends.\n>\n> Hmm, I see your point. A per-db limit *could* be useful even if it's\n> set high enough that you don't expect it to be hit ... but most likely\n> people would try to use it in a way that it wouldn't be very efficient\n> compared to a client-side solution.\n\nAs mentioned in my response to Bruce ... in an ISP situation, a DoS attack\nagainst the database by a single client can be very easy to accomplish in\nour current situation ... all I need to do is setup a perl script that\nopens all the connections I can to the database I have access to until all\nare used up, and nobody else has access to *their* databases ...\n\n",
"msg_date": "Fri, 25 Oct 2002 14:31:59 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "On Fri, 25 Oct 2002, Bruce Momjian wrote:\n\n> Tom Lane wrote:\n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > Basically, total connections is to be set larger than you think you will\n> > > ever need, while you expect per-db to be hit, and if something keeps\n> > > trying to connect and failing, we may get very bad connection\n> > > performance for other backends.\n> >\n> > Hmm, I see your point. A per-db limit *could* be useful even if it's\n> > set high enough that you don't expect it to be hit ... but most likely\n> > people would try to use it in a way that it wouldn't be very efficient\n> > compared to a client-side solution.\n>\n> The only way to do it would be, after a few hits of the limit, to start\n> delaying the connection rejections so you don't get hammered. It could\n> be done, but even then, I am not sure if it would be optimal.\n\nNote that I don't believe there is an \"optimal solution\" for this ... but\nin an environment where there are several clients connecting to several\ndifferent databases, the ability for one client to starve out the others\nis actually very real ...\n\n",
"msg_date": "Fri, 25 Oct 2002 14:34:04 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "On Fri, Oct 25, 2002 at 11:02:48AM -0400, Tom Lane wrote:\n> So? If it hits the installation-wide limit, you'll have the same\n> problem; and at that point the (presumably runaway) app would have\n> sucked up all the connections, denying service to other apps using other\n> databases. I think Marc's point here is to limit his exposure to\n> misbehavior of any one client app, in a database server that is serving\n> multiple clients using multiple databases.\n\nThat would indeed be a useful item. The only way to avoid such\nexposure right now is to run another back end.\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": "Fri, 25 Oct 2002 14:47:01 -0400",
"msg_from": "Andrew Sullivan <andrew@libertyrms.info>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Andrew Sullivan wrote:\n> On Fri, Oct 25, 2002 at 11:02:48AM -0400, Tom Lane wrote:\n> > So? If it hits the installation-wide limit, you'll have the same\n> > problem; and at that point the (presumably runaway) app would have\n> > sucked up all the connections, denying service to other apps using other\n> > databases. I think Marc's point here is to limit his exposure to\n> > misbehavior of any one client app, in a database server that is serving\n> > multiple clients using multiple databases.\n> \n> That would indeed be a useful item. The only way to avoid such\n> exposure right now is to run another back end.\n\nAdded to TODO:\n\n\t* Allow limits on per-db/user connections\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, 25 Oct 2002 15:02:52 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "On Fri, 2002-10-25 at 10:31, Bruce Momjian wrote:\n> \n> Yes, my comments related to useing db/user limits to control the number\n> of persistent connections. From an ISP perspective, I can see value in\n> user/db limits.\n> \n\nYes, this would be amazingly useful. I work for a web hosting provider\nand it happens all too often where a single customer creates a flawed\nscript which consumes all the DB connections. Obviously denying access\nto the rest of our customers.\n\nBeing able to set this per DB connection limit in Postgres itself\nwithout having to restart the backend would also make this feature very\nnice. \n\n-- \nBest Regards,\n \nMike Benoit\nNetNation Communication Inc.\nSystems Engineer\nTel: 604-684-6892 or 888-983-6600\n ---------------------------------------\n \n Disclaimer: Opinions expressed here are my own and not \n necessarily those of my employer\n\n",
"msg_date": "25 Oct 2002 12:16:50 -0700",
"msg_from": "Mike Benoit <mikeb@netnation.com>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Bruce Momjian wrote:\n> Andrew Sullivan wrote:\n> \n>>On Fri, Oct 25, 2002 at 11:02:48AM -0400, Tom Lane wrote:\n>>\n>>>So? If it hits the installation-wide limit, you'll have the same\n>>>problem; and at that point the (presumably runaway) app would have\n>>>sucked up all the connections, denying service to other apps using other\n>>>databases. I think Marc's point here is to limit his exposure to\n>>>misbehavior of any one client app, in a database server that is serving\n>>>multiple clients using multiple databases.\n>>\n>>That would indeed be a useful item. The only way to avoid such\n>>exposure right now is to run another back end.\n> \n> \n> Added to TODO:\n> \n> \t* Allow limits on per-db/user connections\n> \n\nCould I suggest that such a feature falls under the category of \nresource limits, and that the TODO should read something like:\n\nImplement the equivalent of Oracle PROFILEs.\n\nI think this would be a good project for 7.4. I'm not yet \nvolunteering, but if I can wrap up my current project, I might \nbe able to do it, depending upon the 7.4 target date. It would be:\n\n1. A new system table:\n\npg_profile\n\n2. The attributes of the profiles would be:\n\nprofname\nsession_per_user\ncpu_per_session\ncpu_per_call\nconnect_time\nidle_time\nlogical_reads_per_session\nlogical_reads_per_call\n\n3. A new field would be added to pg_user/pg_shadow:\n\nprofileid\n\n4. A 'default' profile would be created when a new database is \ncreated with no resource limits. CREATE/ALTER user would be \nmodified to allow for the specification of the profile. If no \nprofile is provided, 'default' is assumed.\n\n5. A new CREATE PROFILE/ALTER PROFILE/DROP PROFILE command set \nwould be implemented to add/update/remove the tuples in \npg_profiles. And according modification of pg_dump for \ndump/reload and psql for appropriate \\ command.\n\nExample:\n\nCREATE PROFILE clerk\nIDLE_TIME 30;\n\nALTER USER john PROFILE clerk;\nALTER USER bob PROFILE clerk;\n\nor, for an ISP maybe:\n\nALYTER PROFILE default\nIDLE_TIME 30;\n\nIt seems like a nice project, particularly since it wouldn't \naffect anyone that doesn't want to use it. And whenever a new \nresource limitation issue arrises, such as PL/SQL recursion \ndepth, a new attribute would be added to pg_profile to handle \nthe limitation...\n\nMike Mascari\nmascarm@mascari.com\n\n\n\n\n\n\n",
"msg_date": "Fri, 25 Oct 2002 15:31:22 -0400",
"msg_from": "Mike Mascari <mascarm@mascari.com>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "\nI need others wanting this before I can add something of this\nsophistication to TODO.\n\n---------------------------------------------------------------------------\n\nMike Mascari wrote:\n> Bruce Momjian wrote:\n> > Andrew Sullivan wrote:\n> > \n> >>On Fri, Oct 25, 2002 at 11:02:48AM -0400, Tom Lane wrote:\n> >>\n> >>>So? If it hits the installation-wide limit, you'll have the same\n> >>>problem; and at that point the (presumably runaway) app would have\n> >>>sucked up all the connections, denying service to other apps using other\n> >>>databases. I think Marc's point here is to limit his exposure to\n> >>>misbehavior of any one client app, in a database server that is serving\n> >>>multiple clients using multiple databases.\n> >>\n> >>That would indeed be a useful item. The only way to avoid such\n> >>exposure right now is to run another back end.\n> > \n> > \n> > Added to TODO:\n> > \n> > \t* Allow limits on per-db/user connections\n> > \n> \n> Could I suggest that such a feature falls under the category of \n> resource limits, and that the TODO should read something like:\n> \n> Implement the equivalent of Oracle PROFILEs.\n> \n> I think this would be a good project for 7.4. I'm not yet \n> volunteering, but if I can wrap up my current project, I might \n> be able to do it, depending upon the 7.4 target date. It would be:\n> \n> 1. A new system table:\n> \n> pg_profile\n> \n> 2. The attributes of the profiles would be:\n> \n> profname\n> session_per_user\n> cpu_per_session\n> cpu_per_call\n> connect_time\n> idle_time\n> logical_reads_per_session\n> logical_reads_per_call\n> \n> 3. A new field would be added to pg_user/pg_shadow:\n> \n> profileid\n> \n> 4. A 'default' profile would be created when a new database is \n> created with no resource limits. CREATE/ALTER user would be \n> modified to allow for the specification of the profile. If no \n> profile is provided, 'default' is assumed.\n> \n> 5. A new CREATE PROFILE/ALTER PROFILE/DROP PROFILE command set \n> would be implemented to add/update/remove the tuples in \n> pg_profiles. And according modification of pg_dump for \n> dump/reload and psql for appropriate \\ command.\n> \n> Example:\n> \n> CREATE PROFILE clerk\n> IDLE_TIME 30;\n> \n> ALTER USER john PROFILE clerk;\n> ALTER USER bob PROFILE clerk;\n> \n> or, for an ISP maybe:\n> \n> ALYTER PROFILE default\n> IDLE_TIME 30;\n> \n> It seems like a nice project, particularly since it wouldn't \n> affect anyone that doesn't want to use it. And whenever a new \n> resource limitation issue arrises, such as PL/SQL recursion \n> depth, a new attribute would be added to pg_profile to handle \n> the limitation...\n> \n> Mike Mascari\n> mascarm@mascari.com\n> \n> \n> \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": "Fri, 25 Oct 2002 15:54:27 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Mike Mascari <mascarm@mascari.com> writes:\n> [ extensive proposal for PROFILEs ]\n> It seems like a nice project, particularly since it wouldn't \n> affect anyone that doesn't want to use it.\n\n... except in the added overhead to do the resource accounting and check\nto see if there is a restriction ...\n\n> And whenever a new \n> resource limitation issue arrises, such as PL/SQL recursion \n> depth, a new attribute would be added to pg_profile to handle \n> the limitation...\n\nI prefer GUC variables to table entries for setting stuff like recursion\nlimits; they're much lighter-weight to create and access, and you don't\nneed an initdb to add or remove a parameter.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 25 Oct 2002 16:17:46 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "On Fri, 2002-10-25 at 16:17, Tom Lane wrote:\n> Mike Mascari <mascarm@mascari.com> writes:\n> > [ extensive proposal for PROFILEs ]\n> > It seems like a nice project, particularly since it wouldn't \n> > affect anyone that doesn't want to use it.\n> \n> ... except in the added overhead to do the resource accounting and check\n> to see if there is a restriction ...\n\nperhaps you could make a GUC variable \"use_resource_profiles\" that turns\nthe whole thing on/off. \n\n> \n> > And whenever a new \n> > resource limitation issue arrises, such as PL/SQL recursion \n> > depth, a new attribute would be added to pg_profile to handle \n> > the limitation...\n> \n> I prefer GUC variables to table entries for setting stuff like recursion\n> limits; they're much lighter-weight to create and access, and you don't\n> need an initdb to add or remove a parameter.\n> \n\nI don't see an adequate way to store the individual settings as GUC\nvariables per user...\n\nRobert Treat\n\n",
"msg_date": "25 Oct 2002 16:32:58 -0400",
"msg_from": "Robert Treat <xzilla@users.sourceforge.net>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Robert Treat <xzilla@users.sourceforge.net> writes:\n> On Fri, 2002-10-25 at 16:17, Tom Lane wrote:\n>> I prefer GUC variables to table entries for setting stuff like recursion\n>> limits; they're much lighter-weight to create and access, and you don't\n>> need an initdb to add or remove a parameter.\n\n> I don't see an adequate way to store the individual settings as GUC\n> variables per user...\n\nHave you looked at the per-database and per-user GUC facilities in 7.3?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 25 Oct 2002 16:43:41 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ... "
},
{
"msg_contents": "Tom Lane wrote:\n> Robert Treat <xzilla@users.sourceforge.net> writes:\n> > On Fri, 2002-10-25 at 16:17, Tom Lane wrote:\n> >> I prefer GUC variables to table entries for setting stuff like recursion\n> >> limits; they're much lighter-weight to create and access, and you don't\n> >> need an initdb to add or remove a parameter.\n> \n> > I don't see an adequate way to store the individual settings as GUC\n> > variables per user...\n> \n> Have you looked at the per-database and per-user GUC facilities in 7.3?\n\nNice idea. You can now have per-user/db settings that are SET when the\nconnection is made. You can set any GUC variable that way. We just\nneed a variable that can look at all sessions and determine if that user\nhas exceeded their connection quota.\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, 25 Oct 2002 19:03:59 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "On Friday 25 October 2002 07:03 pm, Bruce Momjian wrote:\n> Tom Lane wrote:\n> > Robert Treat <xzilla@users.sourceforge.net> writes:\n> > > On Fri, 2002-10-25 at 16:17, Tom Lane wrote:\n> > >> I prefer GUC variables to table entries for setting stuff like\n> > >> recursion limits; they're much lighter-weight to create and access,\n> > >> and you don't need an initdb to add or remove a parameter.\n> > >\n> > > I don't see an adequate way to store the individual settings as GUC\n> > > variables per user...\n> >\n> > Have you looked at the per-database and per-user GUC facilities in 7.3?\n>\n\nMaybe I haven't looked at them enough ;-)\n\n> Nice idea. You can now have per-user/db settings that are SET when the\n> connection is made. You can set any GUC variable that way. We just\n> need a variable that can look at all sessions and determine if that user\n> has exceeded their connection quota.\n\nI understand how you are able to set those variables per db, but I don't see \nhow you can get those settings to persist between database shutdowns. \nPerhaps someone can point me to the relevant documentation?\n\nRobert Treat\n",
"msg_date": "Sat, 26 Oct 2002 01:01:47 -0400",
"msg_from": "Robert Treat <xzilla@users.sourceforge.net>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Robert Treat wrote:\n> > Nice idea. You can now have per-user/db settings that are SET when the\n> > connection is made. You can set any GUC variable that way. We just\n> > need a variable that can look at all sessions and determine if that user\n> > has exceeded their connection quota.\n> \n> I understand how you are able to set those variables per db, but I don't see \n> how you can get those settings to persist between database shutdowns. \n> Perhaps someone can point me to the relevant documentation?\n\nThe per db/user stuff is stored in the pg_database/pg_shadow tables per\nrow, so they exist in permanent storage.\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, 26 Oct 2002 01:04:55 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "On Sat, Oct 26, 2002 at 01:04:55 -0400,\n Bruce Momjian <pgman@candle.pha.pa.us> wrote:\n> \n> The per db/user stuff is stored in the pg_database/pg_shadow tables per\n> row, so they exist in permanent storage.\n\nI have a question about this. This stuff is per user OR per db right?\nWhen I see per db/user I get the impression that users can have different\nsettings depending on which db they connect to. But looking at alter\ndatabase and alter user it looks like settings are per database or\nper user, but there isn't a way to (in general) set something that\napplies to a particular user when connecting to a particular database.\nIf there is a way to do that, I would be interested in a hint where to\nlook in the documentation.\n",
"msg_date": "Sat, 26 Oct 2002 00:26:39 -0500",
"msg_from": "Bruno Wolff III <bruno@wolff.to>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "On Saturday 26 October 2002 01:04 am, Bruce Momjian wrote:\n> Robert Treat wrote:\n> > > Nice idea. You can now have per-user/db settings that are SET when the\n> > > connection is made. You can set any GUC variable that way. We just\n> > > need a variable that can look at all sessions and determine if that\n> > > user has exceeded their connection quota.\n> >\n> > I understand how you are able to set those variables per db, but I don't\n> > see how you can get those settings to persist between database shutdowns.\n> > Perhaps someone can point me to the relevant documentation?\n>\n> The per db/user stuff is stored in the pg_database/pg_shadow tables per\n> row, so they exist in permanent storage.\n\nAh.. that was the missing link. I don't think the docs mention anything about \nthe information being stored in pg_database or pg_shadow, they always refer \nto the \"session default\" which made me wonder what happened \"between \nsessions\". Perhaps someone (me?) should work up a patch to help clarify \nthis....\n\nRobert Treat\n\n\n",
"msg_date": "Sat, 26 Oct 2002 02:04:15 -0400",
"msg_from": "Robert Treat <xzilla@users.sourceforge.net>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Bruno Wolff III wrote:\n> On Sat, Oct 26, 2002 at 01:04:55 -0400,\n> Bruce Momjian <pgman@candle.pha.pa.us> wrote:\n> > \n> > The per db/user stuff is stored in the pg_database/pg_shadow tables per\n> > row, so they exist in permanent storage.\n> \n> I have a question about this. This stuff is per user OR per db right?\n> When I see per db/user I get the impression that users can have different\n> settings depending on which db they connect to. But looking at alter\n> database and alter user it looks like settings are per database or\n> per user, but there isn't a way to (in general) set something that\n> applies to a particular user when connecting to a particular database.\n> If there is a way to do that, I would be interested in a hint where to\n> look in the documentation.\n\nYou are right, there isn't a per/db-user combination setting. I think\none is done before the other, so you could try to set things that way,\nmaybe in a plpgsql procedure. I think you could do SELECT\ndb_user_set(); and have that function do that sets for you.\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, 26 Oct 2002 16:06:09 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "On Fri, 25 Oct 2002, Bruce Momjian wrote:\n\n>\n> I need others wanting this before I can add something of this\n> sophistication to TODO.\n\nSounds cool to me, and would satisfy what I'm looking for, since it sounds\nlike the same user could have different limits depending on the database\nthey are/were connectin gto ...\n\n >\n> ---------------------------------------------------------------------------\n>\n> Mike Mascari wrote:\n> > Bruce Momjian wrote:\n> > > Andrew Sullivan wrote:\n> > >\n> > >>On Fri, Oct 25, 2002 at 11:02:48AM -0400, Tom Lane wrote:\n> > >>\n> > >>>So? If it hits the installation-wide limit, you'll have the same\n> > >>>problem; and at that point the (presumably runaway) app would have\n> > >>>sucked up all the connections, denying service to other apps using other\n> > >>>databases. I think Marc's point here is to limit his exposure to\n> > >>>misbehavior of any one client app, in a database server that is serving\n> > >>>multiple clients using multiple databases.\n> > >>\n> > >>That would indeed be a useful item. The only way to avoid such\n> > >>exposure right now is to run another back end.\n> > >\n> > >\n> > > Added to TODO:\n> > >\n> > > \t* Allow limits on per-db/user connections\n> > >\n> >\n> > Could I suggest that such a feature falls under the category of\n> > resource limits, and that the TODO should read something like:\n> >\n> > Implement the equivalent of Oracle PROFILEs.\n> >\n> > I think this would be a good project for 7.4. I'm not yet\n> > volunteering, but if I can wrap up my current project, I might\n> > be able to do it, depending upon the 7.4 target date. It would be:\n> >\n> > 1. A new system table:\n> >\n> > pg_profile\n> >\n> > 2. The attributes of the profiles would be:\n> >\n> > profname\n> > session_per_user\n> > cpu_per_session\n> > cpu_per_call\n> > connect_time\n> > idle_time\n> > logical_reads_per_session\n> > logical_reads_per_call\n> >\n> > 3. A new field would be added to pg_user/pg_shadow:\n> >\n> > profileid\n> >\n> > 4. A 'default' profile would be created when a new database is\n> > created with no resource limits. CREATE/ALTER user would be\n> > modified to allow for the specification of the profile. If no\n> > profile is provided, 'default' is assumed.\n> >\n> > 5. A new CREATE PROFILE/ALTER PROFILE/DROP PROFILE command set\n> > would be implemented to add/update/remove the tuples in\n> > pg_profiles. And according modification of pg_dump for\n> > dump/reload and psql for appropriate \\ command.\n> >\n> > Example:\n> >\n> > CREATE PROFILE clerk\n> > IDLE_TIME 30;\n> >\n> > ALTER USER john PROFILE clerk;\n> > ALTER USER bob PROFILE clerk;\n> >\n> > or, for an ISP maybe:\n> >\n> > ALYTER PROFILE default\n> > IDLE_TIME 30;\n> >\n> > It seems like a nice project, particularly since it wouldn't\n> > affect anyone that doesn't want to use it. And whenever a new\n> > resource limitation issue arrises, such as PL/SQL recursion\n> > depth, a new attribute would be added to pg_profile to handle\n> > the limitation...\n> >\n> > Mike Mascari\n> > mascarm@mascari.com\n> >\n> >\n> >\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>\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": "Sat, 26 Oct 2002 23:55:42 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "\nTom's idea of doing this as part of the already added per-user/per-db\nsettings seems like a good direction to take.\n\n---------------------------------------------------------------------------\n\nMarc G. Fournier wrote:\n> On Fri, 25 Oct 2002, Bruce Momjian wrote:\n> \n> >\n> > I need others wanting this before I can add something of this\n> > sophistication to TODO.\n> \n> Sounds cool to me, and would satisfy what I'm looking for, since it sounds\n> like the same user could have different limits depending on the database\n> they are/were connectin gto ...\n> \n> >\n> > ---------------------------------------------------------------------------\n> >\n> > Mike Mascari wrote:\n> > > Bruce Momjian wrote:\n> > > > Andrew Sullivan wrote:\n> > > >\n> > > >>On Fri, Oct 25, 2002 at 11:02:48AM -0400, Tom Lane wrote:\n> > > >>\n> > > >>>So? If it hits the installation-wide limit, you'll have the same\n> > > >>>problem; and at that point the (presumably runaway) app would have\n> > > >>>sucked up all the connections, denying service to other apps using other\n> > > >>>databases. I think Marc's point here is to limit his exposure to\n> > > >>>misbehavior of any one client app, in a database server that is serving\n> > > >>>multiple clients using multiple databases.\n> > > >>\n> > > >>That would indeed be a useful item. The only way to avoid such\n> > > >>exposure right now is to run another back end.\n> > > >\n> > > >\n> > > > Added to TODO:\n> > > >\n> > > > \t* Allow limits on per-db/user connections\n> > > >\n> > >\n> > > Could I suggest that such a feature falls under the category of\n> > > resource limits, and that the TODO should read something like:\n> > >\n> > > Implement the equivalent of Oracle PROFILEs.\n> > >\n> > > I think this would be a good project for 7.4. I'm not yet\n> > > volunteering, but if I can wrap up my current project, I might\n> > > be able to do it, depending upon the 7.4 target date. It would be:\n> > >\n> > > 1. A new system table:\n> > >\n> > > pg_profile\n> > >\n> > > 2. The attributes of the profiles would be:\n> > >\n> > > profname\n> > > session_per_user\n> > > cpu_per_session\n> > > cpu_per_call\n> > > connect_time\n> > > idle_time\n> > > logical_reads_per_session\n> > > logical_reads_per_call\n> > >\n> > > 3. A new field would be added to pg_user/pg_shadow:\n> > >\n> > > profileid\n> > >\n> > > 4. A 'default' profile would be created when a new database is\n> > > created with no resource limits. CREATE/ALTER user would be\n> > > modified to allow for the specification of the profile. If no\n> > > profile is provided, 'default' is assumed.\n> > >\n> > > 5. A new CREATE PROFILE/ALTER PROFILE/DROP PROFILE command set\n> > > would be implemented to add/update/remove the tuples in\n> > > pg_profiles. And according modification of pg_dump for\n> > > dump/reload and psql for appropriate \\ command.\n> > >\n> > > Example:\n> > >\n> > > CREATE PROFILE clerk\n> > > IDLE_TIME 30;\n> > >\n> > > ALTER USER john PROFILE clerk;\n> > > ALTER USER bob PROFILE clerk;\n> > >\n> > > or, for an ISP maybe:\n> > >\n> > > ALYTER PROFILE default\n> > > IDLE_TIME 30;\n> > >\n> > > It seems like a nice project, particularly since it wouldn't\n> > > affect anyone that doesn't want to use it. And whenever a new\n> > > resource limitation issue arrises, such as PL/SQL recursion\n> > > depth, a new attribute would be added to pg_profile to handle\n> > > the limitation...\n> > >\n> > > Mike Mascari\n> > > mascarm@mascari.com\n> > >\n> > >\n> > >\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> >\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> ---------------------------(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": "Sat, 26 Oct 2002 23:12:01 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "On Sat, Oct 26, 2002 at 11:12:01PM -0400, Bruce Momjian wrote:\n> \n> Tom's idea of doing this as part of the already added per-user/per-db\n> settings seems like a good direction to take.\n\nYes, but how? The current implementation has a setting column in\npg_shadow, and another in pg_database. Is the idea to create a separate\npg_settings table or something like that?\n\n-- \nAlvaro Herrera (<alvherre[a]dcc.uchile.cl>)\n\"In Europe they call me Niklaus Wirth; in the US they call me Nickel's worth.\nThat's because in Europe they call me by name, and in the US by value!\"\n",
"msg_date": "Sun, 27 Oct 2002 00:27:18 -0300",
"msg_from": "Alvaro Herrera <alvherre@dcc.uchile.cl>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Alvaro Herrera wrote:\n> On Sat, Oct 26, 2002 at 11:12:01PM -0400, Bruce Momjian wrote:\n> > \n> > Tom's idea of doing this as part of the already added per-user/per-db\n> > settings seems like a good direction to take.\n> \n> Yes, but how? The current implementation has a setting column in\n> pg_shadow, and another in pg_database. Is the idea to create a separate\n> pg_settings table or something like that?\n\nAre you asking how to do per-db-user combination settings, like User A\ncan have only two connections to database B? Seems we should get\nper-user and per-db settings working first and see if anyone needs\ncombination settings.\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, 26 Oct 2002 23:58:35 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "On Sat, Oct 26, 2002 at 11:58:35PM -0400, Bruce Momjian wrote:\n> Alvaro Herrera wrote:\n> > On Sat, Oct 26, 2002 at 11:12:01PM -0400, Bruce Momjian wrote:\n> > > \n> > > Tom's idea of doing this as part of the already added per-user/per-db\n> > > settings seems like a good direction to take.\n> > \n> > Yes, but how? The current implementation has a setting column in\n> > pg_shadow, and another in pg_database. Is the idea to create a separate\n> > pg_settings table or something like that?\n> \n> Are you asking how to do per-db-user combination settings, like User A\n> can have only two connections to database B? Seems we should get\n> per-user and per-db settings working first and see if anyone needs\n> combination settings.\n\nWell, now that I think about it... the \"user local to database\" (those\nwith the @dbname appended and all that) idea may very well cover the\nground for this. But I'm not actually asking, because I don't need it.\n\n-- \nAlvaro Herrera (<alvherre[a]dcc.uchile.cl>)\n\"Aprender sin pensar es inutil; pensar sin aprender, peligroso\" (Confucio)\n",
"msg_date": "Sun, 27 Oct 2002 01:09:11 -0300",
"msg_from": "Alvaro Herrera <alvherre@dcc.uchile.cl>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "On Fri, Oct 25, 2002 at 03:31:22PM -0400, Mike Mascari wrote:\n> Bruce Momjian wrote:\n> > Andrew Sullivan wrote:\n> > \n> >>On Fri, Oct 25, 2002 at 11:02:48AM -0400, Tom Lane wrote:\n> >>\n> >>>So? If it hits the installation-wide limit, you'll have the same\n> >>>problem; and at that point the (presumably runaway) app would have\n> >>>sucked up all the connections, denying service to other apps using other\n> >>>databases. I think Marc's point here is to limit his exposure to\n> >>>misbehavior of any one client app, in a database server that is serving\n> >>>multiple clients using multiple databases.\n> >>\n> >>That would indeed be a useful item. The only way to avoid such\n> >>exposure right now is to run another back end.\n> > \n> > \n> > Added to TODO:\n> > \n> > \t* Allow limits on per-db/user connections\n> > \n> \n> Could I suggest that such a feature falls under the category of \n> resource limits, and that the TODO should read something like:\n> \n> Implement the equivalent of Oracle PROFILEs.\n\n Yes! Please.... it's better than all discussions about some ugly\n variables. The PROFILE is better extendable and it's user \n specific and in the system with ROLEs it really cool and simple\n set user's system options.\n \n I talked about it more times, but is still ignore :-) I don't want \n to maintain my databases by SET command.\n\n> profname\n> session_per_user\n> cpu_per_session\n> cpu_per_call\n> connect_time\n> idle_time\n> logical_reads_per_session\n> logical_reads_per_call\n\n ... and a lot of others things in future.\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": "Tue, 29 Oct 2002 08:11:07 +0100",
"msg_from": "Karel Zak <zakkr@zf.jcu.cz>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
},
{
"msg_contents": "Karel Zak wrote:\n> On Fri, Oct 25, 2002 at 03:31:22PM -0400, Mike Mascari wrote:\n> \n>>Bruce Momjian wrote:\n>>\n>>>Added to TODO:\n>>>\n>>>\t* Allow limits on per-db/user connections\n>>>\n>>\n>>Could I suggest that such a feature falls under the category of \n>>resource limits, and that the TODO should read something like:\n>>\n>>Implement the equivalent of Oracle PROFILEs.\n> \n> \n> Yes! Please.... it's better than all discussions about some ugly\n> variables. The PROFILE is better extendable and it's user \n> specific and in the system with ROLEs it really cool and simple\n> set user's system options.\n> \n> I talked about it more times, but is still ignore :-) I don't want \n> to maintain my databases by SET command.\n\nIt seems we are in the minority. :-(\n\nMike Mascari\nmascarm@mascari.com\n\n\n\n",
"msg_date": "Tue, 29 Oct 2002 05:11:29 -0500",
"msg_from": "Mike Mascari <mascarm@mascari.com>",
"msg_from_op": false,
"msg_subject": "Re: idle connection timeout ..."
}
] |
[
{
"msg_contents": "\nI just reread the patch; is it valid to assume fseek and fseeko have the \nsame failure modes? Or does the call to 'fseek' actually call fseeko?\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n\n",
"msg_date": "Fri, 25 Oct 2002 13:13:02 +1000",
"msg_from": "Philip Warner <pjw@rhyme.com.au>",
"msg_from_op": true,
"msg_subject": "Re: pg_dump and large files - is this a problem?"
},
{
"msg_contents": "Philip Warner wrote:\n> \n> I just reread the patch; is it valid to assume fseek and fseeko have the \n> same failure modes? Or does the call to 'fseek' actually call fseeko?\n\nThe fseek was a typo. It should have been fseeko as you suggested.\nCVS updated.\n\nYour idea of using SEEK_SET is good, except I was concerned that the\ncheckSeek call will move the file pointer. Is that OK? It doesn't seem\nappropriate.\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, 24 Oct 2002 23:51:26 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and large files - is this a problem?"
},
{
"msg_contents": "At 11:51 PM 24/10/2002 -0400, Bruce Momjian wrote:\n>Your idea of using SEEK_SET is good, except I was concerned that the\n>checkSeek call will move the file pointer. Is that OK? It doesn't seem\n>appropriate.\n\nThe call is made just after the file is opened (or it should be!), so \nSEEK_SET, 0 will not be a problem.\n\n\n\n----------------------------------------------------------------\nPhilip Warner | __---_____\nAlbatross Consulting Pty. Ltd. |----/ - \\\n(A.B.N. 75 008 659 498) | /(@) ______---_\nTel: (+61) 0500 83 82 81 | _________ \\\nFax: (+61) 0500 83 82 82 | ___________ |\nHttp://www.rhyme.com.au | / \\|\n | --________--\nPGP key available upon request, | /\nand from pgp5.ai.mit.edu:11371 |/\n\n",
"msg_date": "Fri, 25 Oct 2002 16:43:42 +1000",
"msg_from": "Philip Warner <pjw@rhyme.com.au>",
"msg_from_op": true,
"msg_subject": "Re: pg_dump and large files - is this a problem?"
}
] |
[
{
"msg_contents": "All,\n\nI'm trying to implement row level security using PostgreSQL. I envision\nhaving column on each and every table in the database that will contain a\ndiscriminating value that determines if a given user is able to see that row\nor not. Users do not directly access the database; they get at it only\nthrough client applications.\n\nIs there a way to initially configure PostgreSQL such that each and every\ntime a 'create table foo....' command is executed that, something else would\nhappen instead?\n\nFor example,\n\nCREATE TABLE weather (\n city varchar(80),\n lo int,\n hi int,\n prcp real,\n date date\n);\n\nwould result in the following being executed automatically:\n\nCREATE TABLE t_weather (\n city varchar(80),\n lo int,\n hi int,\n prcp real,\n date date,\n hidden_column int\n);\n\nCREATE VIEW weather AS\n SELECT city, lo, hi, prcp, date\n FROM t_weather;\n\n\nHow the view gets populated is another discussion entirely.\n\nBut is there a way to set things up initially such that each 'create table'\ncommand will exhibit the behavior described above?\n\nThanks.\n\n--\n\nBruce\n\n\n\n\n",
"msg_date": "Fri, 25 Oct 2002 13:50:45 -0400",
"msg_from": "\"Bruce David\" <brucedavid@bigfoot.com>",
"msg_from_op": true,
"msg_subject": "Trigger on 'create table' ?"
},
{
"msg_contents": "\nSorry, I can't think of any to have a view created automatically for\nevery table creation. Interesting idea, though. I wonder if you could\npass the CREATE string into a function that does the proper creation.\n\n---------------------------------------------------------------------------\n\nBruce David wrote:\n> All,\n> \n> I'm trying to implement row level security using PostgreSQL. I envision\n> having column on each and every table in the database that will contain a\n> discriminating value that determines if a given user is able to see that row\n> or not. Users do not directly access the database; they get at it only\n> through client applications.\n> \n> Is there a way to initially configure PostgreSQL such that each and every\n> time a 'create table foo....' command is executed that, something else would\n> happen instead?\n> \n> For example,\n> \n> CREATE TABLE weather (\n> city varchar(80),\n> lo int,\n> hi int,\n> prcp real,\n> date date\n> );\n> \n> would result in the following being executed automatically:\n> \n> CREATE TABLE t_weather (\n> city varchar(80),\n> lo int,\n> hi int,\n> prcp real,\n> date date,\n> hidden_column int\n> );\n> \n> CREATE VIEW weather AS\n> SELECT city, lo, hi, prcp, date\n> FROM t_weather;\n> \n> \n> How the view gets populated is another discussion entirely.\n> \n> But is there a way to set things up initially such that each 'create table'\n> command will exhibit the behavior described above?\n> \n> Thanks.\n> \n> --\n> \n> Bruce\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": "Mon, 28 Oct 2002 21:23:56 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Trigger on 'create table' ?"
},
{
"msg_contents": "We have a similar requirement for PostGIS.\nIn order to meet the OpenGIS specification, we have to maintain a \nGEOMETRY_COLUMNS table which includes an entry for every column in the \ndb which holds a spatial column. It would be ideal if we could have \ntriggers run on CREATE TABLE, DROP TABLE, and ALTER TABLE to watch for \nthe comings and goings of spatial columns and update GEOMETRY_COLUMNS \nappropriately.\nUnfortunately, triggers seem to not be allowed on system tables, or we \nwould just watch pg_class for new geometry columns and update the \nOpenGIS table as necessary. :/\n\n\nBruce Momjian wrote:\n> Sorry, I can't think of any to have a view created automatically for\n> every table creation. Interesting idea, though. I wonder if you could\n> pass the CREATE string into a function that does the proper creation.\n> \n> ---------------------------------------------------------------------------\n> \n> Bruce David wrote:\n> \n>>All,\n>>\n>>I'm trying to implement row level security using PostgreSQL. I envision\n>>having column on each and every table in the database that will contain a\n>>discriminating value that determines if a given user is able to see that row\n>>or not. Users do not directly access the database; they get at it only\n>>through client applications.\n>>\n>>Is there a way to initially configure PostgreSQL such that each and every\n>>time a 'create table foo....' command is executed that, something else would\n>>happen instead?\n>>\n>>For example,\n>>\n>>CREATE TABLE weather (\n>> city varchar(80),\n>> lo int,\n>> hi int,\n>> prcp real,\n>> date date\n>>);\n>>\n>>would result in the following being executed automatically:\n>>\n>>CREATE TABLE t_weather (\n>> city varchar(80),\n>> lo int,\n>> hi int,\n>> prcp real,\n>> date date,\n>> hidden_column int\n>>);\n>>\n>>CREATE VIEW weather AS\n>> SELECT city, lo, hi, prcp, date\n>> FROM t_weather;\n>>\n>>\n>>How the view gets populated is another discussion entirely.\n>>\n>>But is there a way to set things up initially such that each 'create table'\n>>command will exhibit the behavior described above?\n>>\n>>Thanks.\n>>\n>>--\n>>\n>>Bruce\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\n\n",
"msg_date": "Mon, 28 Oct 2002 18:33:21 -0800",
"msg_from": "Paul Ramsey <pramsey@refractions.net>",
"msg_from_op": false,
"msg_subject": "Re: Trigger on 'create table' ?"
},
{
"msg_contents": "Paul Ramsey writes:\n\n> In order to meet the OpenGIS specification, we have to maintain a\n> GEOMETRY_COLUMNS table which includes an entry for every column in the\n> db which holds a spatial column.\n\nWhy not make a view?\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Tue, 29 Oct 2002 23:18:27 +0100 (CET)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: Trigger on 'create table' ?"
},
{
"msg_contents": "Paul Ramsey <pramsey@refractions.net> writes:\n> We have a similar requirement for PostGIS.\n> In order to meet the OpenGIS specification, we have to maintain a \n> GEOMETRY_COLUMNS table which includes an entry for every column in the \n> db which holds a spatial column. It would be ideal if we could have \n> triggers run on CREATE TABLE, DROP TABLE, and ALTER TABLE to watch for \n> the comings and goings of spatial columns and update GEOMETRY_COLUMNS \n> appropriately.\n\nCouldn't you define GEOMETRY_COLUMNS as a view? Seems a lot more\nefficient than firing random triggers on every table operation.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 29 Oct 2002 17:33:07 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Trigger on 'create table' ? "
},
{
"msg_contents": "So close... It is a testament to my blockheadedness that I had not \nconsidered that (Dave probably did). Here is the standards-mandated \ndefinition of the table:\n\nCREATE TABLE GEOMETRY_COLUMNS (\n F_TABLE_CATALOG VARCHAR(256) NOT NULL,\n F_TABLE_SCHEMA VARCHAR(256) NOT NULL,\n F_TABLE_NAME VARCHAR(256) NOT NULL,\n F_GEOMETRY_COLUMN VARCHAR(256) NOT NULL,\n COORD_DIMENSION INTEGER,\n SRID INTEGER REFERENCES SPATIAL_REF_SYS,\n CONSTRAINT GC_PK PRIMARY KEY\n(F_TABLE_CATALOG, F_TABLE_SCHEMA, F_TABLE_NAME, F_GEOMETRY_COLUMN)\n)\n\nEverything except for COORD_DIMENSION and SRID could be extracted from \nthe pg_class table as a view, which would be nice. The COORD_DIMENSION \nis just the dimensionality of the geometries, and the SRID is the \nspatial reference system identifier, which is a key into another \nmandated OpenGIS table, 'SPATIAL_REF_SYS'.\n\nNow, if we changed the geometry type so that when one defined a geometry \ncolumn, one had to include info about what SRID and what dimension it \nwas, (ala varchar(243)) maybe the whole schmeer could reside in pg_class \nand geometry_columns would be a trivial view?\n\nCrazy? Loony?\n\nP.\n\nTom Lane wrote:\n> Paul Ramsey <pramsey@refractions.net> writes:\n> \n>>We have a similar requirement for PostGIS.\n>>In order to meet the OpenGIS specification, we have to maintain a \n>>GEOMETRY_COLUMNS table which includes an entry for every column in the \n>>db which holds a spatial column. It would be ideal if we could have \n>>triggers run on CREATE TABLE, DROP TABLE, and ALTER TABLE to watch for \n>>the comings and goings of spatial columns and update GEOMETRY_COLUMNS \n>>appropriately.\n> \n> \n> Couldn't you define GEOMETRY_COLUMNS as a view? Seems a lot more\n> efficient than firing random triggers on every table operation.\n> \n> \t\t\tregards, tom lane\n> \n\n\n-- \n __\n /\n | Paul Ramsey\n | Refractions Research\n | Email: pramsey@refractions.net\n | Phone: (250) 885-0632\n \\_\n\n",
"msg_date": "Tue, 29 Oct 2002 16:56:31 -0800",
"msg_from": "Paul Ramsey <pramsey@refractions.net>",
"msg_from_op": false,
"msg_subject": "Re: Trigger on 'create table' ?"
},
{
"msg_contents": "Paul Ramsey <pramsey@refractions.net> writes:\n> Now, if we changed the geometry type so that when one defined a geometry \n> column, one had to include info about what SRID and what dimension it \n> was, (ala varchar(243)) maybe the whole schmeer could reside in pg_class \n> and geometry_columns would be a trivial view?\n\natttypmod would be a natural place for that stuff, if you can fit it\ninto 31 bits. The bigger problem is how are you going to get\ntype-specific syntax through the Postgres parser? Currently, all the\ntypes that use typmod have to have special-purpose syntax to set it.\nI'm not eager to see extension datatypes trying to do the same thing...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 30 Oct 2002 12:42:25 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Trigger on 'create table' ? "
}
] |
[
{
"msg_contents": "\nWe are putting out beta3 today or tomorrow. It seems we are ready to\nstart considering an RC1 date, perhaps next Friday, November 1?\n\n---------------------------------------------------------------------------\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://momjian.postgresql.org/pub/postgresql/open_items.\n\nRequired Changes\n-------------------\n\nOptional Changes\n----------------\nFix Linux + Perl 5.8.1 + _GNU_SOURCE problem\nFix AIX + Large File + Flex problem\n\nDocumentation Changes\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, 25 Oct 2002 16:18:22 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Time for RC1 soon?"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> We are putting out beta3 today or tomorrow. It seems we are ready to\n> start considering an RC1 date, perhaps next Friday, November 1?\n\nSeems like someone ought to issue a call for port reports. The\n\"supported platforms\" list hasn't been touched ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 25 Oct 2002 17:06:39 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Time for RC1 soon? "
},
{
"msg_contents": "Tom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > We are putting out beta3 today or tomorrow. It seems we are ready to\n> > start considering an RC1 date, perhaps next Friday, November 1?\n> \n> Seems like someone ought to issue a call for port reports. The\n> \"supported platforms\" list hasn't been touched ...\n\nGood point. Thomas, can you take that on?\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, 25 Oct 2002 19:10:23 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Time for RC1 soon?"
},
{
"msg_contents": ">>Seems like someone ought to issue a call for port reports. The\n>>\"supported platforms\" list hasn't been touched ...\n> Good point. Thomas, can you take that on?\n\nNo, at least not now. I'm not able to communicate reliably with the \nmailing lists, and so can not coordinate anything :( Not sure when or if \nthat will be resolved, but I'll be out of town next week so...\n\n - Thomas\n\n",
"msg_date": "Sat, 26 Oct 2002 00:48:18 +0000",
"msg_from": "Thomas Lockhart <lockhart@fourpalms.org>",
"msg_from_op": false,
"msg_subject": "Re: Time for RC1 soon?"
},
{
"msg_contents": "... and just out of curiosity, why does Bruce's message have a [HACKERS] \ntag in the subject line but my reply does not? It seems to be going \nthrough the same mailer but with different results...\n\n - Thomas\n\n",
"msg_date": "Sat, 26 Oct 2002 00:53:10 +0000",
"msg_from": "Thomas Lockhart <lockhart@fourpalms.org>",
"msg_from_op": false,
"msg_subject": "Re: Time for RC1 soon?"
},
{
"msg_contents": "Thomas Lockhart wrote:\n> ... and just out of curiosity, why does Bruce's message have a [HACKERS] \n> tag in the subject line but my reply does not? It seems to be going \n> through the same mailer but with different results...\n\nYes, I have noticed this happening at random times.\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, 25 Oct 2002 21:56:15 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Time for RC1 soon?"
},
{
"msg_contents": "Thomas Lockhart wrote:\n> >>Seems like someone ought to issue a call for port reports. The\n> >>\"supported platforms\" list hasn't been touched ...\n> > Good point. Thomas, can you take that on?\n> \n> No, at least not now. I'm not able to communicate reliably with the \n> mailing lists, and so can not coordinate anything :( Not sure when or if \n> that will be resolved, but I'll be out of town next week so...\n\nOK, Tom will be away next week, and Thomas will too. I can do it. \nFolks. start sending in those plaform reports, OS name and version\nnumber please.\n\nThe current platform list is:\n\n\thttp://developer.postgresql.org/docs/postgres/supported-platforms.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": "Fri, 25 Oct 2002 22:07:48 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Time for RC1 soon?"
},
{
"msg_contents": "Thomas Lockhart wrote:\n> >>Seems like someone ought to issue a call for port reports. The\n> >>\"supported platforms\" list hasn't been touched ...\n> > Good point. Thomas, can you take that on?\n> \n> No, at least not now. I'm not able to communicate reliably with the \n> mailing lists, and so can not coordinate anything :( Not sure when or if \n> that will be resolved, but I'll be out of town next week so...\n\n[ Reposted with proper subject line.]\n\nOK, Tom will be away next week, and Thomas will too. I can do it. \nFolks. start sending in those plaform reports, OS name and version\nnumber please.\n\nThe current platform list is:\n\n\thttp://developer.postgresql.org/docs/postgres/supported-platforms.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": "Fri, 25 Oct 2002 22:17:22 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Request for supported platforms"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Folks. start sending in those plaform reports, OS name and version\n> number please.\n\nI've checked CVS tip on:\n\tHPUX 10.20, using both gcc and vendor's cc\n\tPPC Linux\n\tMac OS X 10.1\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 26 Oct 2002 10:18:08 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms "
},
{
"msg_contents": "Tom Lane <tgl@sss.pgh.pa.us> writes:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Folks. start sending in those plaform reports, OS name and version\n> > number please.\n> \n> I've checked CVS tip on:\n> \tHPUX 10.20, using both gcc and vendor's cc\n> \tPPC Linux\n> \tMac OS X 10.1\n\nI get the following on Linux/Sparc, Debian 3.0:\n\nmake[3]: Entering directory `/home/doug/src/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\ngcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -fpic\n-I../../../../../../src/include -c -o ascii_and_mic.o\nascii_and_mic.c\nascii_and_mic.c:19: syntax error before `extern'\nascii_and_mic.c:21: syntax error before `extern'\nmake[3]: *** [ascii_and_mic.o] Error 1\nmake[3]: Leaving directory `/home/doug/src/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\nmake[2]: *** [all] Error 2\nmake[2]: Leaving directory `/home/doug/src/pgsql/src/backend/utils/mb/conversion_procs'\nmake[1]: *** [all] Error 2\nmake[1]: Leaving directory `/home/doug/src/pgsql/src'\nmake: *** [all] Error 2\n\nMy gcc version:\n\ndoug@varsoon:~/src/pgsql$ gcc -v\nReading specs from /usr/lib/gcc-lib/sparc-linux/2.95.4/specs\ngcc version 2.95.4 20011002 (Debian prerelease)\n\nThis is CVS tip as of about 11:30 EST Saturday.\n\nLooking into it, we have in ascii_and_mic.c:\n\nPG_FUNCTION_INFO_V1(ascii_to_mic) \nPG_FUNCTION_INFO_V1(mic_to_ascii) \n\nPutting a semicolon after each such line fixes that compile, but there\nare other files under conversion_procs with the same problem. Is my\ngcc not expanding the macro properly?\n\n-Doug\n",
"msg_date": "26 Oct 2002 12:05:24 -0400",
"msg_from": "Doug McNaught <doug@mcnaught.org>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Doug McNaught <doug@mcnaught.org> writes:\n> make[3]: Entering directory `/home/doug/src/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\n> gcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -fpic\n> -I../../../../../../src/include -c -o ascii_and_mic.o\n> ascii_and_mic.c\n> ascii_and_mic.c:19: syntax error before `extern'\n> ascii_and_mic.c:21: syntax error before `extern'\n\nThat should be fixed as of now.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 26 Oct 2002 13:12:49 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms "
},
{
"msg_contents": "Tom Lane <tgl@sss.pgh.pa.us> writes:\n\n> Doug McNaught <doug@mcnaught.org> writes:\n> > make[3]: Entering directory `/home/doug/src/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\n> > gcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -fpic\n> > -I../../../../../../src/include -c -o ascii_and_mic.o\n> > ascii_and_mic.c\n> > ascii_and_mic.c:19: syntax error before `extern'\n> > ascii_and_mic.c:21: syntax error before `extern'\n> \n> That should be fixed as of now.\n\nOK, compile went fine, but I get multiple regression test failures:\n\ntest geometry ... FAILED\n select_views ... FAILED\n foreign_key ... FAILED\n limit ... FAILED\n plpgsql ... FAILED\n copy2 ... FAILED\n temp ... FAILED\n domain ... FAILED\n rangefuncs ... FAILED\n prepare ... FAILED\n without_oid ... FAILED\n conversion ... FAILED\n truncate ... FAILED\n alter_table ... FAILED\n\nI have attached a gzipped copy of \"regression.diffs\". Let me know if\nI can supply any other help.\n\n-Doug",
"msg_date": "26 Oct 2002 13:40:10 -0400",
"msg_from": "Doug McNaught <doug@mcnaught.org>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nOn 26 Oct 2002, Doug McNaught wrote:\n\n> Tom Lane <tgl@sss.pgh.pa.us> writes:\n>\n> > Doug McNaught <doug@mcnaught.org> writes:\n> > > make[3]: Entering directory `/home/doug/src/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\n> > > gcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -fpic\n> > > -I../../../../../../src/include -c -o ascii_and_mic.o\n> > > ascii_and_mic.c\n> > > ascii_and_mic.c:19: syntax error before `extern'\n> > > ascii_and_mic.c:21: syntax error before `extern'\n> >\n> > That should be fixed as of now.\n>\n> OK, compile went fine, but I get multiple regression test failures:\n>\n> test geometry ... FAILED\n> select_views ... FAILED\n> foreign_key ... FAILED\n> limit ... FAILED\n> plpgsql ... FAILED\n> copy2 ... FAILED\n> temp ... FAILED\n> domain ... FAILED\n> rangefuncs ... FAILED\n> prepare ... FAILED\n> without_oid ... FAILED\n> conversion ... FAILED\n> truncate ... FAILED\n> alter_table ... FAILED\n>\n> I have attached a gzipped copy of \"regression.diffs\". Let me know if\n> I can supply any other help.\n\nThe geometry one looked like rounding issues.\n\nDid you run out of space on where the data directory was mounted?\nAt least some of the other errors were complaining about no space\nleft on device.\n\n\n",
"msg_date": "Sat, 26 Oct 2002 12:06:27 -0700 (PDT)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Doug McNaught <doug@mcnaught.org> writes:\n\n> OK, compile went fine, but I get multiple regression test failures:\n> \n> test geometry ... FAILED\n\nAfter realizing that my disk had filled up (thanks Alvaro) I reran the\ntests and 'geometry' is the only failure. I'm guessing this is due to\nfloating-point differences? If this is OK, then\n\nLinux/Sparc (Debian 3.0) \n\nis a PASS.\n\n-Doug\n",
"msg_date": "26 Oct 2002 15:07:41 -0400",
"msg_from": "Doug McNaught <doug@mcnaught.org>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nUpdated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Folks. start sending in those plaform reports, OS name and version\n> > number please.\n> \n> I've checked CVS tip on:\n> \tHPUX 10.20, using both gcc and vendor's cc\n> \tPPC Linux\n> \tMac OS X 10.1\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": "Sat, 26 Oct 2002 16:01:17 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nUpdated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nDoug McNaught wrote:\n> Doug McNaught <doug@mcnaught.org> writes:\n> \n> > OK, compile went fine, but I get multiple regression test failures:\n> > \n> > test geometry ... FAILED\n> \n> After realizing that my disk had filled up (thanks Alvaro) I reran the\n> tests and 'geometry' is the only failure. I'm guessing this is due to\n> floating-point differences? If this is OK, then\n> \n> Linux/Sparc (Debian 3.0) \n> \n> is a PASS.\n> \n> -Doug\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, 26 Oct 2002 16:04:11 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On Sat, Oct 26, 2002 at 04:01:17PM -0400, Bruce Momjian wrote:\n> \n> Updated:\n> \n> http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\nLinux 2.4 on IA32 passes also.\n\n-- \nAlvaro Herrera (<alvherre[a]dcc.uchile.cl>)\n\"La felicidad no es ma�ana. La felicidad es ahora\"\n",
"msg_date": "Sat, 26 Oct 2002 17:05:37 -0300",
"msg_from": "Alvaro Herrera <alvherre@dcc.uchile.cl>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nAlvaro Herrera wrote:\n> On Sat, Oct 26, 2002 at 04:01:17PM -0400, Bruce Momjian wrote:\n> > \n> > Updated:\n> > \n> > http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n> \n> Linux 2.4 on IA32 passes also.\n> \n> -- \n> Alvaro Herrera (<alvherre[a]dcc.uchile.cl>)\n> \"La felicidad no es ma?ana. La felicidad es ahora\"\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, 26 Oct 2002 16:14:51 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On Fri, 25 Oct 2002, Bruce Momjian wrote:\n\n>\n> We are putting out beta3 today or tomorrow. It seems we are ready to\n> start considering an RC1 date, perhaps next Friday, November 1?\n\ntom is away for the week, so I wouldn't see anything earlier hten the week\nfollowing tha t...\n\n\n>\n> ---------------------------------------------------------------------------\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>\n> Current at ftp://momjian.postgresql.org/pub/postgresql/open_items.\n>\n> Required Changes\n> -------------------\n>\n> Optional Changes\n> ----------------\n> Fix Linux + Perl 5.8.1 + _GNU_SOURCE problem\n> Fix AIX + Large File + Flex problem\n>\n> Documentation Changes\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 6: Have you searched our list archives?\n>\n> http://archives.postgresql.org\n>\n\n",
"msg_date": "Sat, 26 Oct 2002 23:57:26 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": false,
"msg_subject": "Re: Time for RC1 soon?"
