threads
listlengths 1
2.99k
|
|---|
[
{
"msg_contents": "Oracle 6, there is NULL into table b:\n\nSQL> select * from a where x in (select * from b);\n\n X\n----------\n 2\n\nSQL> select * from a where x not in (select * from b);\n\nno rows selected\n\nSQL> select * from a where not x in (select * from b);\n\nno rows selected\n\nIs 'not X in' the same as 'X not in' ? Currently we have:\n\nvac=> select * from a where not x in (select * from b);\nx\n-\n1\n(1 row)\n\n: subselect clause is \"atomic\" and NOT-s are never pushed into it.\n\nOnce again - what standard says ?\n\nVadim\n",
"msg_date": "Tue, 17 Feb 1998 13:51:05 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Subselects and NOTs"
},
{
"msg_contents": "> \n> Oracle 6, there is NULL into table b:\n> \n> SQL> select * from a where x in (select * from b);\n> \n> X\n> ----------\n> 2\n> \n> SQL> select * from a where x not in (select * from b);\n> \n> no rows selected\n> \n> SQL> select * from a where not x in (select * from b);\n> \n> no rows selected\n> \n> Is 'not X in' the same as 'X not in' ? Currently we have:\n\nI am not sure about this, but I believe 'not X in subselect' is\nevaluated as 'not (x in subselect)' and not as 'X not in subselect'. Am\nI missing something?\n\nThere is also some interesting stuff about comparisons:\n\n( 1,2,NULL) = (3, NULL,4) false\n( 1,2,NULL) < (3, NULL,4) true\n( 1,2,NULL) = (1, NULL,4) unknown\n( 1,2,NULL) > (NULL, 2,4) unknown\n\nThis happens because the comparisons are:\n\n\tleft < right is true of and only if there exists some j such\nthat Lj < Rj is true and for all i < j, and Li = Ri is true\n\nso it seems it compares these things from left to right, trying to make\nthe comparison. For = and <>, is doesn't matter, but for the <, it does\nmatter.\n\nAlso they show:\n\nselect *\nfrom test\nwhere x <> (select y\n\t from test2)\n\nWhen test2 returns no rows, the query returns no rows because the\nsubquery returns a single row of NULL values.\n\nHope this helps. I can give more detail if you want it.\n\n\n> \n> vac=> select * from a where not x in (select * from b);\n> x\n> -\n> 1\n> (1 row)\n> \n> : subselect clause is \"atomic\" and NOT-s are never pushed into it.\n> \n> Once again - what standard says ?\n> \n> Vadim\n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 17 Feb 1998 10:59:32 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> >\n> > Is 'not X in' the same as 'X not in' ? Currently we have:\n> \n> I am not sure about this, but I believe 'not X in subselect' is\n> evaluated as 'not (x in subselect)' and not as 'X not in subselect'. Am\n> I missing something?\n\nEven 'not (x in subselect)' doesn't help in Oracle! This works just\nas 'x not in subselect'.\n\nOk, if we don't know what standard say then could you test this\nin Informix, etc. Don't forget put NULL into table in subselect\nto get 'no rows selected' for 'x not in subselect'...\n\n> \n> Also they show:\n> \n> select *\n> from test\n> where x <> (select y\n> from test2)\n> \n> When test2 returns no rows, the query returns no rows because the\n> subquery returns a single row of NULL values.\n\nThis is exactly how it works in postgres and Oracle.\n\nVadim\n",
"msg_date": "Wed, 18 Feb 1998 20:38:55 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "> > > Is 'not X in' the same as 'X not in' ? Currently we have:\n> > I am not sure about this, but I believe 'not X in subselect' is\n> > evaluated as 'not (x in subselect)' and not as 'X not in subselect'.\n>\n> Even 'not (x in subselect)' doesn't help in Oracle! This works just\n> as 'x not in subselect'.\n> > Also they show:\n> >\n> > select *\n> > from test\n> > where x <> (select y\n> > from test2)\n> >\n> > When test2 returns no rows, the query returns no rows because the\n> > subquery returns a single row of NULL values.\n\nThe SQL92 standard sez:\n\n 1) If NOT is specified in a <boolean test>, then let BP be the\n contained <boolean primary> and let TV be the contained <truth\n value>. The <boolean test> is equivalent to:\n\n ( NOT ( BP IS TV ) )\n\nHowever, \"a NOT IN b\" is not the same as \"NOT (a IN b)\", and my SQL book\npoints out that \"IN\" is shorthand for \"=ANY\" and \"NOT IN\" is shorthand for\n\"<>ALL\". Also, my Date book sez:\n\n In general, an all-or-any condition evaluates to TRUE if and only\nif\n the corresponding comparison condition without the ALL (or ANY,\n respectively) evaluates to TRUE for ALL (or ANY, respectively) of\n\n the rows in the table represented by the table expression.\n (NOTE: If that table is empty, the ALL conditions return TRUE,\nthe\n ANY conditions return FALSE).\n\nSo, it looks to me that\n\n WHERE x IN (SELECT y FROM empty_table)\n\nevaluates to FALSE, and\n\n WHERE x NOT IN (SELECT y FROM empty_table)\n\nevaluates to TRUE.\n\nI looked through my two reference books and my online draft copy of the SQL92\nstandard, and could not find much mention of operator precendence, and even\nless on \"NOT\" precendence. The only mention I could find was the statement\n\n ... Where\n the precedence of operators is determined by the Formats of this\n International Standard or by parentheses, those operators are ef-\n fectively applied in the order specified by that precedence. Where\n the precedence is not determined by the Formats or by parentheses,\n effective evaluation of expressions is generally performed from\n left to right. However, it is implementation-dependent whether ex-\n pressions are actually evaluated left to right, particularly when\n operands or operators might cause conditions to be raised or if\n the results of the expressions can be determined without completely\n evaluating all parts of the expression.\n\nHowever, it wasn't clear to me whether the above statement was referring to\noperators in expressions or to operators in the BNF notation used to define\nthe language.\n\nAlso, in a completely different place:\n\n Operations on numbers are performed according to the normal rules\n of arithmetic, within implementation-defined limits, except as\n provided for in Subclause 6.12, \"<numeric value expression>\".\n\nI can't believe that the standard isn't more explicit somewhere about\noperator precedence, or that it is strictly a left-to-right evaluation.\n\n - Tom\n\n",
"msg_date": "Wed, 18 Feb 1998 15:33:17 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
}
] |
[
{
"msg_contents": "Now that I have run the regression tests for 6.3 on a NetBSD/i386 v1.3\nbox I am wondering what to do with them. Should they be used as a\nsource for bug fixes before the next release? If so, how should they\nbe used? By me only, or should I post them somewhere for others to\nreview?\n\nCheers,\nBrook\n",
"msg_date": "Tue, 17 Feb 1998 00:19:39 -0700 (MST)",
"msg_from": "Brook Milligan <[email protected]>",
"msg_from_op": true,
"msg_subject": "results of regression tests: NetBSD/i386 v1.3"
},
{
"msg_contents": "> Now that I have run the regression tests for 6.3 on a NetBSD/i386 v1.3\n> box I am wondering what to do with them. Should they be used as a\n> source for bug fixes before the next release? If so, how should they\n> be used? By me only, or should I post them somewhere for others to\n> review?\n\nDo they illustrate platform-specific bugs? If not, then we all can use\nour own platforms for these examples.\n\nIf the differences are platform-specific, but expected, then perhaps Marc\nwould like to get these for his \"OS-specific\" regression test\ncomparisons??\n\n - Tom\n\n",
"msg_date": "Tue, 17 Feb 1998 15:14:45 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] results of regression tests: NetBSD/i386 v1.3"
},
{
"msg_contents": " Do they illustrate platform-specific bugs? If not, then we all can use\n our own platforms for these examples.\n\nI'm not sure how I would recognize platform-specific bugs since I only\nhave one platform. In the past I've ignored them, but I just thought\nthis might be the time to point out problems that need attention in\ncase they are something unusual.\n\nDiffs from the expected output yield things like:\n\n- ERROR: pg_atoi: error reading \"100000\": Math result not representable\n+ ERROR: pg_atoi: error reading \"100000\": Result too large\n\n- ERROR: Bad float8 input format '10e-400'\n+ ERROR: Bad float8 input format '-10e-400'\n\n+ ERROR: check_fkeys2_pkey_exist: tuple references non-existing key in pkeys\n+ ERROR: check_fkeys_pkey_exist: tuple references non-existing key in pkeys\n\n- NOTICE: check_pkeys_fkey_cascade: 1 tuple(s) of fkeys2 are deleted\n+ ERROR: check_fkeys2_fkey_restrict: tuple referenced in fkeys\n\n- ERROR: Cannot insert a duplicate key into a unique index\n\n- NOTICE: Non-functional update, only first update is performed\n\nIf these are general diffs, I'll ignore them; otherwise, I'll be glad\nto give more specific information.\n\nCheers,\nBrook\n",
"msg_date": "Tue, 17 Feb 1998 12:56:25 -0700 (MST)",
"msg_from": "Brook Milligan <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] results of regression tests: NetBSD/i386 v1.3"
},
{
"msg_contents": "> Do they illustrate platform-specific bugs? If not, then we all can use\n> our own platforms for these examples.\n>\n> I'm not sure how I would recognize platform-specific bugs since I only\n> have one platform. In the past I've ignored them, but I just thought\n> this might be the time to point out problems that need attention in\n> case they are something unusual.\n>\n> Diffs from the expected output yield things like:\n> <snip>\n\nThese are typical platform-specific differences, and are expected. The last few\nexamples are from the trigger tests, which are going to be revisited before v6.3\nrelease by Vadim (I hope :).\n\n - Tom\n\n",
"msg_date": "Wed, 18 Feb 1998 02:43:32 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] results of regression tests: NetBSD/i386 v1.3"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> > Do they illustrate platform-specific bugs? If not, then we all can use\n> > our own platforms for these examples.\n> >\n> > I'm not sure how I would recognize platform-specific bugs since I only\n> > have one platform. In the past I've ignored them, but I just thought\n> > this might be the time to point out problems that need attention in\n> > case they are something unusual.\n> >\n> > Diffs from the expected output yield things like:\n> > <snip>\n> \n> These are typical platform-specific differences, and are expected. The last few\n> examples are from the trigger tests, which are going to be revisited before v6.3\n> release by Vadim (I hope :).\n\nResults are the same as in 6.2.1 expected. So, I just put 6.2.1 expected\ninto 6.3 expected (with WARN: --> ERROR: translation). Should be ok now.\nLet's know if not.\n\nVadim\n",
"msg_date": "Wed, 18 Feb 1998 20:22:14 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] results of regression tests: NetBSD/i386 v1.3"
}
] |
[
{
"msg_contents": "Gosh, please leave it in it is superb, great, fantastic ... \nIf somebody defines a different behavior as standard in the future,\nwe will need to tell him that he has a gordian knot in his brains :-)\n\nAndreas \n\n> I understand this. And this is how it works currently:\n> \n> select * from tab where (A,B) >= ANY (select X, Y from tab2);\n> \n> means: select tuples where A >= X _and_ B >= Y for some tuple from tab2.\n> ^^^^^\n> 'AND' is used for all Op-s except for '<>' when 'OR' is used.\n> \n> Question is \"should we drop this feature (?) or leave it as is ?\"\n\n",
"msg_date": "Tue, 17 Feb 1998 09:52:50 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Subselects open issue Nr. NEW"
},
{
"msg_contents": "> \n> Gosh, please leave it in it is superb, great, fantastic ... \n> If somebody defines a different behavior as standard in the future,\n> we will need to tell him that he has a gordian knot in his brains :-)\n> \n> Andreas \n> \n> > I understand this. And this is how it works currently:\n> > \n> > select * from tab where (A,B) >= ANY (select X, Y from tab2);\n> > \n> > means: select tuples where A >= X _and_ B >= Y for some tuple from tab2.\n> > ^^^^^\n> > 'AND' is used for all Op-s except for '<>' when 'OR' is used.\n> > \n> > Question is \"should we drop this feature (?) or leave it as is ?\"\n> \n> \n> \n\nI think my recent posting answers this. You have to comare from\nleft-to-right until you find an answer.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 17 Feb 1998 11:02:49 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: Subselects open issue Nr. NEW"
}
] |
[
{
"msg_contents": "Sorry, I take back my gordian knot, I found it in my own brains ;-(\nFor a <=, <, >=, > a lexical ordering would be more intuitive,\nsince that is how a compound index would sort.\n\nso (a, b) <= (c, d) would resolve to:\n(a <= c) or ((a = c) and (b <= c))\n\nWhat happens to !=~ ? Should also be _OR_ ed.\nI guess that leaves us at a point of no go. Take it out ? *tear drops\nfalling*\n\nAndreas\n\n> Vectors cannot be strictly ordered. \"Partial orderings\" are possible.\n\nI think it should say: an order has to be defined (lexical, vector lenght,\narea size ..... )\n\n> \n> Let A be (1, 2)\n> Let B be (4, 7)\n> Let C be (3, 5)\n> Let D be (5, 10)\n> \n> A is smallest; D is largest; how do B and C relate?\n",
"msg_date": "Tue, 17 Feb 1998 10:34:37 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Subselects open issue Nr. NEW"
}
] |
[
{
"msg_contents": "I just checked the changelogs and Debian has integrated lots of upstream\npatches into glibc-2.0.6. So maybe this wasn't fixed before 2.0.6. I'm\ncurrently running 2.0.7pre1 and that one is fine.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tThe Hermit Hacker [SMTP:[email protected]]\n> Sent:\tMonday, February 16, 1998 5:48 PM\n> To:\tFrederick W. Reimer\n> Cc:\tMeskes, Michael; [email protected]\n> Subject:\tRE: [HACKERS] Re: [PORTS] v6.3 release ToDo list and\n> supported p\n> \n> On Mon, 16 Feb 1998, Frederick W. Reimer wrote:\n> \n> > Could some of the other problems be due to gcc-2.8.0, which I'm not \n> \n> \tI'm using v2.8.0 on the Solaris Sparc tests I do here...but\n> nothing else yet...\n> \n",
"msg_date": "Tue, 17 Feb 1998 15:21:48 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Re: [PORTS] v6.3 release ToDo list and supported p"
},
{
"msg_contents": "> I just checked the changelogs and Debian has integrated lots of upstream\n> patches into glibc-2.0.6. So maybe this wasn't fixed before 2.0.6. I'm\n> currently running 2.0.7pre1 and that one is fine.\n\nAndrew, could you add this to the Linux FAQ? The problem statement is:\n\nQ:\nWhy do I get funny rounding results in some date/time arithmetic, such as\n select '4 hours'::timespan;\nreturning '3 hours 59 minutes 60 seconds'?\n\nA:\nYou are running the new glibc2 libraries and have a version earlier than\n2.0.7. It is a math rounding problem in the library. Upgrade your library.\n\nAnd, it seems to me that we should avoid those platform-specific patches\nwhich were discussed to try getting this to work with the buggy versions of\nthese libraries...\n\n - Tom\n\n",
"msg_date": "Tue, 17 Feb 1998 15:29:11 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [PORTS] v6.3 release ToDo list and supported p"
}
] |
[
{
"msg_contents": "I allready took my statement back, about an hour after I said this.\nBut yes, I agree that the left-right approach would be intuitive (same as\ncompound index).\nIn mathematics it is called lexical order, I only thought this would be hard\nto implement.\nAlso there are a lot of operators (all negated Op's that) will want an _OR_\ne.g. !=~, not only <>\n\nSo all not very easy, I'll try to think some more about it.\nAndreas\n> ----------\n> Von: \tBruce Momjian[SMTP:[email protected]]\n> Gesendet: \tDienstag, 17. Februar 1998 17:15\n> An: \tZeugswetter Andreas SARZ\n> Cc: \[email protected]\n> Betreff: \tRe: [HACKERS] Re: Subselects open issue Nr. NEW\n> \n> > \n> > Gosh, please leave it in it is superb, great, fantastic ... \n> > If somebody defines a different behavior as standard in the future,\n> > we will need to tell him that he has a gordian knot in his brains :-)\n> > \n> > Andreas \n> > \n> > > I understand this. And this is how it works currently:\n> > > \n> > > select * from tab where (A,B) >= ANY (select X, Y from tab2);\n> > > \n> > > means: select tuples where A >= X _and_ B >= Y for some tuple from\n> tab2.\n> > > ^^^^^\n> > > 'AND' is used for all Op-s except for '<>' when 'OR' is used.\n> > > \n> > > Question is \"should we drop this feature (?) or leave it as is ?\"\n> > \n> > \n> > \n> \n> I think my recent posting answers this. You have to comare from\n> left-to-right until you find an answer.\n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n> \n",
"msg_date": "Tue, 17 Feb 1998 17:34:19 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: [HACKERS] Re: Subselects open issue Nr. NEW"
},
{
"msg_contents": "> But yes, I agree that the left-right approach would be intuitive (same as\n> compound index).\n> In mathematics it is called lexical order, I only thought this would be hard\n> to implement.\n> Also there are a lot of operators (all negated Op's that) will want an _OR_\n> e.g. !=~, not only <>\n\nWe had started discussing this back when we implemented the general SQL92\nsyntax for multiple left-hand arguments in expressions.\n\nOne possibility was to make the generalization that any operator with \"!\" or\n\"<>\" (others too?) in the operator string would be handled as a negation, and\nall others otherwise. Then, many operators would \"magically\" work correctly or\nintuitively, while a few, perhaps, would not behave correctly. However, imho\nthis is preferable to not allowing _any_ of these, and perhaps we would learn\nover the next few months a way to tighten it up...\n\n - Tom\n\n",
"msg_date": "Tue, 17 Feb 1998 16:48:27 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Re: Subselects open issue Nr. NEW"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n \n> One possibility was to make the generalization that any operator with \"!\" or\n> \"<>\" (others too?) in the operator string would be handled as a negation, and\n> all others otherwise. Then, many operators would \"magically\" work correctly or\n> intuitively, while a few, perhaps, would not behave correctly. However, imho\n> this is preferable to not allowing _any_ of these, and perhaps we would learn\n> over the next few months a way to tighten it up...\n> \n> - Tom\n\nSince we can't tell by the 'name' of the operator what it does (it\nworks often, but not always) without imposing regulations on the\nnaming of operators...\n...add a column to pg_operator to tell what kind of operator it is:\n\nGroup\tExamples\n-------\t--------\nEQ\t=\nNE\t<>\nLT\t<\nGT\t>\nLE\t<=\nGE\t>=\nOTHER\t<-> <?> @-@ ## *~ !== *=~ etc...\n\n(or give the operators other attributes which can be used to find out\n if we should use OR or AND when rewriting '(x,y) OP (z,q)' queries).\n\n... and then just check what group of operator we are dealing with.\nIf it is \"OTHER\", then produce an ERROR or NOT IMPLEMENTED message.\n\n\nThis could be useful in other places too. I think I have seen something\nabout this some weeks or months ago, when somebody had problems finding\nout how to select an appropriate operator, and just had to *assume*\nthat \"=\" was the equality operator in all cases (which might not always\nbe true, and will create hard-to-find bugs sooner or later, since \nmost people will never think of the possibility that someone will use\nsomething else than '=' for equality...)\n\n/* m */\n",
"msg_date": "Wed, 18 Feb 1998 18:20:51 +0100",
"msg_from": "Mattias Kregert <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Re: Subselects open issue Nr. NEW"
}
] |
[
{
"msg_contents": "The snapshot I downloaded today failed compilation:\ngcc -I../../include -I/usr/include/ncurses -I/usr/include/readline \n-Dlinuxalpha -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -Wall \n-Wmissing-prototypes -I../../interfaces/libpq -c pg_dump.c -o pg_dump.o\nIn file included from ../../include/access/strat.h:17,\n from ../../include/utils/rel.h:18,\n from ../../include/catalog/pg_type.h:21,\n from pg_dump.c:65:\n../../include/access/skey.h:20: fmgr.h: No such file or directory\ngmake[2]: *** [pg_dump.o] Error 1\ngmake[2]: Leaving directory `/usr1/home/olly/mypackages/pgsql/src/bin/pg_dump'\n\n\nI made a link from src/backend/fmgr.h to src/include and the compilation\nthen finished.\n\nI note that there is also a backend/utils/fmgr.h which is the same size\nas backend/fmgr.h but not the same file.\n\n=========\n\nInstall failed:\n\nmake[3]: Entering directory `/usr1/home/olly/mypackages/pgsql/src/interfaces/ec\npg/doc'\nmkdir -p /usr/local/pgsql/man\nmkdir /usr/local/pgsql/man/man1\n#-mkdir /usr/local/pgsql/man/man3\n#-mkdir /usr/local/pgsql/man/man5\n#-mkdir /usr/local/pgsql/man/manl\ninstall -m644 *.1* /usr/local/pgsql/man/man1\ninstall: *.1*: No such file or directory\nmake[3]: *** [install-man] Error 1\n\nPrefixing `install' with `-' in the makefile cured this; the directory\ncontains info files but no man pages.\n\n=====================\n\nThe following files still refer to Postgres95 (not always incorrectly):\n./contrib/pginterface/Makefile -- hardcodes /u/postgres95/include \n./contrib/soundex/soundex.sql -- hardcodes\n /usr/local/postgres/postgres95/src/funcs/soundex.so'\n./doc/manual/admin.html\n./doc/manual/advanced.html\n./doc/manual/appenda.html\n./doc/manual/architec.html\n./doc/manual/copy.html\n./doc/manual/extend.html\n./doc/manual/intro.html\n./doc/manual/libpq.html\n./doc/manual/lobj.html\n./doc/manual/pg95user.html\n./doc/manual/query.html\n./doc/manual/refs.html\n./doc/manual/rules.html\n./doc/manual/start.html\n./doc/manual/xaggr.html\n./doc/manual/xfunc.html\n./doc/manual/xindex.html\n./doc/manual/xoper.html\n./doc/manual/xtypes.html\n./doc/userguide.ps\n./src/backend/commands/version.c -- old comment\n./src/backend/libpq/pg_hba.conf.sample\n./src/backend/utils/adt/like.c -- old comment\n./src/bin/pg_dump/README\n./src/bin/pg_dump/pg_dump.c -- old comment\n./src/bin/pg_dump/pg_dump.h -- old comment\n./src/bin/pgtclsh/README\n./src/interfaces/ecpg/TODO\n./src/interfaces/libpgtcl/README\n./src/interfaces/libpq++/README\n./src/interfaces/libpq++/man/libpq++.3\n./src/interfaces/odbc/CHANGES.TXT\n./src/interfaces/odbc/README.TXT\n./src/interfaces/odbc/src/connect.c\n./src/interfaces/odbc/src/custom.h\n./src/interfaces/odbc/src/dll.c\n./src/interfaces/odbc/src/execute.c\n./src/interfaces/odbc/src/globals.c\n./src/interfaces/odbc/src/globals.h\n./src/interfaces/odbc/src/info.c\n./src/interfaces/odbc/src/options.c\n./src/interfaces/odbc/src/pgtypes.c\n./src/interfaces/odbc/src/pgtypes.h\n./src/interfaces/odbc/src/prepare.c\n./src/interfaces/odbc/src/results.c\n./src/interfaces/odbc/src/socket/compat.h\n./src/interfaces/odbc/src/socket/connect.h\n./src/interfaces/odbc/src/socket/connectp.cpp\n./src/interfaces/odbc/src/socket/errclass.cpp\n./src/interfaces/odbc/src/socket/errclass.h\n./src/interfaces/odbc/src/socket/sockio.cpp\n./src/interfaces/odbc/src/socket/sockio.h\n./src/interfaces/odbc/src/socket/wrapper.cpp\n./src/interfaces/odbc/src/socket/wrapper.h\n./src/interfaces/odbc/src/transact.c\n./src/interfaces/python/tutorial/basics.py\n./src/test/regress/sql/create_index.sql\n./src/tutorial/README\n\n\n-- \nOliver Elphick [email protected]\nIsle of Wight http://www.lfix.co.uk/oliver\n\nPGP key from public servers; key ID 32B8FAA1\n\n\n",
"msg_date": "Tue, 17 Feb 1998 22:05:38 +0000",
"msg_from": "\"Oliver Elphick\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Snapshot downloaded 17Feb does not compile"
},
{
"msg_contents": "Oliver Elphick writes:\n> make[3]: Entering directory `/usr1/home/olly/mypackages/pgsql/src/interfaces/ec\n> pg/doc'\n> mkdir -p /usr/local/pgsql/man\n> mkdir /usr/local/pgsql/man/man1\n> #-mkdir /usr/local/pgsql/man/man3\n> #-mkdir /usr/local/pgsql/man/man5\n> #-mkdir /usr/local/pgsql/man/manl\n> install -m644 *.1* /usr/local/pgsql/man/man1\n> install: *.1*: No such file or directory\n> make[3]: *** [install-man] Error 1\n> \n\nIf the info file has been/will be converted we could remove the doc dir\ncompletely.\n\nMichael\n\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Wed, 18 Feb 1998 09:13:53 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Snapshot downloaded 17Feb does not compile"
},
{
"msg_contents": "> > make[3]: Entering directory `/usr1/home/olly/mypackages/pgsql/src/interfaces/ec\n> > pg/doc'\n> > mkdir -p /usr/local/pgsql/man\n> > mkdir /usr/local/pgsql/man/man1\n> > #-mkdir /usr/local/pgsql/man/man3\n> > #-mkdir /usr/local/pgsql/man/man5\n> > #-mkdir /usr/local/pgsql/man/manl\n> > install -m644 *.1* /usr/local/pgsql/man/man1\n> > install: *.1*: No such file or directory\n> > make[3]: *** [install-man] Error 1\n> >\n>\n> If the info file has been/will be converted we could remove the doc dir\n> completely.\n\nThe info file has been translated into the upcoming hardcopy and html docs. Marc,\ncan you blow away the ecpg/doc directory? I've already disabled the make on the doc\narea in stuff I committed last night so things should build OK now.\n\n - Tom\n\n",
"msg_date": "Wed, 18 Feb 1998 15:39:11 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Snapshot downloaded 17Feb does not compile"
}
] |
[
{
"msg_contents": "Hi All,\n\nI'm getting a SEGV error when testing ecpg using the perftest,or\nany other, example program.\n\nI have tracked this down to a call to PQfinish() in ECPGfinish()\nthat occurs before any connection is established.\n\nbool\nECPGfinish()\n{\n if (simple_connection != NULL)\n {\n ECPGlog(\"ECPGfinish: finishing.\\n\");\n PQfinish(simple_connection);\n }\n else\n ECPGlog(\"ECPGfinish: called an extra time.\\n\");\n return true;\n} \n\nA test is made here for simple_connection != NULL.\n\nIf I apply the following patch all is OK.\n\nKeith.\n\n\n*** src/interfaces/ecpg/lib/ecpglib.c.orig Tue Feb 17 22:01:00 1998\n--- src/interfaces/ecpg/lib/ecpglib.c Tue Feb 17 22:01:30 1998\n***************\n*** 24,30 ****\n #include <libpq-fe.h>\n #include <libpq/pqcomm.h>\n\n! static PGconn *simple_connection;\n static int simple_debug = 0;\n static FILE *debugstream = NULL;\n static int committed = true;\n--- 24,30 ----\n #include <libpq-fe.h>\n #include <libpq/pqcomm.h>\n\n! static PGconn *simple_connection = NULL;\n static int simple_debug = 0;\n static FILE *debugstream = NULL;\n static int committed = true;\n \n \n\n",
"msg_date": "Tue, 17 Feb 1998 22:08:28 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "New ecgp code problem."
},
{
"msg_contents": "Keith Parks writes:\n> If I apply the following patch all is OK.\n\nThanks. It's already in CVS it seems. Are you able to execute programs\nagainst v6.3? I cannot even create a user so I only can test my changes my\nmanually inspecting the C file right now.\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Wed, 18 Feb 1998 09:15:05 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: New ecgp code problem."
}
] |
[
{
"msg_contents": "Hi Hackers,\n(yes I,m here --- I \"unsubscribed\" only on the list, not on the\nmajorodomo...) \n\nI've found a very ugly bug in V6.1, V6.2 and (without testing) I think\nit's still there in V6.3.\n\nTo reproduce it use following script:\n====================================================================\n-- SQL script: acl_crash.sql \n-- Call: psql -d template1 -f acl_crash.sql\n--\n-- SQL script to reproduce the situation where PostgreSQL crashes when the \n-- pg_class is accesed which has an relacl entry, but the corresponding\n-- user has been deleted.\n--\n-- drop datebase testdb;\n\ncreate database testdb;\n\\connect testdb\ncreate table testtab (i4 int4, c20 char(20));\ninsert into pg_user (usename,usesysid,usecreatedb,usetrace,usesuper,usecatupd)\n\tvalues ('user1',990,'t','t','t','t');\ninsert into pg_user (usename,usesysid,usecreatedb,usetrace,usesuper,usecatupd)\n\tvalues ('user2',991,'t','t','t','t');\n\n-- Check pg_class - all OK!\nselect relname,relacl from pg_class where relname not like 'pg_%';\n-- now we create some acl-item entries...\ngrant insert,select,update on testtab to user1;\ngrant insert,select,update on testtab to user2;\nrevoke select,update on testtab from user2;\n\n-- Check pg_class - still everything OK!\nselect relname,relacl from pg_class where relname not like 'pg_%';\n\n-- Delete a user which exists in a relacl-list\ndelete from pg_user where usename='user2';\n\n-- Check pg_class - HERE we crash - with a corrupted pg_class???\nselect relname,relacl from pg_class where relname not like 'pg_%';\n===================================================================\nThe situation is fatal from the following point of view:\n\n1. The server side works still after this situation. Maybe the user finds\n the error just after days or weeks of work.\n\n2. There is NO way to access the pg_class without a core dump. Thus it is\n immpossible to 'dump' out the table because the pg_dump reads the\n pg_class always. It seems to be (from the users sight) a currupted\n pg_class table - which can't be repaired. \n\nI saved my live with a patched pg_dump version (which does not touch the \nrelacl values) dumped the database out and created a new database \n... and crashed a little later of course - because I didn't know the\nreason for the creash - once ago.\n--------------------------------\n\nOf course there is a workaround -- IF the problem is known ;-):\nDo a\n revoke all on <table> from <user>\nfor any table which has an acl-item entry (maybe it's simpler to call it\nfor any table). This removes the acl-item from the table entry in\n'pg_class' and the user can be deleted from the 'pg_user' table, without\ntroubles.\n\nBest regards,\n\tGerhard\n\nPS: I will stay on the list - I never thought that someone remembers my\n name :-)\n \n+-----------------+ +--- [email protected] ---+\n| Technische EDV \\ Reithofer / Technical Sofware Developement |\n| A-2136 Laa/Thaya \\ Gerhard / Tel +43-2522/8726 +-------------+\n| Staatsbahnstr. 100 +-------+ Fax +43-2522/87268 |\n+----- http://members.aon.at/tech-edv/Info -------+\n\n",
"msg_date": "Wed, 18 Feb 1998 00:31:31 +0100 (MET)",
"msg_from": "Gerhard Reithofer <[email protected]>",
"msg_from_op": true,
"msg_subject": "Crash and troubles with acl's"
},
{
"msg_contents": "Please try 6.3. I think this was addressed.\n\n> \n> Hi Hackers,\n> (yes I,m here --- I \"unsubscribed\" only on the list, not on the\n> majorodomo...) \n> \n> I've found a very ugly bug in V6.1, V6.2 and (without testing) I think\n> it's still there in V6.3.\n> \n> To reproduce it use following script:\n> ====================================================================\n> -- SQL script: acl_crash.sql \n> -- Call: psql -d template1 -f acl_crash.sql\n> --\n> -- SQL script to reproduce the situation where PostgreSQL crashes when the \n> -- pg_class is accesed which has an relacl entry, but the corresponding\n> -- user has been deleted.\n> --\n> -- drop datebase testdb;\n> \n> create database testdb;\n> \\connect testdb\n> create table testtab (i4 int4, c20 char(20));\n> insert into pg_user (usename,usesysid,usecreatedb,usetrace,usesuper,usecatupd)\n> \tvalues ('user1',990,'t','t','t','t');\n> insert into pg_user (usename,usesysid,usecreatedb,usetrace,usesuper,usecatupd)\n> \tvalues ('user2',991,'t','t','t','t');\n> \n> -- Check pg_class - all OK!\n> select relname,relacl from pg_class where relname not like 'pg_%';\n> -- now we create some acl-item entries...\n> grant insert,select,update on testtab to user1;\n> grant insert,select,update on testtab to user2;\n> revoke select,update on testtab from user2;\n> \n> -- Check pg_class - still everything OK!\n> select relname,relacl from pg_class where relname not like 'pg_%';\n> \n> -- Delete a user which exists in a relacl-list\n> delete from pg_user where usename='user2';\n> \n> -- Check pg_class - HERE we crash - with a corrupted pg_class???\n> select relname,relacl from pg_class where relname not like 'pg_%';\n> ===================================================================\n> The situation is fatal from the following point of view:\n> \n> 1. The server side works still after this situation. Maybe the user finds\n> the error just after days or weeks of work.\n> \n> 2. There is NO way to access the pg_class without a core dump. Thus it is\n> immpossible to 'dump' out the table because the pg_dump reads the\n> pg_class always. It seems to be (from the users sight) a currupted\n> pg_class table - which can't be repaired. \n> \n> I saved my live with a patched pg_dump version (which does not touch the \n> relacl values) dumped the database out and created a new database \n> ... and crashed a little later of course - because I didn't know the\n> reason for the creash - once ago.\n> --------------------------------\n> \n> Of course there is a workaround -- IF the problem is known ;-):\n> Do a\n> revoke all on <table> from <user>\n> for any table which has an acl-item entry (maybe it's simpler to call it\n> for any table). This removes the acl-item from the table entry in\n> 'pg_class' and the user can be deleted from the 'pg_user' table, without\n> troubles.\n> \n> Best regards,\n> \tGerhard\n> \n> PS: I will stay on the list - I never thought that someone remembers my\n> name :-)\n> \n> +-----------------+ +--- [email protected] ---+\n> | Technische EDV \\ Reithofer / Technical Sofware Developement |\n> | A-2136 Laa/Thaya \\ Gerhard / Tel +43-2522/8726 +-------------+\n> | Staatsbahnstr. 100 +-------+ Fax +43-2522/87268 |\n> +----- http://members.aon.at/tech-edv/Info -------+\n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Tue, 17 Feb 1998 19:24:10 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Crash and troubles with acl's"
}
] |
[
{
"msg_contents": "I am trying to get today's snapshot compiled. The trouble I ran into\nis with ecpg which seems to be missing some header files.\n\nAlso, ecpg.c includes getopt.h which doesn't exist on NetBSD systems.\nThe declarations are in unistd.h. See the patch below.\n\nI wasn't sure how to get the configuration scripts to differentiate\nbetween these cases.\n\n--- interfaces/ecpg/preproc/ecpg.c.orig\tTue Feb 17 01:00:39 1998\n+++ interfaces/ecpg/preproc/ecpg.c\tTue Feb 17 14:22:23 1998\n@@ -3,7 +3,7 @@\n /* Placed under the same copyright as PostgresSQL */\n \n #include <stdio.h>\n-#include <getopt.h>\n+#include <unistd.h>\n #include <stdlib.h>\n #include <strings.h>\n \nCheers,\nBrook\n",
"msg_date": "Tue, 17 Feb 1998 16:40:19 -0700 (MST)",
"msg_from": "Brook Milligan <[email protected]>",
"msg_from_op": true,
"msg_subject": "Feb 17 snapshot and NetBSD"
}
] |
[
{
"msg_contents": "I know this is too simple to be useful, but it was the first test case I\ntried for a regression test :)\n\nregression=> SELECT 1 AS one WHERE 1 IN (SELECT 1);\nPQexec() -- Request was sent to backend, but backend closed the channel\nbefore responding.\n This probably means the backend terminated abnormally before or\nwhile processing the request.\n\nShould we disallow parts of this in or near the parser, or can the\nbackend possibly handle it?\n\n - Tom\n\n",
"msg_date": "Wed, 18 Feb 1998 06:05:53 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "stupid subselect core dumps"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> I know this is too simple to be useful, but it was the first test case I\n> tried for a regression test :)\n> \n> regression=> SELECT 1 AS one WHERE 1 IN (SELECT 1);\n> PQexec() -- Request was sent to backend, but backend closed the channel\n> before responding.\n> This probably means the backend terminated abnormally before or\n> while processing the request.\n> \n> Should we disallow parts of this in or near the parser, or can the\n> backend possibly handle it?\n\nFixed. Patch for nodeResult.c follows.\n\nVadim",
"msg_date": "Wed, 18 Feb 1998 13:27:42 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: stupid subselect core dumps"
}
] |
[
{
"msg_contents": ">\n>Does the large object I/O persist across transactions? If so, then storage\nwould\n>need to be outside of the usual context, which is reset after every\ntransaction.\n>Is there a place where the large object context could be freed, but is not\nat\n>the moment?\n>\n> - Tom\n\n\nLarge object I/O does not persist across transactions in my case.\nBut maybe there are applications which assume that it does. So\n\"fixing\" it might break things. How about some compile time flag\nwhich selects between the old behaviour and new behaviour?\nThe old behaviour could be the default.\n\n(The new behaviour would simply avoid fiddling with MemoryContexts at all.)\nMy current workaround is to reconnect to the database after some\nnumber of transactions.\n\nRegards,\n Maurice\n\n",
"msg_date": "Wed, 18 Feb 1998 08:07:36 +0100",
"msg_from": "\"Maurice Gittens\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Memory leaks for large objects"
},
{
"msg_contents": "> \n> Large object I/O does not persist across transactions in my case.\n> But maybe there are applications which assume that it does. So\n> \"fixing\" it might break things. How about some compile time flag\n> which selects between the old behaviour and new behaviour?\n> The old behaviour could be the default.\n> \n> (The new behaviour would simply avoid fiddling with MemoryContexts at all.)\n> My current workaround is to reconnect to the database after some\n> number of transactions.\n\nLarge object have been broken for quite some time. I say remove the\nmemory context stuff and see what breaks. Can't be worse than earlier\nreleases, and if there is a problem, it will show up for us and we can\nissue a patch.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 18 Feb 1998 10:01:16 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Memory leaks for large objects"
},
{
"msg_contents": "On Wed, 18 Feb 1998, Maurice Gittens wrote:\n\n> > >Does the large object I/O persist across transactions? If so, then\n> storage would >need to be outside of the usual context, which is reset\n> after every transaction. >Is there a place where the large object\n> context could be freed, but is not at >the moment? > > - Tom\n\n> Large object I/O does not persist across transactions in my case.\n\nThey do here when I've tried them.\n\n> But maybe there are applications which assume that it does. So\n> \"fixing\" it might break things. How about some compile time flag\n> which selects between the old behaviour and new behaviour?\n> The old behaviour could be the default.\n> \n> (The new behaviour would simply avoid fiddling with MemoryContexts at all.)\n> My current workaround is to reconnect to the database after some\n> number of transactions.\n\nAt the moment, JDBC defaults to not using transactions. As not many\njava apps are using large objects yet (its a new 6.3 feature), it\nshouldn't be difficult to disable the API's if autoCommit is enabled (aka\nno transaction).\n\nThinking about it, the large object examples in the source tree use\ntransactions, so perhaps this is the original behaviour...\n\n-- \nPeter T Mount [email protected] or [email protected]\nMain Homepage: http://www.demon.co.uk/finder\nWork Homepage: http://www.maidstone.gov.uk Work EMail: [email protected]\n\n",
"msg_date": "Wed, 18 Feb 1998 20:40:13 +0000 (GMT)",
"msg_from": "Peter T Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Memory leaks for large objects"
}
] |
[
{
"msg_contents": "I was updating some utilities in config/ the other day and noticed that\nwe have apparently lost a capability for function overloading on\nbuilt-in functions:\n\n CREATE FUNCTION abstime_datetime(int4)\n RETURNS datetime AS '-' LANGUAGE 'internal';\n\nused to allow me to overload the built-in function abstime_datetime() to\nconvert Unix system times stored as integers into a datetime type. I can\nstill define it, but it never gets a good input value. For example,\nafter defining this function I should be able to say:\n\n SELECT abstime_datetime(0);\n abstime_datetime\n ----------------------------\n Tue Apr 23 13:51:12 1974 GMT\n (1 row)\n\nshould give me 'Thu Jan 01 00:00:00 1970 GMT' instead. Input values\nwhich should be passed by reference are also affected.\n\nDoes this ring a bell with anyone? Could something have changed with the\ninclusion of the PL interface (just speculating here)? It would be nice\nto get this capability back, and if not then to remove my contrib code\nwhich no longer works :(\n\n - Tom\n\nAnother possibility is that I am not remembering the right way to do\nthis, _and_ didn't remember correctly _and_ didn't test anything back\nwhen I make the unixdate.sql contrib file. I don't _think_ I would have\nmissed all of these steps, but...\n\n",
"msg_date": "Wed, 18 Feb 1998 07:49:33 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Lost a function overloading capability?"
},
{
"msg_contents": "\nTom wrote:\n>\n> I was updating some utilities in config/ the other day and noticed that\n> we have apparently lost a capability for function overloading on\n> built-in functions:\n>\n> CREATE FUNCTION abstime_datetime(int4)\n> RETURNS datetime AS '-' LANGUAGE 'internal';\n>\n> used to allow me to overload the built-in function abstime_datetime() to\n> convert Unix system times stored as integers into a datetime type. I can\n> still define it, but it never gets a good input value. For example,\n> after defining this function I should be able to say:\n>\n> SELECT abstime_datetime(0);\n> abstime_datetime\n> ----------------------------\n> Tue Apr 23 13:51:12 1974 GMT\n> (1 row)\n>\n> should give me 'Thu Jan 01 00:00:00 1970 GMT' instead. Input values\n> which should be passed by reference are also affected.\n\n With the Feb/13 cvs sources this still worked (haven't\n checked newer one). It returns \"epoch\" (that is Jan 01 ...).\n\n>\n> Does this ring a bell with anyone? Could something have changed with the\n> inclusion of the PL interface (just speculating here)? It would be nice\n> to get this capability back, and if not then to remove my contrib code\n> which no longer works :(\n\n I don't think the PL interface can cause things like this.\n But I wonder how this could happen at all. The recently fixed\n bug in CREATE FUNCTION (proretset true due to TypeName node\n from parser) isn't a candidate for that - it would have\n returned NO data at all.\n\n>\n> - Tom\n>\n> Another possibility is that I am not remembering the right way to do\n> this, _and_ didn't remember correctly _and_ didn't test anything back\n> when I make the unixdate.sql contrib file. I don't _think_ I would have\n> missed all of these steps, but...\n>\n>\n>\n\n\nA little confused, Jan\n\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 19 Feb 1998 11:26:29 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Lost a function overloading capability?"