},
{
"msg_contents": "On Sat, 26 Oct 2002, Thomas Lockhart wrote:\n\n> >>Seems like someone ought to issue a call for port reports. The\n> >>\"supported platforms\" list hasn't been touched ...\n> > Good point. Thomas, can you take that on?\n>\n> No, at least not now. I'm not able to communicate reliably with the\n> mailing lists, and so can not coordinate anything :( Not sure when or if\n> that will be resolved, but I'll be out of town next week so...\n\nthis should be fixed now tha I've removed the UCE controls ... as for\nbeing out of town next week, so is Tom Lane, so I don't feel tha that is a\nproblem ...\n\n\n",
"msg_date": "Sat, 26 Oct 2002 23:58:34 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": false,
"msg_subject": "Re: Time for RC1 soon?"
},
{
"msg_contents": "On Sat, 26 Oct 2002, Thomas Lockhart wrote:\n\n> ... and just out of curiosity, why does Bruce's message have a [HACKERS]\n> tag in the subject line but my reply does not? It seems to be going\n> through the same mailer but with different results...\n\n*puzzled look* both messages arrived here with the [HACKERS] headin gin\nthe subject ...\n\n\n>\n> - Thomas\n>\n>\n> ---------------------------(end of broadcast)---------------------------\n> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n>\n\n",
"msg_date": "Sat, 26 Oct 2002 23:59:13 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": false,
"msg_subject": "Re: Time for RC1 soon?"
},
{
"msg_contents": "Marc G. Fournier wrote:\n> On Fri, 25 Oct 2002, Bruce Momjian wrote:\n> \n> >\n> > We are putting out beta3 today or tomorrow. It seems we are ready to\n> > start considering an RC1 date, perhaps next Friday, November 1?\n> \n> tom is away for the week, so I wouldn't see anything earlier hten the week\n> following tha t...\n\nI talked to him about that and we both feel this week while he is away\nwill be pretty quiet and the only major thing left is making sure our\ndocs are ready and we have enough platform reports.\n\nIf we can go RC1 soon, we will keep beta to 2 months, which I think is\nacceptable.\n\nTom returns Wednesday so let's see how we are this Thursday/Friday.\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, 26 Oct 2002 23:02:29 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Time for RC1 soon?"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Marc G. Fournier wrote:\n>> tom is away for the week, so I wouldn't see anything earlier hten the week\n>> following tha t...\n\n> I talked to him about that and we both feel this week while he is away\n> will be pretty quiet and the only major thing left is making sure our\n> docs are ready and we have enough platform reports.\n\nUnless we see serious problem reports in the next week, I think RC1 on\nFriday is reasonable. (That assumes people get the platform tests\ndone, of course.)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 27 Oct 2002 12:54:29 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Time for RC1 soon? "
},
{
"msg_contents": "On Saturday 26 October 2002 04:17, Bruce Momjian wrote:\n\n>\n> OK, Tom will be away next week, and Thomas will too. I can do it.\n> Folks. start sending in those plaform reports, OS name and version\n> number please.\n>\n> The current platform list is:\n>\n> \thttp://developer.postgresql.org/docs/postgres/supported-platforms.html\n\nianb > uname -a\nIRIX64 Komma 6.5 07121148 IP27\nianb > gcc -v\nReading specs from /usr/local/lib/gcc-lib/mips-sgi-irix6.5/3.2/specs\nConfigured with: ../configure --with-ld=/usr/bin/ld\nThread model: single\ngcc version 3.2\n\nBuilding 7.3b3; failed on four tests:\nabstime, tinterval, horology, join\n\nThe first three are I presume due to the change to winter time (CET)\nearlier today (27.10), see\nhttp://www.postgresql.org/idocs/index.php?regress-evaluation.html#AEN18382 ;\nthe join failure is a row ordering difference.\n\nShould it be of any use the regression.diffs file is here:\nhttp://home.akademie.de/~IBarwick/IRIX_65_regression.diffs\n\nOtherwise no apparent problems and the server started fine.\n\n\nIan Barwick\nbarwick@gmx.net\n\n",
"msg_date": "Sun, 27 Oct 2002 21:42:26 +0100",
"msg_from": "Ian Barwick <barwick@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nIan Barwick wrote:\n> On Saturday 26 October 2002 04:17, Bruce Momjian wrote:\n> \n> >\n> > OK, Tom will be away next week, and Thomas will too. I can do it.\n> > Folks. start sending in those plaform reports, OS name and version\n> > number please.\n> >\n> > The current platform list is:\n> >\n> > \thttp://developer.postgresql.org/docs/postgres/supported-platforms.html\n> \n> ianb > uname -a\n> IRIX64 Komma 6.5 07121148 IP27\n> ianb > gcc -v\n> Reading specs from /usr/local/lib/gcc-lib/mips-sgi-irix6.5/3.2/specs\n> Configured with: ../configure --with-ld=/usr/bin/ld\n> Thread model: single\n> gcc version 3.2\n> \n> Building 7.3b3; failed on four tests:\n> abstime, tinterval, horology, join\n> \n> The first three are I presume due to the change to winter time (CET)\n> earlier today (27.10), see\n> http://www.postgresql.org/idocs/index.php?regress-evaluation.html#AEN18382 ;\n> the join failure is a row ordering difference.\n> \n> Should it be of any use the regression.diffs file is here:\n> http://home.akademie.de/~IBarwick/IRIX_65_regression.diffs\n> \n> Otherwise no apparent problems and the server started fine.\n> \n> \n> Ian Barwick\n> barwick@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": "Sun, 27 Oct 2002 16:32:07 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Ports list updated:\n\nShouldn't the \"join\" regression test failure be fixed?\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n",
"msg_date": "27 Oct 2002 21:18:15 -0500",
"msg_from": "Neil Conway <neilc@samurai.com>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On an rh6.2, \"Linux gsX 2.4.19-pre6 #4 Thu Apr 11 07:17:39 CEST 2002\nalpha unknown\" (ev56 ruffian) i get this failure.\n\nShould ./configure <options> && gmake && gmake check be the only things\ni need to do?\n\n============== creating database \"regression\" ==============\n/home/mag/d/postgresql/postgresql-7.3b3/src/test/regress/./tmp_check/ins\ntall//usr/local/pgsql/bin/pg_encoding: error in loading shared\nlibraries:\n/home/mag/d/postgresql/postgresql-7.3b3/src/test/regress/./tmp_check/ins\ntall//usr/local/pgsql/bin/pg_encoding: undefined symbol:\npg_char_to_encoding\ncreatedb: \"SQL_ASCII\" is not a valid encoding name\npg_regress: createdb failed\ngmake[2]: *** [check] Error 2\ngmake[2]: Leaving directory\n`/home/mag/d/postgresql/postgresql-7.3b3/src/test/regress'\ngmake[1]: *** [check] Error 2\ngmake[1]: Leaving directory\n`/home/mag/d/postgresql/postgresql-7.3b3/src/test'\ngmake: *** [check] Error 2\n\nMagnus\n\n",
"msg_date": "Mon, 28 Oct 2002 03:34:19 +0100",
"msg_from": "\"Magnus Naeslund(f)\" <mag@fbab.net>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Tom Lane wrote:\n\n> Bruce Momjian writes:\n>\n> >Folks. start sending in those plaform reports, OS name and version\n> >number please.\n>\n>\n> I've checked CVS tip on:\n> \tHPUX 10.20, using both gcc and vendor's cc\n> \tPPC Linux\n> \tMac OS X 10.1\n\n\nHere's the regression.diffs file from Mac OS X 10.2.1 (Jaguar). It \nfailed geometry (rounding) and horology (not sure). gcc -v returns \n\"Apple Computer, Inc. GCC version 1161, based on gcc version 3.1 \n20020420 (prerelease)\".\n\n\n*** ./expected/geometry-powerpc-darwin.out Mon Dec 11 11:45:16 2000\n--- ./results/geometry.out Sun Oct 27 21:27:11 2002\n***************\n*** 127,133 ****\n | (-5,-12) | [(10,-10),(-3,-4)] | \n(-1.60487804878049,-4.64390243902439)\n | (10,10) | [(10,-10),(-3,-4)] | \n(2.39024390243902,-6.48780487804878)\n | (0,0) | [(-1000000,200),(300000,-40)] | \n(0.0028402365895872,15.384614860264)\n! | (-10,0) | [(-1000000,200),(300000,-40)] | \n(-9.99715942258202,15.3864610140472)\n | (-3,4) | [(-1000000,200),(300000,-40)] | \n(-2.99789812267519,15.3851688427303)\n | (5.1,34.5) | [(-1000000,200),(300000,-40)] | \n(5.09647083221496,15.3836744976925)\n | (-5,-12) | [(-1000000,200),(300000,-40)] | \n(-4.99494420845634,15.3855375281616)\n--- 127,133 ----\n | (-5,-12) | [(10,-10),(-3,-4)] | \n(-1.60487804878049,-4.64390243902439)\n | (10,10) | [(10,-10),(-3,-4)] | \n(2.39024390243902,-6.48780487804878)\n | (0,0) | [(-1000000,200),(300000,-40)] | \n(0.0028402365895872,15.384614860264)\n! | (-10,0) | [(-1000000,200),(300000,-40)] | \n(-9.99715942258202,15.3864610140473)\n | (-3,4) | [(-1000000,200),(300000,-40)] | \n(-2.99789812267519,15.3851688427303)\n | (5.1,34.5) | [(-1000000,200),(300000,-40)] | \n(5.09647083221496,15.3836744976925)\n | (-5,-12) | [(-1000000,200),(300000,-40)] | \n(-4.99494420845634,15.3855375281616)\n\n======================================================================\n\n*** ./expected/horology.out Wed Sep 18 17:35:25 2002\n--- ./results/horology.out Sun Oct 27 21:27:12 2002\n***************\n*** 537,549 ****\n SELECT (timestamp with time zone 'today' = (timestamp with time zone \n'tomorrow' - interval '1 day')) as \"True\";\n True\n ------\n! t\n (1 row)\n\n SELECT (timestamp with time zone 'tomorrow' = (timestamp with time \nzone 'yesterday' + interval '2 days')) as \"True\";\n True\n ------\n! t\n (1 row)\n\n SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n--- 537,549 ----\n SELECT (timestamp with time zone 'today' = (timestamp with time zone \n'tomorrow' - interval '1 day')) as \"True\";\n True\n ------\n! f\n (1 row)\n\n SELECT (timestamp with time zone 'tomorrow' = (timestamp with time \nzone 'yesterday' + interval '2 days')) as \"True\";\n True\n ------\n! f\n (1 row)\n\n SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n\n======================================================================\n\n\n-- \nTara Piorkowski\nSystem Administrator, vilaj.com, LLC\n<http://www.vilaj.com/>\n\n",
"msg_date": "Sun, 27 Oct 2002 21:50:38 -0500",
"msg_from": "Tara Piorkowski <tara@vilaj.com>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nI thought the join was part of the timezone change, but I see now that\nit is separate. I will fix it by adding ORDER BY.\n\n---------------------------------------------------------------------------\n\nNeil Conway wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Ports list updated:\n> \n> Shouldn't the \"join\" regression test failure be fixed?\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": "Mon, 28 Oct 2002 09:08:03 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On Fri, Oct 25, 2002 at 10:17:22PM -0400, Bruce Momjian wrote:\n> OK, Tom will be away next week, and Thomas will too. I can do it. \n> Folks. start sending in those plaform reports, OS name and version\n> number please.\n\nSolaris 7, 2xUltraSPARC-II 400 MHz, 2.5 Gig, gcc 2.95.3; all tests\npass. Solaris 8 forthcoming.\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": "Mon, 28 Oct 2002 17:32:04 -0500",
"msg_from": "Andrew Sullivan <andrew@libertyrms.info>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Neil Conway wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Ports list updated:\n> \n> Shouldn't the \"join\" regression test failure be fixed?\n\nOK, I have updated the join regression test to add ORDER BY in the\nqueries Irix differed. Ian, would you retest the CVS version of\nPostgreSQL to see if all the regression tests pass now?\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, 28 Oct 2002 18:08:26 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On Tuesday 29 October 2002 00:08, Bruce Momjian wrote:\n> Neil Conway wrote:\n> > Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > > Ports list updated:\n> >\n> > Shouldn't the \"join\" regression test failure be fixed?\n>\n> OK, I have updated the join regression test to add ORDER BY in the\n> queries Irix differed. Ian, would you retest the CVS version of\n> PostgreSQL to see if all the regression tests pass now?\n\nFor confirmation: IIRC CVS now requires the latest Bison (1.75)?\n(The Irix machine has 1.35).\n\nIan Barwick\nbarwick@gmx.net\n\n \n",
"msg_date": "Tue, 29 Oct 2002 00:25:03 +0100",
"msg_from": "Ian Barwick <barwick@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On Mon, Oct 28, 2002 at 05:32:04PM -0500, Andrew Sullivan wrote:\n> Solaris 7, 2xUltraSPARC-II 400 MHz, 2.5 Gig, gcc 2.95.3; all tests\n> pass. Solaris 8 forthcoming.\n\nI should have mentioned that this is for beta 3; I don't have the\nnecessary bison installed to compile from CVS.\n\nSolaris 8, 2xUltraSPARC-II 400 MHz, 3G RAM, gcc 3.2 also passes.\n\nThe details I'll put in the regression database.\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": "Mon, 28 Oct 2002 18:30:52 -0500",
"msg_from": "Andrew Sullivan <andrew@libertyrms.info>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Sorry, Ian, here is the patch I applied. You can apply this to whatever\nversion you are using and test Irix with that, rather than having to\ngrab CVS.\n\n---------------------------------------------------------------------------\n\nNeil Conway wrote:\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> > Ports list updated:\n> \n> Shouldn't the \"join\" regression test failure be fixed?\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? config.log\n? GNUmakefile\n? config.status\n? autom4te.cache\n? configure.lineno\n? contrib/retep/build\n? contrib/retep/jars\n? src/log\n? src/Makefile.custom\n? src/Makefile.global\n? src/backend/postgres\n? src/backend/catalog/postgres.bki\n? src/backend/catalog/postgres.description\n? src/backend/utils/mb/conversion_procs/conversion_create.sql\n? src/backend/utils/mb/conversion_procs/ascii_and_mic/libascii_and_mic.so.0.0\n? src/backend/utils/mb/conversion_procs/cyrillic_and_mic/libcyrillic_and_mic.so.0.0\n? src/backend/utils/mb/conversion_procs/euc_cn_and_mic/libeuc_cn_and_mic.so.0.0\n? src/backend/utils/mb/conversion_procs/euc_jp_and_sjis/libeuc_jp_and_sjis.so.0.0\n? src/backend/utils/mb/conversion_procs/euc_kr_and_mic/libeuc_kr_and_mic.so.0.0\n? src/backend/utils/mb/conversion_procs/euc_tw_and_big5/libeuc_tw_and_big5.so.0.0\n? src/backend/utils/mb/conversion_procs/latin2_and_win1250/liblatin2_and_win1250.so.0.0\n? src/backend/utils/mb/conversion_procs/latin_and_mic/liblatin_and_mic.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_ascii/libutf8_and_ascii.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_big5/libutf8_and_big5.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_cyrillic/libutf8_and_cyrillic.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_euc_cn/libutf8_and_euc_cn.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_euc_jp/libutf8_and_euc_jp.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_euc_kr/libutf8_and_euc_kr.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_euc_tw/libutf8_and_euc_tw.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_gb18030/libutf8_and_gb18030.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_gbk/libutf8_and_gbk.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_iso8859/libutf8_and_iso8859.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_iso8859_1/libutf8_and_iso8859_1.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_johab/libutf8_and_johab.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_sjis/libutf8_and_sjis.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_tcvn/libutf8_and_tcvn.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_uhc/libutf8_and_uhc.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_win1250/libutf8_and_win1250.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_win1256/libutf8_and_win1256.so.0.0\n? src/backend/utils/mb/conversion_procs/utf8_and_win874/libutf8_and_win874.so.0.0\n? src/bin/initdb/initdb\n? src/bin/initlocation/initlocation\n? src/bin/ipcclean/ipcclean\n? src/bin/pg_config/pg_config\n? src/bin/pg_controldata/pg_controldata\n? src/bin/pg_ctl/pg_ctl\n? src/bin/pg_dump/pg_dump\n? src/bin/pg_dump/pg_restore\n? src/bin/pg_dump/pg_dumpall\n? src/bin/pg_encoding/pg_encoding\n? src/bin/pg_id/pg_id\n? src/bin/pg_resetxlog/pg_resetxlog\n? src/bin/psql/psql\n? src/bin/psql/i\n? src/bin/scripts/createlang\n? src/include/pg_config.h\n? src/include/stamp-h\n? src/interfaces/ecpg/lib/libecpg.so.3.4.0\n? src/interfaces/ecpg/preproc/ecpg\n? src/interfaces/jdbc/build\n? src/interfaces/jdbc/jars\n? src/interfaces/libpq/libpq.so.2.2\n? src/pl/plpgsql/src/libplpgsql.so.1.0\n? src/test/regress/pg_regress\n? src/test/regress/results\n? src/test/regress/regression.out\n? src/test/regress/regression.diffs\n? src/test/regress/expected/copy.out\n? src/test/regress/expected/create_function_1.out\n? src/test/regress/expected/create_function_2.out\n? src/test/regress/expected/misc.out\n? src/test/regress/expected/constraints.out\n? src/test/regress/sql/copy.sql\n? src/test/regress/sql/create_function_1.sql\n? src/test/regress/sql/create_function_2.sql\n? src/test/regress/sql/misc.sql\n? src/test/regress/sql/constraints.sql\nIndex: src/test/regress/expected/join.out\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/test/regress/expected/join.out,v\nretrieving revision 1.11\nretrieving revision 1.12\ndiff -c -c -r1.11 -r1.12\n*** src/test/regress/expected/join.out\t28 Apr 2002 19:54:29 -0000\t1.11\n--- src/test/regress/expected/join.out\t28 Oct 2002 22:54:45 -0000\t1.12\n***************\n*** 1717,1723 ****\n -- Note that OUTER is a noise word\n --\n SELECT '' AS \"xxx\", *\n! FROM J1_TBL LEFT OUTER JOIN J2_TBL USING (i);\n xxx | i | j | t | k \n -----+---+---+-------+----\n | 0 | | zero | \n--- 1717,1724 ----\n -- Note that OUTER is a noise word\n --\n SELECT '' AS \"xxx\", *\n! FROM J1_TBL LEFT OUTER JOIN J2_TBL USING (i)\n! ORDER BY i;\n xxx | i | j | t | k \n -----+---+---+-------+----\n | 0 | | zero | \n***************\n*** 1736,1742 ****\n (13 rows)\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL LEFT JOIN J2_TBL USING (i);\n xxx | i | j | t | k \n -----+---+---+-------+----\n | 0 | | zero | \n--- 1737,1744 ----\n (13 rows)\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL LEFT JOIN J2_TBL USING (i)\n! ORDER BY i;\n xxx | i | j | t | k \n -----+---+---+-------+----\n | 0 | | zero | \n***************\n*** 1785,1791 ****\n (9 rows)\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL FULL OUTER JOIN J2_TBL USING (i);\n xxx | i | j | t | k \n -----+---+---+-------+----\n | 0 | | zero | \n--- 1787,1794 ----\n (9 rows)\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL FULL OUTER JOIN J2_TBL USING (i)\n! ORDER BY i;\n xxx | i | j | t | k \n -----+---+---+-------+----\n | 0 | | zero | \n***************\n*** 1799,1812 ****\n | 6 | 6 | six | \n | 7 | 7 | seven | \n | 8 | 8 | eight | \n | | | null | \n | | 0 | zero | \n | | | | \n- | | | | 0\n (15 rows)\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL FULL JOIN J2_TBL USING (i);\n xxx | i | j | t | k \n -----+---+---+-------+----\n | 0 | | zero | \n--- 1802,1816 ----\n | 6 | 6 | six | \n | 7 | 7 | seven | \n | 8 | 8 | eight | \n+ | | | | 0\n | | | null | \n | | 0 | zero | \n | | | | \n (15 rows)\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL FULL JOIN J2_TBL USING (i)\n! ORDER BY i;\n xxx | i | j | t | k \n -----+---+---+-------+----\n | 0 | | zero | \n***************\n*** 1820,1829 ****\n | 6 | 6 | six | \n | 7 | 7 | seven | \n | 8 | 8 | eight | \n | | | null | \n | | 0 | zero | \n | | | | \n- | | | | 0\n (15 rows)\n \n SELECT '' AS \"xxx\", *\n--- 1824,1833 ----\n | 6 | 6 | six | \n | 7 | 7 | seven | \n | 8 | 8 | eight | \n+ | | | | 0\n | | | null | \n | | 0 | zero | \n | | | | \n (15 rows)\n \n SELECT '' AS \"xxx\", *\nIndex: src/test/regress/sql/join.sql\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/test/regress/sql/join.sql,v\nretrieving revision 1.9\nretrieving revision 1.10\ndiff -c -c -r1.9 -r1.10\n*** src/test/regress/sql/join.sql\t28 Apr 2002 19:54:29 -0000\t1.9\n--- src/test/regress/sql/join.sql\t28 Oct 2002 22:54:45 -0000\t1.10\n***************\n*** 166,175 ****\n --\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL LEFT OUTER JOIN J2_TBL USING (i);\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL LEFT JOIN J2_TBL USING (i);\n \n SELECT '' AS \"xxx\", *\n FROM J1_TBL RIGHT OUTER JOIN J2_TBL USING (i);\n--- 166,177 ----\n --\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL LEFT OUTER JOIN J2_TBL USING (i)\n! ORDER BY i;\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL LEFT JOIN J2_TBL USING (i)\n! ORDER BY i;\n \n SELECT '' AS \"xxx\", *\n FROM J1_TBL RIGHT OUTER JOIN J2_TBL USING (i);\n***************\n*** 178,187 ****\n FROM J1_TBL RIGHT JOIN J2_TBL USING (i);\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL FULL OUTER JOIN J2_TBL USING (i);\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL FULL JOIN J2_TBL USING (i);\n \n SELECT '' AS \"xxx\", *\n FROM J1_TBL LEFT JOIN J2_TBL USING (i) WHERE (k = 1);\n--- 180,191 ----\n FROM J1_TBL RIGHT JOIN J2_TBL USING (i);\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL FULL OUTER JOIN J2_TBL USING (i)\n! ORDER BY i;\n \n SELECT '' AS \"xxx\", *\n! FROM J1_TBL FULL JOIN J2_TBL USING (i)\n! ORDER BY i;\n \n SELECT '' AS \"xxx\", *\n FROM J1_TBL LEFT JOIN J2_TBL USING (i) WHERE (k = 1);",
"msg_date": "Mon, 28 Oct 2002 19:56:29 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nAndrew Sullivan wrote:\n> On Fri, Oct 25, 2002 at 10:17:22PM -0400, Bruce Momjian wrote:\n> > OK, Tom will be away next week, and Thomas will too. I can do it. \n> > Folks. start sending in those plaform reports, OS name and version\n> > number please.\n> \n> Solaris 7, 2xUltraSPARC-II 400 MHz, 2.5 Gig, gcc 2.95.3; all tests\n> pass. Solaris 8 forthcoming.\n> \n> A \n> \n> -- \n> ----\n> Andrew Sullivan 204-4141 Yonge Street\n> Liberty RMS Toronto, Ontario Canada\n> <andrew@libertyrms.info> M2P 2A8\n> +1 416 646 3304 x110\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": "Mon, 28 Oct 2002 21:36:12 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nAndrew Sullivan wrote:\n> On Mon, Oct 28, 2002 at 05:32:04PM -0500, Andrew Sullivan wrote:\n> > Solaris 7, 2xUltraSPARC-II 400 MHz, 2.5 Gig, gcc 2.95.3; all tests\n> > pass. Solaris 8 forthcoming.\n> \n> I should have mentioned that this is for beta 3; I don't have the\n> necessary bison installed to compile from CVS.\n> \n> Solaris 8, 2xUltraSPARC-II 400 MHz, 3G RAM, gcc 3.2 also passes.\n> \n> The details I'll put in the regression database.\n> \n> A\n> \n> -- \n> ----\n> Andrew Sullivan 204-4141 Yonge Street\n> Liberty RMS Toronto, Ontario Canada\n> <andrew@libertyrms.info> M2P 2A8\n> +1 416 646 3304 x110\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": "Mon, 28 Oct 2002 21:38:25 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nStrange. I just got report from another OSX 10.2.1 user saying\nregression tests passed:\n\n\t 10.2.1, Adam Witney (awitney@sghms.ac.uk>\n\nThe proper value seems to be:\n\n\t15.3864610140472\n\nor\n\n\t15.3864610140473\n\nin ./expected/geometry-powerpc-darwin.out. Which is it, folks?\n\n---------------------------------------------------------------------------\n\nTara Piorkowski wrote:\n> Tom Lane wrote:\n> \n> > Bruce Momjian writes:\n> >\n> > >Folks. start sending in those plaform reports, OS name and version\n> > >number please.\n> >\n> >\n> > I've checked CVS tip on:\n> > \tHPUX 10.20, using both gcc and vendor's cc\n> > \tPPC Linux\n> > \tMac OS X 10.1\n> \n> \n> Here's the regression.diffs file from Mac OS X 10.2.1 (Jaguar). It \n> failed geometry (rounding) and horology (not sure). gcc -v returns \n> \"Apple Computer, Inc. GCC version 1161, based on gcc version 3.1 \n> 20020420 (prerelease)\".\n> \n> \n> *** ./expected/geometry-powerpc-darwin.out Mon Dec 11 11:45:16 2000\n> --- ./results/geometry.out Sun Oct 27 21:27:11 2002\n> ***************\n> *** 127,133 ****\n> | (-5,-12) | [(10,-10),(-3,-4)] | \n> (-1.60487804878049,-4.64390243902439)\n> | (10,10) | [(10,-10),(-3,-4)] | \n> (2.39024390243902,-6.48780487804878)\n> | (0,0) | [(-1000000,200),(300000,-40)] | \n> (0.0028402365895872,15.384614860264)\n> ! | (-10,0) | [(-1000000,200),(300000,-40)] | \n> (-9.99715942258202,15.3864610140472)\n> | (-3,4) | [(-1000000,200),(300000,-40)] | \n> (-2.99789812267519,15.3851688427303)\n> | (5.1,34.5) | [(-1000000,200),(300000,-40)] | \n> (5.09647083221496,15.3836744976925)\n> | (-5,-12) | [(-1000000,200),(300000,-40)] | \n> (-4.99494420845634,15.3855375281616)\n> --- 127,133 ----\n> | (-5,-12) | [(10,-10),(-3,-4)] | \n> (-1.60487804878049,-4.64390243902439)\n> | (10,10) | [(10,-10),(-3,-4)] | \n> (2.39024390243902,-6.48780487804878)\n> | (0,0) | [(-1000000,200),(300000,-40)] | \n> (0.0028402365895872,15.384614860264)\n> ! | (-10,0) | [(-1000000,200),(300000,-40)] | \n> (-9.99715942258202,15.3864610140473)\n> | (-3,4) | [(-1000000,200),(300000,-40)] | \n> (-2.99789812267519,15.3851688427303)\n> | (5.1,34.5) | [(-1000000,200),(300000,-40)] | \n> (5.09647083221496,15.3836744976925)\n> | (-5,-12) | [(-1000000,200),(300000,-40)] | \n> (-4.99494420845634,15.3855375281616)\n> \n> ======================================================================\n> \n> *** ./expected/horology.out Wed Sep 18 17:35:25 2002\n> --- ./results/horology.out Sun Oct 27 21:27:12 2002\n> ***************\n> *** 537,549 ****\n> SELECT (timestamp with time zone 'today' = (timestamp with time zone \n> 'tomorrow' - interval '1 day')) as \"True\";\n> True\n> ------\n> ! t\n> (1 row)\n> \n> SELECT (timestamp with time zone 'tomorrow' = (timestamp with time \n> zone 'yesterday' + interval '2 days')) as \"True\";\n> True\n> ------\n> ! t\n> (1 row)\n> \n> SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> --- 537,549 ----\n> SELECT (timestamp with time zone 'today' = (timestamp with time zone \n> 'tomorrow' - interval '1 day')) as \"True\";\n> True\n> ------\n> ! f\n> (1 row)\n> \n> SELECT (timestamp with time zone 'tomorrow' = (timestamp with time \n> zone 'yesterday' + interval '2 days')) as \"True\";\n> True\n> ------\n> ! f\n> (1 row)\n> \n> SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> \n> ======================================================================\n> \n> \n> -- \n> Tara Piorkowski\n> System Administrator, vilaj.com, LLC\n> <http://www.vilaj.com/>\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": "Mon, 28 Oct 2002 22:39:10 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Linux alpha.\n\nAn Redhat 6.2, \"Linux gsX 2.4.19-pre6 #4 Thu Apr 11 07:17:39 CEST 2002\nalpha unknown\" (ev56 ruffian).\n\ngcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)\n\n./configure \\\n --prefix=/usr/local/pgsql-7.3b3 \\\n --with-maxbackends=512 \\\n --with-tcl \\\n --without-tk \\\n --with-perl \\\n --with-openssl \\\n --enable-odbc \\\n --with-CXX\n\nThe box passes every tests except Horology.\nIt's wintertime here aswell, as i read in some other report.\n\nIt installs and runs nicely.\n\nMagnus\n\n\n\n\n\n\n",
"msg_date": "Tue, 29 Oct 2002 04:41:40 +0100",
"msg_from": "\"Magnus Naeslund(f)\" <mag@fbab.net>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nMagnus Naeslund(f) wrote:\n> Linux alpha.\n> \n> An Redhat 6.2, \"Linux gsX 2.4.19-pre6 #4 Thu Apr 11 07:17:39 CEST 2002\n> alpha unknown\" (ev56 ruffian).\n> \n> gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)\n> \n> ./configure \\\n> --prefix=/usr/local/pgsql-7.3b3 \\\n> --with-maxbackends=512 \\\n> --with-tcl \\\n> --without-tk \\\n> --with-perl \\\n> --with-openssl \\\n> --enable-odbc \\\n> --with-CXX\n> \n> The box passes every tests except Horology.\n> It's wintertime here aswell, as i read in some other report.\n> \n> It installs and runs nicely.\n> \n> Magnus\n> \n> \n> \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": "Mon, 28 Oct 2002 22:47:03 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Bruce,\n\nNot sure. I know Tom Lane ran a check on 10.1.x and didn't have any \nerrors, so I figured the differences I saw reflected changes made by \nApple between 10.1.x and 10.2.x. Then, I was surprised when I saw Adam's \nreport, so I thought perhaps he was running on a PPC G3 and there were \ndifferences between G3 and G4 floating point handling (I had run the \noriginal regression tests on a PowerBook G4/400 MHz). So, I ran the same \nexact scenario on my aging iMac (G3/266 Mhz). Those results corroborated \nmy original test (geometry failed; horology cleared up, though, as \nexpected). So, something is different between Adam's system and the two \nI have easy access to. I have confirmed that both my systems are running \n10.2.1. Similarly, the tests were both run on 7.3b3 from the full \ntarball on ftp.us.postgresql.org. The new iMac G3 regression.diffs \noutput is immediately below. Regardless, it would appear that Mac OS X \n10.2.x is well supported.\n\nTara\n\n\n*** ./expected/geometry-powerpc-darwin.out Mon Dec 11 11:45:16 2000\n--- ./results/geometry.out Tue Oct 29 07:59:59 2002\n***************\n*** 127,133 ****\n | (-5,-12) | [(10,-10),(-3,-4)] | \n(-1.60487804878049,-4.64390243902439)\n | (10,10) | [(10,-10),(-3,-4)] | \n(2.39024390243902,-6.48780487804878)\n | (0,0) | [(-1000000,200),(300000,-40)] | \n(0.0028402365895872,15.384614860264)\n! | (-10,0) | [(-1000000,200),(300000,-40)] | \n(-9.99715942258202,15.3864610140472)\n | (-3,4) | [(-1000000,200),(300000,-40)] | \n(-2.99789812267519,15.3851688427303)\n | (5.1,34.5) | [(-1000000,200),(300000,-40)] | \n(5.09647083221496,15.3836744976925)\n | (-5,-12) | [(-1000000,200),(300000,-40)] | \n(-4.99494420845634,15.3855375281616)\n--- 127,133 ----\n | (-5,-12) | [(10,-10),(-3,-4)] | \n(-1.60487804878049,-4.64390243902439)\n | (10,10) | [(10,-10),(-3,-4)] | \n(2.39024390243902,-6.48780487804878)\n | (0,0) | [(-1000000,200),(300000,-40)] | \n(0.0028402365895872,15.384614860264)\n! | (-10,0) | [(-1000000,200),(300000,-40)] | \n(-9.99715942258202,15.3864610140473)\n | (-3,4) | [(-1000000,200),(300000,-40)] | \n(-2.99789812267519,15.3851688427303)\n | (5.1,34.5) | [(-1000000,200),(300000,-40)] | \n(5.09647083221496,15.3836744976925)\n | (-5,-12) | [(-1000000,200),(300000,-40)] | \n(-4.99494420845634,15.3855375281616)\n\n======================================================================\n\n\n\n\nBruce Momjian wrote:\n\n> Strange. I just got report from another OSX 10.2.1 user saying\n> regression tests passed:\n>\n> \t 10.2.1, Adam Witney (awitney@sghms.ac.uk>\n>\n> The proper value seems to be:\n>\n> \t15.3864610140472\n>\n> or\n>\n> \t15.3864610140473\n>\n> in ./expected/geometry-powerpc-darwin.out. Which is it, folks?\n>\n> ---------------------------------------------------------------------------\n>\n> Tara Piorkowski wrote:\n>\n> >Tom Lane wrote:\n> >\n> >\n> >>Bruce Momjian writes:\n> >>\n> >>\n> >>>Folks. start sending in those plaform reports, OS name and version\n> >>>number please.\n> >>\n> >>\n> >>I've checked CVS tip on:\n> >>\tHPUX 10.20, using both gcc and vendor's cc\n> >>\tPPC Linux\n> >>\tMac OS X 10.1\n> >\n> >\n> >Here's the regression.diffs file from Mac OS X 10.2.1 (Jaguar). It\n> >failed geometry (rounding) and horology (not sure). gcc -v returns\n> >\"Apple Computer, Inc. GCC version 1161, based on gcc version 3.1\n> >20020420 (prerelease)\".\n> >\n> >\n> >*** ./expected/geometry-powerpc-darwin.out Mon Dec 11 11:45:16 2000\n> >--- ./results/geometry.out Sun Oct 27 21:27:11 2002\n> >***************\n> >*** 127,133 ****\n> > | (-5,-12) | [(10,-10),(-3,-4)] |\n> >(-1.60487804878049,-4.64390243902439)\n> > | (10,10) | [(10,-10),(-3,-4)] |\n> >(2.39024390243902,-6.48780487804878)\n> > | (0,0) | [(-1000000,200),(300000,-40)] |\n> >(0.0028402365895872,15.384614860264)\n> >! | (-10,0) | [(-1000000,200),(300000,-40)] |\n> >(-9.99715942258202,15.3864610140472)\n> > | (-3,4) | [(-1000000,200),(300000,-40)] |\n> >(-2.99789812267519,15.3851688427303)\n> > | (5.1,34.5) | [(-1000000,200),(300000,-40)] |\n> >(5.09647083221496,15.3836744976925)\n> > | (-5,-12) | [(-1000000,200),(300000,-40)] |\n> >(-4.99494420845634,15.3855375281616)\n> >--- 127,133 ----\n> > | (-5,-12) | [(10,-10),(-3,-4)] |\n> >(-1.60487804878049,-4.64390243902439)\n> > | (10,10) | [(10,-10),(-3,-4)] |\n> >(2.39024390243902,-6.48780487804878)\n> > | (0,0) | [(-1000000,200),(300000,-40)] |\n> >(0.0028402365895872,15.384614860264)\n> >! | (-10,0) | [(-1000000,200),(300000,-40)] |\n> >(-9.99715942258202,15.3864610140473)\n> > | (-3,4) | [(-1000000,200),(300000,-40)] |\n> >(-2.99789812267519,15.3851688427303)\n> > | (5.1,34.5) | [(-1000000,200),(300000,-40)] |\n> >(5.09647083221496,15.3836744976925)\n> > | (-5,-12) | [(-1000000,200),(300000,-40)] |\n> >(-4.99494420845634,15.3855375281616)\n> >\n> >======================================================================\n> >\n> >*** ./expected/horology.out Wed Sep 18 17:35:25 2002\n> >--- ./results/horology.out Sun Oct 27 21:27:12 2002\n> >***************\n> >*** 537,549 ****\n> > SELECT (timestamp with time zone 'today' = (timestamp with time zone\n> >'tomorrow' - interval '1 day')) as \"True\";\n> > True\n> > ------\n> >! t\n> > (1 row)\n> >\n> > SELECT (timestamp with time zone 'tomorrow' = (timestamp with time\n> >zone 'yesterday' + interval '2 days')) as \"True\";\n> > True\n> > ------\n> >! t\n> > (1 row)\n> >\n> > SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> >--- 537,549 ----\n> > SELECT (timestamp with time zone 'today' = (timestamp with time zone\n> >'tomorrow' - interval '1 day')) as \"True\";\n> > True\n> > ------\n> >! f\n> > (1 row)\n> >\n> > SELECT (timestamp with time zone 'tomorrow' = (timestamp with time\n> >zone 'yesterday' + interval '2 days')) as \"True\";\n> > True\n> > ------\n> >! f\n> > (1 row)\n> >\n> > SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> >\n> >======================================================================\n\n\n-- \nTara Piorkowski\nSystem Administrator, vilaj.com, LLC\n<http://www.vilaj.com/>\n\n",
"msg_date": "Tue, 29 Oct 2002 08:22:21 -0500",
"msg_from": "Tara Piorkowski <tara@vilaj.com>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> Strange. I just got report from another OSX 10.2.1 user saying\n> regression tests passed:\n> \t 10.2.1, Adam Witney (awitney@sghms.ac.uk>\n> The proper value seems to be:\n> \t15.3864610140472\n> or\n> \t15.3864610140473\n> in ./expected/geometry-powerpc-darwin.out. Which is it, folks?\n\nThe existing geometry file is exactly correct on my laptop (Powerbook\nG3 using OSX 10.1). I am not sure whether the differences some users\nhave reported are due to hardware or OS version differences. We need\nto figure that out and refine the resultmap, not just change the\nexisting file.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 29 Oct 2002 08:50:46 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms "
},
{
"msg_contents": "On 29/10/02 1:50 pm, \"Tom Lane\" <tgl@sss.pgh.pa.us> wrote:\n\n> Bruce Momjian <pgman@candle.pha.pa.us> writes:\n>> Strange. I just got report from another OSX 10.2.1 user saying\n>> regression tests passed:\n>> 10.2.1, Adam Witney (awitney@sghms.ac.uk>\n>> The proper value seems to be:\n>> 15.3864610140472\n>> or\n>> 15.3864610140473\n>> in ./expected/geometry-powerpc-darwin.out. Which is it, folks?\n> \n> The existing geometry file is exactly correct on my laptop (Powerbook\n> G3 using OSX 10.1). I am not sure whether the differences some users\n> have reported are due to hardware or OS version differences. We need\n> to figure that out and refine the resultmap, not just change the\n> existing file.\n\nI don't have a lot of experience with this stuff, but let me know what to\ntry and I will try it. (Using a Powerbook G4 OSX 10.2.1)\n\nCheers\n\nadam\n\n\n-- \nThis message has been scanned for viruses and\ndangerous content by MailScanner, and is\nbelieved to be clean.\n\n",
"msg_date": "Tue, 29 Oct 2002 13:55:41 +0000",
"msg_from": "Adam Witney <awitney@sghms.ac.uk>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "I noticed that the supported platforms listed 7.2 for linux alpha but\nwith yesterdays date. I figured it was just a typo, but thought I would\ntry to compile myself just to be sure, but I received 4 failures: \nhorology,geometry,opr_sanity, and misc \n\nThis is on alpha running debian 3 linux \nuname -a = Linux usf-cf-alpha-linux-1 2.2.20 #2 Wed Mar 20 19:57:28 EST\n2002 alpha unknown \n\nregression output can be seen at:\nhttp://phppgadmin.sourceforge.net/regression.diffs\nhttp://phppgadmin.sourceforge.net/regression.out \n\nmy alpha history is spotty at best, perhaps someone else can help me\ninterpret things.\n\nRobert Treat\n\nOn Mon, 2002-10-28 at 22:41, Magnus Naeslund(f) wrote:\n> Linux alpha.\n> \n> An Redhat 6.2, \"Linux gsX 2.4.19-pre6 #4 Thu Apr 11 07:17:39 CEST 2002\n> alpha unknown\" (ev56 ruffian).\n> \n> gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)\n> \n> ./configure \\\n> --prefix=/usr/local/pgsql-7.3b3 \\\n> --with-maxbackends=512 \\\n> --with-tcl \\\n> --without-tk \\\n> --with-perl \\\n> --with-openssl \\\n> --enable-odbc \\\n> --with-CXX\n> \n> The box passes every tests except Horology.\n> It's wintertime here aswell, as i read in some other report.\n> \n> It installs and runs nicely.\n> \n> Magnus\n> \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",
"msg_date": "29 Oct 2002 11:49:58 -0500",
"msg_from": "Robert Treat <xzilla@users.sourceforge.net>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Robert Treat <xzilla@users.sourceforge.net> wrote:\n> I noticed that the supported platforms listed 7.2 for linux alpha but\n> with yesterdays date. I figured it was just a typo, but thought I\n> would try to compile myself just to be sure, but I received 4\n> failures: horology,geometry,opr_sanity, and misc\n>\n> This is on alpha running debian 3 linux\n> uname -a = Linux usf-cf-alpha-linux-1 2.2.20 #2 Wed Mar 20 19:57:28\n> EST 2002 alpha unknown\n>\n> regression output can be seen at:\n> http://phppgadmin.sourceforge.net/regression.diffs\n> http://phppgadmin.sourceforge.net/regression.out\n>\n> my alpha history is spotty at best, perhaps someone else can help me\n> interpret things.\n>\n> Robert Treat\n\nI think on Linux Alpha and Sparc alot has to do with the gcc compiler\nversion.\nMy system is handupgraded from Redhat 6.2, kernel 2.2.x to be using a\nmuch newer kernel (2.4.X-aaX).\nBut it's been working flawlessly with almost any kernel from 2.2 up to\nwhere i am now, using postgresql from the 6.x days.\n\nWhat compiler version are you using?\nIs this Debian release a stable branch?\n\nMagnus\n\n\n\n\n",
"msg_date": "Tue, 29 Oct 2002 18:16:15 +0100",
"msg_from": "\"Magnus Naeslund(f)\" <mag@fbab.net>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On Tue, 2002-10-29 at 12:16, Magnus Naeslund(f) wrote:\n> Robert Treat <xzilla@users.sourceforge.net> wrote:\n> > I noticed that the supported platforms listed 7.2 for linux alpha but\n> > with yesterdays date. I figured it was just a typo, but thought I\n> > would try to compile myself just to be sure, but I received 4\n> > failures: horology,geometry,opr_sanity, and misc\n> >\n> > This is on alpha running debian 3 linux\n> > uname -a = Linux usf-cf-alpha-linux-1 2.2.20 #2 Wed Mar 20 19:57:28\n> > EST 2002 alpha unknown\n> >\n> > regression output can be seen at:\n> > http://phppgadmin.sourceforge.net/regression.diffs\n> > http://phppgadmin.sourceforge.net/regression.out\n> >\n> > my alpha history is spotty at best, perhaps someone else can help me\n> > interpret things.\n> >\n> > Robert Treat\n> \n> I think on Linux Alpha and Sparc alot has to do with the gcc compiler\n> version.\n> My system is handupgraded from Redhat 6.2, kernel 2.2.x to be using a\n> much newer kernel (2.4.X-aaX).\n> But it's been working flawlessly with almost any kernel from 2.2 up to\n> where i am now, using postgresql from the 6.x days.\n> \n> What compiler version are you using?\n\nxzilla@usf-cf-alpha-linux-1:~/postgresql-7.3b3/src/test$ gcc --version\n2.95.4\n\n> Is this Debian release a stable branch?\n> \n\nhmm... I'm not an admin on the box, this is one of the boxes from the\nsourceforge compile farm. /proc/version reports this as Debian\nprerelease, and kernel version reports 2.2.20. \n\nRobert Treat\n\n",
"msg_date": "29 Oct 2002 13:23:33 -0500",
"msg_from": "Robert Treat <xzilla@users.sourceforge.net>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On Tuesday 29 October 2002 01:56, Bruce Momjian wrote:\n> Sorry, Ian, here is the patch I applied. You can apply this to whatever\n> version you are using and test Irix with that, rather than having to\n> grab CVS.\n\nOK, I have carried out make check with the updated tests but\ngot FAILED on the same four tests (abstime, tinterval, horology, join).\n\nThe failures are subtly different (see here for new regression diff):\n\nhttp://home.akademie.de/~IBarwick/IRIX_65_1.regression.diffs\n\nHaving looked at them again I see the following:\n- in horology the timestamp tests seem to have succeeded (presumably\n the previous failures were triggered by the change to winter time);\n- the tests which are still failing in abstime, tinterval and horology\n all refer to dates before 1970, where AFAICS they are all out by one hour;\n possibly this explanation?:\n\n\"Some systems using older time zone libraries fail to apply daylight-saving \ncorrections to dates before 1970, causing pre-1970 PDT times to be displayed \nin PST instead. This will result in localized differences in the test \nresults.\"\n\n (cf. http://www.postgresql.org/idocs/index.php?regress-evaluation.html );\n\n- the join tests are failing slightly differently; I would suggest that\n this is because the ORDER BY is still not explicit enough, and for what\n ever reason under IRIX the undefined result row orderings are in a different\n order to every other platform...\n\n e.g. with this statement:\n\nSELECT '' AS \"xxx\", J1_TBL.i, j, t, k\n FROM J1_TBL LEFT OUTER JOIN J2_TBL USING (i)\n ORDER BY i;\n\n xxx | i | j | t | k \n-----+---+---+-------+----\n | 0 | | zero | \n | 1 | 4 | one | -1\n | 2 | 3 | two | 2\n | 2 | 3 | two | 4\n | 3 | 2 | three | -3\n | 4 | 1 | four | \n | 5 | 0 | five | -5\n | 5 | 0 | five | -5\n | 6 | 6 | six | \n | 7 | 7 | seven | \n | 8 | 8 | eight | \n | | 0 | zero | \n | | | null | \n(13 rows)\n\n the order of the last two rows is not defined. The expected order\n according to the regression tests is:\n\n | | | null |\n | | 0 | zero | \n\n\n\nIan Barwick\nbarwick@gmx.net\n\n",
"msg_date": "Tue, 29 Oct 2002 23:52:51 +0100",
"msg_from": "Ian Barwick <barwick@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nI have reviewed your diff and found that it was either timezone changes\nor join.sql which I have recently fixed. Can you grab current CVS\nsnapshot and try again?\n\n---------------------------------------------------------------------------\n\nIan Barwick wrote:\n> On Tuesday 29 October 2002 01:56, Bruce Momjian wrote:\n> > Sorry, Ian, here is the patch I applied. You can apply this to whatever\n> > version you are using and test Irix with that, rather than having to\n> > grab CVS.\n> \n> OK, I have carried out make check with the updated tests but\n> got FAILED on the same four tests (abstime, tinterval, horology, join).\n> \n> The failures are subtly different (see here for new regression diff):\n> \n> http://home.akademie.de/~IBarwick/IRIX_65_1.regression.diffs\n> \n> Having looked at them again I see the following:\n> - in horology the timestamp tests seem to have succeeded (presumably\n> the previous failures were triggered by the change to winter time);\n> - the tests which are still failing in abstime, tinterval and horology\n> all refer to dates before 1970, where AFAICS they are all out by one hour;\n> possibly this explanation?:\n> \n> \"Some systems using older time zone libraries fail to apply daylight-saving \n> corrections to dates before 1970, causing pre-1970 PDT times to be displayed \n> in PST instead. This will result in localized differences in the test \n> results.\"\n> \n> (cf. http://www.postgresql.org/idocs/index.php?regress-evaluation.html );\n> \n> - the join tests are failing slightly differently; I would suggest that\n> this is because the ORDER BY is still not explicit enough, and for what\n> ever reason under IRIX the undefined result row orderings are in a different\n> order to every other platform...\n> \n> e.g. with this statement:\n> \n> SELECT '' AS \"xxx\", J1_TBL.i, j, t, k\n> FROM J1_TBL LEFT OUTER JOIN J2_TBL USING (i)\n> ORDER BY i;\n> \n> xxx | i | j | t | k \n> -----+---+---+-------+----\n> | 0 | | zero | \n> | 1 | 4 | one | -1\n> | 2 | 3 | two | 2\n> | 2 | 3 | two | 4\n> | 3 | 2 | three | -3\n> | 4 | 1 | four | \n> | 5 | 0 | five | -5\n> | 5 | 0 | five | -5\n> | 6 | 6 | six | \n> | 7 | 7 | seven | \n> | 8 | 8 | eight | \n> | | 0 | zero | \n> | | | null | \n> (13 rows)\n> \n> the order of the last two rows is not defined. The expected order\n> according to the regression tests is:\n> \n> | | | null |\n> | | 0 | zero | \n> \n> \n> \n> Ian Barwick\n> barwick@gmx.net\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, 30 Oct 2002 17:24:26 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "> Folks. start sending in those plaform reports, OS name and version\n> number please.\n\nDOESN'T WORK on Digital Unix/Tru64 4.0g, with both cc or gcc compiler.\n\nUsing Compaq C V6.4-216 (dtk) on Digital UNIX V4.0G (Rev. 1530)\nCompiler Driver V6.4-013 (dtk) cc Driver:\n\nmake[3]: Entering directory\n`/usr/local/src/postgresql-7.3b3/src/backend/main'\ncc -std -O4 -Olimit 2000 -I../../../src/include -I/usr/local/include -c\n-o main.o main.c\ncc: Error: main.c, line 83: In this statement, \"errno\" is not declared.\n(undeclared)\n fprintf(stderr, gettext(\"%s: setsysinfo failed: %s\\n\"),\nargv[0], strerror(errno));\n------------------------------------------------------------------------------------------^\nmake[3]: *** [main.o] Error 1\n\nsame with GCC 2.95.1.\n\nmake[1]: Entering directory\n`/usr/local/src/postgresql-7.3b3-gcc/src/backend/main'\ngcc -Wall -Wmissing-prototypes -Wmissing-declarations\n-I../../../src/include -I/usr/local/include -c -o main.o main.c\n\nmain.c: In function `main':\nmain.c:83: `errno' undeclared (first use in this function)\nmake[1]: *** [main.o] Error 1\n\nSo, errno function is undefined.\n\nThis is quite strange, because that section hasn't been changed in the\nlast few months. It's activated from two different #if: one is \n#if defined(__alpha)\nother is \n#if defined(NOFIXADE) || defined(NOPRINTADE)\nMaybe is the setting of NOFIXADE or NOPRINTADE to be changed upstream?\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": "31 Oct 2002 14:06:05 +0200",
"msg_from": "Alessio Bragadini <alessio@albourne.com>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Alessio Bragadini <alessio@albourne.com> writes:\n>> Folks. start sending in those plaform reports, OS name and version\n>> number please.\n\n> DOESN'T WORK on Digital Unix/Tru64 4.0g, with both cc or gcc compiler.\n\nEvidently main.c needs \"#include <errno.h>\" added. Please add that and\nsee if you get any further. There might be other files with the same\nproblem?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 31 Oct 2002 11:53:15 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms "
},
{
"msg_contents": "> > >>Seems like someone ought to issue a call for port reports. The\n> > >>\"supported platforms\" list hasn't been touched ...\n> > > Good point. Thomas, can you take that on?\n> > \n> > No, at least not now. I'm not able to communicate reliably with the \n> > mailing lists, and so can not coordinate anything :( Not sure when or if \n> > that will be resolved, but I'll be out of town next week so...\n> \n> [ Reposted with proper subject line.]\n> \n> OK, Tom will be away next week, and Thomas will too. I can do it. \n> Folks. start sending in those plaform reports, OS name and version\n> number please.\n> \n> The current platform list is:\n> \n> \thttp://developer.postgresql.org/docs/postgres/supported-platforms.html\n\n$ uname -a\n\nFreeBSD avienda.nxad.com 5.0-CURRENT FreeBSD 5.0-CURRENT #1: Mon Oct 28 18:20:14 PST 2002 root@avienda.nxad.com:/usr/obj/usr/src/sys/DELLAPTOP i386\n\n$ gcc -v\nUsing built-in specs.\nConfigured with: FreeBSD/i386 system compiler\nThread model: posix\ngcc version 3.2.1 [FreeBSD] 20021009 (prerelease)\n\n\nLooks like the only problem on beta3 is that the geometry bits are\nfailing, but I'm not 100% if they haven't already been solved. -sc\n\n\n*** ./expected/geometry-positive-zeros-bsd.out\tTue Sep 12 14:07:16 2000\n--- ./results/geometry.out\tThu Oct 31 23:53:31 2002\n***************\n*** 114,120 ****\n | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n | (-5,-12) | [(1,2),(3,4)] | (1,2)\n | (10,10) | [(1,2),(3,4)] | (3,4)\n! | (0,0) | [(0,0),(6,6)] | (0,0)\n | (-10,0) | [(0,0),(6,6)] | (0,0)\n | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n--- 114,120 ----\n | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n | (-5,-12) | [(1,2),(3,4)] | (1,2)\n | (10,10) | [(1,2),(3,4)] | (3,4)\n! | (0,0) | [(0,0),(6,6)] | (-0,0)\n | (-10,0) | [(0,0),(6,6)] | (0,0)\n | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n***************\n*** 224,233 ****\n twentyfour | rotation \n ------------+-----------------------------\n | (0,0),(0,0)\n! | (0,0),(-20,-20)\n! | (0,2),(-14,0)\n | (0,79.2),(-58.8,0)\n! | (14,0),(0,-34)\n | (0,40),(0,0)\n | (0,0),(0,0)\n | (-10,-10),(-30,-30)\n--- 224,233 ----\n twentyfour | rotation \n ------------+-----------------------------\n | (0,0),(0,0)\n! | (-0,0),(-20,-20)\n! | (-0,2),(-14,0)\n | (0,79.2),(-58.8,0)\n! | (14,-0),(0,-34)\n | (0,40),(0,0)\n | (0,0),(0,0)\n | (-10,-10),(-30,-30)\n***************\n*** 254,264 ****\n WHERE (p.f1 <-> point '(0,0)') >= 1;\n twenty | rotation \n --------+-----------------------------------------------------------------------------------\n! | (0,0),(-0.2,-0.2)\n | (-0.1,-0.1),(-0.3,-0.3)\n | (-0.25,-0.25),(-0.25,-0.35)\n | (-0.3,-0.3),(-0.3,-0.3)\n! | (0.08,0),(0,-0.56)\n | (0.12,-0.28),(0.04,-0.84)\n | (0.26,-0.7),(0.1,-0.82)\n | (0.12,-0.84),(0.12,-0.84)\n--- 254,264 ----\n WHERE (p.f1 <-> point '(0,0)') >= 1;\n twenty | rotation \n --------+-----------------------------------------------------------------------------------\n! | (0,-0),(-0.2,-0.2)\n | (-0.1,-0.1),(-0.3,-0.3)\n | (-0.25,-0.25),(-0.25,-0.35)\n | (-0.3,-0.3),(-0.3,-0.3)\n! | (0.08,-0),(0,-0.56)\n | (0.12,-0.28),(0.04,-0.84)\n | (0.26,-0.7),(0.1,-0.82)\n | (0.12,-0.84),(0.12,-0.84)\n***************\n*** 266,272 ****\n | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n! | (0,0.0828402366863905),(-0.201183431952663,0)\n | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n--- 266,272 ----\n | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n! | (-0,0.0828402366863905),(-0.201183431952663,0)\n | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n\n======================================================================\n\n\n\n-- \nSean Chittenden\n",
"msg_date": "Thu, 31 Oct 2002 23:57:08 -0800",
"msg_from": "Sean Chittenden <sean@chittenden.org>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On Thu, 2002-10-31 at 18:53, Tom Lane wrote:\n\n> Evidently main.c needs \"#include <errno.h>\" added. \n\nI wonder what have changed since Beta2 that compiled fine...\n\n> Please add that and see if you get any further.\n\nDone, and now it builds (I've limited the test to native cc compiler for\nnow). But it doesn't pass two regression tests, float8 and alter_table.\n\nThe diffs seem to me non-trivial, I've attached the results.\n\nAny idea?\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",
"msg_date": "01 Nov 2002 10:25:41 +0200",