}
] |
[
{
"msg_contents": "Or we should change the declare/open behaviour in that it ignores the\nembedded SQL declare command and just sends the declare command when it\nfinds an open. But then standard may say that the actual way is correct.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tVadim B. Mikheev [SMTP:[email protected]]\n> Sent:\tWednesday, February 18, 1998 9:56 AM\n> To:\tMichael Meskes\n> Cc:\tThomas G. Lockhart; PostgreSQL Hacker\n> Subject:\tRe: [HACKERS] ecpg stuff\n> \n> Michael Meskes wrote:\n> > \n> > - and check resp. tell me whether the cursor behaviour is correct.\n> Currently\n> > the declare statement is send to the backend via PQexec. The open\n> statement\n> > is ignored and the fetch is executed as fetch via PQexec. I think\n> the data\n> > shouldn't be processed before the cursor is opened. But I do not\n> know\n> > what PostgreSQL does with the declare command.\n> \n> DECLARE: parser + optimizer + ExecutorStart (initializes plan nodes:\n> checks permissions, opens tables & indices). \n> Is OPEN statement in standard ?\n> If yes then we could call ExecutorStart() for the OPEN someday.\n> \n> Vadim\n",
"msg_date": "Wed, 18 Feb 1998 09:52:52 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] ecpg stuff"
},
{
"msg_contents": "Meskes, Michael wrote:\n> \n> Or we should change the declare/open behaviour in that it ignores the\n> embedded SQL declare command and just sends the declare command when it\n> finds an open. But then standard may say that the actual way is correct.\n\nIt seems that OPEN is in standard and this is really bad that we havn't\nOPEN statement: using DECLARE we could plan query once and then \nre-use this plan many times - OPEN, FETCH-es, CLOSE; OPEN, ...\nAlso, we could use queries with parameters (like in SQL-funcs: $1, ...)\nin DECLARE and then assign parameter' values in OPEN and execute\ncursor for different values.\nIsn't this what Oracle allow ?\n\nVadim\n",
"msg_date": "Wed, 18 Feb 1998 16:28:31 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] ecpg stuff"
},
{
"msg_contents": "> \n> Meskes, Michael wrote:\n> > \n> > Or we should change the declare/open behaviour in that it ignores the\n> > embedded SQL declare command and just sends the declare command when it\n> > finds an open. But then standard may say that the actual way is correct.\n> \n> It seems that OPEN is in standard and this is really bad that we havn't\n> OPEN statement: using DECLARE we could plan query once and then \n> re-use this plan many times - OPEN, FETCH-es, CLOSE; OPEN, ...\n> Also, we could use queries with parameters (like in SQL-funcs: $1, ...)\n> in DECLARE and then assign parameter' values in OPEN and execute\n> cursor for different values.\n> Isn't this what Oracle allow ?\n\nYes, I agree we should have it, and allow cursors outside transactions\nwith OPEN/CLOSE behavior.\n\nAdded to TODO.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 18 Feb 1998 10:08:17 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] ecpg stuff"
}
] |
[
{
"msg_contents": "[This email didn't show up on the list when I sent it last monday, so\nhere it is again].\n\nHi,\n\nI found the following bug in sunday's snapshot (I am running Linux\n2.1.85 and gcc 2.8.0):\n\nI created a table with one column and inserted one row:\n\njeroenv=> create table test ( a int4 );\nCREATE\njeroenv=> insert into test values ( 1 );\nINSERT 321708 1\n\nSome checks:\n\njeroenv=> select a from test where (a=1);\na\n-\n1\n(1 row)\n\njeroenv=> select a from test where (a=1) and true;\na\n-\n1\n(1 row)\n\nNow, instead of column 'a' I return the constant 1:\n\njeroenv=> select 1 from test where (a=1);\n?column?\n--------\n 1\n(1 row)\n\njeroenv=> select 1 from test where (a=1) and true;\n?column?\n--------\n(0 rows)\n\nNo rows is not what I would expect.\n\nHowever, with 'or' instead of 'and' returning a const value the query\nworks OK:\n\njeroenv=> select 1 from test where (a=1) or true;\n?column?\n--------\n 1\n(1 row)\n\njeroenv=> select 1 from test where (a=1) or false;\n?column?\n--------\n 1\n(1 row)\n\n\nCheers,\n\nJeroen van Vianen\n",
"msg_date": "Wed, 18 Feb 1998 09:57:21 +0100",
"msg_from": "Jeroen van Vianen <[email protected]>",
"msg_from_op": true,
"msg_subject": "Bug in evaluation of AND"
},
{
"msg_contents": "Jeroen van Vianen wrote:\n> \n> I found the following bug in sunday's snapshot (I am running Linux\n> 2.1.85 and gcc 2.8.0):\n\nPresents in 6.2.1 too...\n\nCaused by bad execution plan:\n\nResult (cost=0.00 size=0 width=0)\n -> Seq Scan on aaa (cost=0.00 size=0 width=0)\n ^^^^^^^^^^^^^^^\nTarget list of SeqScan is empty => ExecProject returns NULL...\nOptimizer should put Constant into Result' childplan target list\nif it's empty. Easy to fix. Anyone ?\n\nVadim\n\n> \n> I created a table with one column and inserted one row:\n> \n> jeroenv=> create table test ( a int4 );\n> CREATE\n> jeroenv=> insert into test values ( 1 );\n> INSERT 321708 1\n> \n> Some checks:\n> \n> jeroenv=> select a from test where (a=1);\n> a\n> -\n> 1\n> (1 row)\n> \n> jeroenv=> select a from test where (a=1) and true;\n> a\n> -\n> 1\n> (1 row)\n> \n> Now, instead of column 'a' I return the constant 1:\n> \n> jeroenv=> select 1 from test where (a=1);\n> ?column?\n> --------\n> 1\n> (1 row)\n> \n> jeroenv=> select 1 from test where (a=1) and true;\n> ?column?\n> --------\n> (0 rows)\n> \n> No rows is not what I would expect.\n",
"msg_date": "Wed, 18 Feb 1998 17:53:11 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Bug in evaluation of AND"
}
] |
[
{
"msg_contents": "I hope that GROUP BY fixed.\nUnfortunately, I can't get any errors under FreeBSD \nwhen using GROUP BY with aggregates. I only got core\nfor something like this: select y, z from tgrp group by y, z.\nTwo different funcs are used for these two cases and so \nplease test new code on your platforms.\nPlease test with different -S XXX to check that all works\nfor both in-memory and on-disk sorts (you'll see pg_psort.xxxx\nfiles in last case).\n\nNew code are for 6.3. This is not patch (sorry, I lost orig version\nof execnodes.h...) and has to be un-tarred in src dir.\n\nIf someone hasn't 6.3 - ok, I'll post patch for 6.2.1 tomorrow.\n\nCVS updated.\n\nVadim",
"msg_date": "Wed, 18 Feb 1998 19:29:24 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "new Group BY code"
},
{
"msg_contents": "> \n> This is a multi-part message in MIME format.\n> --------------0478E3EE3CFC638DB65D6A96\n> Content-Type: text/plain; charset=us-ascii\n> Content-Transfer-Encoding: 7bit\n> \n> I hope that GROUP BY fixed.\n> Unfortunately, I can't get any errors under FreeBSD \n> when using GROUP BY with aggregates. I only got core\n> for something like this: select y, z from tgrp group by y, z.\n> Two different funcs are used for these two cases and so \n> please test new code on your platforms.\n> Please test with different -S XXX to check that all works\n> for both in-memory and on-disk sorts (you'll see pg_psort.xxxx\n> files in last case).\n> \n> New code are for 6.3. This is not patch (sorry, I lost orig version\n> of execnodes.h...) and has to be un-tarred in src dir.\n> \n> If someone hasn't 6.3 - ok, I'll post patch for 6.2.1 tomorrow.\n\n6.3 is only ten days away. I suggest you forget a 6.2.1 patch if it is\nany trouble.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 18 Feb 1998 10:10:25 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] new Group BY code"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> >\n> > If someone hasn't 6.3 - ok, I'll post patch for 6.2.1 tomorrow.\n> \n> 6.3 is only ten days away. I suggest you forget a 6.2.1 patch if it is\n> any trouble.\n\nI'm not sure that I will upgrade my production system to 6.3 in near feature ->\nI need in 6.2.1 patch for myself. And - it's easy.\n\nVadim\n",
"msg_date": "Thu, 19 Feb 1998 08:13:04 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] new Group BY code"
},
{
"msg_contents": "> I hope that GROUP BY fixed.\n> Unfortunately, I can't get any errors under FreeBSD \n> when using GROUP BY with aggregates. I only got core\n> for something like this: select y, z from tgrp group by y, z.\n> Two different funcs are used for these two cases and so \n> please test new code on your platforms.\n> Please test with different -S XXX to check that all works\n> for both in-memory and on-disk sorts (you'll see pg_psort.xxxx\n> files in last case).\n> \n> New code are for 6.3. This is not patch (sorry, I lost orig version\n> of execnodes.h...) and has to be un-tarred in src dir.\n> \n> If someone hasn't 6.3 - ok, I'll post patch for 6.2.1 tomorrow.\n\nVadim,\n\nI downloaded the snapshot of Feb 18 and tried it. I noticed it's much\nharder indeed to reproduce the bug (as compared to 6.2.1p6), but\noccasionaly the bug appeared.\n\nAfter installing your code the bug disappeared completely. I also tried\na couple of group-by queries on my production database, and tried both\nin-memory sorting and on-disk sorting, but everything seems ok now.\n\nSo I guess congratulations are in order.\n\nThanks!\n\nCheers,\nRonald\n",
"msg_date": "Thu, 19 Feb 1998 13:04:04 +0100 (MET)",
"msg_from": "Ronald Baljeu <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] new Group BY code"
}
] |
[
{
"msg_contents": "> Andrew, could you put a new entry into the Linux FAQ suggesting that\n> \n> CFLAGS+= -m486\n> \n> or (for the new compilers that most people are not yet running)\n> \n> CFLAGS+= -mpentium\n> or\n> CFLAGS+= -mpentiumpro\n> \n> be added to Makefile.custom? Otherwise people are going to get 386 code generation\n> for most installations...\n> \n> - Tom\n> \n> \nI've done this, but I guess it's not really Linux specific as I presume it\napplies to BSD as well...\n\nAndrew\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Wed, 18 Feb 1998 12:39:26 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Re: [COMMITTERS] 'pgsql/src/template linux-elf'"
}
] |
[
{
"msg_contents": "I got different results for the second query \n(SELECT * FROM iexit ORDER BY 1, 2;) when using -S 64 (on-disk sorting)\nand -S 8192 (in-memory).\nCould it be caused by this _wrong_ line in pg_operator.h:\n\nDATA(insert OID = 793 ( \"<\" PGUID 0 b t f 602 602 701 ... path_n_lt ...\n ^^^\n this is float8 but path_n_lt() returns bool...\n\nThe same for all path_n_* funcs (may be others too)...\n\nVadim\n",
"msg_date": "Wed, 18 Feb 1998 19:48:48 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "regression: select_views.sql"
},
{
"msg_contents": "> I got different results for the second query\n> (SELECT * FROM iexit ORDER BY 1, 2;) when using -S 64 (on-disk sorting)\n> and -S 8192 (in-memory).\n> Could it be caused by this _wrong_ line in pg_operator.h:\n>\n> DATA(insert OID = 793 ( \"<\" PGUID 0 b t f 602 602 701 ... path_n_lt ...\n> ^^^\n> this is float8 but path_n_lt() returns bool...\n>\n> The same for all path_n_* funcs (may be others too)...\n\nI'll fix this...\n\n",
"msg_date": "Tue, 24 Feb 1998 02:57:49 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] regression: select_views.sql"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> > DATA(insert OID = 793 ( \"<\" PGUID 0 b t f 602 602 701 ... path_n_lt ...\n> > ^^^\n> > this is float8 but path_n_lt() returns bool...\n> >\n> > The same for all path_n_* funcs (may be others too)...\n> \n> I'll fix this...\n\nNice. There are other invalids:\n\nvac=> select oprname, oprcode from pg_operator where oprcanhash and oprname <> '=';\noprname|oprcode \n-------+------------\n~= |intervalsame\n^^^^^^^^^^^^^^^^^^^^ ok, but others are not!\n<> |intervalne \n< |intervallt \n> |intervalgt \n<= |intervalle \n>= |intervalge \n(6 rows)\n\n(BTW, oprcanhash can be used to let know does operator mean EQUALity...\nI'd like to use this for subselects)\n\nVadim\n",
"msg_date": "Tue, 24 Feb 1998 10:39:57 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] regression: select_views.sql"
},
{
"msg_contents": "Vadim B. Mikheev wrote:\n\n> Thomas G. Lockhart wrote:\n> >\n> > > DATA(insert OID = 793 ( \"<\" PGUID 0 b t f 602 602 701 ... path_n_lt ...\n> > > ^^^\n> > > this is float8 but path_n_lt() returns bool...\n> > >\n> > > The same for all path_n_* funcs (may be others too)...\n> >\n> > I'll fix this...\n>\n> Nice. There are other invalids:\n>\n> vac=> select oprname, oprcode from pg_operator where oprcanhash and oprname <> '=';\n> oprname|oprcode\n> -------+------------\n> ~= |intervalsame\n> ^^^^^^^^^^^^^^^^^^^^ ok, but others are not!\n> <> |intervalne\n> < |intervallt\n> > |intervalgt\n> <= |intervalle\n> >= |intervalge\n> (6 rows)\n\nHmm. Those look like they support the \"time interval\" type, which I haven't been updating\nbecause it is based on abstime; I'm more interested in the wider-range/more-precise\ndatetime-based stuff. I would think about rewriting the time interval routines, but\nassume it was used somewhere in the backend to support time travel. Don't know if it is\nstill in there somewhere...\n\n - Tom\n\n",
"msg_date": "Tue, 24 Feb 1998 04:16:49 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] regression: select_views.sql"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> >\n> > vac=> select oprname, oprcode from pg_operator where oprcanhash and oprname <> '=';\n> > oprname|oprcode\n> > -------+------------\n> > ~= |intervalsame\n> > ^^^^^^^^^^^^^^^^^^^^ ok, but others are not!\n> > <> |intervalne\n> > < |intervallt\n> > > |intervalgt\n> > <= |intervalle\n> > >= |intervalge\n> > (6 rows)\n> \n> Hmm. Those look like they support the \"time interval\" type, which I haven't been updating\n> because it is based on abstime; I'm more interested in the wider-range/more-precise\n> datetime-based stuff. I would think about rewriting the time interval routines, but\n> assume it was used somewhere in the backend to support time travel. Don't know if it is\n> still in there somewhere...\n\nI just say that all these <>, <, >, >=, <= must have FALSE in oprcanhash.\nHaving TRUE for them is bug.\n\nVadim\n",
"msg_date": "Tue, 24 Feb 1998 11:26:09 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] regression: select_views.sql"
}
] |
[
{
"msg_contents": "> > I just checked the changelogs and Debian has integrated lots of upstream\n> > patches into glibc-2.0.6. So maybe this wasn't fixed before 2.0.6. I'm\n> > currently running 2.0.7pre1 and that one is fine.\n> \n> Andrew, could you add this to the Linux FAQ? The problem statement is:\n> \n> Q:\n> Why do I get funny rounding results in some date/time arithmetic, such as\n> select '4 hours'::timespan;\n> returning '3 hours 59 minutes 60 seconds'?\n> \n\nHave done....\n\nAndrew\n\n\n\np.s. BRUCE! Could I remind you to make sure the latest FAQs get into the\ndistribution :-)\n\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Wed, 18 Feb 1998 13:03:31 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Re: [PORTS] v6.3 release ToDo list and supported p"
},
{
"msg_contents": "> \n> > > I just checked the changelogs and Debian has integrated lots of upstream\n> > > patches into glibc-2.0.6. So maybe this wasn't fixed before 2.0.6. I'm\n> > > currently running 2.0.7pre1 and that one is fine.\n> > \n> > Andrew, could you add this to the Linux FAQ? The problem statement is:\n> > \n> > Q:\n> > Why do I get funny rounding results in some date/time arithmetic, such as\n> > select '4 hours'::timespan;\n> > returning '3 hours 59 minutes 60 seconds'?\n> > \n> \n> Have done....\n> \n> Andrew\n> \n> \n> \n> p.s. BRUCE! Could I remind you to make sure the latest FAQs get into the\n> distribution :-)\n> \n\nYep, I need that reminder. I am terrible with that.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 18 Feb 1998 10:12:36 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Re: [PORTS] v6.3 release ToDo list and supported p"
}
] |
[
{
"msg_contents": "> Here is my current list for porting status for the v6.3 release. I may\n> have missed at least a few reports, e.g hpux, irix??\n> \n> _ irix5\n> _ irix6/MIPS\n\nWell, I reported (and supplied patches for) the problems with the irix5 \nport and will download and install another version once I hear these \nhave been applied.\n\nI haven't seen any comments as to whether the strange regression test\noutputs are significant...\n\nI think it's difficult to say we fully support irix6 because of the\nbug in Irix ld. However, the Irix-specific FAQ does detail the\nworkaround for this, so I guess it's semi-supported.\n\nAndrew\n\n\np.s. Bruce, the links from the main FAQ on the Web site to the machine-\nspecific ones are still broken - the 'FAQ' should be in lower case.\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Wed, 18 Feb 1998 14:51:35 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Valid ports for v6.3"
},
{
"msg_contents": "> \n> \n> p.s. Bruce, the links from the main FAQ on the Web site to the machine-\n> specific ones are still broken - the 'FAQ' should be in lower case.\n> \n\nOK, I have updated my FAQ to use lowercase names for the files. That\nseems to fix it.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 18 Feb 1998 10:32:18 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Valid ports for v6.3"
}
] |
[
{
"msg_contents": "> \n> Here is my current list for porting status for the v6.3 release. I may\n> have missed at least a few reports, e.g hpux, irix??\n> \n> _ irix5\n> _ irix6/MIPS\n\nOoops, I just posted saying I hadn't heard if the patch had been applied.\n\nNow that I've read the pg-ports mailing list I see it has <BLUSH>.\n\n\n\nWas the change also made to .../src/template/.similar ?\n\n\nI won't have a chance to do another download and compile until Monday, but\nI can't see that there will be any major problems with Irix now.\n\n\nAndrew\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Wed, 18 Feb 1998 14:57:07 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Valid ports for v6.3"
}
] |
[
{
"msg_contents": "Hi, all.\n\nI'm forwarding this message I posted to psql-ports, because I've seen no\nanswer on it, and the D-day approaches...\n\n---------- Forwarded message ----------\nDate: Mon, 16 Feb 1998 12:48:49 +0100 (MET)\nFrom: \"Pedro J. Lobo\" <[email protected]>\nTo: PostgreSQL ports mailing list <[email protected]>\nSubject: Feb. 15 snapshot doesn't compile on alpha / Digital Unix\n\nHi, folks.\n\nThe good news: I compiled 6.3 beta on Digital Unix 3.2c using the DEC C\ncompiler.\n\nThe bad news: It doesn't work :-(\n\nIt compiles and installs well, but initdb fails. This is the output\nmessage:\n\n------------------------------------------------------------\ninitdb: using /usr/local/pgsql.beta/lib/local1_template1.bki.source as\ninput to create the template database.\ninitdb: using /usr/local/pgsql.beta/lib/global1.bki.source as input to\ncreate the global classes.\ninitdb: using /usr/local/pgsql.beta/lib/pg_hba.conf.sample as the\nhost-based authentication control file.\n\nWe are initializing the database system with username pgbeta (uid=104).\nThis user will own all the files and must also own the server process.\n\nCreating Postgres database system directory /usr/local/pgsql.beta/data\n\nCreating Postgres database system directory /usr/local/pgsql.beta/data/base\n\ninitdb: creating template database in /usr/local/pgsql.beta/data/base/template1\nRunning: postgres -boot -C -F -D/usr/local/pgsql.beta/data -Q template1\nERROR: BuildFuncTupleDesc: function mkoidname(opaque, opaque) does not exist\nERROR: BuildFuncTupleDesc: function mkoidname(opaque, opaque) does not exist\nlongjmp or siglongjmp function used outside of saved context\n/usr/local/pgsql.beta/bin/initdb: 420 Abort - core dumped\ninitdb: could not create template database\ninitdb: cleaning up by wiping out /usr/local/pgsql.beta/data/base/template1\n-----------------------------------------------------------\n\nThis is the GDB stack trace from the core file:\n\n\n----------------------------------------------------------\nGDB 4.16 (alpha-dec-osf3.2), Copyright 1996 Free Software Foundation, Inc...\nCore was generated by `postgres'.\nProgram terminated with signal 6, IOT/Abort trap.\nReading symbols from /usr/shlib/libm.so...done.\nReading symbols from /usr/shlib/libcurses.so...done.\nReading symbols from /usr/shlib/libc.so...done.\nReading symbols from /usr/lib/nls/loc//es_ES.ISO8859-1...done.\n#0 0x3ff8010cf28 in kill ()\n(gdb) where\n#0 0x3ff8010cf28 in kill ()\n#1 0x3ff801218f4 in raise ()\n#2 0x3ff8010fd44 in abort ()\n#3 0x3ff80102e20 in _longjmp ()\n#4 0x3ff80105778 in siglongjmp ()\n#5 0x120144d24 in handle_warn (postgres_signal_arg=1) at postgres.c:736\n#6 <signal handler called>\n#7 0x3ff8010cf28 in kill ()\n#8 0x12018deb8 in elog (lev=-1,\n fmt=0x14002f8d8 \"%s: function %s(%s) does not exist\", ...=0x140011428)\n at elog.c:180\n#9 0x12010ddd4 in func_error (caller=0x140011428 \"BuildFuncTupleDesc\",\n funcname=0x11ffff260 \"mkoidname\", nargs=2, argtypes=0x11ffff23c)\n at parse_func.c:1307\n#10 0x120074114 in BuildFuncTupleDesc (funcInfo=0x11ffff238) at index.c:298\n#11 0x120075998 in index_create (heapRelationName=0x1401bba50 \"pg_attribute\",\n indexRelationName=0x1401dc730 \"pg_attribute_relid_attnam_index\",\n funcInfo=0x11ffff238, attributeList=0x0, accessMethodObjectId=403,\n numatts=1, attNums=0x1401dc9c0, classObjectId=0x1401dc9f0,\n parameterCount=0, parameter=0x0, predicate=0x0, islossy=0 '\\000',\n unique=0 '\\000') at index.c:1114\n#12 0x1200860a4 in DefineIndex (heapRelationName=0x1401bba50 \"pg_attribute\",\n indexRelationName=0x1401dc730 \"pg_attribute_relid_attnam_index\",\n accessMethodName=0x1401db830 \"btree\", attributeList=0x1401e0a60,\n parameterList=0x0, unique=0 '\\000', predicate=0x0, rangetable=0x0)\n at defind.c:190\n#13 0x12006b580 in Int_yyparse () at /usr/local/share/bison.simple:234\n#14 0x12006e80c in BootstrapMain (argc=6, argv=0x11ffffcb0) at bootstrap.c:446\n#15 0x1200c51b0 in main (argc=7, argv=0x11ffffca8) at main.c:94\n--------------------------------------------------------------\n\n\nI have no knowledge of postgres internals, but I have experience in C/C++\nprogramming (and debugging) and can help if someone gives me a clue. The\nsame hardware/OS/compiler combination works well with 6.2.1p6.\n\nIn case it matters, this is the configure command that I've used:\n\n./configure --prefix=/usr/local/pgsql.beta --enable-locale\n--with-pgport=5440 --enable-cassert --with-compiler=cc\n\nI have my production database (6.2.1p6) running on the same machine on the\nstandard port (5432), but I don't think it has any adverse effect. In\nfact, postgres -boot doesn't do any socket communications, does it? The\nproduction database is stored completely on another tree\n(/usr/local/pgsql instead of /usr/local/pgsql.beta).\n\nBTW, I had to patch 'configure' to handle --with-compiler correctly. This\nis an old problem, and I will report it more extensively in another\nmessage.\n\n\n-------------------------------------------------------------------\nPedro José Lobo Perea Tel: +34 1 336 78 19\nCentro de Cálculo Fax: +34 1 331 92 29\nEUIT Telecomunicación - UPM e-mail: [email protected]\n\n\n",
"msg_date": "Wed, 18 Feb 1998 16:46:32 +0100 (MET)",
"msg_from": "\"Pedro J. Lobo\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Feb. 15 snapshot doesn't compile on alpha / Digital Unix (fwd)"
},
{
"msg_contents": "OK, where are we with Alpha now? Dec Unix and Alpha Linux users, what\nis your status. Is anyone geting through initdb, and if so, how?\n\n\n> Hi, all.\n> \n> I'm forwarding this message I posted to psql-ports, because I've seen no\n> answer on it, and the D-day approaches...\n> \n> ---------- Forwarded message ----------\n> Date: Mon, 16 Feb 1998 12:48:49 +0100 (MET)\n> From: \"Pedro J. Lobo\" <[email protected]>\n> To: PostgreSQL ports mailing list <[email protected]>\n> Subject: Feb. 15 snapshot doesn't compile on alpha / Digital Unix\n> \n> Hi, folks.\n> \n> The good news: I compiled 6.3 beta on Digital Unix 3.2c using the DEC C\n> compiler.\n> \n> The bad news: It doesn't work :-(\n> \n> It compiles and installs well, but initdb fails. This is the output\n> message:\n> \n> ------------------------------------------------------------\n> initdb: using /usr/local/pgsql.beta/lib/local1_template1.bki.source as\n> input to create the template database.\n> initdb: using /usr/local/pgsql.beta/lib/global1.bki.source as input to\n> create the global classes.\n> initdb: using /usr/local/pgsql.beta/lib/pg_hba.conf.sample as the\n> host-based authentication control file.\n> \n> We are initializing the database system with username pgbeta (uid=104).\n> This user will own all the files and must also own the server process.\n> \n> Creating Postgres database system directory /usr/local/pgsql.beta/data\n> \n> Creating Postgres database system directory /usr/local/pgsql.beta/data/base\n> \n> initdb: creating template database in /usr/local/pgsql.beta/data/base/template1\n> Running: postgres -boot -C -F -D/usr/local/pgsql.beta/data -Q template1\n> ERROR: BuildFuncTupleDesc: function mkoidname(opaque, opaque) does not exist\n> ERROR: BuildFuncTupleDesc: function mkoidname(opaque, opaque) does not exist\n> longjmp or siglongjmp function used outside of saved context\n> /usr/local/pgsql.beta/bin/initdb: 420 Abort - core dumped\n> initdb: could not create template database\n> initdb: cleaning up by wiping out /usr/local/pgsql.beta/data/base/template1\n> -----------------------------------------------------------\n> \n> This is the GDB stack trace from the core file:\n> \n> \n> ----------------------------------------------------------\n> GDB 4.16 (alpha-dec-osf3.2), Copyright 1996 Free Software Foundation, Inc...\n> Core was generated by `postgres'.\n> Program terminated with signal 6, IOT/Abort trap.\n> Reading symbols from /usr/shlib/libm.so...done.\n> Reading symbols from /usr/shlib/libcurses.so...done.\n> Reading symbols from /usr/shlib/libc.so...done.\n> Reading symbols from /usr/lib/nls/loc//es_ES.ISO8859-1...done.\n> #0 0x3ff8010cf28 in kill ()\n> (gdb) where\n> #0 0x3ff8010cf28 in kill ()\n> #1 0x3ff801218f4 in raise ()\n> #2 0x3ff8010fd44 in abort ()\n> #3 0x3ff80102e20 in _longjmp ()\n> #4 0x3ff80105778 in siglongjmp ()\n> #5 0x120144d24 in handle_warn (postgres_signal_arg=1) at postgres.c:736\n> #6 <signal handler called>\n> #7 0x3ff8010cf28 in kill ()\n> #8 0x12018deb8 in elog (lev=-1,\n> fmt=0x14002f8d8 \"%s: function %s(%s) does not exist\", ...=0x140011428)\n> at elog.c:180\n> #9 0x12010ddd4 in func_error (caller=0x140011428 \"BuildFuncTupleDesc\",\n> funcname=0x11ffff260 \"mkoidname\", nargs=2, argtypes=0x11ffff23c)\n> at parse_func.c:1307\n> #10 0x120074114 in BuildFuncTupleDesc (funcInfo=0x11ffff238) at index.c:298\n> #11 0x120075998 in index_create (heapRelationName=0x1401bba50 \"pg_attribute\",\n> indexRelationName=0x1401dc730 \"pg_attribute_relid_attnam_index\",\n> funcInfo=0x11ffff238, attributeList=0x0, accessMethodObjectId=403,\n> numatts=1, attNums=0x1401dc9c0, classObjectId=0x1401dc9f0,\n> parameterCount=0, parameter=0x0, predicate=0x0, islossy=0 '\\000',\n> unique=0 '\\000') at index.c:1114\n> #12 0x1200860a4 in DefineIndex (heapRelationName=0x1401bba50 \"pg_attribute\",\n> indexRelationName=0x1401dc730 \"pg_attribute_relid_attnam_index\",\n> accessMethodName=0x1401db830 \"btree\", attributeList=0x1401e0a60,\n> parameterList=0x0, unique=0 '\\000', predicate=0x0, rangetable=0x0)\n> at defind.c:190\n> #13 0x12006b580 in Int_yyparse () at /usr/local/share/bison.simple:234\n> #14 0x12006e80c in BootstrapMain (argc=6, argv=0x11ffffcb0) at bootstrap.c:446\n> #15 0x1200c51b0 in main (argc=7, argv=0x11ffffca8) at main.c:94\n> --------------------------------------------------------------\n> \n> \n> I have no knowledge of postgres internals, but I have experience in C/C++\n> programming (and debugging) and can help if someone gives me a clue. The\n> same hardware/OS/compiler combination works well with 6.2.1p6.\n> \n> In case it matters, this is the configure command that I've used:\n> \n> ./configure --prefix=/usr/local/pgsql.beta --enable-locale\n> --with-pgport=5440 --enable-cassert --with-compiler=cc\n> \n> I have my production database (6.2.1p6) running on the same machine on the\n> standard port (5432), but I don't think it has any adverse effect. In\n> fact, postgres -boot doesn't do any socket communications, does it? The\n> production database is stored completely on another tree\n> (/usr/local/pgsql instead of /usr/local/pgsql.beta).\n> \n> BTW, I had to patch 'configure' to handle --with-compiler correctly. This\n> is an old problem, and I will report it more extensively in another\n> message.\n> \n> \n> -------------------------------------------------------------------\n> Pedro Jos��� Lobo Perea Tel: +34 1 336 78 19\n> Centro de C���lculo Fax: +34 1 331 92 29\n> EUIT Telecomunicaci���n - UPM e-mail: [email protected]\n> \n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 18 Feb 1998 10:57:35 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Feb. 15 snapshot doesn't compile on alpha / Digital\n\tUnix (fwd)"
}
] |
[
{
"msg_contents": "Here are the items. Can I remove any of them?\n\nLooking over the list, the only major one I see is that pg_user is not\nworld-readable, and with Todd not responding, I am thinking I may need\nto do some ripout. Right now, \\d does not work for non-super-users. \nCan you imagine the problem reports we will get on that?\n\nMany of these items can be moved to the TODO list after the release.\n \n---------------------------------------------------------------------------\n\nORDER BY NULLs problem?\npg_user not world-readable, no initdb password option(Todd Brandys)\nmake psql \\d optionally user pg_user(Bruce)\noptimizer memory exhaustion with many OR's\nmax tuple size patch(Darren)\nIs GROUP BY duplicates fixed?\nDo we have a self-join optimizer performance problem?\nGet interfaces fixed for new protocol\nProfiling improvements?\nDo we have a problem with GROUP BY without ORDER BY?\nProblem with tables >2Gb\nDo we want to add timestamps to elog messages?\nScanKeyData missing initializations\nCan we improve vacuum locking issues?\nAlpha/64-bit issues, mkoidname() problem\n'Can not write block blind' error on createdb and regression collision\nViews on aggregates fail\nlarge objects memory exhaustion\nsubselect issues, NOT IN (empty query)\nCommit sgml document sources(Thomas)\nCommit html documents(Thomas)\nCommit postscript documents(Thomas)\nFix isinf.c irix port problem reported by Andrew(Marc)\nCheck triggers regression test(Vadim)\nCheck select_views regression test\nTest ecpg new struct features(Michael)\nTest big-endian/little-endian operation(Tatsuo?)?\nLinux library issues?\n\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 18 Feb 1998 11:04:31 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "[HACKERS] Open 6.3 issues (fwd)"
},
{
"msg_contents": "> Here are the items. Can I remove any of them?\n>\n> Get interfaces fixed for new protocol\n\nWe should call these out individually...\n\n> Do we want to add timestamps to elog messages?\n\nOne of Massimo's patches did this. Should someone apply it? Won't be me\nthough. Also, will affect many regression tests so someone would have to go\nthrough those...\n\n> 'Can not write block blind' error on createdb and regression collision\n\nThis is a show-stopper...\n\n> Commit sgml document sources(Thomas)\n> Commit html documents(Thomas)\n> Commit postscript documents(Thomas)\n\nWill start on final doc formatting tonight; have a small section to write to\nsummarize the docs. Also, will need to redraw the figures since\nbitmap->Applix->bitmap looks bad, and resizing makes it worse.\n\n> Check select_views regression test\n\nimo, all regression discrepencies are show-stoppers. This must be done, and if\nnoone else picks it up I will try to get to it. All it needs is for someone to\ngo through that test and verify that the new results are valid equivalent\nreorderings of the old results (it's possible that the behavior of the\ngeometry operators has changed for the better, so that the new results are\nbetter too). The byte count of the files is the same, and on first glance the\nresults are equivalent. But, someone should check more carefully.\n\n> Linux library issues?\n\nIt appears that glibc2.0.7 will fix the problem. Andrew has updated the Linux\nFAQ.\n\n",
"msg_date": "Wed, 18 Feb 1998 16:53:17 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Open 6.3 issues (fwd)"
},
{
"msg_contents": "On Wed, 18 Feb 1998, Bruce Momjian wrote:\n\n> Here are the items. Can I remove any of them?\n> \n> Looking over the list, the only major one I see is that pg_user is not\n> world-readable, and with Todd not responding, I am thinking I may need\n> to do some ripout. Right now, \\d does not work for non-super-users. \n> Can you imagine the problem reports we will get on that?\n\nOkay, I just grabbed our \"resident Oracle guru\"s time to see how Oracle\ndoes this, and I'm CCng him in on this...\n\nBasically, from what we can tell, there is one large table that contains\nthe user information, including passwords. If we go in as 'the dba'\nhimself (same as our super user), we can view the table in its\nentirety...if we go in as \"joe blow user\", you can only look at a \"view\"\nof the table.\n\n\"select * from all_users;\" <- no passwords\n\"select * from dba_users;\" <- includes passwords\n\nSo, I think that Todd has the right idea, but where we are lacking is the\nability to have pg_users \"closed\" to non-su users to view all its data,\nbut have it \"open\" through a system view...\n\nSo, what we need is the ability to have a view created, as part of the\nsystem itself, that does:\n\n\"select usename,usesysid,<etc, minus passwd,valuntil>\n from pg_user;\"\n\nSo that that information is available, but the password isn't...\n\n\n\n",
"msg_date": "Wed, 18 Feb 1998 12:48:10 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Open 6.3 issues (fwd)"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> \n> > 'Can not write block blind' error on createdb and regression collision\n> \n> This is a show-stopper...\n\nI'll take a look...\n\n> \n> > Check select_views regression test\n\nImho, it's better to get rid of second query from select_views.sql:\n\nSELECT * FROM iexit ORDER BY 1, 2;\n ^\n2-nd attr is of path type, '<' for path-es compares # of path' points,\nthere are many 2,3,-points pathes there, ordering is valid but we\ncan't get the same results on different platforms/-S XXX.\n\nAnother way is replace the query with:\n\nSELECT name, #thepath FROM iexit ORDER BY 1, 2;\n ^^^^^^^^ path_npoints()\n\nVadim\n",
"msg_date": "Thu, 19 Feb 1998 11:45:00 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Open 6.3 issues (fwd)"
}
] |
[
{
"msg_contents": "This is what I did:\n\tcreate table a (a int, a1 char(8));\n\tcreate table b (b int);\n\tinsert into a values (1, 'one');\n\tinsert into a values (NULL, 'null');\n\nand this is what I got from Informix:\n> select * from a where a not in (select * from b);\n a a1\n 1 one\n null\n2 row(s) retrieved.\n> select * from a where not (a in (select * from b));\n a a1\n 1 one\n null\n2 row(s) retrieved.\n> select * from a where not a in (select * from b);\n a a1\n 1 one\n null\n2 row(s) retrieved.\n> select * from a where a<>(select * from b);\n a a1\nNo rows found.\n> select * from a where a=(select * from b);\n a a1\nNo rows found.\n\nAndreas\n",
"msg_date": "Wed, 18 Feb 1998 18:32:40 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
}
] |
[
{
"msg_contents": "\n\n---------- Forwarded message ----------\nDate: Wed, 18 Feb 1998 12:48:10 -0500 (EST)\nFrom: The Hermit Hacker <[email protected]>\nTo: Bruce Momjian <[email protected]>\nCc: PostgreSQL-development <[email protected]>\nSubject: Re: [HACKERS] Open 6.3 issues (fwd)\n\nOn Wed, 18 Feb 1998, Bruce Momjian wrote:\n\n> Here are the items. Can I remove any of them?\n> \n> Looking over the list, the only major one I see is that pg_user is not\n> world-readable, and with Todd not responding, I am thinking I may need\n> to do some ripout. Right now, \\d does not work for non-super-users. \n> Can you imagine the problem reports we will get on that?\n\nOkay, I just grabbed our \"resident Oracle guru\"s time to see how Oracle\ndoes this, and I'm CCng him in on this...\n\nBasically, from what we can tell, there is one large table that contains\nthe user information, including passwords. If we go in as 'the dba'\nhimself (same as our super user), we can view the table in its\nentirety...if we go in as \"joe blow user\", you can only look at a \"view\"\nof the table.\n\n\"select * from all_users;\" <- no passwords\n\"select * from dba_users;\" <- includes passwords\n\nSo, I think that Todd has the right idea, but where we are lacking is the\nability to have pg_users \"closed\" to non-su users to view all its data,\nbut have it \"open\" through a system view...\n\nSo, what we need is the ability to have a view created, as part of the\nsystem itself, that does:\n\n\"select usename,usesysid,<etc, minus passwd,valuntil>\n from pg_user;\"\n\nSo that that information is available, but the password isn't...\n\n\n\n\n",
"msg_date": "Wed, 18 Feb 1998 12:49:16 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "rights on pg_user (WAs: Re: [HACKERS] Open 6.3 issues (fwd))"
}
] |
[
{
"msg_contents": "Hi,\n\nThings were working upto a point when I was playing yesterday.\n\nThe perftest seemed to run OK\nThe test2 example ran but did not display the varchar (name) field.\n\nI'm afraid I've not got very far debugging.\n\nKeith.\n\n\nMichael Meskes <[email protected]>\n> Keith Parks writes:\n> > If I apply the following patch all is OK.\n> \n> Thanks. It's already in CVS it seems. Are you able to execute programs\n> against v6.3? I cannot even create a user so I only can test my changes my\n> manually inspecting the C file right now.\n> \n\n",
"msg_date": "Wed, 18 Feb 1998 19:27:54 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: New ecgp code problem."
}
] |
[
{
"msg_contents": ">\n>Large object have been broken for quite some time. I say remove the\n>memory context stuff and see what breaks. Can't be worse than earlier\n>releases, and if there is a problem, it will show up for us and we can\n>issue a patch.\n>\n>--\n\n\nI insured that all memory allocations in be-fsstubs.c used the\ncurrent memorycontext for their allocations.\nThe system encounters errors when opening large objects which\nwere just created. Message like: \"ERROR cannot open xinv<number>\".\nThis happens even though all large_object operations are performed\nin a transaction.\n\nI'm now wondering wether in the approach above the files associated\nwith the large object will ever be freed (Or will de virtual file descriptor\nstuff\nhandle this?).\n\nMight it be so that because large objects and are implemented using\nrelations/indexes that information about these must persist until these\nare properly closed by the postgres system?\n\nHow about not changing anything except adding a lo_garbage_collect function,\nwhich frees the MemoryContext used by large objects and does any other\nwork needed? (Like closes indexes/relations?).\n\nThanks,\nMaurice\n\n\n",
"msg_date": "Wed, 18 Feb 1998 20:29:54 +0100",
"msg_from": "\"Maurice Gittens\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Memory leaks for large objects"
},
{
"msg_contents": "Added to TODO list.\n\n> \n> >\n> >Large object have been broken for quite some time. I say remove the\n> >memory context stuff and see what breaks. Can't be worse than earlier\n> >releases, and if there is a problem, it will show up for us and we can\n> >issue a patch.\n> >\n> >--\n> \n> \n> I insured that all memory allocations in be-fsstubs.c used the\n> current memorycontext for their allocations.\n> The system encounters errors when opening large objects which\n> were just created. Message like: \"ERROR cannot open xinv<number>\".\n> This happens even though all large_object operations are performed\n> in a transaction.\n> \n> I'm now wondering wether in the approach above the files associated\n> with the large object will ever be freed (Or will de virtual file descriptor\n> stuff\n> handle this?).\n> \n> Might it be so that because large objects and are implemented using\n> relations/indexes that information about these must persist until these\n> are properly closed by the postgres system?\n> \n> How about not changing anything except adding a lo_garbage_collect function,\n> which frees the MemoryContext used by large objects and does any other\n> work needed? (Like closes indexes/relations?).\n> \n> Thanks,\n> Maurice\n> \n> \n> \n\n\n-- \nBruce Momjian | 830 Blythe Avenue\[email protected] | Drexel Hill, Pennsylvania 19026\n + If your life is a hard drive, | (610) 353-9879(w)\n + Christ can be your backup. | (610) 853-3000(h)\n",
"msg_date": "Mon, 16 Mar 1998 00:10:39 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Memory leaks for large objects"
}
] |
[
{
"msg_contents": "I was reading in my \"Unix Internals: New Frontiers\" book by Vahalia\nyesterday, and he was talking about multi-cpu kernels, and he was\ntalking about read/write locks, and he said:\n\n\n---------------------------------------------------------------------------\n\nWhen the last active reader releases its shared lock, it must wake up a\nsingle waiting writer.\n\nWhen a writer releases a lock, it must choose whether to wake up another\nwriter or the other readers, assuming both readers and writers are\nwaiting. If writers are given preference, the readers could starve\nindefinitely under heavy contention. The preferred solution is to wake\nup all waiting readers when releasing an exclusive lock. If there are\nno waiting readers, wake up a single waiting writer.\n\nThe scheme can lead to writer starvation. If there is a constant stream\nof readers, they will keep the resource read-locked, and the writer\nwill never acquire the lock. To avoid this situation, a lockShared()\nrequest must block if there is any waiting writer, even though the\nresource is currently only read-locked. Such a solution, under heavy\ncontention, will alternate access between individual writer and batches\nof readers.\n\n\n---------------------------------------------------------------------------\n\nThe attached patch will do something similar to what he is suggesting. \nMy change is to keep a single waiting writer at the front of the queue,\neven if the current lock holder is a writer. Any readers arriving\nduring the current lock holder, or when the new writer gets the lock\nwill be bunched together and be given the shared lock after the writer\nreleases the lock. It just doesn't seem fair for the readers to jump in\nfront of a single waiting writer. If multiple writers are waiting, then\nthey will be behind the readers, except in cases where a late-arriving\nreader can't share the readlock, and sits behind the first writer, so in\nthat case it would be alternating.\n\nWhat do people think of this patch? Is his way better, to always put\nthe readers at the front if the current lock holder is a writer?\n\nThe patch basically reverses the queue order, putting the readers before\nthe writers, except that now, a single writer can sit at the front of\nthe queue. This structure also makes it easy for us to check to see if\nwe should allow a new reader to share a readlock. We can quickly check\nthe front of the queue to see if there is a writer waiting.\n\n---------------------------------------------------------------------------\n\n\n*** ./backend/storage/lmgr/proc.c.orig\tWed Feb 18 13:48:11 1998\n--- ./backend/storage/lmgr/proc.c\tWed Feb 18 14:15:09 1998\n***************\n*** 451,465 ****\n \t\t int prio,\n \t\t LOCK *lock)\n {\n! \tint\t\t\ti;\n \tPROC\t *proc;\n \tstruct itimerval timeval,\n \t\t\t\tdummy;\n \n \tproc = (PROC *) MAKE_PTR(waitQueue->links.prev);\n! \tfor (i = 0; i < waitQueue->size; i++)\n \t{\n! \t\tif (proc->prio >= prio)\n \t\t\tproc = (PROC *) MAKE_PTR(proc->links.prev);\n \t\telse\n \t\t\tbreak;\n--- 451,490 ----\n \t\t int prio,\n \t\t LOCK *lock)\n {\n! \tint\t\t\ti = 0;\n \tPROC\t *proc;\n \tstruct itimerval timeval,\n \t\t\t\tdummy;\n \n \tproc = (PROC *) MAKE_PTR(waitQueue->links.prev);\n! \t/*\n! \t *\tIf the first entry in the waitQueue has a greater priority than\n! \t *\twe have, we must be a reader, and he must be a writer, and we\n! \t *\tmust be here because the current holder is a writer or a\n! \t *\treader but we don't share shared locks if a writer is waiting.\n! \t *\tWe put ourselves after the first writer. What this hopefully\n! \t *\tdoes it to bunch up all the waiting readers first, unless there\n! \t *\tis a waiting writer, in which case all the waiting readers are\n! \t *\tafter the first waiting writer. This way, we alternate between\n! \t *\twriters and bunches of readers and don't starve anyone. So the\n! \t *\tqueue becomes:\n! \t *\n! \t *\t[optional writer][reader][reader][reader][writer][writer][writer]\n! \t *\n! \t *\tIn a full queue, we would have a writer holding a lock, then another\n! \t *\twriter gets the lock, then a bunch of readers, it repeats.\n! \t *\tThis may seem strange, but because new readers share the lock even\n! \t *\tif older writers are waiting, this seems only fair. - bjm\n! \t */\n! \tif (i < waitQueue->size)\n! \t\tif (proc->prio > prio)\n! \t\t{\n! \t\t\tproc = (PROC *) MAKE_PTR(proc->links.prev);\n! \t\t\ti++;\n! \t\t}\n! \tfor (; i < waitQueue->size; i++)\n \t{\n! \t\tif (proc->prio <= prio)\n \t\t\tproc = (PROC *) MAKE_PTR(proc->links.prev);\n \t\telse\n \t\t\tbreak;\n***************\n*** 596,603 ****\n \n \t\t/*\n \t\t * ProcWakeup removes proc from the lock waiting process queue and\n! \t\t * returns the next proc in chain.\tIf a writer just dropped its\n! \t\t * lock and there are several waiting readers, wake them all up.\n \t\t */\n \t\tproc = ProcWakeup(proc, NO_ERROR);\n \n--- 621,627 ----\n \n \t\t/*\n \t\t * ProcWakeup removes proc from the lock waiting process queue and\n! \t\t * returns the next proc in chain.\n \t\t */\n \t\tproc = ProcWakeup(proc, NO_ERROR);\n \n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 18 Feb 1998 14:30:56 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "New locking code"
},
{
"msg_contents": "Bruce Momjian wrote:\n\n[snip]\n> What do people think of this patch? Is his way better, to always put\n> the readers at the front if the current lock holder is a writer?\n\nThis way seems to make sense to me because it provides two guarantees:\n\n1) Neither readers or writers can ever be starved when trying toa\naccess an object.\n\n2) The system implements a FIFO queue. I.E. A reader can't \"cut in\nline\" in front of a waiting writer just because the current access\nmode is shared. The operations that have been waiting the longest get\nserviced first.\n\nOcie Mitchell\n\n",
"msg_date": "Wed, 18 Feb 1998 12:31:15 -0800 (PST)",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] New locking code"
},
{
"msg_contents": "> 2) The system implements a FIFO queue. I.E. A reader can't \"cut in\n> line\" in front of a waiting writer just because the current access\n> mode is shared. The operations that have been waiting the longest get\n> serviced first.\n\nI thinking about this FIFO comment, here is a better fix. It implements\na FIFO for the lock wait queue, except it groups readers togther in the\nqueue. The major change is that I now loop past as many writers that\nmay be at the head of the queue, then put the readers in the queue. If\nthere are already readers in the queue after the writers, I add them to\nthe end of the existing readers. The source code comments are changed,\nand provide a more detailed explaination.\n\nI am planning to put this into 6.3, to fix a possible reader starvation\nproblem with the new code.\n\n---------------------------------------------------------------------------\n\n*** ./backend/storage/lmgr/proc.c.orig\tTue Jan 27 21:29:29 1998\n--- ./backend/storage/lmgr/proc.c\tWed Feb 18 20:10:04 1998\n***************\n*** 451,469 ****\n \t\t int prio,\n \t\t LOCK *lock)\n {\n! \tint\t\t\ti;\n \tPROC\t *proc;\n \tstruct itimerval timeval,\n \t\t\t\tdummy;\n \n \tproc = (PROC *) MAKE_PTR(waitQueue->links.prev);\n! \tfor (i = 0; i < waitQueue->size; i++)\n! \t{\n! \t\tif (proc->prio >= prio)\n! \t\t\tproc = (PROC *) MAKE_PTR(proc->links.prev);\n! \t\telse\n! \t\t\tbreak;\n! \t}\n \n \tMyProc->prio = prio;\n \tMyProc->token = token;\n--- 451,492 ----\n \t\t int prio,\n \t\t LOCK *lock)\n {\n! \tint\t\t\ti = 0;\n \tPROC\t *proc;\n \tstruct itimerval timeval,\n \t\t\t\tdummy;\n \n+ \t/*\n+ \t *\tIf the first entries in the waitQueue have a greater priority than\n+ \t *\twe have, we must be a reader, and they must be a writers, and we\n+ \t *\tmust be here because the current holder is a writer or a\n+ \t *\treader but we don't share shared locks if a writer is waiting.\n+ \t *\tWe put ourselves after the writers. This way, we have a FIFO, but\n+ \t *\tkeep the readers together to give them decent priority, and no one\n+ \t *\tstarves. Because we group all readers together, a non-empty queue\n+ \t *\tonly has a few possible configurations:\n+ \t *\n+ \t *\t[readers]\n+ \t *\t[writers]\n+ \t *\t[readers][writers]\n+ \t *\t[writers][readers]\n+ \t *\t[writers][readers][writers]\n+ \t *\n+ \t *\tIn a full queue, we would have a reader holding a lock, then a\n+ \t *\twriter gets the lock, then a bunch of readers, made up of readers\n+ \t *\twho could not share the first readlock because a writer was waiting,\n+ \t *\tand new readers arriving while the writer had the lock.\n+ \t *\n+ \t */\n \tproc = (PROC *) MAKE_PTR(waitQueue->links.prev);\n! \n! \t/* If we are a reader, and they are writers, skip past them */\n! \twhile (i++ < waitQueue->size && proc->prio > prio)\n! \t\tproc = (PROC *) MAKE_PTR(proc->links.prev);\n! \n! \t/* The rest of the queue is FIFO, with readers first, writers last */\n! \twhile (i++ < waitQueue->size && proc->prio <= prio)\n! \t\tproc = (PROC *) MAKE_PTR(proc->links.prev);\n \n \tMyProc->prio = prio;\n \tMyProc->token = token;\n***************\n*** 596,603 ****\n \n \t\t/*\n \t\t * ProcWakeup removes proc from the lock waiting process queue and\n! \t\t * returns the next proc in chain.\tIf a writer just dropped its\n! \t\t * lock and there are several waiting readers, wake them all up.\n \t\t */\n \t\tproc = ProcWakeup(proc, NO_ERROR);\n \n--- 619,625 ----\n \n \t\t/*\n \t\t * ProcWakeup removes proc from the lock waiting process queue and\n! \t\t * returns the next proc in chain.\n \t\t */\n \t\tproc = ProcWakeup(proc, NO_ERROR);\n \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Wed, 18 Feb 1998 20:13:59 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] New locking code"
}
] |
[
{
"msg_contents": "Thomas G. Lockhart <[email protected]> wrote:\n\n> > But yes, I agree that the left-right approach would be intuitive (same as\n> > compound index).\n> > In mathematics it is called lexical order, I only thought this would be hard\n> > to implement.\n> > Also there are a lot of operators (all negated Op's that) will want an _OR_\n> > e.g. !=~, not only <>\n\n> We had started discussing this back when we implemented the general SQL92\n> syntax for multiple left-hand arguments in expressions.\n\n> One possibility was to make the generalization that any operator with \"!\" or\n> \"<>\" (others too?) in the operator string would be handled as a negation, and\n> all others otherwise. Then, many operators would \"magically\" work correctly or\n> intuitively, while a few, perhaps, would not behave correctly. However, imho\n> this is preferable to not allowing _any_ of these, and perhaps we would learn\n> over the next few months a way to tighten it up...\n\n> - Tom\n\nHi,\n\nI think we can handle all operators in one general way,\nsomething like this:\n\n for (i = 0; i < n-1; i++) /* (n-1) iterations here. */\n {\n if (a[i] is NULL || b[i] is NULL)\n return UNKNOWN; /* three-value boolean logic. */\n else if (a[i] == b[i])\n continue;\n else\n return (a[i] op b[i]);\n }\n return (a[n-1] op b[n-1]); /* the last n-th iteration. */\n\nWorks well for all standard ==, !=, <, <=, >, >=.\n\nThe problem appears if someone would define an operator like\n\"approximately equals\". An operator of this sort should\nreplace the operator == in the above pseudo-code.\nAlso someone could have operator \"approximately less than\"\nwhich uses \"approximately equals\" as operator == in the above\nalong with ordinary operators < and ==.\n\nSo the problem seems can be solved if we assume that both\noperator == and operator op in the above are parameters.\n\nThe proposed syntax for tuple comparison predicate could be\n(a1, a2,...) [USING EQUALS OPERATOR [op2]] op (subselect)\nwith op2 defaults to:\n op, if \"USING EQUALS OPERATOR\" is specified,\n \"==\" otherwise.\n\nAny comments ?\n\nTo have much more cool object-oriented spirit we could allow\noperator overloading for tuples. Let's leave it until later...\n\n--\nYurik\n\n",
"msg_date": "18 Feb 1998 21:26:21 GMT",
"msg_from": "[email protected] (Yurik V. Nazaroff)",
"msg_from_op": true,
"msg_subject": "Re: AW: [HACKERS] Re: Subselects open issue Nr. NEW"
}
] |
[
{
"msg_contents": "Hi,\n\nI was having a play with subselects and found this, not very clever,\nnested subselect didn't work as I expected.\n\nAny Ideas?\n\nI also noticed that the subselect regression tests include only\nint and float columns in the tests.\n\nKeith.\n\nNOTICE: QUERY PLAN:\n\nIndex Scan on t1 (cost=2.77 size=16 width=32)\n InitPlan\n -> Unique (cost=188.57 size=0 width=0)\n InitPlan\n -> Aggregate (cost=188.57 size=0 width=0)\n -> Seq Scan on t3 (cost=188.57 size=4199 width=4)\n -> Sort (cost=188.57 size=0 width=0)\n -> Seq Scan on t2 (cost=188.57 size=64 width=12)\n\nEXPLAIN \n\n\ndisks=> select * from tracks t1 where t1.artist = (\ndisks-> select distinct t2.artist from tracks t2 where t2.trackno = (\ndisks-> select max(t3.trackno) from tracks t3\ndisks-> )\ndisks-> );\ndiskid|trackno|artist |song\n------+-------+-------------+-------------\n 51| 1|Elvis Presley|Hound Dog\n 65| 1|Elvis Presley|She's Not You\n(2 rows)\n\ndisks=> select max(t3.trackno) from tracks t3;\nmax\n---\n 26\n(1 row)\n\ndisks=> select distinct t2.artist from tracks t2 where t2.trackno = 26;\nartist\n-------------\nElvis Presley\n(1 row)\n\ndisks=> select * from tracks t1 where t1.artist = 'Elvis Presley';\ndiskid|trackno|artist |song\n------+-------+-------------+-----------------------------------\n 51| 1|Elvis Presley|Hound Dog\n 65| 1|Elvis Presley|She's Not You\n 65| 2|Elvis Presley|Return To Sender\n 65| 3|Elvis Presley|(You're The) Devil In Disguise\n 65| 4|Elvis Presley|Crying In The Chapel\n 65| 5|Elvis Presley|Love Letters\n 65| 6|Elvis Presley|If I Can Dream\n 65| 7|Elvis Presley|In The Ghetto\n 65| 8|Elvis Presley|Suspicious Minds\n 65| 9|Elvis Presley|Don't Cry Daddy\n 65| 10|Elvis Presley|The Wonder Of You\n 65| 11|Elvis Presley|I Just Can't Help Believin'\n 65| 12|Elvis Presley|An American Trilogy\n 65| 13|Elvis Presley|Burning Love\n 65| 14|Elvis Presley|Always On My Mind\n 65| 15|Elvis Presley|My Boy\n 65| 16|Elvis Presley|Suspicion\n 65| 17|Elvis Presley|Moody Blue\n 65| 18|Elvis Presley|Way Down\n 65| 19|Elvis Presley|It's Only Love\n 66| 1|Elvis Presley|Heartbreak Hotel\n 66| 2|Elvis Presley|Blue Suede Shoes\n 66| 3|Elvis Presley|Hound Dog\n 66| 4|Elvis Presley|Love Me Tender\n 66| 5|Elvis Presley|Too Much\n 66| 6|Elvis Presley|All Shook Up\n 66| 7|Elvis Presley|Teddy Bear\n 66| 8|Elvis Presley|Paralysed\n 66| 9|Elvis Presley|Party\n 66| 10|Elvis Presley|Jailhouse Rock\n 66| 11|Elvis Presley|Don't\n 66| 12|Elvis Presley|Wear My Ring Around Your Neck\n 66| 13|Elvis Presley|Hard Headed Woman\n 66| 14|Elvis Presley|King Creole\n 66| 15|Elvis Presley|One Night\n 66| 16|Elvis Presley|A Fool Such As I\n 66| 17|Elvis Presley|A Big Hunk O' Love\n 66| 18|Elvis Presley|Stuck On You\n 66| 19|Elvis Presley|The Girl Of My Best Friend\n 66| 20|Elvis Presley|It's Now Or Never\n 66| 21|Elvis Presley|Are You Lonesome Tonight?\n 66| 22|Elvis Presley|Wooden Heart\n 66| 23|Elvis Presley|Surrender\n 66| 24|Elvis Presley|(Marie's The Name) His Latest Flame\n 66| 25|Elvis Presley|Can't Help Falling In Love\n 66| 26|Elvis Presley|Good Luck Charm\n 231| 11|Elvis Presley|Can't Help Falling In Love\n 248| 1|Elvis Presley|Always On My Mind\n 248| 2|Elvis Presley|I Just Can't Help Believin'\n 248| 3|Elvis Presley|Suspicious Minds\n 248| 4|Elvis Presley|Can't Help Falling In Love\n 248| 5|Elvis Presley|Are You Lonesome Tonight?\n 248| 6|Elvis Presley|The Girl Of My Best Friend\n 248| 7|Elvis Presley|It's Now Or Never\n 248| 8|Elvis Presley|Love Me Tender\n 248| 9|Elvis Presley|Fever\n 248| 10|Elvis Presley|Surrender\n 248| 11|Elvis Presley|Love Me\n 248| 12|Elvis Presley|Loving You\n 248| 13|Elvis Presley|She's Not You\n 248| 14|Elvis Presley|A Fool Such As I\n 248| 15|Elvis Presley|Suspicion\n 248| 16|Elvis Presley|Love Letters\n 248| 17|Elvis Presley|And I Love You So\n 248| 18|Elvis Presley|Help Me Make It Through The Night\n 248| 19|Elvis Presley|You Don't Have To Say You Love Me\n 248| 20|Elvis Presley|Kentucky Rain\n 248| 21|Elvis Presley|Gentle On My Mind\n 248| 22|Elvis Presley|Let It Be Me\n 248| 23|Elvis Presley|Spanish Eyes\n 248| 24|Elvis Presley|It's Only Love\n 248| 25|Elvis Presley|The Wonder Of You\n 248| 26|Elvis Presley|Bridge Over Troubled Water\n(73 rows)\n\ndisks=> \n\n",
"msg_date": "Wed, 18 Feb 1998 22:13:56 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "Unexpected subselect result."