"msg_from": "Alessio Bragadini <alessio@albourne.com>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On Fri, 2002-11-01 at 01:57, Sean Chittenden wrote:\n> > > >>Seems like someone ought to issue a call for port reports. The\n> > > >>\"supported platforms\" list hasn't been touched ...\n> > > > Good point. Thomas, can you take that on?\n> > > \n> > > No, at least not now. I'm not able to communicate reliably with the \n> > > mailing lists, and so can not coordinate anything :( Not sure when or if \n> > > that will be resolved, but I'll be out of town next week so...\n> > \n> > [ Reposted with proper subject line.]\n> > \n> > OK, Tom will be away next week, and Thomas will too. I can do it. \n> > Folks. start sending in those plaform reports, OS name and version\n> > number please.\n> > \n> > The current platform list is:\n> > \n> > \thttp://developer.postgresql.org/docs/postgres/supported-platforms.html\n> \n> $ uname -a\n> \n> FreeBSD avienda.nxad.com 5.0-CURRENT FreeBSD 5.0-CURRENT #1: Mon Oct 28 18:20:14 PST 2002 root@avienda.nxad.com:/usr/obj/usr/src/sys/DELLAPTOP i386\n> \n> $ gcc -v\n> Using built-in specs.\n> Configured with: FreeBSD/i386 system compiler\n> Thread model: posix\n> gcc version 3.2.1 [FreeBSD] 20021009 (prerelease)\n> \n> \n> Looks like the only problem on beta3 is that the geometry bits are\n> failing, but I'm not 100% if they haven't already been solved. -sc\nCan you check it against the geometry-bsd[i]-precision.out file? \n\n(depending on whether you've updated since the weekend). \n\nIf it matches, we need to update the resultmap. \n\nLER\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": "01 Nov 2002 05:54:20 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Alessio Bragadini <alessio@albourne.com> writes:\n> On Thu, 2002-10-31 at 18:53, Tom Lane wrote:\n>> Evidently main.c needs \"#include <errno.h>\" added. \n\n> I wonder what have changed since Beta2 that compiled fine...\n\nThat is odd. main.c itself certainly has not changed.\n\n> Done, and now it builds (I've limited the test to native cc compiler for\n> now). But it doesn't pass two regression tests, float8 and alter_table.\n> The diffs seem to me non-trivial, I've attached the results.\n\nThe float8 diff appears to be because new test cases were added to\nfloat8.sql, but not all the float8 output variants were updated to match\n:-(. I will take care of that.\n\nThe alter_table diff is clearly indicative of a bug. Can you set a\nbreakpoint and discover exactly where calloc is blowing up? (For lack\nof a better idea, I'm going to guess that calloc rejects a zero argument\non your platform.)\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 01 Nov 2002 09:27:47 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms "
},
{
"msg_contents": "I said:\n> The alter_table diff is clearly indicative of a bug. Can you set a\n> breakpoint and discover exactly where calloc is blowing up?\n\nNever mind, I see it: psql is doing calloc(0) when presented with a\ntable of zero columns. Surprising you're the first to try it on a\nplatform where that returns NULL.\n\nI've committed fixes for the other issues too, so CVS tip should now\npass cleanly on your platform.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 01 Nov 2002 10:22:42 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms "
},
{
"msg_contents": "Sean Chittenden <sean@chittenden.org> writes:\n> $ uname -a\n> FreeBSD avienda.nxad.com 5.0-CURRENT FreeBSD 5.0-CURRENT #1: Mon Oct 28 18:20:14 PST 2002 root@avienda.nxad.com:/usr/obj/usr/src/sys/DELLAPTOP i386\n\n> Looks like the only problem on beta3 is that the geometry bits are\n> failing, but I'm not 100% if they haven't already been solved. -sc\n\nHmm. Evidently you now have support for minus-zero. It looks like we\nhave an updated comparison file for that case for FreeBSD, but it's only\nbeing applied for FreeBSD 4.7:\n\ngeometry/i.86-.*-freebsd4.7=geometry-bsd-precision\ngeometry/i.86-.*-freebsd=geometry-positive-zeros-bsd\ngeometry/alpha.*-freebsd=geometry-positive-zeros\n\nOr at least it's *trying* to apply it for 4.7 --- as near as I can tell\nwithout testing, the above scrap of resultmap code is wrong because both\nof the i.86 lines will match on FreeBSD 4.7, and I think the pg_regress\ncoding will take the last match. Larry, did you actually test the\nCVS-tip resultmap to make sure it picks the right comparison file on\nyour box?\n\nWe could possibly do\n\ngeometry/i.86-.*-freebsd=geometry-positive-zeros-bsd\ngeometry/i.86-.*-freebsd4.7=geometry-bsd-precision\ngeometry/i.86-.*-freebsd5=geometry-bsd-precision\ngeometry/alpha.*-freebsd=geometry-positive-zeros\n\nwhich is mighty ugly, but I'm hopeful that by the next PG release we'll\nhave gotten rid of most of the platform-to-platform geometry variants\nanyway.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 01 Nov 2002 10:53:34 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms "
},
{
"msg_contents": "On Fri, 2002-11-01 at 09:53, Tom Lane wrote:\n> Sean Chittenden <sean@chittenden.org> writes:\n> > $ uname -a\n> > FreeBSD avienda.nxad.com 5.0-CURRENT FreeBSD 5.0-CURRENT #1: Mon Oct 28 18:20:14 PST 2002 root@avienda.nxad.com:/usr/obj/usr/src/sys/DELLAPTOP i386\n> \n> > Looks like the only problem on beta3 is that the geometry bits are\n> > failing, but I'm not 100% if they haven't already been solved. -sc\n> \n> Hmm. Evidently you now have support for minus-zero. It looks like we\n> have an updated comparison file for that case for FreeBSD, but it's only\n> being applied for FreeBSD 4.7:\n> \n> geometry/i.86-.*-freebsd4.7=geometry-bsd-precision\n> geometry/i.86-.*-freebsd=geometry-positive-zeros-bsd\n> geometry/alpha.*-freebsd=geometry-positive-zeros\n> \n> Or at least it's *trying* to apply it for 4.7 --- as near as I can tell\n> without testing, the above scrap of resultmap code is wrong because both\n> of the i.86 lines will match on FreeBSD 4.7, and I think the pg_regress\n> coding will take the last match. Larry, did you actually test the\n> CVS-tip resultmap to make sure it picks the right comparison file on\n> your box?\nYes, just did and it *FAILS*. \n\nyou need the order you have below. \n\nSorry...\n\n \n\n> \n> We could possibly do\n> \n> geometry/i.86-.*-freebsd=geometry-positive-zeros-bsd\n> geometry/i.86-.*-freebsd4.7=geometry-bsd-precision\n> geometry/i.86-.*-freebsd5=geometry-bsd-precision\n> geometry/alpha.*-freebsd=geometry-positive-zeros\n> \n> which is mighty ugly, but I'm hopeful that by the next PG release we'll\n> have gotten rid of most of the platform-to-platform geometry variants\n> anyway.\n> \n> \t\t\tregards, tom lane\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": "01 Nov 2002 10:25:28 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Larry Rosenman <ler@lerctr.org> writes:\n> On Fri, 2002-11-01 at 09:53, Tom Lane wrote:\n>> Or at least it's *trying* to apply it for 4.7 --- as near as I can tell\n>> without testing, the above scrap of resultmap code is wrong because both\n>> of the i.86 lines will match on FreeBSD 4.7, and I think the pg_regress\n>> coding will take the last match. Larry, did you actually test the\n>> CVS-tip resultmap to make sure it picks the right comparison file on\n>> your box?\n\n> Yes, just did and it *FAILS*. \n> you need the order you have below. \n> Sorry...\n\nOkay, I've updated the CVS tip to look like this:\n\ngeometry/i.86-.*-freebsd=geometry-positive-zeros-bsd\ngeometry/i.86-.*-freebsd4.7=geometry-bsd-precision\ngeometry/i.86-.*-freebsd5=geometry-bsd-precision\ngeometry/alpha.*-freebsd=geometry-positive-zeros\n\nSean, would you verify this works for you?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 01 Nov 2002 11:35:15 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms "
},
{
"msg_contents": "On Fri, Nov 01, 2002 at 10:22:42AM -0500, Tom Lane wrote:\n\n> I've committed fixes for the other issues too, so CVS tip should now\n> pass cleanly on your platform.\n\nI've built the snapshot from CVS tip and now the regression tests pass\nwith both GCC 2.95.1 and Compaq C V6.4-216 (dtk) on \nDigital UNIX V4.0G (Rev. 1530).\n\nThank you very much for your help!\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",
"msg_date": "Mon, 4 Nov 2002 21:46:34 +0200",
"msg_from": "Alessio Bragadini <alessio@albourne.com>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nAlessio Bragadini wrote:\n> On Fri, Nov 01, 2002 at 10:22:42AM -0500, Tom Lane wrote:\n> \n> > I've committed fixes for the other issues too, so CVS tip should now\n> > pass cleanly on your platform.\n> \n> I've built the snapshot from CVS tip and now the regression tests pass\n> with both GCC 2.95.1 and Compaq C V6.4-216 (dtk) on \n> Digital UNIX V4.0G (Rev. 1530).\n> \n> Thank you very much for your help!\n> \n> -- \n> Alessio F. Bragadini\t\talessio@albourne.com\n> APL Financial Services\t\thttp://village.albourne.com\n> Nicosia, 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> ---------------------------(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": "Mon, 4 Nov 2002 17:40:05 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "> >> Or at least it's *trying* to apply it for 4.7 --- as near as I\n> >> can tell without testing, the above scrap of resultmap code is\n> >> wrong because both of the i.86 lines will match on FreeBSD 4.7,\n> >> and I think the pg_regress coding will take the last match.\n> >> Larry, did you actually test the CVS-tip resultmap to make sure\n> >> it picks the right comparison file on your box?\n> \n> > Yes, just did and it *FAILS*. you need the order you have below.\n> > Sorry...\n> \n> Okay, I've updated the CVS tip to look like this:\n> \n> geometry/i.86-.*-freebsd=geometry-positive-zeros-bsd\n> geometry/i.86-.*-freebsd4.7=geometry-bsd-precision\n> geometry/i.86-.*-freebsd5=geometry-bsd-precision\n> geometry/alpha.*-freebsd=geometry-positive-zeros\n> \n> Sean, would you verify this works for you?\n\nIt does, thank you. I've just updated the -devel port to 7.3b4,\nhopefully the mirrors will pick up the bits soon. -sc\n\n-- \nSean Chittenden\n",
"msg_date": "Mon, 4 Nov 2002 18:14:28 -0800",
"msg_from": "Sean Chittenden <sean@chittenden.org>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nSean Chittenden wrote:\n> > >> Or at least it's *trying* to apply it for 4.7 --- as near as I\n> > >> can tell without testing, the above scrap of resultmap code is\n> > >> wrong because both of the i.86 lines will match on FreeBSD 4.7,\n> > >> and I think the pg_regress coding will take the last match.\n> > >> Larry, did you actually test the CVS-tip resultmap to make sure\n> > >> it picks the right comparison file on your box?\n> > \n> > > Yes, just did and it *FAILS*. you need the order you have below.\n> > > Sorry...\n> > \n> > Okay, I've updated the CVS tip to look like this:\n> > \n> > geometry/i.86-.*-freebsd=geometry-positive-zeros-bsd\n> > geometry/i.86-.*-freebsd4.7=geometry-bsd-precision\n> > geometry/i.86-.*-freebsd5=geometry-bsd-precision\n> > geometry/alpha.*-freebsd=geometry-positive-zeros\n> > \n> > Sean, would you verify this works for you?\n> \n> It does, thank you. I've just updated the -devel port to 7.3b4,\n> hopefully the mirrors will pick up the bits soon. -sc\n> \n> -- \n> Sean Chittenden\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, 4 Nov 2002 22:17:51 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "On Tue, 2002-11-05 at 00:40, Bruce Momjian wrote:\n\n> Ports list updated:\n> \n> http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\nPlease note that you have an entry for Digital Unix and one for Compaq\nTru64 while in fact they are the same OS that went through a whirlwind\nof name changes. Don't know if official name should be now \"HP Tru64\" or\nsome other...\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": "05 Nov 2002 13:05:47 +0200",
"msg_from": "Alessio Bragadini <alessio@albourne.com>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nYes, now it is Digital/HP/Compaq.\n\nUpdated to existing Tru64 entry.\n\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\n\nAlessio Bragadini wrote:\n> On Tue, 2002-11-05 at 00:40, Bruce Momjian wrote:\n> \n> > Ports list updated:\n> > \n> > http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n> \n> Please note that you have an entry for Digital Unix and one for Compaq\n> Tru64 while in fact they are the same OS that went through a whirlwind\n> of name changes. Don't know if official name should be now \"HP Tru64\" or\n> some other...\n> \n> -- \n> Alessio F. Bragadini\t\talessio@albourne.com\n> APL Financial Services\t\thttp://village.albourne.com\n> Nicosia, 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> \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, 5 Nov 2002 12:31:49 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
}
] |
[
{
"msg_contents": "Without specifying the -Xb switch to kill the C99 interpretation of\ninline, I get the following from current CVS:\n\nUX:acomp: ERROR: \"tuplesort.c\", line 1854: \"inline\" functions cannot use\n\"static\" identifier: myFunctionCall2\nUX:acomp: ERROR: \"tuplesort.c\", line 1856: \"inline\" functions cannot use\n\"static\" identifier: myFunctionCall2\nUX:acomp: ERROR: \"tuplesort.c\", line 1870: \"inline\" functions cannot use\n\"static\" identifier: myFunctionCall2\nUX:acomp: ERROR: \"tuplesort.c\", line 1872: \"inline\" functions cannot use\n\"static\" identifier: myFunctionCall2\nUX:acomp: ERROR: \"tuplesort.c\", line 1885: \"inline\" functions cannot use\n\"static\" identifier: myFunctionCall2\nUX:acomp: ERROR: \"tuplesort.c\", line 1897: \"inline\" functions cannot use\n\"static\" identifier: myFunctionCall2\n\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": "26 Oct 2002 09:50:52 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "UnixWare 7.1.3 (BETA), C99 compiler, current CVS, error..."
},
{
"msg_contents": "Larry Rosenman <ler@lerctr.org> writes:\n> Without specifying the -Xb switch to kill the C99 interpretation of\n> inline, I get the following from current CVS:\n\n> UX:acomp: ERROR: \"tuplesort.c\", line 1854: \"inline\" functions cannot use\n> \"static\" identifier: myFunctionCall2\n\nI don't understand what it's unhappy about. My C99 draft sez\n\n [#6] Any function with internal linkage can be an inline\n function.\n\nso the text of the message is surely not what they are really\ncomplaining about? Or is the compiler broken?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 26 Oct 2002 11:07:13 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: UnixWare 7.1.3 (BETA), C99 compiler, current CVS, error... "
},
{
"msg_contents": "On Sat, 2002-10-26 at 10:07, Tom Lane wrote:\n> Larry Rosenman <ler@lerctr.org> writes:\n> > Without specifying the -Xb switch to kill the C99 interpretation of\n> > inline, I get the following from current CVS:\n> \n> > UX:acomp: ERROR: \"tuplesort.c\", line 1854: \"inline\" functions cannot use\n> > \"static\" identifier: myFunctionCall2\n> \n> I don't understand what it's unhappy about. My C99 draft sez\n> \n> [#6] Any function with internal linkage can be an inline\n> function.\n> \n> so the text of the message is surely not what they are really\n> complaining about? Or is the compiler broken?\nI'll ask, it is Beta (although the Compiler has done this since the C99\nfunctionality was added, and it causes a LOT of open source stuff to\nrequire -Xb). \n\n\n> \n> \t\t\tregards, tom lane\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\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": "26 Oct 2002 10:09:47 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "Re: UnixWare 7.1.3 (BETA), C99 compiler, current CVS,"
},
{
"msg_contents": "Larry Rosenman <ler@lerctr.org> writes:\n>> so the text of the message is surely not what they are really\n>> complaining about? Or is the compiler broken?\n> I'll ask, it is Beta (although the Compiler has done this since the C99\n> functionality was added, and it causes a LOT of open source stuff to\n> require -Xb). \n\nAfter reading a little further, it seems that the brain damage is in the\nstandard, not the compiler :-(. It looks like C99's notion of a\nfunction that is both global and inline is that you must provide *two*\ndefinitions of the function, one marked inline and one not; moreover,\nthese must appear in separate translation units. What in the world were\nthose people smoking? That's a recipe for maintenance problems (edit\none definition, forget to edit the other), not to mention completely at\nvariance with the de facto standard behavior of inline that's been\naround for a long time.\n\nMy inclination is to change the code for ApplySortFunction to look like\n\n\t#if defined(__GNUC__)\n\t__inline__\n\t#endif\n\tint32\n\tApplySortFunction\n\nso that the inline optimization only gets done for gcc, which we know\ninterprets inline sanely. Anyone see a better answer?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 26 Oct 2002 11:30:55 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: UnixWare 7.1.3 (BETA), C99 compiler, current CVS, error... "
}
] |
[
{
"msg_contents": "With cc -Xb, we get farther, but still die.... (CURRENT CVS):\n\ncc -Xb -O -g -K PIC -I../../../../../../src/include -I/usr/local/include -c -o ascii_and_mic.o ascii_and_mic.c\nUX:cc: WARNING: debugging and optimization mutually exclusive; -O disabled\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: syntax error, probably missing \",\", \";\" or \"=\"\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: only \"register\" valid as formal parameter storage class\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: parameter not in identifier list: pg_finfo_mic_to_ascii\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: parameter not in identifier list: pg_finfo_mic_to_ascii\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: Syntax error before or at: {\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: only \"register\" valid as formal parameter storage class\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: parameter not in identifier list: my_finfo\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: cannot initialize parameter: my_finfo\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 19: statement not reached\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: Syntax error before or at: return\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: parameter not in identifier list: my_finfo\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 19: only \"register\" valid as formal parameter storage class\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 21: parameter not in identifier list: no_such_variable\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 21: only \"register\" valid as formal parameter storage class\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 21: parameter not in identifier list: ascii_to_mic\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 22: only \"register\" valid as formal parameter storage class\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 22: parameter not in identifier list: mic_to_ascii\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 37: parameter not in identifier list: ascii_to_mic\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 37: Syntax error before or at: {\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 38: parameter not in identifier list: src\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 38: cannot initialize parameter: src\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 38: undefined symbol: fcinfo\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 38: left operand of \"->\" must be pointer to struct/union\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 38: assignment type mismatch\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 39: parameter not in identifier list: dest\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 39: cannot initialize parameter: dest\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 39: left operand of \"->\" must be pointer to struct/union\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 39: assignment type mismatch\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 40: parameter not in identifier list: len\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 40: cannot initialize parameter: len\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 40: left operand of \"->\" must be pointer to struct/union\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 42: Syntax error before or at: do\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 42: Syntax error before or at: &&\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 42: Syntax error before or at: (\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 42: function cannot return function or array\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 42: parameter not in identifier list: int32\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 42: Syntax error before or at: &\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 42: parameter not in identifier list: PG_SQL_ASCII\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 42: parameter not in identifier list: ExceptionalCondition\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 42: syntax error: empty declaration\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 42: syntax error: empty declaration\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 43: Syntax error before or at: &&\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 43: Syntax error before or at: (\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 43: function cannot return function or array\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 43: parameter not in identifier list: int32\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 43: Syntax error before or at: &\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 43: parameter not in identifier list: PG_MULE_INTERNAL\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 43: parameter not in identifier list: ExceptionalCondition\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 43: syntax error: empty declaration\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 43: syntax error: empty declaration\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 44: Syntax error before or at: &&\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 44: parameter not in identifier list: ExceptionalCondition\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 44: syntax error: empty declaration\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 44: syntax error: empty declaration\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 46: function prototype parameters must have types\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 46: parameter not in identifier list: pg_ascii2mic\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 48: Syntax error before or at: return\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 48: parameter not in identifier list: Datum\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 48: Syntax error before or at: 0\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 48: syntax error: empty declaration\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 52: parameter not in identifier list: Datum\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 53: parameter not in identifier list: mic_to_ascii\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 54: parameter not in identifier list: src\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 54: cannot initialize parameter: src\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 54: left operand of \"->\" must be pointer to struct/union\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 54: assignment type mismatch\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 55: parameter not in identifier list: dest\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 55: cannot initialize parameter: dest\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 55: left operand of \"->\" must be pointer to struct/union\nUX:acomp: WARNING: \"ascii_and_mic.c\", line 55: assignment type mismatch\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 56: parameter not in identifier list: len\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 56: cannot initialize parameter: len\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 56: left operand of \"->\" must be pointer to struct/union\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 56: function designator is not of function type\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 56: operands must have integral type: op \"&\"\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 58: left operand of \"->\" must be pointer to struct/union\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 58: function designator is not of function type\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 58: operands must have integral type: op \"&\"\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 59: left operand of \"->\" must be pointer to struct/union\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 59: function designator is not of function type\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 59: operands must have integral type: op \"&\"\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 64: Syntax error before or at: 0\nUX:acomp: ERROR: \"ascii_and_mic.c\", line 65: cannot recover from previous errors\ngmake[3]: *** [ascii_and_mic.o] Error 1\ngmake[3]: Leaving directory `/home/ler/pg-dev/pgsql/src/backend/utils/mb/conversion_procs/ascii_and_mic'\ngmake[2]: *** [all] Error 2\ngmake[2]: Leaving directory `/home/ler/pg-dev/pgsql/src/backend/utils/mb/conversion_procs'\ngmake[1]: *** [all] Error 2\ngmake[1]: Leaving directory `/home/ler/pg-dev/pgsql/src'\ngmake: *** [all] Error 2\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": "26 Oct 2002 09:56:35 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "UnixWare 7.1.3 (BETA), compile error (with cc -Xb)"
},
{
"msg_contents": "Larry Rosenman <ler@lerctr.org> writes:\n> With cc -Xb, we get farther, but still die.... (CURRENT CVS):\n> cc -Xb -O -g -K PIC -I../../../../../../src/include -I/usr/local/include -c -o ascii_and_mic.o ascii_and_mic.c\n> UX:cc: WARNING: debugging and optimization mutually exclusive; -O disabled\n> UX:acomp: ERROR: \"ascii_and_mic.c\", line 19: syntax error, probably missing \",\", \";\" or \"=\"\n\nThis one is my fault: when I hacked up PG_FUNCTION_INFO_V1 yesterday,\nI neglected to check to be sure that *every* use had a semicolon after\nit. Just discovered that a few minutes ago myself. Fix is committed.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 26 Oct 2002 11:09:13 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: UnixWare 7.1.3 (BETA), compile error (with cc -Xb) "
}
] |
[
{
"msg_contents": "Hi\n i would like to use postgresql to store any data (some include chinese), but when i access the chinese data in postgresql, i can't get back correct data, it return any unknow word/data to me.\nHow can i solve it? i am using the RedHat Linux 7.2 and Postgresql 7.1, and i haven't install CLE on Linux.\n\nThanks\nTim\n\n\n\n\n\n\n\nHi\n i would like to use postgresql to store any \ndata (some include chinese), but when i access the chinese data \nin postgresql, i can't get back correct data, it return any unknow \nword/data to me.\nHow can i solve it? i am using the RedHat Linux 7.2 and \nPostgresql 7.1, and i haven't install CLE on Linux.\n \nThanks\nTim",
"msg_date": "Wed, 27 Oct 2032 00:24:21 +0800",
"msg_from": "\"Timothy Chan\" <timtimccy@hotmail.com>",
"msg_from_op": true,
"msg_subject": "How to store chinese using the Postgresql?"
},
{
"msg_contents": "On 27 Oct 2032 at 0:24, Timothy Chan wrote:\n\n> \n> Hi\n> i would like to use postgresql to store any data (some include chinese), \n> but when i access the chinese data inpostgresql, i can't get back correct \n> data, it return any unknow word/data to me.\n> How can i solve it? i am using the RedHat Linux 7.2 and Postgresql 7.1, and i \n> haven't install CLE on Linux.\n\nYou might have done this but just confirrming, did you use initdb -E to enable \nproper locale?\n\nBye\n Shridhar\n\n--\n\"I would rather spend 10 hours reading someone else's source code than10 \nminutes listening to Musak waiting for technical support which isn't.\"(By Dr. \nGreg Wettstein, Roger Maris Cancer Center)\n\n",
"msg_date": "Mon, 28 Oct 2002 19:31:54 +0530",
"msg_from": "\"Shridhar Daithankar\" <shridhar_daithankar@persistent.co.in>",
"msg_from_op": false,
"msg_subject": "Re: How to store chinese using the Postgresql?"
},
{
"msg_contents": "> i would like to use postgresql to store any data (some include chinese), but when i access the chinese data in postgresql, i can't get back correct data, it return any unknow word/data to me.\n> How can i solve it? i am using the RedHat Linux 7.2 and Postgresql 7.1, and i haven't install CLE on Linux.\n\nDefine \"chinese.\" What character set for Chinese are you using?\n--\nTatsuo Ishii\n",
"msg_date": "Mon, 28 Oct 2002 23:21:43 +0900 (JST)",
"msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>",
"msg_from_op": false,
"msg_subject": "Re: How to store chinese using the Postgresql?"
}
] |
[
{
"msg_contents": "We pass all regression test. \n\nthis is UnixWare 7.1.3, with the native cc compiler. \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": "26 Oct 2002 12:50:10 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "UnixWare 7.1.3: with current CVS, and cc -Xb.,.."
},
{
"msg_contents": "\nUpdated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nLarry Rosenman wrote:\n> We pass all regression test. \n> \n> this is UnixWare 7.1.3, with the native cc compiler. \n> \n> \n> -- \n> Larry Rosenman http://www.lerctr.org/~ler\n> Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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": "Sat, 26 Oct 2002 16:11:46 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: UnixWare 7.1.3: with current CVS, and cc -Xb.,.."
},
{
"msg_contents": "On Sat, 2002-10-26 at 15:11, Bruce Momjian wrote:\n> \n> Updated:\n> \n> http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\nthe illustrious marketing folks at The SCO group (nee Caldera) have\nrenamed this OS back to UnixWare. You may want to change the first\ncolumn back to UnixWare. (and footnote that OpenUNIX 8.0.0 is really\nUnixWare 7.1.2). \n\n\n> \n> ---------------------------------------------------------------------------\n> Larry Rosenman wrote:\n> > We pass all regression test. \n> > \n> > this is UnixWare 7.1.3, with the native cc compiler. \n> > \n> > \n> > -- \n> > Larry Rosenman http://www.lerctr.org/~ler\n> > Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> > US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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-- \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": "26 Oct 2002 15:27:57 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "Re: UnixWare 7.1.3: with current CVS, and cc -Xb.,.."
},
{
"msg_contents": "\nUpdate as you suggested:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nLarry Rosenman wrote:\n> On Sat, 2002-10-26 at 15:11, Bruce Momjian wrote:\n> > \n> > Updated:\n> > \n> > http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n> the illustrious marketing folks at The SCO group (nee Caldera) have\n> renamed this OS back to UnixWare. You may want to change the first\n> column back to UnixWare. (and footnote that OpenUNIX 8.0.0 is really\n> UnixWare 7.1.2). \n> \n> \n> > \n> > ---------------------------------------------------------------------------\n> > Larry Rosenman wrote:\n> > > We pass all regression test. \n> > > \n> > > this is UnixWare 7.1.3, with the native cc compiler. \n> > > \n> > > \n> > > -- \n> > > Larry Rosenman http://www.lerctr.org/~ler\n> > > Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> > > US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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> Larry Rosenman http://www.lerctr.org/~ler\n> Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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, 26 Oct 2002 16:45:06 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: UnixWare 7.1.3: with current CVS, and cc -Xb.,.."
}
] |
[
{
"msg_contents": "How concerned are we about assignment type mismatch warnings? \n\nI got a bunch in the mb stuff, and some in other places from the\nUnixWare 7.1.3 compiler. We still pass all regression tests. \n\nLER\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": "26 Oct 2002 13:07:13 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "assignment type mismatch complaints"
},
{
"msg_contents": "\nCan you send over a list of the errors and we will check them out.\n\n---------------------------------------------------------------------------\n\nLarry Rosenman wrote:\n> How concerned are we about assignment type mismatch warnings? \n> \n> I got a bunch in the mb stuff, and some in other places from the\n> UnixWare 7.1.3 compiler. We still pass all regression tests. \n> \n> LER\n> -- \n> Larry Rosenman http://www.lerctr.org/~ler\n> Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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": "Sat, 26 Oct 2002 16:12:28 -0400 (EDT)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: assignment type mismatch complaints"
},
{
"msg_contents": "Larry Rosenman <ler@lerctr.org> writes:\n> How concerned are we about assignment type mismatch warnings? \n\nThey're probably all \"char versus unsigned char\" complaints?\n\nThere are a ton of them on compilers that care about it; most of\n'em in the multibyte support. While it would be nice to clean up\nall that someday, I can't say that I think it's a really profitable\nuse of time.\n\nOne difficulty is that the obvious fix (add a bunch of casts) is\nprobably a net degradation of the code. Explicit casts will hide\nmismatches that are a lot worse than char signedness, and so\ncluttering the code with them makes things more fragile IMHO.\nI think an acceptable fix would involve running around and changing\ndatatype and function declarations; which is much more subtle and\nthought-requiring than throwing in a cast wherever the compiler\nburps.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sat, 26 Oct 2002 19:27:16 -0400",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: assignment type mismatch complaints "
},
{
"msg_contents": "On Sat, 2002-10-26 at 18:27, Tom Lane wrote:\n> Larry Rosenman <ler@lerctr.org> writes:\n> > How concerned are we about assignment type mismatch warnings? \n> \n> They're probably all \"char versus unsigned char\" complaints?\nProbably. The first few I looked at are PG_GETARG_CSTRING to unsigned\nchar assignments. (I can send the whole list to either you, Tom, or the\nlist). \n\n> \n> There are a ton of them on compilers that care about it; most of\n> 'em in the multibyte support. While it would be nice to clean up\n> all that someday, I can't say that I think it's a really profitable\n> use of time.\nOk, I understand that. It seems that there are a bunch, but they are\njust warnings. \n> \n> One difficulty is that the obvious fix (add a bunch of casts) is\n> probably a net degradation of the code. Explicit casts will hide\n> mismatches that are a lot worse than char signedness, and so\n> cluttering the code with them makes things more fragile IMHO.\n> I think an acceptable fix would involve running around and changing\n> datatype and function declarations; which is much more subtle and\n> thought-requiring than throwing in a cast wherever the compiler\n> burps.\nUnderstand, and I don't expect it to happen in a beta test :-). \n\n\n> \n> \t\t\tregards, tom lane\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": "26 Oct 2002 18:34:40 -0500",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "Re: assignment type mismatch complaints"
},
{
"msg_contents": "On Sat, 2002-10-26 at 18:34, Larry Rosenman wrote:\n> On Sat, 2002-10-26 at 18:27, Tom Lane wrote:\n> > Larry Rosenman <ler@lerctr.org> writes:\n> > > How concerned are we about assignment type mismatch warnings? \n> > \n> > They're probably all \"char versus unsigned char\" complaints?\n> Probably. The first few I looked at are PG_GETARG_CSTRING to unsigned\n> char assignments. (I can send the whole list to either you, Tom, or the\n> list). \n> \n> > \n> > There are a ton of them on compilers that care about it; most of\n> > 'em in the multibyte support. While it would be nice to clean up\n> > all that someday, I can't say that I think it's a really profitable\n> > use of time.\n> Ok, I understand that. It seems that there are a bunch, but they are\n> just warnings. \n> > \n> > One difficulty is that the obvious fix (add a bunch of casts) is\n> > probably a net degradation of the code. Explicit casts will hide\n> > mismatches that are a lot worse than char signedness, and so\n> > cluttering the code with them makes things more fragile IMHO.\n> > I think an acceptable fix would involve running around and changing\n> > datatype and function declarations; which is much more subtle and\n> > thought-requiring than throwing in a cast wherever the compiler\n> > burps.\n> Understand, and I don't expect it to happen in a beta test :-). \nIf anyone wants to look at these:\n\nftp://ftp.lerctr.org/pub/pg-dev/gmake.out.txt\n\nThanks,\nLER\n> \n> \n> > \n> > \t\t\tregards, tom lane\n> -- \n> Larry Rosenman http://www.lerctr.org/~ler\n> Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\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 Oct 2002 07:46:24 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "Re: assignment type mismatch complaints"
}
] |
[
{
"msg_contents": "I'm looking at different ways of optimizing queries with a large number of\njoins. I write the same query in a number of different ways and compare\nthe running times. Now the problem is I do not want the optimizer\nchanging the queries. So I explicit state the order of the joins in the\nFROM clause. I also turn off everything I can except for one type of join\n(say hash join), and I've turned off geqo. But I find that the PLANNER\nstill takes an enormous amount of time for some queries. It doesn't look\nlike the Planner is actually optimizing (changing) anything, but just in\ncase, I was wondering if there was a way to turn off the PLANNER.\n\nNote, when I say an enormous amount of time, I mean at least double the\ntime the EXECUTOR takes to actually answer the query.\n\nThanks for your help,\n\nBen McMahan\n\nps. here is a small example of what my queries look like (so you can see\nif there is something else it might be deciding on):\n\nSELECT DISTINCT c0.x1 , c1.x2 , c0.x3 , c0.x4 , c2.x5\nFROM r1 c4 (x4,x2,x5) JOIN (r0 c3 (x2,x3,x5) JOIN (r2 c2 (x3,x1,x5) JOIN (r1 c1\n(x4,x1,x2) JOIN r1 c0 (x1,x3,x4)\nON ( c0.x4 = c1.x4 AND c0.x1 = c1.x1 ))\nON ( c0.x3 = c2.x3 AND c0.x1 = c2.x1 ))\nON ( c1.x2 = c3.x2 AND c0.x3 = c3.x3 AND c2.x5 = c3.x5 ))\nON ( c0.x4 = c4.x4 AND c1.x2 = c4.x2 AND c2.x5 = c4.x5 );\n\nA quick reminder, FROM r1 c4 (x4,x2,x5) just renames a table r1 into c4\nwhere it also renames the columns to x4, x2, and x5 respectively.\n\n\n",
"msg_date": "Sat, 26 Oct 2002 13:28:55 -0500 (CDT)",
"msg_from": "Ben McMahan <mcmahanb@cs.rice.edu>",
"msg_from_op": true,
"msg_subject": "Turning the PLANNER off"
}
] |
[
{
"msg_contents": "\nPlease check it and confirm ... I believe everything, includin the docs,\nshould be right on this ... if not, I'm going to repackage before\nannouncing ...\n\n\n",
"msg_date": "Sun, 27 Oct 2002 00:26:48 -0300 (ADT)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "beta3 packaged ..."
},
{
"msg_contents": "\"Marc G. Fournier\" <scrappy@hub.org> writes:\n> Please check it and confirm ... I believe everything, includin the docs,\n> should be right on this ... if not, I'm going to repackage before\n> announcing ...\n\nCode looks good from here, can't check the docs very easily over this\ndialup connection ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Sun, 27 Oct 2002 12:51:47 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: beta3 packaged ... "
},
{
"msg_contents": "On Sun, 27 Oct 2002, Tom Lane wrote:\n\n> \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > Please check it and confirm ... I believe everything, includin the docs,\n> > should be right on this ... if not, I'm going to repackage before\n> > announcing ...\n>\n> Code looks good from here, can't check the docs very easily over this\n> dialup connection ...\n\n'K, if I haven't heard anything negative by the time I'm finished the\nserver upgrade this evening, I'll put out an announce ...\n\nJust need to confirm, beta3 *does* require an initdb from beta2, right?\n\n",
"msg_date": "Sun, 27 Oct 2002 18:15:28 -0400 (AST)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Re: beta3 packaged ... "
},
{
"msg_contents": "Marc G. Fournier wrote:\n> On Sun, 27 Oct 2002, Tom Lane wrote:\n> \n> > \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > > Please check it and confirm ... I believe everything, includin the docs,\n> > > should be right on this ... if not, I'm going to repackage before\n> > > announcing ...\n> >\n> > Code looks good from here, can't check the docs very easily over this\n> > dialup connection ...\n> \n> 'K, if I haven't heard anything negative by the time I'm finished the\n> server upgrade this evening, I'll put out an announce ...\n> \n> Just need to confirm, beta3 *does* require an initdb from beta2, right?\n\nYes. If we din't need an initdb, we may have just skipped beta3 and\njumped right to RC1 next Friday.\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, 27 Oct 2002 18:47:19 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: beta3 packaged ..."
},
{
"msg_contents": "On Sun, 27 Oct 2002, Bruce Momjian wrote:\n\n> Marc G. Fournier wrote:\n> > On Sun, 27 Oct 2002, Tom Lane wrote:\n> >\n> > > \"Marc G. Fournier\" <scrappy@hub.org> writes:\n> > > > Please check it and confirm ... I believe everything, includin the docs,\n> > > > should be right on this ... if not, I'm going to repackage before\n> > > > announcing ...\n> > >\n> > > Code looks good from here, can't check the docs very easily over this\n> > > dialup connection ...\n> >\n> > 'K, if I haven't heard anything negative by the time I'm finished the\n> > server upgrade this evening, I'll put out an announce ...\n> >\n> > Just need to confirm, beta3 *does* require an initdb from beta2, right?\n>\n> Yes. If we din't need an initdb, we may have just skipped beta3 and\n> jumped right to RC1 next Friday.\n\nWith the amount of chnages that went in between beta2 and no, there is no\nway that an RC1 would go out without a beta3 :)\n\n\n",
"msg_date": "Sun, 27 Oct 2002 19:48:10 -0400 (AST)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Re: beta3 packaged ..."
}
] |
[
{
"msg_contents": "Hi all\n\nI'm trying to compile an old version of Postgres and suddenly I've got the folloging error message:\n\nmake[2]: Cambiando a directorio `/home/jmilan/postgres-r/src/backend/port'\ngcc -I../../include -I../../backend -O2 -Wall -Wmissing-prototypes -I.. -c -o dynloader.o dynloader.c\ngcc -I../../include -I../../backend -O2 -Wall -Wmissing-prototypes -I.. -c -o inet_aton.o inet_aton.c\ninet_aton.c:58: warning: no previous prototype for `inet_aton'\ngcc -I../../include -I../../backend -O2 -Wall -Wmissing-prototypes -I.. -c -o isinf.o isinf.c\nisinf.c:8:20: ieeefp.h: No existe el fichero o el directorio\nmake[2]: *** [isinf.o] Error 1\nmake[2]: Saliendo directorio `/home/jmilan/postgres-r/src/backend/port'\nmake[1]: *** [port.dir] Error 2\nmake[1]: Saliendo directorio `/home/jmilan/postgres-r/src/backend'\nmake: *** [all] Error 2\n\n\nIt is a modified version of Postgres to accept replication. Can anyone help me?\n\nThanks to all\n\nJ.M. Milan\n\n\n\n\n\n\nHi all\n \nI'm trying to compile an old version of Postgres \nand suddenly I've got the folloging error message:\n \nmake[2]: Cambiando a directorio \n`/home/jmilan/postgres-r/src/backend/port'gcc -I../../include \n-I../../backend -O2 -Wall -Wmissing-prototypes \n-I.. -c -o dynloader.o dynloader.cgcc -I../../include \n-I../../backend -O2 -Wall -Wmissing-prototypes \n-I.. -c -o inet_aton.o inet_aton.cinet_aton.c:58: warning: no \nprevious prototype for `inet_aton'gcc -I../../include \n-I../../backend -O2 -Wall -Wmissing-prototypes \n-I.. -c -o isinf.o isinf.cisinf.c:8:20: ieeefp.h: No existe el \nfichero o el directoriomake[2]: *** [isinf.o] Error 1make[2]: Saliendo \ndirectorio `/home/jmilan/postgres-r/src/backend/port'make[1]: *** [port.dir] \nError 2make[1]: Saliendo directorio \n`/home/jmilan/postgres-r/src/backend'make: *** [all] Error \n2\n \nIt is a modified version of Postgres to accept \nreplication. Can anyone help me?\n \nThanks to all\n \nJ.M. Milan",
"msg_date": "Sun, 27 Oct 2002 09:43:05 +0100",
"msg_from": "\"Jesus M. Milan-Franco\" <milanjm@terra.es>",
"msg_from_op": true,
"msg_subject": "Help compiling postgres"
},
{
"msg_contents": ">\n> \n>\n> I'm trying to compile an old version of Postgres and suddenly I've got \n> the folloging error message:\n>\n\nThis looks to be the working model of postgres-r. Did you get it via CVS \nor from\nthe down load area? \n\nThese questions should probably be asked on the pg-replication list \ninstead of hackes.\n\n> \n>\n> make[2]: Cambiando a directorio `/home/jmilan/postgres-r/src/backend/port'\n> gcc -I../../include -I../../backend -O2 -Wall -Wmissing-prototypes \n> -I.. -c -o dynloader.o dynloader.c\n> gcc -I../../include -I../../backend -O2 -Wall -Wmissing-prototypes \n> -I.. -c -o inet_aton.o inet_aton.c\n> inet_aton.c:58: warning: no previous prototype for `inet_aton'\n> gcc -I../../include -I../../backend -O2 -Wall -Wmissing-prototypes \n> -I.. -c -o isinf.o isinf.c\n> isinf.c:8:20: ieeefp.h: No existe el fichero o el directorio\n> make[2]: *** [isinf.o] Error 1\n> make[2]: Saliendo directorio `/home/jmilan/postgres-r/src/backend/port'\n> make[1]: *** [port.dir] Error 2\n> make[1]: Saliendo directorio `/home/jmilan/postgres-r/src/backend'\n> make: *** [all] Error 2\n>\n> \n>\nWhat OS are you using? I know the \"working model\" has worked on several \nversions\nof Red Hat, and Solaris, but I'm not sure it's been tested on many other \nOSs.\n\nDarren\n\n\n",
"msg_date": "Sun, 27 Oct 2002 12:54:57 -0500",
"msg_from": "Darren Johnson <darren@up.hrcoxmail.com>",
"msg_from_op": false,
"msg_subject": "Re: Help compiling postgres"
}
] |
[
{
"msg_contents": "The topic below is quite common on the general list, people tend to ask without checking the documentation. I think most of those questions will\ndisappear if the planner output is modified, so it is clear why a decision is made (maybe with an extra option).\n\ne.g. something like this if an index scan is possible but not used:\n\nEXPLAIN SELECT * FROM copy_of_forum_list_child WHERE f_id = 1 and\nfather_name = 'top';\nSeq Scan on copy_of_forum_list_child (cost=0.00..2.44 rows=1width=100, costs using index: 0.00...9.44 rows=... with=...)\n\nAny comments?\n\nRegards,\n\tMario Weilguni\n\n\n---------- Weitergeleitete Nachricht ----------\n\nSubject: [GENERAL] index not scanned\nDate: Sun, 27 Oct 2002 12:46:39 +0200\nFrom: \"Ben-Nes Michael\" <miki@canaan.co.il>\nTo: \"postgresql\" <pgsql-general@postgresql.org>\n\nHi\n\nI add an Index to table that have only 30 rows.\n\nCREATE INDEX copy_of_forum_l_c_f_id_idx ON copy_of_forum_list_child ( f_id,\nfather_name );\n\nwhen i do:\nEXPLAIN SELECT * FROM copy_of_forum_list_child WHERE f_id = 1 and\nfather_name = 'top';\n\nits return: Seq Scan on copy_of_forum_list_child (cost=0.00..2.44 rows=1\nwidth=100)\n\nwhy its not scanning the index ?\n\ndoes the planner knows that the table is small and it will take more time to\ncheck the index then stright check ?\n\nby the way, the table will get quite big in the future.\n\nCheers\n\n\n---------------------------(end of broadcast)---------------------------\nTIP 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",
"msg_date": "Sun, 27 Oct 2002 12:05:54 +0100",
"msg_from": "Mario Weilguni <mweilguni@sime.com>",
"msg_from_op": true,
"msg_subject": "Fwd: [GENERAL] index not scanned"
}
] |
[
{
"msg_contents": "\nJust a quick note that I'm going to be rebooting mars (the PostgreSQL.Org)\nserver later on this evening to upgrade the OS ... downtime should be no\nmore then 1 hour, unless something unforeseen happens ...\n\n\n\n",
"msg_date": "Sun, 27 Oct 2002 13:25:28 -0400 (AST)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Server downtime ..."
},
{
"msg_contents": "Yo!\n\nWould it be sensible not to use -announce for such things? These bits\nare not really interesting if I'm subscribed to -announce to keep on top\nof what's happening wrt to new releases and such things.\n\n(Yes, I'm also subscribed to a few postgres lists, but I only read them\nwhen I have spare time. I don't filter announce, however, so the\nimportant things won't go unnoticed in case I don't look at list mail\nfor a few weeks).\n\ncheers\n-- vbi\n\n\nOn Sun, 2002-10-27 at 18:25, Marc G. Fournier wrote:\n> \n> Just a quick note that I'm going to be rebooting mars (the PostgreSQL.Org)\n> server later on this evening to upgrade the OS ... downtime should be no\n> more then 1 hour, unless something unforeseen happens ...\n> \n> \n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n-- \nthis email is protected by a digital signature: http://fortytwo.ch/gpg\n\nNOTE: keyserver bugs! get my key here: https://fortytwo.ch/gpg/92082481",
"msg_date": "28 Oct 2002 18:16:27 +0100",
"msg_from": "Adrian 'Dagurashibanipal' von Bidder <avbidder@fortytwo.ch>",
"msg_from_op": false,
"msg_subject": "Re: [ANNOUNCE] Server downtime ..."
},
{
"msg_contents": "\nUmmm .. by the fact that you see [ANNOUNCE] in the above subject, I did\nsend it to -announce :)\n\n\n\n\nOn 28 Oct 2002, Adrian 'Dagurashibanipal' von Bidder wrote:\n\n> Yo!\n>\n> Would it be sensible not to use -announce for such things? These bits\n> are not really interesting if I'm subscribed to -announce to keep on top\n> of what's happening wrt to new releases and such things.\n>\n> (Yes, I'm also subscribed to a few postgres lists, but I only read them\n> when I have spare time. I don't filter announce, however, so the\n> important things won't go unnoticed in case I don't look at list mail\n> for a few weeks).\n>\n> cheers\n> -- vbi\n>\n>\n> On Sun, 2002-10-27 at 18:25, Marc G. Fournier wrote:\n> >\n> > Just a quick note that I'm going to be rebooting mars (the PostgreSQL.Org)\n> > server later on this evening to upgrade the OS ... downtime should be no\n> > more then 1 hour, unless something unforeseen happens ...\n> >\n> >\n> >\n> >\n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 4: Don't 'kill -9' the postmaster\n> --\n> this email is protected by a digital signature: http://fortytwo.ch/gpg\n>\n> NOTE: keyserver bugs! get my key here: https://fortytwo.ch/gpg/92082481\n>\n\n",
"msg_date": "Mon, 28 Oct 2002 13:33:26 -0400 (AST)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Re: [ANNOUNCE] Server downtime ..."
},
{
"msg_contents": "On Mon, 2002-10-28 at 11:33, Marc G. Fournier wrote:\n> \n> Ummm .. by the fact that you see [ANNOUNCE] in the above subject, I did\n> send it to -announce :)\nI think he was saying that this was **NOT** stuff he wants to see on\n-ANNOUNCE. \n\nLER\n\n-- \nLarry Rosenman http://www.lerctr.org/~ler\nPhone: +1 972-414-9812 E-Mail: ler@lerctr.org\nUS Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\n\n",
"msg_date": "28 Oct 2002 11:36:26 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": false,
"msg_subject": "Re: [ANNOUNCE] Server downtime ..."
},
{
"msg_contents": "On 28 Oct 2002, Larry Rosenman wrote:\n\n> On Mon, 2002-10-28 at 11:33, Marc G. Fournier wrote:\n> >\n> > Ummm .. by the fact that you see [ANNOUNCE] in the above subject, I did\n> > send it to -announce :)\n> I think he was saying that this was **NOT** stuff he wants to see on\n> -ANNOUNCE.\n\nOops, just re-read, and you are right, that is what he said ...\n\ni\n\n",
"msg_date": "Mon, 28 Oct 2002 15:31:24 -0400 (AST)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "Re: [ANNOUNCE] Server downtime ..."
}
] |
[
{
"msg_contents": "I have been trying for 3 days to connect to my postgres linux box via vb\n.net on a windows box. I have to make this work somehow.\n\nI found npgsql provider and tried it but could not make it connect either. I\nwould prefer to use the OleDbConnection if possible.\n\nHas anyone been able to connect via the odbc driver and .NET?\n\nI am getting despirate and not sure which way to turn now.\n\nA snip of the code I have been using to try to connect is below.\n\nThanks in advance.\n\nTommy\n\n\nDim connstring As String = \"Server=192.168.5.5;User\nId=myname;Password=mypassword;Database=mydbname\"\n\nDim conn As New NpgsqlConnection(connstring)\n\nconn.Open()\n\nconn.Close()\n\n\n\n",
"msg_date": "Sun, 27 Oct 2002 21:55:54 -0600",
"msg_from": "\"Tommy\" <tmartin1@telocity.com>",
"msg_from_op": true,
"msg_subject": "Pgsql and .NET (Help!)"