},
{
"msg_contents": "Keith Parks wrote:\n> \n> Hi,\n> \n> I was having a play with subselects and found this, not very clever,\n> nested subselect didn't work as I expected.\n> \n> Any Ideas?\n> \n> I also noticed that the subselect regression tests include only\n> int and float columns in the tests.\n\nThanks, Keith!\n\nVadim",
"msg_date": "Thu, 19 Feb 1998 09:28:16 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Unexpected subselect result."
},
{
"msg_contents": "Sorry, I decided to copy subselect' tuple unconditionally...\n\nVadim",
"msg_date": "Thu, 19 Feb 1998 09:55:52 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Unexpected subselect result."
},
{
"msg_contents": "> I also noticed that the subselect regression tests include only\n> int and float columns in the tests.\n\nYeah, if you or someone wants to augment it that would be great. Change and\nadd as much as you want; I did it in a rush...\n\n - Tom\n\n",
"msg_date": "Thu, 19 Feb 1998 05:00:08 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Unexpected subselect result."
}
] |
[
{
"msg_contents": "Bruce Momjian wrote:\n\n>Date: Wed, 18 Feb 1998 10:57:35 -0500 (EST)\n>From: Bruce Momjian <[email protected]>\n>Subject: Re: [HACKERS] Feb. 15 snapshot doesn't compile on alpha / Digital Unix (fwd)\n>\n>OK, where are we with Alpha now? Dec Unix and Alpha Linux users, what\n>is your status. Is anyone geting through initdb, and if so, how?\n>\n>\n>> Hi, all.\n>> \n>> I'm forwarding this message I posted to psql-ports, because I've seen no\n>> answer on it, and the D-day approaches...\n>> \n>> ---------- Forwarded message ----------\n>> Date: Mon, 16 Feb 1998 12:48:49 +0100 (MET)\n>> From: \"Pedro J. Lobo\" <[email protected]>\n>> To: PostgreSQL ports mailing list <[email protected]>\n>> Subject: Feb. 15 snapshot doesn't compile on alpha / Digital Unix\n>> \n>> Hi, folks.\n>> \n>> The good news: I compiled 6.3 beta on Digital Unix 3.2c using the DEC C\n>> compiler.\n>> \n>> The bad news: It doesn't work :-(\n>> \n[...]\n> \n>> \n>\n>\n>- -- \n>Bruce Momjian\n>[email protected]\n>\n\nStatus with V6.3b on Alpha/Unix is that present snapshot 18,19-Feb\nwon't install with some error relating to ecpglib.h. I've posted this\nto ports-pgsql, but the problem has persisted.\n\nI'm still struggling to find out what is wrong with initdb on the\n17-Feb snapshot and I have found that the mkoidname error originates\nfrom the line\n\ndeclare index pg_attribute_relid_attnam_index on pg_attribute using btree(mkoidname(attrelid, attname) oidname_ops)\n\nand that postgres is failing when BuildFuncTupleDesc fails to resolve\nthe tuple with call to SearchSysCacheTuple.\n\nThe lookup methods used in the resolution is not transparently obvious\nto me and I have not yet managed to spot any clear source of the\nproblem.\n\nI am about to try checking the operation of the bootstrap logic at the\ntime of insertion - at the template line\n\ninsert OID = 949 ( mkoidname 1200 11 f t f 2 f 911 \"26 19\" 100 0 0 100 foo bar)\n\nto see if some related value is being corrupted at this stage. Since\nI don't have a lot of experience in the way postgres handles such\nthings, any suggestions will be welcome. Since I have not really used\ngdb previously this is, in itself, a learning experience.\n\nRegards\n\nPeter A. Stockwell\n\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 15:52:58 +1300 (NZDT)",
"msg_from": "Peter Stockwell <[email protected]>",
"msg_from_op": true,
"msg_subject": "Status of Alpha/Unix"
}
] |
[
{
"msg_contents": "In getting 6.3 working on a NetBSD/i386 box I have made up a few\npatches that correct some areas of the\ncode/documentation/configuration. They follow, each preceded by a\ndescription. I hope they are useful; they allowed me to do a complete\ncompilation using the --with-tcl configuration option, something I\ncouldn't originally.\n\nCheers,\nBrook\n\n===========================================================================\n===========================================================================\n\nNOTE: All patches originate at root_of_postgresql_tarball/src.\n\n===========================================================================\n===========================================================================\nDocumentation on at least these configure options is missing from INSTALL.\n\n--- ../INSTALL.orig\tMon Feb 2 01:00:13 1998\n+++ ../INSTALL\tWed Feb 18 16:42:54 1998\n@@ -267,6 +267,15 @@\n listens for incoming connections on. The\n default for this is port 5432.\n \n+ --with-defaults Use default responses to several queries during\n+ configuration.\n+\n+ --with-tcl Enables programs requiring Tcl/Tk and X11,\n+ including pgtclsh and libpgtcl.\n+\n+ --with-perl Enables the perl interface. Note that this\n+ requires an installed version of postgreSQL.\n+\n As an example, here is the configure script I use on a Sparc\n Solaris 2.5 system with /opt/postgres being the install base.\n\n\n===========================================================================\n===========================================================================\nI found that the configuration scripts were not looking for X11/tcl/tk\nsupport and the default built into Makefile.global.in was wrong for my\nsystem. As a result using the --with-tcl bombed when it came to\npgtclsh and libpgtcl. The following three patches change the\nconfiguration system to search for X11/tcl/tk in various spots. I\nhope all the relevant possibilities are included; it's easy to add new\nspots. They also change the pgtclsh/Makefile to use the right\nvariables set in Makefile.global.\n\nNote that a new configure script will need to be made with autoconf if\nthis patch is installed.\n\n--- configure.in.orig\tTue Feb 17 01:00:36 1998\n+++ configure.in\tWed Feb 18 18:18:09 1998\n@@ -229,6 +229,7 @@\n \tUSE_TCL=\n fi\n export USE_TCL\n+USE_X=$USE_TCL\n \n dnl We exclude perl support unless we override it with --with-perl\n if test \"X$with_perl\" = \"Xyes\"\n@@ -514,5 +515,131 @@\n AC_MSG_CHECKING(setting DEF_PGPORT)\n AC_DEFINE_UNQUOTED(DEF_PGPORT, \"${DEF_PGPORT}\")\n AC_MSG_RESULT($DEF_PGPORT)\n+\n+dnl Check for X libraries\n+\n+if test \"$USE_X\" = true; then\n+\n+ice_save_LIBS=\"$LIBS\"\n+ice_save_CFLAGS=\"$CFLAGS\"\n+ice_save_CPPFLAGS=\"$CPPFLAGS\"\n+ice_save_LDFLAGS=\"$LDFLAGS\"\n+\n+AC_PATH_XTRA\n+\n+LIBS=\"$LIBS $X_EXTRA_LIBS\"\n+CFLAGS=\"$CFLAGS $X_CFLAGS\"\n+CPPFLAGS=\"$CPPFLAGS $X_CFLAGS\"\n+LDFLAGS=\"$LDFLAGS $X_LIBS\"\n+\n+dnl Check for X library\n+\n+X11_LIBS=\"\"\n+AC_CHECK_LIB(X11, XOpenDisplay, X11_LIBS=\"-lX11\",,${X_PRE_LIBS})\n+if test \"$X11_LIBS\" = \"\"; then\n+dnl Not having X may be fatal. Let the user fix this.\n+AC_MSG_WARN([The X11 library '-lX11' could not be found.\n+ Please use the configure options '--x-includes=DIR'\n+ and '--x-libraries=DIR' to specify the X location.\n+ See the file 'config.log' for further diagnostics.])\n+fi\n+AC_SUBST(X_LIBS)\n+AC_SUBST(X11_LIBS)\n+AC_SUBST(X_PRE_LIBS)\n+\n+LIBS=\"$ice_save_LIBS\"\n+CFLAGS=\"$ice_save_CFLAGS\"\n+CPPFLAGS=\"$ice_save_CPPFLAGS\"\n+LDFLAGS=\"$ice_save_LDFLAGS\"\n+fi\n+\n+dnl Check for location of Tcl support\n+dnl Disable Tcl support if not found\n+\n+dnl Check for tcl.h\n+if test \"$USE_TCL\" = \"true\"\n+then\n+TCL_INCDIR=no\n+AC_CHECK_HEADER(tcl.h, TCL_INCDIR=)\n+for f in /usr/include /usr/include/tcl8.0 /usr/local/include /usr/local/include/tcl8.0; do\n+if test \"$TCL_INCDIR\" = \"no\"; then\n+AC_CHECK_HEADER($f/tcl.h, TCL_INCDIR=$f)\n+fi\n+done\n+if test \"$TCL_INCDIR\" = \"no\"; then\n+AC_MSG_WARN(tcl support disabled; tcl.h missing)\n+USE_TCL=\n+fi\n+AC_SUBST(TCL_INCDIR)\n+fi\n+\n+dnl Check for Tcl archive\n+if test \"$USE_TCL\" = \"true\"\n+then\n+TCL_LIB=\n+for f in tcl8.0 tcl80; do\n+if test -z \"$TCL_LIB\"; then\n+AC_CHECK_LIB($f, main, TCL_LIB=$f)\n+fi\n+done\n+if test -z \"$TCL_LIB\"; then\n+AC_MSG_WARN(tcl support disabled; Tcl library missing)\n+USE_TCL=\n+else\n+TCL_LIB=-l$TCL_LIB\n+fi\n+AC_SUBST(TCL_LIB)\n+fi\n+\n+dnl Check for location of Tk support (only if Tcl used)\n+dnl Disable Tcl support if Tk not found\n+\n+dnl Check for tk.h\n+if test \"$USE_TCL\" = \"true\"\n+then\n+\n+ice_save_LIBS=\"$LIBS\"\n+ice_save_CFLAGS=\"$CFLAGS\"\n+ice_save_CPPFLAGS=\"$CPPFLAGS\"\n+ice_save_LDFLAGS=\"$LDFLAGS\"\n+\n+CPPFLAGS=\"$CPPFLAGS $X_CFLAGS -I$TCL_INCDIR\"\n+\n+TK_INCDIR=no\n+AC_CHECK_HEADER(tk.h, TK_INCDIR=)\n+for f in /usr/include /usr/include/tk8.0 /usr/local/include /usr/local/include/tk8.0; do\n+if test \"$TK_INCDIR\" = \"no\"; then\n+AC_CHECK_HEADER($f/tk.h, TK_INCDIR=$f)\n+fi\n+done\n+if test \"$TK_INCDIR\" = \"no\"; then\n+AC_MSG_WARN(tcl support disabled; tk.h missing)\n+USE_TCL=\n+fi\n+AC_SUBST(TK_INCDIR)\n+\n+LIBS=\"$ice_save_LIBS\"\n+CFLAGS=\"$ice_save_CFLAGS\"\n+CPPFLAGS=\"$ice_save_CPPFLAGS\"\n+LDFLAGS=\"$ice_save_LDFLAGS\"\n+fi\n+\n+dnl Check for Tk archive\n+if test \"$USE_TCL\" = \"true\"\n+then\n+TK_LIB=\n+for f in tk8.0 tk80; do\n+if test -z \"$TK_LIB\"; then\n+AC_CHECK_LIB($f, main, TK_LIB=$f)\n+fi\n+done\n+if test -z \"$TK_LIB\"; then\n+AC_MSG_WARN(tcl support disabled; Tk library missing)\n+USE_TCL=\n+else\n+TK_LIB=-l$TK_LIB\n+fi\n+AC_SUBST(TK_LIB)\n+fi\n \n AC_OUTPUT(GNUmakefile Makefile.global backend/port/Makefile bin/pg_version/Makefile bin/psql/Makefile bin/pg_dump/Makefile backend/utils/Gen_fmgrtab.sh interfaces/libpq/Makefile interfaces/libpgtcl/Makefile interfaces/ecpg/lib/Makefile ) \n\n===========================================================================\n\n--- Makefile.global.in.orig\tFri Feb 13 01:00:18 1998\n+++ Makefile.global.in\tWed Feb 18 16:49:57 1998\n@@ -160,18 +160,18 @@\n USE_TCL= @USE_TCL@\n # customize these to your site's needs\n #\n-TCL_INCDIR= /usr/local/include\n+TCL_INCDIR= @TCL_INCDIR@\n TCL_LIBDIR= /usr/local/lib\n-TCL_LIB= -ltcl8.0\n-TK_INCDIR= /usr/local/include\n+TCL_LIB= @TCL_LIB@\n+TK_INCDIR= @TK_INCDIR@\n TK_LIBDIR= /usr/local/lib\n-TK_LIB= -ltk8.0\n+TK_LIB= @TK_LIB@\n \n USE_PERL= @USE_PERL@\n \n-X11_INCDIR= /usr/include\n-X11_LIBDIR= /usr/lib\n-X11_LIB= -lX11 @SOCKET_LIB@ @NSL_LIB@\n+X_CFLAGS= @X_CFLAGS@\n+X_LIBS= @X_LIBS@\n+X11_LIBS= -lX11 @X_EXTRA_LIBS@\n \n \n ##############################################################################\n\n===========================================================================\n\n--- bin/pgtclsh/Makefile.orig\tMon Jan 26 01:00:49 1998\n+++ bin/pgtclsh/Makefile\tWed Feb 18 16:05:03 1998\n@@ -14,7 +14,7 @@\n SRCDIR= ../..\n include ../../Makefile.global\n \n-CFLAGS+= -I$(TCL_INCDIR) -I$(TK_INCDIR) -I$(X11_INCDIR) -I$(SRCDIR)/interfaces/libpgtcl\n+CFLAGS+= $(X_CFLAGS) -I$(TCL_INCDIR) -I$(TK_INCDIR) -I$(SRCDIR)/interfaces/libpgtcl\n \n ifdef KRBVERS\n LDFLAGS+= $(KRBLIBS)\n@@ -33,8 +33,8 @@\n \n pgtksh: pgtkAppInit.o\n \t$(CC) $(CFLAGS) -o $@ pgtkAppInit.o \\\n-\t $(LIBPGTCL) $(LIBPQ) -L$(TCL_LIBDIR) -L$(TK_LIBDIR) -L$(X11_LIBDIR) \\\n-\t $(TK_LIB) $(TCL_LIB) -lX11 -lm $(LDFLAGS)\n+\t $(LIBPGTCL) $(LIBPQ) -L$(TCL_LIBDIR) -L$(TK_LIBDIR) $(X_LIBS) \\\n+\t $(TK_LIB) $(TCL_LIB) $(X11_LIBS) -lm $(LDFLAGS)\n \n install: pgtclsh pgtksh\n \t$(INSTALL) $(INSTL_EXE_OPTS) pgtclsh $(DESTDIR)$(BINDIR)/pgtclsh\n \n\n===========================================================================\n===========================================================================\nIn check results, I found the -u (unified) diff option to make some of\nthe differences more intelligable.\n\n--- test/regress/checkresults.orig\tTue Nov 25 15:07:18 1997\n+++ test/regress/checkresults\tTue Feb 17 13:04:30 1998\n@@ -7,7 +7,7 @@\n for file in `cat \"$@\" | grep 'failed$' | cut -d \" \" -f 1`\n do\n \techo \"====== $file ======\"\n-\tdiff -w expected/$file.out results\n+\tdiff -uw expected/$file.out results\n done\n",
"msg_date": "Wed, 18 Feb 1998 20:18:12 -0700 (MST)",
"msg_from": "Brook Milligan <[email protected]>",
"msg_from_op": true,
"msg_subject": "a few patches for the upcoming 6.3 release"
},
{
"msg_contents": "Applied, with autoconf.\n\n\n> \n> In getting 6.3 working on a NetBSD/i386 box I have made up a few\n> patches that correct some areas of the\n> code/documentation/configuration. They follow, each preceded by a\n> description. I hope they are useful; they allowed me to do a complete\n> compilation using the --with-tcl configuration option, something I\n> couldn't originally.\n> \n> Cheers,\n> Brook\n> \n> ===========================================================================\n> ===========================================================================\n> \n> NOTE: All patches originate at root_of_postgresql_tarball/src.\n> \n> ===========================================================================\n> ===========================================================================\n> Documentation on at least these configure options is missing from INSTALL.\n> \n> --- ../INSTALL.orig\tMon Feb 2 01:00:13 1998\n> +++ ../INSTALL\tWed Feb 18 16:42:54 1998\n> @@ -267,6 +267,15 @@\n> listens for incoming connections on. The\n> default for this is port 5432.\n> \n> + --with-defaults Use default responses to several queries during\n> + configuration.\n> +\n> + --with-tcl Enables programs requiring Tcl/Tk and X11,\n> + including pgtclsh and libpgtcl.\n> +\n> + --with-perl Enables the perl interface. Note that this\n> + requires an installed version of postgreSQL.\n> +\n> As an example, here is the configure script I use on a Sparc\n> Solaris 2.5 system with /opt/postgres being the install base.\n> \n> \n> ===========================================================================\n> ===========================================================================\n> I found that the configuration scripts were not looking for X11/tcl/tk\n> support and the default built into Makefile.global.in was wrong for my\n> system. As a result using the --with-tcl bombed when it came to\n> pgtclsh and libpgtcl. The following three patches change the\n> configuration system to search for X11/tcl/tk in various spots. I\n> hope all the relevant possibilities are included; it's easy to add new\n> spots. They also change the pgtclsh/Makefile to use the right\n> variables set in Makefile.global.\n> \n> Note that a new configure script will need to be made with autoconf if\n> this patch is installed.\n> \n> --- configure.in.orig\tTue Feb 17 01:00:36 1998\n> +++ configure.in\tWed Feb 18 18:18:09 1998\n> @@ -229,6 +229,7 @@\n> \tUSE_TCL=\n> fi\n> export USE_TCL\n> +USE_X=$USE_TCL\n> \n> dnl We exclude perl support unless we override it with --with-perl\n> if test \"X$with_perl\" = \"Xyes\"\n> @@ -514,5 +515,131 @@\n> AC_MSG_CHECKING(setting DEF_PGPORT)\n> AC_DEFINE_UNQUOTED(DEF_PGPORT, \"${DEF_PGPORT}\")\n> AC_MSG_RESULT($DEF_PGPORT)\n> +\n> +dnl Check for X libraries\n> +\n> +if test \"$USE_X\" = true; then\n> +\n> +ice_save_LIBS=\"$LIBS\"\n> +ice_save_CFLAGS=\"$CFLAGS\"\n> +ice_save_CPPFLAGS=\"$CPPFLAGS\"\n> +ice_save_LDFLAGS=\"$LDFLAGS\"\n> +\n> +AC_PATH_XTRA\n> +\n> +LIBS=\"$LIBS $X_EXTRA_LIBS\"\n> +CFLAGS=\"$CFLAGS $X_CFLAGS\"\n> +CPPFLAGS=\"$CPPFLAGS $X_CFLAGS\"\n> +LDFLAGS=\"$LDFLAGS $X_LIBS\"\n> +\n> +dnl Check for X library\n> +\n> +X11_LIBS=\"\"\n> +AC_CHECK_LIB(X11, XOpenDisplay, X11_LIBS=\"-lX11\",,${X_PRE_LIBS})\n> +if test \"$X11_LIBS\" = \"\"; then\n> +dnl Not having X may be fatal. Let the user fix this.\n> +AC_MSG_WARN([The X11 library '-lX11' could not be found.\n> + Please use the configure options '--x-includes=DIR'\n> + and '--x-libraries=DIR' to specify the X location.\n> + See the file 'config.log' for further diagnostics.])\n> +fi\n> +AC_SUBST(X_LIBS)\n> +AC_SUBST(X11_LIBS)\n> +AC_SUBST(X_PRE_LIBS)\n> +\n> +LIBS=\"$ice_save_LIBS\"\n> +CFLAGS=\"$ice_save_CFLAGS\"\n> +CPPFLAGS=\"$ice_save_CPPFLAGS\"\n> +LDFLAGS=\"$ice_save_LDFLAGS\"\n> +fi\n> +\n> +dnl Check for location of Tcl support\n> +dnl Disable Tcl support if not found\n> +\n> +dnl Check for tcl.h\n> +if test \"$USE_TCL\" = \"true\"\n> +then\n> +TCL_INCDIR=no\n> +AC_CHECK_HEADER(tcl.h, TCL_INCDIR=)\n> +for f in /usr/include /usr/include/tcl8.0 /usr/local/include /usr/local/include/tcl8.0; do\n> +if test \"$TCL_INCDIR\" = \"no\"; then\n> +AC_CHECK_HEADER($f/tcl.h, TCL_INCDIR=$f)\n> +fi\n> +done\n> +if test \"$TCL_INCDIR\" = \"no\"; then\n> +AC_MSG_WARN(tcl support disabled; tcl.h missing)\n> +USE_TCL=\n> +fi\n> +AC_SUBST(TCL_INCDIR)\n> +fi\n> +\n> +dnl Check for Tcl archive\n> +if test \"$USE_TCL\" = \"true\"\n> +then\n> +TCL_LIB=\n> +for f in tcl8.0 tcl80; do\n> +if test -z \"$TCL_LIB\"; then\n> +AC_CHECK_LIB($f, main, TCL_LIB=$f)\n> +fi\n> +done\n> +if test -z \"$TCL_LIB\"; then\n> +AC_MSG_WARN(tcl support disabled; Tcl library missing)\n> +USE_TCL=\n> +else\n> +TCL_LIB=-l$TCL_LIB\n> +fi\n> +AC_SUBST(TCL_LIB)\n> +fi\n> +\n> +dnl Check for location of Tk support (only if Tcl used)\n> +dnl Disable Tcl support if Tk not found\n> +\n> +dnl Check for tk.h\n> +if test \"$USE_TCL\" = \"true\"\n> +then\n> +\n> +ice_save_LIBS=\"$LIBS\"\n> +ice_save_CFLAGS=\"$CFLAGS\"\n> +ice_save_CPPFLAGS=\"$CPPFLAGS\"\n> +ice_save_LDFLAGS=\"$LDFLAGS\"\n> +\n> +CPPFLAGS=\"$CPPFLAGS $X_CFLAGS -I$TCL_INCDIR\"\n> +\n> +TK_INCDIR=no\n> +AC_CHECK_HEADER(tk.h, TK_INCDIR=)\n> +for f in /usr/include /usr/include/tk8.0 /usr/local/include /usr/local/include/tk8.0; do\n> +if test \"$TK_INCDIR\" = \"no\"; then\n> +AC_CHECK_HEADER($f/tk.h, TK_INCDIR=$f)\n> +fi\n> +done\n> +if test \"$TK_INCDIR\" = \"no\"; then\n> +AC_MSG_WARN(tcl support disabled; tk.h missing)\n> +USE_TCL=\n> +fi\n> +AC_SUBST(TK_INCDIR)\n> +\n> +LIBS=\"$ice_save_LIBS\"\n> +CFLAGS=\"$ice_save_CFLAGS\"\n> +CPPFLAGS=\"$ice_save_CPPFLAGS\"\n> +LDFLAGS=\"$ice_save_LDFLAGS\"\n> +fi\n> +\n> +dnl Check for Tk archive\n> +if test \"$USE_TCL\" = \"true\"\n> +then\n> +TK_LIB=\n> +for f in tk8.0 tk80; do\n> +if test -z \"$TK_LIB\"; then\n> +AC_CHECK_LIB($f, main, TK_LIB=$f)\n> +fi\n> +done\n> +if test -z \"$TK_LIB\"; then\n> +AC_MSG_WARN(tcl support disabled; Tk library missing)\n> +USE_TCL=\n> +else\n> +TK_LIB=-l$TK_LIB\n> +fi\n> +AC_SUBST(TK_LIB)\n> +fi\n> \n> AC_OUTPUT(GNUmakefile Makefile.global backend/port/Makefile bin/pg_version/Makefile bin/psql/Makefile bin/pg_dump/Makefile backend/utils/Gen_fmgrtab.sh interfaces/libpq/Makefile interfaces/libpgtcl/Makefile interfaces/ecpg/lib/Makefile ) \n> \n> ===========================================================================\n> \n> --- Makefile.global.in.orig\tFri Feb 13 01:00:18 1998\n> +++ Makefile.global.in\tWed Feb 18 16:49:57 1998\n> @@ -160,18 +160,18 @@\n> USE_TCL= @USE_TCL@\n> # customize these to your site's needs\n> #\n> -TCL_INCDIR= /usr/local/include\n> +TCL_INCDIR= @TCL_INCDIR@\n> TCL_LIBDIR= /usr/local/lib\n> -TCL_LIB= -ltcl8.0\n> -TK_INCDIR= /usr/local/include\n> +TCL_LIB= @TCL_LIB@\n> +TK_INCDIR= @TK_INCDIR@\n> TK_LIBDIR= /usr/local/lib\n> -TK_LIB= -ltk8.0\n> +TK_LIB= @TK_LIB@\n> \n> USE_PERL= @USE_PERL@\n> \n> -X11_INCDIR= /usr/include\n> -X11_LIBDIR= /usr/lib\n> -X11_LIB= -lX11 @SOCKET_LIB@ @NSL_LIB@\n> +X_CFLAGS= @X_CFLAGS@\n> +X_LIBS= @X_LIBS@\n> +X11_LIBS= -lX11 @X_EXTRA_LIBS@\n> \n> \n> ##############################################################################\n> \n> ===========================================================================\n> \n> --- bin/pgtclsh/Makefile.orig\tMon Jan 26 01:00:49 1998\n> +++ bin/pgtclsh/Makefile\tWed Feb 18 16:05:03 1998\n> @@ -14,7 +14,7 @@\n> SRCDIR= ../..\n> include ../../Makefile.global\n> \n> -CFLAGS+= -I$(TCL_INCDIR) -I$(TK_INCDIR) -I$(X11_INCDIR) -I$(SRCDIR)/interfaces/libpgtcl\n> +CFLAGS+= $(X_CFLAGS) -I$(TCL_INCDIR) -I$(TK_INCDIR) -I$(SRCDIR)/interfaces/libpgtcl\n> \n> ifdef KRBVERS\n> LDFLAGS+= $(KRBLIBS)\n> @@ -33,8 +33,8 @@\n> \n> pgtksh: pgtkAppInit.o\n> \t$(CC) $(CFLAGS) -o $@ pgtkAppInit.o \\\n> -\t $(LIBPGTCL) $(LIBPQ) -L$(TCL_LIBDIR) -L$(TK_LIBDIR) -L$(X11_LIBDIR) \\\n> -\t $(TK_LIB) $(TCL_LIB) -lX11 -lm $(LDFLAGS)\n> +\t $(LIBPGTCL) $(LIBPQ) -L$(TCL_LIBDIR) -L$(TK_LIBDIR) $(X_LIBS) \\\n> +\t $(TK_LIB) $(TCL_LIB) $(X11_LIBS) -lm $(LDFLAGS)\n> \n> install: pgtclsh pgtksh\n> \t$(INSTALL) $(INSTL_EXE_OPTS) pgtclsh $(DESTDIR)$(BINDIR)/pgtclsh\n> \n> \n> ===========================================================================\n> ===========================================================================\n> In check results, I found the -u (unified) diff option to make some of\n> the differences more intelligable.\n> \n> --- test/regress/checkresults.orig\tTue Nov 25 15:07:18 1997\n> +++ test/regress/checkresults\tTue Feb 17 13:04:30 1998\n> @@ -7,7 +7,7 @@\n> for file in `cat \"$@\" | grep 'failed$' | cut -d \" \" -f 1`\n> do\n> \techo \"====== $file ======\"\n> -\tdiff -w expected/$file.out results\n> +\tdiff -uw expected/$file.out results\n> done\n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sun, 22 Feb 1998 14:57:47 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] a few patches for the upcoming 6.3 release"
}
] |
[
{
"msg_contents": "Now that I have libpgtcl compiled and installed I tried pgaccess-0.76\nand got the following error:\n\n /usr/libexec/ld.so: Undefined symbol \"_pgresStatus\" in wish8.0:/usr/local/pgsql/lib/libpgtcl.so.1.0\n\nPresumably this results from the following line in\ninterfaces/libpgtcl/pgtclCmds.c, which is the only reference to\npgresStatus in that directory:\n\npgtclCmds.c:538:\tTcl_AppendResult(interp, pgresStatus[PQresultStatus(result)], 0);\n\nIs this the result of some new changes to libpgtcl or should I be\nincluding another library to resolve this? It seems that we should\nmake the 6.3 distribution work with pgaccess since it is often\nreferred to as a useful graphical interface on these lists.\n\nCheers,\nBrook\n",
"msg_date": "Wed, 18 Feb 1998 20:38:19 -0700 (MST)",
"msg_from": "Brook Milligan <[email protected]>",
"msg_from_op": true,
"msg_subject": "libpgtcl undefined symbol error with pgaccess-0.76"
}
] |
[
{
"msg_contents": "> I am trying to track down the actual text of the SQL92 standard.\n> In a HotBot search I found a page that declared the standard is\n> only available in hardcopy (no Web page) on the page:\n>\n> http://www.naiua.org/std-orgs.html\n>\n> It went on to say that the standard could be ordered from ANSI\n> or ISO. Do you know if this is true (only in hardcopy)?\n\nYes, I believe this is likely true. The standards organizations pay for\nsome of their operation by selling hardcopies. Probably not extremely\ncheap, either...\n\n> Also,\n> are the ISO and ANSI standards identical? I searched both the\n> www.ansi.org and www.iso.ch sites and all I could find were an\n> interface to order the books and no mention as to whether or\n> not the standards were identical.\n\nI believe that they are, but don't know with absolute certainty. I had\nfound a text file of a draft of the standard somewhere on the web\n(perhaps at berkeley?), which I hope is close to the final version.\n\n - Tom\n\n\n",
"msg_date": "Thu, 19 Feb 1998 04:51:37 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [QUESTIONS] constraint clause"
}
] |
[
{
"msg_contents": "Dates & times:\nI confirm that libc6 2.0.7pre1 cures all the time and date problems\nthat I was having with glibc2. So no patches are needed; just undefine\nHAVE_INT_TIMEZONE for glibc2.\n\nInt2, int4, oidint2, oidint4:\nIn the regression tests, for glibc2 we should expect to see 'Numerical\nresult out of range' rather than `Math result not representable'.\n\nNew with libc6 2.0.7:\nFloat4: some of the tests failed because glibc2 works _better_ than libc5!\nOne test involves inserting values of 10e-40. libc5 fails, but glibc2\nsays that 10e-40 is equivalent to 1e-39 and successfully inserts it.\nIt also stores larger numbers as Inf and -Inf and inserts these values rather\nthan failing. So many of the tests report 9 rows instead of 5.\n\nFloat8:\nAgain some tests work which didn't with libc5:\nexpected:\nQUERY: UPDATE FLOAT8_TBL \n SET f1 = FLOAT8_TBL.f1 * '-1'\n WHERE FLOAT8_TBL.f1 > '0.0';\nQUERY: SELECT '' AS bad, f.f1 * '1e200' from FLOAT8_TBL f;\nERROR: Bad float8 input format -- overflow\n\nglibc2:\nQUERY: SELECT '' AS bad, f.f1 * '1e200' from FLOAT8_TBL f;\nbad|?column? \n---+----------------\n |0\n |-3.484e+201\n |-1.0043e+203 \n |Infinity \n |-1.2345678901234\n(5 rows)\n\nOn the other hand, this no longer works:\nexpected:\nQUERY: SELECT '' AS bad, : (f.f1) from FLOAT8_TBL f;\nbad| ?column?\n---+--------------------\n | 1\n |7.39912306090513e-16\n | 0 \n | 0 \n | 1\n(5 rows) \n\nglibc2:\nQUERY: SELECT '' AS bad, : (f.f1) from FLOAT8_TBL f;\nERROR: exp() result is out of range\n\n\n\ngeometry:\nresults from converting circles to polygons are still different by about\n3e-12 on some co-ordinates.\n\ntriggers:\nI think the expected results are wrong:\n I get (among other differences): \nQUERY: insert into fkeys2 values (70, '5', 3);\nERROR: check_fkeys2_pkey_exist: tuple references non-existing key in pkeys\n\nwhich is not in the expected results but looks right.\n\n\nselect-views:\nmany differences, but I believe these are also seen on other platforms.\n\n\n-- \nOliver Elphick [email protected]\nIsle of Wight http://www.lfix.co.uk/oliver\n\nPGP key from public servers; key ID 32B8FAA1\n\n\n",
"msg_date": "Thu, 19 Feb 1998 08:02:35 +0000",
"msg_from": "\"Oliver Elphick\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Status of glibc2 problems"
}
] |
[
{
"msg_contents": "Do you mean you didn't get any output? Or you got just the numbers? If\nyou didn't get anything that's the old problem and I believe the same\nproblem holds for ODBC (Julie?). The first fetch returns no row, so the\nprogram ends. Unfortunately I haven't changed anything I expect to be\nable to cause that problem. It seems to be a backend/library problem.\n\nAnyway, which version are you using? I cannot connect to a database\nsince last week or so. I tried re-initdb'ing it but get a seg fault in\nthe backend when doing a createuser. So my only chance of testing right\nnow is to manually inspect the code created.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tKeith Parks [SMTP:[email protected]]\n> Sent:\tWednesday, February 18, 1998 8:28 PM\n> To:\[email protected]\n> Cc:\[email protected]\n> Subject:\tRe: New ecgp code problem.\n> \n> Hi,\n> \n> Things were working upto a point when I was playing yesterday.\n> \n> The perftest seemed to run OK\n> The test2 example ran but did not display the varchar (name) field.\n> \n> I'm afraid I've not got very far debugging.\n> \n> Keith.\n> \n> \n> \n",
"msg_date": "Thu, 19 Feb 1998 09:51:25 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: New ecgp code problem."
},
{
"msg_contents": "On Thu, 19 Feb 1998, Meskes, Michael wrote:\n\n> Do you mean you didn't get any output? Or you got just the numbers? If\n> you didn't get anything that's the old problem and I believe the same\n> problem holds for ODBC (Julie?). The first fetch returns no row, so the\n> program ends. Unfortunately I haven't changed anything I expect to be\n> able to cause that problem. It seems to be a backend/library problem.\n\n\tI spent time with Julie last night working on this, and the\nproblem comes back down to \"pg_user: Permission Denied\"...\n\n\tI think this is a high priority problem, since it affects so many\nthings.\n\n\tWhat is the chance of (how hard would it be to?) having a hard\ncoded view setup that overrides the permissions lock on pg_user? At what\npoint is the lock implemented, software level, or file system?\n\n\tAll we would really need was:\n\n\tcreate view all_users as select usename,usesysid from pg_user;\n\n\tWhich psql/odbc/ecgp/jdbc could call instead of what its doing\nnow...\n\n\n\n",
"msg_date": "Thu, 19 Feb 1998 08:02:19 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "pg_user permissions problem (Was: Re: [HACKERS] RE: New ecgp code\n\tproblem.)"
},
{
"msg_contents": "The Hermit Hacker writes:\n> \tWhat is the chance of (how hard would it be to?) having a hard\n> coded view setup that overrides the permissions lock on pg_user? At what\n> point is the lock implemented, software level, or file system?\n\nI wasn't able to read data from a DB that I created with the very same user.\nBut I was able to connect, insert data etc. Why was that possible?\n\n> \tAll we would really need was:\n> \n> \tcreate view all_users as select usename,usesysid from pg_user;\n> \n> \tWhich psql/odbc/ecgp/jdbc could call instead of what its doing\n> now...\n\necpg doesn't call anything like that. All I use is some PQ... functions. No\ndirect access to whatever user permission system we have. So does this have\nto be changed in libpq?\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Thu, 19 Feb 1998 14:13:06 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: pg_user permissions problem (Was: Re: [HACKERS] RE: New ecgp code\n\tproblem.)"
},
{
"msg_contents": "> \tI spent time with Julie last night working on this, and the\n> problem comes back down to \"pg_user: Permission Denied\"...\n> \n> \tI think this is a high priority problem, since it affects so many\n> things.\n> \n> \tWhat is the chance of (how hard would it be to?) having a hard\n> coded view setup that overrides the permissions lock on pg_user? At what\n> point is the lock implemented, software level, or file system?\n> \n\nI believe the only way to fix it at this point is to remove the REVOKE\npg_user from initdb, and add a check in user.c to make sure the\npermissions on pg_user are not NULL if they try and use passwords.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 09:31:50 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: pg_user permissions problem (Was: Re: [HACKERS] RE: New ecgp code\n\tproblem.)"
}
] |
[
{
"msg_contents": "Exactly the same with the latest Oracle7 version.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tZeugswetter Andreas SARZ [SMTP:[email protected]]\n> Sent:\tWednesday, February 18, 1998 6:33 PM\n> To:\t'[email protected]'\n> Subject:\tRe: [HACKERS] Subselects and NOTs\n> \n> This is what I did:\n> \tcreate table a (a int, a1 char(8));\n> \tcreate table b (b int);\n> \tinsert into a values (1, 'one');\n> \tinsert into a values (NULL, 'null');\n> \n> \n> and this is what I got from Informix:\n> > select * from a where a not in (select * from b);\n> a a1\n> 1 one\n> null\n> 2 row(s) retrieved.\n> > select * from a where not (a in (select * from b));\n> a a1\n> 1 one\n> null\n> 2 row(s) retrieved.\n> > select * from a where not a in (select * from b);\n> a a1\n> 1 one\n> null\n> 2 row(s) retrieved.\n> > select * from a where a<>(select * from b);\n> a a1\n> No rows found.\n> > select * from a where a=(select * from b);\n> a a1\n> No rows found.\n> \n> Andreas\n",
"msg_date": "Thu, 19 Feb 1998 09:58:07 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "Meskes, Michael wrote:\n> \n> Exactly the same with the latest Oracle7 version.\n\nI also get the same results with Sybasem, with one interesting\nexception. Sybase will not let me select * in a subquery! It gives\nme the following:\n\nThe symbol '*' can only be used for a subquery select list when the subquery is\nintroduced with EXISTS or NOT EXISTS.\n\nchanging this to \"select b from b\" seems to work though.\n\nOcie\n\n> \n> Michael\n> \n> --\n> Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> [email protected] | Europark A2, Adenauerstr. 20\n> [email protected] | 52146 Wuerselen\n> Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n> \n> > -----Original Message-----\n> > From:\tZeugswetter Andreas SARZ [SMTP:[email protected]]\n> > Sent:\tWednesday, February 18, 1998 6:33 PM\n> > To:\t'[email protected]'\n> > Subject:\tRe: [HACKERS] Subselects and NOTs\n> > \n> > This is what I did:\n> > \tcreate table a (a int, a1 char(8));\n> > \tcreate table b (b int);\n> > \tinsert into a values (1, 'one');\n> > \tinsert into a values (NULL, 'null');\n> > \n> > \n> > and this is what I got from Informix:\n> > > select * from a where a not in (select * from b);\n> > a a1\n> > 1 one\n> > null\n> > 2 row(s) retrieved.\n> > > select * from a where not (a in (select * from b));\n> > a a1\n> > 1 one\n> > null\n> > 2 row(s) retrieved.\n> > > select * from a where not a in (select * from b);\n> > a a1\n> > 1 one\n> > null\n> > 2 row(s) retrieved.\n> > > select * from a where a<>(select * from b);\n> > a a1\n> > No rows found.\n> > > select * from a where a=(select * from b);\n> > a a1\n> > No rows found.\n> > \n> > Andreas\n> \n> \n\n",
"msg_date": "Thu, 19 Feb 1998 10:21:58 -0800 (PST)",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "[email protected] wrote:\n> \n> Meskes, Michael wrote:\n> >\n> > Exactly the same with the latest Oracle7 version.\n> \n> I also get the same results with Sybasem, with one interesting\n> exception. Sybase will not let me select * in a subquery! It gives\n> me the following:\n> \n> The symbol '*' can only be used for a subquery select list when the subquery is\n> introduced with EXISTS or NOT EXISTS.\n\nIt's funny... SyBases could check subquery' target list length...\n\nOk, thanks. Pg returns the same results for these queries. But I asked\nAndreas to run other queries and they show that both Oracle & Informix\npush NOT into subquery clause (and so, 'NOT x IN' is equal 'x NOT IN'), \nPg doesn't, currently. \n\nOcie, could you run this in SyBase:\n\ncreate table a (a int, a1 char(8));\ncreate table b (b int);\ninsert into a values (1, 'one');\ninsert into a values (NULL, 'null');\n\ninsert into b values (1);\ninsert into b values (NULL);\n\nselect * from a where a in (select * from b);\n-- 1 row with a == 1 expected\nselect * from a where a not in (select * from b);\n-- 0 row expected\nselect * from a where not a in (select * from b);\n-- 0 row in Oracle & Informix, 1 row in Pg (with a == NULL), SyBase ???\nselect * from a where not (a in (select * from b));\n-- the same. SyBase ???\n\nVadim\n",
"msg_date": "Fri, 20 Feb 1998 08:26:51 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "Vadim B. Mikheev wrote:\n> \n> [email protected] wrote:\n> > \n> > Meskes, Michael wrote:\n> > >\n> > > Exactly the same with the latest Oracle7 version.\n> > \n> > I also get the same results with Sybasem, with one interesting\n> > exception. Sybase will not let me select * in a subquery! It gives\n> > me the following:\n> > \n> > The symbol '*' can only be used for a subquery select list when the subquery is\n> > introduced with EXISTS or NOT EXISTS.\n> \n> It's funny... SyBases could check subquery' target list length...\n> \n> Ok, thanks. Pg returns the same results for these queries. But I asked\n> Andreas to run other queries and they show that both Oracle & Informix\n> push NOT into subquery clause (and so, 'NOT x IN' is equal 'x NOT IN'), \n> Pg doesn't, currently. \n> \n> Ocie, could you run this in SyBase:\n\nHere's my output. BTW, this reminds me of a small request (perhaps I\nwill write a patch for it myself). In Postgres, a column can be\ndesignated \"not null\", the default being to allow nulls. In the\ndefault Sybase configuration, it is the other way around. In the\ninterest of writing cross database compatible code, I try to always\ncall out columns as either \"null\" (nulls allowed), or \"not null\"\n(nulls not allowed). Unfortunately, Postgres does not support this\n(Please forgive me if this has been added recently, as I have been too\nbusy to try out the latest code). \n\nIn short, it would be nice if Postgres would take \"null\" as a type\nspecifier as well as \"not null\".\n\nAnd now for the Sybase output (I think it agrees with the rest):\n\n1> select @@version\n2> go\n \n ----------------------------------------------------------------------------- \n SQL Server/11.0.2/P/Sun_svr4/OS 5.4/EBF 6536/OPT/Sat Aug 17 11:54:59 PDT 1996 \n \n(1 row affected)\n1> create table a (a int null,a1 char(8) not null)\n2> go\n1> create table b (b int null)\n2> go\n1> insert into a values (1, 'one')\n2> insert into a values (NULL, 'null')\n3> insert into b values (1)\n4> insert into b values (NULL)\n5> go\n(1 row affected)\n(1 row affected)\n(1 row affected)\n(1 row affected)\n1> select * from a where a in (select b from b)\n2> go\n a a1 \n ----------- -------- \n 1 one \n \n(1 row affected)\n1> select * from a where a not in (select b from b)\n2> go\n a a1 \n ----------- -------- \n \n(0 rows affected)\n1> select * from a where not a in (select b from b)\n2> go\n a a1 \n ----------- -------- \n \n(0 rows affected)\n1> select * from a where not (a in (select b from b))\n2> go\n a a1 \n ----------- -------- \n \n(0 rows affected)\n\n\nOn another note, I have been following this \"not a in b\" vs \"not a in\nb\" discussion and it seems to me that the two statements are logically\nequivalent. Testing for a's membership in the set b and then negating\nshould be equivalent to testing for a's membership in the compliment\nof set b. In these tests, nulls seem to be treated just like any\nother value.\n\nOcie\n\n",
"msg_date": "Thu, 19 Feb 1998 17:46:09 -0800 (PST)",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "Added to TODO.\n\n> Here's my output. BTW, this reminds me of a small request (perhaps I\n> will write a patch for it myself). In Postgres, a column can be\n> designated \"not null\", the default being to allow nulls. In the\n> default Sybase configuration, it is the other way around. In the\n> interest of writing cross database compatible code, I try to always\n> call out columns as either \"null\" (nulls allowed), or \"not null\"\n> (nulls not allowed). Unfortunately, Postgres does not support this\n> (Please forgive me if this has been added recently, as I have been too\n> busy to try out the latest code). \n> \n> In short, it would be nice if Postgres would take \"null\" as a type\n> specifier as well as \"not null\".\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 09:33:26 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "> Added to TODO.\n> > ... In Postgres, a column can be\n> > designated \"not null\", the default being to allow nulls. In the\n> > default Sybase configuration, it is the other way around. In the\n> > interest of writing cross database compatible code, I try to always\n> > call out columns as either \"null\" (nulls allowed), or \"not null\"\n> > (nulls not allowed). Unfortunately, Postgres does not support this\n> > In short, it would be nice if Postgres would take \"null\" as a type\n> > specifier as well as \"not null\".\n\nWe currently get a shift/reduce parsing conflict on this since NULL can be\nspecified in other constraint clauses and since the constraint clauses are\nonly whitespace delimited. It might be that this part of the parser can be\nredone, or perhaps the only way around is to restrict the ordering of the\nconstraints. But NULL constraint is not SQL92 and free ordering is...\n\n - Tom\n\n",
"msg_date": "Fri, 20 Feb 1998 15:42:05 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "I think my original patch about NOT NULL constraint allows this type of\nconstruction. However, thinks have changed in the parser since I made\nthis patch, and now seems that a construction like \nCREATE TABLE a (b int constraint not_null NOT NULL)\nis now valid (I only saw gram.y - maybe I'm wrong). I can make the\npatches to allow NULL, but I only want a litte help: is the name of this\ntype of constraint beeing saved anywere?\n\n\tRobson.\n\n\nBruce Momjian wrote:\n> \n> Added to TODO.\n> \n> > Here's my output. BTW, this reminds me of a small request (perhaps I\n> > will write a patch for it myself). In Postgres, a column can be\n> > designated \"not null\", the default being to allow nulls. In the\n> > default Sybase configuration, it is the other way around. In the\n> > interest of writing cross database compatible code, I try to always\n> > call out columns as either \"null\" (nulls allowed), or \"not null\"\n> > (nulls not allowed). Unfortunately, Postgres does not support this\n> > (Please forgive me if this has been added recently, as I have been too\n> > busy to try out the latest code).\n> >\n> > In short, it would be nice if Postgres would take \"null\" as a type\n> > specifier as well as \"not null\".\n> \n> --\n> Bruce Momjian\n> [email protected]\n",
"msg_date": "Fri, 20 Feb 1998 13:51:22 -0300",
"msg_from": "Robson Paniago de Miranda <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "> I think my original patch about NOT NULL constraint allows this type of\n> construction. However, thinks have changed in the parser since I made\n> this patch, and now seems that a construction like\n> CREATE TABLE a (b int constraint not_null NOT NULL)\n> is now valid (I only saw gram.y - maybe I'm wrong). I can make the\n> patches to allow NULL, but I only want a litte help: is the name of this\n> type of constraint beeing saved anywere?\n\nYes, I think so.\n\n",
"msg_date": "Fri, 20 Feb 1998 16:51:31 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "> \n> > Added to TODO.\n> > > ... In Postgres, a column can be\n> > > designated \"not null\", the default being to allow nulls. In the\n> > > default Sybase configuration, it is the other way around. In the\n> > > interest of writing cross database compatible code, I try to always\n> > > call out columns as either \"null\" (nulls allowed), or \"not null\"\n> > > (nulls not allowed). Unfortunately, Postgres does not support this\n> > > In short, it would be nice if Postgres would take \"null\" as a type\n> > > specifier as well as \"not null\".\n> \n> We currently get a shift/reduce parsing conflict on this since NULL can be\n> specified in other constraint clauses and since the constraint clauses are\n> only whitespace delimited. It might be that this part of the parser can be\n> redone, or perhaps the only way around is to restrict the ordering of the\n> constraints. But NULL constraint is not SQL92 and free ordering is...\n\nOK, removed from TODO.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 11:55:23 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> > \n> > > Added to TODO.\n> > > > ... In Postgres, a column can be\n> > > > designated \"not null\", the default being to allow nulls. In the\n> > > > default Sybase configuration, it is the other way around. In the\n> > > > interest of writing cross database compatible code, I try to always\n> > > > call out columns as either \"null\" (nulls allowed), or \"not null\"\n> > > > (nulls not allowed). Unfortunately, Postgres does not support this\n> > > > In short, it would be nice if Postgres would take \"null\" as a type\n> > > > specifier as well as \"not null\".\n> > \n> > We currently get a shift/reduce parsing conflict on this since NULL can be\n> > specified in other constraint clauses and since the constraint clauses are\n> > only whitespace delimited. It might be that this part of the parser can be\n> > redone, or perhaps the only way around is to restrict the ordering of the\n> > constraints. But NULL constraint is not SQL92 and free ordering is...\n> \n> OK, removed from TODO.\n\nI'll look into this since I requested it. I'll see if I can add this\nwithout breaking everything else.\n\nOcie Mitchell\n",
"msg_date": "Fri, 20 Feb 1998 10:36:27 -0800 (PST)",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "[email protected] wrote:\n>\n> On another note, I have been following this \"not a in b\" vs \"not a in\n> b\" discussion and it seems to me that the two statements are logically\n> equivalent. Testing for a's membership in the set b and then negating\n> should be equivalent to testing for a's membership in the compliment\n> of set b. In these tests, nulls seem to be treated just like any\n> other value.\n> \n> Ocie\n\nAccording to the SQL standard: Where 'NOT' and 'IN' are written next to\neach other, this is an alias for \"<>ALL\", and 'IN' is an alias for\n\"=ANY\". Therefore:\n\n\"a NOT IN b\" evaluates as: (a) <>ALL (b)\n\"NOT a IN b\" evaluates as: NOT ( (a) =ANY (b) )\n\n...which give these results:\n\n NOT 1 IN 2\ttrue\n 1 NOT IN 2\ttrue\n\n NOT 1 IN NULL\ttrue [NOT (1 =ANY NULL)]\n 1 NOT IN NULL\tfalse [1 <>ALL NULL]\n\nUsing \"NOT IN\" is a little confusing, since you might not think about\nthe two words as only one operator, which cannot be split in two.\n\n/* m */\n",
"msg_date": "Mon, 23 Feb 1998 18:11:50 +0100",
"msg_from": "Mattias Kregert <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "Mattias Kregert wrote:\n> \n> [email protected] wrote:\n> >\n> > On another note, I have been following this \"not a in b\" vs \"not a in\n> > b\" discussion and it seems to me that the two statements are logically\n> > equivalent. Testing for a's membership in the set b and then negating\n> > should be equivalent to testing for a's membership in the compliment\n> > of set b. In these tests, nulls seem to be treated just like any\n> > other value.\n> >\n> > Ocie\n> \n> According to the SQL standard: Where 'NOT' and 'IN' are written next to\n> each other, this is an alias for \"<>ALL\", and 'IN' is an alias for\n> \"=ANY\". Therefore:\n> \n> \"a NOT IN b\" evaluates as: (a) <>ALL (b)\n> \"NOT a IN b\" evaluates as: NOT ( (a) =ANY (b) )\n> \n> ...which give these results:\n> \n> NOT 1 IN 2 true\n> 1 NOT IN 2 true\n> \n> NOT 1 IN NULL true [NOT (1 =ANY NULL)]\n> 1 NOT IN NULL false [1 <>ALL NULL]\n\nThis is exactly how Postgres works now and differ from 3 \"big boys\".\nIf there are no objections then I'll leave this as is. We can return\nto this issue latter.\n\nVadim\n",
"msg_date": "Tue, 24 Feb 1998 08:30:27 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "> This is exactly how Postgres works now and differ from 3 \"big boys\".\n> If there are no objections then I'll leave this as is. We can return\n> to this issue latter.\n\nI interprete SQL92 the same way. Let's leave as-is indefinitely/forever...\n\n - Tom\n\n",
"msg_date": "Tue, 24 Feb 1998 02:16:00 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
},
{
"msg_contents": "Thomas G. Lockhart wrote:\n> \n> > This is exactly how Postgres works now and differ from 3 \"big boys\".\n> > If there are no objections then I'll leave this as is. We can return\n> > to this issue latter.\n> \n> I interprete SQL92 the same way. Let's leave as-is indefinitely/forever...\n\nNice for me. There are so many more important issues :)\n\nVadim\n",
"msg_date": "Tue, 24 Feb 1998 09:48:41 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
}
] |
[
{
"msg_contents": "Where do I find the pgaccess source?\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tBrook Milligan [SMTP:[email protected]]\n> Sent:\tThursday, February 19, 1998 4:38 AM\n> To:\[email protected]\n> Subject:\t[HACKERS] libpgtcl undefined symbol error with\n> pgaccess-0.76\n> \n> Now that I have libpgtcl compiled and installed I tried pgaccess-0.76\n> and got the following error:\n> \n> /usr/libexec/ld.so: Undefined symbol \"_pgresStatus\" in\n> wish8.0:/usr/local/pgsql/lib/libpgtcl.so.1.0\n> \n> Presumably this results from the following line in\n> interfaces/libpgtcl/pgtclCmds.c, which is the only reference to\n> pgresStatus in that directory:\n> \n> pgtclCmds.c:538:\tTcl_AppendResult(interp,\n> pgresStatus[PQresultStatus(result)], 0);\n> \n> Is this the result of some new changes to libpgtcl or should I be\n> including another library to resolve this? It seems that we should\n> make the 6.3 distribution work with pgaccess since it is often\n> referred to as a useful graphical interface on these lists.\n> \n> Cheers,\n> Brook\n",
"msg_date": "Thu, 19 Feb 1998 10:16:46 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] libpgtcl undefined symbol error with pgaccess-0.76"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Meskes, Michael wrote:\n\n> Where do I find the pgaccess source?\n\n\tsrc/bin/pgaccess\n\n\n",
"msg_date": "Thu, 19 Feb 1998 08:03:00 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] libpgtcl undefined symbol error with pgaccess-0.76"
},
{
"msg_contents": " On Thu, 19 Feb 1998, Meskes, Michael wrote:\n\n > Where do I find the pgaccess source?\n\n\t src/bin/pgaccess\n\nActually, the most recent version is\n\n\tftp://ftp.flex.ro/pub/pgaccess/pgaccess-0.76.tar.gz\n\nThis, not the old version, really should ship with 6.3.\n\nCheers,\nBrook\n",
"msg_date": "Thu, 19 Feb 1998 08:52:15 -0700 (MST)",
"msg_from": "Brook Milligan <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] libpgtcl undefined symbol error with pgaccess-0.76"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Brook Milligan wrote:\n\n> On Thu, 19 Feb 1998, Meskes, Michael wrote:\n> \n> > Where do I find the pgaccess source?\n> \n> \t src/bin/pgaccess\n> \n> Actually, the most recent version is\n> \n> \tftp://ftp.flex.ro/pub/pgaccess/pgaccess-0.76.tar.gz\n> \n> This, not the old version, really should ship with 6.3.\n\n\tSomeone needs to tells us of these upgrades, eh? :) Updating\nnow...\n\n\n\n",
"msg_date": "Thu, 19 Feb 1998 11:42:45 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] libpgtcl undefined symbol error with pgaccess-0.76"
},
{
"msg_contents": "Brook Milligan writes:\n> Actually, the most recent version is\n> \n> \tftp://ftp.flex.ro/pub/pgaccess/pgaccess-0.76.tar.gz\n> \n> This, not the old version, really should ship with 6.3.\n\nWhy isn't it in CVS?\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Fri, 20 Feb 1998 12:20:21 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] libpgtcl undefined symbol error with pgaccess-0.76"
},
{
"msg_contents": "On Fri, 20 Feb 1998, Michael Meskes wrote:\n\n> Brook Milligan writes:\n> > Actually, the most recent version is\n> > \n> > \tftp://ftp.flex.ro/pub/pgaccess/pgaccess-0.76.tar.gz\n> > \n> > This, not the old version, really should ship with 6.3.\n> \n> Why isn't it in CVS?\n\n\tCause nobody had told me it was upgraded :( Done now...\n\n",
"msg_date": "Fri, 20 Feb 1998 08:03:14 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] libpgtcl undefined symbol error with pgaccess-0.76"
}
] |
[
{
"msg_contents": "Ouch! :-)\n\nI thought it was an additional program.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tThe Hermit Hacker [SMTP:[email protected]]\n> Sent:\tThursday, February 19, 1998 2:03 PM\n> To:\tMeskes, Michael\n> Cc:\t'Brook Milligan'; [email protected]\n> Subject:\tRE: [HACKERS] libpgtcl undefined symbol error with\n> pgaccess-0.76\n> \n> On Thu, 19 Feb 1998, Meskes, Michael wrote:\n> \n> > Where do I find the pgaccess source?\n> \n> \tsrc/bin/pgaccess\n> \n",
"msg_date": "Thu, 19 Feb 1998 14:02:50 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] libpgtcl undefined symbol error with pgaccess-0.76"
}
] |
[
{
"msg_contents": "Hi all,\n\nWhat about:\ngrant select on pg_user to public;\ncreate rule pg_user_hide_pw as on\nselect to pg_user.passwd\ndo instead select '********' as passwd;\n\nThen if I do:\nselect * from pg_user;\nusename |usesysid|usecreatedb|usetrace|usesuper|usecatupd|passwd |valuntil\n--------+--------+-----------+--------+--------+---------+--------+---------\n-------------------\npostgres| 6|t |t |t |t |********|Sat Jan\n31 07:00:00 2037 NFT\nzeus | 60|t |t |f |t |********|\n(2 rows)\n\nAlso the \\d works for all users !\n\nOnly \"disadvantage\" is that noone can read passwd without first dropping the\nrule pg_user_hide_pw,\nI consider this a feature though ;-)\n\nSince the userauthentication bypasses the rewrite mechanism the logins,\nalter user .. and others do work !\n\nCan all of you try to crack this ?\n\n(c) Andreas Zeugswetter\n\nCopyright by Andreas Zeugswetter 1998 contributed to the postgresql project\n;-)\nWow, I am actually proud of this (so far, and hope it holds what I think it\ndoes)\n\n",
"msg_date": "Thu, 19 Feb 1998 14:59:48 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "\nI like it!!! I'm going to do a fresh buildl and try this out, and see if\nJulie/I can get that ODBC driver working with this...*cross fingers*\n\nBruce...do see anyting particularly bad about this?\n\n\n\nOn Thu, 19 Feb 1998, Zeugswetter Andreas SARZ wrote:\n\n> Hi all,\n> \n> What about:\n> grant select on pg_user to public;\n> create rule pg_user_hide_pw as on\n> select to pg_user.passwd\n> do instead select '********' as passwd;\n> \n> Then if I do:\n> select * from pg_user;\n> usename |usesysid|usecreatedb|usetrace|usesuper|usecatupd|passwd |valuntil\n> --------+--------+-----------+--------+--------+---------+--------+---------\n> -------------------\n> postgres| 6|t |t |t |t |********|Sat Jan\n> 31 07:00:00 2037 NFT\n> zeus | 60|t |t |f |t |********|\n> (2 rows)\n> \n> Also the \\d works for all users !\n> \n> Only \"disadvantage\" is that noone can read passwd without first dropping the\n> rule pg_user_hide_pw,\n> I consider this a feature though ;-)\n> \n> Since the userauthentication bypasses the rewrite mechanism the logins,\n> alter user .. and others do work !\n> \n> Can all of you try to crack this ?\n> \n> (c) Andreas Zeugswetter\n> \n> Copyright by Andreas Zeugswetter 1998 contributed to the postgresql project\n> ;-)\n> Wow, I am actually proud of this (so far, and hope it holds what I think it\n> does)\n> \n> \n\n",
"msg_date": "Thu, 19 Feb 1998 09:14:36 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> Hi all,\n> \n> What about:\n> grant select on pg_user to public;\n> create rule pg_user_hide_pw as on\n> select to pg_user.passwd\n> do instead select '********' as passwd;\n\nWhen I see this, the word 'genius' comes to mind. What a brilliantly\nelegant solution to the problem.\n\n> \n> Then if I do:\n> select * from pg_user;\n> usename |usesysid|usecreatedb|usetrace|usesuper|usecatupd|passwd |valuntil\n> --------+--------+-----------+--------+--------+---------+--------+---------\n> -------------------\n> postgres| 6|t |t |t |t |********|Sat Jan\n> 31 07:00:00 2037 NFT\n> zeus | 60|t |t |f |t |********|\n> (2 rows)\n> \n> Also the \\d works for all users !\n> \n> Only \"disadvantage\" is that noone can read passwd without first dropping the\n> rule pg_user_hide_pw,\n> I consider this a feature though ;-)\n> \n> Since the userauthentication bypasses the rewrite mechanism the logins,\n> alter user .. and others do work !\n> \n> Can all of you try to crack this ?\n> \n> (c) Andreas Zeugswetter\n> \n> Copyright by Andreas Zeugswetter 1998 contributed to the postgresql project\n> ;-)\n> Wow, I am actually proud of this (so far, and hope it holds what I think it\n> does)\n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 09:50:32 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": ">\n> Hi all,\n>\n> What about:\n> grant select on pg_user to public;\n> create rule pg_user_hide_pw as on\n> select to pg_user.passwd\n> do instead select '********' as passwd;\n>\n> Then if I do:\n> select * from pg_user;\n> usename |usesysid|usecreatedb|usetrace|usesuper|usecatupd|passwd |valuntil\n> --------+--------+-----------+--------+--------+---------+--------+---------\n> -------------------\n> postgres| 6|t |t |t |t |********|Sat Jan\n> 31 07:00:00 2037 NFT\n> zeus | 60|t |t |f |t |********|\n> (2 rows)\n>\n> Also the \\d works for all users !\n>\n> Only \"disadvantage\" is that noone can read passwd without first dropping the\n> rule pg_user_hide_pw,\n> I consider this a feature though ;-)\n>\n> Since the userauthentication bypasses the rewrite mechanism the logins,\n> alter user .. and others do work !\n>\n> Can all of you try to crack this ?\n\n Cracked!\n\n create table get_passwds (usename name, passwd text);\n insert into get_passwds select usename, passwd from pg_user;\n select * from get_passwds;\n usename|passwd\n -------+------\n pgsql |\n wieck |test\n (2 rows)\n\n\n\nSorry, Jan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 19 Feb 1998 15:55:07 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Jan Wieck wrote:\n\n> >\n> > Hi all,\n> >\n> > What about:\n> > grant select on pg_user to public;\n> > create rule pg_user_hide_pw as on\n> > select to pg_user.passwd\n> > do instead select '********' as passwd;\n> >\n> > Then if I do:\n> > select * from pg_user;\n> > usename |usesysid|usecreatedb|usetrace|usesuper|usecatupd|passwd |valuntil\n> > --------+--------+-----------+--------+--------+---------+--------+---------\n> > -------------------\n> > postgres| 6|t |t |t |t |********|Sat Jan\n> > 31 07:00:00 2037 NFT\n> > zeus | 60|t |t |f |t |********|\n> > (2 rows)\n> >\n> > Also the \\d works for all users !\n> >\n> > Only \"disadvantage\" is that noone can read passwd without first dropping the\n> > rule pg_user_hide_pw,\n> > I consider this a feature though ;-)\n> >\n> > Since the userauthentication bypasses the rewrite mechanism the logins,\n> > alter user .. and others do work !\n> >\n> > Can all of you try to crack this ?\n> \n> Cracked!\n> \n> create table get_passwds (usename name, passwd text);\n> insert into get_passwds select usename, passwd from pg_user;\n> select * from get_passwds;\n> usename|passwd\n> -------+------\n> pgsql |\n> wieck |test\n> (2 rows)\n\nOkay, its moving in the right direction though...is there a flaw in the\nrules system that allows this to get around this in this manner? Nice\nthing about this is that it doesn't require any 'front end' rewrites,\nwhich is a good thing...\n\nIMHO, this is an acceptable solution to the problem for v6.3s\nrelease...better then ripping out the code altogether...\n\n\n",
"msg_date": "Thu, 19 Feb 1998 09:58:53 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "\nHave we considering using the unix crypt function for passwords? That\nway it wouldn't matter (as much) if people saw the password, and would\nstill be (somewhat less) secure.\n\nOn Thu, 19 February 1998, at 15:55:07, Jan Wieck wrote:\n\n> Cracked!\n> \n> create table get_passwds (usename name, passwd text);\n> insert into get_passwds select usename, passwd from pg_user;\n> select * from get_passwds;\n> usename|passwd\n> -------+------\n> pgsql |\n> wieck |test\n> (2 rows)\n> \n> \n> \n> Sorry, Jan\n> \n> --\n> \n> #======================================================================#\n> # It's easier to get forgiveness for being wrong than for being right. #\n> # Let's break this rule - forgive me. #\n> #======================================== [email protected] (Jan Wieck) #\n> \n> \n",
"msg_date": "Thu, 19 Feb 1998 09:47:47 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "Thus spake Brett McCormick\n> Have we considering using the unix crypt function for passwords? That\n> way it wouldn't matter (as much) if people saw the password, and would\n> still be (somewhat less) secure.\n\nThat's what I do. I have both a Tcl and a Python function to encrypt\nthe password and that's how I store them. You still want to protect\nthem somewhat but it isn't quite as critical as plain tex passwords.\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 424 2871 (DoD#0082) (eNTP) | what's for dinner.\n",
"msg_date": "Thu, 19 Feb 1998 12:55:45 -0500 (EST)",
"msg_from": "[email protected] (D'Arcy J.M. Cain)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> \n> Have we considering using the unix crypt function for passwords? That\n> way it wouldn't matter (as much) if people saw the password, and would\n> still be (somewhat less) secure.\n> \n> On Thu, 19 February 1998, at 15:55:07, Jan Wieck wrote:\n\nI don't know what the problem with using crypt was. It may be because\nhe passes a random salt to the user, and the user makes the password\npacket with the given salt and returns it to the backend. If we use\ncrypt, we have to send a plaintext password over the network, don't we?\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 13:16:23 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > \n> > Have we considering using the unix crypt function for passwords? That\n> > way it wouldn't matter (as much) if people saw the password, and would\n> > still be (somewhat less) secure.\n> > \n> > On Thu, 19 February 1998, at 15:55:07, Jan Wieck wrote:\n> \n> I don't know what the problem with using crypt was. It may be because\n> he passes a random salt to the user, and the user makes the password\n> packet with the given salt and returns it to the backend. If we use\n> crypt, we have to send a plaintext password over the network, don't we?\n\n\tBut, aren't we doing that now? \n\n\n",
"msg_date": "Thu, 19 Feb 1998 13:34:01 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> \n> > > \n> > > \n> > > Have we considering using the unix crypt function for passwords? That\n> > > way it wouldn't matter (as much) if people saw the password, and would\n> > > still be (somewhat less) secure.\n> > > \n> > > On Thu, 19 February 1998, at 15:55:07, Jan Wieck wrote:\n> > \n> > I don't know what the problem with using crypt was. It may be because\n> > he passes a random salt to the user, and the user makes the password\n> > packet with the given salt and returns it to the backend. If we use\n> > crypt, we have to send a plaintext password over the network, don't we?\n> \n> \tBut, aren't we doing that now? \n\nYes, we are using crypt. We are picking a random salt, using crypt to\nencrypt the cleartext password, then sending the salt to the frontend,\nand asking them to supply a password crypted with our requested salt.\n\nAnyway to do this while storing encrypted passwords?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 13:56:18 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> >\n> > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> >\n> > > >\n> > > >\n> > > > Have we considering using the unix crypt function for passwords? That\n> > > > way it wouldn't matter (as much) if people saw the password, and would\n> > > > still be (somewhat less) secure.\n> > > >\n> > > > On Thu, 19 February 1998, at 15:55:07, Jan Wieck wrote:\n> > >\n> > > I don't know what the problem with using crypt was. It may be because\n> > > he passes a random salt to the user, and the user makes the password\n> > > packet with the given salt and returns it to the backend. If we use\n> > > crypt, we have to send a plaintext password over the network, don't we?\n> >\n> > But, aren't we doing that now?\n> \n> Yes, we are using crypt. We are picking a random salt, using crypt to\n> encrypt the cleartext password, then sending the salt to the frontend,\n> and asking them to supply a password crypted with our requested salt.\n\nA benefit of this approach (which I think is valuable) is that, as far\nas sniffing the network is concerned, one-time passwords are used.\n\n> Anyway to do this while storing encrypted passwords?\n\nHow UNIX specific is crypt()? At the moment the same password can be\nused with clients that support crypt(), and send the password encrypted,\nand those that don't, and send the password in clear.\n\nIf you want to store encrypted passwords then the encryption method used\nmust be supported by all types of client - or be included as part of the\nPostgreSQL distribution.\n\nPhil\n",
"msg_date": "Thu, 19 Feb 1998 22:47:59 +0000",
"msg_from": "Phil Thompson <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> > \n> > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> > \n> > > > \n> > > > \n> > > > Have we considering using the unix crypt function for passwords? That\n> > > > way it wouldn't matter (as much) if people saw the password, and would\n> > > > still be (somewhat less) secure.\n> > > > \n> > > > On Thu, 19 February 1998, at 15:55:07, Jan Wieck wrote:\n> > > \n> > > I don't know what the problem with using crypt was. It may be because\n> > > he passes a random salt to the user, and the user makes the password\n> > > packet with the given salt and returns it to the backend. If we use\n> > > crypt, we have to send a plaintext password over the network, don't we?\n> > \n> > \tBut, aren't we doing that now? \n> \n> Yes, we are using crypt. We are picking a random salt, using crypt to\n> encrypt the cleartext password, then sending the salt to the frontend,\n> and asking them to supply a password crypted with our requested salt.\n> \n> Anyway to do this while storing encrypted passwords?\n\nStandard salt is two characters, so an adversary might be able to\nwatch and record which salts produced which replies. Even with a\nsingle login, a brute force attack might still be able to get the\nuser's password. A stronger challenge-response system might be more\nsecure. It should be possible for the server to authenticate a user\nwithout having to store the user's password.\n\nThen again, this is all starting to sound like Kerberos, so if\nPostgres had Kerberos authentication (which I think it does), then\nthis could be used for the ultra-high security authentication system.\n\nOcie Mitchell\n",
"msg_date": "Thu, 19 Feb 1998 15:25:56 -0800 (PST)",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "\nWhat about a public/private key mechanism, like ssh?\n\nOn Thu, 19 February 1998, at 15:25:56, [email protected] wrote:\n\n> Standard salt is two characters, so an adversary might be able to\n> watch and record which salts produced which replies. Even with a\n> single login, a brute force attack might still be able to get the\n> user's password. A stronger challenge-response system might be more\n> secure. It should be possible for the server to authenticate a user\n> without having to store the user's password.\n> \n> Then again, this is all starting to sound like Kerberos, so if\n> Postgres had Kerberos authentication (which I think it does), then\n> this could be used for the ultra-high security authentication system.\n> \n> Ocie Mitchell\n",
"msg_date": "Thu, 19 Feb 1998 19:03:55 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Phil Thompson wrote:\n\n> How UNIX specific is crypt()? At the moment the same password can be\n> used with clients that support crypt(), and send the password encrypted,\n> and those that don't, and send the password in clear.\n> \n> If you want to store encrypted passwords then the encryption method used\n> must be supported by all types of client - or be included as part of the\n> PostgreSQL distribution.\n\nTo implement the current crypt authentication in the JDBC driver, we had\nto supply a crypt class.\n\n\tsrc/interfaces/jdbc/postgresql/util/UnixCrypt.java\n\n> \n> Phil\n> \n\n-- \nPeter T Mount [email protected] or [email protected]\nMain Homepage: http://www.demon.co.uk/finder\nWork Homepage: http://www.maidstone.gov.uk Work EMail: [email protected]\n\n",
"msg_date": "Sat, 21 Feb 1998 10:40:24 +0000 (GMT)",
"msg_from": "Peter T Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "\n> What about a public/private key mechanism, like ssh?\n\nYes, just use ssh as-is and run psql etc in a ssh tunnel\nwhen running over insecure nets.\n\nMaking cryptography part of pgsql is not good idea for two reasons:\n1. It is bloat\n2. It make us targets of legal-problems (US-export-control etc).\n\nIMHO, even Kerberous should be removed.\n\ninfo about ssh is available at:\n\nhttp://www.cs.hut.fi/ssh/ free version\nhttp://www.ssh.fi/ commercial (+mac/windows) versions\n\n regards,\n-- \n---------------------------------------------\nG�ran Thyni, sysadm, JMS Bildbasen, Kiruna\n\n",
"msg_date": "26 Feb 1998 16:35:18 -0000",
"msg_from": "Goran Thyni <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, Feb 26, 1998 at 04:35:18PM -0000, Goran Thyni wrote:\n\n> Yes, just use ssh as-is and run psql etc in a ssh tunnel\n> when running over insecure nets.\n\nIt is not the case that ssh is a complete replacement for anything and\neverything in the way of authentication, authorization and encryption.\nAmong other things, \"just use an ssh tunnel\" means you have to have\nhad a previous out-of-band exchange of keys between the end-points for\nthe authentication to be completely secure. Thus, ssh does not by\nitself scale very well. (This is, of course, the reason why Kerberos\nis supported as an authentication method in recent versions of ssh.\nThis is good for people who use Kerberos 5, the version that ssh now\nsupports, but only people inside the US can legally do that, and for\nthose of us who still use Kerberos IV (I legally run the version from\nKTH, in Sweden, both at home and at work), ssh won't always cut it.)\n\n> IMHO, even Kerberous should be removed.\n\nI strongly disagree. On the off chance that the ability to utilize\ncryptographic software not even included in the release somehow makes\nPosgreSQL a weapon in the eyes of the US government, I guess it would\nbe necessary to remove it, but I can't believe that their laws can be\nso mind-bogglingly stupid? If this is the case, shouldn't /bin/cat be\nexport restricted, because you can use it to stuff text into software\nthat encrypts it?\n\nKerberos itself is export restricted, because it contains software\nthat implements cryptographic functions (specifically, the Top Secret\nalgorithm called DES, which it would mean the end of civilization as\nwe know it if anyone outside the US found out how it works), but this\nsurely cannot mean that anything that can use Kerberos is tainted?\n\nThe operating system I use the most, NetBSD, comes with the ability to\nuse Kerberos authentication present in various program sources, but\nthe actual Kerberos IV source code is in a separate, export restricted\npackage. At a US site, when this is installed, you can compile the\nsources in question with proper -D flags to enable the calls to the\nKerberos functions. I've always understood this to be entirely legal.\n\nAnyway, if it should at some point be decided to strip PostgreSQL of\nany way of even calling user supplied cryptographic functions, PLEASE\ndon't do this without creating a clean, documented framework for\nadding authentication methods to the system, so that those of us who\nneed secure authentication can reintegrate Kerberos and the like on\nour own...\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n",
"msg_date": "Thu, 26 Feb 1998 22:58:36 +0100",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> Anyway, if it should at some point be decided to strip PostgreSQL of\n> any way of even calling user supplied cryptographic functions, PLEASE\n> don't do this without creating a clean, documented framework for\n> adding authentication methods to the system, so that those of us who\n> need secure authentication can reintegrate Kerberos and the like on\n> our own...\n\nDon't worry. Kerberos support is not going to be change, only improved\nif we can. I always liked Kerberos.\n\n-- \nBruce Momjian | 830 Blythe Avenue\[email protected] | Drexel Hill, Pennsylvania 19026\n + If your life is a hard drive, | (610) 353-9879(w)\n + Christ can be your backup. | (610) 853-3000(h)\n",
"msg_date": "Thu, 26 Feb 1998 22:14:54 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "Brett McCormick <[email protected]> writes:\n\n> Have we considering using the unix crypt function for passwords? That\n> way it wouldn't matter (as much) if people saw the password, and would\n> still be (somewhat less) secure.\n\nI would MUCH rather see MD5, salted with the username.\n\n--Michael\n",
"msg_date": "04 Mar 1998 16:58:55 -0800",
"msg_from": "Michael Graff <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "Goran Thyni <[email protected]> writes:\n\n> IMHO, even Kerberous should be removed.\n\nNo, it should be fixed, not removed. I would very much like to use krb5\nauthentication.\n\nMy main problem with the whole authentication system is that there can only\nbe one global method. That is, everything is password based, on one\npassword per user.\n\nI would think a good method would have the (username, auth-data) stored\ntogether, but you could have multiple (username, auth-data) tuples. For\ninstance, I might have a Kerberos5 authentication:\n\n\tusename\t\tauthtype\tauthdata\n\t---------------\t---------------\t------------------------------\n\texplorer\tkrb5\t\[email protected]\n\texplorer\tmd5\t\t4f1929cbca91deadbeef0123...\n\texplorer\tcryptpw\t\tUNIX-standard crypted string\n\texplorer\tplainpw\t\tpassword\n\nThen I can control how someone connects down to a method. I could\nallow different methods for different trust levels. I might trust my\nlocal, server machine with just a password, but deny access to the rest\nof the net unless they use Kerberos5.\n\n--Michael\n",
"msg_date": "04 Mar 1998 17:05:36 -0800",
"msg_from": "Michael Graff <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
}
] |
[
{
"msg_contents": "Okay :-(\n\nBut: I think this is an error in the rewrite system. I think this query\nshould get rewritten !\nCan we fix this ?\n\nAndreas\n> ----------\n> Von: \tJan Wieck[SMTP:[email protected]]\n> Antwort an: \tJan Wieck\n> Gesendet: \tDonnerstag, 19. Februar 1998 15:53\n> An: \tZeugswetter Andreas SARZ\n> Cc: \[email protected]\n> Betreff: \tRe: [HACKERS] Solution to the pg_user passwd problem !?? (c)\n> \n> >\n> > Hi all,\n> >\n> > What about:\n> > grant select on pg_user to public;\n> > create rule pg_user_hide_pw as on\n> > select to pg_user.passwd\n> > do instead select '********' as passwd;\n> >\n> > Then if I do:\n> > select * from pg_user;\n> > usename |usesysid|usecreatedb|usetrace|usesuper|usecatupd|passwd\n> |valuntil\n> >\n> --------+--------+-----------+--------+--------+---------+--------+-------\n> --\n> > -------------------\n> > postgres| 6|t |t |t |t |********|Sat\n> Jan\n> > 31 07:00:00 2037 NFT\n> > zeus | 60|t |t |f |t |********|\n> > (2 rows)\n> >\n> > Also the \\d works for all users !\n> >\n> > Only \"disadvantage\" is that noone can read passwd without first dropping\n> the\n> > rule pg_user_hide_pw,\n> > I consider this a feature though ;-)\n> >\n> > Since the userauthentication bypasses the rewrite mechanism the logins,\n> > alter user .. and others do work !\n> >\n> > Can all of you try to crack this ?\n> \n> Cracked!\n> \n> create table get_passwds (usename name, passwd text);\n> insert into get_passwds select usename, passwd from pg_user;\n> select * from get_passwds;\n> usename|passwd\n> -------+------\n> pgsql |\n> wieck |test\n> (2 rows)\n> \n> \n> \n> Sorry, Jan\n> \n> --\n> \n> #======================================================================#\n> # It's easier to get forgiveness for being wrong than for being right. #\n> # Let's break this rule - forgive me. #\n> #======================================== [email protected] (Jan Wieck) #\n> \n> \n> \n",
"msg_date": "Thu, 19 Feb 1998 15:58:08 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "Well, seeing as Jan is one of the rewrite/rules system experts, let's\nask him.\n\n> \n> Okay :-(\n> \n> But: I think this is an error in the rewrite system. I think this query\n> should get rewritten !\n> Can we fix this ?\n> \n> Andreas\n> > ----------\n> > Von: \tJan Wieck[SMTP:[email protected]]\n> > Antwort an: \tJan Wieck\n> > Gesendet: \tDonnerstag, 19. Februar 1998 15:53\n> > An: \tZeugswetter Andreas SARZ\n> > Cc: \[email protected]\n> > Betreff: \tRe: [HACKERS] Solution to the pg_user passwd problem !?? (c)\n> > \n> > >\n> > > Hi all,\n> > >\n> > > What about:\n> > > grant select on pg_user to public;\n> > > create rule pg_user_hide_pw as on\n> > > select to pg_user.passwd\n> > > do instead select '********' as passwd;\n> > >\n> > > Then if I do:\n> > > select * from pg_user;\n> > > usename |usesysid|usecreatedb|usetrace|usesuper|usecatupd|passwd\n> > |valuntil\n> > >\n> > --------+--------+-----------+--------+--------+---------+--------+-------\n> > --\n> > > -------------------\n> > > postgres| 6|t |t |t |t |********|Sat\n> > Jan\n> > > 31 07:00:00 2037 NFT\n> > > zeus | 60|t |t |f |t |********|\n> > > (2 rows)\n> > >\n> > > Also the \\d works for all users !\n> > >\n> > > Only \"disadvantage\" is that noone can read passwd without first dropping\n> > the\n> > > rule pg_user_hide_pw,\n> > > I consider this a feature though ;-)\n> > >\n> > > Since the userauthentication bypasses the rewrite mechanism the logins,\n> > > alter user .. and others do work !\n> > >\n> > > Can all of you try to crack this ?\n> > \n> > Cracked!\n> > \n> > create table get_passwds (usename name, passwd text);\n> > insert into get_passwds select usename, passwd from pg_user;\n> > select * from get_passwds;\n> > usename|passwd\n> > -------+------\n> > pgsql |\n> > wieck |test\n> > (2 rows)\n> > \n> > \n> > \n> > Sorry, Jan\n> > \n> > --\n> > \n> > #======================================================================#\n> > # It's easier to get forgiveness for being wrong than for being right. #\n> > # Let's break this rule - forgive me. #\n> > #======================================== [email protected] (Jan Wieck) #\n> > \n> > \n> > \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 10:07:44 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "\nAndreas wrote:\n>\n> Okay :-(\n>\n> But: I think this is an error in the rewrite system. I think this query\n> should get rewritten !\n> Can we fix this ?\n>\n> > > Can all of you try to crack this ?\n> >\n> > Cracked!\n> >\n> > create table get_passwds (usename name, passwd text);\n> > insert into get_passwds select usename, passwd from pg_user;\n> > select * from get_passwds;\n> > usename|passwd\n> > -------+------\n> > pgsql |\n> > wieck |test\n> > (2 rows)\n\n Right - it's an error in the rewrite system. The retrieve\n (SELECT) rule rewriting is only done if the command in the\n query is a SELECT. But this time it's an INSERT and the\n rewrite system only looks for rules on the result relation\n (as far as I understood the coding). It doesn't process the\n complete rangetable and no rewriting is done at all.\n\n As a side effect it is also impossible to\n\n INSERT INTO tab SELECT anything FROM view;\n\n And this should really get fixed. I'll take a deep breath and\n then a look at it. But that will take some time.\n\n\nUntil later, Jan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 19 Feb 1998 16:48:40 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
}
] |
[
{
"msg_contents": "I want to run pgindent before the final release. Who has outstanding\npatches they are sitting on that would be affected by this? Only places\nwhere we have added non-conforming code would be affected.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 09:59:52 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Running pgindent"
},
{
"msg_contents": "Bruce Momjian wrote:\n\n> I want to run pgindent before the final release. Who has outstanding\n> patches they are sitting on that would be affected by this? Only places\n> where we have added non-conforming code would be affected.\n\nAck! Can you make at least three changes to pgindent before doing this?\n\n1) for functions which look like\n datetime *\n abstime_datetime(int4 x)\n {\n ...\ndo not put a tab (or extra spaces) between \"datetime\" and \"*\".\n\nFor the end of functions which look like\n ...\n } /* abstime_datetime() */\n\ndo not put any tabs (or just one) between the \"{\" and the comment.\n\nAlso, some comments get wrapped to the next line, making some ugly sources;\nperhaps pgindent should not touch comment lines at all? Or at least keep\nthings on the same line, messing with the tabbing only??\n\n\n\n",
"msg_date": "Thu, 19 Feb 1998 15:32:39 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> \n> Bruce Momjian wrote:\n> \n> > I want to run pgindent before the final release. Who has outstanding\n> > patches they are sitting on that would be affected by this? Only places\n> > where we have added non-conforming code would be affected.\n> \n> Ack! Can you make at least three changes to pgindent before doing this?\n> \n> 1) for functions which look like\n> datetime *\n> abstime_datetime(int4 x)\n> {\n> ...\n> do not put a tab (or extra spaces) between \"datetime\" and \"*\".\n> \n> For the end of functions which look like\n> ...\n> } /* abstime_datetime() */\n> \n> do not put any tabs (or just one) between the \"{\" and the comment.\n> \n> Also, some comments get wrapped to the next line, making some ugly sources;\n> perhaps pgindent should not touch comment lines at all? Or at least keep\n> things on the same line, messing with the tabbing only??\n\nWell, that is what indent does. I can try and work around it, but I am\nafraid I will mess it up somehow. Right now, all the stuff is indented\nconsistently, and running it again will only make the changed stuff look\nlike the rest.\n\nMaking a change will change all the code, even the unchanged stuff. \nAlso, for comments, it will not change comments that are block-style. \nSee the indent manual page and the flags I use for help. Keep in mind I\nuse BSD indent, which does not have the bugs in GNU indent.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 10:59:08 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> I want to run pgindent before the final release. Who has outstanding\n> patches they are sitting on that would be affected by this? Only places\n> where we have added non-conforming code would be affected.\n\nI'm changing bufmgr.c & elog.c to fix problems with smgrblind.\n\nVadim\n",
"msg_date": "Fri, 20 Feb 1998 09:03:12 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> \n> Bruce Momjian wrote:\n> > \n> > I want to run pgindent before the final release. Who has outstanding\n> > patches they are sitting on that would be affected by this? Only places\n> > where we have added non-conforming code would be affected.\n> \n> I'm changing bufmgr.c & elog.c to fix problems with smgrblind.\n> \n> Vadim\n> \n\nLet me know when you would like it done.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 22:02:44 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "Bruce Momjian wrote:\n\n> >\n> > Bruce Momjian wrote:\n> >\n> > > I want to run pgindent before the final release. Who has outstanding\n> > > patches they are sitting on that would be affected by this? Only places\n> > > where we have added non-conforming code would be affected.\n> >\n> > Ack! Can you make at least three changes to pgindent before doing this?\n> >\n> > 1) for functions which look like\n> > datetime *\n> > abstime_datetime(int4 x)\n> > {\n> > ...\n> > do not put a tab (or extra spaces) between \"datetime\" and \"*\".\n> >\n> > For the end of functions which look like\n> > ...\n> > } /* abstime_datetime() */\n> >\n> > do not put any tabs (or just one) between the \"{\" and the comment.\n> >\n> > Also, some comments get wrapped to the next line, making some ugly sources;\n> > perhaps pgindent should not touch comment lines at all? Or at least keep\n> > things on the same line, messing with the tabbing only??\n>\n> Well, that is what indent does. I can try and work around it, but I am\n> afraid I will mess it up somehow. Right now, all the stuff is indented\n> consistently, and running it again will only make the changed stuff look\n> like the rest.\n>\n> Making a change will change all the code, even the unchanged stuff.\n> Also, for comments, it will not change comments that are block-style.\n> See the indent manual page and the flags I use for help. Keep in mind I\n> use BSD indent, which does not have the bugs in GNU indent.\n\nWell, *&^*^#$! I wasted hours cleaning up some of what I considered damage from\nthe last pass through.\nWould it be possible to pass back through (i.e. pipe indent output to a filter)\nand fix at least points (1) and (2)? _That_ would be pretty easy to automate\nsince the heuristics can be pretty simple:\n\nif (first char is nonwhitespace) && (next word is \"*\") && (next line starts with\nword+paren)\nthen compress whitespace\n\nelse if (first char is \"}\") && (next word is \"/*\")\nthen compress whitespace\n\nelse\nwrite line as-is\n\nI'll write the perl if you would be willing to use it?\n\n - Tom\n\n",