}
] |
[
{
"msg_contents": "\nIt has take just over a month to get to this point, but Beta3 is now ready\nand available for testing. Depending on the changes generated by this\nBeta, we are planning on release our first Release Candidate out within\nthe week, with a full release shortly after.\n\nThis Beta requires an initdb, so a dump/reload is required from previous\nbetas.\n\nWith this beta, we are looking to complete our list of supported platforms\nthat is viewable at:\n\n\thttp://developer.postgresql.org/docs/postgres/supported-platforms.html\n\nIf you've been able to successful compile, and run, the regression tests\non any of the platforms not listed as being supported for v7.3, please let\nus know by sending a note to pgsql-hackers@postgresql.org.\n\nIf you've been unable to do so, please send us in a report of the problem\nyou hit, and, if its just the regression tests that failed, which tests\nfailed, as well as a diff between your actual results and what is listed\nas the expected results.\n\nThe list of changes between beta2 and beta3 is quite extensive,\nencompassing over 2000 lines in a ChangeLog format. A large portion of\nthis is cosmetic changes (silencing compiler warnings) or documentation\nchanges, but some of the more prominent changes consist of:\n\n - Function-call-style type coercions should be treated as explicit\n coercions, not implicit ones.\n\n - Perform transaction cleanup operations in a less ad-hoc, more\n principled order; in particular ensure that all shared resources are\n released before we release transaction locks.\n\n - Allow 8-byte off_t to properly pg_dump\n\n - Fix places that were using IsTransactionBlock() as an (inadequate)\n check that they'd get to commit immediately on finishing.\n\n - Fix ALTER TABLE ... ADD COLUMN for inheritance cases.\n\n - Make CREATE/ALTER/DROP USER/GROUP transaction-safe, or at least\n pretty nearly so, by postponing write of flat password file until\n transaction commit.\n\n - Fix potential problem with btbulkdelete deleting an indexscan's\n current item, if the page containing the current item is split while\n the indexscan is stopped and holds no read-lock on the page.\n\n - Rule rewriter was doing the wrong thing with conditional INSTEAD\n rules whose conditions might yield NULL.\n\n - Disallow aggregate functions in rule WHERE clauses.\n\n - Fix within-function memory leaks in the various PLs' interfaces to\n SPI_prepare\n\n - Fix case where a function in FROM returns a scalar type, but is\n referred to with whole-tuple syntax.\n\n - Invert logic in pg_exec_query_string() so that we set a snapshot for\n all utility statement types *except* a short list\n\n - Fix rewrite code so that rules are in fact executed in order by name,\n rather than being reordered according to INSTEAD attribute for\n implementation convenience.\n\n - Fix range-query estimation to not double-exclude NULLs\n\n - adds some missing functions for float8 math operations, specifically\n ceil(), floor(), and sign().\n\n - And many many more ...\n\nCopies of beta3 are available on all the mirror sites, as well as at the\nmain site under:\n\n\tftp://ftp.postgresql.org/pub/beta\n\nA complete copy of the ChangeLog file is also available at the same\nlocation.\n\nAny bugs/problems with this release, please report it to\npgsql-hackers@postgresql.org.\n\n\nMarc G. Fournier\nCo-Ordinator\nPostgreSQL Global Development Group\n\n\n\n\n",
"msg_date": "Sun, 27 Oct 2002 23:59:48 -0400 (AST)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": true,
"msg_subject": "PostgreSQL v7.3beta3 Released"
}
] |
[
{
"msg_contents": "I'm seeing this on my Linux box (kernel 2.4.18, glibc 2.2.4):\n\n*** ./expected/horology.out\tThu Sep 19 06:35:25 2002\n--- ./results/horology.out\tMon Oct 28 13:42:39 2002\n***************\n*** 537,549 ****\n SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n True \n ------\n! t\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n True \n ------\n! t\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n--- 537,549 ----\n SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n True \n ------\n! f\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n True \n ------\n! f\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n***************\n*** 790,796 ****\n + interval '02:01' AS time with time zone) AS time) AS \"03:31:00\";\n 03:31:00 \n ----------\n! 03:31:00\n (1 row)\n \n SELECT CAST(cast(date 'today' + time with time zone '03:30'\n--- 790,796 ----\n + interval '02:01' AS time with time zone) AS time) AS \"03:31:00\";\n 03:31:00 \n ----------\n! 02:31:00\n (1 row)\n \n SELECT CAST(cast(date 'today' + time with time zone '03:30'\n\n======================================================================\n\n",
"msg_date": "Mon, 28 Oct 2002 13:48:39 +0900 (JST)",
"msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>",
"msg_from_op": true,
"msg_subject": "7.3 beta3 regression failure"
},
{
"msg_contents": "> I'm seeing this on my Linux box (kernel 2.4.18, glibc 2.2.4):\n> \n> *** ./expected/horology.out\tThu Sep 19 06:35:25 2002\n> --- ./results/horology.out\tMon Oct 28 13:42:39 2002\n\n[snip]\n\nSorry, this must be due to summer time...\n--\nTatsuo Ishii\n",
"msg_date": "Mon, 28 Oct 2002 15:05:17 +0900 (JST)",
"msg_from": "Tatsuo Ishii <t-ishii@sra.co.jp>",
"msg_from_op": true,
"msg_subject": "Re: 7.3 beta3 regression failure"
}
] |
[
{
"msg_contents": "Regression Diffs:\n\n*** ./expected/geometry-positive-zeros-bsd.out\tTue Sep 12 16:07:16 2000\n--- ./results/geometry.out\tMon Oct 28 00:06:41 2002\n***************\n*** 114,120 ****\n | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n | (-5,-12) | [(1,2),(3,4)] | (1,2)\n | (10,10) | [(1,2),(3,4)] | (3,4)\n! | (0,0) | [(0,0),(6,6)] | (0,0)\n | (-10,0) | [(0,0),(6,6)] | (0,0)\n | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n--- 114,120 ----\n | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n | (-5,-12) | [(1,2),(3,4)] | (1,2)\n | (10,10) | [(1,2),(3,4)] | (3,4)\n! | (0,0) | [(0,0),(6,6)] | (-0,0)\n | (-10,0) | [(0,0),(6,6)] | (0,0)\n | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n***************\n*** 224,233 ****\n twentyfour | rotation \n ------------+-----------------------------\n | (0,0),(0,0)\n! | (0,0),(-20,-20)\n! | (0,2),(-14,0)\n | (0,79.2),(-58.8,0)\n! | (14,0),(0,-34)\n | (0,40),(0,0)\n | (0,0),(0,0)\n | (-10,-10),(-30,-30)\n--- 224,233 ----\n twentyfour | rotation \n ------------+-----------------------------\n | (0,0),(0,0)\n! | (-0,0),(-20,-20)\n! | (-0,2),(-14,0)\n | (0,79.2),(-58.8,0)\n! | (14,-0),(0,-34)\n | (0,40),(0,0)\n | (0,0),(0,0)\n | (-10,-10),(-30,-30)\n***************\n*** 254,264 ****\n WHERE (p.f1 <-> point '(0,0)') >= 1;\n twenty | rotation \n --------+-----------------------------------------------------------------------------------\n! | (0,0),(-0.2,-0.2)\n | (-0.1,-0.1),(-0.3,-0.3)\n | (-0.25,-0.25),(-0.25,-0.35)\n | (-0.3,-0.3),(-0.3,-0.3)\n! | (0.08,0),(0,-0.56)\n | (0.12,-0.28),(0.04,-0.84)\n | (0.26,-0.7),(0.1,-0.82)\n | (0.12,-0.84),(0.12,-0.84)\n--- 254,264 ----\n WHERE (p.f1 <-> point '(0,0)') >= 1;\n twenty | rotation \n --------+-----------------------------------------------------------------------------------\n! | (0,-0),(-0.2,-0.2)\n | (-0.1,-0.1),(-0.3,-0.3)\n | (-0.25,-0.25),(-0.25,-0.35)\n | (-0.3,-0.3),(-0.3,-0.3)\n! | (0.08,-0),(0,-0.56)\n | (0.12,-0.28),(0.04,-0.84)\n | (0.26,-0.7),(0.1,-0.82)\n | (0.12,-0.84),(0.12,-0.84)\n***************\n*** 266,272 ****\n | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n! | (0,0.0828402366863905),(-0.201183431952663,0)\n | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n--- 266,272 ----\n | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n! | (-0,0.0828402366863905),(-0.201183431952663,0)\n | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n\n======================================================================\n\n*** ./expected/horology.out\tWed Sep 18 16:35:25 2002\n--- ./results/horology.out\tMon Oct 28 00:06:41 2002\n***************\n*** 537,549 ****\n SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n True \n ------\n! t\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n True \n ------\n! t\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n--- 537,549 ----\n SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n True \n ------\n! f\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n True \n ------\n! f\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n\n======================================================================\n\n*** ./expected/copy.out\tMon Oct 28 00:06:11 2002\n--- ./results/copy.out\tMon Oct 28 00:06:49 2002\n***************\n*** 7,14 ****\n--- 7,16 ----\n COPY aggtest FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/data/agg.data';\n COPY onek FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/data/onek.data';\n COPY onek TO '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data';\n+ ERROR: COPY command, running in backend with effective uid 70, could not open file '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data' for writing. Errno = Permission denied (13).\n DELETE FROM onek;\n COPY onek FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data';\n+ ERROR: COPY command, running in backend with effective uid 70, could not open file '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data' for reading. Errno = No such file or directory (2).\n COPY tenk1 FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/data/tenk.data';\n COPY slow_emp4000 FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/data/rect.data';\n COPY person FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/data/person.data';\n\n======================================================================\n\n*** ./expected/select.out\tThu Jul 18 12:11:49 2002\n--- ./results/select.out\tMon Oct 28 00:07:02 2002\n***************\n*** 8,24 ****\n ORDER BY onek.unique1;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 0 | 998 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | AAAAAA | KMBAAA | OOOOxx\n! 1 | 214 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 2 | 3 | BAAAAA | GIAAAA | OOOOxx\n! 2 | 326 | 0 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 4 | 5 | CAAAAA | OMAAAA | OOOOxx\n! 3 | 431 | 1 | 3 | 3 | 3 | 3 | 3 | 3 | 3 | 3 | 6 | 7 | DAAAAA | PQAAAA | VVVVxx\n! 4 | 833 | 0 | 0 | 4 | 4 | 4 | 4 | 4 | 4 | 4 | 8 | 9 | EAAAAA | BGBAAA | HHHHxx\n! 5 | 541 | 1 | 1 | 5 | 5 | 5 | 5 | 5 | 5 | 5 | 10 | 11 | FAAAAA | VUAAAA | HHHHxx\n! 6 | 978 | 0 | 2 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 12 | 13 | GAAAAA | QLBAAA | OOOOxx\n! 7 | 647 | 1 | 3 | 7 | 7 | 7 | 7 | 7 | 7 | 7 | 14 | 15 | HAAAAA | XYAAAA | VVVVxx\n! 8 | 653 | 0 | 0 | 8 | 8 | 8 | 8 | 8 | 8 | 8 | 16 | 17 | IAAAAA | DZAAAA | HHHHxx\n! 9 | 49 | 1 | 1 | 9 | 9 | 9 | 9 | 9 | 9 | 9 | 18 | 19 | JAAAAA | XBAAAA | HHHHxx\n! (10 rows)\n \n --\n -- awk '{if($1<20){print $1,$14;}else{next;}}' onek.data | sort +0nr -1\n--- 8,14 ----\n ORDER BY onek.unique1;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n --\n -- awk '{if($1<20){print $1,$14;}else{next;}}' onek.data | sort +0nr -1\n***************\n*** 28,54 ****\n ORDER BY unique1 using >;\n unique1 | stringu1 \n ---------+----------\n! 19 | TAAAAA\n! 18 | SAAAAA\n! 17 | RAAAAA\n! 16 | QAAAAA\n! 15 | PAAAAA\n! 14 | OAAAAA\n! 13 | NAAAAA\n! 12 | MAAAAA\n! 11 | LAAAAA\n! 10 | KAAAAA\n! 9 | JAAAAA\n! 8 | IAAAAA\n! 7 | HAAAAA\n! 6 | GAAAAA\n! 5 | FAAAAA\n! 4 | EAAAAA\n! 3 | DAAAAA\n! 2 | CAAAAA\n! 1 | BAAAAA\n! 0 | AAAAAA\n! (20 rows)\n \n --\n -- awk '{if($1>980){print $1,$14;}else{next;}}' onek.data | sort +1d -2\n--- 18,24 ----\n ORDER BY unique1 using >;\n unique1 | stringu1 \n ---------+----------\n! (0 rows)\n \n --\n -- awk '{if($1>980){print $1,$14;}else{next;}}' onek.data | sort +1d -2\n***************\n*** 58,83 ****\n ORDER BY stringu1 using <;\n unique1 | stringu1 \n ---------+----------\n! 988 | AMAAAA\n! 989 | BMAAAA\n! 990 | CMAAAA\n! 991 | DMAAAA\n! 992 | EMAAAA\n! 993 | FMAAAA\n! 994 | GMAAAA\n! 995 | HMAAAA\n! 996 | IMAAAA\n! 997 | JMAAAA\n! 998 | KMAAAA\n! 999 | LMAAAA\n! 981 | TLAAAA\n! 982 | ULAAAA\n! 983 | VLAAAA\n! 984 | WLAAAA\n! 985 | XLAAAA\n! 986 | YLAAAA\n! 987 | ZLAAAA\n! (19 rows)\n \n \t\n --\n--- 28,34 ----\n ORDER BY stringu1 using <;\n unique1 | stringu1 \n ---------+----------\n! (0 rows)\n \n \t\n --\n***************\n*** 89,114 ****\n ORDER BY string4 using <, unique1 using >;\n unique1 | string4 \n ---------+---------\n! 999 | AAAAxx\n! 995 | AAAAxx\n! 983 | AAAAxx\n! 982 | AAAAxx\n! 981 | AAAAxx\n! 998 | HHHHxx\n! 997 | HHHHxx\n! 993 | HHHHxx\n! 990 | HHHHxx\n! 986 | HHHHxx\n! 996 | OOOOxx\n! 991 | OOOOxx\n! 988 | OOOOxx\n! 987 | OOOOxx\n! 985 | OOOOxx\n! 994 | VVVVxx\n! 992 | VVVVxx\n! 989 | VVVVxx\n! 984 | VVVVxx\n! (19 rows)\n \n \t\n --\n--- 40,46 ----\n ORDER BY string4 using <, unique1 using >;\n unique1 | string4 \n ---------+---------\n! (0 rows)\n \n \t\n --\n***************\n*** 120,145 ****\n ORDER BY string4 using >, unique1 using <;\n unique1 | string4 \n ---------+---------\n! 984 | VVVVxx\n! 989 | VVVVxx\n! 992 | VVVVxx\n! 994 | VVVVxx\n! 985 | OOOOxx\n! 987 | OOOOxx\n! 988 | OOOOxx\n! 991 | OOOOxx\n! 996 | OOOOxx\n! 986 | HHHHxx\n! 990 | HHHHxx\n! 993 | HHHHxx\n! 997 | HHHHxx\n! 998 | HHHHxx\n! 981 | AAAAxx\n! 982 | AAAAxx\n! 983 | AAAAxx\n! 995 | AAAAxx\n! 999 | AAAAxx\n! (19 rows)\n \n \t\n --\n--- 52,58 ----\n ORDER BY string4 using >, unique1 using <;\n unique1 | string4 \n ---------+---------\n! (0 rows)\n \n \t\n --\n***************\n*** 151,177 ****\n ORDER BY unique1 using >, string4 using <;\n unique1 | string4 \n ---------+---------\n! 19 | OOOOxx\n! 18 | VVVVxx\n! 17 | HHHHxx\n! 16 | OOOOxx\n! 15 | VVVVxx\n! 14 | AAAAxx\n! 13 | OOOOxx\n! 12 | AAAAxx\n! 11 | OOOOxx\n! 10 | AAAAxx\n! 9 | HHHHxx\n! 8 | HHHHxx\n! 7 | VVVVxx\n! 6 | OOOOxx\n! 5 | HHHHxx\n! 4 | HHHHxx\n! 3 | VVVVxx\n! 2 | OOOOxx\n! 1 | OOOOxx\n! 0 | OOOOxx\n! (20 rows)\n \n --\n -- awk '{if($1<20){print $1,$16;}else{next;}}' onek.data |\n--- 64,70 ----\n ORDER BY unique1 using >, string4 using <;\n unique1 | string4 \n ---------+---------\n! (0 rows)\n \n --\n -- awk '{if($1<20){print $1,$16;}else{next;}}' onek.data |\n***************\n*** 182,208 ****\n ORDER BY unique1 using <, string4 using >;\n unique1 | string4 \n ---------+---------\n! 0 | OOOOxx\n! 1 | OOOOxx\n! 2 | OOOOxx\n! 3 | VVVVxx\n! 4 | HHHHxx\n! 5 | HHHHxx\n! 6 | OOOOxx\n! 7 | VVVVxx\n! 8 | HHHHxx\n! 9 | HHHHxx\n! 10 | AAAAxx\n! 11 | OOOOxx\n! 12 | AAAAxx\n! 13 | OOOOxx\n! 14 | AAAAxx\n! 15 | VVVVxx\n! 16 | OOOOxx\n! 17 | HHHHxx\n! 18 | VVVVxx\n! 19 | OOOOxx\n! (20 rows)\n \n --\n -- test partial btree indexes\n--- 75,81 ----\n ORDER BY unique1 using <, string4 using >;\n unique1 | string4 \n ---------+---------\n! (0 rows)\n \n --\n -- test partial btree indexes\n***************\n*** 217,233 ****\n SELECT onek2.* WHERE onek2.unique1 < 10;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 0 | 998 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | AAAAAA | KMBAAA | OOOOxx\n! 1 | 214 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 2 | 3 | BAAAAA | GIAAAA | OOOOxx\n! 2 | 326 | 0 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 4 | 5 | CAAAAA | OMAAAA | OOOOxx\n! 3 | 431 | 1 | 3 | 3 | 3 | 3 | 3 | 3 | 3 | 3 | 6 | 7 | DAAAAA | PQAAAA | VVVVxx\n! 4 | 833 | 0 | 0 | 4 | 4 | 4 | 4 | 4 | 4 | 4 | 8 | 9 | EAAAAA | BGBAAA | HHHHxx\n! 5 | 541 | 1 | 1 | 5 | 5 | 5 | 5 | 5 | 5 | 5 | 10 | 11 | FAAAAA | VUAAAA | HHHHxx\n! 6 | 978 | 0 | 2 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 12 | 13 | GAAAAA | QLBAAA | OOOOxx\n! 7 | 647 | 1 | 3 | 7 | 7 | 7 | 7 | 7 | 7 | 7 | 14 | 15 | HAAAAA | XYAAAA | VVVVxx\n! 8 | 653 | 0 | 0 | 8 | 8 | 8 | 8 | 8 | 8 | 8 | 16 | 17 | IAAAAA | DZAAAA | HHHHxx\n! 9 | 49 | 1 | 1 | 9 | 9 | 9 | 9 | 9 | 9 | 9 | 18 | 19 | JAAAAA | XBAAAA | HHHHxx\n! (10 rows)\n \n --\n -- awk '{if($1<20){print $1,$14;}else{next;}}' onek.data | sort +0nr -1\n--- 90,96 ----\n SELECT onek2.* WHERE onek2.unique1 < 10;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n --\n -- awk '{if($1<20){print $1,$14;}else{next;}}' onek.data | sort +0nr -1\n***************\n*** 237,263 ****\n ORDER BY unique1 using >;\n unique1 | stringu1 \n ---------+----------\n! 19 | TAAAAA\n! 18 | SAAAAA\n! 17 | RAAAAA\n! 16 | QAAAAA\n! 15 | PAAAAA\n! 14 | OAAAAA\n! 13 | NAAAAA\n! 12 | MAAAAA\n! 11 | LAAAAA\n! 10 | KAAAAA\n! 9 | JAAAAA\n! 8 | IAAAAA\n! 7 | HAAAAA\n! 6 | GAAAAA\n! 5 | FAAAAA\n! 4 | EAAAAA\n! 3 | DAAAAA\n! 2 | CAAAAA\n! 1 | BAAAAA\n! 0 | AAAAAA\n! (20 rows)\n \n --\n -- awk '{if($1>980){print $1,$14;}else{next;}}' onek.data | sort +1d -2\n--- 100,106 ----\n ORDER BY unique1 using >;\n unique1 | stringu1 \n ---------+----------\n! (0 rows)\n \n --\n -- awk '{if($1>980){print $1,$14;}else{next;}}' onek.data | sort +1d -2\n***************\n*** 266,291 ****\n WHERE onek2.unique1 > 980;\n unique1 | stringu1 \n ---------+----------\n! 981 | TLAAAA\n! 982 | ULAAAA\n! 983 | VLAAAA\n! 984 | WLAAAA\n! 985 | XLAAAA\n! 986 | YLAAAA\n! 987 | ZLAAAA\n! 988 | AMAAAA\n! 989 | BMAAAA\n! 990 | CMAAAA\n! 991 | DMAAAA\n! 992 | EMAAAA\n! 993 | FMAAAA\n! 994 | GMAAAA\n! 995 | HMAAAA\n! 996 | IMAAAA\n! 997 | JMAAAA\n! 998 | KMAAAA\n! 999 | LMAAAA\n! (19 rows)\n \n SELECT two, stringu1, ten, string4\n INTO TABLE tmp\n--- 109,115 ----\n WHERE onek2.unique1 > 980;\n unique1 | stringu1 \n ---------+----------\n! (0 rows)\n \n SELECT two, stringu1, ten, string4\n INTO TABLE tmp\n\n======================================================================\n\n*** ./expected/select_distinct.out\tThu Jan 6 00:40:54 2000\n--- ./results/select_distinct.out\tMon Oct 28 00:07:03 2002\n***************\n*** 7,15 ****\n SELECT DISTINCT two FROM tmp;\n two \n -----\n! 0\n! 1\n! (2 rows)\n \n --\n -- awk '{print $5;}' onek.data | sort -n | uniq\n--- 7,13 ----\n SELECT DISTINCT two FROM tmp;\n two \n -----\n! (0 rows)\n \n --\n -- awk '{print $5;}' onek.data | sort -n | uniq\n***************\n*** 17,33 ****\n SELECT DISTINCT ten FROM tmp;\n ten \n -----\n! 0\n! 1\n! 2\n! 3\n! 4\n! 5\n! 6\n! 7\n! 8\n! 9\n! (10 rows)\n \n --\n -- awk '{print $16;}' onek.data | sort -d | uniq\n--- 15,21 ----\n SELECT DISTINCT ten FROM tmp;\n ten \n -----\n! (0 rows)\n \n --\n -- awk '{print $16;}' onek.data | sort -d | uniq\n***************\n*** 35,45 ****\n SELECT DISTINCT string4 FROM tmp;\n string4 \n ---------\n! AAAAxx\n! HHHHxx\n! OOOOxx\n! VVVVxx\n! (4 rows)\n \n --\n -- awk '{print $3,$16,$5;}' onek.data | sort -d | uniq |\n--- 23,29 ----\n SELECT DISTINCT string4 FROM tmp;\n string4 \n ---------\n! (0 rows)\n \n --\n -- awk '{print $3,$16,$5;}' onek.data | sort -d | uniq |\n***************\n*** 50,96 ****\n ORDER BY two using <, string4 using <, ten using <;\n two | string4 | ten \n -----+---------+-----\n! 0 | AAAAxx | 0\n! 0 | AAAAxx | 2\n! 0 | AAAAxx | 4\n! 0 | AAAAxx | 6\n! 0 | AAAAxx | 8\n! 0 | HHHHxx | 0\n! 0 | HHHHxx | 2\n! 0 | HHHHxx | 4\n! 0 | HHHHxx | 6\n! 0 | HHHHxx | 8\n! 0 | OOOOxx | 0\n! 0 | OOOOxx | 2\n! 0 | OOOOxx | 4\n! 0 | OOOOxx | 6\n! 0 | OOOOxx | 8\n! 0 | VVVVxx | 0\n! 0 | VVVVxx | 2\n! 0 | VVVVxx | 4\n! 0 | VVVVxx | 6\n! 0 | VVVVxx | 8\n! 1 | AAAAxx | 1\n! 1 | AAAAxx | 3\n! 1 | AAAAxx | 5\n! 1 | AAAAxx | 7\n! 1 | AAAAxx | 9\n! 1 | HHHHxx | 1\n! 1 | HHHHxx | 3\n! 1 | HHHHxx | 5\n! 1 | HHHHxx | 7\n! 1 | HHHHxx | 9\n! 1 | OOOOxx | 1\n! 1 | OOOOxx | 3\n! 1 | OOOOxx | 5\n! 1 | OOOOxx | 7\n! 1 | OOOOxx | 9\n! 1 | VVVVxx | 1\n! 1 | VVVVxx | 3\n! 1 | VVVVxx | 5\n! 1 | VVVVxx | 7\n! 1 | VVVVxx | 9\n! (40 rows)\n \n --\n -- awk '{print $2;}' person.data |\n--- 34,40 ----\n ORDER BY two using <, string4 using <, ten using <;\n two | string4 | ten \n -----+---------+-----\n! (0 rows)\n \n --\n -- awk '{print $2;}' person.data |\n\n======================================================================\n\n*** ./expected/select_distinct_on.out\tThu Jan 27 12:11:50 2000\n--- ./results/select_distinct_on.out\tMon Oct 28 00:07:03 2002\n***************\n*** 6,16 ****\n ORDER BY string4 using <, two using >, ten using <;\n string4 | two | ten \n ---------+-----+-----\n! AAAAxx | 1 | 1\n! HHHHxx | 1 | 1\n! OOOOxx | 1 | 1\n! VVVVxx | 1 | 1\n! (4 rows)\n \n -- this will fail due to conflict of ordering requirements\n SELECT DISTINCT ON (string4, ten) string4, two, ten\n--- 6,12 ----\n ORDER BY string4 using <, two using >, ten using <;\n string4 | two | ten \n ---------+-----+-----\n! (0 rows)\n \n -- this will fail due to conflict of ordering requirements\n SELECT DISTINCT ON (string4, ten) string4, two, ten\n***************\n*** 22,66 ****\n ORDER BY string4 using <, ten using >, two using <;\n string4 | ten | two \n ---------+-----+-----\n! AAAAxx | 9 | 1\n! AAAAxx | 8 | 0\n! AAAAxx | 7 | 1\n! AAAAxx | 6 | 0\n! AAAAxx | 5 | 1\n! AAAAxx | 4 | 0\n! AAAAxx | 3 | 1\n! AAAAxx | 2 | 0\n! AAAAxx | 1 | 1\n! AAAAxx | 0 | 0\n! HHHHxx | 9 | 1\n! HHHHxx | 8 | 0\n! HHHHxx | 7 | 1\n! HHHHxx | 6 | 0\n! HHHHxx | 5 | 1\n! HHHHxx | 4 | 0\n! HHHHxx | 3 | 1\n! HHHHxx | 2 | 0\n! HHHHxx | 1 | 1\n! HHHHxx | 0 | 0\n! OOOOxx | 9 | 1\n! OOOOxx | 8 | 0\n! OOOOxx | 7 | 1\n! OOOOxx | 6 | 0\n! OOOOxx | 5 | 1\n! OOOOxx | 4 | 0\n! OOOOxx | 3 | 1\n! OOOOxx | 2 | 0\n! OOOOxx | 1 | 1\n! OOOOxx | 0 | 0\n! VVVVxx | 9 | 1\n! VVVVxx | 8 | 0\n! VVVVxx | 7 | 1\n! VVVVxx | 6 | 0\n! VVVVxx | 5 | 1\n! VVVVxx | 4 | 0\n! VVVVxx | 3 | 1\n! VVVVxx | 2 | 0\n! VVVVxx | 1 | 1\n! VVVVxx | 0 | 0\n! (40 rows)\n \n--- 18,22 ----\n ORDER BY string4 using <, ten using >, two using <;\n string4 | ten | two \n ---------+-----+-----\n! (0 rows)\n \n\n======================================================================\n\n*** ./expected/aggregates.out\tWed Oct 2 14:21:26 2002\n--- ./results/aggregates.out\tMon Oct 28 00:07:04 2002\n***************\n*** 3,10 ****\n --\n SELECT avg(four) AS avg_1 FROM onek;\n avg_1 \n! ---------------------\n! 1.50000000000000000\n (1 row)\n \n SELECT avg(a) AS avg_32 FROM aggtest WHERE a < 100;\n--- 3,10 ----\n --\n SELECT avg(four) AS avg_1 FROM onek;\n avg_1 \n! -------\n! \n (1 row)\n \n SELECT avg(a) AS avg_32 FROM aggtest WHERE a < 100;\n***************\n*** 30,36 ****\n SELECT sum(four) AS sum_1500 FROM onek;\n sum_1500 \n ----------\n! 1500\n (1 row)\n \n SELECT sum(a) AS sum_198 FROM aggtest;\n--- 30,36 ----\n SELECT sum(four) AS sum_1500 FROM onek;\n sum_1500 \n ----------\n! \n (1 row)\n \n SELECT sum(a) AS sum_198 FROM aggtest;\n***************\n*** 54,60 ****\n SELECT max(four) AS max_3 FROM onek;\n max_3 \n -------\n! 3\n (1 row)\n \n SELECT max(a) AS max_100 FROM aggtest;\n--- 54,60 ----\n SELECT max(four) AS max_3 FROM onek;\n max_3 \n -------\n! \n (1 row)\n \n SELECT max(a) AS max_100 FROM aggtest;\n***************\n*** 78,137 ****\n SELECT count(four) AS cnt_1000 FROM onek;\n cnt_1000 \n ----------\n! 1000\n (1 row)\n \n SELECT count(DISTINCT four) AS cnt_4 FROM onek;\n cnt_4 \n -------\n! 4\n (1 row)\n \n select ten, count(*), sum(four) from onek group by ten;\n ten | count | sum \n -----+-------+-----\n! 0 | 100 | 100\n! 1 | 100 | 200\n! 2 | 100 | 100\n! 3 | 100 | 200\n! 4 | 100 | 100\n! 5 | 100 | 200\n! 6 | 100 | 100\n! 7 | 100 | 200\n! 8 | 100 | 100\n! 9 | 100 | 200\n! (10 rows)\n \n select ten, count(four), sum(DISTINCT four) from onek group by ten;\n ten | count | sum \n -----+-------+-----\n! 0 | 100 | 2\n! 1 | 100 | 4\n! 2 | 100 | 2\n! 3 | 100 | 4\n! 4 | 100 | 2\n! 5 | 100 | 4\n! 6 | 100 | 2\n! 7 | 100 | 4\n! 8 | 100 | 2\n! 9 | 100 | 4\n! (10 rows)\n \n SELECT newavg(four) AS avg_1 FROM onek;\n avg_1 \n! ---------------------\n! 1.50000000000000000\n (1 row)\n \n SELECT newsum(four) AS sum_1500 FROM onek;\n sum_1500 \n ----------\n! 1500\n (1 row)\n \n SELECT newcnt(four) AS cnt_1000 FROM onek;\n cnt_1000 \n ----------\n! 1000\n (1 row)\n \n--- 78,117 ----\n SELECT count(four) AS cnt_1000 FROM onek;\n cnt_1000 \n ----------\n! 0\n (1 row)\n \n SELECT count(DISTINCT four) AS cnt_4 FROM onek;\n cnt_4 \n -------\n! 0\n (1 row)\n \n select ten, count(*), sum(four) from onek group by ten;\n ten | count | sum \n -----+-------+-----\n! (0 rows)\n \n select ten, count(four), sum(DISTINCT four) from onek group by ten;\n ten | count | sum \n -----+-------+-----\n! (0 rows)\n \n SELECT newavg(four) AS avg_1 FROM onek;\n avg_1 \n! -------\n! \n (1 row)\n \n SELECT newsum(four) AS sum_1500 FROM onek;\n sum_1500 \n ----------\n! 0\n (1 row)\n \n SELECT newcnt(four) AS cnt_1000 FROM onek;\n cnt_1000 \n ----------\n! 0\n (1 row)\n \n\n======================================================================\n\n*** ./expected/random.out\tThu Jan 6 00:40:54 2000\n--- ./results/random.out\tMon Oct 28 00:07:05 2002\n***************\n*** 6,12 ****\n SELECT count(*) FROM onek;\n count \n -------\n! 1000\n (1 row)\n \n -- select roughly 1/10 of the tuples\n--- 6,12 ----\n SELECT count(*) FROM onek;\n count \n -------\n! 0\n (1 row)\n \n -- select roughly 1/10 of the tuples\n***************\n*** 25,35 ****\n GROUP BY random HAVING count(random) > 1;\n random | count \n --------+-------\n! (0 rows)\n \n SELECT random FROM RANDOM_TBL\n WHERE random NOT BETWEEN 80 AND 120;\n random \n --------\n! (0 rows)\n \n--- 25,38 ----\n GROUP BY random HAVING count(random) > 1;\n random | count \n --------+-------\n! 0 | 2\n! (1 row)\n \n SELECT random FROM RANDOM_TBL\n WHERE random NOT BETWEEN 80 AND 120;\n random \n --------\n! 0\n! 0\n! (2 rows)\n \n\n======================================================================\n\n*** ./expected/misc.out\tMon Oct 28 00:06:11 2002\n--- ./results/misc.out\tMon Oct 28 00:07:06 2002\n***************\n*** 40,73 ****\n -- copy\n --\n COPY onek TO '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data';\n DELETE FROM onek;\n COPY onek FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data';\n SELECT unique1 FROM onek WHERE unique1 < 2 ORDER BY unique1;\n unique1 \n ---------\n! 0\n! 1\n! (2 rows)\n \n DELETE FROM onek2;\n COPY onek2 FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data';\n SELECT unique1 FROM onek2 WHERE unique1 < 2 ORDER BY unique1;\n unique1 \n ---------\n! 0\n! 1\n! (2 rows)\n \n COPY BINARY stud_emp TO '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/stud_emp.data';\n DELETE FROM stud_emp;\n COPY BINARY stud_emp FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/stud_emp.data';\n SELECT * FROM stud_emp;\n name | age | location | salary | manager | gpa | percent \n! -------+-----+------------+--------+---------+-----+---------\n! jeff | 23 | (8,7.7) | 600 | sharon | 3.5 | \n! cim | 30 | (10.5,4.7) | 400 | | 3.4 | \n! linda | 19 | (0.9,6.1) | 100 | | 2.9 | \n! (3 rows)\n \n -- COPY aggtest FROM stdin;\n -- 56\t7.8\n--- 40,71 ----\n -- copy\n --\n COPY onek TO '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data';\n+ ERROR: COPY command, running in backend with effective uid 70, could not open file '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data' for writing. Errno = Permission denied (13).\n DELETE FROM onek;\n COPY onek FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data';\n+ ERROR: COPY command, running in backend with effective uid 70, could not open file '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data' for reading. Errno = No such file or directory (2).\n SELECT unique1 FROM onek WHERE unique1 < 2 ORDER BY unique1;\n unique1 \n ---------\n! (0 rows)\n \n DELETE FROM onek2;\n COPY onek2 FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data';\n+ ERROR: COPY command, running in backend with effective uid 70, could not open file '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/onek.data' for reading. Errno = No such file or directory (2).\n SELECT unique1 FROM onek2 WHERE unique1 < 2 ORDER BY unique1;\n unique1 \n ---------\n! (0 rows)\n \n COPY BINARY stud_emp TO '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/stud_emp.data';\n+ ERROR: COPY command, running in backend with effective uid 70, could not open file '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/stud_emp.data' for writing. Errno = Permission denied (13).\n DELETE FROM stud_emp;\n COPY BINARY stud_emp FROM '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/stud_emp.data';\n+ ERROR: COPY command, running in backend with effective uid 70, could not open file '/usr/ports/databases/postgresql-devel/work/postgresql-7.3b3/src/test/regress/results/stud_emp.data' for reading. Errno = No such file or directory (2).\n SELECT * FROM stud_emp;\n name | age | location | salary | manager | gpa | percent \n! ------+-----+----------+--------+---------+-----+---------\n! (0 rows)\n \n -- COPY aggtest FROM stdin;\n -- 56\t7.8\n***************\n*** 460,467 ****\n mike | posthacking\n joe | basketball\n sally | basketball\n! jeff | posthacking\n! (4 rows)\n \n --\n -- the next two queries demonstrate how functions generate bogus duplicates.\n--- 458,464 ----\n mike | posthacking\n joe | basketball\n sally | basketball\n! (3 rows)\n \n --\n -- the next two queries demonstrate how functions generate bogus duplicates.\n***************\n*** 512,520 ****\n mike | posthacking | peet's coffee\n joe | basketball | hightops\n sally | basketball | hightops\n! jeff | posthacking | advil\n! jeff | posthacking | peet's coffee\n! (6 rows)\n \n --\n -- just like the last two, but make sure that the target list fixup and\n--- 509,515 ----\n mike | posthacking | peet's coffee\n joe | basketball | hightops\n sally | basketball | hightops\n! (4 rows)\n \n --\n -- just like the last two, but make sure that the target list fixup and\n***************\n*** 536,544 ****\n peet's coffee | mike | posthacking\n hightops | joe | basketball\n hightops | sally | basketball\n! advil | jeff | posthacking\n! peet's coffee | jeff | posthacking\n! (6 rows)\n \n SELECT (p.hobbies).equipment.name, name(p.hobbies), p.name FROM ONLY person p;\n name | name | name \n--- 531,537 ----\n peet's coffee | mike | posthacking\n hightops | joe | basketball\n hightops | sally | basketball\n! (4 rows)\n \n SELECT (p.hobbies).equipment.name, name(p.hobbies), p.name FROM ONLY person p;\n name | name | name \n***************\n*** 556,564 ****\n peet's coffee | posthacking | mike\n hightops | basketball | joe\n hightops | basketball | sally\n! advil | posthacking | jeff\n! peet's coffee | posthacking | jeff\n! (6 rows)\n \n SELECT user_relns() AS user_relns\n ORDER BY user_relns;\n--- 549,555 ----\n peet's coffee | posthacking | mike\n hightops | basketball | joe\n hightops | basketball | sally\n! (4 rows)\n \n SELECT user_relns() AS user_relns\n ORDER BY user_relns;\n\n======================================================================\n\n*** ./expected/portals_p2.out\tMon Jul 16 00:07:00 2001\n--- ./results/portals_p2.out\tMon Oct 28 00:07:10 2002\n***************\n*** 31,110 ****\n FETCH all in foo13;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 50 | 253 | 0 | 2 | 0 | 10 | 0 | 50 | 50 | 50 | 50 | 0 | 1 | YBAAAA | TJAAAA | HHHHxx\n! (1 row)\n \n FETCH all in foo14;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 51 | 76 | 1 | 3 | 1 | 11 | 1 | 51 | 51 | 51 | 51 | 2 | 3 | ZBAAAA | YCAAAA | AAAAxx\n! (1 row)\n \n FETCH all in foo15;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 52 | 985 | 0 | 0 | 2 | 12 | 2 | 52 | 52 | 52 | 52 | 4 | 5 | ACAAAA | XLBAAA | HHHHxx\n! (1 row)\n \n FETCH all in foo16;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 53 | 196 | 1 | 1 | 3 | 13 | 3 | 53 | 53 | 53 | 53 | 6 | 7 | BCAAAA | OHAAAA | AAAAxx\n! (1 row)\n \n FETCH all in foo17;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 54 | 356 | 0 | 2 | 4 | 14 | 4 | 54 | 54 | 54 | 54 | 8 | 9 | CCAAAA | SNAAAA | AAAAxx\n! (1 row)\n \n FETCH all in foo18;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 55 | 627 | 1 | 3 | 5 | 15 | 5 | 55 | 55 | 55 | 55 | 10 | 11 | DCAAAA | DYAAAA | VVVVxx\n! (1 row)\n \n FETCH all in foo19;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 56 | 54 | 0 | 0 | 6 | 16 | 6 | 56 | 56 | 56 | 56 | 12 | 13 | ECAAAA | CCAAAA | OOOOxx\n! (1 row)\n \n FETCH all in foo20;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 57 | 942 | 1 | 1 | 7 | 17 | 7 | 57 | 57 | 57 | 57 | 14 | 15 | FCAAAA | GKBAAA | OOOOxx\n! (1 row)\n \n FETCH all in foo21;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 58 | 114 | 0 | 2 | 8 | 18 | 8 | 58 | 58 | 58 | 58 | 16 | 17 | GCAAAA | KEAAAA | OOOOxx\n! (1 row)\n \n FETCH all in foo22;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 59 | 593 | 1 | 3 | 9 | 19 | 9 | 59 | 59 | 59 | 59 | 18 | 19 | HCAAAA | VWAAAA | HHHHxx\n! (1 row)\n \n FETCH all in foo23;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 60 | 483 | 0 | 0 | 0 | 0 | 0 | 60 | 60 | 60 | 60 | 0 | 1 | ICAAAA | PSAAAA | VVVVxx\n! (1 row)\n \n FETCH all in foo24;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 50 | 253 | 0 | 2 | 0 | 10 | 0 | 50 | 50 | 50 | 50 | 0 | 1 | YBAAAA | TJAAAA | HHHHxx\n! (1 row)\n \n FETCH all in foo25;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! 60 | 483 | 0 | 0 | 0 | 0 | 0 | 60 | 60 | 60 | 60 | 0 | 1 | ICAAAA | PSAAAA | VVVVxx\n! (1 row)\n \n CLOSE foo13;\n CLOSE foo14;\n--- 31,97 ----\n FETCH all in foo13;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo14;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo15;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo16;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo17;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo18;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo19;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo20;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo21;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo22;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo23;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo24;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n FETCH all in foo25;\n unique1 | unique2 | two | four | ten | twenty | hundred | thousand | twothousand | fivethous | tenthous | odd | even | stringu1 | stringu2 | string4 \n ---------+---------+-----+------+-----+--------+---------+----------+-------------+-----------+----------+-----+------+----------+----------+---------\n! (0 rows)\n \n CLOSE foo13;\n CLOSE foo14;\n\n======================================================================\n\n*** ./expected/limit.out\tFri Oct 19 21:55:39 2001\n--- ./results/limit.out\tMon Oct 28 00:07:19 2002\n***************\n*** 7,46 ****\n \t\tORDER BY unique1 LIMIT 2;\n two | unique1 | unique2 | stringu1 \n -----+---------+---------+----------\n! | 51 | 76 | ZBAAAA\n! | 52 | 985 | ACAAAA\n! (2 rows)\n \n SELECT ''::text AS five, unique1, unique2, stringu1 \n \t\tFROM onek WHERE unique1 > 60 \n \t\tORDER BY unique1 LIMIT 5;\n five | unique1 | unique2 | stringu1 \n ------+---------+---------+----------\n! | 61 | 560 | JCAAAA\n! | 62 | 633 | KCAAAA\n! | 63 | 296 | LCAAAA\n! | 64 | 479 | MCAAAA\n! | 65 | 64 | NCAAAA\n! (5 rows)\n \n SELECT ''::text AS two, unique1, unique2, stringu1 \n \t\tFROM onek WHERE unique1 > 60 AND unique1 < 63\n \t\tORDER BY unique1 LIMIT 5;\n two | unique1 | unique2 | stringu1 \n -----+---------+---------+----------\n! | 61 | 560 | JCAAAA\n! | 62 | 633 | KCAAAA\n! (2 rows)\n \n SELECT ''::text AS three, unique1, unique2, stringu1 \n \t\tFROM onek WHERE unique1 > 100 \n \t\tORDER BY unique1 LIMIT 3 OFFSET 20;\n three | unique1 | unique2 | stringu1 \n -------+---------+---------+----------\n! | 121 | 700 | REAAAA\n! | 122 | 519 | SEAAAA\n! | 123 | 777 | TEAAAA\n! (3 rows)\n \n SELECT ''::text AS zero, unique1, unique2, stringu1 \n \t\tFROM onek WHERE unique1 < 50 \n--- 7,34 ----\n \t\tORDER BY unique1 LIMIT 2;\n two | unique1 | unique2 | stringu1 \n -----+---------+---------+----------\n! (0 rows)\n \n SELECT ''::text AS five, unique1, unique2, stringu1 \n \t\tFROM onek WHERE unique1 > 60 \n \t\tORDER BY unique1 LIMIT 5;\n five | unique1 | unique2 | stringu1 \n ------+---------+---------+----------\n! (0 rows)\n \n SELECT ''::text AS two, unique1, unique2, stringu1 \n \t\tFROM onek WHERE unique1 > 60 AND unique1 < 63\n \t\tORDER BY unique1 LIMIT 5;\n two | unique1 | unique2 | stringu1 \n -----+---------+---------+----------\n! (0 rows)\n \n SELECT ''::text AS three, unique1, unique2, stringu1 \n \t\tFROM onek WHERE unique1 > 100 \n \t\tORDER BY unique1 LIMIT 3 OFFSET 20;\n three | unique1 | unique2 | stringu1 \n -------+---------+---------+----------\n! (0 rows)\n \n SELECT ''::text AS zero, unique1, unique2, stringu1 \n \t\tFROM onek WHERE unique1 < 50 \n***************\n*** 54,110 ****\n \t\tORDER BY unique1 DESC LIMIT 20 OFFSET 39;\n eleven | unique1 | unique2 | stringu1 \n --------+---------+---------+----------\n! | 10 | 520 | KAAAAA\n! | 9 | 49 | JAAAAA\n! | 8 | 653 | IAAAAA\n! | 7 | 647 | HAAAAA\n! | 6 | 978 | GAAAAA\n! | 5 | 541 | FAAAAA\n! | 4 | 833 | EAAAAA\n! | 3 | 431 | DAAAAA\n! | 2 | 326 | CAAAAA\n! | 1 | 214 | BAAAAA\n! | 0 | 998 | AAAAAA\n! (11 rows)\n \n SELECT ''::text AS ten, unique1, unique2, stringu1 \n \t\tFROM onek\n \t\tORDER BY unique1 OFFSET 990;\n ten | unique1 | unique2 | stringu1 \n -----+---------+---------+----------\n! | 990 | 369 | CMAAAA\n! | 991 | 426 | DMAAAA\n! | 992 | 363 | EMAAAA\n! | 993 | 661 | FMAAAA\n! | 994 | 695 | GMAAAA\n! | 995 | 144 | HMAAAA\n! | 996 | 258 | IMAAAA\n! | 997 | 21 | JMAAAA\n! | 998 | 549 | KMAAAA\n! | 999 | 152 | LMAAAA\n! (10 rows)\n \n SELECT ''::text AS five, unique1, unique2, stringu1 \n \t\tFROM onek\n \t\tORDER BY unique1 OFFSET 990 LIMIT 5;\n five | unique1 | unique2 | stringu1 \n ------+---------+---------+----------\n! | 990 | 369 | CMAAAA\n! | 991 | 426 | DMAAAA\n! | 992 | 363 | EMAAAA\n! | 993 | 661 | FMAAAA\n! | 994 | 695 | GMAAAA\n! (5 rows)\n \n SELECT ''::text AS five, unique1, unique2, stringu1 \n \t\tFROM onek\n \t\tORDER BY unique1 LIMIT 5 OFFSET 900;\n five | unique1 | unique2 | stringu1 \n ------+---------+---------+----------\n! | 900 | 913 | QIAAAA\n! | 901 | 931 | RIAAAA\n! | 902 | 702 | SIAAAA\n! | 903 | 641 | TIAAAA\n! | 904 | 793 | UIAAAA\n! (5 rows)\n \n--- 42,67 ----\n \t\tORDER BY unique1 DESC LIMIT 20 OFFSET 39;\n eleven | unique1 | unique2 | stringu1 \n --------+---------+---------+----------\n! (0 rows)\n \n SELECT ''::text AS ten, unique1, unique2, stringu1 \n \t\tFROM onek\n \t\tORDER BY unique1 OFFSET 990;\n ten | unique1 | unique2 | stringu1 \n -----+---------+---------+----------\n! (0 rows)\n \n SELECT ''::text AS five, unique1, unique2, stringu1 \n \t\tFROM onek\n \t\tORDER BY unique1 OFFSET 990 LIMIT 5;\n five | unique1 | unique2 | stringu1 \n ------+---------+---------+----------\n! (0 rows)\n \n SELECT ''::text AS five, unique1, unique2, stringu1 \n \t\tFROM onek\n \t\tORDER BY unique1 LIMIT 5 OFFSET 900;\n five | unique1 | unique2 | stringu1 \n ------+---------+---------+----------\n! (0 rows)\n \n\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": "28 Oct 2002 00:09:09 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "FreeBSD 4.7: BETA3 (from port): regression failures..."
},
{
"msg_contents": "On Mon, 2002-10-28 at 00:09, Larry Rosenman wrote:\nIgnore the first one, stupid me forgot about permissions. \n\nHere is the real one (horology and geometry):\n(and I expect horology is due to DST change day). \n\nLER\n\n\n\n*** ./expected/geometry-positive-zeros-bsd.out\tTue Sep 12 16:07:16 2000\n--- ./results/geometry.out\tMon Oct 28 00:15:32 2002\n***************\n*** 114,120 ****\n | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n | (-5,-12) | [(1,2),(3,4)] | (1,2)\n | (10,10) | [(1,2),(3,4)] | (3,4)\n! | (0,0) | [(0,0),(6,6)] | (0,0)\n | (-10,0) | [(0,0),(6,6)] | (0,0)\n | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n--- 114,120 ----\n | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n | (-5,-12) | [(1,2),(3,4)] | (1,2)\n | (10,10) | [(1,2),(3,4)] | (3,4)\n! | (0,0) | [(0,0),(6,6)] | (-0,0)\n | (-10,0) | [(0,0),(6,6)] | (0,0)\n | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n***************\n*** 224,233 ****\n twentyfour | rotation \n ------------+-----------------------------\n | (0,0),(0,0)\n! | (0,0),(-20,-20)\n! | (0,2),(-14,0)\n | (0,79.2),(-58.8,0)\n! | (14,0),(0,-34)\n | (0,40),(0,0)\n | (0,0),(0,0)\n | (-10,-10),(-30,-30)\n--- 224,233 ----\n twentyfour | rotation \n ------------+-----------------------------\n | (0,0),(0,0)\n! | (-0,0),(-20,-20)\n! | (-0,2),(-14,0)\n | (0,79.2),(-58.8,0)\n! | (14,-0),(0,-34)\n | (0,40),(0,0)\n | (0,0),(0,0)\n | (-10,-10),(-30,-30)\n***************\n*** 254,264 ****\n WHERE (p.f1 <-> point '(0,0)') >= 1;\n twenty | rotation \n --------+-----------------------------------------------------------------------------------\n! | (0,0),(-0.2,-0.2)\n | (-0.1,-0.1),(-0.3,-0.3)\n | (-0.25,-0.25),(-0.25,-0.35)\n | (-0.3,-0.3),(-0.3,-0.3)\n! | (0.08,0),(0,-0.56)\n | (0.12,-0.28),(0.04,-0.84)\n | (0.26,-0.7),(0.1,-0.82)\n | (0.12,-0.84),(0.12,-0.84)\n--- 254,264 ----\n WHERE (p.f1 <-> point '(0,0)') >= 1;\n twenty | rotation \n --------+-----------------------------------------------------------------------------------\n! | (0,-0),(-0.2,-0.2)\n | (-0.1,-0.1),(-0.3,-0.3)\n | (-0.25,-0.25),(-0.25,-0.35)\n | (-0.3,-0.3),(-0.3,-0.3)\n! | (0.08,-0),(0,-0.56)\n | (0.12,-0.28),(0.04,-0.84)\n | (0.26,-0.7),(0.1,-0.82)\n | (0.12,-0.84),(0.12,-0.84)\n***************\n*** 266,272 ****\n | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n! | (0,0.0828402366863905),(-0.201183431952663,0)\n | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n--- 266,272 ----\n | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n! | (-0,0.0828402366863905),(-0.201183431952663,0)\n | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n\n======================================================================\n\n*** ./expected/horology.out\tWed Sep 18 16:35:25 2002\n--- ./results/horology.out\tMon Oct 28 00:15:33 2002\n***************\n*** 537,549 ****\n SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n True \n ------\n! t\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n True \n ------\n! t\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n--- 537,549 ----\n SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n True \n ------\n! f\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n True \n ------\n! f\n (1 row)\n \n SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n\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": "28 Oct 2002 00:20:10 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "Re: FreeBSD 4.7: BETA3 (from port): regression failures..."