"msg_date": "Fri, 20 Feb 1998 03:34:24 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> \n> > \n> > Bruce Momjian wrote:\n> > > \n> > > I want to run pgindent before the final release. Who has outstanding\n> > > patches they are sitting on that would be affected by this? Only places\n> > > where we have added non-conforming code would be affected.\n> > \n> > I'm changing bufmgr.c & elog.c to fix problems with smgrblind.\n> > \n> > Vadim\n> > \n> \n> Let me know when you would like it done.\n\n Got very close to the view permission override - just a vew hours\n for me :-)\n\n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n> \n\n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Fri, 20 Feb 1998 11:52:06 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> \n> > \n> > > \n> > > Bruce Momjian wrote:\n> > > > \n> > > > I want to run pgindent before the final release. Who has outstanding\n> > > > patches they are sitting on that would be affected by this? Only places\n> > > > where we have added non-conforming code would be affected.\n> > > \n> > > I'm changing bufmgr.c & elog.c to fix problems with smgrblind.\n> > > \n> > > Vadim\n> > > \n> > \n> > Let me know when you would like it done.\n> \n> Got very close to the view permission override - just a vew hours\n> for me :-)\n\nGreat. I was thinking about this, and I think it would be best if the\nview permissions were checked as the view owner at view USE time, not\nview CREATION time, that way if permissions on the base tables change,\nthe permissions are properly honored.\n\nI think the range table idea is good, so there is an OWNER field on each\nrange table which defaults to the current user. As views are replaced\nby base tables in the rewrite system, the owner can be changed to the\nowner of the view. The issue is whether the range table entry will be\navailable in the executor for you to access that owner field. But at\nthis point, any fix for this would be great. People are asking for this\nview permissions thing.\n\nAlso, this makes non-super-user created views even harder, because if\npeople can create their own views, they can change the owner field to\nanyone they want to, but that is for later.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 10:29:43 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "\nBruce wrote:\n> >\n> > Got very close to the view permission override - just a vew hours\n> > for me :-)\n>\n> Great. I was thinking about this, and I think it would be best if the\n> view permissions were checked as the view owner at view USE time, not\n> view CREATION time, that way if permissions on the base tables change,\n> the permissions are properly honored.\n\n Did it that way as it looked better for me too. :-)\n\n>\n> I think the range table idea is good, so there is an OWNER field on each\n> range table which defaults to the current user. As views are replaced\n> by base tables in the rewrite system, the owner can be changed to the\n> owner of the view. The issue is whether the range table entry will be\n> available in the executor for you to access that owner field. But at\n> this point, any fix for this would be great. People are asking for this\n> view permissions thing.\n\n Done much easier. Appended a bool field aclSkip to\n RangeTblEntry that defaults to false due to makenode(). When\n the rewriting system finds a rule on select on the relation\n level, with exactly one action, the actions command type is a\n select which is instead (wow) it is a view. This time the\n rewrite handler checks acl for the range table entries in the\n rules action against the owner of the table the rule is on\n (the view itself). On success it sets aclSkip to true.\n\n Later the executor in ExecCheckPerms() just skips these\n entries. Since one range table entry for the view itself is\n left without the aclSkip set the executor checks if the\n current user has access rights for the view. But it skips\n those entries appended to the range table by the rewrite\n handler accessing the real tables in question.\n\n One little thing isn't covered then. User A creates a view on\n a table he revoked from world and grants access to the view\n only to user B. Now user B can create another view selecting\n A's view and grant that to world and this would do it. But\n since any user that can read a view could simply copy the\n data into another table and grant that to world I don't see a\n really security problem here. Granting access to user implies\n IMHO you can trust that user.\n\n>\n> Also, this makes non-super-user created views even harder, because if\n> people can create their own views, they can change the owner field to\n> anyone they want to, but that is for later.\n\n Do they - hmmm that's not good. But there could be a way\n round. Really for later but let's keep the solution in mind.\n\n We add a bool to pg_class that let's the rewrite handler know\n if he really should set the aclSkip defaulting to false. On\n ownership changes this flag is reset to false and only the\n owner or superusers might set it.\n\n\nUntil later, Jan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Fri, 20 Feb 1998 18:22:33 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> \n> \n> Bruce wrote:\n> > >\n> > > Got very close to the view permission override - just a vew hours\n> > > for me :-)\n> >\n> > Great. I was thinking about this, and I think it would be best if the\n> > view permissions were checked as the view owner at view USE time, not\n> > view CREATION time, that way if permissions on the base tables change,\n> > the permissions are properly honored.\n> \n> Did it that way as it looked better for me too. :-)\n\nGood.\n\n> \n> >\n> > I think the range table idea is good, so there is an OWNER field on each\n> > range table which defaults to the current user. As views are replaced\n> > by base tables in the rewrite system, the owner can be changed to the\n> > owner of the view. The issue is whether the range table entry will be\n> > available in the executor for you to access that owner field. But at\n> > this point, any fix for this would be great. People are asking for this\n> > view permissions thing.\n> \n> Done much easier. Appended a bool field aclSkip to\n> RangeTblEntry that defaults to false due to makenode(). When\n> the rewriting system finds a rule on select on the relation\n> level, with exactly one action, the actions command type is a\n> select which is instead (wow) it is a view. This time the\n> rewrite handler checks acl for the range table entries in the\n> rules action against the owner of the table the rule is on\n> (the view itself). On success it sets aclSkip to true.\n> \n> Later the executor in ExecCheckPerms() just skips these\n> entries. Since one range table entry for the view itself is\n> left without the aclSkip set the executor checks if the\n> current user has access rights for the view. But it skips\n> those entries appended to the range table by the rewrite\n> handler accessing the real tables in question.\n> \n> One little thing isn't covered then. User A creates a view on\n> a table he revoked from world and grants access to the view\n> only to user B. Now user B can create another view selecting\n> A's view and grant that to world and this would do it. But\n> since any user that can read a view could simply copy the\n> data into another table and grant that to world I don't see a\n> really security problem here. Granting access to user implies\n> IMHO you can trust that user.\n\nThis sounds great to me. As you have added to RangeTableEntry, I assume\nyou also added the needed fields to copyfuncs.c/outfuncs.c/makefuncs.c\nand anywhere else that needs it. I will add this to the developers FAQ.\n\n> \n> >\n> > Also, this makes non-super-user created views even harder, because if\n> > people can create their own views, they can change the owner field to\n> > anyone they want to, but that is for later.\n> \n> Do they - hmmm that's not good. But there could be a way\n> round. Really for later but let's keep the solution in mind.\n> \n> We add a bool to pg_class that let's the rewrite handler know\n> if he really should set the aclSkip defaulting to false. On\n> ownership changes this flag is reset to false and only the\n> owner or superusers might set it.\n\nNo, that is not what I was saying. If they can create views, the can\nchange pg_rewrite, and because we now take the view as the owners\npermission granting, someone could change anything in pg_rewrite and\nmake it look like it is a view of someone else. They could change the\nview text to look at pg_user, for example.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 14:12:46 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "\nBruce wrote:\n> This sounds great to me. As you have added to RangeTableEntry, I assume\n> you also added the needed fields to copyfuncs.c/outfuncs.c/makefuncs.c\n> and anywhere else that needs it. I will add this to the developers FAQ.\n\n Only to copyfuncs.c now. Is it possible that they get\n output/reread after deepRewriteQuery()?. I don't think so.\n And since makenode() initializes the allocated memory to\n nulls aclSkip defaults to false.\n\n> > We add a bool to pg_class that let's the rewrite handler know\n> > if he really should set the aclSkip defaulting to false. On\n> > ownership changes this flag is reset to false and only the\n> > owner or superusers might set it.\n>\n> No, that is not what I was saying. If they can create views, the can\n> change pg_rewrite, and because we now take the view as the owners\n> permission granting, someone could change anything in pg_rewrite and\n> make it look like it is a view of someone else. They could change the\n> view text to look at pg_user, for example.\n\n Instead of granting users access to pg_rewrite we should use\n again some mechanism in rewriteDefine/ExecCheckPerms to let\n users only create rules through the CREATE RULE command. Not\n by INSERT INTO pg_rewrite. Then we can check during the\n creation of a rule that the user is the owner of the relation\n the rule is on. Totally safe then.\n\n\nUntil later, Jan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Fri, 20 Feb 1998 21:18:30 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> \n> \n> Bruce wrote:\n> > This sounds great to me. As you have added to RangeTableEntry, I assume\n> > you also added the needed fields to copyfuncs.c/outfuncs.c/makefuncs.c\n> > and anywhere else that needs it. I will add this to the developers FAQ.\n> \n> Only to copyfuncs.c now. Is it possible that they get\n> output/reread after deepRewriteQuery()?. I don't think so.\n> And since makenode() initializes the allocated memory to\n> nulls aclSkip defaults to false.\n\nIt doesn't matter whether we think it will never be used. If it says it\ncopies a structure, we have to make it work. Never know how it will be\nused in the future. I went through all the copy/make/read/out files to\nmake sure every element of every structure was output, where possible.\n\nI have added this to the developers FAQ, and it is on our web page.\n\n> \n> > > We add a bool to pg_class that let's the rewrite handler know\n> > > if he really should set the aclSkip defaulting to false. On\n> > > ownership changes this flag is reset to false and only the\n> > > owner or superusers might set it.\n> >\n> > No, that is not what I was saying. If they can create views, the can\n> > change pg_rewrite, and because we now take the view as the owners\n> > permission granting, someone could change anything in pg_rewrite and\n> > make it look like it is a view of someone else. They could change the\n> > view text to look at pg_user, for example.\n> \n> Instead of granting users access to pg_rewrite we should use\n> again some mechanism in rewriteDefine/ExecCheckPerms to let\n> users only create rules through the CREATE RULE command. Not\n> by INSERT INTO pg_rewrite. Then we can check during the\n> creation of a rule that the user is the owner of the relation\n> the rule is on. Totally safe then.\n\nYep. Need this for pg_database too.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 15:51:36 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> > Only to copyfuncs.c now. Is it possible that they get\n> > output/reread after deepRewriteQuery()?. I don't think so.\n> > And since makenode() initializes the allocated memory to\n> > nulls aclSkip defaults to false.\n> \n> It doesn't matter whether we think it will never be used. If it says it\n> copies a structure, we have to make it work. Never know how it will be\n> used in the future. I went through all the copy/make/read/out files to\n> make sure every element of every structure was output, where possible.\n\n Correct. I'll do it along with the flag in pg_class.\n\n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Fri, 20 Feb 1998 21:58:15 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> \n> > > Only to copyfuncs.c now. Is it possible that they get\n> > > output/reread after deepRewriteQuery()?. I don't think so.\n> > > And since makenode() initializes the allocated memory to\n> > > nulls aclSkip defaults to false.\n> > \n> > It doesn't matter whether we think it will never be used. If it says it\n> > copies a structure, we have to make it work. Never know how it will be\n> > used in the future. I went through all the copy/make/read/out files to\n> > make sure every element of every structure was output, where possible.\n> \n> Correct. I'll do it along with the flag in pg_class.\n\nCan you remind me of the pg_class flag's purpose?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 16:21:19 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "If I ever get to run pgindent. I will make these changes, Tom.\n\n> Well, *&^*^#$! I wasted hours cleaning up some of what I considered damage from\n> the last pass through.\n> Would it be possible to pass back through (i.e. pipe indent output to a filter)\n> and fix at least points (1) and (2)? _That_ would be pretty easy to automate\n> since the heuristics can be pretty simple:\n> \n> if (first char is nonwhitespace) && (next word is \"*\") && (next line starts with\n> word+paren)\n> then compress whitespace\n> \n> else if (first char is \"}\") && (next word is \"/*\")\n> then compress whitespace\n> \n> else\n> write line as-is\n> \n> I'll write the perl if you would be willing to use it?\n> \n> - Tom\n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sat, 21 Feb 1998 21:04:49 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> If I ever get to run pgindent. I will make these changes, Tom.\n\nAhhhhh. Thx. :)\n\n - Tom\n\n",
"msg_date": "Mon, 23 Feb 1998 14:37:44 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> Well, *&^*^#$! I wasted hours cleaning up some of what I considered damage from\n> the last pass through.\n> Would it be possible to pass back through (i.e. pipe indent output to a filter)\n> and fix at least points (1) and (2)? _That_ would be pretty easy to automate\n> since the heuristics can be pretty simple:\n> \n> if (first char is nonwhitespace) && (next word is \"*\") && (next line starts with\n> word+paren)\n> then compress whitespace\n> \n> else if (first char is \"}\") && (next word is \"/*\")\n> then compress whitespace\n> \n> else\n> write line as-is\n> \n> I'll write the perl if you would be willing to use it?\n\nOK, I have changed pgindent to reflect the new structure names, and have\nmade your requested changes. For #1, if the line began with an alpha,\nand ends with a *, I remove tabs/spaces before the * and make it only\none space. Let me know how you like the changes.\n\nI have not run it on the source yet because people are still holding\npatches.\n\n\n-- \nBruce Momjian | 830 Blythe Avenue\[email protected] | Drexel Hill, Pennsylvania 19026\n + If your life is a hard drive, | (610) 353-9879(w)\n + Christ can be your backup. | (610) 853-3000(h)\n",
"msg_date": "Tue, 24 Feb 1998 19:25:28 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "On Tue, 24 Feb 1998, Bruce Momjian wrote:\n\n> OK, I have changed pgindent to reflect the new structure names, and have\n> made your requested changes. For #1, if the line began with an alpha,\n> and ends with a *, I remove tabs/spaces before the * and make it only\n> one space. Let me know how you like the changes.\n> \n> I have not run it on the source yet because people are still holding\n> patches.\n\nBruce, do you run pgindent on the java sources aswell?\n\nIf you do, a couple of points:\n\n\t/* */ are as in C\n\t/** same as /* but also marks the start of a comment (can be \n\t html) that javadoc will place in documentation.\n\t// Single line comment\n\nYou may want to check that it doesn't mangle the last two.\n\n-- \nPeter T Mount [email protected] or [email protected]\nMain Homepage: http://www.demon.co.uk/finder\nWork Homepage: http://www.maidstone.gov.uk Work EMail: [email protected]\n\n",
"msg_date": "Wed, 25 Feb 1998 06:49:43 +0000 (GMT)",
"msg_from": "Peter T Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Running pgindent"
},
{
"msg_contents": "> \n> On Tue, 24 Feb 1998, Bruce Momjian wrote:\n> \n> > OK, I have changed pgindent to reflect the new structure names, and have\n> > made your requested changes. For #1, if the line began with an alpha,\n> > and ends with a *, I remove tabs/spaces before the * and make it only\n> > one space. Let me know how you like the changes.\n> > \n> > I have not run it on the source yet because people are still holding\n> > patches.\n> \n> Bruce, do you run pgindent on the java sources aswell?\n> \n> If you do, a couple of points:\n> \n> \t/* */ are as in C\n> \t/** same as /* but also marks the start of a comment (can be \n> \t html) that javadoc will place in documentation.\n> \t// Single line comment\n> \n> You may want to check that it doesn't mangle the last two.\n\nNo, no run of C++ or Java.\n\n-- \nBruce Momjian | 830 Blythe Avenue\[email protected] | Drexel Hill, Pennsylvania 19026\n + If your life is a hard drive, | (610) 353-9879(w)\n + Christ can be your backup. | (610) 853-3000(h)\n",
"msg_date": "Wed, 25 Feb 1998 09:17:19 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Running pgindent\\"
}
] |
[
{
"msg_contents": "The command \ncopy pg_user to stdout;\n\nwill also show the cleartext password and I think it is hard to do a rewrite\nhere,\nsince this would also affect the pg_dump ?\n\n* Teardrops keep falling on my head ... * :-(\n\nAndreas\n> ----------\n> Von: \tJan Wieck[SMTP:[email protected]]\n> Antwort an: \tJan Wieck\n> Gesendet: \tDonnerstag, 19. Februar 1998 15:53\n> An: \tZeugswetter Andreas SARZ\n> Cc: \[email protected]\n> Betreff: \tRe: [HACKERS] Solution to the pg_user passwd problem !?? (c)\n> \n> >\n> > Hi all,\n> >\n> > What about:\n> > grant select on pg_user to public;\n> > create rule pg_user_hide_pw as on\n> > select to pg_user.passwd\n> > do instead select '********' as passwd;\n> >\n> > Then if I do:\n> > select * from pg_user;\n> > usename |usesysid|usecreatedb|usetrace|usesuper|usecatupd|passwd\n> |valuntil\n> >\n> --------+--------+-----------+--------+--------+---------+--------+-------\n> --\n> > -------------------\n> > postgres| 6|t |t |t |t |********|Sat\n> Jan\n> > 31 07:00:00 2037 NFT\n> > zeus | 60|t |t |f |t |********|\n> > (2 rows)\n> >\n> > Also the \\d works for all users !\n> >\n> > Only \"disadvantage\" is that noone can read passwd without first dropping\n> the\n> > rule pg_user_hide_pw,\n> > I consider this a feature though ;-)\n> >\n> > Since the userauthentication bypasses the rewrite mechanism the logins,\n> > alter user .. and others do work !\n> >\n> > Can all of you try to crack this ?\n> \n> Cracked!\n> \n> create table get_passwds (usename name, passwd text);\n> insert into get_passwds select usename, passwd from pg_user;\n> select * from get_passwds;\n> usename|passwd\n> -------+------\n> pgsql |\n> wieck |test\n> (2 rows)\n> \n> \n> \n> Sorry, Jan\n> \n> --\n> \n> #======================================================================#\n> # It's easier to get forgiveness for being wrong than for being right. #\n> # Let's break this rule - forgive me. #\n> #======================================== [email protected] (Jan Wieck) #\n> \n> \n> \n",
"msg_date": "Thu, 19 Feb 1998 16:08:41 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> The command \n> copy pg_user to stdout;\n> \n> will also show the cleartext password and I think it is hard to do a rewrite\n> here,\n> since this would also affect the pg_dump ?\n> \n> * Teardrops keep falling on my head ... * :-(\n\nIt was brilliant, even if it doesn't fully solve the problem.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 10:15:35 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > The command \n> > copy pg_user to stdout;\n> > \n> > will also show the cleartext password and I think it is hard to do a rewrite\n> > here,\n> > since this would also affect the pg_dump ?\n> > \n> > * Teardrops keep falling on my head ... * :-(\n> \n> It was brilliant, even if it doesn't fully solve the problem.\n\n\tIt solves enough of the problem for v6.3's release...I still think\nits important to get it so that a 'grant select' can be imposed on a view\nthat looks at a table that as a 'revoke all' on it...this would fix the\nproblem cleanly, and completely...\n\n\tThen you could do a 'copy all_users to stdout;' vs a 'copy\npg_users to stdout;'\n\n\n\n",
"msg_date": "Thu, 19 Feb 1998 10:59:45 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": ">\n> >\n> > The command\n> > copy pg_user to stdout;\n> >\n> > will also show the cleartext password and I think it is hard to do a rewrite\n> > here,\n> > since this would also affect the pg_dump ?\n> >\n> > * Teardrops keep falling on my head ... * :-(\n>\n> It was brilliant, even if it doesn't fully solve the problem.\n>\n> --\n> Bruce Momjian\n> [email protected]\n>\n\n Not only that it was brilliant - as stated in my last post it\n also showed that there is a bug in the rewrite system that\n must get fixed so we can use views in INSERT...SELECT!\n Please add this to the TODO.\n\n For the pg_user permission problem I can think about another\n solution. It's a feature still on my personal wishlist -\n setuid views.\n\n\nUntil later, Jan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 19 Feb 1998 17:06:45 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": ">\n> On Thu, 19 Feb 1998, Bruce Momjian wrote:\n>\n> > >\n> > > The command\n> > > copy pg_user to stdout;\n> > >\n> > > will also show the cleartext password and I think it is hard to do a rewrite\n> > > here,\n> > > since this would also affect the pg_dump ?\n> > >\n> > > * Teardrops keep falling on my head ... * :-(\n> >\n> > It was brilliant, even if it doesn't fully solve the problem.\n>\n> It solves enough of the problem for v6.3's release...I still think\n> its important to get it so that a 'grant select' can be imposed on a view\n> that looks at a table that as a 'revoke all' on it...this would fix the\n> problem cleanly, and completely...\n>\n> Then you could do a 'copy all_users to stdout;' vs a 'copy\n> pg_users to stdout;'\n\n Copy on a view doesn't return anything! I think it's in the\n coding of the copy command - as far as I remember the copy\n directly accesses the table bypassing the rule system. But\n the table underlying the view is empty and the values are\n faked into the view via the rewrite rule system.\n\n The 'grant select' on views is a IMHO urgent required\n feature. I'll take a look on the code checking permissions\n and the rewrite system.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 19 Feb 1998 17:24:06 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> >\n> > >\n> > > The command\n> > > copy pg_user to stdout;\n> > >\n> > > will also show the cleartext password and I think it is hard to do a rewrite\n> > > here,\n> > > since this would also affect the pg_dump ?\n> > >\n> > > * Teardrops keep falling on my head ... * :-(\n> >\n> > It was brilliant, even if it doesn't fully solve the problem.\n> >\n> > --\n> > Bruce Momjian\n> > [email protected]\n> >\n> \n> Not only that it was brilliant - as stated in my last post it\n> also showed that there is a bug in the rewrite system that\n> must get fixed so we can use views in INSERT...SELECT!\n> Please add this to the TODO.\n> \n> For the pg_user permission problem I can think about another\n> solution. It's a feature still on my personal wishlist -\n> setuid views.\n\nAdded to TODO:\n\n* Allow INSERT INTO ... SELECT ... FROM view to work\n\nWe already had:\n\n* Allow VIEW permissions to be set separately from the underlying tables\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 11:31:26 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> The command \n> copy pg_user to stdout;\n> \n> will also show the cleartext password and I think it is hard to do a rewrite\n> here,\n> since this would also affect the pg_dump ?\n\nOK, I have committed code that removes the REVOKE from initdb, and does\nnot allow them to do any adding or altering of users if there is a\npassword involved AND the ACL for pg_user is null. It prints a nice\nmessage telling them they need to issue the REVOKE command so normal\nusers can't read the passwords.\n\nIf they use user passwords psql \\d does fail if you do this. Do we want\nto duplicate the query without the pg_user reference if the \\d query\nfails?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 12:19:59 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > The command \n> > copy pg_user to stdout;\n> > \n> > will also show the cleartext password and I think it is hard to do a rewrite\n> > here,\n> > since this would also affect the pg_dump ?\n> \n> OK, I have committed code that removes the REVOKE from initdb, and does\n> not allow them to do any adding or altering of users if there is a\n> password involved AND the ACL for pg_user is null. It prints a nice\n> message telling them they need to issue the REVOKE command so normal\n> users can't read the passwords.\n\n\tI put the REVOKE back in, with the appropriate rule rewrite...I've\ntried it here and it works cleanly, and just masks out the passwd\nentry...doesn't compensate for the 'copy' problem, but its better then\nexpecting the admin to go do the revoke on his own :(\n\n\n",
"msg_date": "Thu, 19 Feb 1998 12:29:08 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "\nI wrote:\n> The 'grant select' on views is a IMHO urgent required\n> feature. I'll take a look on the code checking permissions\n> and the rewrite system.\n\n Interesting - first of all an unprivileged user cannot create\n any view \"pg_rewrite: Permission denied\". I think this is\n absolutely wrong.\n\n Anyway - if we add a flag to the rangetable entry that tells\n the executor in ExecCheckPerms() if this rte came from the\n rewriting due to a view or not, it can skip the permission\n check on that and the tests will pass.\n\n But then we'll run into a little security hole problem. If\n the permissions only rely on access to the view, the view\n owner (or public as long as ACL_WORLD_DEFAULT contains\n ACL_RD) can select throug the view. So we must check on view\n creation that the creator of the view has proper permissions\n to what the view selects. And in addition if not all objects\n the view selects are granted to public, we should\n automagically revoke public from the view so the creator must\n explicitly grant access to the view.\n\n Anything forgotten?\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 19 Feb 1998 19:02:26 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> \n> I wrote:\n> > The 'grant select' on views is a IMHO urgent required\n> > feature. I'll take a look on the code checking permissions\n> > and the rewrite system.\n> \n> Interesting - first of all an unprivileged user cannot create\n> any view \"pg_rewrite: Permission denied\". I think this is\n> absolutely wrong.\n> \n> Anyway - if we add a flag to the rangetable entry that tells\n> the executor in ExecCheckPerms() if this rte came from the\n> rewriting due to a view or not, it can skip the permission\n> check on that and the tests will pass.\n> \n> But then we'll run into a little security hole problem. If\n> the permissions only rely on access to the view, the view\n> owner (or public as long as ACL_WORLD_DEFAULT contains\n> ACL_RD) can select throug the view. So we must check on view\n> creation that the creator of the view has proper permissions\n> to what the view selects. And in addition if not all objects\n> the view selects are granted to public, we should\n> automagically revoke public from the view so the creator must\n> explicitly grant access to the view.\n> \n> Anything forgotten?\n\nNo, I think these are the valid issues.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 13:17:49 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> \n> > > \n> > > The command \n> > > copy pg_user to stdout;\n> > > \n> > > will also show the cleartext password and I think it is hard to do a rewrite\n> > > here,\n> > > since this would also affect the pg_dump ?\n> > \n> > OK, I have committed code that removes the REVOKE from initdb, and does\n> > not allow them to do any adding or altering of users if there is a\n> > password involved AND the ACL for pg_user is null. It prints a nice\n> > message telling them they need to issue the REVOKE command so normal\n> > users can't read the passwords.\n> \n> \tI put the REVOKE back in, with the appropriate rule rewrite...I've\n> tried it here and it works cleanly, and just masks out the passwd\n> entry...doesn't compensate for the 'copy' problem, but its better then\n> expecting the admin to go do the revoke on his own :(\n\nSorry, don't like it. First, by doing a REVOKE ALL and GRANT SELECT,\nyou have the same permissions as default, except the pg_user permissions\nare not null and therefore my check allows it.\n\nSecond, if COPY works, then passwords are not secure, and there is no\nreason for the feature. Either a feature is secure and valuable, or\nunsecure and worse then unvaluable because people think it is secure,\nand it is not.\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 13:20:06 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> > \n> > > > \n> > > > The command \n> > > > copy pg_user to stdout;\n> > > > \n> > > > will also show the cleartext password and I think it is hard to do a rewrite\n> > > > here,\n> > > > since this would also affect the pg_dump ?\n> > > \n> > > OK, I have committed code that removes the REVOKE from initdb, and does\n> > > not allow them to do any adding or altering of users if there is a\n> > > password involved AND the ACL for pg_user is null. It prints a nice\n> > > message telling them they need to issue the REVOKE command so normal\n> > > users can't read the passwords.\n> > \n> > \tI put the REVOKE back in, with the appropriate rule rewrite...I've\n> > tried it here and it works cleanly, and just masks out the passwd\n> > entry...doesn't compensate for the 'copy' problem, but its better then\n> > expecting the admin to go do the revoke on his own :(\n> \n> Sorry, don't like it. First, by doing a REVOKE ALL and GRANT SELECT,\n> you have the same permissions as default, except the pg_user permissions\n> are not null and therefore my check allows it.\n> \n> Second, if COPY works, then passwords are not secure, and there is no\n> reason for the feature. Either a feature is secure and valuable, or\n> unsecure and worse then unvaluable because people think it is secure,\n> and it is not.\n\n\tJust curious, but why don't the copy command fall under the same\ngrant/revoke restrictions in the first place? It sounds to me like we are\nbacking off of the problem instead of addressing it...\n\n\tThe problem being that it appears that 'copy' overrides/ignores\nthe rewrite rules, which kind of invalidates having them, doesn't it?\nWhat would it take to have copy follow them as select does?\n\n\n\n",
"msg_date": "Thu, 19 Feb 1998 13:36:03 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \tJust curious, but why don't the copy command fall under the same\n> grant/revoke restrictions in the first place? It sounds to me like we are\n> backing off of the problem instead of addressing it...\n> \n> \tThe problem being that it appears that 'copy' overrides/ignores\n> the rewrite rules, which kind of invalidates having them, doesn't it?\n> What would it take to have copy follow them as select does?\n\n Copy does a heap scan by itself. Doesn't care about rules.\n Instead copy must do a SELECT internal. Don't know if that\n is really what copy should do.\n \n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Thu, 19 Feb 1998 20:02:23 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \tJust curious, but why don't the copy command fall under the same\n> grant/revoke restrictions in the first place? It sounds to me like we are\n> backing off of the problem instead of addressing it...\n\ngrant/revoke works for copy.\n\n> \n> \tThe problem being that it appears that 'copy' overrides/ignores\n> the rewrite rules, which kind of invalidates having them, doesn't it?\n> What would it take to have copy follow them as select does?\n\nHard, because COPY does not do joins and does not go through the\noptimizer and executor. It dumps single tables. Tough to yank that\nthrough rewrite.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 14:13:30 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> > \tJust curious, but why don't the copy command fall under the same\n> > grant/revoke restrictions in the first place? It sounds to me like we are\n> > backing off of the problem instead of addressing it...\n> \n> grant/revoke works for copy.\n\n\tAh, okay, so when we have it setup so that a view overrides the\n'grant' of a select, then we're fine?\n\n\n",
"msg_date": "Thu, 19 Feb 1998 14:21:45 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> \n> > > \tJust curious, but why don't the copy command fall under the same\n> > > grant/revoke restrictions in the first place? It sounds to me like we are\n> > > backing off of the problem instead of addressing it...\n> > \n> > grant/revoke works for copy.\n> \n> \tAh, okay, so when we have it setup so that a view overrides the\n> 'grant' of a select, then we're fine?\n\nYep, but can we do that in nine days, and be sure it is tested?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 14:26:03 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> > \n> > > > \tJust curious, but why don't the copy command fall under the same\n> > > > grant/revoke restrictions in the first place? It sounds to me like we are\n> > > > backing off of the problem instead of addressing it...\n> > > \n> > > grant/revoke works for copy.\n> > \n> > \tAh, okay, so when we have it setup so that a view overrides the\n> > 'grant' of a select, then we're fine?\n> \n> Yep, but can we do that in nine days, and be sure it is tested?\n\n\tI don't think so...but I'rather have the obviuos \"select * from\npg_user\" closed off, and the more obscure \"copy pg_user to stdout\" still\nthere then have both wide open...its a half measure, but its better then\nno measure...\n\n\n",
"msg_date": "Thu, 19 Feb 1998 14:30:54 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> > \n> > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> > \n> > > > \tJust curious, but why don't the copy command fall under the same\n> > > > grant/revoke restrictions in the first place? It sounds to me like we are\n> > > > backing off of the problem instead of addressing it...\n> > > \n> > > grant/revoke works for copy.\n> > \n> > \tAh, okay, so when we have it setup so that a view overrides the\n> > 'grant' of a select, then we're fine?\n> \n> Yep, but can we do that in nine days, and be sure it is tested?\n\n Don't think so - not really hard tested.\n\n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n\n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Thu, 19 Feb 1998 20:45:46 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> \n> > > \n> > > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> > > \n> > > > > \tJust curious, but why don't the copy command fall under the same\n> > > > > grant/revoke restrictions in the first place? It sounds to me like we are\n> > > > > backing off of the problem instead of addressing it...\n> > > > \n> > > > grant/revoke works for copy.\n> > > \n> > > \tAh, okay, so when we have it setup so that a view overrides the\n> > > 'grant' of a select, then we're fine?\n> > \n> > Yep, but can we do that in nine days, and be sure it is tested?\n> \n> \tI don't think so...but I'rather have the obviuos \"select * from\n> pg_user\" closed off, and the more obscure \"copy pg_user to stdout\" still\n> there then have both wide open...its a half measure, but its better then\n> no measure...\n\nBut it is not secure. Why have passwords then?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 14:46:08 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> > \n> > > \n> > > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> > > \n> > > > > \tJust curious, but why don't the copy command fall under the same\n> > > > > grant/revoke restrictions in the first place? It sounds to me like we are\n> > > > > backing off of the problem instead of addressing it...\n> > > > \n> > > > grant/revoke works for copy.\n> > > \n> > > \tAh, okay, so when we have it setup so that a view overrides the\n> > > 'grant' of a select, then we're fine?\n> > \n> > Yep, but can we do that in nine days, and be sure it is tested?\n> \n> Don't think so - not really hard tested.\n\nIf you think you can get it in with light testing by the beta people,\nMarc might bite on it.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 14:47:33 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> > \n> > > > \n> > > > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> > > > \n> > > > > > \tJust curious, but why don't the copy command fall under the same\n> > > > > > grant/revoke restrictions in the first place? It sounds to me like we are\n> > > > > > backing off of the problem instead of addressing it...\n> > > > > \n> > > > > grant/revoke works for copy.\n> > > > \n> > > > \tAh, okay, so when we have it setup so that a view overrides the\n> > > > 'grant' of a select, then we're fine?\n> > > \n> > > Yep, but can we do that in nine days, and be sure it is tested?\n> > \n> > \tI don't think so...but I'rather have the obviuos \"select * from\n> > pg_user\" closed off, and the more obscure \"copy pg_user to stdout\" still\n> > there then have both wide open...its a half measure, but its better then\n> > no measure...\n> \n> But it is not secure. Why have passwords then?\n\n\tpassswords had to get in there at *some* point...they are there\nnow, now we have to extend the security to the next level. Better to move\nforward 1 step at a time. If we remove the REVOKE altogether, the\npasswords are still there, but there is *0* security instead of 50%\nsecurity...\n\n\tSo, I think we should leave the REVOKE/GRANT in initdb, and work\nat having grant/revoke work on a view (such that a view overrides the\nrevoke of all on pg_user) so that it is appliable *after* v6.3 is\nreleased, and available as (if possible) a patch for just after...\n\n\tWe aren't hurting anything by leaving the REVOKE/GRANT in place,\nbut I think we are if we remove it and just leave it wide open...\n\n\n",
"msg_date": "Thu, 19 Feb 1998 15:01:29 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > > \n> > > > \n> > > > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> > > > \n> > > > > > \tJust curious, but why don't the copy command fall under the same\n> > > > > > grant/revoke restrictions in the first place? It sounds to me like we are\n> > > > > > backing off of the problem instead of addressing it...\n> > > > > \n> > > > > grant/revoke works for copy.\n> > > > \n> > > > \tAh, okay, so when we have it setup so that a view overrides the\n> > > > 'grant' of a select, then we're fine?\n> > > \n> > > Yep, but can we do that in nine days, and be sure it is tested?\n> > \n> > Don't think so - not really hard tested.\n> \n> If you think you can get it in with light testing by the beta people,\n> Marc might bite on it.\n\n\tThere is nothing wrong with doing what we did last time...we have,\nwhat, 7 post-release patches for v6.2.1?\n\n\n",
"msg_date": "Thu, 19 Feb 1998 15:04:20 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \tpassswords had to get in there at *some* point...they are there\n> now, now we have to extend the security to the next level. Better to move\n> forward 1 step at a time. If we remove the REVOKE altogether, the\n> passwords are still there, but there is *0* security instead of 50%\n> security...\n\nIf we remove the REVOKE, then people will not use passwords by mistake,\nthinking they are secure. To use them, they have to issue a REVOKE, and\nthen they are secure.\n\nWhat am I missing here?\n\n> \n> \tSo, I think we should leave the REVOKE/GRANT in initdb, and work\n> at having grant/revoke work on a view (such that a view overrides the\n> revoke of all on pg_user) so that it is appliable *after* v6.3 is\n> released, and available as (if possible) a patch for just after...\n> \n> \tWe aren't hurting anything by leaving the REVOKE/GRANT in place,\n> but I think we are if we remove it and just leave it wide open...\n\nAgain, am I missing something?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 15:13:12 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "[Marc]\n\n> \tI don't think so...but I'rather have the obviuos \"select * from\n> pg_user\" closed off, and the more obscure \"copy pg_user to stdout\" still\n> there then have both wide open...its a half measure, but its better then\n> no measure...\n\n[Bruce]\n\n> But it is not secure. Why have passwords then?\n\n[Marc]\n\n> \tpassswords had to get in there at *some* point...they are there\n> now, now we have to extend the security to the next level. Better to move\n> forward 1 step at a time. If we remove the REVOKE altogether, the\n> passwords are still there, but there is *0* security instead of 50%\n> security...\n\nWrong. It's still *0* security, but with the illusion of working\nsecurity in the eyes of anyone who doesn't know better -- and you're\ntrying to keep them from knowing better. If you go this way, cases\n*will* occur where people think their data secure, and then someone\ngains access to it who shouldn't. Security by obscurity never was,\nand never will be a good idea.\n\nLeave wide open looking wide open, and document it. Say something\nlike \"This release has a password field in the pg_user table, but it\nisn't actually useful as a security measure. It's there because we\nintend to use it in a secure manner in future. Meanwhile, a secure\ninstallation of the current version can be achieved by ...\".\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n",
"msg_date": "Thu, 19 Feb 1998 21:29:50 +0100",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "Bruce said:\n> > > > Ah, okay, so when we have it setup so that a view overrides the\n> > > > 'grant' of a select, then we're fine?\n> > >\n> > > Yep, but can we do that in nine days, and be sure it is tested?\n> >\n> > Don't think so - not really hard tested.\n>\n> If you think you can get it in with light testing by the beta people,\n> Marc might bite on it.\n\n So I'll give it a try...\n\n\nHacking, Jan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Thu, 19 Feb 1998 22:27:15 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> Bruce said:\n> > > > > Ah, okay, so when we have it setup so that a view overrides the\n> > > > > 'grant' of a select, then we're fine?\n> > > >\n> > > > Yep, but can we do that in nine days, and be sure it is tested?\n> > >\n> > > Don't think so - not really hard tested.\n> >\n> > If you think you can get it in with light testing by the beta people,\n> > Marc might bite on it.\n> \n> So I'll give it a try...\n\nI have been hoping to get you back into the rewrite system so we can get\na feeling of what does and doesn't work properly there. Here is a good\nexcuse.\n\nI did some work on it with aggregates and subselects, and cleaned up\nsome stuff.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 16:35:59 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Tom I Helbekkmo wrote:\n\n> [Marc]\n> \n> > \tI don't think so...but I'rather have the obviuos \"select * from\n> > pg_user\" closed off, and the more obscure \"copy pg_user to stdout\" still\n> > there then have both wide open...its a half measure, but its better then\n> > no measure...\n> \n> [Bruce]\n> \n> > But it is not secure. Why have passwords then?\n> \n> [Marc]\n> \n> > \tpassswords had to get in there at *some* point...they are there\n> > now, now we have to extend the security to the next level. Better to move\n> > forward 1 step at a time. If we remove the REVOKE altogether, the\n> > passwords are still there, but there is *0* security instead of 50%\n> > security...\n> \n> Wrong. It's still *0* security, but with the illusion of working\n> security in the eyes of anyone who doesn't know better -- and you're\n> trying to keep them from knowing better. If you go this way, cases\n> *will* occur where people think their data secure, and then someone\n> gains access to it who shouldn't. Security by obscurity never was,\n> and never will be a good idea.\n> \n> Leave wide open looking wide open, and document it. Say something\n> like \"This release has a password field in the pg_user table, but it\n> isn't actually useful as a security measure. It's there because we\n> intend to use it in a secure manner in future. Meanwhile, a secure\n> installation of the current version can be achieved by ...\".\n\n\tI concede the argument...you guys win...*groan*\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Thu, 19 Feb 1998 19:08:39 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> > Leave wide open looking wide open, and document it. Say something\n> > like \"This release has a password field in the pg_user table, but it\n> > isn't actually useful as a security measure. It's there because we\n> > intend to use it in a secure manner in future. Meanwhile, a secure\n> > installation of the current version can be achieved by ...\".\n> \n> \tI concede the argument...you guys win...*groan*\n\nOK, back out those patches, nice and slowly. Keep your hands where I\ncan see them. :-)\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 18:23:08 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> > > Leave wide open looking wide open, and document it. Say something\n> > > like \"This release has a password field in the pg_user table, but it\n> > > isn't actually useful as a security measure. It's there because we\n> > > intend to use it in a secure manner in future. Meanwhile, a secure\n> > > installation of the current version can be achieved by ...\".\n> > \n> > \tI concede the argument...you guys win...*groan*\n> \n> OK, back out those patches, nice and slowly. Keep your hands where I\n> can see them. :-)\n\n\t*rofl* ... its done now :)\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Thu, 19 Feb 1998 21:44:08 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
}
] |
[
{
"msg_contents": "Here is my current list of supported platforms. The outstanding issues\nare:\n\naix/3.5 - anyone able to test? Success on aix4.x good enough??\nhpux/9.0.x - anyone testing? there are installation differences with\nhpux10.20. Supported?\nirix5 - is irix6 testing enough? Should we just say irix is supported?\ndec/alpha - any progress? Someone able to help on the inlining assembler\nproblems?\nlinux/alpha - any progress?\nmklinux/ppc - still core dumping on regression tests? why??\nsco - anyone able to test? If not, moving to unsupported list :(\nsvr4 - anyone able to test? If not, ...\nultrix4 - any users anymore? If not, ...\nnextstep - moving to unsupported list unless someone claims v6.3\nsuccess.\n\nAny platforms missing from the complete list below?\n\nI'd like to wrap this up by Feb. 25. Actually, I must finish by about\nthen to be able to put this info into new docs.\n\n - Tom\n\n* aix/4.1.4.0,4.2 - confirmed working when built on 4.1.4.0 (Darren\nKing)\n* aix/4.1.5 - confirmed working (Andreas Zeugswetter)\n_ aix/3.5 - not yet tested? (Frank Dana?)\n* bsdi - confirmed working (Bruce)\n* FreeBSD/2.2.x-3.x - confirmed working (Marc)\nx NetBSD/arm32 - not yet working (Dave Millen <[email protected]>)\n* NetBSD/i386 version? - confirmed working (Brook Milligan)\nx NetBSD/m68k Amiga, HP300, Mac - not yet working (Henry Hotz)\n* NetBSD/sparc/1.3 - confirmed working (Tom Helbekkmo)\n* NetBSD/vax/1.3 - confirmed working (Tom Helbekkmo)\n* dgux/5.4R4.11 - patches submitted (Brian Gallew)\n? hpux/9.0.x\n* hpux/10.20 - confirmed working (Stan Brown)\n_ irix5\n? irix6/MIPS - patches submitted (Andrew)\n_ dec/alpha - currently broken? confirmed working on v6.2.1 (Pedro)\n_ linux/alpha - partial success? (Ryan,Jeff)\n* linux/i386 - confirmed working (Thomas)\n* linux/sparc - confirmed working (Tom Szybist)\n_ mklinux/ppc - in progress (Tatsuo)\n_ sco\n* solaris/i386 - confirmed working (Marc)\n* solaris/sparc/2.5.1 - confirmed working (Marc)\n* solaris/sparc/2.6 - confirmed working (Marc)\n_ sunos/sparc/4.1.4 - in progress (Tatsuo)\n_ svr4/MIPS - dcosx and sinix/seimens-nixdorf worked on v6.1 (Frank\nRidderbusch?)\n_ ultrix4 - no recent reports? obsolete port??\nx univel - not working now; in progress? (Billy G. Allie)\n_ nextstep - worked with patches on v1.0.9; not working now?\nx MacOS - not library compatible; use ODBC/JDBC instead\nx nt - not library compatible; client side maybe; use ODBC/JDBC instead\n\nx windows - not library compatible; client side maybe; use ODBC/JDBC\ninstead\n\n",
"msg_date": "Thu, 19 Feb 1998 16:06:01 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Platform status"
},
{
"msg_contents": "> From: \"Thomas G. Lockhart\" <[email protected]>\n(...)\n> nextstep - moving to unsupported list unless someone claims v6.3\n> success.\n\nNS does only work as client. I have an EOAdaptor to use PostgreSQL from \nNeXT's Enterprise Objects Framework.\n\n---\n _ _\n _(_)(_)_ David Wetzel, Turbocat's Development,\n(_) __ (_) Buchhorster Strasse, D-16567 Muehlenbeck/Berlin, FRG,\n _/ \\_ Fax +49 33056 82835 NeXTmail [email protected]\n (______) http://www.turbocat.de/\n DEVELOPMENT * CONSULTING * ADMINISTRATION\n WATCH OUT FOR TURBOFAX for OPENSTEP!\n",
"msg_date": "Thu, 19 Feb 98 17:41:49 +0100",
"msg_from": "David Wetzel <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PORTS] Platform status"
},
{
"msg_contents": "-----BEGIN PGP SIGNED MESSAGE-----\n\nOn Thu, 19 Feb 1998, Thomas G. Lockhart wrote:\n\n> Here is my current list of supported platforms. The outstanding issues\n> are:\n> \n> aix/3.5 - anyone able to test? Success on aix4.x good enough??\n> hpux/9.0.x - anyone testing? there are installation differences with\n> hpux10.20. Supported?\n> irix5 - is irix6 testing enough? Should we just say irix is supported?\n> dec/alpha - any progress? Someone able to help on the inlining assembler\n> problems?\n> linux/alpha - any progress?\n> mklinux/ppc - still core dumping on regression tests? why??\n> sco - anyone able to test? If not, moving to unsupported list :(\n> svr4 - anyone able to test? If not, ...\n> ultrix4 - any users anymore? If not, ...\n> nextstep - moving to unsupported list unless someone claims v6.3\n> success.\n\nI've been strictly \"read only\" on these lists up to now, but I'd\nlike to offer my assistance on the DEC/Alpha and Ultrix4 ports,\nboth of which I use regularly (although I'm still at back revs on\nboth, right now.) I've also got access to AIX 4.1.4 and 4.2\n(maybe) as well as SunOS 5.5.1, if those would be of help.\n\nI'll pull down the latest sources, and see what I can do.\n\n- --\nDwayne Bailey + WHAT is your name? Sir Galahad\nMIKA Systems, Bingham Farms, MI + WHAT is your quest? I Seek the Holy Grail\[email protected] + What is your favorite color?\nhttp://www.mika.com/~dwayne + Blue ... no, Yelloooooooooooooooooow\n finger [email protected] for PGP Public Key\n\n-----BEGIN PGP SIGNATURE-----\nVersion: 2.6.2\n\niQB1AwUBNOxnEKA2uleK7maRAQHx4AL/ejbnE9qQmJZLrYaV8IavEzbsr/SW0xad\nQ7BuOtXguwy3p0HNBt3wQrjKycc5EYPEvblJHN6DYqIinrAew61guXcIWJljoId0\nHJeDtN1UgP9HuKTvfOS16Mym5g7hBOj8\n=gBAq\n-----END PGP SIGNATURE-----\n\n",
"msg_date": "Thu, 19 Feb 1998 12:08:18 -0500 (EST)",
"msg_from": "Dwayne Bailey <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Platform status"
},
{
"msg_contents": "\nWhat exactly are the problems, aside from the mkoidname deal in\ninitdb? I haven't heard anything about the inline assembler problems\n(isn't that for linux/alpha)?\n\nI can help.\n\nOn Thu, 19 February 1998, at 12:08:18, Dwayne Bailey wrote:\n\n> > dec/alpha - any progress? Someone able to help on the inlining assembler\n> > problems?\n",
"msg_date": "Thu, 19 Feb 1998 10:03:48 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Platform status"
},
{
"msg_contents": " Any platforms missing from the complete list below?\n\n * NetBSD/i386 version? - confirmed working (Brook Milligan)\n\nNetBSD/i386 v1.3 (the current release version)\n\nCheers,\nBrook\n",
"msg_date": "Thu, 19 Feb 1998 13:09:40 -0700 (MST)",
"msg_from": "Brook Milligan <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Platform status"
},
{
"msg_contents": "Thomas G. Lockhart writes:\n > Here is my current list of supported platforms. The outstanding issues\n > are:\n > \n > aix/3.5 - anyone able to test? Success on aix4.x good enough??\n > hpux/9.0.x - anyone testing? there are installation differences with\n > hpux10.20. Supported?\n > irix5 - is irix6 testing enough? Should we just say irix is supported?\n > dec/alpha - any progress? Someone able to help on the inlining assembler\n > problems?\n > linux/alpha - any progress?\n > mklinux/ppc - still core dumping on regression tests? why??\n > sco - anyone able to test? If not, moving to unsupported list :(\n > svr4 - anyone able to test? If not, ...\n ^^^^\n\nI tried with last Sundays snapshot (I think). I got it to compile,\nwith some minor fixes necessary. However I was yet unable to run the\nregression tests yet. It appears, I already have problems during the\ninitdb stage. I'm going to write a more detailed report next morning (my\nlocal time). \n\n > ultrix4 - any users anymore? If not, ...\n > nextstep - moving to unsupported list unless someone claims v6.3\n > success.\n > \n\n\nOne major problem is, that I can't connect with psql to the postmaster\nvia the UNIX domain socket. Judging from netstat (below) output the\npostmaster is listening ok.\n\n$ netstat -f unix\nActive UNIX domain sockets\nAddress Type Vnode Conn Addr\n805c6f9c stream 2204 0 /tmp/.s.PGSQL.5432\n\nAre there any SOLARIS specific ifdefs in the UNIX domain socket stuff?\nIf there are, I think it could very well be, that they also apply for\nSVR4. \n-- \nMfG/Regards\n\t\t\t\t Siemens Nixdorf Informationssysteme AG\n /==== Abt.: LoB OEC HES QM4\n / Ridderbusch / , Heinz Nixdorf Ring\n / /./ 33106 Paderborn, Germany\n /=== /,== ,===/ /,==, // Tel.: (49) 05251-8-15211\n / // / / // / / \\ NERV: ridderbusch.pad\n/ / `==/\\ / / / \\ Email: [email protected]\n\nSince I have taken all the Gates out of my computer, it finally works!!\n",
"msg_date": "Thu, 19 Feb 1998 20:32:14 GMT",
"msg_from": "Frank Ridderbusch <[email protected]>",
"msg_from_op": false,
"msg_subject": "[HACKERS] Platform status"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Thomas G. Lockhart wrote:\n\n>Here is my current list of supported platforms. The outstanding issues\n>are:\n>\n\n[...]\n\n>dec/alpha - any progress? Someone able to help on the inlining assembler\n>problems?\n\nNo progress yet, as far as I can tell. I've been unable to find the time\nto investigate the problem, and probably won't be able until\ntuesday/wednesday next week (too much \"real\" work :-( ).\n\n-------------------------------------------------------------------\nPedro José Lobo Perea Tel: +34 1 336 78 19\nCentro de Cálculo Fax: +34 1 331 92 29\nEUIT Telecomunicación - UPM e-mail: [email protected]\n\n",
"msg_date": "Fri, 20 Feb 1998 11:47:22 +0100 (MET)",
"msg_from": "\"Pedro J. Lobo\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PORTS] Platform status"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Thomas G. Lockhart wrote:\n\n> Here is my current list of supported platforms. The outstanding issues\n> are:\n> \n> linux/alpha - any progress?\n\n\tWe have the system up and running, though not anywhere near stable\nor production quality. Might put \"In progress, working, but only for the\nbrave of heart!'\n\tIt is still very much for developers/debuggers only (and those that\nabsolutely must have pgsql for Linux/Alpha ASAP). There are a number of\nfailed regression tests and compiler defines to still sort out. \n\tHope this answers your question.\n\n----------------------------------------------------------------------------\n| \"For to me to live is Christ, and to die is gain.\" |\n| --- Philippians 1:21 (KJV) |\n----------------------------------------------------------------------------\n| Ryan Kirkpatrick | Boulder, Colorado | [email protected] |\n----------------------------------------------------------------------------\n| http://www-ugrad.cs.colorado.edu/~rkirkpat/ |\n----------------------------------------------------------------------------\n\n",
"msg_date": "Fri, 20 Feb 1998 18:54:46 -0600 (CST)",
"msg_from": "Ryan Kirkpatrick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Platform status"
}
] |
[
{
"msg_contents": "Crypt sounds good. I like that Idea. I also use the unix crypt\nfunction to password protect httpd.\n\nRight now this is done with PERL doing the encryption and sticking\nit into the database in a char field. Then, another prog dumps\nthe passwords periodically into the httpd password file.\n\nThis would allow me to export the crypted passwords to the webserver\nas well? It would be convenient to be able to use the same password\nin both places.\n\n------- OTHER WISE ---- CAN WE STEAL IDEAS FROM OTHERS? --------\n\nI'm also running MySQL (trying to switch to PG, but waiting for ODBC\nto catch up), I like the mysql feature of inserting passords by:\n\n insert into user (host,user,password)\n values (\"domain.com\",\"dlorenso\",password('psswrd');\n\nthen if I\n\n select (host,user,password) * from user;\n\nI will get something like:\n\n...\n| domain.com | dlorenso | 2cfcb63e23e2d463 |\n...\n\nDante\n\n\n.------------------------------------------.-----------------------.\n| _ [email protected] - D. Dante Lorenso | Network Administrator |\n| | | ___ _ _ ___ __ _ ___ ___ | |\n| | |__ / o \\| '_|/ o_\\| \\ |\\_ _\\/ o \\ | Accounting Firms |\n| |____|\\___/|_| \\___/|_|\\_|\\___|\\___/ | Associated, inc. |\n| http://www.afai.com/~dlorenso | http://www.afai.com/ |\n'------------------------------------------'-----------------------'\n-----Original Message-----\nFrom: Brett McCormick <[email protected]>\nTo: Jan Wieck <[email protected]>\nCc: Zeugswetter Andreas SARZ <[email protected]>;\[email protected] <[email protected]>\nDate: Thursday, February 19, 1998 12:53 PM\nSubject: Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)\n\n\n>\n>Have we considering using the unix crypt function for passwords? That\n>way it wouldn't matter (as much) if people saw the password, and would\n>still be (somewhat less) secure.\n>\n>On Thu, 19 February 1998, at 15:55:07, Jan Wieck wrote:\n>\n>> Cracked!\n>>\n>> create table get_passwds (usename name, passwd text);\n>> insert into get_passwds select usename, passwd from pg_user;\n>> select * from get_passwds;\n>> usename|passwd\n>> -------+------\n>> pgsql |\n>> wieck |test\n>> (2 rows)\n>>\n>>\n>>\n>> Sorry, Jan\n>>\n>> --\n>>\n>> #======================================================================#\n>> # It's easier to get forgiveness for being wrong than for being right. #\n>> # Let's break this rule - forgive me. #\n>> #======================================== [email protected] (Jan Wieck) #\n>>\n>>\n>\n\n",
"msg_date": "Thu, 19 Feb 1998 13:08:37 -0500",
"msg_from": "\"D. Dante Lorenso\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> ------- OTHER WISE ---- CAN WE STEAL IDEAS FROM OTHERS? --------\n> \n> I'm also running MySQL (trying to switch to PG, but waiting for ODBC\n> to catch up), I like the mysql feature of inserting passords by:\n> \n> insert into user (host,user,password)\n> values (\"domain.com\",\"dlorenso\",password('psswrd');\n\nHmmm... When I read this, another (related) idea pops into my head.\nIt's probably nothing, but:\n\nwhat about defining a new type 'passwd', with\nan input (or output) routine which does the crypting?\n\nCheers,\nRonald\n",
"msg_date": "Fri, 20 Feb 1998 10:46:31 +0100 (MET)",
"msg_from": "Ronald Baljeu <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> > ------- OTHER WISE ---- CAN WE STEAL IDEAS FROM OTHERS? --------\n> > \n> > I'm also running MySQL (trying to switch to PG, but waiting for ODBC\n> > to catch up), I like the mysql feature of inserting passords by:\n> > \n> > insert into user (host,user,password)\n> > values (\"domain.com\",\"dlorenso\",password('psswrd');\n> \n> Hmmm... When I read this, another (related) idea pops into my head.\n> It's probably nothing, but:\n> \n> what about defining a new type 'passwd', with\n> an input (or output) routine which does the crypting?\n\nIf they do a binary cursor, does the output function get called?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 10:02:50 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> > \n> > > ------- OTHER WISE ---- CAN WE STEAL IDEAS FROM OTHERS? --------\n> > > \n> > > I'm also running MySQL (trying to switch to PG, but waiting for ODBC\n> > > to catch up), I like the mysql feature of inserting passords by:\n> > > \n> > > insert into user (host,user,password)\n> > > values (\"domain.com\",\"dlorenso\",password('psswrd');\n> > \n> > Hmmm... When I read this, another (related) idea pops into my head.\n> > It's probably nothing, but:\n> > \n> > what about defining a new type 'passwd', with\n> > an input (or output) routine which does the crypting?\n> \n> If they do a binary cursor, does the output function get called?\n\nOops, I haven't thought about that. I guess it doesn't :(\nOk, what about an input routine that does the crypting?\n\nOne down, one to go...\n\nCheers,\nRonald\n",
"msg_date": "Fri, 20 Feb 1998 16:21:22 +0100 (MET)",
"msg_from": "Ronald Baljeu <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
},
{
"msg_contents": "> \n> > > \n> > > > ------- OTHER WISE ---- CAN WE STEAL IDEAS FROM OTHERS? --------\n> > > > \n> > > > I'm also running MySQL (trying to switch to PG, but waiting for ODBC\n> > > > to catch up), I like the mysql feature of inserting passords by:\n> > > > \n> > > > insert into user (host,user,password)\n> > > > values (\"domain.com\",\"dlorenso\",password('psswrd');\n> > > \n> > > Hmmm... When I read this, another (related) idea pops into my head.\n> > > It's probably nothing, but:\n> > > \n> > > what about defining a new type 'passwd', with\n> > > an input (or output) routine which does the crypting?\n> > \n> > If they do a binary cursor, does the output function get called?\n> \n> Oops, I haven't thought about that. I guess it doesn't :(\n> Ok, what about an input routine that does the crypting?\n> \n> One down, one to go...\n> \n> Cheers,\n> Ronald\n> \n\nThen, the backend can't find the original password to test the random\nsalt on.\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 11:16:17 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
}
] |
[
{
"msg_contents": "does it matter if people are able to see the passwords? I mean,\nif the passwords are stored in a table (preferable encrypted), and\nthe table is only readable (select, insert, etc...) by the superuser\nor those of equal grant rights), then who cares?\n\nDante\n\n.------------------------------------------.-----------------------.\n| _ [email protected] - D. Dante Lorenso | Network Administrator |\n| | | ___ _ _ ___ __ _ ___ ___ | |\n| | |__ / o \\| '_|/ o_\\| \\ |\\_ _\\/ o \\ | Accounting Firms |\n| |____|\\___/|_| \\___/|_|\\_|\\___|\\___/ | Associated, inc. |\n| http://www.afai.com/~dlorenso | http://www.afai.com/ |\n'------------------------------------------'-----------------------'\n\n-----Original Message-----\nFrom: Brett McCormick <[email protected]>\nTo: Jan Wieck <[email protected]>\nCc: Zeugswetter Andreas SARZ <[email protected]>;\[email protected] <[email protected]>\nDate: Thursday, February 19, 1998 12:53 PM\nSubject: Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)\n\n\n>\n>Have we considering using the unix crypt function for passwords? That\n>way it wouldn't matter (as much) if people saw the password, and would\n>still be (somewhat less) secure.\n>\n>On Thu, 19 February 1998, at 15:55:07, Jan Wieck wrote:\n>\n>> Cracked!\n>>\n>> create table get_passwds (usename name, passwd text);\n>> insert into get_passwds select usename, passwd from pg_user;\n>> select * from get_passwds;\n>> usename|passwd\n>> -------+------\n>> pgsql |\n>> wieck |test\n>> (2 rows)\n>>\n>>\n>>\n>> Sorry, Jan\n>>\n>> --\n>>\n>> #======================================================================#\n>> # It's easier to get forgiveness for being wrong than for being right. #\n>> # Let's break this rule - forgive me. #\n>> #======================================== [email protected] (Jan Wieck) #\n>>\n>>\n>\n\n",
"msg_date": "Thu, 19 Feb 1998 13:15:18 -0500",
"msg_from": "\"D. Dante Lorenso\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
}
] |
[
{
"msg_contents": "Hi,\n\nThe problem is just the missing name column.\n\nI've just trashed my last build, which passed the regression tests\nperfectly (SPARC/Linux on an IPX), to do a fresh build from CVS.\n\nOnce this is complete I'll do some more tests.\n\nBasically the output looks something like.\n\n was born 19661202 (age=32)\n was born 19881202 (age=54)\n etc.\n etc.\n^^^^^^^^\n Just a blank field!! \n\nMaybe it's something in the structure handling?\n\nKeith.\n\n\"Meskes, Michael\" <[email protected]>\n> \n> Do you mean you didn't get any output? Or you got just the numbers? If\n> you didn't get anything that's the old problem and I believe the same\n> problem holds for ODBC (Julie?). The first fetch returns no row, so the\n> program ends. Unfortunately I haven't changed anything I expect to be\n> able to cause that problem. It seems to be a backend/library problem.\n> \n> Anyway, which version are you using? I cannot connect to a database\n> since last week or so. I tried re-initdb'ing it but get a seg fault in\n> the backend when doing a createuser. So my only chance of testing right\n> now is to manually inspect the code created.\n> \n> Michael\n> \n> --\n> Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> [email protected] | Europark A2, Adenauerstr. 20\n> [email protected] | 52146 Wuerselen\n> Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n> \n> > -----Original Message-----\n> > From:\tKeith Parks [SMTP:[email protected]]\n> > Sent:\tWednesday, February 18, 1998 8:28 PM\n> > To:\[email protected]\n> > Cc:\[email protected]\n> > Subject:\tRe: New ecgp code problem.\n> > \n> > Hi,\n> > \n> > Things were working upto a point when I was playing yesterday.\n> > \n> > The perftest seemed to run OK\n> > The test2 example ran but did not display the varchar (name) field.\n> > \n> > I'm afraid I've not got very far debugging.\n> > \n> > Keith.\n> > \n> > \n> > \n\n",
"msg_date": "Thu, 19 Feb 1998 19:52:57 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: New ecgp code problem."