},
{
"msg_contents": "On Mon, 2002-10-28 at 00:20, Larry Rosenman wrote:\n> On Mon, 2002-10-28 at 00:09, Larry Rosenman wrote:\n> Ignore the first one, stupid me forgot about permissions. \n> \n> Here is the real one (horology and geometry):\n> (and I expect horology is due to DST change day). \nThe geometry on FreeBSD4.7 MATCHES EXACTLY geometry-bsdi-precision.out. \n$ diff -c ../expected/geometry-bsdi-precision.out geometry.out \n$ \n\nSo, how can we change the resultmap to take the 4.7 stuff into account?\n\n\n\n> \n> LER\n> \n> \n> \n> *** ./expected/geometry-positive-zeros-bsd.out\tTue Sep 12 16:07:16 2000\n> --- ./results/geometry.out\tMon Oct 28 00:15:32 2002\n> ***************\n> *** 114,120 ****\n> | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n> | (-5,-12) | [(1,2),(3,4)] | (1,2)\n> | (10,10) | [(1,2),(3,4)] | (3,4)\n> ! | (0,0) | [(0,0),(6,6)] | (0,0)\n> | (-10,0) | [(0,0),(6,6)] | (0,0)\n> | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n> | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n> --- 114,120 ----\n> | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n> | (-5,-12) | [(1,2),(3,4)] | (1,2)\n> | (10,10) | [(1,2),(3,4)] | (3,4)\n> ! | (0,0) | [(0,0),(6,6)] | (-0,0)\n> | (-10,0) | [(0,0),(6,6)] | (0,0)\n> | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n> | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n> ***************\n> *** 224,233 ****\n> twentyfour | rotation \n> ------------+-----------------------------\n> | (0,0),(0,0)\n> ! | (0,0),(-20,-20)\n> ! | (0,2),(-14,0)\n> | (0,79.2),(-58.8,0)\n> ! | (14,0),(0,-34)\n> | (0,40),(0,0)\n> | (0,0),(0,0)\n> | (-10,-10),(-30,-30)\n> --- 224,233 ----\n> twentyfour | rotation \n> ------------+-----------------------------\n> | (0,0),(0,0)\n> ! | (-0,0),(-20,-20)\n> ! | (-0,2),(-14,0)\n> | (0,79.2),(-58.8,0)\n> ! | (14,-0),(0,-34)\n> | (0,40),(0,0)\n> | (0,0),(0,0)\n> | (-10,-10),(-30,-30)\n> ***************\n> *** 254,264 ****\n> WHERE (p.f1 <-> point '(0,0)') >= 1;\n> twenty | rotation \n> --------+-----------------------------------------------------------------------------------\n> ! | (0,0),(-0.2,-0.2)\n> | (-0.1,-0.1),(-0.3,-0.3)\n> | (-0.25,-0.25),(-0.25,-0.35)\n> | (-0.3,-0.3),(-0.3,-0.3)\n> ! | (0.08,0),(0,-0.56)\n> | (0.12,-0.28),(0.04,-0.84)\n> | (0.26,-0.7),(0.1,-0.82)\n> | (0.12,-0.84),(0.12,-0.84)\n> --- 254,264 ----\n> WHERE (p.f1 <-> point '(0,0)') >= 1;\n> twenty | rotation \n> --------+-----------------------------------------------------------------------------------\n> ! | (0,-0),(-0.2,-0.2)\n> | (-0.1,-0.1),(-0.3,-0.3)\n> | (-0.25,-0.25),(-0.25,-0.35)\n> | (-0.3,-0.3),(-0.3,-0.3)\n> ! | (0.08,-0),(0,-0.56)\n> | (0.12,-0.28),(0.04,-0.84)\n> | (0.26,-0.7),(0.1,-0.82)\n> | (0.12,-0.84),(0.12,-0.84)\n> ***************\n> *** 266,272 ****\n> | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n> | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n> | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n> ! | (0,0.0828402366863905),(-0.201183431952663,0)\n> | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n> | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n> | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n> --- 266,272 ----\n> | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n> | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n> | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n> ! | (-0,0.0828402366863905),(-0.201183431952663,0)\n> | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n> | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n> | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n> \n> ======================================================================\n> \n> *** ./expected/horology.out\tWed Sep 18 16:35:25 2002\n> --- ./results/horology.out\tMon Oct 28 00:15:33 2002\n> ***************\n> *** 537,549 ****\n> SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n> True \n> ------\n> ! t\n> (1 row)\n> \n> SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n> True \n> ------\n> ! t\n> (1 row)\n> \n> SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> --- 537,549 ----\n> SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n> True \n> ------\n> ! f\n> (1 row)\n> \n> SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n> True \n> ------\n> ! f\n> (1 row)\n> \n> SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> \n> ======================================================================\n> \n> -- \n> Larry Rosenman http://www.lerctr.org/~ler\n> Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\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 Oct 2002 21:22:03 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "Re: FreeBSD 4.7: BETA3 (from port): regression failures..."
},
{
"msg_contents": "\nSee the resultmap file in the regression directory. I see:\n\n\tgeometry/.*-bsdi=geometry-bsdi-precision\n\tgeometry/.*-darwin=geometry-powerpc-darwin\n\tgeometry/i.86-.*-freebsd=geometry-positive-zeros-bsd \n\tgeometry/alpha.*-freebsd=geometry-positive-zeros\n\tgeometry/i.86-.*-openbsd=geometry-positive-zeros-bsd\n\tgeometry/sparc-.*-openbsd=geometry-positive-zeros\n\tgeometry/.*-irix6=geometry-irix\n\tgeometry/.*-netbsd=geometry-positive-zeros\n\tgeometry/.*-sysv5.*:cc=geometry-uw7-cc\n\tgeometry/.*-sysv5.*:gcc=geometry-uw7-gcc \n\nI assume we need to modify the FreeBSD entries. Once you give me a\nstring to match your OS, I will rename geometry-bsdi-precision to\nsomething non-OS specific. I assume we need a string that is 4.7-only.\n\n\n---------------------------------------------------------------------------\n\nLarry Rosenman wrote:\n> On Mon, 2002-10-28 at 00:20, Larry Rosenman wrote:\n> > On Mon, 2002-10-28 at 00:09, Larry Rosenman wrote:\n> > Ignore the first one, stupid me forgot about permissions. \n> > \n> > Here is the real one (horology and geometry):\n> > (and I expect horology is due to DST change day). \n> The geometry on FreeBSD4.7 MATCHES EXACTLY geometry-bsdi-precision.out. \n> $ diff -c ../expected/geometry-bsdi-precision.out geometry.out \n> $ \n> \n> So, how can we change the resultmap to take the 4.7 stuff into account?\n> \n> \n> \n> > \n> > LER\n> > \n> > \n> > \n> > *** ./expected/geometry-positive-zeros-bsd.out\tTue Sep 12 16:07:16 2000\n> > --- ./results/geometry.out\tMon Oct 28 00:15:32 2002\n> > ***************\n> > *** 114,120 ****\n> > | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n> > | (-5,-12) | [(1,2),(3,4)] | (1,2)\n> > | (10,10) | [(1,2),(3,4)] | (3,4)\n> > ! | (0,0) | [(0,0),(6,6)] | (0,0)\n> > | (-10,0) | [(0,0),(6,6)] | (0,0)\n> > | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n> > | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n> > --- 114,120 ----\n> > | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n> > | (-5,-12) | [(1,2),(3,4)] | (1,2)\n> > | (10,10) | [(1,2),(3,4)] | (3,4)\n> > ! | (0,0) | [(0,0),(6,6)] | (-0,0)\n> > | (-10,0) | [(0,0),(6,6)] | (0,0)\n> > | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n> > | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n> > ***************\n> > *** 224,233 ****\n> > twentyfour | rotation \n> > ------------+-----------------------------\n> > | (0,0),(0,0)\n> > ! | (0,0),(-20,-20)\n> > ! | (0,2),(-14,0)\n> > | (0,79.2),(-58.8,0)\n> > ! | (14,0),(0,-34)\n> > | (0,40),(0,0)\n> > | (0,0),(0,0)\n> > | (-10,-10),(-30,-30)\n> > --- 224,233 ----\n> > twentyfour | rotation \n> > ------------+-----------------------------\n> > | (0,0),(0,0)\n> > ! | (-0,0),(-20,-20)\n> > ! | (-0,2),(-14,0)\n> > | (0,79.2),(-58.8,0)\n> > ! | (14,-0),(0,-34)\n> > | (0,40),(0,0)\n> > | (0,0),(0,0)\n> > | (-10,-10),(-30,-30)\n> > ***************\n> > *** 254,264 ****\n> > WHERE (p.f1 <-> point '(0,0)') >= 1;\n> > twenty | rotation \n> > --------+-----------------------------------------------------------------------------------\n> > ! | (0,0),(-0.2,-0.2)\n> > | (-0.1,-0.1),(-0.3,-0.3)\n> > | (-0.25,-0.25),(-0.25,-0.35)\n> > | (-0.3,-0.3),(-0.3,-0.3)\n> > ! | (0.08,0),(0,-0.56)\n> > | (0.12,-0.28),(0.04,-0.84)\n> > | (0.26,-0.7),(0.1,-0.82)\n> > | (0.12,-0.84),(0.12,-0.84)\n> > --- 254,264 ----\n> > WHERE (p.f1 <-> point '(0,0)') >= 1;\n> > twenty | rotation \n> > --------+-----------------------------------------------------------------------------------\n> > ! | (0,-0),(-0.2,-0.2)\n> > | (-0.1,-0.1),(-0.3,-0.3)\n> > | (-0.25,-0.25),(-0.25,-0.35)\n> > | (-0.3,-0.3),(-0.3,-0.3)\n> > ! | (0.08,-0),(0,-0.56)\n> > | (0.12,-0.28),(0.04,-0.84)\n> > | (0.26,-0.7),(0.1,-0.82)\n> > | (0.12,-0.84),(0.12,-0.84)\n> > ***************\n> > *** 266,272 ****\n> > | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n> > | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n> > | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n> > ! | (0,0.0828402366863905),(-0.201183431952663,0)\n> > | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n> > | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n> > | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n> > --- 266,272 ----\n> > | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n> > | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n> > | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n> > ! | (-0,0.0828402366863905),(-0.201183431952663,0)\n> > | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n> > | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n> > | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n> > \n> > ======================================================================\n> > \n> > *** ./expected/horology.out\tWed Sep 18 16:35:25 2002\n> > --- ./results/horology.out\tMon Oct 28 00:15:33 2002\n> > ***************\n> > *** 537,549 ****\n> > SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n> > True \n> > ------\n> > ! t\n> > (1 row)\n> > \n> > SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n> > True \n> > ------\n> > ! t\n> > (1 row)\n> > \n> > SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> > --- 537,549 ----\n> > SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n> > True \n> > ------\n> > ! f\n> > (1 row)\n> > \n> > SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n> > True \n> > ------\n> > ! f\n> > (1 row)\n> > \n> > SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> > \n> > ======================================================================\n> > \n> > -- \n> > Larry Rosenman http://www.lerctr.org/~ler\n> > Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> > US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\n> > \n> > \n> > ---------------------------(end of broadcast)---------------------------\n> > TIP 6: Have you searched our list archives?\n> > \n> > http://archives.postgresql.org\n> > \n> -- \n> Larry Rosenman http://www.lerctr.org/~ler\n> Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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": "Mon, 28 Oct 2002 22:33:45 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: FreeBSD 4.7: BETA3 (from port): regression failures..."
},
{
"msg_contents": "On Mon, 2002-10-28 at 21:33, Bruce Momjian wrote:\n> \n> See the resultmap file in the regression directory. I see:\n> \n> \tgeometry/.*-bsdi=geometry-bsdi-precision\n> \tgeometry/.*-darwin=geometry-powerpc-darwin\n> \tgeometry/i.86-.*-freebsd=geometry-positive-zeros-bsd \n> \tgeometry/alpha.*-freebsd=geometry-positive-zeros\n> \tgeometry/i.86-.*-openbsd=geometry-positive-zeros-bsd\n> \tgeometry/sparc-.*-openbsd=geometry-positive-zeros\n> \tgeometry/.*-irix6=geometry-irix\n> \tgeometry/.*-netbsd=geometry-positive-zeros\n> \tgeometry/.*-sysv5.*:cc=geometry-uw7-cc\n> \tgeometry/.*-sysv5.*:gcc=geometry-uw7-gcc \n> \n> I assume we need to modify the FreeBSD entries. Once you give me a\n> string to match your OS, I will rename geometry-bsdi-precision to\n> something non-OS specific. I assume we need a string that is 4.7-only.\ngeometry/i.86-.*-freebsd4.7= \n\nI think...\n\nI don't see where it picks out the release? \n\nLER\n\n> \n> \n> ---------------------------------------------------------------------------\n> \n> Larry Rosenman wrote:\n> > On Mon, 2002-10-28 at 00:20, Larry Rosenman wrote:\n> > > On Mon, 2002-10-28 at 00:09, Larry Rosenman wrote:\n> > > Ignore the first one, stupid me forgot about permissions. \n> > > \n> > > Here is the real one (horology and geometry):\n> > > (and I expect horology is due to DST change day). \n> > The geometry on FreeBSD4.7 MATCHES EXACTLY geometry-bsdi-precision.out. \n> > $ diff -c ../expected/geometry-bsdi-precision.out geometry.out \n> > $ \n> > \n> > So, how can we change the resultmap to take the 4.7 stuff into account?\n> > \n> > \n> > \n> > > \n> > > LER\n> > > \n> > > \n> > > \n> > > *** ./expected/geometry-positive-zeros-bsd.out\tTue Sep 12 16:07:16 2000\n> > > --- ./results/geometry.out\tMon Oct 28 00:15:32 2002\n> > > ***************\n> > > *** 114,120 ****\n> > > | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n> > > | (-5,-12) | [(1,2),(3,4)] | (1,2)\n> > > | (10,10) | [(1,2),(3,4)] | (3,4)\n> > > ! | (0,0) | [(0,0),(6,6)] | (0,0)\n> > > | (-10,0) | [(0,0),(6,6)] | (0,0)\n> > > | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n> > > | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n> > > --- 114,120 ----\n> > > | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n> > > | (-5,-12) | [(1,2),(3,4)] | (1,2)\n> > > | (10,10) | [(1,2),(3,4)] | (3,4)\n> > > ! | (0,0) | [(0,0),(6,6)] | (-0,0)\n> > > | (-10,0) | [(0,0),(6,6)] | (0,0)\n> > > | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n> > > | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n> > > ***************\n> > > *** 224,233 ****\n> > > twentyfour | rotation \n> > > ------------+-----------------------------\n> > > | (0,0),(0,0)\n> > > ! | (0,0),(-20,-20)\n> > > ! | (0,2),(-14,0)\n> > > | (0,79.2),(-58.8,0)\n> > > ! | (14,0),(0,-34)\n> > > | (0,40),(0,0)\n> > > | (0,0),(0,0)\n> > > | (-10,-10),(-30,-30)\n> > > --- 224,233 ----\n> > > twentyfour | rotation \n> > > ------------+-----------------------------\n> > > | (0,0),(0,0)\n> > > ! | (-0,0),(-20,-20)\n> > > ! | (-0,2),(-14,0)\n> > > | (0,79.2),(-58.8,0)\n> > > ! | (14,-0),(0,-34)\n> > > | (0,40),(0,0)\n> > > | (0,0),(0,0)\n> > > | (-10,-10),(-30,-30)\n> > > ***************\n> > > *** 254,264 ****\n> > > WHERE (p.f1 <-> point '(0,0)') >= 1;\n> > > twenty | rotation \n> > > --------+-----------------------------------------------------------------------------------\n> > > ! | (0,0),(-0.2,-0.2)\n> > > | (-0.1,-0.1),(-0.3,-0.3)\n> > > | (-0.25,-0.25),(-0.25,-0.35)\n> > > | (-0.3,-0.3),(-0.3,-0.3)\n> > > ! | (0.08,0),(0,-0.56)\n> > > | (0.12,-0.28),(0.04,-0.84)\n> > > | (0.26,-0.7),(0.1,-0.82)\n> > > | (0.12,-0.84),(0.12,-0.84)\n> > > --- 254,264 ----\n> > > WHERE (p.f1 <-> point '(0,0)') >= 1;\n> > > twenty | rotation \n> > > --------+-----------------------------------------------------------------------------------\n> > > ! | (0,-0),(-0.2,-0.2)\n> > > | (-0.1,-0.1),(-0.3,-0.3)\n> > > | (-0.25,-0.25),(-0.25,-0.35)\n> > > | (-0.3,-0.3),(-0.3,-0.3)\n> > > ! | (0.08,-0),(0,-0.56)\n> > > | (0.12,-0.28),(0.04,-0.84)\n> > > | (0.26,-0.7),(0.1,-0.82)\n> > > | (0.12,-0.84),(0.12,-0.84)\n> > > ***************\n> > > *** 266,272 ****\n> > > | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n> > > | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n> > > | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n> > > ! | (0,0.0828402366863905),(-0.201183431952663,0)\n> > > | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n> > > | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n> > > | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n> > > --- 266,272 ----\n> > > | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n> > > | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n> > > | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n> > > ! | (-0,0.0828402366863905),(-0.201183431952663,0)\n> > > | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n> > > | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n> > > | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n> > > \n> > > ======================================================================\n> > > \n> > > *** ./expected/horology.out\tWed Sep 18 16:35:25 2002\n> > > --- ./results/horology.out\tMon Oct 28 00:15:33 2002\n> > > ***************\n> > > *** 537,549 ****\n> > > SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n> > > True \n> > > ------\n> > > ! t\n> > > (1 row)\n> > > \n> > > SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n> > > True \n> > > ------\n> > > ! t\n> > > (1 row)\n> > > \n> > > SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> > > --- 537,549 ----\n> > > SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n> > > True \n> > > ------\n> > > ! f\n> > > (1 row)\n> > > \n> > > SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n> > > True \n> > > ------\n> > > ! f\n> > > (1 row)\n> > > \n> > > SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> > > \n> > > ======================================================================\n> > > \n> > > -- \n> > > Larry Rosenman http://www.lerctr.org/~ler\n> > > Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> > > US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\n> > > \n> > > \n> > > ---------------------------(end of broadcast)---------------------------\n> > > TIP 6: Have you searched our list archives?\n> > > \n> > > http://archives.postgresql.org\n> > > \n> > -- \n> > Larry Rosenman http://www.lerctr.org/~ler\n> > Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> > US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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-- \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 Oct 2002 21:36:32 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "Re: FreeBSD 4.7: BETA3 (from port): regression failures..."
},
{
"msg_contents": "\nOK, I just updated CVS to add a freebsd 4.7 entry to use the bsdi\noutput. Please update cvs and let me know if it passes now. Thanks.\n\n---------------------------------------------------------------------------\n\nLarry Rosenman wrote:\n> On Mon, 2002-10-28 at 21:33, Bruce Momjian wrote:\n> > \n> > See the resultmap file in the regression directory. I see:\n> > \n> > \tgeometry/.*-bsdi=geometry-bsdi-precision\n> > \tgeometry/.*-darwin=geometry-powerpc-darwin\n> > \tgeometry/i.86-.*-freebsd=geometry-positive-zeros-bsd \n> > \tgeometry/alpha.*-freebsd=geometry-positive-zeros\n> > \tgeometry/i.86-.*-openbsd=geometry-positive-zeros-bsd\n> > \tgeometry/sparc-.*-openbsd=geometry-positive-zeros\n> > \tgeometry/.*-irix6=geometry-irix\n> > \tgeometry/.*-netbsd=geometry-positive-zeros\n> > \tgeometry/.*-sysv5.*:cc=geometry-uw7-cc\n> > \tgeometry/.*-sysv5.*:gcc=geometry-uw7-gcc \n> > \n> > I assume we need to modify the FreeBSD entries. Once you give me a\n> > string to match your OS, I will rename geometry-bsdi-precision to\n> > something non-OS specific. I assume we need a string that is 4.7-only.\n> geometry/i.86-.*-freebsd4.7= \n> \n> I think...\n> \n> I don't see where it picks out the release? \n> \n> LER\n> \n> > \n> > \n> > ---------------------------------------------------------------------------\n> > \n> > Larry Rosenman wrote:\n> > > On Mon, 2002-10-28 at 00:20, Larry Rosenman wrote:\n> > > > On Mon, 2002-10-28 at 00:09, Larry Rosenman wrote:\n> > > > Ignore the first one, stupid me forgot about permissions. \n> > > > \n> > > > Here is the real one (horology and geometry):\n> > > > (and I expect horology is due to DST change day). \n> > > The geometry on FreeBSD4.7 MATCHES EXACTLY geometry-bsdi-precision.out. \n> > > $ diff -c ../expected/geometry-bsdi-precision.out geometry.out \n> > > $ \n> > > \n> > > So, how can we change the resultmap to take the 4.7 stuff into account?\n> > > \n> > > \n> > > \n> > > > \n> > > > LER\n> > > > \n> > > > \n> > > > \n> > > > *** ./expected/geometry-positive-zeros-bsd.out\tTue Sep 12 16:07:16 2000\n> > > > --- ./results/geometry.out\tMon Oct 28 00:15:32 2002\n> > > > ***************\n> > > > *** 114,120 ****\n> > > > | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n> > > > | (-5,-12) | [(1,2),(3,4)] | (1,2)\n> > > > | (10,10) | [(1,2),(3,4)] | (3,4)\n> > > > ! | (0,0) | [(0,0),(6,6)] | (0,0)\n> > > > | (-10,0) | [(0,0),(6,6)] | (0,0)\n> > > > | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n> > > > | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n> > > > --- 114,120 ----\n> > > > | (5.1,34.5) | [(1,2),(3,4)] | (3,4)\n> > > > | (-5,-12) | [(1,2),(3,4)] | (1,2)\n> > > > | (10,10) | [(1,2),(3,4)] | (3,4)\n> > > > ! | (0,0) | [(0,0),(6,6)] | (-0,0)\n> > > > | (-10,0) | [(0,0),(6,6)] | (0,0)\n> > > > | (-3,4) | [(0,0),(6,6)] | (0.5,0.5)\n> > > > | (5.1,34.5) | [(0,0),(6,6)] | (6,6)\n> > > > ***************\n> > > > *** 224,233 ****\n> > > > twentyfour | rotation \n> > > > ------------+-----------------------------\n> > > > | (0,0),(0,0)\n> > > > ! | (0,0),(-20,-20)\n> > > > ! | (0,2),(-14,0)\n> > > > | (0,79.2),(-58.8,0)\n> > > > ! | (14,0),(0,-34)\n> > > > | (0,40),(0,0)\n> > > > | (0,0),(0,0)\n> > > > | (-10,-10),(-30,-30)\n> > > > --- 224,233 ----\n> > > > twentyfour | rotation \n> > > > ------------+-----------------------------\n> > > > | (0,0),(0,0)\n> > > > ! | (-0,0),(-20,-20)\n> > > > ! | (-0,2),(-14,0)\n> > > > | (0,79.2),(-58.8,0)\n> > > > ! | (14,-0),(0,-34)\n> > > > | (0,40),(0,0)\n> > > > | (0,0),(0,0)\n> > > > | (-10,-10),(-30,-30)\n> > > > ***************\n> > > > *** 254,264 ****\n> > > > WHERE (p.f1 <-> point '(0,0)') >= 1;\n> > > > twenty | rotation \n> > > > --------+-----------------------------------------------------------------------------------\n> > > > ! | (0,0),(-0.2,-0.2)\n> > > > | (-0.1,-0.1),(-0.3,-0.3)\n> > > > | (-0.25,-0.25),(-0.25,-0.35)\n> > > > | (-0.3,-0.3),(-0.3,-0.3)\n> > > > ! | (0.08,0),(0,-0.56)\n> > > > | (0.12,-0.28),(0.04,-0.84)\n> > > > | (0.26,-0.7),(0.1,-0.82)\n> > > > | (0.12,-0.84),(0.12,-0.84)\n> > > > --- 254,264 ----\n> > > > WHERE (p.f1 <-> point '(0,0)') >= 1;\n> > > > twenty | rotation \n> > > > --------+-----------------------------------------------------------------------------------\n> > > > ! | (0,-0),(-0.2,-0.2)\n> > > > | (-0.1,-0.1),(-0.3,-0.3)\n> > > > | (-0.25,-0.25),(-0.25,-0.35)\n> > > > | (-0.3,-0.3),(-0.3,-0.3)\n> > > > ! | (0.08,-0),(0,-0.56)\n> > > > | (0.12,-0.28),(0.04,-0.84)\n> > > > | (0.26,-0.7),(0.1,-0.82)\n> > > > | (0.12,-0.84),(0.12,-0.84)\n> > > > ***************\n> > > > *** 266,272 ****\n> > > > | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n> > > > | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n> > > > | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n> > > > ! | (0,0.0828402366863905),(-0.201183431952663,0)\n> > > > | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n> > > > | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n> > > > | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n> > > > --- 266,272 ----\n> > > > | (0.0976764836465887,-0.0241724631246608),(0.0325588278821962,-0.0725173893739825)\n> > > > | (0.109762715208919,-0.0562379754328844),(0.0813970697054906,-0.0604311578116521)\n> > > > | (0.0976764836465887,-0.0725173893739825),(0.0976764836465887,-0.0725173893739825)\n> > > > ! | (-0,0.0828402366863905),(-0.201183431952663,0)\n> > > > | (-0.100591715976331,0.124260355029586),(-0.301775147928994,0.0414201183431953)\n> > > > | (-0.251479289940828,0.103550295857988),(-0.322485207100592,0.0739644970414201)\n> > > > | (-0.301775147928994,0.124260355029586),(-0.301775147928994,0.124260355029586)\n> > > > \n> > > > ======================================================================\n> > > > \n> > > > *** ./expected/horology.out\tWed Sep 18 16:35:25 2002\n> > > > --- ./results/horology.out\tMon Oct 28 00:15:33 2002\n> > > > ***************\n> > > > *** 537,549 ****\n> > > > SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n> > > > True \n> > > > ------\n> > > > ! t\n> > > > (1 row)\n> > > > \n> > > > SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n> > > > True \n> > > > ------\n> > > > ! t\n> > > > (1 row)\n> > > > \n> > > > SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> > > > --- 537,549 ----\n> > > > SELECT (timestamp with time zone 'today' = (timestamp with time zone 'tomorrow' - interval '1 day')) as \"True\";\n> > > > True \n> > > > ------\n> > > > ! f\n> > > > (1 row)\n> > > > \n> > > > SELECT (timestamp with time zone 'tomorrow' = (timestamp with time zone 'yesterday' + interval '2 days')) as \"True\";\n> > > > True \n> > > > ------\n> > > > ! f\n> > > > (1 row)\n> > > > \n> > > > SELECT (timestamp with time zone 'tomorrow' > 'now') as \"True\";\n> > > > \n> > > > ======================================================================\n> > > > \n> > > > -- \n> > > > Larry Rosenman http://www.lerctr.org/~ler\n> > > > Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> > > > US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\n> > > > \n> > > > \n> > > > ---------------------------(end of broadcast)---------------------------\n> > > > TIP 6: Have you searched our list archives?\n> > > > \n> > > > http://archives.postgresql.org\n> > > > \n> > > -- \n> > > Larry Rosenman http://www.lerctr.org/~ler\n> > > Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> > > US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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> Larry Rosenman http://www.lerctr.org/~ler\n> Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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, 28 Oct 2002 22:44:20 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: FreeBSD 4.7: BETA3 (from port): regression failures..."
},
{
"msg_contents": "Larry Rosenman wrote:\n> On Mon, 2002-10-28 at 21:33, Bruce Momjian wrote:\n> > \n> > See the resultmap file in the regression directory. I see:\n> > \n> > \tgeometry/.*-bsdi=geometry-bsdi-precision\n> > \tgeometry/.*-darwin=geometry-powerpc-darwin\n> > \tgeometry/i.86-.*-freebsd=geometry-positive-zeros-bsd \n> > \tgeometry/alpha.*-freebsd=geometry-positive-zeros\n> > \tgeometry/i.86-.*-openbsd=geometry-positive-zeros-bsd\n> > \tgeometry/sparc-.*-openbsd=geometry-positive-zeros\n> > \tgeometry/.*-irix6=geometry-irix\n> > \tgeometry/.*-netbsd=geometry-positive-zeros\n> > \tgeometry/.*-sysv5.*:cc=geometry-uw7-cc\n> > \tgeometry/.*-sysv5.*:gcc=geometry-uw7-gcc \n> > \n> > I assume we need to modify the FreeBSD entries. Once you give me a\n> > string to match your OS, I will rename geometry-bsdi-precision to\n> > something non-OS specific. I assume we need a string that is 4.7-only.\n> geometry/i.86-.*-freebsd4.7= \n> \n> I think...\n> \n> I don't see where it picks out the release? \n\nresultmap is read by pg_regress.sh.\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, 28 Oct 2002 22:45:06 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: FreeBSD 4.7: BETA3 (from port): regression failures..."
},
{
"msg_contents": "On Mon, 2002-10-28 at 21:45, Bruce Momjian wrote:\n> Larry Rosenman wrote:\n> > On Mon, 2002-10-28 at 21:33, Bruce Momjian wrote:\n> > > \n> > > See the resultmap file in the regression directory. I see:\n> > > \n> > > \tgeometry/.*-bsdi=geometry-bsdi-precision\n> > > \tgeometry/.*-darwin=geometry-powerpc-darwin\n> > > \tgeometry/i.86-.*-freebsd=geometry-positive-zeros-bsd \n> > > \tgeometry/alpha.*-freebsd=geometry-positive-zeros\n> > > \tgeometry/i.86-.*-openbsd=geometry-positive-zeros-bsd\n> > > \tgeometry/sparc-.*-openbsd=geometry-positive-zeros\n> > > \tgeometry/.*-irix6=geometry-irix\n> > > \tgeometry/.*-netbsd=geometry-positive-zeros\n> > > \tgeometry/.*-sysv5.*:cc=geometry-uw7-cc\n> > > \tgeometry/.*-sysv5.*:gcc=geometry-uw7-gcc \n> > > \n> > > I assume we need to modify the FreeBSD entries. Once you give me a\n> > > string to match your OS, I will rename geometry-bsdi-precision to\n> > > something non-OS specific. I assume we need a string that is 4.7-only.\n> > geometry/i.86-.*-freebsd4.7= \n> > \n> > I think...\n> > \n> > I don't see where it picks out the release? \n> \n> resultmap is read by pg_regress.sh.\nI made the following addition:\ngeometry/i.86-.*-freebsd4.7=geometry-bsdi-precision\n\n(The anon cvs server hasn't updated yet), and we pass geometry.\n\nStill fail horology, but I suspect that's DST change day fubar. \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-- \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 Oct 2002 21:52:28 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "Re: FreeBSD 4.7: BETA3 (from port): regression failures..."
}
] |
[
{
"msg_contents": "> -----Original Message-----\n> From: Bruce Momjian [mailto:pgman@candle.pha.pa.us] \n> Sent: 26 October 2002 03:17\n> To: PostgreSQL-development\n> Cc: Thomas Lockhart; Tom Lane\n> Subject: [HACKERS] Request for supported platforms\n> \n>\n> Folks. start sending in those plaform reports, OS name and \n> version number please.\n\nCYGWIN_NT-5.1 PC9 1.3.10(0.51/3/2) 2002-02-25 11:14 i686 unknown\n\nMake check failed with the normal spurious errors.\nMake installcheck also failed on horology, copy2 and domain - see\nattached output. \n\nThe clocks changed here on Saturday night, so I guess that shouldn't\nhave caused the first error (or should the docs be updated?).\n\nThe second 2 errors are both with copys - related to the problem with\nthe listen() backlog queue in the parallel test perhaps?\n\nRegards, Dave.",
"msg_date": "Mon, 28 Oct 2002 10:31:00 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Dave,\n\nThanks for the heads up...\n\nOn Mon, Oct 28, 2002 at 10:31:00AM -0000, Dave Page wrote:\n> > -----Original Message-----\n> > From: Bruce Momjian [mailto:pgman@candle.pha.pa.us] \n> > Sent: 26 October 2002 03:17\n> > Subject: [HACKERS] Request for supported platforms\n> >\n> > Folks. start sending in those plaform reports, OS name and \n> > version number please.\n> \n> CYGWIN_NT-5.1 PC9 1.3.10(0.51/3/2) 2002-02-25 11:14 i686 unknown\n ^^^^^^\n\nPlease try with Cygwin 1.3.14-1 while I attempt to deal with at least\nthe following Cygwin build issues with PostgreSQL CVS as of today at\nabout 7:00 AM EST:\n\n1. pg_config.h.in HAVE_FSEEKO ifdef:\n\nmake[4]: Entering directory `/home/jt/src/pgsql/src/backend/access/common'\ngcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -I../../../../src/include -DBUILDING_DLL -c -o heaptuple.o heaptuple.c\nIn file included from ../../../../src/include/c.h:56,\n from ../../../../src/include/postgres.h:48,\n from heaptuple.c:21:\n/usr/include/stdio.h:207: parse error before `('\n\n2. Cygwin bison limit exceeded:\n\nmake[4]: Entering directory `/home/jt/src/pgsql/src/interfaces/ecpg/preproc'\n[snip]\nbison -y -d preproc.y\npreproc.y:5560: fatal error: maximum table size (32767) exceeded\n\n> Make check failed with the normal spurious errors.\n\nI would stick with make installcheck due to the Cygwin (i.e., Windows)\nbacklog issue.\n\n> Make installcheck also failed on horology, copy2 and domain - see\n> attached output. \n> \n> The clocks changed here on Saturday night, so I guess that shouldn't\n> have caused the first error (or should the docs be updated?).\n> \n> The second 2 errors are both with copys - related to the problem with\n> the listen() backlog queue in the parallel test perhaps?\n\nI haven't looked into the above yet due to the build problems. Any help\nregarding these issues is gratefully appreciated.\n\nThanks,\nJason\n",
"msg_date": "Mon, 28 Oct 2002 08:32:55 -0500",
"msg_from": "Jason Tishler <jason@tishler.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": ">> Make installcheck also failed on horology, copy2 and domain - see\n>> attached output. \n>> \n>> The clocks changed here on Saturday night, so I guess that shouldn't\n>> have caused the first error (or should the docs be updated?).\n\nHorology failures are normal for a day or so on either side of a DST\nchange --- see the \"regression tests interpretation\" docs. I have\nno time right now to examine the other diffs.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 28 Oct 2002 10:52:41 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms "
},
{
"msg_contents": "Are you compiling from CVS or from a released tarball? The bison\nrequirement was recently raised to bison 1.5 or above (1.75 was recently\nreleased also.) This is an issue only when compiling from CVS, since\nthe bison stuff is preprocessed for released tarballs. So you might\nwant to try the just release beta3.\n\nOn Mon, 2002-10-28 at 08:32, Jason Tishler wrote:\n> Dave,\n> \n> Thanks for the heads up...\n> \n> On Mon, Oct 28, 2002 at 10:31:00AM -0000, Dave Page wrote:\n> > > -----Original Message-----\n> > > From: Bruce Momjian [mailto:pgman@candle.pha.pa.us] \n> > > Sent: 26 October 2002 03:17\n> > > Subject: [HACKERS] Request for supported platforms\n> > >\n> > > Folks. start sending in those plaform reports, OS name and \n> > > version number please.\n> > \n> > CYGWIN_NT-5.1 PC9 1.3.10(0.51/3/2) 2002-02-25 11:14 i686 unknown\n> ^^^^^^\n> \n> Please try with Cygwin 1.3.14-1 while I attempt to deal with at least\n> the following Cygwin build issues with PostgreSQL CVS as of today at\n> about 7:00 AM EST:\n> \n> 1. pg_config.h.in HAVE_FSEEKO ifdef:\n> \n> make[4]: Entering directory `/home/jt/src/pgsql/src/backend/access/common'\n> gcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -I../../../../src/include -DBUILDING_DLL -c -o heaptuple.o heaptuple.c\n> In file included from ../../../../src/include/c.h:56,\n> from ../../../../src/include/postgres.h:48,\n> from heaptuple.c:21:\n> /usr/include/stdio.h:207: parse error before `('\n> \n> 2. Cygwin bison limit exceeded:\n> \n> make[4]: Entering directory `/home/jt/src/pgsql/src/interfaces/ecpg/preproc'\n> [snip]\n> bison -y -d preproc.y\n> preproc.y:5560: fatal error: maximum table size (32767) exceeded\n> \n> > Make check failed with the normal spurious errors.\n> \n> I would stick with make installcheck due to the Cygwin (i.e., Windows)\n> backlog issue.\n> \n> > Make installcheck also failed on horology, copy2 and domain - see\n> > attached output. \n> > \n> > The clocks changed here on Saturday night, so I guess that shouldn't\n> > have caused the first error (or should the docs be updated?).\n> > \n> > The second 2 errors are both with copys - related to the problem with\n> > the listen() backlog queue in the parallel test perhaps?\n> \n> I haven't looked into the above yet due to the build problems. Any help\n> regarding these issues is gratefully appreciated.\n> \n> Thanks,\n> Jason\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": "28 Oct 2002 22:50:40 -0500",
"msg_from": "\"Matthew T. O'Connor\" <matthew@zeut.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "Matthew,\n\nOn Mon, Oct 28, 2002 at 10:50:40PM -0500, Matthew T. O'Connor wrote:\n> Are you compiling from CVS or from a released tarball?\n\nCVS.\n\n> The bison requirement was recently raised to bison 1.5 or above (1.75\n> was recently released also.) This is an issue only when compiling\n> from CVS, since the bison stuff is preprocessed for released tarballs.\n> So you might want to try the just release beta3.\n\nThanks for the above, but see my recent, related posts (if interested).\n\nJason\n",
"msg_date": "Tue, 29 Oct 2002 09:54:20 -0500",
"msg_from": "Jason Tishler <jason@tishler.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Bruce Momjian [mailto:pgman@candle.pha.pa.us] \n> Sent: 26 October 2002 03:17\n> To: PostgreSQL-development\n> Cc: Thomas Lockhart; Tom Lane\n> Subject: [HACKERS] Request for supported platforms\n> \n> \n> Folks. start sending in those plaform reports, OS name and \n> version number please.\n> \n\nWindows XP Professional SP1\n\nClient build fails :-(\n\nRegards, Dave.\n\nC:\\cygwin\\usr\\local\\src\\postgresql-7.3b3\\src>nmake /f win32.mak\n\nMicrosoft (R) Program Maintenance Utility Version 6.00.8168.0\nCopyright (C) Microsoft Corp 1988-1998. All rights reserved.\n\n cd include\n if not exist pg_config.h copy pg_config.h.win32 pg_config.h\n 1 file(s) copied.\n cd ..\n cd interfaces\\libpq\n nmake /f win32.mak\n\nMicrosoft (R) Program Maintenance Utility Version 6.00.8168.0\nCopyright (C) Microsoft Corp 1988-1998. All rights reserved.\n\nBuilding the Win32 static library...\n\n if not exist \".\\Release/\" mkdir \".\\Release\"\n cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nma04188.\ndllist.c\n cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmb04188.\nmd5.c\n cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmc04188.\nwchar.c\n cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmd04188.\nencnames.c\n cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nme04188.\nwin32.c\nfe-auth.c\nfe-connect.c\nfe-exec.c\nfe-lobj.c\nfe-misc.c\nfe-print.c\nfe-secure.c\npqexpbuffer.c\n link.exe -lib @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmf04188.\n cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmg04188.\nlibpqdll.c\n rc.exe /l 0x409 /fo\".\\Release\\libpq.res\" libpq.rc\n link.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmh04188.\n Creating library .\\Release\\libpqdll.lib and object\n.\\Release\\libpqdll.exp\n cd ..\\..\\bin\\psql\n nmake /f win32.mak\n\nMicrosoft (R) Program Maintenance Utility Version 6.00.8168.0\nCopyright (C) Microsoft Corp 1988-1998. All rights reserved.\n\n if not exist \".\\Release/\" mkdir \".\\Release\"\nNMAKE : fatal error U1073: don't know how to make '..\\..\\utils\\getopt.c'\nStop.\nNMAKE : fatal error U1077: '\"C:\\Program Files\\Microsoft Visual\nStudio\\VC98\\bin\\N\nMAKE.EXE\"' : return code '0x2'\nStop.\n\nC:\\cygwin\\usr\\local\\src\\postgresql-7.3b3\\src>\n",
"msg_date": "Mon, 28 Oct 2002 10:46:05 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "\nI have updated CVS and re-added getopt.c, now in /port, and updated\nwin32.mak. That should help.\n\n---------------------------------------------------------------------------\n\nDave Page wrote:\n> \n> \n> > -----Original Message-----\n> > From: Bruce Momjian [mailto:pgman@candle.pha.pa.us] \n> > Sent: 26 October 2002 03:17\n> > To: PostgreSQL-development\n> > Cc: Thomas Lockhart; Tom Lane\n> > Subject: [HACKERS] Request for supported platforms\n> > \n> > \n> > Folks. start sending in those plaform reports, OS name and \n> > version number please.\n> > \n> \n> Windows XP Professional SP1\n> \n> Client build fails :-(\n> \n> Regards, Dave.\n> \n> C:\\cygwin\\usr\\local\\src\\postgresql-7.3b3\\src>nmake /f win32.mak\n> \n> Microsoft (R) Program Maintenance Utility Version 6.00.8168.0\n> Copyright (C) Microsoft Corp 1988-1998. All rights reserved.\n> \n> cd include\n> if not exist pg_config.h copy pg_config.h.win32 pg_config.h\n> 1 file(s) copied.\n> cd ..\n> cd interfaces\\libpq\n> nmake /f win32.mak\n> \n> Microsoft (R) Program Maintenance Utility Version 6.00.8168.0\n> Copyright (C) Microsoft Corp 1988-1998. All rights reserved.\n> \n> Building the Win32 static library...\n> \n> if not exist \".\\Release/\" mkdir \".\\Release\"\n> cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nma04188.\n> dllist.c\n> cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmb04188.\n> md5.c\n> cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmc04188.\n> wchar.c\n> cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmd04188.\n> encnames.c\n> cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nme04188.\n> win32.c\n> fe-auth.c\n> fe-connect.c\n> fe-exec.c\n> fe-lobj.c\n> fe-misc.c\n> fe-print.c\n> fe-secure.c\n> pqexpbuffer.c\n> link.exe -lib @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmf04188.\n> cl.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmg04188.\n> libpqdll.c\n> rc.exe /l 0x409 /fo\".\\Release\\libpq.res\" libpq.rc\n> link.exe @C:\\DOCUME~1\\dpage\\LOCALS~1\\Temp\\nmh04188.\n> Creating library .\\Release\\libpqdll.lib and object\n> .\\Release\\libpqdll.exp\n> cd ..\\..\\bin\\psql\n> nmake /f win32.mak\n> \n> Microsoft (R) Program Maintenance Utility Version 6.00.8168.0\n> Copyright (C) Microsoft Corp 1988-1998. All rights reserved.\n> \n> if not exist \".\\Release/\" mkdir \".\\Release\"\n> NMAKE : fatal error U1073: don't know how to make '..\\..\\utils\\getopt.c'\n> Stop.\n> NMAKE : fatal error U1077: '\"C:\\Program Files\\Microsoft Visual\n> Studio\\VC98\\bin\\N\n> MAKE.EXE\"' : return code '0x2'\n> Stop.\n> \n> C:\\cygwin\\usr\\local\\src\\postgresql-7.3b3\\src>\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, 28 Oct 2002 23:24:01 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
}
] |
[
{
"msg_contents": "Is there a planned string freeze for postgresql?\n\nThere was a (good) change some weeks ago that altered lots of witespace.\nRemember that it's enough to change a single space to make that\ntranslation of that string not work. To do \"psql --help\" and get half the\nentries in swedish and half in english would be ugly. I sent an update a\ncouple of days ago that fixes the swedish translation. I'm not sure about\nthe status of the other languages. Now I'm just planning for the release\nand to prevent us from getting into a situation where strings are altered\nthe day before the release or something.\n\nThere have been a lot work with the translations and would be nice if it\nworks in the release.\n\n-- \n/Dennis\n\n",
"msg_date": "Mon, 28 Oct 2002 13:15:45 +0100 (CET)",
"msg_from": "=?ISO-8859-1?Q?Dennis_Bj=F6rklund?= <db@zigo.dhs.org>",
"msg_from_op": true,
"msg_subject": "string freeze"
},
{
"msg_contents": "Dennis Bj�rklund wrote:\n> Is there a planned string freeze for postgresql?\n> \n> There was a (good) change some weeks ago that altered lots of witespace.\n> Remember that it's enough to change a single space to make that\n> translation of that string not work. To do \"psql --help\" and get half the\n> entries in swedish and half in english would be ugly. I sent an update a\n> couple of days ago that fixes the swedish translation. I'm not sure about\n> the status of the other languages. Now I'm just planning for the release\n> and to prevent us from getting into a situation where strings are altered\n> the day before the release or something.\n> \n> There have been a lot work with the translations and would be nice if it\n> works in the release.\n\nThis is a good question. We only added translations in 7.2 so we are\nstill learning how to get those languages updated. There should be\n_very_ few changes from now until 7.3 final. We need to keep in mind\nthat changes affect the translators.\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, 28 Oct 2002 22:01:02 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: string freeze"
},
{
"msg_contents": "Dennis Bjï¿œrklund writes:\n\n> Is there a planned string freeze for postgresql?\n\nLet's say right now. ;-)\n\nSeriously, beta 3 was the last beta (one hopes), so this is a good time.\nIf there are more changes required due to code fixes, I'll make sure\neveryone is notified.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Tue, 29 Oct 2002 23:19:22 +0100 (CET)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: string freeze"
}
] |
[
{
"msg_contents": "\nJust to update the list of supported platforms, 7.3b3 compiles and passes\nall the regression tests on MacOSX 10.2.1\n\nAlthough don't know if this is relevant but this appears when running the\ntests:\n\nparallel group (20 tests): ./pg_regress: fork: Resource temporarily\nunavailable\n./pg_regress: fork: Resource temporarily unavailable\n comments lseg box path timetz point circle reltime tinterval date inet\ninterval timestamp time abstime polygon timestamptz oidjoins\n\nCheers\n\nAdam\n\n\n-- \nThis message has been scanned for viruses and\ndangerous content by MailScanner, and is\nbelieved to be clean.\n\n",
"msg_date": "Mon, 28 Oct 2002 12:34:18 +0000",
"msg_from": "Adam Witney <awitney@sghms.ac.uk>",
"msg_from_op": true,
"msg_subject": "7.3b3 on MacOSX 10.2.1"
},
{
"msg_contents": "\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nAdam Witney wrote:\n> \n> Just to update the list of supported platforms, 7.3b3 compiles and passes\n> all the regression tests on MacOSX 10.2.1\n> \n> Although don't know if this is relevant but this appears when running the\n> tests:\n> \n> parallel group (20 tests): ./pg_regress: fork: Resource temporarily\n> unavailable\n> ./pg_regress: fork: Resource temporarily unavailable\n> comments lseg box path timetz point circle reltime tinterval date inet\n> interval timestamp time abstime polygon timestamptz oidjoins\n> \n> Cheers\n> \n> Adam\n> \n> \n> -- \n> This message has been scanned for viruses and\n> dangerous content by MailScanner, and is\n> believed to be clean.\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": "Mon, 28 Oct 2002 22:06:29 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 on MacOSX 10.2.1"
},
{
"msg_contents": "Adam Witney <awitney@sghms.ac.uk> writes:\n> Just to update the list of supported platforms, 7.3b3 compiles and passes\n> all the regression tests on MacOSX 10.2.1\n\n> Although don't know if this is relevant but this appears when running the\n> tests:\n\n> parallel group (20 tests): ./pg_regress: fork: Resource temporarily\n> unavailable\n> ./pg_regress: fork: Resource temporarily unavailable\n> comments lseg box path timetz point circle reltime tinterval date inet\n> interval timestamp time abstime polygon timestamptz oidjoins\n\nThis suggests that you are hitting the per-user limit on the number of\nprocesses you can have; try raising that limit.\n\nI'd expect one of the tests not to have been run when that message\nappears; did you really get successful matches for all tests?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 29 Oct 2002 13:02:54 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 on MacOSX 10.2.1 "
},
{
"msg_contents": "So first off, what's the best way to tell from a cvs snapshot which \"release\" (if any) that snapshot is?\n\n\nI just (2pm) grabbed a full clean cvs checkout, and tried to run the regression tests. This is on a Macintosh G4/450 dual CPU with 512MB.\n\n'make runcheck' in src/test/regress/ fails with:\n\n<snip from src/test/regress/log/install.log>\n\nrm -f libecpg.so && \\\nln -s libecpg.so.3.4.0 libecpg.so\nmake -C preproc install\nmake -C ../../../../src/port all\nmake[6]: Nothing to be done for `all'.\nbison -y -d preproc.y\npreproc.y:5560: fatal error: maximum table size (32767) exceeded\nmake[5]: *** [preproc.h] Error 1\nmake[4]: *** [install] Error 2\nmake[3]: *** [install] Error 2\nmake[2]: *** [install] Error 2\nmake[1]: *** [install] Error 2\n\n</snip>\n\n-pmb\n\n\n",
"msg_date": "Tue, 29 Oct 2002 14:38:40 -0800",
"msg_from": "Peter Bierman <bierman@apple.com>",
"msg_from_op": false,
"msg_subject": "CVS TOT fails to build on MacOSX 10.2.1"
},
{
"msg_contents": "Peter Bierman <bierman@apple.com> writes:\n> So first off, what's the best way to tell from a cvs snapshot which\n> \"release\" (if any) that snapshot is?\n\nconfigure.in, perhaps?\n\n> 'make runcheck' in src/test/regress/ fails with:\n> \n> bison -y -d preproc.y\n> preproc.y:5560: fatal error: maximum table size (32767) exceeded\n\nYou need bison 1.50 or greater to build the new ecpg (due to a bison\nlimitation).\n\nCheers,\n\nNeil\n\n-- \nNeil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC\n\n",
"msg_date": "29 Oct 2002 18:11:00 -0500",
"msg_from": "Neil Conway <neilc@samurai.com>",
"msg_from_op": false,
"msg_subject": "Re: CVS TOT fails to build on MacOSX 10.2.1"
},
{
"msg_contents": "At 6:11 PM -0500 10/29/02, Neil Conway wrote:\n>Peter Bierman <bierman@apple.com> writes:\n>> So first off, what's the best way to tell from a cvs snapshot which\n>> \"release\" (if any) that snapshot is?\n>\n>configure.in, perhaps?\n\nAh, thanks. 7.3b3 it is then.\n\n\n\n>> 'make runcheck' in src/test/regress/ fails with:\n>> \n>> bison -y -d preproc.y\n>> preproc.y:5560: fatal error: maximum table size (32767) exceeded\n>\n>You need bison 1.50 or greater to build the new ecpg (due to a bison\n>limitation).\n\n\nOk. After upgrading to bison-1.75, all regression tests pass except:\n\n\n*** ./expected/geometry-powerpc-darwin.out Mon Dec 11 08:45:16 2000\n--- ./results/geometry.out Tue Oct 29 15:40:56 2002\n***************\n*** 127,133 ****\n! | (-10,0) | [(-1000000,200),(300000,-40)] | (-9.99715942258202,15.3864610140472)\n--- 127,133 ----\n! | (-10,0) | [(-1000000,200),(300000,-40)] | (-9.99715942258202,15.3864610140473)\n======================================================================\n\n\nPerhaps the change from gcc2.x to 3.x changed floats a bit?\n\n-pmb\n\n\n",
"msg_date": "Tue, 29 Oct 2002 15:47:42 -0800",
"msg_from": "Peter Bierman <bierman@apple.com>",
"msg_from_op": false,
"msg_subject": "7.3b3 passes on MacOSX 10.2.1"
},
{
"msg_contents": "\nYes, OSX 10.2.X seems to have this output on _some_ machines, but not\nothers, and we can't seem to figure out why. Can you tell us more about\nyour machine and cpu?\n\n---------------------------------------------------------------------------\n\nPeter Bierman wrote:\n> At 6:11 PM -0500 10/29/02, Neil Conway wrote:\n> >Peter Bierman <bierman@apple.com> writes:\n> >> So first off, what's the best way to tell from a cvs snapshot which\n> >> \"release\" (if any) that snapshot is?\n> >\n> >configure.in, perhaps?\n> \n> Ah, thanks. 7.3b3 it is then.\n> \n> \n> \n> >> 'make runcheck' in src/test/regress/ fails with:\n> >> \n> >> bison -y -d preproc.y\n> >> preproc.y:5560: fatal error: maximum table size (32767) exceeded\n> >\n> >You need bison 1.50 or greater to build the new ecpg (due to a bison\n> >limitation).\n> \n> \n> Ok. After upgrading to bison-1.75, all regression tests pass except:\n> \n> \n> *** ./expected/geometry-powerpc-darwin.out Mon Dec 11 08:45:16 2000\n> --- ./results/geometry.out Tue Oct 29 15:40:56 2002\n> ***************\n> *** 127,133 ****\n> ! | (-10,0) | [(-1000000,200),(300000,-40)] | (-9.99715942258202,15.3864610140472)\n> --- 127,133 ----\n> ! | (-10,0) | [(-1000000,200),(300000,-40)] | (-9.99715942258202,15.3864610140473)\n> ======================================================================\n> \n> \n> Perhaps the change from gcc2.x to 3.x changed floats a bit?\n> \n> -pmb\n> \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": "Tue, 29 Oct 2002 21:19:15 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 passes on MacOSX 10.2.1"
},
{
"msg_contents": ">> *** ./expected/geometry-powerpc-darwin.out Mon Dec 11 08:45:16 2000\n>> --- ./results/geometry.out Tue Oct 29 15:40:56 2002\n>> ***************\n>> *** 127,133 ****\n>> ! | (-10,0) | [(-1000000,200),(300000,-40)] | (-9.99715942258202,15.3864610140472)\n>> --- 127,133 ----\n>> ! | (-10,0) | [(-1000000,200),(300000,-40)] | (-9.99715942258202,15.3864610140473)\n>> ======================================================================\n\n\nAt 9:19 PM -0500 10/29/02, Bruce Momjian wrote:\n>Yes, OSX 10.2.X seems to have this output on _some_ machines, but not\n>others, and we can't seem to figure out why. Can you tell us more about\n>your machine and cpu?\n\n\nIt's a dual 450MHz G4 with 512MB RAM, running 10.2.1.\n\nIs there any particular type of Mac you'd like me to try it on?\n\n-pmb\n\n\n",
"msg_date": "Tue, 29 Oct 2002 18:47:46 -0800",
"msg_from": "Peter Bierman <bierman@apple.com>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 passes on MacOSX 10.2.1"
},
{
"msg_contents": "Peter Bierman wrote:\n> >> *** ./expected/geometry-powerpc-darwin.out Mon Dec 11 08:45:16 2000\n> >> --- ./results/geometry.out Tue Oct 29 15:40:56 2002\n> >> ***************\n> >> *** 127,133 ****\n> >> ! | (-10,0) | [(-1000000,200),(300000,-40)] | (-9.99715942258202,15.3864610140472)\n> >> --- 127,133 ----\n> >> ! | (-10,0) | [(-1000000,200),(300000,-40)] | (-9.99715942258202,15.3864610140473)\n> >> ======================================================================\n> \n> \n> At 9:19 PM -0500 10/29/02, Bruce Momjian wrote:\n> >Yes, OSX 10.2.X seems to have this output on _some_ machines, but not\n> >others, and we can't seem to figure out why. Can you tell us more about\n> >your machine and cpu?\n> \n> \n> It's a dual 450MHz G4 with 512MB RAM, running 10.2.1.\n> \n> Is there any particular type of Mac you'd like me to try it on?\n\nYea, one that gets the original value. We know they are out there, but\ncan't figure out which ones they are. Sorry I can't be more specific.\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, 29 Oct 2002 21:56:58 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 passes on MacOSX 10.2.1"
},
{
"msg_contents": "On 29/10/02 6:02 pm, \"Tom Lane\" <tgl@sss.pgh.pa.us> wrote:\n\n> Adam Witney <awitney@sghms.ac.uk> writes:\n>> Just to update the list of supported platforms, 7.3b3 compiles and passes\n>> all the regression tests on MacOSX 10.2.1\n> \n>> Although don't know if this is relevant but this appears when running the\n>> tests:\n> \n>> parallel group (20 tests): ./pg_regress: fork: Resource temporarily\n>> unavailable\n>> ./pg_regress: fork: Resource temporarily unavailable\n>> comments lseg box path timetz point circle reltime tinterval date inet\n>> interval timestamp time abstime polygon timestamptz oidjoins\n> \n> This suggests that you are hitting the per-user limit on the number of\n> processes you can have; try raising that limit.\n> \n> I'd expect one of the tests not to have been run when that message\n> appears; did you really get successful matches for all tests?\n> \n> regards, tom lane\n\nIt appears that my ignorance got the better of me.... It was the first time\nI had run the regression tests on any PostgreSQL installation. But I think I\nam getting the same problems as others. below is the last part of the\nregression tests (I had taken the \"All 15 tests passed\" as a success!)\n\nLet me know if I can be of any assistance in further checking this out\n\n============== running regression test queries ==============\nparallel group (13 tests): char int4 boolean name varchar float8 bit text\nint2 oid int8 float4 numeric\n boolean ... ok\n char ... ok\n name ... ok\n varchar ... ok\n text ... ok\n int2 ... ok\n int4 ... ok\n int8 ... ok\n oid ... ok\n float4 ... ok\n float8 ... ok\n bit ... ok\n numeric ... ok\ntest strings ... ok\ntest numerology ... ok\nparallel group (20 tests): ./pg_regress: fork: Resource temporarily\nunavailable\n./pg_regress: fork: Resource temporarily unavailable\n comments lseg box path timetz point circle reltime tinterval date inet\ninterval timestamp time abstime polygon timestamptz oidjoins==============\nshutting down postmaster ==============\n\n======================\n All 15 tests passed.\n======================\n\nrm regress.o\n\n\n\n-- \nThis message has been scanned for viruses and\ndangerous content by MailScanner, and is\nbelieved to be clean.\n\n",
"msg_date": "Wed, 30 Oct 2002 09:25:51 +0000",
"msg_from": "Adam Witney <awitney@sghms.ac.uk>",
"msg_from_op": true,
"msg_subject": "Re: 7.3b3 on MacOSX 10.2.1"
},
{
"msg_contents": "Adam Witney <awitney@sghms.ac.uk> writes:\n> It appears that my ignorance got the better of me.... It was the first time\n> I had run the regression tests on any PostgreSQL installation. But I think I\n> am getting the same problems as others. below is the last part of the\n> regression tests (I had taken the \"All 15 tests passed\" as a success!)\n\nNo, there should be ~80 tests in all. I'm not sure why the pg_regress\nscript is failing to process the remaining tests when this happens; any\nideas out there?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 30 Oct 2002 12:53:32 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 on MacOSX 10.2.1 "
},
{
"msg_contents": "Peter Bierman <bierman@apple.com> writes:\n> Ok. After upgrading to bison-1.75, all regression tests pass except:\n> *** ./expected/geometry-powerpc-darwin.out Mon Dec 11 08:45:16 2000\n\n> Perhaps the change from gcc2.x to 3.x changed floats a bit?\n\nCould be. We had previous reports of the same diff on OSX 10.2 with\na G4 processor, so I was wondering if it was hardware or software\ndifferences (geometry-powerpc-darwin.out matches exactly on my G3\nlaptop running OSX 10.1). I have a 10.2 CD and am planning to update\nsometime soon to verify this by experiment. Or, if you can confirm\nthat you had passes before on the same hardware you're using, we could\nconclude that it's the OS update that counts.\n\nIn any case, what's the output from config.guess on your machine?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 30 Oct 2002 13:11:09 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 passes on MacOSX 10.2.1 "
},
{
"msg_contents": "Tom Lane writes:\n\n> No, there should be ~80 tests in all. I'm not sure why the pg_regress\n> script is failing to process the remaining tests when this happens; any\n> ideas out there?\n\nIt appears that the shell simply aborts on fork failure. Example:\n\npeter ~$ cat test.sh\nulimit -u 30\n\nfor i in $(seq 1 25); do\n echo $i\n ( sleep 5 ) &\n echo ok\ndone\n\nwait\necho \"all done\"\npeter ~$ sh test.sh\n1\nok\n2\ntest.sh: fork: Resource temporarily unavailable\npeter ~$\n\nIn pg_regress, the big loop is probably done in a subshell, because it's\nin a pipe, so the outer script can complete. Detecting failures in\ncommands in a pipe is of course a bit tricky.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Wed, 30 Oct 2002 20:35:52 +0100 (CET)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 on MacOSX 10.2.1 "
},
{
"msg_contents": "I said:\n> Peter Bierman <bierman@apple.com> writes:\n>> Perhaps the change from gcc2.x to 3.x changed floats a bit?\n\n> Could be. We had previous reports of the same diff on OSX 10.2 with\n> a G4 processor, so I was wondering if it was hardware or software\n> differences (geometry-powerpc-darwin.out matches exactly on my G3\n> laptop running OSX 10.1). I have a 10.2 CD and am planning to update\n> sometime soon to verify this by experiment.\n\nI have done the update and now I get the \"...40473\" output on the same\nhardware that useta pass. So it's clearly a software-version issue.\n\nIs it worth carrying two expected files for OS X 10.1 and 10.2? I'm\ninclined to think not, and am leaning towards updating the expected\nfile. Comments?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 01 Nov 2002 01:30:28 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 passes on MacOSX 10.2.1 "
},
{
"msg_contents": "At 1:30 AM -0500 11/1/02, Tom Lane wrote:\n>I said:\n>> Peter Bierman <bierman@apple.com> writes:\n>>> Perhaps the change from gcc2.x to 3.x changed floats a bit?\n>\n>> Could be. We had previous reports of the same diff on OSX 10.2 with\n>> a G4 processor, so I was wondering if it was hardware or software\n>> differences (geometry-powerpc-darwin.out matches exactly on my G3\n>> laptop running OSX 10.1). I have a 10.2 CD and am planning to update\n>> sometime soon to verify this by experiment.\n>\n>I have done the update and now I get the \"...40473\" output on the same\n>hardware that useta pass. So it's clearly a software-version issue.\n>\n>Is it worth carrying two expected files for OS X 10.1 and 10.2? I'm\n>inclined to think not, and am leaning towards updating the expected\n>file. Comments?\n\n\nI'm 90% certain that the difference is caused by GCC 2.95 vs 3.1.\n(10.0 & 10.1 vs 10.2)\n\nIf you can easily pick the right file based which GCC compiled it, that'd be ideal.\n\n-pmb\n\n\n",
"msg_date": "Fri, 1 Nov 2002 13:11:58 -0800",
"msg_from": "Peter Bierman <bierman@apple.com>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 passes on MacOSX 10.2.1"
},
{
"msg_contents": "Peter Bierman <bierman@apple.com> writes:\n> At 1:30 AM -0500 11/1/02, Tom Lane wrote:\n>> Is it worth carrying two expected files for OS X 10.1 and 10.2? I'm\n>> inclined to think not, and am leaning towards updating the expected\n>> file. Comments?\n\n> I'm 90% certain that the difference is caused by GCC 2.95 vs 3.1.\n\nProbably.\n\n> If you can easily pick the right file based which GCC compiled it, that'd be ideal.\n\nNo, we can't easily do that. We could conditionalize it on the OS\nversion, but I don't think it's worth the trouble. I've committed a\nchange to the expected file so that OSX 10.2 will pass cleanly, and\nolder versions will have the one-digit difference instead.\n\nThis whole issue should go away in PG 7.4, unless someone objects to the\ncurrent plan for making float output precision adjustable. We'll back\noff the number of displayed digits in the geometry test by one or two\nplaces, and hopefully need only one or a very few geometry comparison\nfiles.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 01 Nov 2002 16:31:48 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 passes on MacOSX 10.2.1 "
},
{
"msg_contents": "At 4:31 PM -0500 11/1/02, Tom Lane wrote:\n>Peter Bierman <bierman@apple.com> writes:\n>> At 1:30 AM -0500 11/1/02, Tom Lane wrote:\n>>> Is it worth carrying two expected files for OS X 10.1 and 10.2? I'm\n>>> inclined to think not, and am leaning towards updating the expected\n>>> file. Comments?\n>\n>> I'm 90% certain that the difference is caused by GCC 2.95 vs 3.1.\n>\n>Probably.\n\n\nI had to do a bunch of updates to my 10.1.x system, but I can now verify that 10.1.5 builds and runs 7.3b3 regression test with no failures.\n\n\n>> If you can easily pick the right file based which GCC compiled it, that'd be ideal.\n>\n>No, we can't easily do that. We could conditionalize it on the OS\n>version, but I don't think it's worth the trouble. I've committed a\n>change to the expected file so that OSX 10.2 will pass cleanly, and\n>older versions will have the one-digit difference instead.\n\nThat's fine. If someone gets ambitious, the uname -a from the two differing versions are:\n\n10.1.5\nDarwin bierpe3 5.5 Darwin Kernel Version 5.5: Thu May 30 14:51:26 PDT 2002; root:xnu/xnu-201.42.3.obj~1/RELEASE_PPC Power Macintosh powerpc\n\n10.2.1\nDarwin cmos.apple.com 6.1 Darwin Kernel Version 6.1: Fri Sep 6 23:24:34 PDT 2002; root:xnu/xnu-344.2.obj~2/RELEASE_PPC Power Macintosh powerpc\n\n\n>This whole issue should go away in PG 7.4, unless someone objects to the\n>current plan for making float output precision adjustable. We'll back\n>off the number of displayed digits in the geometry test by one or two\n>places, and hopefully need only one or a very few geometry comparison\n>files.\n\nExcellent.\n\n-pmb\n\n\n",
"msg_date": "Fri, 1 Nov 2002 14:59:42 -0800",
"msg_from": "Peter Bierman <bierman@apple.com>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 passes on MacOSX 10.2.1"
}
] |
[
{
"msg_contents": "RH 8.0 has perl 5.8, here are the details:\n\n From config.log:\n\nconfigure:4159: checking for perl\nconfigure:4177: found /usr/bin/perl\nconfigure:4189: result: /usr/bin/perl\nconfigure:4198: checking for Perl archlibexp\nconfigure:4201: result: /usr/lib/perl5/5.8.0/i386-linux-thread-multi\nconfigure:4203: checking for Perl privlibexp\nconfigure:4206: result: /usr/lib/perl5/5.8.0\nconfigure:4208: checking for Perl useshrplib\nconfigure:4211: result: true\nconfigure:4214: checking for flags to link embedded Perl\nconfigure:4219: result: -L/usr/local/lib \n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/auto/DynaLoader/DynaLoader.a \n-L/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE -lperl -lnsl -ldl \n-lm -lpthread -lc -lcrypt -lutil\n\nfrom gmake:\ngmake[3]: Entering directory \n`/localhome/rovero/postgresql-7.3b3/src/pl/plperl'\ngcc -O2 -fpic -I. -I/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE \n-I../../..\n/src/include -c -o plperl.o plperl.c\nIn file included from \n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/op.h:480\n,\n from \n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/perl.h:2\n209,\n from plperl.c:61:\n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/reentr.h:602: field \n`_crypt_st\nruct' has incomplete type\nIn file included from \n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/perl.h:3\n368,\n from plperl.c:61:\n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:246: parse \nerror befor\ne \"off64_t\"\n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:248: parse \nerror befor\ne \"Perl_do_sysseek\"\n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:248: parse \nerror befor\ne \"off64_t\"\n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:248: warning: \ndata def\ninition has no type or storage class\n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:249: parse \nerror befor\ne \"Perl_do_tell\"\n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:249: warning: \ndata def\ninition has no type or storage class\n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:1378: parse \nerror befo\nre \"Perl_PerlIO_tell\"\n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:1378: warning: \ndata de\nfinition has no type or storage class\n/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:1379: parse \nerror befo\nre \"off64_t\"\ngmake[3]: *** [plperl.o] Error 1\ngmake[3]: Leaving directory \n`/localhome/rovero/postgresql-7.3b3/src/pl/plperl'\ngmake[2]: *** [all] Error 2\ngmake[2]: Leaving directory `/localhome/rovero/postgresql-7.3b3/src/pl'\ngmake[1]: *** [all] Error 2\ngmake[1]: Leaving directory `/localhome/rovero/postgresql-7.3b3/src'\ngmake: *** [all] Error 2\n\n\n-- \nP. J. \"Josh\" Rovero Sonalysts, Inc.\nEmail: rovero@sonalysts.com www.sonalysts.com 215 Parkway North\nWork: (860)326-3671 or 442-4355 Waterford CT 06385\n***********************************************************************\n\n",
"msg_date": "Mon, 28 Oct 2002 07:41:54 -0500",
"msg_from": "\"P.J. \\\"Josh\\\" Rovero\" <rovero@sonalysts.com>",
"msg_from_op": true,
"msg_subject": "7.3b3 build --with-perl fails on RH 8.0"
},
{
"msg_contents": "Change directory in src/pl/plperl\nand execute gcc command with option -D_GNU_SOURCE \n(2 files)\nand then continue make form base directory\nYou must add /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/libperl.so in \n/etc/ld/so/conf or in LD_LIBRARY_PATH\n\nregards\nHaris Peco\n\nOn Monday 28 October 2002 12:41 pm, P.J. \\\"Josh\\\" Rovero wrote:\n> RH 8.0 has perl 5.8, here are the details:\n>\n> From config.log:\n>\n> configure:4159: checking for perl\n> configure:4177: found /usr/bin/perl\n> configure:4189: result: /usr/bin/perl\n> configure:4198: checking for Perl archlibexp\n> configure:4201: result: /usr/lib/perl5/5.8.0/i386-linux-thread-multi\n> configure:4203: checking for Perl privlibexp\n> configure:4206: result: /usr/lib/perl5/5.8.0\n> configure:4208: checking for Perl useshrplib\n> configure:4211: result: true\n> configure:4214: checking for flags to link embedded Perl\n> configure:4219: result: -L/usr/local/lib\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/auto/DynaLoader/DynaLoader.a\n> -L/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE -lperl -lnsl -ldl\n> -lm -lpthread -lc -lcrypt -lutil\n>\n> from gmake:\n> gmake[3]: Entering directory\n> `/localhome/rovero/postgresql-7.3b3/src/pl/plperl'\n> gcc -O2 -fpic -I. -I/usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE\n> -I../../..\n> /src/include -c -o plperl.o plperl.c\n> In file included from\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/op.h:480\n> ,\n> from\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/perl.h:2\n> 209,\n> from plperl.c:61:\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/reentr.h:602: field\n> `_crypt_st\n> ruct' has incomplete type\n> In file included from\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/perl.h:3\n> 368,\n> from plperl.c:61:\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:246: parse\n> error befor\n> e \"off64_t\"\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:248: parse\n> error befor\n> e \"Perl_do_sysseek\"\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:248: parse\n> error befor\n> e \"off64_t\"\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:248: warning:\n> data def\n> inition has no type or storage class\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:249: parse\n> error befor\n> e \"Perl_do_tell\"\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:249: warning:\n> data def\n> inition has no type or storage class\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:1378: parse\n> error befo\n> re \"Perl_PerlIO_tell\"\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:1378: warning:\n> data de\n> finition has no type or storage class\n> /usr/lib/perl5/5.8.0/i386-linux-thread-multi/CORE/proto.h:1379: parse\n> error befo\n> re \"off64_t\"\n> gmake[3]: *** [plperl.o] Error 1\n> gmake[3]: Leaving directory\n> `/localhome/rovero/postgresql-7.3b3/src/pl/plperl'\n> gmake[2]: *** [all] Error 2\n> gmake[2]: Leaving directory `/localhome/rovero/postgresql-7.3b3/src/pl'\n> gmake[1]: *** [all] Error 2\n> gmake[1]: Leaving directory `/localhome/rovero/postgresql-7.3b3/src'\n> gmake: *** [all] Error 2\n\n",
"msg_date": "Mon, 28 Oct 2002 17:14:24 +0000",
"msg_from": "snpe <snpe@snpe.co.yu>",
"msg_from_op": false,
"msg_subject": "Re: 7.3b3 build --with-perl fails on RH 8.0"
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Tommy [mailto:tmartin1@telocity.com] \n> Sent: 28 October 2002 03:56\n> To: pgsql-hackers@postgresql.org\n> Subject: [HACKERS] Pgsql and .NET (Help!)\n> \n> \n> I have been trying for 3 days to connect to my postgres linux \n> box via vb .net on a windows box. I have to make this work somehow.\n> \n> I found npgsql provider and tried it but could not make it \n> connect either. I would prefer to use the OleDbConnection if possible.\n> \n> Has anyone been able to connect via the odbc driver and .NET?\n\nYes, instructions are on http://odbc.postgresql.org, though they are for\nC#, but they are using .Net. Npgsql does work quite well as well\nthough...\n\nRegards, Dave\n",
"msg_date": "Mon, 28 Oct 2002 13:28:09 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: Pgsql and .NET (Help!)"
}
] |
[
{
"msg_contents": "Dave,\n Thanks for the details, I've copied this reply back to the PostgreSQL\nguys as well. \n\nLER\n\nOn Mon, 2002-10-28 at 09:00, Dave Prosser wrote:\n> Larry Rosenman wrote:\n> > From: Tom Lane <tgl@sss.pgh.pa.us>\n> > To: Larry Rosenman <ler@lerctr.org>\n> > Cc: pgsql-hackers@postgresql.org\n> > Subject: Re: [HACKERS] UnixWare 7.1.3 (BETA), C99 compiler, current CVS, error...\n> > Date: 26 Oct 2002 11:07:13 -0400\n> > \n> > Larry Rosenman <ler@lerctr.org> writes:\n> > > Without specifying the -Xb switch to kill the C99 interpretation of\n> > > inline, I get the following from current CVS:\n> > \n> > > UX:acomp: ERROR: \"tuplesort.c\", line 1854: \"inline\" functions cannot use\n> > > \"static\" identifier: myFunctionCall2\n> > \n> > I don't understand what it's unhappy about. My C99 draft sez\n> > \n> > [#6] Any function with internal linkage can be an inline\n> > function.\n> > \n> > so the text of the message is surely not what they are really\n> > complaining about? Or is the compiler broken?\n> \n> There is a contraint (i.e., a diagnostic is required) in 6.7.4 Function Specifiers\n> that says:\n> \n> An inline definition of a function with external linkage shall not contain a\n> definition of a modifiable object with static storage duration, and shall not\n> contain a reference to an identifier with internal linkage.\n> \n> Line 1854 is\n> \tif (DatumGetBool(myFunctionCall2(sortFunction, datum1, datum2)))\n> where myFunctionCall2() is a static function defined above ApplySortFunction().\n> It's not the inlinedness--a word?--of myFunctionCall2() that's the problem,\n> it's that myFunctionCall2() is static and that ApplySortFunction() is inline'd.\n> \n> You wrote in your follow up:\n> >After reading a little further, it seems that the brain damage is in the\n> >standard, not the compiler :-(. It looks like C99's notion of a\n> >function that is both global and inline is that you must provide *two*\n> >definitions of the function, one marked inline and one not; moreover,\n> >these must appear in separate translation units. What in the world were\n> >those people smoking? That's a recipe for maintenance problems (edit\n> >one definition, forget to edit the other), not to mention completely at\n> >variance with the de facto standard behavior of inline that's been\n> >around for a long time.\n> \n> The C committee's view of inline does not match the historic GCC one.\n> They were trying to find a middle ground that was fully compatible with\n> the C++ inline, while not requiring any fancy code generation tricks.\n> In other words, that C could still be compiled with a one-pass compiler.\n> \n> The motivation for this restriction is to make sure that all instances\n> of an inline function that's visible outside of the compilation unit\n> are identical. Having the same sequence of tokens isn't good enough\n> if there are references to identifiers that could well be different in\n> differing compilation units.\n> \n> Until the open source base (and GCC) get around to matching the C99\n> inline model, I generally attempt to compile open source with \"cc -Xb\"\n> as that eliminates recognition of inline as a keyword, and thus doesn't\n> get into the issues with the clashes between the two models.\n> \n> >My inclination is to change the code for ApplySortFunction to look like\n> >\n> > #if defined(__GNUC__)\n> > __inline__\n> > #endif\n> > int32\n> > ApplySortFunction\n> >\n> >so that the inline optimization only gets done for gcc, which we know\n> >interprets inline sanely. Anyone see a better answer?\n> \n> You've given us one source file. Is ApplySortFunction() really called\n> from other files? Another approach, assuming both this and that the\n> inlining is viewed as important for its three calls within this file,\n> to have a front end version of an internal function. To wit:\n> \n> static inline int32\n> StaticApplySortFunction(FmgrInfo *sortFunction, SortFunctionKind kind,\n> Datum datum1, bool isNull1,\n> Datum datum2, bool isNull2)\n> {\n> //etc.\n> }\n> \n> int32\n> ApplySortFunction(FmgrInfo *sortFunction, SortFunctionKind kind,\n> Datum datum1, bool isNull1,\n> Datum datum2, bool isNull2)\n> {\n> \treturn StaticApplySortFunction(sortFunction, kind, datum1, isNull1, datum2, isNull2);\n> }\n> \n> and change all the existing calls within tuplesort.c to use\n> StaticApplySortFunction(). This approach requires no #ifdef's and\n> should have the same effect as the existing source with GCC's view\n> of inline.\n> \n> -- \n> Dave Prosser dfp@caldera.com (908)790-2358 The SCO Group, Murray Hill, NJ\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 Oct 2002 08:54:20 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "Re: [Fwd: Re: UnixWare 7.1.3 (BETA), C99 compiler,"
},
{
"msg_contents": "Larry Rosenman <ler@lerctr.org> forwards:\n>>> My inclination is to change the code for ApplySortFunction to look like\n>>> \n>>> #if defined(__GNUC__)\n>>> __inline__\n>>> #endif\n>>> int32\n>>> ApplySortFunction\n>>> \n>>> so that the inline optimization only gets done for gcc, which we know\n>>> interprets inline sanely. Anyone see a better answer?\n>> \n>> You've given us one source file. Is ApplySortFunction() really called\n>> from other files? Another approach, assuming both this and that the\n>> inlining is viewed as important for its three calls within this file,\n\nYup, both of those are correct: the performance-critical calls are in\nthe same file, but there are others.\n\n>> to have a front end version of an internal function. To wit:\n>> \n>> static inline int32\n>> StaticApplySortFunction(FmgrInfo *sortFunction, SortFunctionKind kind,\n>> Datum datum1, bool isNull1,\n>> Datum datum2, bool isNull2)\n>> {\n>> //etc.\n>> }\n>> \n>> int32\n>> ApplySortFunction(FmgrInfo *sortFunction, SortFunctionKind kind,\n>> Datum datum1, bool isNull1,\n>> Datum datum2, bool isNull2)\n\nGrumble. I suppose we have to do it that way, but it's really pretty\nstupid. Also, won't this fall foul of the original restriction\n(ApplySortFunction referencing the static function myFunctionCall2)?\nIf not, why not?\n\n> Until the open source base (and GCC) get around to matching the C99\n> inline model,\n\nDon't hold your breath... it looks like a net loss in functionality\nfor no gain, from where I sit.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 28 Oct 2002 11:07:09 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: [Fwd: Re: UnixWare 7.1.3 (BETA), C99 compiler, "
},
{
"msg_contents": "Tom Lane wrote:\n> >> static inline int32\n> >> StaticApplySortFunction(FmgrInfo *sortFunction, SortFunctionKind kind,\n> >> Datum datum1, bool isNull1,\n> >> Datum datum2, bool isNull2)\n> >> {\n> >> //etc.\n> >> }\n> >>\n> >> int32\n> >> ApplySortFunction(FmgrInfo *sortFunction, SortFunctionKind kind,\n> >> Datum datum1, bool isNull1,\n> >> Datum datum2, bool isNull2)\n> \n> Grumble. I suppose we have to do it that way, but it's really pretty\n> stupid. Also, won't this fall foul of the original restriction\n> (ApplySortFunction referencing the static function myFunctionCall2)?\n> If not, why not?\n\nThe change is that the inline function referencing the identifiers with\ninternal linkage also has internal linkage -- i.e., they're both static.\nSince there can be only one definition of StaticApplySortFunction(),\nthere's no reason to restrict it's contents.\n\n> > Until the open source base (and GCC) get around to matching the C99\n> > inline model,\n> \n> Don't hold your breath... it looks like a net loss in functionality\n> for no gain, from where I sit.\n\nThe GCC owners are now much more standards-aware and -driven than they\nwere 10+ years ago. I believe that they are going to (if not already)\nhave a way of interpreting \"inline\" according to the C99 standard. It\nwill probably not be the default, given the incompatibilities, but it'll\nbe there, somehow, sometime.\n\nAs far as I'm concerned, the C99 inline doesn't do \"me\" much good either,\nas I cannot use it in headers to take the place of function-like macros\nin standard headers, either. But, this is what you get with Committees--\nspecifications that make everyone unhappy at roughly equivalent levels. :-)\n\nSince \"inline\" (as specified in C99) is pretty much just a hint to the\ncompiler in much the same way that \"register\" was in the good old days,\nit really shouldn't matter too much one way or another, but feel free\nto use it as you desire.\n\n-- \nDave Prosser dfp@caldera.com (908)790-2358 The SCO Group, Murray Hill, NJ\n",
"msg_date": "Mon, 28 Oct 2002 11:32:45 -0500",
"msg_from": "Dave Prosser <dfp@caldera.com>",
"msg_from_op": false,
"msg_subject": "Re: [Fwd: Re: UnixWare 7.1.3 (BETA), C99 compiler,"
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Jason Tishler [mailto:jason@tishler.net] \n> Sent: 28 October 2002 13:33\n> To: Dave Page\n> Cc: Bruce Momjian; PostgreSQL-development; Thomas Lockhart; \n> Tom Lane; Pgsql-Cygwin\n> Subject: Re: [HACKERS] Request for supported platforms\n> \n> \n> Dave,\n> \n> Thanks for the heads up...\n> \n> On Mon, Oct 28, 2002 at 10:31:00AM -0000, Dave Page wrote:\n> > > -----Original Message-----\n> > > From: Bruce Momjian [mailto:pgman@candle.pha.pa.us]\n> > > Sent: 26 October 2002 03:17\n> > > Subject: [HACKERS] Request for supported platforms\n> > >\n> > > Folks. start sending in those plaform reports, OS name and\n> > > version number please.\n> > \n> > CYGWIN_NT-5.1 PC9 1.3.10(0.51/3/2) 2002-02-25 11:14 i686 unknown\n> ^^^^^^\n> \n> Please try with Cygwin 1.3.14-1 while I attempt to deal with \n> at least the following Cygwin build issues with PostgreSQL \n> CVS as of today at about 7:00 AM EST:\n\nOk, but this is going to take a while as few of the mirrors seem to have\nthis release yet. I also need to download a new set of everything for\nreasons I won't go into.\n\nIs there actually a reason for this though or are you just trying to\nkeep me busy? :-) It can't be a good thing for us to require that people\nupgrade to the latest release of their OS.\n\n> 2. Cygwin bison limit exceeded:\n> \n> make[4]: Entering directory \n> `/home/jt/src/pgsql/src/interfaces/ecpg/preproc'\n> [snip]\n> bison -y -d preproc.y\n> preproc.y:5560: fatal error: maximum table size (32767) exceeded\n\nI believe a new bison is required now. Don't know much about it other\nthan ecpg hit some limit or other and much discussion followed. Iirc,\nit's only an issue when compiling from CVS, not a tarball.\n\nRegards, Dave.\n",
"msg_date": "Mon, 28 Oct 2002 14:56:01 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "On Mon, 2002-10-28 at 10:20, Jason Tishler wrote:\n\n> > > 2. Cygwin bison limit exceeded:\n> > > \n> > > make[4]: Entering directory \n> > > `/home/jt/src/pgsql/src/interfaces/ecpg/preproc'\n> > > [snip]\n> > > bison -y -d preproc.y\n> > > preproc.y:5560: fatal error: maximum table size (32767) exceeded\n> > \n> > I believe a new bison is required now. Don't know much about it other\n> > than ecpg hit some limit or other and much discussion followed. Iirc,\n> > it's only an issue when compiling from CVS, not a tarball.\n> \n> The above should help save me some time.\n1.50 of Bison fixed the issue, 1.75 works as well (current public\nrelease). \n\nJust my $.02. \n\n\n> \n> Thanks,\n> Jason\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 6: Have you searched our list archives?\n> \n> http://archives.postgresql.org\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 Oct 2002 10:19:15 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "Dave,\n\nOn Mon, Oct 28, 2002 at 02:56:01PM -0000, Dave Page wrote:\n> > -----Original Message-----\n> > From: Jason Tishler [mailto:jason@tishler.net] \n> > Sent: 28 October 2002 13:33\n> > Subject: Re: [HACKERS] Request for supported platforms\n> > \n> > Please try with Cygwin 1.3.14-1 while I attempt to deal with at\n> > least the following Cygwin build issues with PostgreSQL CVS as of\n> > today at about 7:00 AM EST:\n> \n> Ok, but this is going to take a while as few of the mirrors seem to\n> have this release yet. I also need to download a new set of everything\n> for reasons I won't go into.\n\nMy WAG is that you will be able to upgrade your Cygwin installation\nbefore I fix the Cygwin build issues. :,)\n\n> Is there actually a reason for this though or are you just trying to\n> keep me busy? :-) It can't be a good thing for us to require that\n> people upgrade to the latest release of their OS.\n\nAgreed, but sometimes a new Cygwin release fixes some problems and\nbreaks others...\n\n> > 2. Cygwin bison limit exceeded:\n> > \n> > make[4]: Entering directory \n> > `/home/jt/src/pgsql/src/interfaces/ecpg/preproc'\n> > [snip]\n> > bison -y -d preproc.y\n> > preproc.y:5560: fatal error: maximum table size (32767) exceeded\n> \n> I believe a new bison is required now. Don't know much about it other\n> than ecpg hit some limit or other and much discussion followed. Iirc,\n> it's only an issue when compiling from CVS, not a tarball.\n\nThe above should help save me some time.\n\nThanks,\nJason\n",
"msg_date": "Mon, 28 Oct 2002 11:20:16 -0500",
"msg_from": "Jason Tishler <jason@tishler.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "Dave,\n\nOn Mon, Oct 28, 2002 at 11:20:16AM -0500, Jason Tishler wrote:\n> On Mon, Oct 28, 2002 at 02:56:01PM -0000, Dave Page wrote:\n> > Ok, but this is going to take a while as few of the mirrors seem to\n> > have this release yet. I also need to download a new set of everything\n> > for reasons I won't go into.\n> \n> My WAG is that you will be able to upgrade your Cygwin installation\n> before I fix the Cygwin build issues. :,)\n\nI guess my WAG was wrong... :,)\n\nUnder the following platform:\n\n $ uname -a\n CYGWIN_NT-5.0 TISHLERJASON 1.3.14(0.62/3/2) 2002-10-23 14:47 i686 unknown\n\nall make installcheck tests pass except for horology. However, the\nhorology failure is to be expected due to the recent time change.\n\nUnfortunately, there are some Cygwin build issues that I need to\naddress:\n\n1. Cygwin bison needs to be upgraded from 1.35 to 1.75 (i.e., 1.50+) to\nprocess src/interfaces/ecpg/preproc/preproc.y successfully. I will post\nto the Cygwin mailing list asking the maintainer for this upgrade.\n\n2. The following fseeko/ftello ifdef in src/include/pg_config.h.in:\n\n #ifndef HAVE_FSEEKO\n #define fseeko(a, b, c) fseek((a), (b), (c))\n #define ftello(a) ftell((a))\n #endif\n\nconflicts with the following Cygwin /usr/include/stdio.h entries:\n\n int _EXFUN(fseeko, (FILE *, off_t, int));\n off_t _EXFUN(ftello, ( FILE *));\n\nUnfortunately, I'm not sure what is the best way to solve this one yet.\nAny suggestions would be appreciated.\n\nThanks,\nJason\n",
"msg_date": "Mon, 28 Oct 2002 15:41:49 -0500",
"msg_from": "Jason Tishler <jason@tishler.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Tom Lane [mailto:tgl@sss.pgh.pa.us] \n> Sent: 28 October 2002 15:53\n> To: Jason Tishler\n> Cc: Dave Page; Bruce Momjian; PostgreSQL-development; Thomas \n> Lockhart; Pgsql-Cygwin\n> Subject: Re: [HACKERS] Request for supported platforms \n> \n> \n> >> Make installcheck also failed on horology, copy2 and domain - see \n> >> attached output.\n> >> \n> >> The clocks changed here on Saturday night, so I guess that \n> shouldn't \n> >> have caused the first error (or should the docs be updated?).\n> \n> Horology failures are normal for a day or so on either side \n> of a DST change --- see the \"regression tests interpretation\" \n> docs. I have no time right now to examine the other diffs.\n\nThe docs say:\n\n====\nSome of the queries in the timestamp test will fail if you run the test\non\nthe day of a daylight-savings time changeover, or the day before or\nafter\none.\n====\n\nClocks changed at midnight Saturday so I figured a Monday morning run\nshould be OK. Do they actually change on Sunday at 00:00:00 I wonder?\nI'll try again tomorrow.\n\nRegards, Dave.\n",
"msg_date": "Mon, 28 Oct 2002 15:58:09 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Request for supported platforms "
},
{
"msg_contents": "\"Dave Page\" <dpage@vale-housing.co.uk> writes:\n>> Horology failures are normal for a day or so on either side \n>> of a DST change --- see the \"regression tests interpretation\" \n>> docs. I have no time right now to examine the other diffs.\n\n> The docs say:\n\n> Some of the queries in the timestamp test will fail if you run the test\n> on\n> the day of a daylight-savings time changeover, or the day before or\n> after\n> one.\n\n> Clocks changed at midnight Saturday so I figured a Monday morning run\n> should be OK. Do they actually change on Sunday at 00:00:00 I wonder?\n> I'll try again tomorrow.\n\nIn the US, DST changes occur at 02:00 Sunday, so the affected queries\nactually fail starting at 00:00 Sunday and ending 00:00 Tuesday ---\nbut that's local time in PST8PDT. The docs are vague because your local\ntime might vary considerably from that ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 28 Oct 2002 13:22:00 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms "
}
] |
[
{
"msg_contents": "\n> > > The question is *which* seek APIs we need to support. Are there any\n> > > besides fseeko() and fgetpos()?\n> > \n> > On AIX we have \n> > int fseeko64 (FILE* Stream, off64_t Offset, int Whence);\n> > which is intended for large file access for programs that do NOT\n> > #define _LARGE_FILES\n> > \n> > It is functionality that is available if _LARGE_FILE_API is defined,\n> > which is the default if _LARGE_FILES is not defined.\n> > \n> > That would have been my preferred way of handling large files on AIX\n> > in the two/three? places that need it (pg_dump/restore, psql and backend COPY).\n> > This would have had the advantage that off_t is not 64 bit in all other places\n> > where it is actually not needed, no ?\n> \n> OK, I am focusing on AIX now. I don't think we can go down the road of\n> saying where large file support is needed or not needed. I think for\n> each platform either we support large files or we don't. Is there a way\n> to have off_t be 64 bits everywhere, and if it is, why wouldn't we just\n> enable that rather than poke around figuring out where it is needed?\n\nif _LARGE_FILES is defined, off_t is 64 bits on AIX (and fseeko works). \nThe problem with flex is, that the generated c file does #include <unistd.h>\nbefore we #include \"postgres.h\".\nIn this situation _LARGE_FILES is not defined for unistd.h and unistd.h\nchooses to define _LARGE_FILE_API, those two are not compatible.\n\nIf a general off_t of 64 bits is no performance problem, we should focus\non fixing the #include <unistd.h> issue, and forget what I wanted/hinted.\nPeter E. has a patch for this in his pipeline. I can give it a second try \ntomorrow.\n\nSorry for the late answer, I am very pressed currently :-(\nAndreas\n",
"msg_date": "Mon, 28 Oct 2002 17:35:57 +0100",
"msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>",
"msg_from_op": true,
"msg_subject": "Re: pg_dump and large files - is this a problem?"
},
{
"msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n> The problem with flex is, that the generated c file does #include <unistd.h>\n> before we #include \"postgres.h\".\n> In this situation _LARGE_FILES is not defined for unistd.h and unistd.h\n> chooses to define _LARGE_FILE_API, those two are not compatible.\n\nYeah. AFAICS the only way around this is to avoid doing any I/O\noperations in the flex-generated files. Fortunately, that's not much\nof a restriction.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Mon, 28 Oct 2002 13:15:50 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and large files - is this a problem? "
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Jason Tishler [mailto:jason@tishler.net] \n> Sent: 28 October 2002 20:42\n> To: Dave Page\n> Cc: Bruce Momjian; PostgreSQL-development; Thomas Lockhart; \n> Tom Lane; Pgsql-Cygwin\n> Subject: Re: [HACKERS] Request for supported platforms\n> \n> \n> Dave,\n> \n> On Mon, Oct 28, 2002 at 11:20:16AM -0500, Jason Tishler wrote:\n> >\n> > My WAG is that you will be able to upgrade your Cygwin installation \n> > before I fix the Cygwin build issues. :,)\n> \n> I guess my WAG was wrong... :,)\n\nI've been meaning to ask this for a while - what exactly is a WAG? :-)\n\n> 1. Cygwin bison needs to be upgraded from 1.35 to 1.75 (i.e., \n> 1.50+) to process src/interfaces/ecpg/preproc/preproc.y \n> successfully. I will post to the Cygwin mailing list asking \n> the maintainer for this upgrade.\n\nOK. This shouldn't stop a release though I assume, only a build from\nCVS.\n\n> 2. The following fseeko/ftello ifdef in src/include/pg_config.h.in:\n> \n> #ifndef HAVE_FSEEKO\n> #define fseeko(a, b, c) fseek((a), (b), (c))\n> #define ftello(a) ftell((a))\n> #endif\n> \n> conflicts with the following Cygwin /usr/include/stdio.h entries:\n> \n> int _EXFUN(fseeko, (FILE *, off_t, int));\n> off_t _EXFUN(ftello, ( FILE *));\n> \n> Unfortunately, I'm not sure what is the best way to solve \n> this one yet. Any suggestions would be appreciated.\n\nYes, I'm seeing errors with this on my updated Cygwin very early in the\nbuild. I did think it was my hacked about installation, but I guess not!\n\nUnfortunately though, I don't know the answer either, but I guess the\nchange was in Cygwin as I didn't see it before you asked me to upgrade.\n\nRegards, Dave.\n",
"msg_date": "Mon, 28 Oct 2002 20:58:12 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "On Mon, 2002-10-28 at 14:58, Dave Page wrote:\n> \n> > >\n> > > My WAG is that you will be able to upgrade your Cygwin installation \n> > > before I fix the Cygwin build issues. :,)\n> > \n> > I guess my WAG was wrong... :,)\n> \n> I've been meaning to ask this for a while - what exactly is a WAG? :-)\nWild-A**ed-Guess, I would presume. \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": "28 Oct 2002 15:00:15 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "Dave,\n\nOn Mon, Oct 28, 2002 at 08:58:12PM -0000, Dave Page wrote:\n> > -----Original Message-----\n> > From: Jason Tishler [mailto:jason@tishler.net] \n> > Sent: 28 October 2002 20:42\n> > Subject: Re: [HACKERS] Request for supported platforms\n> > \n> > On Mon, Oct 28, 2002 at 11:20:16AM -0500, Jason Tishler wrote:\n> > > My WAG is that you will be able to upgrade your Cygwin\n> > > installation before I fix the Cygwin build issues. :,)\n> > \n> > I guess my WAG was wrong... :,)\n> \n> I've been meaning to ask this for a while - what exactly is a WAG? :-)\n\nLarry was kind enough to answer this one for me. :,)\n\n> > 1. Cygwin bison needs to be upgraded from 1.35 to 1.75 (i.e., \n> > 1.50+) to process src/interfaces/ecpg/preproc/preproc.y \n> > successfully. I will post to the Cygwin mailing list asking \n> > the maintainer for this upgrade.\n> \n> OK. This shouldn't stop a release though I assume, only a build from\n> CVS.\n\nYes. Nevertheless, I have posted my request:\n\n http://cygwin.com/ml/cygwin/2002-10/msg01740.html\n\n> > 2. The following fseeko/ftello ifdef in src/include/pg_config.h.in:\n> > \n> > #ifndef HAVE_FSEEKO\n> > #define fseeko(a, b, c) fseek((a), (b), (c))\n> > #define ftello(a) ftell((a))\n> > #endif\n> > \n> > conflicts with the following Cygwin /usr/include/stdio.h entries:\n> > \n> > int _EXFUN(fseeko, (FILE *, off_t, int));\n> > off_t _EXFUN(ftello, ( FILE *));\n> > \n> > Unfortunately, I'm not sure what is the best way to solve \n> > this one yet. Any suggestions would be appreciated.\n\nI found a solution to the above which will hopefully find its way into\nthe next Cygwin release:\n\n http://cygwin.com/ml/cygwin-patches/2002-q4/msg00042.html\n \n> Yes, I'm seeing errors with this on my updated Cygwin very early in\n> the build. I did think it was my hacked about installation, but I\n> guess not!\n\nA quick and *dirty* fix for this problem is to temporarily delete the\nabove two entries from your stdio.h file.\n\nJason\n",
"msg_date": "Tue, 29 Oct 2002 09:47:35 -0500",
"msg_from": "Jason Tishler <jason@tishler.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "This post is just for closure -- both of the issues below have been\nresolved:\n\nOn Tue, Oct 29, 2002 at 09:47:35AM -0500, Jason Tishler wrote:\n> > > 1. Cygwin bison needs to be upgraded from 1.35 to 1.75 (i.e.,\n> > > 1.50+) to process src/interfaces/ecpg/preproc/preproc.y\n> > > successfully. I will post to the Cygwin mailing list asking the\n> > > maintainer for this upgrade.\n> > \n> > OK. This shouldn't stop a release though I assume, only a build from\n> > CVS.\n> \n> Yes. Nevertheless, I have posted my request:\n> \n> http://cygwin.com/ml/cygwin/2002-10/msg01740.html\n\nhttp://cygwin.com/ml/cygwin-announce/2002-10/msg00016.html\n\n> > > 2. The following fseeko/ftello ifdef in src/include/pg_config.h.in:\n> > > \n> > > #ifndef HAVE_FSEEKO\n> > > #define fseeko(a, b, c) fseek((a), (b), (c))\n> > > #define ftello(a) ftell((a))\n> > > #endif\n> > > \n> > > conflicts with the following Cygwin /usr/include/stdio.h entries:\n> > > \n> > > int _EXFUN(fseeko, (FILE *, off_t, int));\n> > > off_t _EXFUN(ftello, ( FILE *));\n> > > \n> > > Unfortunately, I'm not sure what is the best way to solve this one\n> > > yet. Any suggestions would be appreciated.\n> \n> I found a solution to the above which will hopefully find its way into\n> the next Cygwin release:\n> \n> http://cygwin.com/ml/cygwin-patches/2002-q4/msg00042.html\n\nhttp://cygwin.com/ml/cygwin-patches/2002-q4/msg00089.html\n\nJason\n\n-- \nPGP/GPG Key: http://www.tishler.net/jason/pubkey.asc or key servers\nFingerprint: 7A73 1405 7F2B E669 C19D 8784 1AFD E4CC ECF4 8EF6\n",
"msg_date": "Thu, 07 Nov 2002 08:42:48 -0500",
"msg_from": "Jason Tishler <jason@tishler.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
}
] |
[
{
"msg_contents": "Is there i faq item or maybe some other way that we can tell users\nthat --disable-rpath should be used when checking regressiontests if you\nhave an install already on the box?\nThe rpath defaults to the installdir even in the temporary install that\nis made when doing \"gmake check\", and that confuses some programs (like\npg_encoding) because it loads the installed libpq. In my case that's an\nold install running in production state.\n\nThere are tools that set the rpath on an already linked executable, but\ni dunno if that feels appropriate. But maybe an warning in \"gmake check\"\nwould be nice.\n\nMagnus\n\n--\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n Programmer/Networker [|] Magnus Naeslund\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n\n",
"msg_date": "Tue, 29 Oct 2002 03:38:24 +0100",
"msg_from": "\"Magnus Naeslund(f)\" <mag@fbab.net>",
"msg_from_op": true,
"msg_subject": "About the Request for supported platforms"
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Bruce Momjian [mailto:pgman@candle.pha.pa.us] \n> Sent: 29 October 2002 04:24\n> To: Dave Page\n> Cc: PostgreSQL-development; Thomas Lockhart; Tom Lane\n> Subject: Re: [HACKERS] Request for supported platforms\n> \n> \n> \n> I have updated CVS and re-added getopt.c, now in /port, and \n> updated win32.mak. That should help.\n\nThanks. Unfortunately not quite there though:\n\nprint.c(1038) : warning C4013: 'pclose' undefined; assuming extern\nreturning int\n\ndescribe.c\ntab-complete.c\ndescribe.c(1462) : warning C4761: integral size mismatch in argument;\nconversion supplied\nmbprint.c\n link.exe @.\\nmc01556.\nprint.obj : error LNK2001: unresolved external symbol _pclose\n.\\Release\\psql.exe : fatal error LNK1120: 1 unresolved externals\nNMAKE : fatal error U1077: 'link.exe' : return code '0x460'\nStop.\nNMAKE : fatal error U1077: '\"c:\\Program Files\\Microsoft Visual\nStudio\\VC98\\bin\\nmake.exe\"' : return code '0x2'\nStop.\n\nAny ideas?\n\nRegards, Dave.\n",
"msg_date": "Tue, 29 Oct 2002 08:21:17 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
},
{
"msg_contents": "Attached is a diff to fix the pclose problem. It turns out there was\ncode in there to make popen/pclose be _popen/_pclose, but it was only in\ncommon.c, even in 7.2.3 (print.c). Not sure how it would compile in the\npast with that. Maybe it didn't. Anyway, this is committed and should\n_help_ with the compile.\n\n---------------------------------------------------------------------------\n\nDave Page wrote:\n> \n> \n> > -----Original Message-----\n> > From: Bruce Momjian [mailto:pgman@candle.pha.pa.us] \n> > Sent: 29 October 2002 04:24\n> > To: Dave Page\n> > Cc: PostgreSQL-development; Thomas Lockhart; Tom Lane\n> > Subject: Re: [HACKERS] Request for supported platforms\n> > \n> > \n> > \n> > I have updated CVS and re-added getopt.c, now in /port, and \n> > updated win32.mak. That should help.\n> \n> Thanks. Unfortunately not quite there though:\n> \n> print.c(1038) : warning C4013: 'pclose' undefined; assuming extern\n> returning int\n> \n> describe.c\n> tab-complete.c\n> describe.c(1462) : warning C4761: integral size mismatch in argument;\n> conversion supplied\n> mbprint.c\n> link.exe @.\\nmc01556.\n> print.obj : error LNK2001: unresolved external symbol _pclose\n> .\\Release\\psql.exe : fatal error LNK1120: 1 unresolved externals\n> NMAKE : fatal error U1077: 'link.exe' : return code '0x460'\n> Stop.\n> NMAKE : fatal error U1077: '\"c:\\Program Files\\Microsoft Visual\n> Studio\\VC98\\bin\\nmake.exe\"' : return code '0x2'\n> Stop.\n> \n> Any ideas?\n> \n> Regards, Dave.\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\nIndex: src/bin/psql/common.c\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/bin/psql/common.c,v\nretrieving revision 1.50\ndiff -c -c -r1.50 common.c\n*** src/bin/psql/common.c\t24 Oct 2002 01:33:50 -0000\t1.50\n--- src/bin/psql/common.c\t29 Oct 2002 19:15:38 -0000\n***************\n*** 6,12 ****\n * $Header: /cvsroot/pgsql-server/src/bin/psql/common.c,v 1.50 2002/10/24 01:33:50 momjian Exp $\n */\n #include \"postgres_fe.h\"\n- \n #include \"common.h\"\n \n #include <errno.h>\n--- 6,11 ----\n***************\n*** 27,35 ****\n \n #ifndef WIN32\n #include <sys/ioctl.h>\t\t\t/* for ioctl() */\n- #else\n- #define popen(x,y) _popen(x,y)\n- #define pclose(x) _pclose(x)\n #endif\n \n #ifdef HAVE_TERMIOS_H\n--- 26,31 ----\nIndex: src/bin/psql/common.h\n===================================================================\nRCS file: /cvsroot/pgsql-server/src/bin/psql/common.h,v\nretrieving revision 1.20\ndiff -c -c -r1.20 common.h\n*** src/bin/psql/common.h\t23 Oct 2002 19:23:56 -0000\t1.20\n--- src/bin/psql/common.h\t29 Oct 2002 19:15:38 -0000\n***************\n*** 42,45 ****\n--- 42,51 ----\n /* sprompt.h */\n extern char *simple_prompt(const char *prompt, int maxlen, bool echo);\n \n+ /* Used for all Win32 popen/pclose calls */\n+ #ifdef WIN32\n+ #define popen(x,y) _popen(x,y)\n+ #define pclose(x) _pclose(x)\n+ #endif\n+ \n #endif /* COMMON_H */",
"msg_date": "Tue, 29 Oct 2002 14:33:58 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Request for supported platforms"
}
] |
[
{
"msg_contents": "Hi,\n\nI am not subscribed (and I would prefer to continue read\nthe list through web interface only).\n\nDoes the developers team follow the\nhttp://developer.postgresql.org/regress/report.php ?\nMay be it's worthless to report there?\n\nBest regards,\nVladimir Chukharev\n",
"msg_date": "Tue, 29 Oct 2002 10:46:02 +0200",
"msg_from": "Vladimir Chukharev <chu@kempc17.me.tut.fi>",
"msg_from_op": true,
"msg_subject": "Is regress/report.php in use?"
}
] |
[
{
"msg_contents": "Hi,\n\nCan anybody explain the following results when using EXPLAIN, \none time with enable_seqscan=on and one time with enable_seqscan=off.\nWhat I don't understand is that the nodes created are the same \n(index scan, seq scan), but the costs differ. \n\nset enable_seqscan=on;\nexplain (SELECT alias96.ID FROM JOB AS alias96 WHERE NOT EXISTS\n((SELECT alias97.DEPENDENT_ID FROM JOB_MISSING_DEPENDENCIES AS alias97\nWHERE alias97.DEPENDENT_ID = alias96.ID)) AND NOT EXISTS ((SELECT\nalias98.JOB_ID FROM ACTIVE_JOBS AS alias98 WHERE alias98.JOB_ID =\nalias96.ID)) AND ((alias96.STATUS IN (-4, -1))) FOR UPDATE OF alias96\nOFFSET 10 LIMIT 10);\nNOTICE: QUERY PLAN:\n\nLimit (cost=458.18..916.35 rows=10 width=10)\n -> Seq Scan on job alias96 (cost=0.00..2185013.05 rows=47689\nwidth=10)\n SubPlan\n -> Index Scan using job_mis6970_dependent_id_idx on\njob_missing_dependencies alias97 (cost=0.00..893.88 rows=288 width=4)\n -> Index Scan using active_jobs_job_id_key on active_jobs\nalias98 (cost=0.00..4.48 rows=1 width=4)\n\nset enable_seqscan=off;\nexplain (SELECT alias96.ID FROM JOB AS alias96 WHERE NOT EXISTS\n((SELECT alias97.DEPENDENT_ID FROM JOB_MISSING_DEPENDENCIES AS alias97\nWHERE alias97.DEPENDENT_ID = alias96.ID)) AND NOT EXISTS ((SELECT\nalias98.JOB_ID FROM ACTIVE_JOBS AS alias98 WHERE alias98.JOB_ID =\nalias96.ID)) AND ((alias96.STATUS IN (-4, -1))) FOR UPDATE OF alias96\nOFFSET 10 LIMIT 10);\nNOTICE: QUERY PLAN:\n\nLimit (cost=100000458.18..100000916.35 rows=10 width=10)\n -> Seq Scan on job alias96 (cost=100000000.00..102185013.05\nrows=47689 width=10)\n SubPlan\n -> Index Scan using job_mis6970_dependent_id_idx on\njob_missing_dependencies alias97 (cost=0.00..893.88 rows=288 width=4)\n -> Index Scan using active_jobs_job_id_key on active_jobs\nalias98 (cost=0.00..4.48 rows=1 width=4)\n\n\nIves\n",
"msg_date": "29 Oct 2002 02:11:14 -0800",
"msg_from": "ives.landrieu@ua.ac.be (Ives Landrieu)",
"msg_from_op": true,
"msg_subject": "Strange cost computation?"