},
{
"msg_contents": "Keith Parks writes:\n> Hi,\n> \n> The problem is just the missing name column.\n\nI think I know the problem. I will send a patch later (again) but I cannot\ntest it myself as I still cannot create a user. Hmm, maybe I should try as\npostgres user.\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Fri, 20 Feb 1998 09:41:13 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] RE: New ecgp code problem."
}
] |
[
{
"msg_contents": "VAdim,\n\nGreat work.\n\nKeith.\n\ndisks=> select * from tracks t1 where t1.artist = (\ndisks-> select distinct t2.artist from tracks t2 where t2.trackno = (\ndisks-> select max(t3.trackno) from tracks t3\ndisks-> )\ndisks-> );\ndiskid|trackno|artist |song\n------+-------+-------------+-----------------------------------\n 51| 1|Elvis Presley|Hound Dog \n .\n .\n .\n .\n .\n\n 248| 25|Elvis Presley|The Wonder Of You\n 248| 26|Elvis Presley|Bridge Over Troubled Water\n(73 rows)\n\ndisks=>\n\n\n\nVadim B. Mikheev <[email protected]>\n> \n> Sorry, I decided to copy subselect' tuple unconditionally...\n> \n> Vadim\n\n",
"msg_date": "Thu, 19 Feb 1998 20:33:42 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Unexpected subselect result."
}
] |
[
{
"msg_contents": "Bruce Momjian wrote:\n> \n> >\n> > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> >\n> > > >\n> > > > On Thu, 19 Feb 1998, Bruce Momjian wrote:\n> > > >\n> > > > > > Just curious, but why don't the copy command fall under the same\n> > > > > > grant/revoke restrictions in the first place? It sounds to me like we are\n> > > > > > backing off of the problem instead of addressing it...\n> > > > >\n> > > > > grant/revoke works for copy.\n> > > >\n> > > > Ah, okay, so when we have it setup so that a view overrides the\n> > > > 'grant' of a select, then we're fine?\n> > >\n> > > Yep, but can we do that in nine days, and be sure it is tested?\n> >\n> > I don't think so...but I'rather have the obviuos \"select * from\n> > pg_user\" closed off, and the more obscure \"copy pg_user to stdout\" still\n> > there then have both wide open...its a half measure, but its better then\n> > no measure...\n> \n> But it is not secure. Why have passwords then?\n> \n\tI think is better have the encrypted passwords and the salt in pg_user.\nI don't know if this will be bing a security hole :(\n\n\tRobson.\n\n> --\n> Bruce Momjian\n> [email protected]",
"msg_date": "Thu, 19 Feb 1998 18:07:07 -0300",
"msg_from": "Robson Paniago de Miranda <[email protected]>",
"msg_from_op": true,
"msg_subject": "[Fwd: Re: AW: [HACKERS] Solution to the pg_user passwd problem !??\n\t(c)]"
}
] |
[
{
"msg_contents": "Bruce Momjian wrote:\n> \n> > > But it is not secure. Why have passwords then?\n> > >\n> > I think is better have the encrypted passwords and the salt in pg_user.\n> > I don't know if this will be bing a security hole :(\n> >\n> \n> If we do this, then what does the frontend pass us?\n> \n> --\n> Bruce Momjian\n> [email protected]\n\n\tI was thinking in the backend pass the salt stored in pg_user to the\nfrontend, but doing that is (almost) the same as having the password\nstored in clear text. It was a bad idea :(\n\n\tRobson.\n",
"msg_date": "Thu, 19 Feb 1998 18:20:41 -0300",
"msg_from": "Robson Paniago de Miranda <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: AW: [HACKERS] Solution to the pg_user passwd problem !?? (c)"
}
] |
[
{
"msg_contents": "Hello,\n\tI'm newly into PostgreSQL from a SYBASE world and I haven't had a lot of\ntime to get into the code, so bare with me please.\n\tIs there a query optimizer in PostgreSQL? or if not are there plans for\none? And are there plans for anything like clustered indexes?\n\tI'm curious because one of the jobs I almost took a few years ago was\nworking on the query optimizer itself. And please forgive me current\nignorance of the architecture of your engine.\n\nThanks, Rick.\n+-------------------------------------------+\n| \"No matter where you go...there you are.\" |\n|�������������� -Buckaroo Banzai����������� |\n+-------------------------------------------+---------------------\nChapWin Consulting Inc.�������������������������� [email protected]\nRed Lake Internet / Earfalls Internet������������� www.chapwin.net\nBox 882�������������������������������������������� (807) 727-2606\nRed Lake, ON P0V 2M0��������������������������� FAX (807) 727-3594\n------------------------------------------------------------------\n\n",
"msg_date": "Thu, 19 Feb 1998 15:35:16 -0600",
"msg_from": "\"Rick Windham\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "New to PostgreSQL Question (Query Optimizer)"
},
{
"msg_contents": "> \n> Hello,\n> \tI'm newly into PostgreSQL from a SYBASE world and I haven't had a lot of\n> time to get into the code, so bare with me please.\n> \tIs there a query optimizer in PostgreSQL? or if not are there plans for\n> one? And are there plans for anything like clustered indexes?\n> \tI'm curious because one of the jobs I almost took a few years ago was\n> working on the query optimizer itself. And please forgive me current\n> ignorance of the architecture of your engine.\n\nWe have a first-class optimizer. If you look in the current sources,\nthe optimizer directory has a README that describes some of it. You\nwill need the 6.3/snapshot release from ftp.postgresql.org to see the\nREADME, or go to www.postgresql.org, and go to support/documentation,\nand choose the nice program flowchart. You can cruise around in there,\nand see the source and README's.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 16:43:21 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] New to PostgreSQL Question (Query Optimizer)"
}
] |
[
{
"msg_contents": "Hi,\n\nI've a new problem with ecpg.\n\nThe latest source fails to process test2.pgc.\n\nSeems like it's choking on the new \"whenever\" clauses.\n\n[postgres@sparclinux test]$ make\necpg test2.pgc\nparse error in line 8\nmake: *** [test2.c] Error 1 \n\nKeith.\n \n\n",
"msg_date": "Thu, 19 Feb 1998 21:40:14 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "ecpg error (19th Feb cvsup)"
},
{
"msg_contents": "Keith Parks writes:\n> Hi,\n> \n> I've a new problem with ecpg.\n> \n> The latest source fails to process test2.pgc.\n> \n> Seems like it's choking on the new \"whenever\" clauses.\n> \n> [postgres@sparclinux test]$ make\n> ecpg test2.pgc\n> parse error in line 8\n> make: *** [test2.c] Error 1 \n\nOops! ecpg is correct the statement in line 8 isn't. The correct syntax is \n\nexec sql whenever sqlerror do db_error(msg);\n\nNote the 'do'.\n\nI will send a patch later.\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Fri, 20 Feb 1998 09:39:25 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: ecpg error (19th Feb cvsup)"
}
] |
[
{
"msg_contents": "Just to let people know, pg_attribute.atttypmod is now passed as the\nthird parameter to all in and out functions, so it is ready to go for\npeople who need this functionality.\n\nWe did this by passing it through the backend from TypName and Resdom to\nTupleDesc.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 16:48:26 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "atttypmod"
},
{
"msg_contents": "\nWhich functionality is that?\n\nOn Thu, 19 February 1998, at 16:48:26, Bruce Momjian wrote:\n\n> Just to let people know, pg_attribute.atttypmod is now passed as the\n> third parameter to all in and out functions, so it is ready to go for\n> people who need this functionality.\n",
"msg_date": "Thu, 19 Feb 1998 18:59:43 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] atttypmod"
},
{
"msg_contents": "> \n> \n> Which functionality is that?\n> \n> On Thu, 19 February 1998, at 16:48:26, Bruce Momjian wrote:\n> \n> > Just to let people know, pg_attribute.atttypmod is now passed as the\n> > third parameter to all in and out functions, so it is ready to go for\n> > people who need this functionality.\n> \n\nThings like decimal types with defined precision. decimal(10,2).\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 22:04:34 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] atttypmod"
},
{
"msg_contents": "\nHow about having output options for types? I know this can be\nimplemented with functions and views, but it would be nice to able to\noutput values in different formats by default (such as datetime =>\nseconds since the epoch)\n\nOn Thu, 19 February 1998, at 22:04:34, Bruce Momjian wrote:\n\n> > Which functionality is that?\n> > \n> > On Thu, 19 February 1998, at 16:48:26, Bruce Momjian wrote:\n> > \n> > > Just to let people know, pg_attribute.atttypmod is now passed as the\n> > > third parameter to all in and out functions, so it is ready to go for\n> > > people who need this functionality.\n> > \n> \n> Things like decimal types with defined precision. decimal(10,2).\n> \n> -- \n> Bruce Momjian\n> [email protected]\n",
"msg_date": "Thu, 19 Feb 1998 19:09:55 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] atttypmod"
},
{
"msg_contents": "You could use atttypmod to have a type that has a default output format\nthat you specify at type creation time.\n\n> \n> \n> How about having output options for types? I know this can be\n> implemented with functions and views, but it would be nice to able to\n> output values in different formats by default (such as datetime =>\n> seconds since the epoch)\n> \n> On Thu, 19 February 1998, at 22:04:34, Bruce Momjian wrote:\n> \n> > > Which functionality is that?\n> > > \n> > > On Thu, 19 February 1998, at 16:48:26, Bruce Momjian wrote:\n> > > \n> > > > Just to let people know, pg_attribute.atttypmod is now passed as the\n> > > > third parameter to all in and out functions, so it is ready to go for\n> > > > people who need this functionality.\n> > > \n> > \n> > Things like decimal types with defined precision. decimal(10,2).\n> > \n> > -- \n> > Bruce Momjian\n> > [email protected]\n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 22:19:57 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] atttypmod"
},
{
"msg_contents": "Brett McCormick wrote:\n\n> How about having output options for types? I know this can be\n> implemented with functions and views, but it would be nice to able to\n> output values in different formats by default (such as datetime =>\n> seconds since the epoch)\n\nThis is a slippery slope. If you have various output options, then you also\nwould want to cleanly input those same outputs. There are already some output\noptions for date/time, to accomodate the world-wide variations in same. Having\na fully-general input/output capability for all data types which bypasses\nfunctions and views seems to be ignoring the power of those existing\nmechanisms...\n\notoh, we _do_ have to figure out how to implement the decimal/numeric type\nsomeday, so dive in :)\n\n - Tom\n\n",
"msg_date": "Fri, 20 Feb 1998 03:24:54 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] atttypmod"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Brett McCormick wrote:\n\n> \n> How about having output options for types? I know this can be\n> implemented with functions and views, but it would be nice to able to\n> output values in different formats by default (such as datetime =>\n> seconds since the epoch)\n\n\tdate_part('epoch', 'now'::datetime);\n\n\n> \n> On Thu, 19 February 1998, at 22:04:34, Bruce Momjian wrote:\n> \n> > > Which functionality is that?\n> > > \n> > > On Thu, 19 February 1998, at 16:48:26, Bruce Momjian wrote:\n> > > \n> > > > Just to let people know, pg_attribute.atttypmod is now passed as the\n> > > > third parameter to all in and out functions, so it is ready to go for\n> > > > people who need this functionality.\n> > > \n> > \n> > Things like decimal types with defined precision. decimal(10,2).\n> > \n> > -- \n> > Bruce Momjian\n> > [email protected]\n> \n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Thu, 19 Feb 1998 23:36:58 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] atttypmod"
},
{
"msg_contents": "\ner, at table creation time?\n\nOn Thu, 19 February 1998, at 22:19:57, Bruce Momjian wrote:\n\n> You could use atttypmod to have a type that has a default output format\n> that you specify at type creation time.\n",
"msg_date": "Thu, 19 Feb 1998 19:59:35 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] atttypmod"
},
{
"msg_contents": "> \n> \n> er, at table creation time?\n> \n> On Thu, 19 February 1998, at 22:19:57, Bruce Momjian wrote:\n> \n> > You could use atttypmod to have a type that has a default output format\n> > that you specify at type creation time.\n> \n\n\nOops, yep, table creation time.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 23:14:42 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] atttypmod"
}
] |
[
{
"msg_contents": "I have been working with many of you for over a hear, and would like to\nknow more details about you.\n\nI am 36, and live with my wife of eight years and two boys in a house\njust outside of Philadelphia, Pennsylvania. My boys are 5 and 2 1/2. I\nam a consultant who programs custom reports and applications for law\nfirms around the country. My boss and other seven other employees work\nin an office 10 minutes from here, but I work from my house almost all\nthe time. Perhaps twice a month, I go downtown to Philadelphia to work\nfor the day at the law firms. I have done this same work for the past\neight years. I have a BSDI Unix machine in my basement, and dumb\nterminals on the first and second floors. My office is in the basement.\n\nCan others tell us about themselves?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 22:47:20 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Who is everyone?"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> I have been working with many of you for over a hear, and would like to\n> know more details about you.\n> \n> I am 36, and live with my wife of eight years and two boys in a house\n> just outside of Philadelphia, Pennsylvania. My boys are 5 and 2 1/2. I\n> am a consultant who programs custom reports and applications for law\n> firms around the country. My boss and other seven other employees work\n> in an office 10 minutes from here, but I work from my house almost all\n> the time. Perhaps twice a month, I go downtown to Philadelphia to work\n> for the day at the law firms. I have done this same work for the past\n> eight years. I have a BSDI Unix machine in my basement, and dumb\n> terminals on the first and second floors. My office is in the basement.\n> \n> Can others tell us about themselves?\n\nI am 29, and live alone in a two bedroom apartment in Wolfville, Nova\nScotia, Canada. I work as Systems Administrator at Acadia University for\nthe \"steady paycheck\", and do contract programming and web hosting for the\n\"real money\".\n\nI run FreeBSD sitting in front of my TV so that I don't have to miss any\nof the good shows (what few are left), and have 220watt speakers hanging\noff of it as my \"stereo system\" so that I have good sound.\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Fri, 20 Feb 1998 00:06:11 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "\nI've just joined and haven't yet done much development, but I hope to\nsoon! I'm 19 and I work as a programmer and system administrator at\nthe Speakeasy Cafe (internet cafe) in Seattle. I developed our\naccounting system for keeping track of our members (we're an ISP,\nessentially), which uses PostgreSQL. As an aside, I just upgraded to\n6.2.1 but we had to revert to 1.09 (which has been very stable for\nmonths) due to backend core dumping problems. I didn't report them\nbecause 6.3 was being worked on, but at the same time, my boss doesn't\nthink it is a good idea to go to 6.3beta for our production system,\nespecially since we can't compile it on our Alpha (waste of hardware,\nyada yada). This is all totally irrelevant to who I am, but I figured\nyou guys would appreciate knowing all the same.\n\nAnd I'm a high school dropout who's been using computers since 1st\ngrade. I'm currently going to community college with plans of\nattending a university at some point (which is where cool things like\npostgresql get developed). I hope to become actively involved in the\nfree software community and refine my programming skills.\n\nAnd, I must say, you guys rock. You've done a great amount of\nwonderful work on PostgreSQL, and you deserve recognition (and a whole\nlot more).\n\nOn Thu, 19 February 1998, at 22:47:20, Bruce Momjian wrote:\n\n> I have been working with many of you for over a hear, and would like to\n> know more details about you.\n> \n> I am 36, and live with my wife of eight years and two boys in a house\n> just outside of Philadelphia, Pennsylvania. My boys are 5 and 2 1/2. I\n> am a consultant who programs custom reports and applications for law\n> firms around the country. My boss and other seven other employees work\n> in an office 10 minutes from here, but I work from my house almost all\n> the time. Perhaps twice a month, I go downtown to Philadelphia to work\n> for the day at the law firms. I have done this same work for the past\n> eight years. I have a BSDI Unix machine in my basement, and dumb\n> terminals on the first and second floors. My office is in the basement.\n> \n> Can others tell us about themselves?\n",
"msg_date": "Thu, 19 Feb 1998 21:26:14 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "> I have been working with many of you for over a year, and would like to\n> know more details about you.\n\nI'm 40 (yuck) and live with my wife and two cats near Pasadena, California.\nWe have five horses; my wife does show jumping and I do some trail riding\nand some polo. Used to do a lot of backpacking, cross-country skiing,\nwhitewater rafting and kayaking, but as we've accumulated horses we spend\nmost of our time on them nowadays. As you might guess, we don't have\nchildren.\n\nI went to school at Occidental College and Caltech, working part time at\nNASA's Jet Propulsion Laboratory until out of school, and full time ever\nsince. My background is in astronomy, and I've worked on VLBI (radio\ninterferometric) systems in the distant past. More recently, I designed and\nbuilt a world-wide network of GPS ground tracking stations, collecting data\nback to JPL for distribution to the international geodetic community. We\nalso used the network for some earth-orbiting spacecraft tracking. Just\nchanged jobs within JPL to work on optical interferometers, hopefully\nleading toward flying one in space to look for planets around other stars.\n\nOh, and I spend _way_ too much time working on Postgres :)\n\n - Tom\n\n",
"msg_date": "Fri, 20 Feb 1998 06:57:03 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n>I have been working with many of you for over a hear, and would like to\n>know more details about you.\n\nI am 26, and still live with my parents in Madrid. I work at the\nuniversity, in a telecommunications school (the same where I have\nstudied). I am a systems administrator, network administrator,\nprogrammer, consultant, and almost anything you can think of. I also work\n(sometimes) doing systems administration and programming out of the\nuniversity. My home computer has Windows 95, NT and FreeBSD installed, and\nI really really *HATE* Micro$oft :-). At the university, I administer a\nDigital Unix machine and a Windows NT network (it sucks).\n\nI use PostgreSQL for a database that contains all the information about\nour students. Currently, I am programming an account management system to\nkeep accounts in sync in Digital Unix and NT. The system contains a daemon\nin DU, a service in NT and a front-end written in Java, and it uses\nPostgreSQL to store the account data.\n\nMy contributions to PostgreSQL have been very small (just a few patches to\nget a clean compile in DU), but it's better than nothing :-) I'd like go\nget more involved when time permits. Anyway, the PostgreSQL team is doing\na *great* job. We have already a great DBMS, and it is getting better day\nby day.\n\n-------------------------------------------------------------------\nPedro José Lobo Perea Tel: +34 1 336 78 19\nCentro de Cálculo Fax: +34 1 331 92 29\nEUIT Telecomunicación - UPM e-mail: [email protected]\n\n",
"msg_date": "Fri, 20 Feb 1998 12:28:32 +0100 (MET)",
"msg_from": "\"Pedro J. Lobo\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "> I have been working with many of you for over a year, and would like to\n> know more details about you.\n\nWell, I am 47 and have been playing with computers for years. I own\na company that does contract software as well as running an ISP.\nI wrote my first program in Fortran IV in 1969 and have been published\nin a book on C programming. I ride a motorcycle for fun and I play\nguitar. My wife and I are foster parents for cats.\n\nMore on my web page if you are interested.\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 424 2871 (DoD#0082) (eNTP) | what's for dinner.\n",
"msg_date": "Fri, 20 Feb 1998 06:40:36 -0500 (EST)",
"msg_from": "[email protected] (D'Arcy J.M. Cain)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": ">\n> I have been working with many of you for over a hear, and would like to\n> know more details about you.\n\n I'm 33 and live with my wife and two boys aged 8 and 9 in\n Hamburg, Germany. In april our house will get finished and\n we'll live then in Harsefeld (small town of about 17000\n people 50km from Hamburg). I like beer, scotch and any hot\n and spicy meal. I'm driving Vespa and when the damn\n hack/compile/regress/hack cycle never seems to end I play\n some backgammon matches on FIBS.com (telnet on port 4321).\n\n I work for 8 years now as an SAP base consultant, programmer\n and administrator.\n\n I'm using and hacking on Postgres since the good old version\n 4.2 days. That was the last UCB release prior to Postgres95\n with the good old postquel query language where many of the\n newbe-confusing terms like tuples, attributes, append and\n retrieve come from. I still have such a dino running :-).\n\n\nUntil later, Jan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Fri, 20 Feb 1998 12:47:59 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "> \n> Oh, and I spend _way_ too much time working on Postgres :)\n> \n\nMany of us fit into that category. Problem is, PostgreSQL is way more\ninteresting than my day job. Not that my day work is boring, but\nPostgreSQL is very interesting for me. I am learning so much, almost\nlike a Comp Sci masters degree.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 09:58:09 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "> I have been working with many of you for over a hear, and would like to\n> know more details about you.\n\nOk,\n\nI haven't contributed to Postgres much, but here goes.\n\nI'm 26, live with my mom and little brother in the Netherlands.\nAnd ehm... I'm still single <grmph!>.\n\nAbout 12 years ago I started programming on a TRS-80, after that\nI bought me a MSX (the M from Micros*ft, little did I know...).\nThen, in 1989, I did informatics and got familiar with Unix\nand Internet. The next thing I did, was installing a Unix-clone\n(MiNT OS + GNU software) on my Atari-ST, which I had then. I graduated\nin 1995 and started working at the university hospital in Rotterdam.\n\nAfter 1-2 years I managed to convince the people at my department to\ngo completely freeware. We run an archiving system for patient studies\n(images from gamma camera's), which is built on a Linux PC that runs\nPostgres. PHP/FI is used to access Postgres' data from a\nWWW-browser. Apache is used as WWW-server. The browser is usually\nNetscape. We also experiment with Java, using JDK from Sun and Peter Mount's\nJDBC driver. Using a callback facility doctors can view images at home,\nin case of an emergency/urgent call. I wrote the PHP/FI-scripts, the Java\nprogram's and the C-program's to access the database, to update it\nautomatically, to convert proprierty format images to, for example, GIF,\nto adjust the colors, contrast etc...\n\nOther hobby's I have, besides computers, are electronics, making music\nwith MIDI, riding a bicycle, but best of all: just doing nothing (besides\nlistening to music) on a sunny day... It's just that I haven't much spare\ntime left to do so. And of course you need to have good weather...\n\nCheers,\nRonald\n",
"msg_date": "Fri, 20 Feb 1998 16:03:53 +0100 (MET)",
"msg_from": "Ronald Baljeu <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "Hi:\n\nI'm D. Jay Newman (the \"D.\" is silent) :).\n\nI'm 40, I earn money as a Research Programmer for Penn State, and\nthow money away by running Sprucegrove Internet Marketing, Inc.\n(a general web/net consulting firm).\n\nMy wife and I live in State College, PA, USA, with two dogs (Chaos and\nAmber), and a parrot (Random).\n\nI use PostgreSQL for as much of my Penn State work as possible (I'm\nworking on a communications package <http://projects.cac.psu.edu/ct/> called\nCourseTalk; this package uses PostgreSQL to store all of the information,\nwhich makes my life easier.\n\nI'm also working on a (very different) communications package for\nSprucegrove, which is designed for interactive fiction/gaming.\n\nI'm also trying to be an author of technical books. We'll see how that\ngoes.\n-- \nD. Jay Newman ! For the pleasure and the profit it derives\[email protected] ! I arrange things, like furniture, and\nhttp://www.sprucegrove.com/~jay/ ! daffodils, and ...lives. -- Hello Dolly\n",
"msg_date": "Fri, 20 Feb 1998 11:08:37 -0500 (EST)",
"msg_from": "\"D. Jay Newman\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everybody?"
},
{
"msg_contents": "-----BEGIN PGP SIGNED MESSAGE-----\n\n> I have been working with many of you for over a hear, and would like to\n> know more details about you.\n\nWell, unlike most of the others who have responded, I haven't\ndone a blessed thing for the project. At least, not yet. I'm\nhoping that will change, hence my introduction now.\n\nI'm currently a programmer/SysAdmin/webmaster/release\nengineer/etc/etc/etc \"on consignment\" on Ford Motor Company, just\noutside of Detroit, Michigan. Our project involves \"webifying\"\nsome of Ford's CAE tools, specifically the simulation of the\nstamping of body parts. I do some of the programming, and most\nof the admin work on our IBM RS/6000's and Sun Ultra 30's\n\nMy use of PostgreSQL has been limited to a couple of small \"in\nhouse\" projects at the company who actually employs me, MIKA\nSystems, plus a project or two I'm doing on my own, on DEC Alpha\nand Ultrix boxes, and a PC running Linux.\n\nOn the personal side, I'm 37, married, with one child, a\ndaughter, who's 8 years old. In my free time, I work at training\nmy dogs (collies) for obedience competitions. (We're just\nstarting out, so we haven't earned any titles yet.) Also, from\nmy .sig, you may be able to deduce that I'm a fan of Monty\nPython.\n\n- --\nDwayne Bailey + WHAT is your name? Sir Galahad\nMIKA Systems, Bingham Farms, MI + WHAT is your quest? I Seek the Holy Grail\[email protected] + What is your favorite color?\nhttp://www.mika.com/~dwayne + Blue ... no, Yelloooooooooooooooooow\n finger [email protected] for PGP Public Key\n\n\n-----BEGIN PGP SIGNATURE-----\nVersion: 2.6.2\n\niQB1AwUBNO2s7aA2uleK7maRAQHLcwL/fYUcxiMcbqnP4xMyBA97RZQeqRkuJTkf\nzZ0xEyEabgBNdDXG0CV43zRE4M0ZfVSYBQ76r5AoLogKIVKFGJex0xtqRcaqeCzo\nQKt61mpw5uGffzYK6y7wDvbhC4EvGRBh\n=xeG0\n-----END PGP SIGNATURE-----\n\n",
"msg_date": "Fri, 20 Feb 1998 11:18:45 -0500 (EST)",
"msg_from": "Dwayne Bailey <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "> On the personal side, I'm 37, married, with one child, a\n> daughter, who's 8 years old. In my free time, I work at training\n> my dogs (collies) for obedience competitions. (We're just\n> starting out, so we haven't earned any titles yet.) Also, from\n> my .sig, you may be able to deduce that I'm a fan of Monty\n> Python.\n\nI am noticing a lot of animal lovers in the group.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 12:00:15 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "On Fri, 20 Feb 1998, Bruce Momjian wrote:\n\n> > On the personal side, I'm 37, married, with one child, a\n> > daughter, who's 8 years old. In my free time, I work at training\n> > my dogs (collies) for obedience competitions. (We're just\n> > starting out, so we haven't earned any titles yet.) Also, from\n> > my .sig, you may be able to deduce that I'm a fan of Monty\n> > Python.\n> \n> I am noticing a lot of animal lovers in the group.\n\n\tWe have to get our affections somewhere, since SO's don't handle\nour long hours very well? :)\n\n\n",
"msg_date": "Fri, 20 Feb 1998 12:11:20 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "\n\nHi, Although I do not contribute to code writting, im there in spirte.\nI am a jack of all trades. I have done Unix Admin and DBA consulting before\nI joined my current employer. I currently am working for a company that \ndoes work for the NFL. So my interests in postgres are for a fanasty football\ndatabase, amongst other things. I have been working on/in/with Linux since\nDec 93, and Postgress for about a year and a half now. Keep up the good work\nguys. I am currently trying to start a DBA consulting company, based on\nmy Oracle, DB2, Postgres, and Msql expertise. Oh ya, Im married and I have\none son(Andrew 2yrs old) and a little girl is on the way. My son, also likes\nlinux( especially xfishtank)!!!\n\n\nj.heil\n-- \nJoseph A. Heil, Jr.\n\nSuperStat, Inc.\t\t\temail: [email protected]\n7500 Market Place Drive\t\tvoice: 612-943-8400\nEden Prairie, MN 55344\t\tfax: 612-943-8300\n\nKey fingerprint = 95 FC 3A F4 8A 10 05 85 3F 53 01 86 AD DB DB 51\n",
"msg_date": "Fri, 20 Feb 1998 11:20:44 -0600",
"msg_from": "Joseph Heil <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "I wasn't going to join this thread, until Bruce wrote:\n\n> I am noticing a lot of animal lovers in the group.\n\nI've noticed that as well, with pleasure. My wife and I have six cats\nliving with us (and a seventh sort of living in our yard, being an old\nhomeless stray). Might a fondness for animals be another trait shared\nby computer oriented people?\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n",
"msg_date": "Fri, 20 Feb 1998 21:01:06 +0100",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> I have been working with many of you for over a hear, and would like to\n> know more details about you.\n\n\tOk, I haven't done a whole lot to help out pgsql, just some\nsupport/encouragment for the Linux/alpha port and a few patches, but...\n\tI am 21 years old, and a undergrad college student at a small,\nchristian, engineering university called LeTourneau University in\nLongview, TX. I am originally from Boulder, Colorado (a much more\ninteresting town than Longview). I spend most of my time doing homework,\nand the rest hacking around on one of my four Linux boxes (P100, 386, UDB\nAlpha, 486 Thinkpad) or trying to convince the school's information\ntechnology department to install a Linux server for academic student use.\nFor some reason they think Novell & Microsoft are God, and Unix is evil!\nBut I am winning, slowly but surely. I do web page design, C++ programing,\nperl programming, web/cgi programming, system admin, and I find myself\ndoing more and more database programming as well (hence my like of pgsql).\nI also have an electrical engineering side, where I play with circuits and\nsee how many different ways I can blow them up! :) Between computers and\nelectronics (not to mention Linux), I keep quite busy. If you want to know\nmore, visit my web page. Thanks.\n\n----------------------------------------------------------------------------\n| \"For to me to live is Christ, and to die is gain.\" |\n| --- Philippians 1:21 (KJV) |\n----------------------------------------------------------------------------\n| Ryan Kirkpatrick | Boulder, Colorado | [email protected] |\n----------------------------------------------------------------------------\n| http://www-ugrad.cs.colorado.edu/~rkirkpat/ |\n----------------------------------------------------------------------------\n\n",
"msg_date": "Fri, 20 Feb 1998 19:02:23 -0600 (CST)",
"msg_from": "Ryan Kirkpatrick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "On Thu, 19 Feb 1998, Bruce Momjian wrote:\n\n> I have been working with many of you for over a hear, and would like to\n> know more details about you.\n\nI'm 27, single :-( , and currently live with my parents just outside\nMaidstone, Kent in England. \n\nI work for the Maidstone Borough Council (local government), officially as\nfirst line support for our users. However, I tend to second line support\nand development most of my time. (I'm currently trying to get my home\nnetwork connected to the works network). \n\nI've setup part of our Intranet (based around two Linux boxes), setup\nsecure dial up for the Mainframe & Unix boxes (again using Linux). Also,\nthe authorities web site is currently down to me. \n \nI'm using PostgreSQL at work currently to hold our purchase ordering, and\ndetails on the 400+ PC's we currently have. \n\nAt home I write Astronomical utilities mainly in Java, sometimes in C or\nPerl. Postgres is used to store Astronomical data (mainly catalogues at\nthe moment, but soon images as well), research on the Vikings, and to\ndevelop the JDBC driver. \n\nI also lurk on the TASS list, where they are using Postgresql to store\nobservations made by CCD cameras from around the world.\n\nI have networked here a Linux box (P133, 48Mb ram), a Windows 95 machine,\na 486 Laptop and a A1200 Amiga. A second linux box is being planned to\nhandle postgres & cdr work.\n\nOther pastimes include being a member of CAMRA (Campaign for Real Ale),\nsampling various real ales, and being a trainee Viking Warrior (basically\ngoing all over the country, dressing up in tenth century costume, and\nhaving large battles - and no the weapons are not foam ;-) )\n\n-- \nPeter T Mount [email protected] or [email protected]\nPage me via SMS on: http://www.demon.co.uk/finder/misc/sms.html\nMain Homepage: http://www.demon.co.uk/finder\nWork Homepage: http://www.maidstone.gov.uk Work EMail: [email protected]\n\n",
"msg_date": "Sat, 21 Feb 1998 10:31:55 +0000 (GMT)",
"msg_from": "Peter T Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "Thus spake Tom I Helbekkmo\n> > I am noticing a lot of animal lovers in the group.\n> I've noticed that as well, with pleasure. My wife and I have six cats\n> living with us (and a seventh sort of living in our yard, being an old\n> homeless stray). Might a fondness for animals be another trait shared\n> by computer oriented people?\n\nCould be. Over the last two years we have had over 50 cats in our\nhouse and we can name each one of them. We have had as many as 10\nat one time. We foster them for a local group that rescues them\nfrom lab research.\n\nSee http://www.druid.net/cats/ for details.\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 424 2871 (DoD#0082) (eNTP) | what's for dinner.\n",
"msg_date": "Sat, 21 Feb 1998 07:00:12 -0500 (EST)",
"msg_from": "[email protected] (D'Arcy J.M. Cain)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "\nI am 35, and live in the very north of Sweden in a small village\ncalled Jukkasj�rvi, 25 km outside the mining town of Kiruna.\nThis is about 200 km north of the arctic circle.\n\nI am currently (more or less) single in my little house near the Torne river \nand enjoys sauna, skiing, fishing and dancing (especially tango).\n\nI work as development and system engineer and in a small outfit\nas ours it means doing all sort of things. Our main business is\non-line image databases, tied to the mother companies printing\nand prepress shops.\nhttp://www.bildbasen.se/\nhttp://www.jms-gruppen.com/\n\nMy main areas of knowledges is \nunix and communication, \nC and perl programming.\n\nWe are slowly shutting down our old DG/Aviion/Informix-system and\nporting our software to linux/pgsql.\nI am responsable for the new unix socket code in 6.3 and hope to\nhelp out some more on the following releases.\n\n regards,\n-- \n---------------------------------------------\nG�ran Thyni, sysadm, JMS Bildbasen, Kiruna\n\n",
"msg_date": "26 Feb 1998 17:12:06 -0000",
"msg_from": "Goran Thyni <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Who is everyone?"
},
{
"msg_contents": "hi everybody,\n\na couple of days ago i stopped receiveing mail from {questions,hackers}\nlists. i resubscribed, but no answer at all. i did it several times, my\nmailer reported the email reached destination, but no efffect.\n\nwhat can i do?\n\ni remeber this is the second or even third time i got this problem, but\nnever so radical and persistent.\n\n\nCostin\n\n",
"msg_date": "Thu, 12 Mar 1998 11:10:43 +0200 (EET)",
"msg_from": "PostgreSQL <[email protected]>",
"msg_from_op": false,
"msg_subject": "Failing to get email from list or to susbscribe"
},
{
"msg_contents": "On Thu, 12 Mar 1998, PostgreSQL wrote:\n\n> hi everybody,\n> \n> a couple of days ago i stopped receiveing mail from {questions,hackers}\n> lists. i resubscribed, but no answer at all. i did it several times, my\n> mailer reported the email reached destination, but no efffect.\n> \n> what can i do?\n> \n> i remeber this is the second or even third time i got this problem, but\n> never so radical and persistent.\n\n\tThis mail won't reach you either, most likely:\n\nMar 11 16:15:28 hub sendmail[8151]: QAA08125: \\\n\[email protected], delay=00:00:01, \\\n\txdelay=00:00:00, mailer=smtp, relay=linux.tpd.deuroconsult.ro, \\\n\tstat=Host unknown (Name server: linux.tpd.deuroconsult.ro: host \\\n\tnot found) \n\n# nslookup linux.tpd.deuroconsult.ro\nServer: hub.org\nAddress: 209.47.148.200\n\n*** hub.org can't find linux.tpd.deuroconsult.ro: Non-existent host/domain\n\n\n",
"msg_date": "Thu, 12 Mar 1998 07:40:46 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Failing to get email from list or to susbscribe"
}
] |
[
{
"msg_contents": "> > Maybe it is used a lot on that chat thing. I can't remember the name\n> > right now. Not Usenet, but that thing where people discuss in real\n> > time.\n> \n> \tIRC? :)\n> \n\nYep, that's it.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Thu, 19 Feb 1998 22:49:06 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [CORE] Al Dev"
}
] |
[
{
"msg_contents": "\nSomeone want to tell me why the below CREATE INDEX fails? \n\nI've tried everything I can think of, and can't get that INDEX to be\ncreated :(\n\n-- $Id: groups.psql,v 1.2 1997/08/14 20:05:08 shevett Exp $\n-- postgresql 6.1\n--\n-- Host: localhost Database: pts\n----------------------------------------------------------\n\n\n--\n-- Table structure for table 'groups'\n--\nCREATE TABLE groups (\n id CHAR(10) NOT NULL,\n detail CHAR(30)\n) \\g\n\nCREATE INDEX groups_indx ON groups ( id ) \\g\n\n\n--\n-- Dumping data for table 'groups'\n--\n\nINSERT INTO groups (id, detail) VALUES ('PTS','PTS Development')\\g\n\n\n",
"msg_date": "Fri, 20 Feb 1998 00:58:08 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "odd error creating index in -current..."
},
{
"msg_contents": "\nI may be missing something, but because char_ops has been left out?\n\nOn Fri, 20 February 1998, at 00:58:08, The Hermit Hacker wrote:\n\n> Someone want to tell me why the below CREATE INDEX fails? \n> \n> I've tried everything I can think of, and can't get that INDEX to be\n> created :(\n> \n> -- $Id: groups.psql,v 1.2 1997/08/14 20:05:08 shevett Exp $\n> -- postgresql 6.1\n> --\n> -- Host: localhost Database: pts\n> ----------------------------------------------------------\n> \n> \n> --\n> -- Table structure for table 'groups'\n> --\n> CREATE TABLE groups (\n> id CHAR(10) NOT NULL,\n> detail CHAR(30)\n> ) \\g\n> \n> CREATE INDEX groups_indx ON groups ( id ) \\g\n> \n> \n> --\n> -- Dumping data for table 'groups'\n> --\n> \n> INSERT INTO groups (id, detail) VALUES ('PTS','PTS Development')\\g\n> \n> \n",
"msg_date": "Thu, 19 Feb 1998 21:28:00 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] odd error creating index in -current..."
},
{
"msg_contents": "On Thu, 19 Feb 1998, Brett McCormick wrote:\n\n> \n> I may be missing something, but because char_ops has been left out?\n\n\tLast I knew, it wasn't required...it auto-detects it, and is, in\nfact, the recommended method of creating an index...\n\n\n> \n> On Fri, 20 February 1998, at 00:58:08, The Hermit Hacker wrote:\n> \n> > Someone want to tell me why the below CREATE INDEX fails? \n> > \n> > I've tried everything I can think of, and can't get that INDEX to be\n> > created :(\n> > \n> > -- $Id: groups.psql,v 1.2 1997/08/14 20:05:08 shevett Exp $\n> > -- postgresql 6.1\n> > --\n> > -- Host: localhost Database: pts\n> > ----------------------------------------------------------\n> > \n> > \n> > --\n> > -- Table structure for table 'groups'\n> > --\n> > CREATE TABLE groups (\n> > id CHAR(10) NOT NULL,\n> > detail CHAR(30)\n> > ) \\g\n> > \n> > CREATE INDEX groups_indx ON groups ( id ) \\g\n> > \n> > \n> > --\n> > -- Dumping data for table 'groups'\n> > --\n> > \n> > INSERT INTO groups (id, detail) VALUES ('PTS','PTS Development')\\g\n> > \n> > \n> \n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Fri, 20 Feb 1998 01:35:13 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] odd error creating index in -current..."
},
{
"msg_contents": "The Hermit Hacker wrote:\n> \n> Someone want to tell me why the below CREATE INDEX fails?\n> \n> I've tried everything I can think of, and can't get that INDEX to be\n> created :(\n> \n> -- $Id: groups.psql,v 1.2 1997/08/14 20:05:08 shevett Exp $\n> -- postgresql 6.1\n ^^^\n??? NOT NULL appeared in 6.2...\n\n6.3:\nvac=> CREATE TABLE groups (\nvac-> id CHAR(10) NOT NULL,\nvac-> detail CHAR(30)\nvac-> ) \\g\nCREATE\nvac=> \nvac=> CREATE INDEX groups_indx ON groups ( id ) \\g\nCREATE\nvac=> \nvac=> INSERT INTO groups (id, detail) VALUES ('PTS','PTS Development')\\g\nINSERT 43725 1\n\nVadim\n",
"msg_date": "Fri, 20 Feb 1998 12:38:26 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] odd error creating index in -current..."
},
{
"msg_contents": "Brett McCormick wrote:\n> \n> I may be missing something, but because char_ops has been left out?\n\nIt's good in _most_ cases to let all XXX_ops be left out :)\nIndices know what to use...\n\nVadim\n",
"msg_date": "Fri, 20 Feb 1998 12:53:52 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] odd error creating index in -current..."
},
{
"msg_contents": "\nI'm still living in pre 6.0 days, then.. are indices on type oid\natttributes used as well?\n\nOn Fri, 20 February 1998, at 12:53:52, Vadim B. Mikheev wrote:\n\n> It's good in _most_ cases to let all XXX_ops be left out :)\n> Indices know what to use...\n> \n> Vadim\n",
"msg_date": "Thu, 19 Feb 1998 22:24:23 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] odd error creating index in -current..."
},
{
"msg_contents": "Brett McCormick wrote:\n> \n> I'm still living in pre 6.0 days, then.. are indices on type oid\n> atttributes used as well?\n> \n> On Fri, 20 February 1998, at 12:53:52, Vadim B. Mikheev wrote:\n> \n> > It's good in _most_ cases to let all XXX_ops be left out :)\n> > Indices know what to use...\n\nThey should... Just try.\n\nVadim\n",
"msg_date": "Fri, 20 Feb 1998 14:08:27 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] odd error creating index in -current..."
},
{
"msg_contents": "On Fri, 20 Feb 1998, Vadim B. Mikheev wrote:\n\n> The Hermit Hacker wrote:\n> > \n> > Someone want to tell me why the below CREATE INDEX fails?\n> > \n> > I've tried everything I can think of, and can't get that INDEX to be\n> > created :(\n> > \n> > -- $Id: groups.psql,v 1.2 1997/08/14 20:05:08 shevett Exp $\n> > -- postgresql 6.1\n> ^^^\n> ??? NOT NULL appeared in 6.2...\n> \n> 6.3:\n> vac=> CREATE TABLE groups (\n> vac-> id CHAR(10) NOT NULL,\n> vac-> detail CHAR(30)\n> vac-> ) \\g\n> CREATE\n> vac=> \n> vac=> CREATE INDEX groups_indx ON groups ( id ) \\g\n> CREATE\n> vac=> \n> vac=> INSERT INTO groups (id, detail) VALUES ('PTS','PTS Development')\\g\n> INSERT 43725 1\n\nStrange...I'm going to re-check this all through tonight when I get home,\nsince I've obviously got something messed up on my computer...\n\nAs for the 'NOT NULL'...I'm trying to get the PostgreSQL stuff up to date\nin PTS for v6.3 ... I added in the NOT NULL last night...\n\n\n",
"msg_date": "Fri, 20 Feb 1998 07:48:49 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] odd error creating index in -current..."
}
] |
[
{
"msg_contents": "\nBackend crashes on this one too...?\n\n\nCREATE TABLE projects (\n pid INT NOT NULL PRIMARY KEY,\n ppid INT,\n grp VARCHAR(10),\n o_date INT,\n o_tech VARCHAR(10),\n sc_date INT,\n c_date INT,\n c_tech VARCHAR(5),\n currtech VARCHAR(10),\n state VARCHAR(1),\n status VARCHAR(1),\n priority INT,\n c_name VARCHAR(20),\n c_phone VARCHAR(20),\n problem VARCHAR(100),\n summary VARCHAR(80)\n) \\g\n\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Fri, 20 Feb 1998 01:12:32 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Another one I thought should work..."
},
{
"msg_contents": "> Backend crashes on this one too...?\n\nMy 980217 cvs tree does not crash on this.\n\n - Tom\n\npostgres=> CREATE TABLE projects (\npostgres-> pid INT NOT NULL PRIMARY KEY,\npostgres-> ppid INT,\npostgres-> grp VARCHAR(10),\npostgres-> o_date INT,\npostgres-> o_tech VARCHAR(10),\npostgres-> sc_date INT,\npostgres-> c_date INT,\npostgres-> c_tech VARCHAR(5),\npostgres-> currtech VARCHAR(10),\npostgres-> state VARCHAR(1),\npostgres-> status VARCHAR(1),\npostgres-> priority INT,\npostgres-> c_name VARCHAR(20),\npostgres-> c_phone VARCHAR(20),\npostgres-> problem VARCHAR(100),\npostgres-> summary VARCHAR(80)\npostgres-> ) \\g\nNOTICE: CREATE TABLE/PRIMARY KEY will create implicit index projects_pkey for\ntable projects\nCREATE\npostgres=> insert into projects values\n(1,2,'hi',3,'there',4,5,'five','six','y','n',99,'tom','123-4567','crashes?','no\nway!');\nINSERT 1018395 1\n\n",
"msg_date": "Fri, 20 Feb 1998 05:29:39 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Another one I thought should work..."
},
{
"msg_contents": "The Hermit Hacker wrote:\n> \n> Backend crashes on this one too...?\n> \n> CREATE TABLE projects (\n> pid INT NOT NULL PRIMARY KEY,\n ^^^^^^^^^^^^^^^^^^^^\nNo probs here, in 6.3...\n\nVadim\n",
"msg_date": "Fri, 20 Feb 1998 12:39:46 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Another one I thought should work..."
},
{
"msg_contents": "On Fri, 20 Feb 1998, Thomas G. Lockhart wrote:\n\n> > Backend crashes on this one too...?\n> \n> My 980217 cvs tree does not crash on this.\n\nAck...wait, that is 3 days old...okay, mine is based on todays CVS...\n\nscrappy=> CREATE TABLE projects (\nscrappy-> pid INT NOT NULL PRIMARY KEY,\nscrappy-> ppid INT,\nscrappy-> grp VARCHAR(10),\nscrappy-> o_date INT,\nscrappy-> o_tech VARCHAR(10),\nscrappy-> sc_date INT,\nscrappy-> c_date INT,\nscrappy-> c_tech VARCHAR(5),\nscrappy-> currtech VARCHAR(10),\nscrappy-> state VARCHAR(1),\nscrappy-> status VARCHAR(1),\nscrappy-> priority INT,\nscrappy-> c_name VARCHAR(20),\nscrappy-> c_phone VARCHAR(20),\nscrappy-> problem VARCHAR(100),\nscrappy-> summary VARCHAR(80)\nscrappy-> ) \\g\nNOTICE: CREATE TABLE/PRIMARY KEY will create implicit index projects_pkey\nfor table projects\nNOTICE: _outNode: don't know how to print type 715\nNOTICE: _outNode: don't know how to print type 715\n\n\nBruce, can you try yours?\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Fri, 20 Feb 1998 01:40:48 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Another one I thought should work..."
},
{
"msg_contents": "> > > Backend crashes on this one too...?\n> >\n> > My 980217 cvs tree does not crash on this.\n\nbtw, I'm not running with asserts on. Also, this will be easy to fix; it is\ntrying to print a node which does not have an associated print routine. Actually,\nare you sure the table did not get created? That is a NOTICE, not ERROR...\n\n - Tom\n\n> Ack...wait, that is 3 days old...okay, mine is based on todays CVS...\n>\n> scrappy=> CREATE TABLE projects (\n> scrappy-> pid INT NOT NULL PRIMARY KEY,\n> scrappy-> ppid INT,\n> scrappy-> grp VARCHAR(10),\n> scrappy-> o_date INT,\n> scrappy-> o_tech VARCHAR(10),\n> scrappy-> sc_date INT,\n> scrappy-> c_date INT,\n> scrappy-> c_tech VARCHAR(5),\n> scrappy-> currtech VARCHAR(10),\n> scrappy-> state VARCHAR(1),\n> scrappy-> status VARCHAR(1),\n> scrappy-> priority INT,\n> scrappy-> c_name VARCHAR(20),\n> scrappy-> c_phone VARCHAR(20),\n> scrappy-> problem VARCHAR(100),\n> scrappy-> summary VARCHAR(80)\n> scrappy-> ) \\g\n> NOTICE: CREATE TABLE/PRIMARY KEY will create implicit index projects_pkey\n> for table projects\n> NOTICE: _outNode: don't know how to print type 715\n> NOTICE: _outNode: don't know how to print type 715\n\n",
"msg_date": "Fri, 20 Feb 1998 06:10:43 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Another one I thought should work..."
},
{
"msg_contents": "> \n> On Fri, 20 Feb 1998, Thomas G. Lockhart wrote:\n> \n> > > Backend crashes on this one too...?\n> > \n> > My 980217 cvs tree does not crash on this.\n> \n> Ack...wait, that is 3 days old...okay, mine is based on todays CVS...\n> \n> scrappy=> CREATE TABLE projects (\n> scrappy-> pid INT NOT NULL PRIMARY KEY,\n> scrappy-> ppid INT,\n> scrappy-> grp VARCHAR(10),\n> scrappy-> o_date INT,\n> scrappy-> o_tech VARCHAR(10),\n> scrappy-> sc_date INT,\n> scrappy-> c_date INT,\n> scrappy-> c_tech VARCHAR(5),\n> scrappy-> currtech VARCHAR(10),\n> scrappy-> state VARCHAR(1),\n> scrappy-> status VARCHAR(1),\n> scrappy-> priority INT,\n> scrappy-> c_name VARCHAR(20),\n> scrappy-> c_phone VARCHAR(20),\n> scrappy-> problem VARCHAR(100),\n> scrappy-> summary VARCHAR(80)\n> scrappy-> ) \\g\n> NOTICE: CREATE TABLE/PRIMARY KEY will create implicit index projects_pkey\n> for table projects\n> NOTICE: _outNode: don't know how to print type 715\n> NOTICE: _outNode: don't know how to print type 715\n> \n> \n> Bruce, can you try yours?\n\nWorks here, my snapshot is from Feb 19.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 09:52:09 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Another one I thought should work..."
}
] |
[
{
"msg_contents": "Unfortunately createdb as user postgres doesn't work either, so I have\nto keep debugging without being able to run the binary.\n\nMichael\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tMichael Meskes [SMTP:[email protected]]\n> Sent:\tFriday, February 20, 1998 9:41 AM\n> To:\[email protected]\n> Cc:\[email protected]\n> Subject:\tRe: [HACKERS] RE: New ecgp code problem.\n> \n> Keith Parks writes:\n> > Hi,\n> > \n> > The problem is just the missing name column.\n> \n> I think I know the problem. I will send a patch later (again) but I\n> cannot\n> test it myself as I still cannot create a user. Hmm, maybe I should\n> try as\n> postgres user.\n> \n> Michael\n> \n> -- \n> Dr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\n> [email protected] | Europark A2, Adenauerstr. 20\n> [email protected] | 52146 Wuerselen\n> Go SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\n> Use Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Fri, 20 Feb 1998 09:58:41 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] RE: New ecgp code problem."
},
{
"msg_contents": "On Fri, 20 Feb 1998, Meskes, Michael wrote:\n\n> Unfortunately createdb as user postgres doesn't work either, so I have\n> to keep debugging without being able to run the binary.\n\n\tI'm curious here as to why nobody has looked into this\nproblem...why is it that you can't create a database/user? *raised\neyebrow*\n\n\n",
"msg_date": "Fri, 20 Feb 1998 07:58:33 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "RE: [HACKERS] RE: New ecgp code problem."
},
{
"msg_contents": "> > Unfortunately createdb as user postgres doesn't work either, so I have\n> > to keep debugging without being able to run the binary.\n>\n> I'm curious here as to why nobody has looked into this\n> problem...\n\nBecause no one else sees the problem? I've forgotten what your symptom\nactually is, but from the 980217 cvs tree:\n\npostgres=> \\q\ngolem$ createuser tgl\nEnter user's postgres ID or RETURN to use unix user ID: 106 ->\nIs user \"tgl\" allowed to create databases (y/n) y\nIs user \"tgl\" allowed to add users? (y/n) y\ncreateuser: tgl was successfully added\n...\ngolem> createdb\ngolem> psql\nWelcome to the POSTGRESQL interactive sql monitor:\n...\nYou are currently connected to the database: tgl\n\ntgl=>\n\nHave you done a complete \"make clean; make install; initdb\" cycle??\n\n - Tom\n\n",
"msg_date": "Fri, 20 Feb 1998 13:26:50 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] RE: New ecgp code problem."
}
] |
[
{
"msg_contents": ">I have been working with many of you for over a hear, and would like to\n>know more details about you.\n>\n\n\nI'm 32 married with 3 children. My girl is 4 (Michelle), and my sons\n(Michael-Alan and Daniel), are 3, and 1.\nWe live in the Netherlands.\n\nI sell myself as a consultant usually to banks.\nI have experience building a keymanagment system, internet payment system,\nencrypted TCP tunnels, userauthentication systems based on smartcards\nand other applications in cryptography (SSL, SET, IKP).\nI run a _small_ ISP which helps me to learn new things like postgresql and\nother stuff like TCP/IP networking in general..\n\nI also works as a C/C++ trainer (since 1992 about) and I'm now completing\nmy first Java course with a group of about 10 poeple.\n\nI used to be a real OS/2 / Windows fan but since I discovered Linux in 1995\nI'm\nall for UNIX. (I have the privilige to be able to work with 4 Unix\nvariants).\nI now only use W95/NT when I'm there is some reason to.\n\nMy personal interest is in Chomsky phrase structure grammars, and as such\nI hope to find time to find out more about the rewriting system in\npostgresql.\nI'm wondering if it's a grammar based term rewriting system.\n\nMy ideal system combines a distributed OO database with object interaction\ndefined by some formal rewriting system (I like dreaming).\n\nAnyway I have a hole lot to learn about the postgresql system.\nI like the neat coding style and the modular organisation\nof the system. All we have to do now is to make it the best (and fastest)\ndatabase\nin the world -:).\n\nThanks,\n Maurice\n\n\n",
"msg_date": "Fri, 20 Feb 1998 10:04:44 +0100",
"msg_from": "\"Maurice Gittens\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Who is everyone?"
}
] |
[
{
"msg_contents": "I know that postgres originally was able to do recursive queries like\n\nretrieve* into a from a ...\n\nIs this still possible with PostgreSQL? If so, is it just for one statement\nor even as a block?\n\nThis happens to be the area I worked on for several years.\n\nMichael\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Fri, 20 Feb 1998 10:14:35 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": true,
"msg_subject": "Recursive queries?"
},
{
"msg_contents": "Michael Meskes wrote:\n> \n> I know that postgres originally was able to do recursive queries like\n> \n> retrieve* into a from a ...\n> \n> Is this still possible with PostgreSQL? If so, is it just for one statement\n> or even as a block?\n\nYes, it's possible. This is from spi.txt:\n---\n Data changes visibility\n\n PostgreSQL data changes visibility rule: during a query execution, data\nchanges made by the query itself (via SQL-function, SPI-function, triggers)\nare invisible to the query scan. For example, in query\n\n INSERT INTO a SELECT * FROM a\n\n tuples inserted are invisible for SELECT' scan. In effect, this\nduplicates the database table within itself (subject to unique index\nrules, of course) without recursing.\n\n Changes made by query Q are visible by queries which are started after\nquery Q, no matter whether they are started inside Q (during the execution\nof Q) or after Q is done.\n---\n\nSecond query 'INSERT INTO a SELECT * FROM a' inside BEGIN/END\nwill see tuples inserted by first one. Pg uses special CommandCounter\nto distinguish changes made in the same transaction.\n\n> This happens to be the area I worked on for several years.\n\nI also like such areas :)\n\nVadim\n",
"msg_date": "Fri, 20 Feb 1998 16:59:17 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Recursive queries?"
}
] |
[
{
"msg_contents": "But such a statement is executed only once, isn't it?\n\nPostquel did execute a retrieve* into as long as it was able to generate\nnew data.\n\nMichael\n\nP.S.: Glad I'm not the only one interested in this. :-)\n\n--\nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n\n> -----Original Message-----\n> From:\tVadim B. Mikheev [SMTP:[email protected]]\n> Sent:\tFriday, February 20, 1998 10:59 AM\n> To:\tMichael Meskes\n> Cc:\tPostgreSQL Hacker\n> Subject:\tRe: [HACKERS] Recursive queries?\n> \n> Michael Meskes wrote:\n> > \n> > I know that postgres originally was able to do recursive queries\n> like\n> > \n> > retrieve* into a from a ...\n> > \n> > Is this still possible with PostgreSQL? If so, is it just for one\n> statement\n> > or even as a block?\n> \n> Yes, it's possible. This is from spi.txt:\n> ---\n> Data changes visibility\n> \n> PostgreSQL data changes visibility rule: during a query execution,\n> data\n> changes made by the query itself (via SQL-function, SPI-function,\n> triggers)\n> are invisible to the query scan. For example, in query\n> \n> INSERT INTO a SELECT * FROM a\n> \n> tuples inserted are invisible for SELECT' scan. In effect, this\n> duplicates the database table within itself (subject to unique index\n> rules, of course) without recursing.\n> \n> Changes made by query Q are visible by queries which are started\n> after\n> query Q, no matter whether they are started inside Q (during the\n> execution\n> of Q) or after Q is done.\n> ---\n> \n> Second query 'INSERT INTO a SELECT * FROM a' inside BEGIN/END\n> will see tuples inserted by first one. Pg uses special CommandCounter\n> to distinguish changes made in the same transaction.\n> \n> > This happens to be the area I worked on for several years.\n> \n> I also like such areas :)\n> \n> Vadim\n",
"msg_date": "Fri, 20 Feb 1998 11:35:18 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Recursive queries?"
},
{
"msg_contents": "> But such a statement is executed only once, isn't it?\n>\n> Postquel did execute a retrieve* into as long as it was able to generate\n> new data.\n\nOK, I'll bite :)\nCan you give an example of this? How does new data become available in the\nmiddle of a transaction? Just curious...\n\n",
"msg_date": "Fri, 20 Feb 1998 16:44:15 +0000",
"msg_from": "\"Thomas G. Lockhart\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Recursive queries?"
}
] |
[
{
"msg_contents": "Since the copy statement is behaving differently than the normal select\nstuff,\nI think we should eighter introduce a new permission (name it copy or dump)\nor include the copy into the rewrite system.\n\nI would vote for the first and implement a new command: \n\tunload to <filename> [delimiter '|'] <select statement>; -- and\n\tload from <filename> [delimiter '|'] <insert statement>;\nthat does behave like the select. (please forgive my Informix\nbackground)\n\nAs to the topic with setuid triggers and others, I think setuid procedures\nwould be sufficient.\nThese are implemented in Informix with the following simple syntax:\ncreate dba procedure ....\n\nAndreas\n",
"msg_date": "Fri, 20 Feb 1998 12:02:52 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "Permissions on copy"
},
{
"msg_contents": "\nAndreas wrote:\n>\n> Since the copy statement is behaving differently than the normal select\n> stuff,\n> I think we should eighter introduce a new permission (name it copy or dump)\n> or include the copy into the rewrite system.\n>\n> I would vote for the first and implement a new command:\n> unload to <filename> [delimiter '|'] <select statement>; -- and\n> load from <filename> [delimiter '|'] <insert statement>;\n> that does behave like the select. (please forgive my Informix\n> background)\n\n Since the copy command checks for ACL_RD or ACL_WR on the\n relation a user can use copy on any table, where he has\n propper permissions.\n\n And triggers are fired too on COPY FROM (thanks for the push\n - found another place relevant for setuid triggers).\n\n The only things not working for copy are rewrite rules. But I\n think we should restrict rules to the view handling in the\n future and move forward by implementing a pure and really\n powerful procedural language.\n\n I vote for leaving copy as it is and take a look if we can\n fire triggers on SCAN which then could modify tuples instead\n of using rules at all. If the time an ON SCAN trigger is\n fired is just before the evaluation of any query\n qualification and put down into the heap access methods,\n anything should be fine except for performance maybe (but as\n I usually say: don't force it - use a bigger hammer).\n\n>\n> As to the topic with setuid triggers and others, I think setuid procedures\n> would be sufficient.\n> These are implemented in Informix with the following simple syntax:\n> create dba procedure ....\n\n Looks good. Alternatively/additionally something like\n\n ALTER FUNCTION funcname (argtypes) { SETUID | NOSETUID }\n\n might be useful. Parser etc. steps could mostly be copied\n from the DROP FUNCTION ... statement. And we have an unused\n boolean attribute proistrusted in the pg_proc struct. If we\n rename that to proissetuid anything is fine.\n\n As I said I already fiddled around with that and got setuid\n functions and triggers working (don't ask for code - usually\n I make a copy of the source tree, hack in totally ugly things\n until I know how to do it and throw anything away to be sure\n only development not hacking get's into PostgreSQL).\n\n\nUntil later, Jan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Fri, 20 Feb 1998 13:30:13 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Permissions on copy"
},
{
"msg_contents": "> \n> Since the copy statement is behaving differently than the normal select\n> stuff,\n> I think we should eighter introduce a new permission (name it copy or dump)\n> or include the copy into the rewrite system.\n> \n> I would vote for the first and implement a new command: \n> \tunload to <filename> [delimiter '|'] <select statement>; -- and\n> \tload from <filename> [delimiter '|'] <insert statement>;\n> that does behave like the select. (please forgive my Informix\n> background)\n\nYes, I agree the Informix way of having load/unload, and having a SELECT\ncapability so you can dump any data/join you want, not just a single\ntable. Do I have votes to put this on the TODO list?\n\n> \n> As to the topic with setuid triggers and others, I think setuid procedures\n> would be sufficient.\n> These are implemented in Informix with the following simple syntax:\n> create dba procedure ....\n> \n> Andreas\n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 10:45:30 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Permissions on copy"
},
{
"msg_contents": "On Fri, 20 Feb 1998, Bruce Momjian wrote:\n\n> > \n> > Since the copy statement is behaving differently than the normal select\n> > stuff,\n> > I think we should eighter introduce a new permission (name it copy or dump)\n> > or include the copy into the rewrite system.\n> > \n> > I would vote for the first and implement a new command: \n> > \tunload to <filename> [delimiter '|'] <select statement>; -- and\n> > \tload from <filename> [delimiter '|'] <insert statement>;\n> > that does behave like the select. (please forgive my Informix\n> > background)\n> \n> Yes, I agree the Informix way of having load/unload, and having a SELECT\n> capability so you can dump any data/join you want, not just a single\n> table. Do I have votes to put this on the TODO list?\n\n\tI'm not quite sure what we are voting on here...is it to implement\npermissions on a copy, like we do on 'select/delete/insert/etc'?\n\n\tIf so, count me in...\n\n\n",
"msg_date": "Fri, 20 Feb 1998 11:56:47 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Permissions on copy"
},
{
"msg_contents": "> \n> On Fri, 20 Feb 1998, Bruce Momjian wrote:\n> \n> > > \n> > > Since the copy statement is behaving differently than the normal select\n> > > stuff,\n> > > I think we should eighter introduce a new permission (name it copy or dump)\n> > > or include the copy into the rewrite system.\n> > > \n> > > I would vote for the first and implement a new command: \n> > > \tunload to <filename> [delimiter '|'] <select statement>; -- and\n> > > \tload from <filename> [delimiter '|'] <insert statement>;\n> > > that does behave like the select. (please forgive my Informix\n> > > background)\n> > \n> > Yes, I agree the Informix way of having load/unload, and having a SELECT\n> > capability so you can dump any data/join you want, not just a single\n> > table. Do I have votes to put this on the TODO list?\n> \n> \tI'm not quite sure what we are voting on here...is it to implement\n> permissions on a copy, like we do on 'select/delete/insert/etc'?\n> \n> \tIf so, count me in...\n\nTwo things. First was a separate COPY priviledge, which I vote against.\nI see no real value to it, except to work around the problem that COPY\ndoesn't use rules.\n\nSecond, there was the idea of making copy allow a real select statement\nand not just a table name. If we do that, all goes through the\nexecutor, and you get view and rules working properly. May have some\nperformance penalty, though it probabably will be minor.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 12:03:10 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Permissions on copy"
},
{
"msg_contents": "On Fri, 20 Feb 1998, Bruce Momjian wrote:\n\n> Two things. First was a separate COPY priviledge, which I vote against.\n> I see no real value to it, except to work around the problem that COPY\n> doesn't use rules.\n\n\tOkay, I may be totally out in left field here (ie. unrelated), but\nwhat stops a user from doing a 'COPY out' on a table that they don't have\nSELECT privileges on? Kind of negates 'REVOKE ALL...', no?\n\n> Second, there was the idea of making copy allow a real select statement\n> and not just a table name. If we do that, all goes through the\n> executor, and you get view and rules working properly. May have some\n> performance penalty, though it probabably will be minor.\n\n\tThis sounds reasonable...\n\n\n",
"msg_date": "Fri, 20 Feb 1998 12:10:29 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Permissions on copy"
},
{
"msg_contents": "> \n> On Fri, 20 Feb 1998, Bruce Momjian wrote:\n> \n> > Two things. First was a separate COPY priviledge, which I vote against.\n> > I see no real value to it, except to work around the problem that COPY\n> > doesn't use rules.\n> \n> \tOkay, I may be totally out in left field here (ie. unrelated), but\n> what stops a user from doing a 'COPY out' on a table that they don't have\n> SELECT privileges on? Kind of negates 'REVOKE ALL...', no?\n\nYes I think a separate COPY permission makes no sense.\n\n> \n\n> > Second, there was the idea of making copy allow a real select statement\n> > and not just a table name. If we do that, all goes through the\n> > executor, and you get view and rules working properly. May have some\n> > performance penalty, though it probabably will be minor.\n> \n> \tThis sounds reasonable...\n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 12:22:03 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Permissions on copy"
},
{
"msg_contents": "\neither-or or just a select, so it is forced to go through the\nexecutor? If we do that, what about reload the database.. it seems\nlike copy should just dump the table as-is and not do funny tricks\nthat make it impossible to COPY FROM...\n\nOn Fri, 20 February 1998, at 12:10:29, The Hermit Hacker wrote:\n\n> > Second, there was the idea of making copy allow a real select statement\n> > and not just a table name. If we do that, all goes through the\n> > executor, and you get view and rules working properly. May have some\n> > performance penalty, though it probabably will be minor.\n> \n> \tThis sounds reasonable...\n> \n> \n",
"msg_date": "Fri, 20 Feb 1998 14:11:44 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Permissions on copy"
},
{
"msg_contents": "> either-or or just a select, so it is forced to go through the\n> executor? If we do that, what about reload the database.. it seems\n> like copy should just dump the table as-is and not do funny tricks\n> that make it impossible to COPY FROM...\n\n Good point!\n\n Currently COPY FROM does fire insert triggers. Pg_dump uses\n that and if there are triggers a restore from a pg_dump\n output with triggers active might not recreate the database\n dumped. This does not only lead to triggers, constraints are\n also object to this because reference checks implemented by\n triggers or constraints must fail if the data isn't dumped by\n pg_dump in the right order of tables. For constraints it may\n be possible (if not yet done) for pg_dump, to analyze them\n and dump the tables in the right order. But pg_dump cannot\n know what a trigger checks or what it inserts/updates/deletes\n if fired.\n\n So we need at least a switch for the COPY command restricted\n to superusers or the DB owner telling COPY to suppress\n trigger firing. Then have a look at pg_dump if it analyzes\n constraints.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Sat, 21 Feb 1998 12:34:38 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Permissions on copy"
},
{
"msg_contents": "Jan Wieck wrote:\n> \n> > either-or or just a select, so it is forced to go through the\n> > executor? If we do that, what about reload the database.. it seems\n> > like copy should just dump the table as-is and not do funny tricks\n> > that make it impossible to COPY FROM...\n> \n> Good point!\n> \n> Currently COPY FROM does fire insert triggers. Pg_dump uses\n> that and if there are triggers a restore from a pg_dump\n> output with triggers active might not recreate the database\n> dumped. This does not only lead to triggers, constraints are\n> also object to this because reference checks implemented by\n> triggers or constraints must fail if the data isn't dumped by\n> pg_dump in the right order of tables. For constraints it may\n> be possible (if not yet done) for pg_dump, to analyze them\n> and dump the tables in the right order. But pg_dump cannot\n> know what a trigger checks or what it inserts/updates/deletes\n> if fired.\n> \n> So we need at least a switch for the COPY command restricted\n> to superusers or the DB owner telling COPY to suppress\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nTable owber ?\n\n> trigger firing. Then have a look at pg_dump if it analyzes\n> constraints.\n\n1. usage: pg_dump [options] [dbname]\n -a dump out only the data, no schema\n -d dump data as proper insert strings\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nNot only COPY should be satisfied.\n\n2. pg_dump put triggers at the end of output -> triggers will not be\n fired on restoring.\n\n3. As for referential constraints, it's better for pg_dump to create \n all tables without any of these, save data and then just use ALTER TABLE to\n add constraints.\n\nVadim\n",
"msg_date": "Sun, 22 Feb 1998 17:12:32 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Permissions on copy"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> Yes, I agree the Informix way of having load/unload, and having a SELECT\n> capability so you can dump any data/join you want, not just a single\n> table. Do I have votes to put this on the TODO list?\n\n\nI guess this means that it would be possible to load (for example)\na file with 2 columns into a table with 3 columns? Like this:\n load \"my2columnsfile\" insert into mytable (mycol13, mycol1);\n\nAnother thing which would be nice is to be able to select only some\nof the columns from the file, like this:\n load myfile(1,2,5) insert into mytable (x, y, z);\n\nThis would be very useful (for me, at least).\n\n\nThe 'raw' COPY should be restricted to superuser only, so that no user\ncan bypass views, triggers and so on.\n\n/* m */\n",
"msg_date": "Mon, 23 Feb 1998 12:22:24 +0100",
"msg_from": "Mattias Kregert <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Permissions on copy"
}
] |
[
{
"msg_contents": "Hi,\n\nI've changed the large object memory managment strategy to ensure that\neach large object has it's own memory context.\nThis way I can free all memory for a large object when I lo_close\nis called for it's oid. This seems to work.\n\nI've noticed that the heap_close function used to close the heap used\nby a large object doesn't really do any thing. (It calls RelationClose\nwhich decrements some reference count).\n\nSomehow I have to free the relation from the cache in the following\nsituations:\n1. In a transaction I must free the stuff when the transaction is\ncommited/aborted.\n2. Otherwise it must happen when lo_close is called.\n\nIs this right?\n\nCan anyone point me in the right direction?\n\nThanks Maurice.\n\n\n",
"msg_date": "Fri, 20 Feb 1998 14:14:22 +0100",
"msg_from": "\"Maurice Gittens\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "How To free resources used by large object Relations?"
},
{
"msg_contents": "> \n> Hi,\n> \n> I've changed the large object memory managment strategy to ensure that\n> each large object has it's own memory context.\n> This way I can free all memory for a large object when I lo_close\n> is called for it's oid. This seems to work.\n\nThis is a good idea.\n\n> \n> I've noticed that the heap_close function used to close the heap used\n> by a large object doesn't really do any thing. (It calls RelationClose\n> which decrements some reference count).\n> \n> Somehow I have to free the relation from the cache in the following\n> situations:\n> 1. In a transaction I must free the stuff when the transaction is\n> commited/aborted.\n> 2. Otherwise it must happen when lo_close is called.\n> \n> Is this right?\n> \n\nBeats me.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 11:04:16 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] How To free resources used by large object Relations?"
},
{
"msg_contents": "\nSince we have so little documentation on the rules, I think we should save\nevery \nlittle word describing them, so could you simply put the following into a\nrules.readme \n(undigested is still better than not adding it)\n\n> > Why I like the rewrite system is:\n> > 1. select rewrite -- select trigger would be no good (optimizer)\n> \n> Exactly that's what is done if you create a view. Postgres\n> creates a regular table (look at pg_class and into the\n> database directory) and then sets up a relation level instead\n> rewrite rule on select.\n> \n> > 2. The client can be really dumb, like MS Access or some other\n> > standard ODBC tool\n> > which does not know anything about funcs procs and the like\n> > (even without using passthrough)\n> \n> Yupp - the client must not know why and how and where the\n> data is left and coming from. But that's true in any case - a\n> trigger for each row on insert can do anything different and\n> push the data wherever it wants.\n> \n> > 3. it is a lot more powerful than views\n> \n> As said - views are only one special rule case in Postgres.\n> \n> > 4. it allows the optimizer to get involved (this is where triggers\n> > fail per definition)\n> > 5. once understood it is very easy to use\n> > easier than trigger with c stored procedure at least\n> \n> Optimizing again and again. If the rules aren't instead, the\n> querytree get's additional queries for every rule appended.\n> Have a table field that references an entry in another table\n> and this entry should have a refcount. So on update you must\n> decrease the refcount from the old ref and increase it on the\n> new. You create two rules so the UPDATE will result in 1\n> scan and 2 nestloops with scans inside - really optimized if\n> the referenced value doesn't change. And don't think that a\n> rule qual of NEW != CURRENT might help - that will result in\n> 2 mergejoins where the scanned tuples are compared.\n> \nI fought that like a windmill, I guess it would be better to kill the\nCURRENT keyword\nwith this meaning alltogether, since it only has the same meaning as the\ntablename itself.\nI have already crossed it out of my mind and don't miss anything.\nI think there should instead be an OLD and NEW keyword\nlike in triggers:\n\treferencing old as <oldname> new as <newname>\nthat only reference the tuples in memory.\n\n> BTW, this sample doesn't work currently because the rules\n> queries are appended at the end of the querytree, thus the\n> decrement scan having the same qual will not find the old\n> tuple at all because it's already outdated\n> (command_counter_increment between processing the queries).\n> Referencing CURRENT in a rule is not what most people think\n> it is.\n> \n> The old 4.2 postgres had a second, instance level rule system\n> (prs2 stubs) that fired the rules actions when actually the\n> old tuple and the new projected tuple where handy. There you\n> could have made also things like 'UPDATE NEW SET a = 4' that\n> really modified the in memory tuple in the executors\n> expression context. Who the hell removed all that? It was so\n> nice :-(\n> \nAbsolutely ! I did cry up when that was done, but nobody responded :-(\nWell to be honest Vadim did respond with the trigger code, which made me\nfeel comfortable again.\n\n> A really simple to write trigger can compare old != new and\n> only if send down the other two queries. This time they wont\n> be nestloops, they are simple scans. And the trigger can\n> arrange that the queries it uses are only parsed on it's\n> first of all calls and store the generated execution plans\n> permanently for quick execution (look at SPI_prepare).\n> \n> For the stored C procedures you're totally right. I don't\n> like the C functions because it requires postgres superuser\n> rights to develop them and thus I created PL/Tcl where joe\n> user can hack around without having complete access to the\n> whole database (look at src/pl/tcl). And someday after 6.3\n> release I'll really start on a plain PL/pgSQL implementation\n> that would give a normal user the opportunity to create\n> functions and triggers on a high level. There is light at the\n> end of the tunnel - hope that it isn't the coming train :-)\n> \n> >\n> > I guess if triggers could also trigger simple select statements, I could\n> do\n> > most of what I want using triggers except of course the select stuff.\n> > But as I said I like the rules system very much, especially after your\n> > recent\n> > fixes Jan :-) So please stick to supporting all 3: triggers, views and\n> > rules. Wow :-)\n> \n> Well - a trigger cannot build a view. The relation underlying\n> the view doesn't contain any tuples and a select trigger will\n> never be fired. As long as there is no possibility to return\n> tuple sets from non-SQL functions. But a trigger can do\n> things like the pg_hide_passwd stuff much more powerful. You\n> could define the trigger so that it checks if the user is a\n> superuser and overwrite the passwd value only in the case\n> where he/she isn't. If fired at the right place it would too\n> work for things like the copy command etc.\n> \n> We must stay with all 3 features. And I will take a look at\n> the INSERT ... SELECT view problem really soon as it is a\n> rule system problem that breaks views. But this is only the\n> SELECT rewriting part of the rule system which I really like\n> (optimizable). The other areas (insert, update, delete) of\n> the rule system are dangerous and I really think a powerful\n> PL/pgSQL language could make them obsolete.\n> \n> Jan\n> \nOk, to sum it up:\n\t1. We need and want the select part of the rewrite rules.\n\t2. for the insert/update/delete rules the old instance rules system\n\t was much more appropriate. TODO: dig up the old code\n\t and merge it with the current trigger Implementation\n\t\t it must be pretty much the wanted functionality (it\nsupported sql)\n\t\t3. the CURRENT keyword in the i/u/d rewrite rules is stupid\nand should be disabled\n\t\t destroyed and burned in hell\n\t\t4. To stick to the mainstream we should enhance the trigger\nsyntax,\n\t\t and forget the rule stuff for i/u/d\n\t\t\n\t\tcreate trigger passwd_utr\n\t\t..........\n\t\treferencing old as o new as n\n\t\t for each row (statement, statement, statement, procedure,\n...... all PL/pgSQL syntax allowed );\n\t\t-- with a syntax to modify the new tuple in memory\n\t\t\n\nAndreas\n\n\n",
"msg_date": "Fri, 20 Feb 1998 18:51:42 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": false,
"msg_subject": "AW: [HACKERS] triggers, views and rules (not instead)"
},
{
"msg_contents": "> > \n> Ok, to sum it up:\n> \t1. We need and want the select part of the rewrite rules.\n> \t2. for the insert/update/delete rules the old instance rules system\n> \t was much more appropriate. TODO: dig up the old code\n> \t and merge it with the current trigger Implementation\n> \t\t it must be pretty much the wanted functionality (it\n> supported sql)\n> \t\t3. the CURRENT keyword in the i/u/d rewrite rules is stupid\n> and should be disabled\n> \t\t destroyed and burned in hell\n> \t\t4. To stick to the mainstream we should enhance the trigger\n> syntax,\n> \t\t and forget the rule stuff for i/u/d\n> \t\t\n> \t\tcreate trigger passwd_utr\n> \t\t..........\n> \t\treferencing old as o new as n\n> \t\t for each row (statement, statement, statement, procedure,\n> ...... all PL/pgSQL syntax allowed );\n> \t\t-- with a syntax to modify the new tuple in memory\n\nThis all sounds good to me. Let's do it soon. I like the removal of\ni/u/d rewrite so we can give people something that will work, and not\nhave all those gray areas of 'it works here, but not here.'\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 14:15:09 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] triggers, views and rules (not instead)"
},
{
"msg_contents": "Maurice Gittens wrote:\n> \n> Hi,\n> \n> I've changed the large object memory managment strategy to ensure that\n> each large object has it's own memory context.\n> This way I can free all memory for a large object when I lo_close\n> is called for it's oid. This seems to work.\n> \n> I've noticed that the heap_close function used to close the heap used\n> by a large object doesn't really do any thing. (It calls RelationClose\n> which decrements some reference count).\n> \n> Somehow I have to free the relation from the cache in the following\n> situations:\n> 1. In a transaction I must free the stuff when the transaction is\n> commited/aborted.\n\nBackend does it, don't worry.\n\n> 2. Otherwise it must happen when lo_close is called.\n\nIt seems that you can't remove relation from cache untill\ncommit/abort, currently: backend uses local cache to unlink\nfiles of relations created in transaction if abort...\nWe could change relcache.c:RelationPurgeLocalRelation()\nto read from pg_class directly...\n\nBut how many LO do you create in single xact ?\nIs memory allocated for cache so big ?\n\nVadim\n",
"msg_date": "Sun, 22 Feb 1998 17:47:07 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] How To free resources used by large object Relations?"