},
{
"msg_contents": "On 29 Oct 2002, Ives Landrieu wrote:\n\n> Hi,\n>\n> Can anybody explain the following results when using EXPLAIN,\n> one time with enable_seqscan=on and one time with enable_seqscan=off.\n> What I don't understand is that the nodes created are the same\n> (index scan, seq scan), but the costs differ.\n\nEnable seq scan actually just sets a large cost disbenefit to seq scans.\nIs alias96.STATUS indexed?\n\n",
"msg_date": "Wed, 30 Oct 2002 07:59:56 -0800 (PST)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: Strange cost computation?"
},
{
"msg_contents": "On Wed, 30 Oct 2002, Stephan Szabo wrote:\n\n> On 29 Oct 2002, Ives Landrieu wrote:\n>\n> > Hi,\n> >\n> > Can anybody explain the following results when using EXPLAIN,\n> > one time with enable_seqscan=on and one time with enable_seqscan=off.\n> > What I don't understand is that the nodes created are the same\n> > (index scan, seq scan), but the costs differ.\n>\n> Enable seq scan actually just sets a large cost disbenefit to seq scans.\n\nBoy, that sentence sure didn't make sense. enable_seqscan=off actually\ngives a large cost disbenefit to seqscans.\n\n",
"msg_date": "Wed, 30 Oct 2002 08:12:12 -0800 (PST)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: Strange cost computation?"
}
] |
[
{
"msg_contents": "Hi,\n\nDoes the developers team follow the\nhttp://developer.postgresql.org/regress/report.php ?\nMay be it's worthless to report there?\n\nBest regards,\nVladimir Chukharev\n",
"msg_date": "Tue, 29 Oct 2002 12:32:01 +0200",
"msg_from": "Vladimir Chukharev <chu@kempc17.me.tut.fi>",
"msg_from_op": true,
"msg_subject": "Is regress/report.php in use?"
},
{
"msg_contents": "\nGee. I didn't know that existed. We are probably better off doing it\nvia mailing list so we can discuss the results.\n\n---------------------------------------------------------------------------\n\nVladimir Chukharev wrote:\n> Hi,\n> \n> Does the developers team follow the\n> http://developer.postgresql.org/regress/report.php ?\n> May be it's worthless to report there?\n> \n> Best regards,\n> Vladimir Chukharev\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": "Tue, 29 Oct 2002 13:10:45 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Is regress/report.php in use?"
},
{
"msg_contents": "On Tue, 29 Oct 2002, Bruce Momjian wrote:\n\n>\n> Gee. I didn't know that existed. We are probably better off doing it\n> via mailing list so we can discuss the results.\n\nWhat do you mean you didn't know it existed? It's been there for the\nlast few releases (since 7.1). You've even submitted to it!\n\nVince.\n-- \n http://www.meanstreamradio.com http://www.unknown-artists.com\n Internet radio: It's not file sharing, it's just radio.\n\n",
"msg_date": "Tue, 29 Oct 2002 14:36:14 -0500 (EST)",
"msg_from": "Vince Vielhaber <vev@michvhf.com>",
"msg_from_op": false,
"msg_subject": "Re: Is regress/report.php in use?"
},
{
"msg_contents": "Vince Vielhaber wrote:\n> On Tue, 29 Oct 2002, Bruce Momjian wrote:\n> \n> >\n> > Gee. I didn't know that existed. We are probably better off doing it\n> > via mailing list so we can discuss the results.\n> \n> What do you mean you didn't know it existed? It's been there for the\n> last few releases (since 7.1). You've even submitted to it!\n\nI didn't _remember_ it existed.\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, 29 Oct 2002 14:37:08 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: Is regress/report.php in use?"
},
{
"msg_contents": "On Tue, 29 Oct 2002, Bruce Momjian wrote:\n\n> Vince Vielhaber wrote:\n> > On Tue, 29 Oct 2002, Bruce Momjian wrote:\n> >\n> > >\n> > > Gee. I didn't know that existed. We are probably better off doing it\n> > > via mailing list so we can discuss the results.\n> >\n> > What do you mean you didn't know it existed? It's been there for the\n> > last few releases (since 7.1). You've even submitted to it!\n>\n> I didn't _remember_ it existed.\n\nMuch better. Now you should use it. :) It keeps an archive of the\ntest results, etc.\n\nVince.\n-- \n http://www.meanstreamradio.com http://www.unknown-artists.com\n Internet radio: It's not file sharing, it's just radio.\n\n",
"msg_date": "Tue, 29 Oct 2002 14:55:22 -0500 (EST)",
"msg_from": "Vince Vielhaber <vev@michvhf.com>",
"msg_from_op": false,
"msg_subject": "Re: Is regress/report.php in use?"
},
{
"msg_contents": "\nto be honest ... I forgot it was there too :(\n\nOn Tue, 29 Oct 2002, Vince Vielhaber wrote:\n\n> On Tue, 29 Oct 2002, Bruce Momjian wrote:\n>\n> >\n> > Gee. I didn't know that existed. We are probably better off doing it\n> > via mailing list so we can discuss the results.\n>\n> What do you mean you didn't know it existed? It's been there for the\n> last few releases (since 7.1). You've even submitted to it!\n>\n> Vince.\n> --\n> http://www.meanstreamradio.com http://www.unknown-artists.com\n> Internet radio: It's not file sharing, it's just radio.\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": "Tue, 29 Oct 2002 16:12:56 -0400 (AST)",
"msg_from": "\"Marc G. Fournier\" <scrappy@hub.org>",
"msg_from_op": false,
"msg_subject": "Re: Is regress/report.php in use?"
},
{
"msg_contents": "On Tue, 29 Oct 2002, Marc G. Fournier wrote:\n\n>\n> to be honest ... I forgot it was there too :(\n\nJust means I'm gonna have to do a bunch of popup ads!\n\n** ducking and running\n\n>\n> On Tue, 29 Oct 2002, Vince Vielhaber wrote:\n>\n> > On Tue, 29 Oct 2002, Bruce Momjian wrote:\n> >\n> > >\n> > > Gee. I didn't know that existed. We are probably better off doing it\n> > > via mailing list so we can discuss the results.\n> >\n> > What do you mean you didn't know it existed? It's been there for the\n> > last few releases (since 7.1). You've even submitted to it!\n> >\n> > Vince.\n> > --\n> > http://www.meanstreamradio.com http://www.unknown-artists.com\n> > Internet radio: It's not file sharing, it's just radio.\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\nVince.\n-- \n http://www.meanstreamradio.com http://www.unknown-artists.com\n Internet radio: It's not file sharing, it's just radio.\n\n",
"msg_date": "Tue, 29 Oct 2002 15:54:43 -0500 (EST)",
"msg_from": "Vince Vielhaber <vev@michvhf.com>",
"msg_from_op": false,
"msg_subject": "Re: Is regress/report.php in use?"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> Gee. I didn't know that existed. We are probably better off doing it\n> via mailing list so we can discuss the results.\n\nThe similar systems I know usually post a copy to a corresponding\nlist automatically. Example - FreeBSD send-pr online.\n\nV.Chukharev\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, 30 Oct 2002 08:50:12 +0200",
"msg_from": "Vladimir Chukharev <chu@kempc17.me.tut.fi>",
"msg_from_op": true,
"msg_subject": "Re: Is regress/report.php in use?"
}
] |
[
{
"msg_contents": "With Bruce's resultmap change, and now that we are beyond the DST stuff:\n\n======================\n All 89 tests passed. \n======================\nlerlaptop# uname -a\nFreeBSD lerlaptop.lerctr.org 4.7-STABLE FreeBSD 4.7-STABLE #13: Fri Oct\n25 01:32:16 CDT 2002 \nler@lerlaptop.lerctr.org:/usr/obj/usr/src/sys/LERLAPTOP i386\nlerlaptop# \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 Oct 2002 04:44:14 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "good news: FreeBSD 4.7: all passed (with Bruce's resultmap change)."
},
{
"msg_contents": "\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nLarry Rosenman wrote:\n> With Bruce's resultmap change, and now that we are beyond the DST stuff:\n> \n> ======================\n> All 89 tests passed. \n> ======================\n> lerlaptop# uname -a\n> FreeBSD lerlaptop.lerctr.org 4.7-STABLE FreeBSD 4.7-STABLE #13: Fri Oct\n> 25 01:32:16 CDT 2002 \n> ler@lerlaptop.lerctr.org:/usr/obj/usr/src/sys/LERLAPTOP i386\n> lerlaptop# \n> \n> -- \n> Larry Rosenman http://www.lerctr.org/~ler\n> Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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, 29 Oct 2002 13:13:05 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: good news: FreeBSD 4.7: all passed (with Bruce's resultmap"
},
{
"msg_contents": "On Tue, 2002-10-29 at 12:13, Bruce Momjian wrote:\n> \n> Ports list updated:\n> \n> http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\nexcept for the version of PG :-)\n\n\n> \n> ---------------------------------------------------------------------------\n> Larry Rosenman wrote:\n> > With Bruce's resultmap change, and now that we are beyond the DST stuff:\n> > \n> > ======================\n> > All 89 tests passed. \n> > ======================\n> > lerlaptop# uname -a\n> > FreeBSD lerlaptop.lerctr.org 4.7-STABLE FreeBSD 4.7-STABLE #13: Fri Oct\n> > 25 01:32:16 CDT 2002 \n> > ler@lerlaptop.lerctr.org:/usr/obj/usr/src/sys/LERLAPTOP i386\n> > lerlaptop# \n> > \n> > -- \n> > Larry Rosenman http://www.lerctr.org/~ler\n> > Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> > US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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-- \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 Oct 2002 12:27:42 -0600",
"msg_from": "Larry Rosenman <ler@lerctr.org>",
"msg_from_op": true,
"msg_subject": "Re: good news: FreeBSD 4.7: all passed (with Bruce's"
},
{
"msg_contents": "\nOops, I forgot in a few places. Fortunately, I always updated the date\nso I have fixed them all. We have a pretty good 7.3 list already.\n\n---------------------------------------------------------------------------\n\nLarry Rosenman wrote:\n> On Tue, 2002-10-29 at 12:13, Bruce Momjian wrote:\n> > \n> > Ports list updated:\n> > \n> > http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n> except for the version of PG :-)\n> \n> \n> > \n> > ---------------------------------------------------------------------------\n> > Larry Rosenman wrote:\n> > > With Bruce's resultmap change, and now that we are beyond the DST stuff:\n> > > \n> > > ======================\n> > > All 89 tests passed. \n> > > ======================\n> > > lerlaptop# uname -a\n> > > FreeBSD lerlaptop.lerctr.org 4.7-STABLE FreeBSD 4.7-STABLE #13: Fri Oct\n> > > 25 01:32:16 CDT 2002 \n> > > ler@lerlaptop.lerctr.org:/usr/obj/usr/src/sys/LERLAPTOP i386\n> > > lerlaptop# \n> > > \n> > > -- \n> > > Larry Rosenman http://www.lerctr.org/~ler\n> > > Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> > > US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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> -- \n> Larry Rosenman http://www.lerctr.org/~ler\n> Phone: +1 972-414-9812 E-Mail: ler@lerctr.org\n> US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749\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, 29 Oct 2002 13:59:57 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: good news: FreeBSD 4.7: all passed (with Bruce's"
}
] |
[
{
"msg_contents": "Hi All,\n\nI wrote a Matlab interface to PostgreSQL based on libpq. It is working \nfine, supports multiple connections, supports all data types and arrays. \nIt is a C program to do the interface and some Matlab wrapper functions \naround it to do the job at application level.\n\nMatlab has an ODBC toolbox but we dont want to buy it since the \ninterface does the job and we have been using PostgreSQL for long time.\nWe want to use PostgreSQl to store numeric data from simulation, \ncomputation and data acquisition sources. Basically a big bunch of float \nnumbers.\n\nThere is still one problem, regarding float output formatting in querys \nand dumps, which essential for this type of application.\n\nIf I have a float8 field (named real8 below) in a table and insert the \nfollowing,\n\ninsert into test(real8) values (4503599627370496);\nINSERT 21192 1\n\nA select produces,\n\nselect real8 from test;\n real8\n---------------------\n 4.5035996273705e+15\n(1 row)\n\nThis is the string I would get from libpq's PQgetvalue(). PQftype() \ncorrectly returns float8. pg_dump produces the same result. If I convert \nPQgetvalue() to a C double I wont get the same number I inserted. If I \ndo a restore from a dump I also wont have the same number. I can see \nthat the original number is correctly stored because,\n\nselect to_char(real8,'9999999999999999999.99999') from test;\n to_char\n----------------------\n 4503599627370496\n(1 row)\n\nThis way PQftype wont tell this is a float8.\n\nIs there a way to set query output precision to maximum precision ?\nFor the type of application I mentioned this is crucial. People want to \nget the 'same' numbers, from querys or dumps, as they inserted them.\n\nMatlab has a toolbox fucntion, claiming maximum precision, to convert \nfrom its double type (PostgreSQL float8) to string which does a \nsprintf('%25.18g',number).\n\nWould it be possible to have a a parameter which could be changed by a \nSET command, in order to control output precision ? I searched the docs \nbut could not find this.\n\n-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Tue, 29 Oct 2002 11:34:58 +0000",
"msg_from": "Pedro Miguel Frazao Fernandes Ferreira <pfrazao@ualg.pt>",
"msg_from_op": true,
"msg_subject": "float output precision questions"
},
{
"msg_contents": "Sorry. I forgot to thank for any help from all of you in the previous \nmessage. Thanks! :)\n\nJust one more thing:\n\nI now I can go to the source and change the code which converts floats \nto strings, and have my problem solved. But this wont be general. Others \nmight need this kind of application.\n\nI could post this interface to postgres interfaces site. Do you think \nthis is ok ?\n\nThanks again !\n\nPedro M. Ferreira\n\nPedro Miguel Frazao Fernandes Ferreira wrote:\n> Hi All,\n> \n> I wrote a Matlab interface to PostgreSQL based on libpq. It is working \n> fine, supports multiple connections, supports all data types and arrays. \n> It is a C program to do the interface and some Matlab wrapper functions \n> around it to do the job at application level.\n> \n> Matlab has an ODBC toolbox but we dont want to buy it since the \n> interface does the job and we have been using PostgreSQL for long time.\n> We want to use PostgreSQl to store numeric data from simulation, \n> computation and data acquisition sources. Basically a big bunch of float \n> numbers.\n> \n> There is still one problem, regarding float output formatting in querys \n> and dumps, which essential for this type of application.\n> \n> If I have a float8 field (named real8 below) in a table and insert the \n> following,\n> \n> insert into test(real8) values (4503599627370496);\n> INSERT 21192 1\n> \n> A select produces,\n> \n> select real8 from test;\n> real8\n> ---------------------\n> 4.5035996273705e+15\n> (1 row)\n> \n> This is the string I would get from libpq's PQgetvalue(). PQftype() \n> correctly returns float8. pg_dump produces the same result. If I convert \n> PQgetvalue() to a C double I wont get the same number I inserted. If I \n> do a restore from a dump I also wont have the same number. I can see \n> that the original number is correctly stored because,\n> \n> select to_char(real8,'9999999999999999999.99999') from test;\n> to_char\n> ----------------------\n> 4503599627370496\n> (1 row)\n> \n> This way PQftype wont tell this is a float8.\n> \n> Is there a way to set query output precision to maximum precision ?\n> For the type of application I mentioned this is crucial. People want to \n> get the 'same' numbers, from querys or dumps, as they inserted them.\n> \n> Matlab has a toolbox fucntion, claiming maximum precision, to convert \n> from its double type (PostgreSQL float8) to string which does a \n> sprintf('%25.18g',number).\n> \n> Would it be possible to have a a parameter which could be changed by a \n> SET command, in order to control output precision ? I searched the docs \n> but could not find this.\n> \n\n\n-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Tue, 29 Oct 2002 11:39:18 +0000",
"msg_from": "Pedro Miguel Frazao Fernandes Ferreira <pfrazao@ualg.pt>",
"msg_from_op": true,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "\nTODO has:\n\n o Add SET REAL_FORMAT and SET DOUBLE_PRECISION_FORMAT\n\t using printf args\n\nso we have not implemented it yet.\n\n---------------------------------------------------------------------------\n\nPedro Miguel Frazao Fernandes Ferreira wrote:\n> Hi All,\n> \n> I wrote a Matlab interface to PostgreSQL based on libpq. It is working \n> fine, supports multiple connections, supports all data types and arrays. \n> It is a C program to do the interface and some Matlab wrapper functions \n> around it to do the job at application level.\n> \n> Matlab has an ODBC toolbox but we dont want to buy it since the \n> interface does the job and we have been using PostgreSQL for long time.\n> We want to use PostgreSQl to store numeric data from simulation, \n> computation and data acquisition sources. Basically a big bunch of float \n> numbers.\n> \n> There is still one problem, regarding float output formatting in querys \n> and dumps, which essential for this type of application.\n> \n> If I have a float8 field (named real8 below) in a table and insert the \n> following,\n> \n> insert into test(real8) values (4503599627370496);\n> INSERT 21192 1\n> \n> A select produces,\n> \n> select real8 from test;\n> real8\n> ---------------------\n> 4.5035996273705e+15\n> (1 row)\n> \n> This is the string I would get from libpq's PQgetvalue(). PQftype() \n> correctly returns float8. pg_dump produces the same result. If I convert \n> PQgetvalue() to a C double I wont get the same number I inserted. If I \n> do a restore from a dump I also wont have the same number. I can see \n> that the original number is correctly stored because,\n> \n> select to_char(real8,'9999999999999999999.99999') from test;\n> to_char\n> ----------------------\n> 4503599627370496\n> (1 row)\n> \n> This way PQftype wont tell this is a float8.\n> \n> Is there a way to set query output precision to maximum precision ?\n> For the type of application I mentioned this is crucial. People want to \n> get the 'same' numbers, from querys or dumps, as they inserted them.\n> \n> Matlab has a toolbox fucntion, claiming maximum precision, to convert \n> from its double type (PostgreSQL float8) to string which does a \n> sprintf('%25.18g',number).\n> \n> Would it be possible to have a a parameter which could be changed by a \n> SET command, in order to control output precision ? I searched the docs \n> but could not find this.\n> \n> -- \n> ----------------------------------------------------------------------\n> Pedro Miguel Frazao Fernandes Ferreira\n> Universidade do Algarve\n> Faculdade de Ciencias e Tecnologia\n> Campus de Gambelas\n> 8000-117 Faro\n> Portugal\n> Tel./Fax: (+351) 289 800950 / 289 819403\n> http://w3.ualg.pt/~pfrazao\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": "Tue, 29 Oct 2002 13:14:51 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Pedro Miguel Frazao Fernandes Ferreira writes:\n\n> Is there a way to set query output precision to maximum precision ?\n> For the type of application I mentioned this is crucial. People want to\n> get the 'same' numbers, from querys or dumps, as they inserted them.\n\nThere isn't a way right now, but it's planned to be able to dump\nfloating-point numbers in some binary form (like printf(\"%A\")) to be able\nto restore them exactly. Not sure how that would satisfy the needs of\nclient interfaces, though.\n\n> Matlab has a toolbox fucntion, claiming maximum precision, to convert\n> from its double type (PostgreSQL float8) to string which does a\n> sprintf('%25.18g',number).\n\nDo we have some mathematical guarantee that this is sufficient and\nnecessary? If so, then it might do.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Tue, 29 Oct 2002 23:19:05 +0100 (CET)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Bruce Momjian <pgman@candle.pha.pa.us> writes:\n> TODO has:\n> o Add SET REAL_FORMAT and SET DOUBLE_PRECISION_FORMAT\n> \t using printf args\n> so we have not implemented it yet.\n\nIIRC, the last time it was discussed there was disagreement about how\nit should work; check the pghackers archives for details. The feature\nprobably won't go anywhere until those issues are resolved.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 29 Oct 2002 17:21:27 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions "
},
{
"msg_contents": "On Tue, 29 Oct 2002, Peter Eisentraut wrote:\n\n> Pedro Miguel Frazao Fernandes Ferreira writes:\n>\n> > Is there a way to set query output precision to maximum precision ?\n> > For the type of application I mentioned this is crucial. People want to\n> > get the 'same' numbers, from querys or dumps, as they inserted them.\n\nHow do you define maximum precision and \"same\"? With simple test programs\nin C, using two digits more than DBL_DIG for printf specifier, it's easy\nfor me to find numbers that \"change\" decimal string representation in the\ndecimal representation -> double -> decimal representation conversion(*).\nThe final double you get from the second conversion should be the same as\nthe first, but is that what you need or do you need a stronger guarantee\nthan that?\n\n\n",
"msg_date": "Tue, 29 Oct 2002 16:22:31 -0800 (PST)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "On Tue, Oct 29, 2002 at 23:19:05 +0100,\n Peter Eisentraut <peter_e@gmx.net> wrote:\n> \n> There isn't a way right now, but it's planned to be able to dump\n> floating-point numbers in some binary form (like printf(\"%A\")) to be able\n> to restore them exactly. Not sure how that would satisfy the needs of\n> client interfaces, though.\n\nWhy not print it as a floating binary number instead of a floating decimal\nnumber? I would think that would give you better portability than a system\nspecific binary representation.\n",
"msg_date": "Tue, 29 Oct 2002 20:15:06 -0600",
"msg_from": "Bruno Wolff III <bruno@wolff.to>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "\n\nPeter Eisentraut wrote:\n> \n>>Matlab has a toolbox fucntion, claiming maximum precision, to convert\n>>from its double type (PostgreSQL float8) to string which does a\n>>sprintf('%25.18g',number).\n> \n> \n> Do we have some mathematical guarantee that this is sufficient and\n> necessary? If so, then it might do.\n\nIt is necessary if you want to do this type of (huge amount of number \nstorage) application:\n\n [Some client] (insert) [PostgreSQL] (query) [Some client]\n(double number a)-------->(float8 number)------->(double number b)\n\nIn order for a=b this is necessary. With current float8 output this is \nnot allways true.\n\nHere is the help for that particular Matlab function:\n\n NUM2MSTR Convert number to string in maximum precision.\n S = NUM2MSTR(N) converts real numbers of input\n matrix N to string output vector S, in\n maximum precision.\n\n See also NUM2STR.\n\nIf you want I can try to contact the guys who coded this function to know\nif this is sufficient.\n\nThanks,\nPedro M. Ferreira\n-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Wed, 30 Oct 2002 10:22:01 +0000",
"msg_from": "Pedro Miguel Frazao Fernandes Ferreira <pfrazao@ualg.pt>",
"msg_from_op": true,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Stephan Szabo wrote:\n> On Tue, 29 Oct 2002, Peter Eisentraut wrote:\n> \n> \n>>Pedro Miguel Frazao Fernandes Ferreira writes:\n>>\n>>\n>>>Is there a way to set query output precision to maximum precision ?\n>>>For the type of application I mentioned this is crucial. People want to\n>>>get the 'same' numbers, from querys or dumps, as they inserted them.\n>>\n> \n> How do you define maximum precision and \"same\"? With simple test programs\n> in C, using two digits more than DBL_DIG for printf specifier, it's easy\n> for me to find numbers that \"change\" decimal string representation in the\n> decimal representation -> double -> decimal representation conversion(*).\n> The final double you get from the second conversion should be the same as\n> the first, but is that what you need or do you need a stronger guarantee\n> than that?\n\nWhen I say \"same\" I am talking about having a number 'stored' in double \nformat in some client, inserting it in PostgreSQL float8 field and get \nit to the client as it was before:\n\n [Some client] (insert) [PostgreSQL] (query) [Some client]\n(double number a)-------->(float8 number)------->(double number b)\n\n\"same\" is so that a==b is true.\nWith current float8 output this is not allways true.\n\nI believe this should allways be true for numbers which are originally \nstored in double format.\n\nThanks,\nPedro M. Ferreira\n\n> \n> \n> \n> \n\n\n-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Wed, 30 Oct 2002 10:28:42 +0000",
"msg_from": "Pedro Miguel Frazao Fernandes Ferreira <pfrazao@ualg.pt>",
"msg_from_op": true,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Bruno Wolff III wrote:\n> On Tue, Oct 29, 2002 at 23:19:05 +0100,\n> Peter Eisentraut <peter_e@gmx.net> wrote:\n> \n>>There isn't a way right now, but it's planned to be able to dump\n>>floating-point numbers in some binary form (like printf(\"%A\")) to be able\n>>to restore them exactly. Not sure how that would satisfy the needs of\n>>client interfaces, though.\n> \n> \n> Why not print it as a floating binary number instead of a floating decimal\n> number? I would think that would give you better portability than a system\n> specific binary representation.\n\nHaving a way to get the binary storage representation for float numbers \nwould be good and would suffice within the same float number standard.\n\nThanks,\nPedro M. Ferreira\n\n-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Wed, 30 Oct 2002 10:35:09 +0000",
"msg_from": "Pedro Miguel Frazao Fernandes Ferreira <pfrazao@ualg.pt>",
"msg_from_op": true,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "\nOn Wed, 30 Oct 2002, Pedro Miguel Frazao Fernandes Ferreira wrote:\n\n> Stephan Szabo wrote:\n> > On Tue, 29 Oct 2002, Peter Eisentraut wrote:\n> >\n> >\n> >>Pedro Miguel Frazao Fernandes Ferreira writes:\n> >>\n> >>\n> >>>Is there a way to set query output precision to maximum precision ?\n> >>>For the type of application I mentioned this is crucial. People want to\n> >>>get the 'same' numbers, from querys or dumps, as they inserted them.\n> >>\n> >\n> > How do you define maximum precision and \"same\"? With simple test programs\n> > in C, using two digits more than DBL_DIG for printf specifier, it's easy\n> > for me to find numbers that \"change\" decimal string representation in the\n> > decimal representation -> double -> decimal representation conversion(*).\n> > The final double you get from the second conversion should be the same as\n> > the first, but is that what you need or do you need a stronger guarantee\n> > than that?\n>\n> When I say \"same\" I am talking about having a number 'stored' in double\n> format in some client, inserting it in PostgreSQL float8 field and get\n> it to the client as it was before:\n>\n> [Some client] (insert) [PostgreSQL] (query) [Some client]\n> (double number a)-------->(float8 number)------->(double number b)\n>\n> \"same\" is so that a==b is true.\n> With current float8 output this is not allways true.\n>\n> I believe this should allways be true for numbers which are originally\n> stored in double format.\n\nThe problem is that there are two competing needs here. One is the above,\nthe other other is that you get something that has the same decimal\nrepresentation (within the float's ability to store the number). Right now\nthe system does the latter since for most people, that seems to be the\nguarantee they want.\n\nThis would probably make sense as an option, so why don't you look at the\npast discussions and see if you can come up with a solution that keeps\neveryone happy (and preferably implement it, but...) :)\n\n\n\n",
"msg_date": "Wed, 30 Oct 2002 07:53:22 -0800 (PST)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Stephan Szabo wrote:\n> On Wed, 30 Oct 2002, Pedro Miguel Frazao Fernandes Ferreira wrote:\n>>Stephan Szabo wrote:\n>>>On Tue, 29 Oct 2002, Peter Eisentraut wrote:\n>>>>Pedro Miguel Frazao Fernandes Ferreira writes:\n>>>>>Is there a way to set query output precision to maximum precision ?\n>>>>>For the type of application I mentioned this is crucial. People want to\n>>>>>get the 'same' numbers, from querys or dumps, as they inserted them.\n>>>>\n>>>How do you define maximum precision and \"same\"? With simple test programs\n>>>in C, using two digits more than DBL_DIG for printf specifier, it's easy\n>>>for me to find numbers that \"change\" decimal string representation in the\n>>>decimal representation -> double -> decimal representation conversion(*).\n>>>The final double you get from the second conversion should be the same as\n>>>the first, but is that what you need or do you need a stronger guarantee\n>>>than that?\n>>\n>>When I say \"same\" I am talking about having a number 'stored' in double\n>>format in some client, inserting it in PostgreSQL float8 field and get\n>>it to the client as it was before:\n>>\n>> [Some client] (insert) [PostgreSQL] (query) [Some client]\n>>(double number a)-------->(float8 number)------->(double number b)\n>>\n>>\"same\" is so that a==b is true.\n>>With current float8 output this is not allways true.\n>>\n>>I believe this should allways be true for numbers which are originally\n>>stored in double format.\n> \n> The problem is that there are two competing needs here. One is the above,\n> the other other is that you get something that has the same decimal\n> representation (within the float's ability to store the number). Right now\n> the system does the latter since for most people, that seems to be the\n> guarantee they want.\n\nLook at this example:\n\n1.79769313486231571e+308 is the largest floating point number \nrepresentable by a C double in x86.\n\nIn C this is possible:\n\n#include <stdio.h>\n#include <stdlib.h>\n\nint main(void)\n{\n double v;\n char a[30];\n\n v=1.79769313486231571e+308;\n\n printf(\" Stored double number: %25.18g\\n\",v);\n sprintf(a,\"%25.18g\",v);\n printf(\" Converted to string: %s\\n\",a);\n v=atof(a);\n printf(\"Converted from string to double: %25.18g\\n\",v);\n}\n\nUsing standard PostgreSQL query output it would not be possible to get \nthis number, which has representation as a double.\n\nI fetched the PostgreSQL source from Debian, changed \nsrc/backend/utils/adt/float.c to do sprintf(ascii, \"%25.18g\", num) \ninstead of sprintf(ascii, \"%.*g\", DBL_DIG, num), compiled and installed. \nNow I can get the number as is.\n\nI understand that if people insert a value of 1.1 in a double, they want \nto get 1.1 without knowing that in fact the stored number is \n1.10000000000000009. But do you understand that if some people insert, \nfor example, a value of 1.79769313486231571e+308 they dont want to get \n1.79769313486232e+308 which does not compare equal (in Matlab or C) to \nthe first ? This is a bug.\n\n> This would probably make sense as an option, so why don't you look at the\n> past discussions and see if you can come up with a solution that keeps\n> everyone happy (and preferably implement it, but...) :)\n\nbut ???\n\nI have a sugestion:\n\nTo have parameters, say DOUBLE_FORMAT and FLOAT_FORMAT, which could have \n option values of 'SHORT' and 'LONG'.\nOption 'SHORT' would be default and produce the standard sprintf(ascii,...\nOption 'LONG' would produce sprintf(ascii, \"%25.18g\", num).\n\nOther way would be to have number parameters to be used in the sprintf \ncalls, in place of 25 and 18, in the format string.\n\n> \n> \n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 4: Don't 'kill -9' the postmaster\n> \n> \n\n\n-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Wed, 30 Oct 2002 17:01:19 +0000",
"msg_from": "Pedro Miguel Frazao Fernandes Ferreira <pfrazao@ualg.pt>",
"msg_from_op": true,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Bruno Wolff III writes:\n\n> On Tue, Oct 29, 2002 at 23:19:05 +0100,\n> Peter Eisentraut <peter_e@gmx.net> wrote:\n> >\n> > There isn't a way right now, but it's planned to be able to dump\n> > floating-point numbers in some binary form (like printf(\"%A\")) to be able\n> > to restore them exactly. Not sure how that would satisfy the needs of\n> > client interfaces, though.\n>\n> Why not print it as a floating binary number instead of a floating decimal\n> number? I would think that would give you better portability than a system\n> specific binary representation.\n\nThe printf(\"%A\") output is not system-specific.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Wed, 30 Oct 2002 19:27:57 +0100 (CET)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "On Wed, 30 Oct 2002, Pedro Miguel Frazao Fernandes Ferreira wrote:\n\n> In C this is possible:\n>\n> #include <stdio.h>\n> #include <stdlib.h>\n>\n> int main(void)\n> {\n> double v;\n> char a[30];\n>\n> v=1.79769313486231571e+308;\n>\n> printf(\" Stored double number: %25.18g\\n\",v);\n> sprintf(a,\"%25.18g\",v);\n> printf(\" Converted to string: %s\\n\",a);\n> v=atof(a);\n> printf(\"Converted from string to double: %25.18g\\n\",v);\n> }\n\nAFAICT, this is not guaranteed to give you the same representation that\nyou're using in the initializer however.\n\n> Using standard PostgreSQL query output it would not be possible to get\n> this number, which has representation as a double.\n\n> I fetched the PostgreSQL source from Debian, changed\n> src/backend/utils/adt/float.c to do sprintf(ascii, \"%25.18g\", num)\n> instead of sprintf(ascii, \"%.*g\", DBL_DIG, num), compiled and installed.\n> Now I can get the number as is.\n>\n> I understand that if people insert a value of 1.1 in a double, they want\n> to get 1.1 without knowing that in fact the stored number is\n> 1.10000000000000009. But do you understand that if some people insert,\n> for example, a value of 1.79769313486231571e+308 they dont want to get\n> 1.79769313486232e+308 which does not compare equal (in Matlab or C) to\n> the first ? This is a bug.\n\nI disagree to some extent. I'm not sure it's meaningful to expect that\n(what if the database and the client are on different architectures) in\ngeneral. In any case, you're effectively going from decimal\nrepresentation to double to decimal representation (the string you used\nto insert it -> internal representation -> string used to output it) and\nthat's only guaranteed to be correct up to DBL_DIG digits as far as I can\ntell. I think it'd be nice to have an option to get more digits for those\nsorts of applications, however.\n\n> > This would probably make sense as an option, so why don't you look at the\n> > past discussions and see if you can come up with a solution that keeps\n> > everyone happy (and preferably implement it, but...) :)\n>\n> but ???\n\n, but I realize that you might not be interested in doing such. (I figured\nthe last part was implied)\n\n> I have a sugestion:\n>\n> To have parameters, say DOUBLE_FORMAT and FLOAT_FORMAT, which could have\n> option values of 'SHORT' and 'LONG'.\n> Option 'SHORT' would be default and produce the standard sprintf(ascii,...\n> Option 'LONG' would produce sprintf(ascii, \"%25.18g\", num).\n>\n> Other way would be to have number parameters to be used in the sprintf\n> calls, in place of 25 and 18, in the format string.\n\n From what Tom said, something similar was suggested and there were issues\nbrought up. I don't know what they were, since I wasn't personally\nterribly interested, but it should be in the archives. If there were any\nconcerns, you'll probably need to deal with those as well.\n\n",
"msg_date": "Wed, 30 Oct 2002 10:49:38 -0800 (PST)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "On Wed, Oct 30, 2002 at 19:27:57 +0100,\n Peter Eisentraut <peter_e@gmx.net> wrote:\n> \n> The printf(\"%A\") output is not system-specific.\n\nJust out of curiosity, can you tell me a web page or keywords to use\nin a search to see what that format does? I tried using google, but\nsearching for printf with and \"A\" format didn't go to well. I only\nfound one correct reference and it didn't explain what %A did.\nThanks.\n",
"msg_date": "Wed, 30 Oct 2002 13:07:01 -0600",
"msg_from": "Bruno Wolff III <bruno@wolff.to>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Stephan Szabo wrote:\n> On Wed, 30 Oct 2002, Pedro Miguel Frazao Fernandes Ferreira wrote:\n>>\n>>I understand that if people insert a value of 1.1 in a double, they want\n>>to get 1.1 without knowing that in fact the stored number is\n>>1.10000000000000009. But do you understand that if some people insert,\n>>for example, a value of 1.79769313486231571e+308 they dont want to get\n>>1.79769313486232e+308 which does not compare equal (in Matlab or C) to\n>>the first ? This is a bug.\n> \n> \n> I disagree to some extent. I'm not sure it's meaningful to expect that\n> (what if the database and the client are on different architectures) in\n> general. In any case, you're effectively going from decimal\n> representation to double to decimal representation (the string you used\n> to insert it -> internal representation -> string used to output it) and\n> that's only guaranteed to be correct up to DBL_DIG digits as far as I can\n> tell. I think it'd be nice to have an option to get more digits for those\n> sorts of applications, however.\n\nIn the previous email example, in C, I was going from decimal to double \nand so on, but this is not the case when I do some simulation. In this \ncase it will allways be from Matlab double to PostgreSQL float8 and from \nlibpq PQgetvalue() string to Matlab double. The example was just a x86 \nnumber example where (got the string from Matlab double) query output \nwould fail.\n\n>>>This would probably make sense as an option, so why don't you look at the\n>>>past discussions and see if you can come up with a solution that keeps\n>>>everyone happy (and preferably implement it, but...) :)\n>>\n>>but ???\n> , but I realize that you might not be interested in doing such. (I figured\n> the last part was implied)\n\nok! :)\n\nNo problem. I have seen the GUC thing in the source (guc.c etc...) and it\ndoes not look too dificult. It has lots of examples in the code itself.\nWhat I am saying is that I can do it if pg-people agree on the (some) \nway to do it.\n\n>>To have parameters, say DOUBLE_FORMAT and FLOAT_FORMAT, which could have\n>> option values of 'SHORT' and 'LONG'.\n>>Option 'SHORT' would be default and produce the standard sprintf(ascii,...\n>>Option 'LONG' would produce sprintf(ascii, \"%25.18g\", num).\n>>\n>>Other way would be to have number parameters to be used in the sprintf\n>>calls, in place of 25 and 18, in the format string.\n> \n> \n>>From what Tom said, something similar was suggested and there were issues\n> brought up. I don't know what they were, since I wasn't personally\n> terribly interested, but it should be in the archives. If there were any\n> concerns, you'll probably need to deal with those as well.\n\nI looked at some of these emails and it seemed to me that the problem \nwas that Tom did'nt want a parameter that would force people to know \nabout printf number formatting. I think the first solution above (the \nSHORT and LONG way) is simple, maintains usual output as default and \nenables 'maximum' precision at request.\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-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Wed, 30 Oct 2002 19:29:55 +0000",
"msg_from": "\"Pedro M. Ferreira\" <pfrazao@ualg.pt>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "\nOn Wed, 30 Oct 2002, Pedro M. Ferreira wrote:\n\n> Stephan Szabo wrote:\n> > On Wed, 30 Oct 2002, Pedro Miguel Frazao Fernandes Ferreira wrote:\n> >>\n> >>I understand that if people insert a value of 1.1 in a double, they want\n> >>to get 1.1 without knowing that in fact the stored number is\n> >>1.10000000000000009. But do you understand that if some people insert,\n> >>for example, a value of 1.79769313486231571e+308 they dont want to get\n> >>1.79769313486232e+308 which does not compare equal (in Matlab or C) to\n> >>the first ? This is a bug.\n> >\n> >\n> > I disagree to some extent. I'm not sure it's meaningful to expect that\n> > (what if the database and the client are on different architectures) in\n> > general. In any case, you're effectively going from decimal\n> > representation to double to decimal representation (the string you used\n> > to insert it -> internal representation -> string used to output it) and\n> > that's only guaranteed to be correct up to DBL_DIG digits as far as I can\n> > tell. I think it'd be nice to have an option to get more digits for those\n> > sorts of applications, however.\n>\n> In the previous email example, in C, I was going from decimal to double\n> and so on, but this is not the case when I do some simulation. In this\n> case it will allways be from Matlab double to PostgreSQL float8 and from\n> libpq PQgetvalue() string to Matlab double. The example was just a x86\n> number example where (got the string from Matlab double) query output\n> would fail.\n\nI was just responding to it being a bug. I don't think that expecting\na float->db->float (double->db->double) giving the same double is always\nsafe when you assume that the PostgreSQL machine might be on a system with\ndifferent guarantees about precision. In practice, it's probably not a\nbig deal.\n\n> > , but I realize that you might not be interested in doing such. (I figured\n> > the last part was implied)\n>\n> ok! :)\n>\n> No problem. I have seen the GUC thing in the source (guc.c etc...) and it\n> does not look too dificult. It has lots of examples in the code itself.\n> What I am saying is that I can do it if pg-people agree on the (some)\n> way to do it.\n\nYeah, I didn't think it'd be hard, but sometimes people are unable or\nunwilling to do C code for things.\n\n> I looked at some of these emails and it seemed to me that the problem\n> was that Tom did'nt want a parameter that would force people to know\n> about printf number formatting. I think the first solution above (the\n> SHORT and LONG way) is simple, maintains usual output as default and\n> enables 'maximum' precision at request.\n\nThat seems reasonable then, Tom'll probably give any other objections he\nmight have if he has any.\n\n",
"msg_date": "Wed, 30 Oct 2002 13:26:59 -0800 (PST)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Stephan Szabo <sszabo@megazone23.bigpanda.com> writes:\n> On Wed, 30 Oct 2002, Pedro M. Ferreira wrote:\n>> I looked at some of these emails and it seemed to me that the problem\n>> was that Tom did'nt want a parameter that would force people to know\n>> about printf number formatting. I think the first solution above (the\n>> SHORT and LONG way) is simple, maintains usual output as default and\n>> enables 'maximum' precision at request.\n\n> That seems reasonable then, Tom'll probably give any other objections he\n> might have if he has any.\n\nMy recollection is that other people (perhaps Peter?) were the ones\nobjecting before. However I'd be somewhat unhappy with the proposal\nas given:\n\n>>Option 'SHORT' would be default and produce the standard sprintf(ascii,...\n>>Option 'LONG' would produce sprintf(ascii, \"%25.18g\", num).\n\nsince this seems to me to hardwire inappropriate assumptions about the\nnumber of significant digits in a double. (Yes, I know practically\neveryone uses IEEE floats these days. But it's inappropriate for PG\nto assume that.)\n\nAFAICT the real issue here is that binary float representations will\nhave a fractional decimal digit of precision beyond what DBL_DIG claims.\nI think I could support adding an option that switches between the\ncurrent output format:\n\tsprintf(ascii, \"%.*g\", DBL_DIG, num);\nand:\n\tsprintf(ascii, \"%.*g\", DBL_DIG+1, num);\nand similarly for float4. Given carefully written float I/O routines,\nreading the latter output should reproduce the originally stored value.\n(And if the I/O routines are not carefully written, you probably lose\nanyway.) I don't see a need for allowing more flexibility than that.\n\nComments?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Wed, 30 Oct 2002 22:45:44 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions "
},
{
"msg_contents": "On Wed, 30 Oct 2002, Tom Lane wrote:\n\n> Stephan Szabo <sszabo@megazone23.bigpanda.com> writes:\n> > On Wed, 30 Oct 2002, Pedro M. Ferreira wrote:\n> >> I looked at some of these emails and it seemed to me that the problem\n> >> was that Tom did'nt want a parameter that would force people to know\n> >> about printf number formatting. I think the first solution above (the\n> >> SHORT and LONG way) is simple, maintains usual output as default and\n> >> enables 'maximum' precision at request.\n>\n> > That seems reasonable then, Tom'll probably give any other objections he\n> > might have if he has any.\n>\n> My recollection is that other people (perhaps Peter?) were the ones\n> objecting before. However I'd be somewhat unhappy with the proposal\n> as given:\n>\n> >>Option 'SHORT' would be default and produce the standard sprintf(ascii,...\n> >>Option 'LONG' would produce sprintf(ascii, \"%25.18g\", num).\n>\n> since this seems to me to hardwire inappropriate assumptions about the\n> number of significant digits in a double. (Yes, I know practically\n> everyone uses IEEE floats these days. But it's inappropriate for PG\n> to assume that.)\n\nTrue (which I actually was trying to get at in my messages as well). I'll\nadmit to having not read the precise proposal. It's really pretty outside\nwhat I work with in any case.\n\n> AFAICT the real issue here is that binary float representations will\n> have a fractional decimal digit of precision beyond what DBL_DIG claims.\n> I think I could support adding an option that switches between the\n> current output format:\n> \tsprintf(ascii, \"%.*g\", DBL_DIG, num);\n> and:\n> \tsprintf(ascii, \"%.*g\", DBL_DIG+1, num);\n> and similarly for float4. Given carefully written float I/O routines,\n> reading the latter output should reproduce the originally stored value.\n> (And if the I/O routines are not carefully written, you probably lose\n> anyway.) I don't see a need for allowing more flexibility than that.\n\nWell, on my system, it doesn't look like doing the above sprintfs will\nactually work for all numbers. I did a simple program using an arbitrary\nbig number and the DBL_DIG+1 output when stuck into another double\nactually was a different double value. DBL_DIG+2 worked on my system,\nbut...\n\n",
"msg_date": "Wed, 30 Oct 2002 22:09:40 -0800 (PST)",
"msg_from": "Stephan Szabo <sszabo@megazone23.bigpanda.com>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions "
},
{
"msg_contents": "Tom Lane wrote:\n\n> My recollection is that other people (perhaps Peter?) were the ones\n> objecting before. However I'd be somewhat unhappy with the proposal\n> as given:\n>\n>>>Option 'SHORT' would be default and produce the standard sprintf(ascii,...\n>>>Option 'LONG' would produce sprintf(ascii, \"%25.18g\", num).\n >\n> since this seems to me to hardwire inappropriate assumptions about the\n> number of significant digits in a double. (Yes, I know practically\n> everyone uses IEEE floats these days. But it's inappropriate for PG\n> to assume that.)\n\nI understand this. Unfortunately I only have IEEE compliant stuff.\n\n> AFAICT the real issue here is that binary float representations will\n> have a fractional decimal digit of precision beyond what DBL_DIG claims.\n\nIn fact, for some numbers I have been testing with, the double \nrepresentation can distinguish up to DBL_BIG+2.\n\n> I think I could support adding an option that switches between the\n> current output format:\n> \tsprintf(ascii, \"%.*g\", DBL_DIG, num);\n> and:\n> \tsprintf(ascii, \"%.*g\", DBL_DIG+1, num);\n\nEasy to find numbers with double representation which would need DBL_BIG+2.\n\n> and similarly for float4. Given carefully written float I/O routines,\n> reading the latter output should reproduce the originally stored value.\n\nFor some numbers it does not. Not true as I said above.\n\n> (And if the I/O routines are not carefully written, you probably lose\n> anyway.) I don't see a need for allowing more flexibility than that.\n\nTests like a==b will fail for some numbers with DBL_BIG+1.\nIts like I said before, the guys from matlab (in x86 IEEE float) go to \nDBL_BIG+3 to have 'maximum precision'.\n\n> \n> Comments?\n\nYes. I think there are several options.\nI checked the sprintf(ascii, \"%A\", num) output format and all the \nnumbers that would fail because of DBL_DIG=15 are ok. After insertion on \na table and conversion to double after a query, comparison a==b holds.\nAFAICT \"%A\" is system independent.\n\nI would (if I may) propose the following:\n\nHave two parameters, say DOUBLE_OUTPUT and EXTRA_DIGITS. DOUBLE_OUTPUT \nwould select from decimal output or normalized output. EXTRA_DIGITS \nwould add the required extra digits, from 0 (default) to 3, when output \nis decimal.\n\nEXTRA_DIGITS:\nin the range [0:3]. 0 as defualt.\n\nDOUBLE_OUTPUT:\n\n'DECIMAL': sprintf(ascii, \"%.*g\", DBL_DIG+EXTRA_DIGITS, num); (default)\n'NORMALIZED': sprintf(ascii, \"%A\", num);\n\nThe same could be done for floats (float4).\n\nThis way PG does not assume anything (DOUBLE_OUTPUT as 'NORMALIZED'), it \ndoes not hardwire 'inappropriate' assumptions about the number of \nsignificant digits in a double (default EXTRA_DIGITS=0), and it gives \nflexibility (EXTRA_DIGITS!=0) if needed.\nI think this is functional and reasonable.\n\nRegards,\nPedro M. Ferreira\n\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-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Thu, 31 Oct 2002 12:31:47 +0000",
"msg_from": "\"Pedro M. Ferreira\" <pfrazao@ualg.pt>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Just one more note,\n\nMaybe it makes sense that in the proposal below the parameter \nEXTRA_DIGITS could be SIGNIFICANT_DIGITS with a default value of 15 and \nmaximum 18.\nIts more 'documentable' and maybe easy to understand in general.\n\n\nPedro M. Ferreira wrote:\n\n> Yes. I think there are several options.\n> I checked the sprintf(ascii, \"%A\", num) output format and all the \n> numbers that would fail because of DBL_DIG=15 are ok. After insertion on \n> a table and conversion to double after a query, comparison a==b holds.\n> AFAICT \"%A\" is system independent.\n> \n> I would (if I may) propose the following:\n> \n> Have two parameters, say DOUBLE_OUTPUT and EXTRA_DIGITS. DOUBLE_OUTPUT \n> would select from decimal output or normalized output. EXTRA_DIGITS \n> would add the required extra digits, from 0 (default) to 3, when output \n> is decimal.\n> \n> EXTRA_DIGITS:\n> in the range [0:3]. 0 as defualt.\n> \n> DOUBLE_OUTPUT:\n> \n> 'DECIMAL': sprintf(ascii, \"%.*g\", DBL_DIG+EXTRA_DIGITS, num); (default)\n> 'NORMALIZED': sprintf(ascii, \"%A\", num);\n> \n> The same could be done for floats (float4).\n> \n> This way PG does not assume anything (DOUBLE_OUTPUT as 'NORMALIZED'), it \n> does not hardwire 'inappropriate' assumptions about the number of \n> significant digits in a double (default EXTRA_DIGITS=0), and it gives \n> flexibility (EXTRA_DIGITS!=0) if needed.\n> I think this is functional and reasonable.\n> \n> Regards,\n> Pedro M. Ferreira\n> \n>>\n>> regards, 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\n\n-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Thu, 31 Oct 2002 13:47:35 +0000",
"msg_from": "\"Pedro M. Ferreira\" <pfrazao@ualg.pt>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Stephan Szabo <sszabo@megazone23.bigpanda.com> writes:\n> On Wed, 30 Oct 2002, Tom Lane wrote:\n>> sprintf(ascii, \"%.*g\", DBL_DIG+1, num);\n>> and similarly for float4. Given carefully written float I/O routines,\n>> reading the latter output should reproduce the originally stored value.\n\n> Well, on my system, it doesn't look like doing the above sprintfs will\n> actually work for all numbers. I did a simple program using an arbitrary\n> big number and the DBL_DIG+1 output when stuck into another double\n> actually was a different double value. DBL_DIG+2 worked on my system,\n> but...\n\nOh, you're right; I had forgotten about the effects of scale.\nDBL_DIG=15 means that the system claims to distinguish all 15-digit\nvalues, but in a binary system there's more headroom at the bottom end\nof a decimal order of magnitude. For example, 15-digit values are fine:\n\nregression=# select 100000000000001::float8 - 100000000000000::float8;\n ?column?\n----------\n 1\n(1 row)\n\nregression=# select 999999999999999::float8 - 999999999999998::float8;\n ?column?\n----------\n 1\n(1 row)\n\nbut the 9-etc values are over three binary orders of magnitude larger\nthan the 1-etc values, and so they have three less spare bits at the\nright end. The system would be lying to claim DBL_DIG=16:\n\nregression=# select 9999999999999999::float8 - 9999999999999998::float8;\n ?column?\n----------\n 2\n(1 row)\n\neven though values a little over 1e15 are represented perfectly\naccurately:\n\nregression=# select 1000000000000001::float8 - 1000000000000000::float8;\n ?column?\n----------\n 1\n(1 row)\n\nIf you experiment with 17-digit values, you find that the representable\nvalues are about 2 counts apart near 1e16:\n\nregression=# select 10000000000000001::float8 - 10000000000000000::float8;\n ?column?\n----------\n 0\n(1 row)\n\nregression=# select 10000000000000002::float8 - 10000000000000000::float8;\n ?column?\n----------\n 2\n(1 row)\n\nbut they're about 16 counts apart near 9e16:\n\nregression=# select 99999999999999992::float8 - 99999999999999990::float8;\n ?column?\n----------\n 16\n(1 row)\n\nregression=# select 99999999999999991::float8 - 99999999999999990::float8;\n ?column?\n----------\n 0\n(1 row)\n\nwhich is exactly what you'd expect seeing that the values are about a\nfactor of 8 apart.\n\nBottom line: if DBL_DIG=15 and the float arithmetic is binary, then\nthere are some double values that require 17 displayed digits to\ndistinguish, even though not all 16-digit numbers are distinct.\n\nSo I retract my original proposal and instead suggest that we offer\na switch to display either DBL_DIG or DBL_DIG+2 significant digits\n(and correspondingly increase the digits for float4). The DBL_DIG+2\ncase should handle the need for exact dump/reload.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 31 Oct 2002 10:38:59 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions "
},
{
"msg_contents": "\"Pedro M. Ferreira\" <pfrazao@ualg.pt> writes:\n> Its like I said before, the guys from matlab (in x86 IEEE float) go to \n> DBL_BIG+3 to have 'maximum precision'.\n\nApparently they have not read the canonical papers in the field.\n[ googles for a moment... ] See\n\n How to read floating point numbers accurately\n William D. Clinger\n \n How to print floating-point numbers accurately\n Guy L. Steele, Jr., Jon L. White \n\nboth published at the 1990 ACM Conference on Programming Language Design\nand Implementation and subsequently reprinted in ACM SIGPLAN Notices\nVolume 25, Issue 6 (June 1990).\n\nI was misremembering these papers to claim DBL_DIG+1 is enough, but\nactually they prove that DBL_DIG+2 is necessary and sufficient (and\ngive code to do it correctly, too).\n\nPrinting DBL_DIG+3 is just producing an extra garbage digit; it won't\nhelp matters. Any reasonably well-written C library is going to be\nable to reproduce a double value with DBL_DIG+2 digits of I/O; and if\nit's not well-written, I would have no confidence in its ability to do\nso with DBL_DIG+3 digits...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 31 Oct 2002 11:01:21 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions "
},
{
"msg_contents": "Tom Lane wrote:\n> \n> I was misremembering these papers to claim DBL_DIG+1 is enough, but\n> actually they prove that DBL_DIG+2 is necessary and sufficient (and\n> give code to do it correctly, too).\n\nYeahh! If there's a proof its safe to implement.\nI also Googled a bit and found another paper saying that 17 is the\nminimum number of significant digits guaranteed to distinguish among\nIEEE double-precision floating point numbers:\n\n\"Robert G. Burger and R. Kent Dybvig. Printing floating-point numbers \nquickly and accurately. In Proceedings of the ACM SIGPLAN '96 Conference \non Programming Language Design and Implementation, pages 108--116\"\nhttp://citeseer.nj.nec.com/28233.html\n\n> \n> Printing DBL_DIG+3 is just producing an extra garbage digit; it won't\n> help matters. Any reasonably well-written C library is going to be\n> able to reproduce a double value with DBL_DIG+2 digits of I/O; and if\n> it's not well-written, I would have no confidence in its ability to do\n> so with DBL_DIG+3 digits...\n\nOff course. This is also good in terms of dump storage for big float8 \ndatabases. Its one byte less for every float8.\n\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\n-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Thu, 31 Oct 2002 17:41:58 +0000",
"msg_from": "\"Pedro M. Ferreira\" <pfrazao@ualg.pt>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "\"Pedro M. Ferreira\" <pfrazao@ualg.pt> writes:\n> Have two parameters, say DOUBLE_OUTPUT and EXTRA_DIGITS. DOUBLE_OUTPUT \n> would select from decimal output or normalized output. EXTRA_DIGITS \n> would add the required extra digits, from 0 (default) to 3, when output \n> is decimal.\n\nI'm not happy with adding the hex-output option, since it's not\nvery portable and doesn't seem necessary to solve the problem anyway.\n\nBut I think an EXTRA_DIGITS setting might be interesting. In\nparticular, suppose we allowed EXTRA_DIGITS to be negative? Setting\nit to -1 or -2 would go a long way towards eliminating our problems\nwith platform variations in the geometry regression test.\n\nPerhaps something like\n\nextra_float_digits\tint\trange -2 to 2, default 0\n\nextra_float_digits adjusts the number of digits displayed for float4 and\nfloat8 output; the base value of 0 means we output FLT_DIG or DBL_DIG\ndigits respectively.\n\nPer discussion, there's no reason to allow a value greater than 2, but\nI'm not as sure what the lower limit should be --- maybe there's some\nuse in setting it less than -2?\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 31 Oct 2002 12:58:21 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions "
},
{
"msg_contents": "Tom Lane wrote:\n> \n> Bottom line: if DBL_DIG=15 and the float arithmetic is binary, then\n> there are some double values that require 17 displayed digits to\n> distinguish, even though not all 16-digit numbers are distinct.\n> \n> So I retract my original proposal and instead suggest that we offer\n> a switch to display either DBL_DIG or DBL_DIG+2 significant digits\n> (and correspondingly increase the digits for float4). The DBL_DIG+2\n> case should handle the need for exact dump/reload.\n\nNice. This will be good for number storage purposes.\n\nShall it be done with two parameters, 'DOUBLE_FORMAT' and \n'SINGLE_FORMAT', with options 'SHORT' and 'LONG' controlling how the \nsprintf's are done ?\n\nWill someone from pg-people do it or shall I do it for you ?\nAs I said previously, I have seen the GUC stuff and it seem's ok for me \nto do it. I really do not know if there are any restrictions on who \nimplements what respecting PostgreSQL.\nTomorrow we have an holyday in Portugal and I shall leave for the whole \nweek-end, but I can do it on monday.\n\nBest regards,\nPedro M. Ferreira\n\n> \n> \t\t\tregards, tom lane\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----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Thu, 31 Oct 2002 17:58:51 +0000",
"msg_from": "\"Pedro M. Ferreira\" <pfrazao@ualg.pt>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "I sent an email before receiving the one below. I am happier also with \nthe extra_digits way (from the previous email I thought the options were \nDBL_DIG or DBL_DIG+2).\n\n> I'm not happy with adding the hex-output option, since it's not\n> very portable and doesn't seem necessary to solve the problem anyway.\n\nAgree.\n\n> Perhaps something like\n> \n> extra_float_digits\tint\trange -2 to 2, default 0\n> \n> extra_float_digits adjusts the number of digits displayed for float4 and\n> float8 output; the base value of 0 means we output FLT_DIG or DBL_DIG\n> digits respectively.\n\nAgree.\n\n> \n> Per discussion, there's no reason to allow a value greater than 2, but\n> I'm not as sure what the lower limit should be --- maybe there's some\n> use in setting it less than -2?\n\nI could see some use. At least in my type of application. When people \nare shure they only need p significant digits, they can set \nextra_float_digits to an apropriate negative value and spare a lot in \nstorage for dumps and backups. In this case it would make sense to let\nextra_float_digits go to -13.\n\nRegards,\nPedro\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-- \n----------------------------------------------------------------------\nPedro Miguel Frazao Fernandes Ferreira\nUniversidade do Algarve\nFaculdade de Ciencias e Tecnologia\nCampus de Gambelas\n8000-117 Faro\nPortugal\nTel./Fax: (+351) 289 800950 / 289 819403\nhttp://w3.ualg.pt/~pfrazao\n\n",
"msg_date": "Thu, 31 Oct 2002 18:15:03 +0000",
"msg_from": "\"Pedro M. Ferreira\" <pfrazao@ualg.pt>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "On Thu, Oct 31, 2002 at 12:58:21 -0500,\n Tom Lane <tgl@sss.pgh.pa.us> wrote:\n> \n> But I think an EXTRA_DIGITS setting might be interesting. In\n> particular, suppose we allowed EXTRA_DIGITS to be negative? Setting\n> it to -1 or -2 would go a long way towards eliminating our problems\n> with platform variations in the geometry regression test.\n\nMy attempt to avoid this problem with the earthdistance regression used\na cast to numeric to limit the number of digits to the right of the\ndecimal point.\n\nIf the normal number of digits displayed is different between two systems,\nthan displaying a fixed number less digits is still going to result in\ndifferences.\n",
"msg_date": "Thu, 31 Oct 2002 17:33:28 -0600",
"msg_from": "Bruno Wolff III <bruno@wolff.to>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
},
{
"msg_contents": "Bruno Wolff III <bruno@wolff.to> writes:\n> If the normal number of digits displayed is different between two systems,\n> than displaying a fixed number less digits is still going to result in\n> differences.\n\nThat is, however, not the problem we have with the geometry test.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 31 Oct 2002 18:55:03 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions "
},
{
"msg_contents": "\nAdded to TODO:\n\n\t* Add GUC variables extra_float_digits and extra_double_digits\n\tto control output digits\n\n---------------------------------------------------------------------------\n\nTom Lane wrote:\n> \"Pedro M. Ferreira\" <pfrazao@ualg.pt> writes:\n> > Have two parameters, say DOUBLE_OUTPUT and EXTRA_DIGITS. DOUBLE_OUTPUT \n> > would select from decimal output or normalized output. EXTRA_DIGITS \n> > would add the required extra digits, from 0 (default) to 3, when output \n> > is decimal.\n> \n> I'm not happy with adding the hex-output option, since it's not\n> very portable and doesn't seem necessary to solve the problem anyway.\n> \n> But I think an EXTRA_DIGITS setting might be interesting. In\n> particular, suppose we allowed EXTRA_DIGITS to be negative? Setting\n> it to -1 or -2 would go a long way towards eliminating our problems\n> with platform variations in the geometry regression test.\n> \n> Perhaps something like\n> \n> extra_float_digits\tint\trange -2 to 2, default 0\n> \n> extra_float_digits adjusts the number of digits displayed for float4 and\n> float8 output; the base value of 0 means we output FLT_DIG or DBL_DIG\n> digits respectively.\n> \n> Per discussion, there's no reason to allow a value greater than 2, but\n> I'm not as sure what the lower limit should be --- maybe there's some\n> use in setting it less than -2?\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": "Thu, 7 Nov 2002 23:27:22 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: float output precision questions"
}
] |
[
{
"msg_contents": "\n> > The problem with flex is, that the generated c file does #include <unistd.h>\n> > before we #include \"postgres.h\".\n> > In this situation _LARGE_FILES is not defined for unistd.h and unistd.h\n> > chooses to define _LARGE_FILE_API, those two are not compatible.\n> \n> Yeah. AFAICS the only way around this is to avoid doing any I/O\n> operations in the flex-generated files. Fortunately, that's not much\n> of a restriction.\n\nUnfortunately I do not think that is sufficient, since the problem is already \nat the #include level. The compiler barfs on the second #include <unistd.h>\nfrom postgres.h\n\nAndreas\n",
"msg_date": "Tue, 29 Oct 2002 14:17:25 +0100",
"msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>",
"msg_from_op": true,
"msg_subject": "Re: pg_dump and large files - is this a problem? "
},
{
"msg_contents": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at> writes:\n>> Yeah. AFAICS the only way around this is to avoid doing any I/O\n>> operations in the flex-generated files. Fortunately, that's not much\n>> of a restriction.\n\n> Unfortunately I do not think that is sufficient, since the problem is already\n> at the #include level. The compiler barfs on the second #include <unistd.h>\n> from postgres.h\n\nAIX is too stupid to wrap unistd.h in an \"#ifndef\" to protect against\ndouble inclusion? I suppose we could do that for them...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Tue, 29 Oct 2002 08:37:29 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and large files - is this a problem? "
}
] |
[
{
"msg_contents": "\n> >> Yeah. AFAICS the only way around this is to avoid doing any I/O\n> >> operations in the flex-generated files. Fortunately, \n> that's not much\n> >> of a restriction.\n> \n> > Unfortunately I do not think that is sufficient, since the problem is already\n> > at the #include level. The compiler barfs on the second #include <unistd.h>\n> > from postgres.h\n> \n> AIX is too stupid to wrap unistd.h in an \"#ifndef\" to protect against\n> double inclusion? I suppose we could do that for them...\n\nI guess that is exactly not wanted, since that would hide the actual\nproblem, namely that _LARGE_FILE_API gets defined (off_t --> 32bit).\nThus I think IBM did not protect unistd.h on purpose.\n\nAndreas\n",
"msg_date": "Tue, 29 Oct 2002 17:08:14 +0100",
"msg_from": "\"Zeugswetter Andreas SB SD\" <ZeugswetterA@spardat.at>",
"msg_from_op": true,
"msg_subject": "Re: pg_dump and large files - is this a problem? "
},
{
"msg_contents": "Zeugswetter Andreas SB SD writes:\n\n> > AIX is too stupid to wrap unistd.h in an \"#ifndef\" to protect against\n> > double inclusion? I suppose we could do that for them...\n>\n> I guess that is exactly not wanted, since that would hide the actual\n> problem, namely that _LARGE_FILE_API gets defined (off_t --> 32bit).\n> Thus I think IBM did not protect unistd.h on purpose.\n\nI think the problem is more accurately described thus: Flex generated\nfiles include <stdio.h> before \"postgres.h\" due to the way it lays out the\ncode in the output. stdio.h does something which prevents switching to\nthe large file model later on in postgres.h. (This manifests itself in\nunistd.h, but unistd.h itself is not the problem per se.)\n\nThe proposed fix was to include the flex output in some other file (such\nas the corresponding grammar file) rather than to compile it separately.\nThe patch just needs to be tried out.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Tue, 29 Oct 2002 23:19:32 +0100 (CET)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and large files - is this a problem? "
},
{
"msg_contents": "Peter Eisentraut <peter_e@gmx.net> writes:\n> The proposed fix was to include the flex output in some other file (such\n> as the corresponding grammar file) rather than to compile it separately.\n\nSeems like a reasonable solution. Can you make that happen in the next\nday or two? If not, I'll take a whack at it ...\n\n\t\t\tregards, tom lane\n",
"msg_date": "Thu, 31 Oct 2002 17:07:28 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and large files - is this a problem? "
},
{
"msg_contents": "Peter Eisentraut <peter_e@gmx.net> writes:\n> I think the problem is more accurately described thus: Flex generated\n> files include <stdio.h> before \"postgres.h\" due to the way it lays out the\n> code in the output. stdio.h does something which prevents switching to\n> the large file model later on in postgres.h. (This manifests itself in\n> unistd.h, but unistd.h itself is not the problem per se.)\n\n> The proposed fix was to include the flex output in some other file (such\n> as the corresponding grammar file) rather than to compile it separately.\n\nI have made this change. CVS tip should compile cleanly now on machines\nwhere this is an issue.\n\n\t\t\tregards, tom lane\n",
"msg_date": "Fri, 01 Nov 2002 19:31:44 -0500",
"msg_from": "Tom Lane <tgl@sss.pgh.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: pg_dump and large files - is this a problem? "
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Jason Tishler [mailto:jason@tishler.net] \n> Sent: 29 October 2002 14:48\n> To: Dave Page\n> Cc: Bruce Momjian; PostgreSQL-development; Thomas Lockhart; \n> Tom Lane; Pgsql-Cygwin\n> Subject: Re: [HACKERS] Request for supported platforms\n> \n> \n> I found a solution to the above which will hopefully find its way \n> into the next Cygwin release:\n> \n> http://cygwin.com/ml/cygwin-patches/2002-q4/msg00042.html\n> \n> > Yes, I'm seeing errors with this on my updated Cygwin very early in \n> > the build. I did think it was my hacked about installation, but I \n> > guess not!\n> \n> A quick and *dirty* fix for this problem is to temporarily delete the \n> above two entries from your stdio.h file.\n\nHi Jason,\n\nAll regression tests pass with the above hack on:\n\nCYGWIN_NT-5.1 PC9 1.3.14(0.62/3/2) 2002-10-24 10:48 i686 unknown\n\nHackers: As the Cygwin release that is actively supported is the binary\ndistribution that Jason builds, I would think this is OK to be listed as\nsupported if no-one disagrees...\n\nRegards, Dave.\n",
"msg_date": "Tue, 29 Oct 2002 16:57:58 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "Dave,\n\nOn Tue, Oct 29, 2002 at 04:57:58PM -0000, Dave Page wrote:\n> All regression tests pass with the above hack on:\n> \n> CYGWIN_NT-5.1 PC9 1.3.14(0.62/3/2) 2002-10-24 10:48 i686 unknown\n\nThanks for the above.\n\n> Hackers: As the Cygwin release that is actively supported is the\n> binary distribution that Jason builds, I would think this is OK to be\n> listed as supported if no-one disagrees...\n\nUmm... Should I disagree? :,)\n\nJason\n",
"msg_date": "Tue, 29 Oct 2002 13:57:58 -0500",
"msg_from": "Jason Tishler <jason@tishler.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "\nI kept Dave and Jason's name on the report.\n\nPorts list updated:\n\n http://candle.pha.pa.us/main/writings/pgsql/sgml/supported-platforms.html\n\n---------------------------------------------------------------------------\nDave Page wrote:\n> \n> \n> > -----Original Message-----\n> > From: Jason Tishler [mailto:jason@tishler.net] \n> > Sent: 29 October 2002 14:48\n> > To: Dave Page\n> > Cc: Bruce Momjian; PostgreSQL-development; Thomas Lockhart; \n> > Tom Lane; Pgsql-Cygwin\n> > Subject: Re: [HACKERS] Request for supported platforms\n> > \n> > \n> > I found a solution to the above which will hopefully find its way \n> > into the next Cygwin release:\n> > \n> > http://cygwin.com/ml/cygwin-patches/2002-q4/msg00042.html\n> > \n> > > Yes, I'm seeing errors with this on my updated Cygwin very early in \n> > > the build. I did think it was my hacked about installation, but I \n> > > guess not!\n> > \n> > A quick and *dirty* fix for this problem is to temporarily delete the \n> > above two entries from your stdio.h file.\n> \n> Hi Jason,\n> \n> All regression tests pass with the above hack on:\n> \n> CYGWIN_NT-5.1 PC9 1.3.14(0.62/3/2) 2002-10-24 10:48 i686 unknown\n> \n> Hackers: As the Cygwin release that is actively supported is the binary\n> distribution that Jason builds, I would think this is OK to be listed as\n> supported if no-one disagrees...\n> \n> Regards, Dave.\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, 29 Oct 2002 14:09:03 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "Dave Page writes:\n\n> Hackers: As the Cygwin release that is actively supported is the binary\n> distribution that Jason builds, I would think this is OK to be listed as\n> supported if no-one disagrees...\n\nI disagree. We document as supported those platforms that build out of\nthe box, not those that build somehow, somewhere, by someone.\n\nRather than advocating methods to manually edit your system headers we\nshould try to find out what the problem really is, such as by analyzing\nconfig.log.\n\n-- \nPeter Eisentraut peter_e@gmx.net\n\n",
"msg_date": "Wed, 30 Oct 2002 19:36:40 +0100 (CET)",
"msg_from": "Peter Eisentraut <peter_e@gmx.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "Peter,\n\nOn Wed, Oct 30, 2002 at 07:36:40PM +0100, Peter Eisentraut wrote:\n> Dave Page writes:\n> > Hackers: As the Cygwin release that is actively supported is the\n> > binary distribution that Jason builds, I would think this is OK to\n> > be listed as supported if no-one disagrees...\n> \n> I disagree. We document as supported those platforms that build out\n> of the box, not those that build somehow, somewhere, by someone.\n> \n> Rather than advocating methods to manually edit your system headers we\n> should try to find out what the problem really is, such as by\n> analyzing config.log.\n\nDid you miss the following?\n\n http://archives.postgresql.org/pgsql-hackers/2002-10/msg01303.php\n\nAs you can see, I have already performed root cause analysis of theses\nproblems *and* have taken the proper steps to ensure that PostgreSQL\nbuilds OOTB under Cygwin (after the next Cygwin release).\n\nJason\n",
"msg_date": "Wed, 30 Oct 2002 14:04:08 -0500",
"msg_from": "Jason Tishler <jason@tishler.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Jason Tishler [mailto:jason@tishler.net] \n> Sent: 29 October 2002 18:58\n> To: Dave Page\n> Cc: Bruce Momjian; PostgreSQL-development; Thomas Lockhart; \n> Tom Lane; Pgsql-Cygwin\n> Subject: Re: [HACKERS] Request for supported platforms\n> \n> > Hackers: As the Cygwin release that is actively supported is the \n> > binary distribution that Jason builds, I would think this \n> is OK to be \n> > listed as supported if no-one disagrees...\n> \n> Umm... Should I disagree? :,)\n> \n\nEntirely up to you - you do the build :-). I'm just pointing out that\nthe line wrt to Cygwin is generally \"use the standard package\", so if we\nstick to that, as long as you build it OK, most people needn't worry\n'bout hacking stdio.h.\n\nRegards, Dave.\n",
"msg_date": "Tue, 29 Oct 2002 21:00:20 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
},
{
"msg_contents": "Dave,\n\nOn Tue, Oct 29, 2002 at 09:00:20PM -0000, Dave Page wrote:\n> > -----Original Message-----\n> > From: Jason Tishler [mailto:jason@tishler.net] \n> > Sent: 29 October 2002 18:58\n> > \n> > > Hackers: As the Cygwin release that is actively supported is the\n> > > binary distribution that Jason builds, I would think this is OK to\n> > > be listed as supported if no-one disagrees...\n ^^^^^^^^^\n *********\n\n> > Umm... Should I disagree? :,)\n> \n> Entirely up to you - you do the build :-).\n\nI was meekly trying to voice my concern about who supplies the \"support\"\nmentioned above.\n\n> I'm just pointing out that the line wrt to Cygwin is generally \"use\n> the standard package\", so if we stick to that, as long as you build it\n> OK, most people needn't worry 'bout hacking stdio.h.\n\nIf my patch gets accepted, then the stdio.h hack won't be necessary for\nanyone (after the next Cygwin release).\n\nJason\n",
"msg_date": "Tue, 29 Oct 2002 16:17:38 -0500",
"msg_from": "Jason Tishler <jason@tishler.net>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Request for supported platforms"
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Bruce Momjian [mailto:pgman@candle.pha.pa.us] \n> Sent: 29 October 2002 19:34\n> To: Dave Page\n> Cc: PostgreSQL-development; Thomas Lockhart; Tom Lane\n> Subject: Re: [HACKERS] Request for supported platforms\n> \n> \n> \n> Attached is a diff to fix the pclose problem. It turns out \n> there was code in there to make popen/pclose be \n> _popen/_pclose, but it was only in common.c, even in 7.2.3 \n> (print.c). Not sure how it would compile in the past with \n> that. Maybe it didn't. Anyway, this is committed and should \n> _help_ with the compile.\n\nThanks Bruce, builds fine now.\n\nActually, I'm sure it did just a few weeks ago - We've been developing\npgAdmin III in C++ using libpq on Windows, Linux and FreeBSD for a few\nweeks now and we've been using the CVS code.\n\nRegards, Dave\n",
"msg_date": "Tue, 29 Oct 2002 21:10:52 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: Request for supported platforms"
}
] |
[
{
"msg_contents": "Dear Sir,\nI use SCO Open Server 5.0.5 on an intel box. Although I have installed and\nused PostgreSQL on Linux, setting it on SCO has not been smooth :)\nI have downloaded the latest version ie \"Postgresql-7.2.3.\"\nI also installed \"ant\" package for using java. I have \"tcl8.0,\" \"tk8.0,\"\n\"itclsh3.0\" and \"itkwish3.0\" installed in my system.\n\"gcc version 2.7.2.1\"\n\" Java 2 SDK, Standard Edition, v. 1.2.1\"\n\nMy \"configure\" command was as following ...\n./configure --prefix=/data/pgsql --with-perl --with-tcl\n--with-tclconfig=/data/tcl/lib --with-tkconfig=/data/tk/lib --with-java\n\nThe last output were ....\n...............\nupdating cache ./config.cache\ncreating ./config.status\ncreating GNUmakefile\ncreating src/GNUmakefile\ncreating src/Makefile.global\ncreating src/backend/port/Makefile\ncreating src/include/pg_config.h\nlinking ./src/backend/port/dynloader/sco.c to src/backend/port/dynloader.c\nlinking ./src/backend/port/dynloader/sco.h to src/include/dynloader.h\nlinking ./src/include/port/sco.h to src/include/pg_config_os.h\nlinking ./src/makefiles/Makefile.sco to src/Makefile.port\nlinking ./src/backend/port/tas/dummy.s to src/backend/port/tas.s\n\nThen i give the \"gmake\" command. The compiling stops on an error and\nexits after some time. The \"messages\" in this step 2 (step2mesg.txt) and\nthe \"errors\" (step2err.txt) are reproduced in the 2 text files attached.\n\nI'd be thankful to you if you can help me out sort the problem. I got your\nmail-ids from the net and came to know that you are working on the similar\nlines.\n\nWaiting for a quick response from your end. kindly inform me if you have\nalready solved the problem, or whether any patch is available. Any\ndocumentation or url will be highly helpful.\n\nThanking You in anticipation.\n\nwith regards\nShibashish Satpathy\nshiba@it.iitb.ac.in\nSoftware Engineer\nIIT Bombay, India",
"msg_date": "Wed, 30 Oct 2002 02:49:16 +0530",
"msg_from": "Shibashish <shiba@it.iitb.ac.in>",
"msg_from_op": true,
"msg_subject": "PostgreSQL Installation on SCO"
},
{
"msg_contents": "\nHave you read the SCO FAQ?\n\n\thttp://www.us.postgresql.org/users-lounge/docs/faq.html\n\nAlso, we are about to release 7.3. Would you please try 7.3beta3 and\nlet us know how that works?\n\n---------------------------------------------------------------------------\n\nShibashish wrote:\n> Dear Sir,\n> I use SCO Open Server 5.0.5 on an intel box. Although I have installed and\n> used PostgreSQL on Linux, setting it on SCO has not been smooth :)\n> I have downloaded the latest version ie \"Postgresql-7.2.3.\"\n> I also installed \"ant\" package for using java. I have \"tcl8.0,\" \"tk8.0,\"\n> \"itclsh3.0\" and \"itkwish3.0\" installed in my system.\n> \"gcc version 2.7.2.1\"\n> \" Java 2 SDK, Standard Edition, v. 1.2.1\"\n> \n> My \"configure\" command was as following ...\n> ./configure --prefix=/data/pgsql --with-perl --with-tcl\n> --with-tclconfig=/data/tcl/lib --with-tkconfig=/data/tk/lib --with-java\n> \n> The last output were ....\n> ...............\n> updating cache ./config.cache\n> creating ./config.status\n> creating GNUmakefile\n> creating src/GNUmakefile\n> creating src/Makefile.global\n> creating src/backend/port/Makefile\n> creating src/include/pg_config.h\n> linking ./src/backend/port/dynloader/sco.c to src/backend/port/dynloader.c\n> linking ./src/backend/port/dynloader/sco.h to src/include/dynloader.h\n> linking ./src/include/port/sco.h to src/include/pg_config_os.h\n> linking ./src/makefiles/Makefile.sco to src/Makefile.port\n> linking ./src/backend/port/tas/dummy.s to src/backend/port/tas.s\n> \n> Then i give the \"gmake\" command. The compiling stops on an error and\n> exits after some time. The \"messages\" in this step 2 (step2mesg.txt) and\n> the \"errors\" (step2err.txt) are reproduced in the 2 text files attached.\n> \n> I'd be thankful to you if you can help me out sort the problem. I got your\n> mail-ids from the net and came to know that you are working on the similar\n> lines.\n> \n> Waiting for a quick response from your end. kindly inform me if you have\n> already solved the problem, or whether any patch is available. Any\n> documentation or url will be highly helpful.\n> \n> Thanking You in anticipation.\n> \n> with regards\n> Shibashish Satpathy\n> shiba@it.iitb.ac.in\n> Software Engineer\n> IIT Bombay, India\n> \n\nContent-Description: \n\n[ Attachment, skipping... ]\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": "Fri, 1 Nov 2002 17:28:30 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: PostgreSQL Installation on SCO"
}
] |
[
{
"msg_contents": "I have a client that I am developing a web-based business application for\nusing Perl / PHP, DHTML, and postgreSQL on a FreeBSD server with Apache...\n\nThe client realized (or hopes) that this application may become mission\ncritical (to his clients / end-users) and was asking about the possibility\nof having a replicated database using postgreSQL in case the database server\ntakes a dive (paranoia). He is somewhat familiar with MS Access (I gathered\nhe was thinking of the MS Access replication / front-end / back-end /\nsynchronized database concept).\n\nI have read through all 7 chapters of the PG documentation and didn't see\nanything about replicating a postgreSQL database on a secondary database\nserver and having the primary server push all data manipulations on through\nto the secondary server. The secondary server would be available (through\nthe connect script) if the primary server died.\n\nDoes anyone know of a reasonable means of synchronizing two (or more)\npostgreSQL database servers in order to maintain two COMPLETE concurrent\ncopies of the same database ?\n\nSince I can already see REFINT problems in maintaining the integrity of\nserial values, trigger problems, and stored procedure problems (UDF's),\nperhaps I should just suggest a robust database backup and restore scheme\nwith a standby postgreSQL server ready to accept a pgdumpall.\n\nAlthough the documentation claims that CREATE DATABASE is not intended as a\n\"copy database\" utility, maybe a CRON event (say every 4 hours) on the\nsecondary server to:\n\nDROP database 'backupdatabase';\nCREATE DATABASE 'backupdatabase' WITH TEMPLATE = 'livedatabase';\n\nwould do the trick....\n\nI am completely open to suggestions here...\n\nTIA.\n\nGP\n\n\n",
"msg_date": "Tue, 29 Oct 2002 15:56:46 -0800",
"msg_from": "\"Greg Patnude\" <GPatnude@adelphia.net>",
"msg_from_op": true,
"msg_subject": "Database replication... - Mission Critical DBMS's -- Taking\n\tpostgreSQL to the Enterprise Computing level..."
},
{
"msg_contents": "On Tue, 2002-10-29 at 17:56, Greg Patnude wrote:\n> Does anyone know of a reasonable means of synchronizing two (or more)\n> postgreSQL database servers in order to maintain two COMPLETE concurrent\n> copies of the same database ?\n\nI'm not sure how complete it is (just starting to look at this myself)\nbut 'dbbalancer' at least makes some claims to this. It acts as\n(basically) a postgresql proxy server, and can send queries to multiple\nservers. My presumption would be that if you initialize 2 databases to\na known identical start, have all the same triggers and rules on both,\nthen send all queries to both databases, you will have 2 identical\ndatabases at the end. \n\nDon't know how well that will work in practice tho. I should know more\nin the next couple of weeks. \n\nb.g.\n\n\n\n",
"msg_date": "30 Oct 2002 08:24:28 -0600",
"msg_from": "Bill Gribble <grib@linuxdevel.com>",
"msg_from_op": false,
"msg_subject": "Re: Database replication... - Mission Critical DBMS's --"
},
{
"msg_contents": "You might want to check out\nhttp://gborg.postgresql.org/genpage?replication_research which has\ninformation and links to several replication solutions for postgresql. \n\nthe techdocs.postgresql.org website also has a good number of papers\nregarding replication, as well as a \"high availability how-to\" that\nwould probably be worth reading.\n\nRobert Treat\n\nOn Wed, 2002-10-30 at 09:24, Bill Gribble wrote:\n> On Tue, 2002-10-29 at 17:56, Greg Patnude wrote:\n> > Does anyone know of a reasonable means of synchronizing two (or more)\n> > postgreSQL database servers in order to maintain two COMPLETE concurrent\n> > copies of the same database ?\n> \n> I'm not sure how complete it is (just starting to look at this myself)\n> but 'dbbalancer' at least makes some claims to this. It acts as\n> (basically) a postgresql proxy server, and can send queries to multiple\n> servers. My presumption would be that if you initialize 2 databases to\n> a known identical start, have all the same triggers and rules on both,\n> then send all queries to both databases, you will have 2 identical\n> databases at the end. \n> \n> Don't know how well that will work in practice tho. I should know more\n> in the next couple of weeks. \n> \n> b.g.\n> \n> \n> \n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n\n\n\n",
"msg_date": "30 Oct 2002 10:18:47 -0500",
"msg_from": "Robert Treat <xzilla@users.sourceforge.net>",
"msg_from_op": false,
"msg_subject": "Re: Database replication... - Mission Critical DBMS's --"
},
{
"msg_contents": "On Tue, Oct 29, 2002 at 03:56:46PM -0800, Greg Patnude wrote:\n\n> I have read through all 7 chapters of the PG documentation and didn't see\n> anything about replicating a postgreSQL database on a secondary database\n> server and having the primary server push all data manipulations on through\n> to the secondary server. The secondary server would be available (through\n> the connect script) if the primary server died.\n> \n> Does anyone know of a reasonable means of synchronizing two (or more)\n> postgreSQL database servers in order to maintain two COMPLETE concurrent\n> copies of the same database ?\n\nI think what you want is that any change, including schema changes,\n&c., on the master database are echoed to the slave. I know rserv\ncan't do that, and I don't _think_ dbmirror can, either. But both of\nthose are possibilities. AFAIK, the contrib/rserv code is completely\nbroken in the 7.3 series, so dbmirror might be the free answer to\npick; rserv has a commercial cousin which we use.\n\nThe problem with these is that they send _data_ to the slave, and use\nstandard SQL statements. So sequences &c. are not carried over. You\nneed a script to do that. It's not an instantaneous failover.\n\nOn the other hand, if your client is comparing with Access, the\nreliability will be so much better that perhaps some extra work in\nthe rare case of failure will be acceptable. It is for us, and we\nhave extremely stringent SLAs which constrain how long we could be\ndown if (heaven forfend!) our master database ever died.\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, 30 Oct 2002 10:30:45 -0500",
"msg_from": "Andrew Sullivan <andrew@libertyrms.info>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Database replication... - Mission Critical DBMS's --\n\tTaking postgreSQL to the Enterprise Computing level..."
},
{
"msg_contents": "On Wed, Oct 30, 2002 at 10:18:47AM -0500, Robert Treat wrote:\n> regarding replication, as well as a \"high availability how-to\" that\n> would probably be worth reading.\n\nThe high availability howto suggests using rsync to synchronise the\ndata areas of two data servers. That is an _extremely bad_ idea. \nI've suggested before that the link be removed, because it recommends\nsomething almost guaranteed to introduce massive database corruption\nat some point. If there's no load and you have a fast network, you\nmight get lucky. But it is an extremely dangerous plan.\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, 30 Oct 2002 10:33:09 -0500",
"msg_from": "Andrew Sullivan <andrew@libertyrms.info>",
"msg_from_op": false,
"msg_subject": "Re: Database replication... - Mission Critical DBMS's --"
},
{
"msg_contents": "Andrew Sullivan wrote:\n> \n> On Wed, Oct 30, 2002 at 10:18:47AM -0500, Robert Treat wrote:\n> > regarding replication, as well as a \"high availability how-to\" that\n> > would probably be worth reading.\n> \n> The high availability howto suggests using rsync to synchronise the\n> data areas of two data servers. That is an _extremely bad_ idea.\n> I've suggested before that the link be removed, because it recommends\n> something almost guaranteed to introduce massive database corruption\n> at some point. If there's no load and you have a fast network, you\n> might get lucky. But it is an extremely dangerous plan.\n\nOk, have just removed the link. Sorry for not getting around to it\nbefore Andrew.\n\n(Bruce pointed out your email, otherwise I would have missed it again\ntoo).\n\n:-/\n\nRegards and best wishes,\n\nJustin Clift\n\n\n> A\n> \n> --\n> ----\n> Andrew Sullivan 204-4141 Yonge Street\n> Liberty RMS Toronto, Ontario Canada\n> <andrew@libertyrms.info> M2P 2A8\n> +1 416 646 3304 x110\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": "Sat, 02 Nov 2002 12:42:47 +1100",
"msg_from": "Justin Clift <justin@postgresql.org>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Database replication... - Mission Critical "
}
] |
[
{
"msg_contents": "\nMay I be allowed to use '\\n' as a delimiter for \"copy from\"?\n\nI was able to use it to \"copy to\" a file, and it worked as\nexpected, but when I tried to \"copy from\" the file, I got an\nerror because it treated the end of line as the end of the\ntuple (which I recognize is documented behavior). If the\ncopy command expected a 1:1 relationship between the fields\nin the \"copy from\" file and the fields in the target table,\nthen it seemed to me it could read the file correctly?\nI can see that it might be considered too dicey to allow.\n\n\nLarry\n\n",
"msg_date": "30 Oct 2002 02:40:57 -0000",
"msg_from": "Larry Riedel <larryr@saturn.sdsu.edu>",
"msg_from_op": true,
"msg_subject": "copy from ... with delimiter '\\n'"
}
] |
[
{
"msg_contents": "Karel Zak kirjutas K, 30.10.2002 kell 10:08:\n> \n> Hi, \n> \n> I read a presentation about Object-Oriented features in relation DBs.\n> The nice are UDT (user defined type):\n> \n> CREATE TABLE person (\n> name varchar(32),\n> address ROW( street varchar(32),\n> town varchar(32)),\n> age int\n> );\n> \n> INSERT INTO person VALUES ('Bill', ('Somestreet', 'Sometown'), 33);\n> \n> SELECT name, address.town FROM person;\n> \n> \n> We have composite types in PostgreSQL and I think we can use it for this:\n> \n> CREATE TYPE addr AS (street varchar(32), town varchar(32));\n> CREATE TABLE person (\n> name varchar(32),\n> address addr,\n> age int\n> );\n> \n> \n> Comments? I nothinig found about OO in the current TODO.\n\nI'm writing a small proposal for evoving inheritance and other OO\nfeatures in 7.4 and beyond. Will post once 7.3 is out.\n\n> BTW, my\n> examples are only small part of possible OO features, the others\n> ideas are for example define PRIVATE/PUBLIC attributes in composite\n> types\n\nAt least the \"Third Manifesto\" by Date et.al. claims that PRIVATE/PUBLIC\nis better left to standard access control mechanisms (GRANT/REVOKE). \n\nI agree to that.\n\n> and methods, \"SELECT p.name FROM person p WHERE p.pay->tax() > 100;\"\n\nThe methods will probably have problems with syntax clashes with\nexisting stuff.\n\n--------------\nHannu\n\n\n",
"msg_date": "30 Oct 2002 08:35:09 +0200",
"msg_from": "Hannu Krosing <hannu@tm.ee>",
"msg_from_op": true,
"msg_subject": "Re: OO future"
},
{
"msg_contents": "\n Hi, \n\n I read a presentation about Object-Oriented features in relation DBs.\n The nice are UDT (user defined type):\n\n CREATE TABLE person (\n name varchar(32),\n address ROW( street varchar(32),\n town varchar(32)),\n age int\n );\n\n INSERT INTO person VALUES ('Bill', ('Somestreet', 'Sometown'), 33);\n\n SELECT name, address.town FROM person;\n\n\n We have composite types in PostgreSQL and I think we can use it for this:\n\n CREATE TYPE addr AS (street varchar(32), town varchar(32));\n CREATE TABLE person (\n name varchar(32),\n address addr,\n age int\n );\n \n\n Comments? I nothinig found about OO in the current TODO. BTW, my\n examples are only small part of possible OO features, the others\n ideas are for example define PRIVATE/PUBLIC attributes in composite\n types and methods, \"SELECT p.name FROM person p WHERE p.pay->tax() > 100;\"\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, 30 Oct 2002 09:08:45 +0100",
"msg_from": "Karel Zak <zakkr@zf.jcu.cz>",
"msg_from_op": false,
"msg_subject": "OO future"
},
{
"msg_contents": "On Wed, Oct 30, 2002 at 08:35:09AM +0200, Hannu Krosing wrote:\n> > Comments? I nothinig found about OO in the current TODO.\n> \n> I'm writing a small proposal for evoving inheritance and other OO\n> features in 7.4 and beyond. Will post once 7.3 is out.\n\n Good! I look forward.\n\n> > BTW, my\n> > examples are only small part of possible OO features, the others\n> > ideas are for example define PRIVATE/PUBLIC attributes in composite\n> > types\n> \n> At least the \"Third Manifesto\" by Date et.al. claims that PRIVATE/PUBLIC\n> is better left to standard access control mechanisms (GRANT/REVOKE). \n> \n> I agree to that.\n\n Yes, but it expect access control pre-column and for per composite type\n attribute. I understand PRIVATE as some internal data for methods\n and it needn't a speciffic access control, because control must be\n define for methods, and other way access PRIVATE data is not possible.\n IMHO it's better.\n \n The other important thing is possibility create table from type: \n CREATE TABLE adresses AS address_t;\n\n> > and methods, \"SELECT p.name FROM person p WHERE p.pay->tax() > 100;\"\n> \n> The methods will probably have problems with syntax clashes with\n> existing stuff.\n\n Hmm, p.pay.tax() ?\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, 30 Oct 2002 09:51:34 +0100",
"msg_from": "Karel Zak <zakkr@zf.jcu.cz>",
"msg_from_op": false,
"msg_subject": "Re: OO future"
},
{
"msg_contents": "There's a nice simple book from 1999 by Stonebreaker and a technologist \nform Informix about object-relational features. PostgreSQL has \ndefinately started to lag on that front, while shoring up other aspects \nof the RDBMS. A simple (simple?) start might just be supporting dot \nnotation and other syntactical niceties around the table-as-column concept.\n\nWe have recently had an opportunity to evaluate the object relational \ncapabilities of Oracle9i, and sad to say, they have finally surpassed \nPgSQL on the OO front in this release. Very easy composite type creation \non the command line, composite types are easily indexable based on their \nattributes (a custom type which wraps a spatial type can be indexed \nspatially, etc), function adding is easy too. Now, adding completely new \ntypes in C might be hell, but we haven't checked that yet :)\n\nP.\n\nKarel Zak wrote:\n> Hi, \n> \n> I read a presentation about Object-Oriented features in relation DBs.\n> The nice are UDT (user defined type):\n> \n> CREATE TABLE person (\n> name varchar(32),\n> address ROW( street varchar(32),\n> town varchar(32)),\n> age int\n> );\n> \n> INSERT INTO person VALUES ('Bill', ('Somestreet', 'Sometown'), 33);\n> \n> SELECT name, address.town FROM person;\n> \n> \n> We have composite types in PostgreSQL and I think we can use it for this:\n> \n> CREATE TYPE addr AS (street varchar(32), town varchar(32));\n> CREATE TABLE person (\n> name varchar(32),\n> address addr,\n> age int\n> );\n> \n> \n> Comments? I nothinig found about OO in the current TODO. BTW, my\n> examples are only small part of possible OO features, the others\n> ideas are for example define PRIVATE/PUBLIC attributes in composite\n> types and methods, \"SELECT p.name FROM person p WHERE p.pay->tax() > 100;\"\n> \n> Karel\n> \n\n\n-- \n __\n /\n | Paul Ramsey\n | Refractions Research\n | Email: pramsey@refractions.net\n | Phone: (250) 885-0632\n \\_\n\n",
"msg_date": "Wed, 30 Oct 2002 08:24:56 -0800",
"msg_from": "Paul Ramsey <pramsey@refractions.net>",
"msg_from_op": false,
"msg_subject": "Re: OO future"
},
{
"msg_contents": "\nHere is an O'Reilly article talking about Oracle's OO capabilities:\n\n\thttp://www.oreillynet.com/pub/a/network/2002/10/29/Feuerstein.html\n\n---------------------------------------------------------------------------\n\nHannu Krosing wrote:\n> Karel Zak kirjutas K, 30.10.2002 kell 10:08:\n> > \n> > Hi, \n> > \n> > I read a presentation about Object-Oriented features in relation DBs.\n> > The nice are UDT (user defined type):\n> > \n> > CREATE TABLE person (\n> > name varchar(32),\n> > address ROW( street varchar(32),\n> > town varchar(32)),\n> > age int\n> > );\n> > \n> > INSERT INTO person VALUES ('Bill', ('Somestreet', 'Sometown'), 33);\n> > \n> > SELECT name, address.town FROM person;\n> > \n> > \n> > We have composite types in PostgreSQL and I think we can use it for this:\n> > \n> > CREATE TYPE addr AS (street varchar(32), town varchar(32));\n> > CREATE TABLE person (\n> > name varchar(32),\n> > address addr,\n> > age int\n> > );\n> > \n> > \n> > Comments? I nothinig found about OO in the current TODO.\n> \n> I'm writing a small proposal for evoving inheritance and other OO\n> features in 7.4 and beyond. Will post once 7.3 is out.\n> \n> > BTW, my\n> > examples are only small part of possible OO features, the others\n> > ideas are for example define PRIVATE/PUBLIC attributes in composite\n> > types\n> \n> At least the \"Third Manifesto\" by Date et.al. claims that PRIVATE/PUBLIC\n> is better left to standard access control mechanisms (GRANT/REVOKE). \n> \n> I agree to that.\n> \n> > and methods, \"SELECT p.name FROM person p WHERE p.pay->tax() > 100;\"\n> \n> The methods will probably have problems with syntax clashes with\n> existing stuff.\n> \n> --------------\n> Hannu\n> \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": "Fri, 1 Nov 2002 18:31:53 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: OO future"
}
] |
[
{
"msg_contents": "\n\n> -----Original Message-----\n> From: Greg Copeland [mailto:greg@copelandconsulting.net] \n> Sent: 30 October 2002 01:08\n> To: Dave Page\n> Subject: Re: [HACKERS] Request for supported platforms\n> \n> \n> C++? Really? What GUI toolkit is being used?\n> \n> Just curious.\n\nwxWindows. The CVS is online if anyone is interested in taking a peek -\nit's at http://cvs.pgadmin.org/. The code is in the pgadmin3 module.\n\nPlease bear in mind though, that Mark & I have only been using C++ for\nabout a month (though things are coming along very quickly), so there\nmay be some nasties in our code. Any constructive comments from any of\nthe -hackers would be more than welcome.\n\nRegards, Dave.\n",
"msg_date": "Wed, 30 Oct 2002 08:19:11 -0000",
"msg_from": "\"Dave Page\" <dpage@vale-housing.co.uk>",
"msg_from_op": true,
"msg_subject": "Re: pgAdmin III (Was: Request for supported platforms)"
},
{
"msg_contents": "Since you're using wxWindows, I *HIGHLY* recommend obtaining a license\nto wxDesigner from http://www.roebling.de/. It allows for very rapid\nGUI design. It also understands various sizers and makes it SOOO much\neasier to make use of them. Once you understand sizers, you'll love\nthem but they are somewhat hard to use without a tool to assist.\n\nAlso, for the about box, I also suggest that you make use of a wxHTML\ndialog. The cool thing about doing this is that you can create links\nand even place nice images within the about box. Plus, maintaining the\nabout box can be easily done via an HTML resource file versus having to\nupdate code and recompile.\n\nIf you have questions on wxWindows and/or wxPython, please let me know. \nI've been a long time user. Also, if you're willing, I also have some\nconstructive criticism on the code.\n\nSince this is obviously going off topic as it relates to PostgreSQL, it\nprobably wouldn't be appropriate to followup on the mailing list.\n\n\nBest Regards,\n\n\tGreg Copeland\n\n\nOn Wed, 2002-10-30 at 02:19, Dave Page wrote:\n> \n> \n> > -----Original Message-----\n> > From: Greg Copeland [mailto:greg@copelandconsulting.net] \n> > Sent: 30 October 2002 01:08\n> > To: Dave Page\n> > Subject: Re: [HACKERS] Request for supported platforms\n> > \n> > \n> > C++? Really? What GUI toolkit is being used?\n> > \n> > Just curious.\n> \n> wxWindows. The CVS is online if anyone is interested in taking a peek -\n> it's at http://cvs.pgadmin.org/. The code is in the pgadmin3 module.\n> \n> Please bear in mind though, that Mark & I have only been using C++ for\n> about a month (though things are coming along very quickly), so there\n> may be some nasties in our code. Any constructive comments from any of\n> the -hackers would be more than welcome.\n> \n> Regards, Dave.\n> \n> ---------------------------(end of broadcast)---------------------------\n> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org\n\n\n",
"msg_date": "30 Oct 2002 13:33:35 -0600",
"msg_from": "Greg Copeland <greg@CopelandConsulting.Net>",
"msg_from_op": false,
"msg_subject": "Re: pgAdmin III (Was: Request for supported platforms)"
}
] |
[
{
"msg_contents": "Hi,\n\nI just installed 7.3b3 on my server (SuSE linux 8.1 on x86 with gcc 3.2) and \ndiscovered a problem with tabcompletition in pgsql. It doesn't work with \nschema. I created a schema 's' and a table 'tab' in this schema. When typing \n'select * from s.' followed by tab nothing happens. A second tab should list \nme all possible values, but the list ist empty. Without schema it works fine.\n\n\nTommi\n\n\n-- \nDr. Eckhardt + Partner GmbH\nhttp://www.epgmbh.de\n",
"msg_date": "Wed, 30 Oct 2002 10:03:34 +0100",
"msg_from": "Tommi Maekitalo <t.maekitalo@epgmbh.de>",
"msg_from_op": true,
"msg_subject": "tabcompletition and schema"
},
{
"msg_contents": "Tommi Maekitalo wrote:\n> Hi,\n> \n> I just installed 7.3b3 on my server (SuSE linux 8.1 on x86 with gcc 3.2) and \n> discovered a problem with tabcompletition in pgsql. It doesn't work with \n> schema. I created a schema 's' and a table 'tab' in this schema. When typing \n> 'select * from s.' followed by tab nothing happens. A second tab should list \n> me all possible values, but the list ist empty. Without schema it works fine.\n\nAdded to TODO:\n\n\t* Allow psql to do table completion for SELECT * FROM schema_part and\n table completion for SELECT * FROM schema_name.\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, 1 Nov 2002 19:15:12 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: tabcompletition and schema"
}
] |
[
{
"msg_contents": "People,\n\nI posted this to the general list with no response so this is the next \nstep . .\n\nI have a client that I have just changed over to PG 7.2.3 and despite \nthe early teething problems things are going pretty well. There are, \nhowever, a few more difficult problems to do with reporting via Crystal \nReports that could easily be solved by Stored Procedures but these \ndon't exist in PG. The User Definable Functions, which would be able \nto do the job, are not visible to CR.\n\nThere is a white paper from Crystal Decisions about how Oracle Stored \nProcedures are packaged to enable them to be used by CR.\n\nhttp://support.crystaldecisions.com/library/kbase/articles/c2010910.asp\n\nMaybe the issue really lies with psqlODBC?\n\nDoes anyone know anything about this area? If you do, do you think it \nmight be possible to also package PG functions to look like stored \nprocedures to CR? If, so is anyone interested in doing the (paid) \nwork? I am quite happy for the result to be GPLd.\n\nThanks,\n\nPhil.\n-- \nPhilip Rhoades\n\nPricom Pty Limited (ACN 003 252 275)\nGPO Box 3411\nSydney NSW\t2001\nAustralia\nMobile: +61:0411-185-652\nFax: +61:2:8923-5363\nE-mail: pri@chu.com.au\n----- End Forwarded Message -----\n\n-- \nPhilip Rhoades\n\nPricom Pty Limited (ACN 003 252 275)\nGPO Box 3411\nSydney NSW\t2001\nAustralia\nMobile: +61:0411-185-652\nFax: +61:2:8923-5363\nE-mail: pri@chu.com.au\n",
"msg_date": "Wed, 30 Oct 2002 20:20:26 +1100",
"msg_from": "Philip Rhoades <pri@chu.com.au>",
"msg_from_op": true,
"msg_subject": "PG, Crystal Reports, Stored Procedures and User Defined Functions"
},
{
"msg_contents": "\n7.3beta has much more powerful stored procedures. Basically, you can\nput function names in the FROM clause of a SELECT. I recommend you try\nthat and then let us know exactly what is failing with Crystal Reports.\nIt may be just a syntax issue, or something more major.\n\n---------------------------------------------------------------------------\n\nPhilip Rhoades wrote:\n> People,\n> \n> I posted this to the general list with no response so this is the next \n> step . .\n> \n> I have a client that I have just changed over to PG 7.2.3 and despite \n> the early teething problems things are going pretty well. There are, \n> however, a few more difficult problems to do with reporting via Crystal \n> Reports that could easily be solved by Stored Procedures but these \n> don't exist in PG. The User Definable Functions, which would be able \n> to do the job, are not visible to CR.\n> \n> There is a white paper from Crystal Decisions about how Oracle Stored \n> Procedures are packaged to enable them to be used by CR.\n> \n> http://support.crystaldecisions.com/library/kbase/articles/c2010910.asp\n> \n> Maybe the issue really lies with psqlODBC?\n> \n> Does anyone know anything about this area? If you do, do you think it \n> might be possible to also package PG functions to look like stored \n> procedures to CR? If, so is anyone interested in doing the (paid) \n> work? I am quite happy for the result to be GPLd.\n> \n> Thanks,\n> \n> Phil.\n> -- \n> Philip Rhoades\n> \n> Pricom Pty Limited (ACN 003 252 275)\n> GPO Box 3411\n> Sydney NSW\t2001\n> Australia\n> Mobile: +61:0411-185-652\n> Fax: +61:2:8923-5363\n> E-mail: pri@chu.com.au\n> ----- End Forwarded Message -----\n> \n> -- \n> Philip Rhoades\n> \n> Pricom Pty Limited (ACN 003 252 275)\n> GPO Box 3411\n> Sydney NSW\t2001\n> Australia\n> Mobile: +61:0411-185-652\n> Fax: +61:2:8923-5363\n> E-mail: pri@chu.com.au\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, 1 Nov 2002 19:18:59 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: PG, Crystal Reports, Stored Procedures and User Defined"
}
] |
[
{
"msg_contents": "Hello,\n\nOK\nselect 1 as ccc where 1=1\n\nERROR\nselect 1 as ccc where ccc=1\nPostgreSQL said: ERROR: Attribute 'ccc' not found\n\nIs there any way to set conditions on calculated fields values?\n\nBest regards,\nYar\n\n\n",
"msg_date": "Wed, 30 Oct 2002 14:48:19 +0300 (MSK)",
"msg_from": "\"Yaroslav Dmitriev\" <yar@warlock.ru>",
"msg_from_op": true,
"msg_subject": "calculated fields are not seen in the WHERE clause"
},
{
"msg_contents": "Yaroslav Dmitriev kirjutas K, 30.10.2002 kell 13:48:\n> Hello,\n> \n> OK\n> select 1 as ccc where 1=1\n> \n> ERROR\n> select 1 as ccc where ccc=1\n> PostgreSQL said: ERROR: Attribute 'ccc' not found\n> \n> Is there any way to set conditions on calculated fields values?\n\nYou could try using a subquery\n\nselect *\n from\n (select 1 as ccc) sub\n where ccc=1\n\n---------------\nHannu\n\n",
"msg_date": "30 Oct 2002 14:03:25 +0200",
"msg_from": "Hannu Krosing <hannu@tm.ee>",
"msg_from_op": false,
"msg_subject": "Re: calculated fields are not seen in the WHERE clause"
},
{
"msg_contents": "\nAS strings are labels, not columns that can be referenced in a where\nclause:\n\n\ttest=> select lanname as xx from pg_language where xx = 'yy';\n\nHannu is correct that making a FROM entry works.\n\n---------------------------------------------------------------------------\n\nHannu Krosing wrote:\n> Yaroslav Dmitriev kirjutas K, 30.10.2002 kell 13:48:\n> > Hello,\n> > \n> > OK\n> > select 1 as ccc where 1=1\n> > \n> > ERROR\n> > select 1 as ccc where ccc=1\n> > PostgreSQL said: ERROR: Attribute 'ccc' not found\n> > \n> > Is there any way to set conditions on calculated fields values?\n> \n> You could try using a subquery\n> \n> select *\n> from\n> (select 1 as ccc) sub\n> where ccc=1\n> \n> ---------------\n> Hannu\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": "Fri, 1 Nov 2002 19:39:07 -0500 (EST)",
"msg_from": "Bruce Momjian <pgman@candle.pha.pa.us>",
"msg_from_op": false,
"msg_subject": "Re: calculated fields are not seen in the WHERE clause"
}
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.