},
{
"msg_contents": "Zeugswetter Andreas SARZ wrote:\n> Ok, to sum it up:\n> 1. We need and want the select part of the rewrite rules.\n\nAgreed.\n\n> 2. for the insert/update/delete rules the old instance rules system\n> was much more appropriate. TODO: dig up the old code\n> and merge it with the current trigger Implementation\n> it must be pretty much the wanted functionality (it\n> supported sql)\n\n??? Old instance rules system was removed by Jolly & Andrew and so\nit never supported SQL. I hope that Jan will give us PL/pgSQL soon\nand it will be used for triggers, without changing current trigger \nimplementation...\n\n> 3. the CURRENT keyword in the i/u/d rewrite rules is stupid\n> and should be disabled, destroyed and burned in hell\n\nAgreed, if standard hasn't it. I know that OLD & NEW are in standard,\nfor triggers atleast.\n\n> 4. To stick to the mainstream we should enhance the trigger\n> syntax, and forget the rule stuff for i/u/d\n\nYes. Statement level triggers give the same functionality as rewrite \ni/u/d rules. We could let them to return something special to skip \nuser' i/u/d itself, isn't it the same as INSTEAD ?\n\nVadim\n",
"msg_date": "Sun, 22 Feb 1998 18:26:45 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] triggers, views and rules (not instead)"
},
{
"msg_contents": "\nOn Sun, 22 February 1998, at 18:26:45, Vadim B. Mikheev wrote:\n\n> ??? Old instance rules system was removed by Jolly & Andrew and so\n> it never supported SQL. I hope that Jan will give us PL/pgSQL soon\n> and it will be used for triggers, without changing current trigger \n> implementation...\n\nIs develemopment being done for PL/pgSQL? What are peoples ideas for\nthis? I've never used a commercial db before, and the free ones don't\nusualle have a stored PL language. What sort of things are you guys\nanticipating? In the writing of PL/perl i've been tempted to give\nlots of access to the backend internals from perl (why not, it should\nhave all the facilities C function programmers have!) What do you think?\n\nAlso, as far as argument passing goes: strings & numbers get passed as\nperl scalars, and most other types get passed as a Posgres::Type\nobject (with methods for conversion etc). Right now I've got a switch\nblock on the type oid and I have many case statements and a few bodies\nfor these conversions.\n\nThe conversions are hard-coded in the .c file (via the case\nstatements). The only reason any particular input type shows up in\nperl any particular way (as a scalar (string/int) or Postgres::Type)\nis because there's a hardcoded case statement for it. Of course, the\ndefault is a Postgres::Type. Which means new integer types show up as\na Postgres::Type, which could be considered a bad thing. Right now\npart of what i'm doing is checking the typbyval and then passing that\ntype as an integer scalar (excluding selected types that don't make\nsense as ints).. Maybe the default case should be the types ouptut\nfunction?\n\nI hope I make sense..\n\n--brett\n\n> \n> > 3. the CURRENT keyword in the i/u/d rewrite rules is stupid\n> > and should be disabled, destroyed and burned in hell\n> \n> Agreed, if standard hasn't it. I know that OLD & NEW are in standard,\n> for triggers atleast.\n> \n> > 4. To stick to the mainstream we should enhance the trigger\n> > syntax, and forget the rule stuff for i/u/d\n> \n> Yes. Statement level triggers give the same functionality as rewrite \n> i/u/d rules. We could let them to return something special to skip \n> user' i/u/d itself, isn't it the same as INSTEAD ?\n> \n> Vadim\nOn Sun, 22 February 1998, at 03:33:07, [email protected] wrote:\n\n> X-VM-v5-Data: ([t nil nil nil nil nil nil nil nil]\n> \t[\"9840\" \"Sun\" \"22\" \"February\" \"1998\" \"03:33:07\" \"-0800\" \"[email protected]\" \"[email protected]\" nil \"273\" \"cron: /etc/dailyback \" nil nil nil \"2\" nil nil (number \" \" mark \"N [email protected] Feb 22 273/9840 \" thread-indent \"\\\"cron: /etc/dailyback \\\"\\n\") nil nil]\n> \tnil)\n> Return-Path: <[email protected]>\n> Received: from eve.speakeasy.org ([email protected] [199.238.226.1])\n> \tby abraxas.scene.com (8.8.8/8.8.5) with ESMTP id DAA16378\n> \tfor <[email protected]>; Sun, 22 Feb 1998 03:27:21 -0800\n> Received: from bigfoot.speakeasy.org (bigfoot.speakeasy.org [199.238.226.54]) by eve.speakeasy.org (8.8.5/8.7.3) with ESMTP id DAA03076; Sun, 22 Feb 1998 03:27:09 -0800 (PST)\n> From: [email protected]\n> Received: (from root@localhost)\n> \tby bigfoot.speakeasy.org (8.8.7/8.8.7) id DAA23856;\n> \tSun, 22 Feb 1998 03:33:07 -0800\n> Date: Sun, 22 Feb 1998 03:33:07 -0800\n> Message-Id: <[email protected]>\n> To: [email protected]\n> Subject: cron: /etc/dailyback \n> \n> ARCHIVING: eve\n> Begin time: Sun Feb 22 02:14:21 PST 1998\n> \n> DUMP: Date of this level 7 dump: Sun Feb 22 02:14:21 1998\n> DUMP: Date of last level 3 dump: Sat Feb 21 04:26:20 1998\n> DUMP: Dumping /dev/rsd3a (/home) to /dev/nrmt0 on host bigfoot\n> DUMP: mapping (Pass I) [regular files]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: estimated 127906 blocks (62.45MB) on 0.01 tape(s).\n> DUMP: dumping (Pass III) [directories]\n> DUMP: dumping (Pass IV) [regular files]\n> DUMP: level 7 dump on Sun Feb 22 02:14:21 1998\n> DUMP: Tape rewinding\n> DUMP: 128490 blocks (62.74MB) on 1 volume\n> DUMP: DUMP IS DONE\n> DUMP: Date of this level 7 dump: Sun Feb 22 02:17:28 1998\n> DUMP: Date of last level 3 dump: Sat Feb 21 04:31:51 1998\n> DUMP: Dumping /dev/rsd3d (/sp1) to /dev/nrmt0 on host bigfoot\n> DUMP: mapping (Pass I) [regular files]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: estimated 84136 blocks (41.08MB) on 0.00 tape(s).\n> DUMP: dumping (Pass III) [directories]\n> DUMP: dumping (Pass IV) [regular files]\n> DUMP: level 7 dump on Sun Feb 22 02:17:28 1998\n> DUMP: Tape rewinding\n> DUMP: 84222 blocks (41.12MB) on 1 volume\n> DUMP: DUMP IS DONE\n> DUMP: Date of this level 7 dump: Sun Feb 22 02:20:38 1998\n> DUMP: Date of last level 3 dump: Sat Feb 21 04:38:31 1998\n> DUMP: Dumping /dev/rsd2b (/sp2) to /dev/nrmt0 on host bigfoot\n> DUMP: mapping (Pass I) [regular files]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: estimated 109538 blocks (53.49MB) on 0.01 tape(s).\n> DUMP: dumping (Pass III) [directories]\n> DUMP: dumping (Pass IV) [regular files]\n> DUMP: level 7 dump on Sun Feb 22 02:20:38 1998\n> DUMP: Tape rewinding\n> DUMP: 109562 blocks (53.50MB) on 1 volume\n> DUMP: DUMP IS DONE\n> DUMP: Date of this level 7 dump: Sun Feb 22 02:25:28 1998\n> DUMP: Date of last level 3 dump: Sat Feb 21 04:48:03 1998\n> DUMP: Dumping /dev/rsd3e (/sp3) to /dev/nrmt0 on host bigfoot\n> DUMP: mapping (Pass I) [regular files]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: estimated 121746 blocks (59.45MB) on 0.01 tape(s).\n> DUMP: dumping (Pass III) [directories]\n> DUMP: dumping (Pass IV) [regular files]\n> DUMP: level 7 dump on Sun Feb 22 02:25:28 1998\n> DUMP: Tape rewinding\n> DUMP: 121760 blocks (59.45MB) on 1 volume\n> DUMP: DUMP IS DONE\n> DUMP: Date of this level 7 dump: Sun Feb 22 02:31:30 1998\n> DUMP: Date of last level 3 dump: Sat Feb 21 05:01:18 1998\n> DUMP: Dumping /dev/rsd3f (/usr/spool/mail) to /dev/nrmt0 on host bigfoot\n> DUMP: mapping (Pass I) [regular files]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: estimated 709654 blocks (346.51MB) on 0.04 tape(s).\n> DUMP: dumping (Pass III) [directories]\n> DUMP: dumping (Pass IV) [regular files]\n> DUMP: 42.60% done, finished in 0:06\n> DUMP: 84.00% done, finished in 0:01\n> DUMP: level 7 dump on Sun Feb 22 02:31:30 1998\n> DUMP: Tape rewinding\n> DUMP: 710144 blocks (346.75MB) on 1 volume\n> DUMP: DUMP IS DONE\n> DUMP: Date of this level 7 dump: Sun Feb 22 02:45:06 1998\n> DUMP: Date of last level 3 dump: Sat Feb 21 05:17:49 1998\n> DUMP: Dumping /dev/rsd1a (/usr/local/etc/httpd) to /dev/nrmt0 on host bigfoot\n> DUMP: mapping (Pass I) [regular files]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: estimated 12236 blocks (5.97MB) on 0.00 tape(s).\n> DUMP: dumping (Pass III) [directories]\n> DUMP: dumping (Pass IV) [regular files]\n> DUMP: level 7 dump on Sun Feb 22 02:45:06 1998\n> DUMP: Tape rewinding\n> DUMP: 12454 blocks (6.08MB) on 1 volume\n> DUMP: DUMP IS DONE\n> DUMP: Date of this level 7 dump: Sun Feb 22 02:46:59 1998\n> DUMP: Date of last level 3 dump: Sat Feb 21 05:20:04 1998\n> DUMP: Dumping /dev/rsd2a (/usr/local) to /dev/nrmt0 on host bigfoot\n> DUMP: mapping (Pass I) [regular files]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: estimated 576498 blocks (281.49MB) on 0.03 tape(s).\n> DUMP: dumping (Pass III) [directories]\n> DUMP: dumping (Pass IV) [regular files]\n> DUMP: 97.17% done, finished in 0:00\n> DUMP: level 7 dump on Sun Feb 22 02:46:59 1998\n> DUMP: Tape rewinding\n> DUMP: 577062 blocks (281.77MB) on 1 volume\n> DUMP: DUMP IS DONE\n> DUMP: Date of this level 7 dump: Sun Feb 22 02:54:18 1998\n> DUMP: Date of last level 3 dump: Sat Feb 21 05:28:41 1998\n> DUMP: Dumping /dev/rsd0a (/) to /dev/nrmt0 on host bigfoot\n> DUMP: mapping (Pass I) [regular files]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: estimated 4658 blocks (2.27MB) on 0.00 tape(s).\n> DUMP: dumping (Pass III) [directories]\n> DUMP: dumping (Pass IV) [regular files]\n> DUMP: level 7 dump on Sun Feb 22 02:54:18 1998\n> DUMP: Tape rewinding\n> DUMP: 4666 blocks (2.28MB) on 1 volume\n> DUMP: DUMP IS DONE\n> DUMP: Date of this level 7 dump: Sun Feb 22 02:54:28 1998\n> DUMP: Date of last level 3 dump: Sat Feb 21 05:29:03 1998\n> DUMP: Dumping /dev/rsd0g (/usr) to /dev/nrmt0 on host bigfoot\n> DUMP: mapping (Pass I) [regular files]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: mapping (Pass II) [directories]\n> DUMP: estimated 4516 blocks (2.21MB) on 0.00 tape(s).\n> DUMP: dumping (Pass III) [directories]\n> DUMP: dumping (Pass IV) [regular files]\n> DUMP: level 7 dump on Sun Feb 22 02:54:28 1998\n> DUMP: Tape rewinding\n> DUMP: 4512 blocks (2.20MB) on 1 volume\n> DUMP: DUMP IS DONE\n> End time: Sun Feb 22 02:54:59 PST 1998\n> ARCHIVING: gemini\n> gemini:/\n> \n> **** bru: execution summary ****\n> \n> Started:\t\tSun Feb 22 02:57:09 1998\n> Completed:\t\tSun Feb 22 02:57:50 1998\n> Archive id:\t\t34f004852588\n> Messages:\t\t0 warnings, 0 errors\n> Archive I/O:\t\t6920 blocks (13840Kb) written\n> Archive I/O:\t\t0 blocks (0Kb) read\n> Files written:\t\t18 files (15 regular, 3 other)\n> Files read:\t\t0 files (0 regular, 0 other)\n> Write errors:\t\t0 soft, 0 hard\n> Read errors:\t\t0 soft, 0 hard\n> Checksum errors:\t0\n> gemini:/usr\n> \n> **** bru: execution summary ****\n> \n> Started:\t\tSun Feb 22 02:57:50 1998\n> Completed:\t\tSun Feb 22 02:58:34 1998\n> Archive id:\t\t34f004ae2595\n> Messages:\t\t0 warnings, 0 errors\n> Archive I/O:\t\t10 blocks (20Kb) written\n> Archive I/O:\t\t0 blocks (0Kb) read\n> Files written:\t\t1 files (1 regular, 0 other)\n> Files read:\t\t0 files (0 regular, 0 other)\n> Write errors:\t\t0 soft, 0 hard\n> Read errors:\t\t0 soft, 0 hard\n> Checksum errors:\t0\n> gemini:/usr/local\n> \n> **** bru: execution summary ****\n> \n> Started:\t\tSun Feb 22 02:58:34 1998\n> Completed:\t\tSun Feb 22 03:00:45 1998\n> Archive id:\t\t34f004da259d\n> Messages:\t\t0 warnings, 0 errors\n> Archive I/O:\t\t22980 blocks (45960Kb) written\n> Archive I/O:\t\t0 blocks (0Kb) read\n> Files written:\t\t419 files (419 regular, 0 other)\n> Files read:\t\t0 files (0 regular, 0 other)\n> Write errors:\t\t0 soft, 0 hard\n> Read errors:\t\t0 soft, 0 hard\n> Checksum errors:\t0\n> gemini:/usr/local/apache/sites\n> \n> **** bru: execution summary ****\n> \n> Started:\t\tSun Feb 22 03:00:45 1998\n> Completed:\t\tSun Feb 22 03:14:29 1998\n> Archive id:\t\t34f0055d25ba\n> Messages:\t\t0 warnings, 0 errors\n> Archive I/O:\t\t48900 blocks (97800Kb) written\n> Archive I/O:\t\t0 blocks (0Kb) read\n> Files written:\t\t814 files (814 regular, 0 other)\n> Files read:\t\t0 files (0 regular, 0 other)\n> Write errors:\t\t0 soft, 0 hard\n> Read errors:\t\t0 soft, 0 hard\n> Checksum errors:\t0\n> gemini:/mnt/logs\n> \n> **** bru: execution summary ****\n> \n> Started:\t\tSun Feb 22 03:14:29 1998\n> Completed:\t\tSun Feb 22 03:17:10 1998\n> Archive id:\t\t34f00895260f\n> Messages:\t\t0 warnings, 0 errors\n> Archive I/O:\t\t20 blocks (40Kb) written\n> Archive I/O:\t\t0 blocks (0Kb) read\n> Files written:\t\t5 files (5 regular, 0 other)\n> Files read:\t\t0 files (0 regular, 0 other)\n> Write errors:\t\t0 soft, 0 hard\n> Read errors:\t\t0 soft, 0 hard\n> Checksum errors:\t0\n> ARCHIVING: betty\n> betty:/usr\n> \n> **** bru: execution summary ****\n> \n> Started:\t\tSun Feb 22 03:15:42 1998\n> Completed:\t\tSun Feb 22 03:16:48 1998\n> Archive id:\t\t34f008de0465\n> Messages:\t\t0 warnings, 0 errors\n> Archive I/O:\t\t3980 blocks (7960Kb) written\n> Archive I/O:\t\t0 blocks (0Kb) read\n> Files written:\t\t8 files (8 regular, 0 other)\n> Files read:\t\t0 files (0 regular, 0 other)\n> Write errors:\t\t0 soft, 0 hard\n> Read errors:\t\t0 soft, 0 hard\n> Checksum errors:\t0\n> betty:/usr/local/samba\n> \n> **** bru: execution summary ****\n> \n> Started:\t\tSun Feb 22 03:16:48 1998\n> Completed:\t\tSun Feb 22 03:17:46 1998\n> Archive id:\t\t34f00920046c\n> Messages:\t\t0 warnings, 0 errors\n> Archive I/O:\t\t3630 blocks (7260Kb) written\n> Archive I/O:\t\t0 blocks (0Kb) read\n> Files written:\t\t55 files (55 regular, 0 other)\n> Files read:\t\t0 files (0 regular, 0 other)\n> Write errors:\t\t0 soft, 0 hard\n> Read errors:\t\t0 soft, 0 hard\n> Checksum errors:\t0\n> ARCHIVING: bigfoot\n> bigfoot:/var/log0\n> \n> **** bru: execution summary ****\n> \n> Started:\t\tSun Feb 22 03:23:13 1998\n> Completed:\t\tSun Feb 22 03:30:26 1998\n> Archive id:\t\t34f00aa15d23\n> Messages:\t\t0 warnings, 0 errors\n> Archive I/O:\t\t272880 blocks (545760Kb) written\n> Archive I/O:\t\t0 blocks (0Kb) read\n> Files written:\t\t937 files (937 regular, 0 other)\n> Files read:\t\t0 files (0 regular, 0 other)\n> Write errors:\t\t0 soft, 0 hard\n> Read errors:\t\t0 soft, 0 hard\n> Checksum errors:\t0\n> ARCHIVING: ella.pscs.org\n> ella:/var\n> \n> **** bru: execution summary ****\n> \n> Started:\t\tSun Feb 22 03:27:39 1998\n> Completed:\t\tSun Feb 22 03:28:46 1998\n> Archive id:\t\t34f00bab08e0\n> Messages:\t\t0 warnings, 0 errors\n> Archive I/O:\t\t13830 blocks (27660Kb) written\n> Archive I/O:\t\t0 blocks (0Kb) read\n> Files written:\t\t58 files (58 regular, 0 other)\n> Files read:\t\t0 files (0 regular, 0 other)\n> Write errors:\t\t0 soft, 0 hard\n> Read errors:\t\t0 soft, 0 hard\n> Checksum errors:\t0\nOn Sun, 22 February 1998, at 17:47:07, Vadim B. Mikheev wrote:\n\n> X-VM-v5-Data: ([t nil nil nil nil nil nil nil nil]\n> \t[\"1080\" \"Sun\" \"22\" \"February\" \"1998\" \"17:47:07\" \"+0700\" \"Vadim B. Mikheev\" \"[email protected]\" nil \"32\" \"Re: [HACKERS] How To free resources used by large object Relations?\" nil nil nil \"2\" nil nil (number \" \" mark \"N Vadim B. Mikheev Feb 22 32/1080 \" thread-indent \"\\\"Re: [HACKERS] How To free resources used by large object Relations?\\\"\\n\") nil nil]\n> \tnil)\n> Return-Path: <[email protected]>\n> Received: from hub.org (hub.org [209.47.148.200])\n> \tby abraxas.scene.com (8.8.8/8.8.5) with ESMTP id CAA16288\n> \tfor <[email protected]>; Sun, 22 Feb 1998 02:51:20 -0800\n> Received: from localhost (majordom@localhost) by hub.org (8.8.8/8.7.5) with SMTP id FAA28144; Sun, 22 Feb 1998 05:45:06 -0500 (EST)\n> Received: by hub.org (TLB v0.10a (1.23 tibbs 1997/01/09 00:29:32)); Sun, 22 Feb 1998 05:44:57 -0500 (EST)\n> Received: (from majordom@localhost) by hub.org (8.8.8/8.7.5) id FAA28023 for pgsql-hackers-outgoing; Sun, 22 Feb 1998 05:44:49 -0500 (EST)\n> Received: from dune.krasnet.ru (dune.krasnet.ru [193.125.44.86]) by hub.org (8.8.8/8.7.5) with ESMTP id FAA27920 for <[email protected]>; Sun, 22 Feb 1998 05:44:29 -0500 (EST)\n> Received: from sable.krasnoyarsk.su (dune.krasnet.ru [193.125.44.86])\n> \tby dune.krasnet.ru (8.8.7/8.8.7) with ESMTP id RAA00809;\n> \tSun, 22 Feb 1998 17:47:19 +0700 (KRS)\n> \t(envelope-from [email protected])\n> Message-ID: <[email protected]>\n> Date: Sun, 22 Feb 1998 17:47:07 +0700\n> From: \"Vadim B. Mikheev\" <[email protected]>\n> Organization: ITTS (Krasnoyarsk)\n> X-Mailer: Mozilla 4.04 [en] (X11; I; FreeBSD 2.2.5-RELEASE i386)\n> MIME-Version: 1.0\n> To: Maurice Gittens <[email protected]>\n> CC: [email protected]\n> Subject: Re: [HACKERS] How To free resources used by large object Relations?\n> References: <[email protected]>\n> Content-Type: text/plain; charset=us-ascii\n> Content-Transfer-Encoding: 7bit\n> Sender: [email protected]\n> Precedence: bulk\n> \n> Maurice Gittens wrote:\n> > \n> > Hi,\n> > \n> > I've changed the large object memory managment strategy to ensure that\n> > each large object has it's own memory context.\n> > This way I can free all memory for a large object when I lo_close\n> > is called for it's oid. This seems to work.\n> > \n> > I've noticed that the heap_close function used to close the heap used\n> > by a large object doesn't really do any thing. (It calls RelationClose\n> > which decrements some reference count).\n> > \n> > Somehow I have to free the relation from the cache in the following\n> > situations:\n> > 1. In a transaction I must free the stuff when the transaction is\n> > commited/aborted.\n> \n> Backend does it, don't worry.\n> \n> > 2. Otherwise it must happen when lo_close is called.\n> \n> It seems that you can't remove relation from cache untill\n> commit/abort, currently: backend uses local cache to unlink\n> files of relations created in transaction if abort...\n> We could change relcache.c:RelationPurgeLocalRelation()\n> to read from pg_class directly...\n> \n> But how many LO do you create in single xact ?\n> Is memory allocated for cache so big ?\n> \n> Vadim\nOn Sun, 22 February 1998, at 02:49:37, Eve Arden wrote:\n\n> X-VM-v5-Data: ([t nil nil nil nil nil nil nil nil]\n> \t[\"129\" \"Sun\" \"22\" \"February\" \"1998\" \"02:49:37\" \"-0800\" \"Eve Arden\" \"[email protected]\" nil \"2\" \"\" nil nil nil \"2\" nil nil (number \" \" mark \"N Eve Arden Feb 22 2/129 \" thread-indent \"\\\"\\\"\\n\") nil nil]\n> \tnil)\n> Return-Path: <[email protected]>\n> Received: from eve.speakeasy.org ([email protected] [199.238.226.1])\n> \tby abraxas.scene.com (8.8.8/8.8.5) with ESMTP id CAA16276\n> \tfor <[email protected]>; Sun, 22 Feb 1998 02:49:50 -0800\n> Received: (from root@localhost) by eve.speakeasy.org (8.8.5/8.7.3) id CAA01899; Sun, 22 Feb 1998 02:49:37 -0800 (PST)\n> Date: Sun, 22 Feb 1998 02:49:37 -0800 (PST)\n> From: Eve Arden <[email protected]>\n> Message-Id: <[email protected]>\n> \n> # Checking accounts from NIS.\n> >--WARN-- [acc001w] Login ID a-ron is disabled, but still has a valid shell (/usr/local/bin/tcsh).\n",
"msg_date": "Sun, 22 Feb 1998 21:33:27 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "pl/{perl, pgsql} (was Re: AW: [HACKERS] triggers,\n\tviews and rules (not instead))"
},
{
"msg_contents": "\nVadim wrote:\n>\n> ??? Old instance rules system was removed by Jolly & Andrew and so\n> it never supported SQL. I hope that Jan will give us PL/pgSQL soon\n> and it will be used for triggers, without changing current trigger\n> implementation...\n\n I'll start on PL/pgSQL as soon as the view-aggregate, insert-\n select-view and the pg_user-freed-cc_tupdesc problems are\n fixed. But due to time I think PL/pgSQL will be an add on\n after 6.3 and I'm very sure it will not require any more\n changes to the trigger implementation.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 23 Feb 1998 09:19:58 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: AW: [HACKERS] triggers, views and rules (not instead)"
},
{
"msg_contents": "\nBrett wrote:\n>\n>\n> On Sun, 22 February 1998, at 18:26:45, Vadim B. Mikheev wrote:\n>\n> > ??? Old instance rules system was removed by Jolly & Andrew and so\n> > it never supported SQL. I hope that Jan will give us PL/pgSQL soon\n> > and it will be used for triggers, without changing current trigger\n> > implementation...\n>\n> Is develemopment being done for PL/pgSQL? What are peoples ideas for\n> this? I've never used a commercial db before, and the free ones don't\n> usualle have a stored PL language. What sort of things are you guys\n> anticipating? In the writing of PL/perl i've been tempted to give\n> lots of access to the backend internals from perl (why not, it should\n> have all the facilities C function programmers have!) What do you think?\n\n No actual development - just have something in mind how I\n would implement it. I'll get into details after 6.3 release.\n PL/pgSQL will have at least the following capabilities:\n\n - local variable\n - local records\n - access to the database over SPI\n - control structures (if/else/while/loop)\n - elog messages\n - triggers can modify new tuple\n - triggers can skip operation\n\n Why not handing many backend internals through a PL? Just to\n let ordinary users use the language without breaking\n security. The implementation of PL/Tcl uses a safe Tcl\n interpreter for the evaluation of the Tcl functions/trigger-\n procedures. A safe Tcl interpreter has very limited command\n set. No access to filesystem, no access to networking, no\n loading of other packages so nothing dangerous.\n\n If perl doesn't have such a restricted interpreter facility,\n then perl might never become a TRUSTED procedural language\n like Tcl is. Remember, if the PL/perl implementation gives\n an ordinary user the right just to open a file for writing, a\n function independent who created it could damage database\n files directly. This MUST be restricted to superusers and\n this is the reason why the C language isn't TRUSTED.\n\n>\n> Also, as far as argument passing goes: strings & numbers get passed as\n> perl scalars, and most other types get passed as a Posgres::Type\n> object (with methods for conversion etc). Right now I've got a switch\n> block on the type oid and I have many case statements and a few bodies\n> for these conversions.\n>\n> The conversions are hard-coded in the .c file (via the case\n> statements). The only reason any particular input type shows up in\n> perl any particular way (as a scalar (string/int) or Postgres::Type)\n> is because there's a hardcoded case statement for it. Of course, the\n> default is a Postgres::Type. Which means new integer types show up as\n> a Postgres::Type, which could be considered a bad thing. Right now\n> part of what i'm doing is checking the typbyval and then passing that\n> type as an integer scalar (excluding selected types that don't make\n> sense as ints).. Maybe the default case should be the types ouptut\n> function?\n\n Whatever perl scalars are - don't know. Tcl is happy with\n string representation of anything on the evaluation level and\n Tcl_Eval() leaves a string in the interpreter result. So I\n used the registered input/output functions to convert\n anything from/into strings to communicate with the\n interpreter.\n\n Whatever is given to or returned from a Tcl function in\n PL/Tcl will be in the external representation. This is what\n the user sees if he selects from a table in psql and what has\n to be given on insert/update. So on the PL/Tcl level,\n anything is the external representation as defined by the\n types registered input/output functions. This is the values\n format humans deal with best.\n\n I know that the way I choose isn't the performance optimized\n one. But blowing up the functions code by handling binary\n data wouldn't be either.\n\n When implementing perl or Tcl as a procedural language, we\n must satisfy the need of perl/Tcl programmers. These\n programmers must not have any C skill, but they sure will\n know how to read/create a string in the external\n representation. For some types (especially for user created\n types) it might be hard to figure out what each bit in the\n binary Datum is for if you aren't familiar with C.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 23 Feb 1998 10:41:12 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: pl/{perl, pgsql} (was Re: AW: [HACKERS] triggers,\n\tviews and rules (not instead))"
},
{
"msg_contents": "\nPlease don't CC the root users (or the mailing list twice) on this.\nI've removed them from the headers. Sorry, I don't know how the\naddresses got in there.\n\n> Brett wrote:\n> >\n> >\n> > On Sun, 22 February 1998, at 18:26:45, Vadim B. Mikheev wrote:\n> >\n> > Is develemopment being done for PL/pgSQL? What are peoples ideas for\n> > this? I've never used a commercial db before, and the free ones don't\n> > usualle have a stored PL language. What sort of things are you guys\n> > anticipating? In the writing of PL/perl i've been tempted to give\n> > lots of access to the backend internals from perl (why not, it should\n> > have all the facilities C function programmers have!) What do you think?\n> \n",
"msg_date": "Mon, 23 Feb 1998 02:41:45 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: pl/{perl, pgsql} (was Re: AW: [HACKERS] triggers,\n\tviews and rules (not instead))"
}
] |
[
{
"msg_contents": "Jan wrote:\n\n The only things not working for copy are rewrite rules. But I\n think we should restrict rules to the view handling in the\n future and move forward by implementing a pure and really\n powerful procedural language.\n\nHm, it looks like you are not really a fan of the rewrite system,\neventhough you seem to have the most insight in these matters. I wonder why?\n\nWhy I like the rewrite system is:\n\t1. select rewrite -- select trigger would be no good (optimizer)\n\t2. The client can be really dumb, like MS Access or some other\nstandard ODBC tool\n\t\twhich does not know anything about funcs procs and the like\n\t\t(even without using passthrough)\n\t3. it is a lot more powerful than views\n\t4. it allows the optimizer to get involved (this is where triggers\nfail per definition)\n\t5. once understood it is very easy to use\n\t\teasier than trigger with c stored procedure at least\n\t\nI guess if triggers could also trigger simple select statements, I could do\nmost of what I want using triggers except of course the select stuff.\nBut as I said I like the rules system very much, especially after your\nrecent\nfixes Jan :-) So please stick to supporting all 3: triggers, views and\nrules. Wow :-)\n\nAndreas\n",
"msg_date": "Fri, 20 Feb 1998 15:46:46 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "triggers, views and rules (not instead)"
},
{
"msg_contents": "\nAndreas wrote:\n>\n> Jan wrote:\n>\n> The only things not working for copy are rewrite rules. But I\n> think we should restrict rules to the view handling in the\n> future and move forward by implementing a pure and really\n> powerful procedural language.\n>\n> Hm, it looks like you are not really a fan of the rewrite system,\n> eventhough you seem to have the most insight in these matters. I wonder why?\n\n Confusing - eh? Well I know much about the internals of the\n postgres rule system and due to this I know where the limits\n are. Especially in the case of qualifications it somtimes\n gets totally confused about what to compare against what. Try\n to add a delete rule on a view that is simply a select * from\n another table and then delete some tuples :-)\n\n>\n> Why I like the rewrite system is:\n> 1. select rewrite -- select trigger would be no good (optimizer)\n\n Exactly that's what is done if you create a view. Postgres\n creates a regular table (look at pg_class and into the\n database directory) and then sets up a relation level instead\n rewrite rule on select.\n\n> 2. The client can be really dumb, like MS Access or some other\n> standard ODBC tool\n> which does not know anything about funcs procs and the like\n> (even without using passthrough)\n\n Yupp - the client must not know why and how and where the\n data is left and coming from. But that's true in any case - a\n trigger for each row on insert can do anything different and\n push the data wherever it wants.\n\n> 3. it is a lot more powerful than views\n\n As said - views are only one special rule case in Postgres.\n\n> 4. it allows the optimizer to get involved (this is where triggers\n> fail per definition)\n> 5. once understood it is very easy to use\n> easier than trigger with c stored procedure at least\n\n Optimizing again and again. If the rules aren't instead, the\n querytree get's additional queries for every rule appended.\n Have a table field that references an entry in another table\n and this entry should have a refcount. So on update you must\n decrease the refcount from the old ref and increase it on the\n new. You create two rules so the UPDATE will result in 1\n scan and 2 nestloops with scans inside - really optimized if\n the referenced value doesn't change. And don't think that a\n rule qual of NEW != CURRENT might help - that will result in\n 2 mergejoins where the scanned tuples are compared.\n\n BTW, this sample doesn't work currently because the rules\n queries are appended at the end of the querytree, thus the\n decrement scan having the same qual will not find the old\n tuple at all because it's already outdated\n (command_counter_increment between processing the queries).\n Referencing CURRENT in a rule is not what most people think\n it is.\n\n The old 4.2 postgres had a second, instance level rule system\n (prs2 stubs) that fired the rules actions when actually the\n old tuple and the new projected tuple where handy. There you\n could have made also things like 'UPDATE NEW SET a = 4' that\n really modified the in memory tuple in the executors\n expression context. Who the hell removed all that? It was so\n nice :-(\n\n A really simple to write trigger can compare old != new and\n only if send down the other two queries. This time they wont\n be nestloops, they are simple scans. And the trigger can\n arrange that the queries it uses are only parsed on it's\n first of all calls and store the generated execution plans\n permanently for quick execution (look at SPI_prepare).\n\n For the stored C procedures you're totally right. I don't\n like the C functions because it requires postgres superuser\n rights to develop them and thus I created PL/Tcl where joe\n user can hack around without having complete access to the\n whole database (look at src/pl/tcl). And someday after 6.3\n release I'll really start on a plain PL/pgSQL implementation\n that would give a normal user the opportunity to create\n functions and triggers on a high level. There is light at the\n end of the tunnel - hope that it isn't the coming train :-)\n\n>\n> I guess if triggers could also trigger simple select statements, I could do\n> most of what I want using triggers except of course the select stuff.\n> But as I said I like the rules system very much, especially after your\n> recent\n> fixes Jan :-) So please stick to supporting all 3: triggers, views and\n> rules. Wow :-)\n\n Well - a trigger cannot build a view. The relation underlying\n the view doesn't contain any tuples and a select trigger will\n never be fired. As long as there is no possibility to return\n tuple sets from non-SQL functions. But a trigger can do\n things like the pg_hide_passwd stuff much more powerful. You\n could define the trigger so that it checks if the user is a\n superuser and overwrite the passwd value only in the case\n where he/she isn't. If fired at the right place it would too\n work for things like the copy command etc.\n\n We must stay with all 3 features. And I will take a look at\n the INSERT ... SELECT view problem really soon as it is a\n rule system problem that breaks views. But this is only the\n SELECT rewriting part of the rule system which I really like\n (optimizable). The other areas (insert, update, delete) of\n the rule system are dangerous and I really think a powerful\n PL/pgSQL language could make them obsolete.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Fri, 20 Feb 1998 17:52:38 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] triggers, views and rules (not instead)"
}
] |
[
{
"msg_contents": "> I have been working with many of you for over a hear, and would like to\n> know more details about you.\n> \n> ...\n> \n> Can others tell us about themselves?\n\nGreat idea, Bruce. Best thread in the last two weeks.\n\nI'm 29, single, and live in Bel Air, Maryland (only a touch over an hour\nfrom you Bruce...we have to do lunch someday. :) Have a math degree from\nIllinois, working on a masters from Johns Hopkins. Weekdays, I work at\nInsight Distribution Systems as a programmer writing software for beverage\ndistributors. Hobbies include traveling to professional dart tournaments\nand riding my motorcycle at twice the legal speed limit. To relax, I\ncollect baseball cards, fold origami, work out at the local Gold's Gym,\nand play any musical instrument I can find...\n\nSomewhere in there I try to keep the aix port alive and kicking.\n\nDarren King\n\n",
"msg_date": "Fri, 20 Feb 1998 10:13:36 -0500",
"msg_from": "[email protected] (Darren King)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Who is everyone?"
}
] |
[
{
"msg_contents": "> I have been working with many of you for over a year, and would like to\n> know more details about you.\n\nI am currently 27, to be honest 28 tomorrow. I studied Business Informatics\nat the TU Wien,\nafter a technically oriented high school. I work for an IT Company for\nbanks, actually \"the\" IT in Austria\nsoon called IT Austria (we are supposed to be proud to be one of the biggest\nin Europe).\nI am the System Expert for Informix and recently also Oracle on UNIX and NT\nPlatforms here.\nWe run ~1 Terra byte of Informix DB's here for SAP, Datawarehousing, Image\nScanning with OCR and the like. \nI (well actually my wife Maria) am expecting my first child in September,\nand looking forward to it.\nWe live 1/2 h south of Vienna, Austria. I also lived in New Canaan,\nConnecticut for 2 years when I was 14. \nI am writing a diploma paper on \"Integration of Neural Nets into the\nPostgreSQL Database System\".\nActually the program \"quelnns\" can already learn from a select statement,\nand then give answers\nto input that is not in the training select. My biggest problem now is to\nreturn the answer as a tuple from \na C function, right now I return text. The software uses postgresql, pvm3\nand SNNS.\nI like skiing, hiking, horses, swimming and travelling around the world\n(or should I already say I liked the recent?).\n\nYours sincerely\nAndreas\n",
"msg_date": "Fri, 20 Feb 1998 16:38:43 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Who is everyone?"
}
] |
[
{
"msg_contents": "\n> > On Fri, 20 Feb 1998, Bruce Momjian wrote:\n> > \n> > > Two things. First was a separate COPY priviledge, which I vote\n> against.\n> > > I see no real value to it, except to work around the problem that COPY\n> > > doesn't use rules.\n> > \n> > \tOkay, I may be totally out in left field here (ie. unrelated), but\n> > what stops a user from doing a 'COPY out' on a table that they don't\n> have\n> > SELECT privileges on? Kind of negates 'REVOKE ALL...', no?\n> \n> Yes I think a separate COPY permission makes no sense.\n> \n> > > Second, there was the idea of making copy allow a real select\n> statement\n> > > and not just a table name. If we do that, all goes through the\n> > > executor, and you get view and rules working properly. May have some\n> > > performance penalty, though it probabably will be minor.\n> > \n> > \tThis sounds reasonable...\n> >\n> \n> \nJust to make you comfortable, yes I also completely agree\n\nAndreas\n\n\n",
"msg_date": "Fri, 20 Feb 1998 18:56:12 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "AW: [HACKERS] Permissions on copy"
}
] |
[
{
"msg_contents": "Hi,\n\nI get the foll. error for a group by query:\nThis is with the snapshot of Feb 20, and I am running on sparc-solaris 2.6\n\ninsert into FTemp1 select item as item1, count(*) as cnt from data group by \nitem;\nERROR: The field being grouped by must appear in the target list\n\nThanks\n--shiby\n\n\n",
"msg_date": "Fri, 20 Feb 1998 13:55:29 -0500",
"msg_from": "Shiby Thomas <[email protected]>",
"msg_from_op": true,
"msg_subject": "group by problem"
},
{
"msg_contents": "> \n> Hi,\n> \n> I get the foll. error for a group by query:\n> This is with the snapshot of Feb 20, and I am running on sparc-solaris 2.6\n> \n> insert into FTemp1 select item as item1, count(*) as cnt from data group by \n> item;\n> ERROR: The field being grouped by must appear in the target list\n> \n> Thanks\n> --shiby\n\nSay:\n\ngroup by item1.\n\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 14:16:14 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] group by problem"
}
] |
[
{
"msg_contents": "Hey,\n\n the current snapshot dumps core on the 4th time doing\n\n REVOKE ALL ON pg_user FROM public;\n\n It does too in other situations but this is the simplest to\n reproduce. The segmentation fault happens in nocachegetattr()\n due to a destroyed tuple descriptor (natts = 0!!! and the\n others don't look good either) for the syscache 21 (USENAME).\n But the destruction must happen somewhere else.\n\n With the 02/13 snapshot I haven't got any problems on it.\n But cannot find the error with diff.\n\n BTW: Doing last checks on view permissions - sending a patch\n soon.\n\n\nUntil later, Jan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Fri, 20 Feb 1998 20:06:31 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Backend crashes - what's going on here???"
},
{
"msg_contents": "> \n> Hey,\n> \n> the current snapshot dumps core on the 4th time doing\n> \n> REVOKE ALL ON pg_user FROM public;\n> \n> It does too in other situations but this is the simplest to\n> reproduce. The segmentation fault happens in nocachegetattr()\n> due to a destroyed tuple descriptor (natts = 0!!! and the\n> others don't look good either) for the syscache 21 (USENAME).\n> But the destruction must happen somewhere else.\n> \n> With the 02/13 snapshot I haven't got any problems on it.\n> But cannot find the error with diff.\n> \n> BTW: Doing last checks on view permissions - sending a patch\n> soon.\n\nYep, I saw this too when testing my password acl null patch. Couldn't\nreproduce it, so I thought it was a fluke.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 14:27:08 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Backend crashes - what's going on here???"
},
{
"msg_contents": "\nWhow - gdb is a nice tool\n\n>\n> >\n> > Hey,\n> >\n> > the current snapshot dumps core on the 4th time doing\n> >\n> > REVOKE ALL ON pg_user FROM public;\n> >\n> > It does too in other situations but this is the simplest to\n> > reproduce. The segmentation fault happens in nocachegetattr()\n> > due to a destroyed tuple descriptor (natts = 0!!! and the\n> > others don't look good either) for the syscache 21 (USENAME).\n> > But the destruction must happen somewhere else.\n> >\n> > With the 02/13 snapshot I haven't got any problems on it.\n> > But cannot find the error with diff.\n> >\n> > BTW: Doing last checks on view permissions - sending a patch\n> > soon.\n>\n> Yep, I saw this too when testing my password acl null patch. Couldn't\n> reproduce it, so I thought it was a fluke.\n>\n> --\n> Bruce Momjian\n> [email protected]\n>\n\n Have a clue now what causes the crash. It happens when\n pg_user is looked up in the syscache. It must have to do with\n the fact that during initialization in miscinit.c on\n SetUserId() the user tuple is fetched using\n SearchSysCacheTuple(). Due to this the SysCache entry 21\n gets initialized but later on start transaction through the\n cache reset the memory for the cc_tupdesc in the cache is\n freed. So I assume when SetUserId() is called, the syscache\n is not ready for use yet.\n\n I don't have a solution right now. Is someone more familiar\n with the handling of the syscache during startup? Is\n SetUserId() just called a little too early or is the syscache\n unusable during InitPostgres at all?\n\n But the fact that CatalogCacheInitializeCache() is called\n only for pg_user during startup makes me feel sure that the\n lookup of the user using SearchSysCacheTuple() is wrong at\n this time. I think it sould be done without using the\n syscache.\n\n Back on monday - maybe with a solution.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Sat, 21 Feb 1998 16:29:39 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Backend crashes - what's going on here???"
},
{
"msg_contents": "\nUhhh - much more ugly than I thought first :-(\n\nI wrote:\n>\n>\n> Whow - gdb is a nice tool\n>\n> >\n> > >\n> > > Hey,\n> > >\n> > > the current snapshot dumps core on the 4th time doing\n> > >\n> > > REVOKE ALL ON pg_user FROM public;\n> > >\n> > > It does too in other situations but this is the simplest to\n> > > reproduce. The segmentation fault happens in nocachegetattr()\n> > > due to a destroyed tuple descriptor (natts = 0!!! and the\n> > > others don't look good either) for the syscache 21 (USENAME).\n> > > But the destruction must happen somewhere else.\n> > >\n> > > With the 02/13 snapshot I haven't got any problems on it.\n> > > But cannot find the error with diff.\n> > >\n> > > BTW: Doing last checks on view permissions - sending a patch\n> > > soon.\n> >\n> > Yep, I saw this too when testing my password acl null patch. Couldn't\n> > reproduce it, so I thought it was a fluke.\n> >\n> > --\n> > Bruce Momjian\n> > [email protected]\n> >\n>\n> Have a clue now what causes the crash. It happens when\n> pg_user is looked up in the syscache. It must have to do with\n> the fact that during initialization in miscinit.c on\n> SetUserId() the user tuple is fetched using\n> SearchSysCacheTuple(). Due to this the SysCache entry 21\n> gets initialized but later on start transaction through the\n> cache reset the memory for the cc_tupdesc in the cache is\n> freed. So I assume when SetUserId() is called, the syscache\n> is not ready for use yet.\n>\n> I don't have a solution right now. Is someone more familiar\n> with the handling of the syscache during startup? Is\n> SetUserId() just called a little too early or is the syscache\n> unusable during InitPostgres at all?\n>\n> But the fact that CatalogCacheInitializeCache() is called\n> only for pg_user during startup makes me feel sure that the\n> lookup of the user using SearchSysCacheTuple() is wrong at\n> this time. I think it sould be done without using the\n> syscache.\n>\n> Back on monday - maybe with a solution.\n\n The crash is due to the cache invalidations on updates to\n pg_class (and can happen too on updates to pg_attribute and\n others).\n\n When a tuple in pg_class or the others is modified, its cache\n invalidation causes a RelationFlushRelation() for the\n affected relation. revoking from pg_user e.g. means that\n RelationFlushRelation() is called for pg_user but this frees\n the tuple desctiptor. The tuple descriptor is also used in\n the SysCache, and this isn't flushed/freed!\n\n There are more possible errors on this. A simple\n\n UPDATE pg_class SET relname = relname;\n\n let's the backend crash on the very next command. And\n\n REVOKE ALL ON pg_class FROM public;\n\n crashes immediately because the cache invalidation needs the\n just invalidated heap tuple for pg_class in pg_class. Sounds\n a bit hairy.\n\n I think this is also the reason for backend crashes I had\n when defining rewrite rules on relations that already exist\n (where I expect others that already noticed them).\n\n I still don't have the solution. But this must get fixed\n before releasing 6.3. I think a walk through the SysCache on\n RelationFlushRelation() looking if this relation is in the\n SysCache and if found resetting this cache can help (except\n for the revoke on pg_class).\n\n Append this to TODO!\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 23 Feb 1998 13:06:21 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Backend crashes - what's going on here???"
}
] |
[
{
"msg_contents": "> \n> \n> => > insert into FTemp1 select item as item1, count(*) as cnt from data group by \n> => > item;\n> => > ERROR: The field being grouped by must appear in the target list\n> => > \n> => > Thanks\n> => > --shiby\n> => \n> => Say:\n> => \n> => group by item1.\n> It works.\n> => \n> But, just wanted to check this also.\n> I get the same error for:\n> insert into FTemp1 select item, count(*) as cnt from data group by item\n> also.\n> \n> So, is it mandatory now to give alias names for all the selected columns if\n> we are using group by\n\nNo, I am sorry. Looks like we have a bug where GROUP BY can't be used\nwith INSERT ... SELECT.\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 14:29:08 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] group by problem"
}
] |
[
{
"msg_contents": "I'm just listening in on the list, but I'm hoping I can dive in and\nhelp once I'm up-to-speed on the education part. I'm using MySQL to\ndevelop my Visual Basic 5 programming skills, but once PostgreSQL\ncatches up on ODBC, I'll be moving to PG.\n\nI am a 21-yr-old native Floridian (here in the US) and have been married\nfor over a year. I will graduate this semester with a degree in CIS from\nthe University of Florida. Going along with the whole pet theme, I have\ntwo cats (well kittens actually), named Spartacus and Cervantes.\n\nIn college, my senior project was to design a mini-DBMS using the PERL\nlanguage, but set it up such that it could be installed and run from a\nnormal user's account of 4Mb quota. I used DBM files to store each\nrecord, with the 'KEY' acting as the Primary Key for the tables. This\nworked rather well, and I was able to then add a small subset of SQL to\nselect, insert, delete, and update tables. Of course, no security,\ndatatypes, transactions, or anything that makes a database functional\nexisted, but the whole program (DBMS) fit in just 60k and a 10 tables of\n10,000 tuples manages to stay under 1 Mb altogether..., so I accomplished\nmy goal ;)\n\nFor the past two years, I have worked for Accounting Firms Associated, inc.,\nwhere I am the Network Admin. I run a couple redhat 5 linux servers\n(connected to the internet through ISDN), two NT servers, one Novell\nserver, and about 35+ workstations. I prefer Linux of the three servers\nand have spent several years working with it since the 1.0 kernels and\nslackware (remember when you had to make all those disk sets...YUK!)\n\nAt home, I have three machines, one Redhat 5 server, and two Win95\nworkstations. My home server dial-up to the internet and connects my home\nPCs with IP masquerading (so my wife and I can share a phone line), and I'm\nusing Samba to allow the Redhat 5 server to authenticate logins and store\nuser profiles. I'm running MySQL and PostgreSQl 6.2.1 until I finish\ndesigning my current project.\n\nMy immediate goal is to integrate all of this firms information pools\ninto large databases which can be accessed via internal and external\nmethods. My ideal solution is to have Win95 on all desktops (for Office\nand stuff) with VB5 applications interfacing with the DB. Then, on the\nlinux\nmachines I'll have PERL and DBI/DBD interfaces generating pages for Apache\nto serve up. With that, costs stay low, and I learn a whole lot about\ntons of systems and languages and protocols etc...\n\nDante\n\nPS - Who needs graduate school? It'll only interfere with my education!\n\n.------------------------------------------.-----------------------.\n| _ [email protected] - D. Dante Lorenso | Network Administrator |\n| | | ___ _ _ ___ __ _ ___ ___ | |\n| | |__ / o \\| '_|/ o_\\| \\ |\\_ _\\/ o \\ | Accounting Firms |\n| |____|\\___/|_| \\___/|_|\\_|\\___|\\___/ | Associated, inc. |\n| http://www.afai.com/~dlorenso | http://www.afai.com/ |\n'------------------------------------------'-----------------------'\n\n\n",
"msg_date": "Fri, 20 Feb 1998 15:04:09 -0500",
"msg_from": "\"D. Dante Lorenso\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Who is everyone?"
}
] |
[
{
"msg_contents": "First step done,\n\n below is the patch to have views to override the permission\n checks for the accessed tables. Now we can do the following:\n\n CREATE VIEW db_user AS SELECT\n usename,\n usesysid,\n usecreatedb,\n usetrace,\n usecatupd,\n '**********'::text as passwd,\n valuntil\n FROM pg_user;\n\n REVOKE ALL ON pg_user FROM public;\n REVOKE ALL ON db_user FROM public;\n GRANT SELECT ON db_user TO public;\n\n And after setting up a non-privileged user account the user\n does\n\n wieck=> select usename, usesysid, passwd from pg_user;\n ERROR: pg_user: Permission denied.\n wieck=> select usename, usesysid, passwd from db_user;\n usename|usesysid|passwd\n -------+--------+----------\n pgsql | 24|**********\n wieck | 201|**********\n (2 rows)\n\n wieck=>\n\n For now and as long as only superusers are permitted to\n create rules or views due to the pg_rewrite permissions this\n should be ok. But we should change it soon by adding the\n flag to pg_class as I said earlier and have only the view\n owner and superusers permitted to change this flag.\n\n This flag is useful anyway. We change the world now that ALL\n views are overriding the acl checks. If we have this flag we\n can setup views that behave like before (user cannot see\n anything through the view he cannot see without) and use the\n overriding only in cases we need it (like for the pg_user\n permissions problem). I'll work on it.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\ndiff -r -c /usr/local/src/pgsql.orig/src/backend/executor/execMain.c ./src/backend/executor/execMain.c\n*** /usr/local/src/pgsql.orig/src/backend/executor/execMain.c\tFri Feb 13 10:17:15 1998\n--- ./src/backend/executor/execMain.c\tFri Feb 20 04:43:39 1998\n***************\n*** 299,304 ****\n--- 299,315 ----\n \t{\n \t\tRangeTblEntry *rte = lfirst(lp);\n \n+ \t\tif (rte->skipAcl)\n+ \t\t{\n+ \t\t\t/*\n+ \t\t\t * This happens if the access to this table is due\n+ \t\t\t * to a view query rewriting - the rewrite handler\n+ \t\t\t * checked the permissions against the view owner,\n+ \t\t\t * so we just skip this entry.\n+ \t\t\t */\n+ \t\t\tcontinue;\n+ \t\t}\n+ \n \t\trelid = rte->relid;\n \t\thtp = SearchSysCacheTuple(RELOID,\n \t\t\t\t\t\t\t\t ObjectIdGetDatum(relid),\ndiff -r -c /usr/local/src/pgsql.orig/src/backend/nodes/copyfuncs.c ./src/backend/nodes/copyfuncs.c\n*** /usr/local/src/pgsql.orig/src/backend/nodes/copyfuncs.c\tFri Feb 13 10:17:30 1998\n--- ./src/backend/nodes/copyfuncs.c\tFri Feb 20 02:38:13 1998\n***************\n*** 1495,1500 ****\n--- 1495,1501 ----\n \tnewnode->relid = from->relid;\n \tnewnode->inh = from->inh;\n \tnewnode->inFromCl = from->inFromCl;\n+ \tnewnode->skipAcl = from->skipAcl;\n \n \t\t\n \treturn newnode;\ndiff -r -c /usr/local/src/pgsql.orig/src/backend/rewrite/rewriteHandler.c ./src/backend/rewrite/rewriteHandler.c\n*** /usr/local/src/pgsql.orig/src/backend/rewrite/rewriteHandler.c\tWed Feb 4 09:33:56 1998\n--- ./src/backend/rewrite/rewriteHandler.c\tFri Feb 20 04:39:37 1998\n***************\n*** 10,15 ****\n--- 10,16 ----\n *\n *-------------------------------------------------------------------------\n */\n+ #include <string.h>\n #include \"postgres.h\"\n #include \"miscadmin.h\"\n #include \"utils/palloc.h\"\n***************\n*** 29,41 ****\n #include \"commands/creatinh.h\"\n #include \"access/heapam.h\"\n \n static void ApplyRetrieveRule(Query *parsetree, RewriteRule *rule,\n! \t\t\t\t int rt_index, int relation_level, int *modified);\n static List *fireRules(Query *parsetree, int rt_index, CmdType event,\n \t\t bool *instead_flag, List *locks, List **qual_products);\n static void QueryRewriteSubLink(Node *node);\n static List\t *QueryRewriteOne(Query *parsetree);\n static List *deepRewriteQuery(Query *parsetree);\n \n /*\n * gatherRewriteMeta -\n--- 30,48 ----\n #include \"commands/creatinh.h\"\n #include \"access/heapam.h\"\n \n+ #include \"utils/syscache.h\"\n+ #include \"utils/acl.h\"\n+ #include \"catalog/pg_user.h\"\n+ \n static void ApplyRetrieveRule(Query *parsetree, RewriteRule *rule,\n! \t\t\t\t int rt_index, int relation_level,\n! \t\t\t\t Relation relation, int *modified);\n static List *fireRules(Query *parsetree, int rt_index, CmdType event,\n \t\t bool *instead_flag, List *locks, List **qual_products);\n static void QueryRewriteSubLink(Node *node);\n static List\t *QueryRewriteOne(Query *parsetree);\n static List *deepRewriteQuery(Query *parsetree);\n+ static void CheckViewPerms(Relation view, List *rtable);\n \n /*\n * gatherRewriteMeta -\n***************\n*** 219,225 ****\n \t\t\t\t*instead_flag = TRUE;\n \t\t\t\treturn rule_lock->actions;\n \t\t\t}\n! \t\t\tApplyRetrieveRule(parsetree, rule_lock, rt_index, relation_level,\n \t\t\t\t\t\t\t &modified);\n \t\t\tif (modified)\n \t\t\t{\n--- 226,232 ----\n \t\t\t\t*instead_flag = TRUE;\n \t\t\t\treturn rule_lock->actions;\n \t\t\t}\n! \t\t\tApplyRetrieveRule(parsetree, rule_lock, rt_index, relation_level, relation,\n \t\t\t\t\t\t\t &modified);\n \t\t\tif (modified)\n \t\t\t{\n***************\n*** 247,252 ****\n--- 254,260 ----\n \t\t\t\t RewriteRule *rule,\n \t\t\t\t int rt_index,\n \t\t\t\t int relation_level,\n+ \t\t\t\t Relation relation,\n \t\t\t\t int *modified)\n {\n \tQuery\t *rule_action = NULL;\n***************\n*** 256,271 ****\n \tint\t\t\tnothing,\n \t\t\t\trt_length;\n \tint\t\t\tbadsql = FALSE;\n \n \trule_qual = rule->qual;\n \tif (rule->actions)\n \t{\n \t\tif (length(rule->actions) > 1)\t/* ??? because we don't handle\n! \t\t\t\t\t\t\t\t\t\t * rules with more than one\n! \t\t\t\t\t\t\t\t\t\t * action? -ay */\n \t\t\treturn;\n \t\trule_action = copyObject(lfirst(rule->actions));\n \t\tnothing = FALSE;\n \t}\n \telse\n \t{\n--- 264,304 ----\n \tint\t\t\tnothing,\n \t\t\t\trt_length;\n \tint\t\t\tbadsql = FALSE;\n+ \tint\t\t\tviewAclOverride = FALSE;\n \n \trule_qual = rule->qual;\n \tif (rule->actions)\n \t{\n \t\tif (length(rule->actions) > 1)\t/* ??? because we don't handle\n! \t\t\t\t\t\t * rules with more than one\n! \t\t\t\t\t\t * action? -ay */\n! \n! \t\t\t\t\t\t/* WARNING!!!\n! \t\t\t\t\t\t * If we sometimes handle\n! \t\t\t\t\t\t * rules with more than one\n! \t\t\t\t\t\t * action, the view acl checks\n! \t\t\t\t\t\t * might get broken. \n! \t\t\t\t\t\t * viewAclOverride should only\n! \t\t\t\t\t\t * become true (below) if this\n! \t\t\t\t\t\t * is a relation_level, instead,\n! \t\t\t\t\t\t * select query - Jan\n! \t\t\t\t\t\t */\n \t\t\treturn;\n \t\trule_action = copyObject(lfirst(rule->actions));\n \t\tnothing = FALSE;\n+ \n+ \t\t/*\n+ \t\t * If this rule is on the relation level, the rule action\n+ \t\t * is a select and the rule is instead then it must be\n+ \t\t * a view. Permissions for views now follow the owner of\n+ \t\t * the view, not the current user.\n+ \t\t */\n+ \t\tif (relation_level && rule_action->commandType == CMD_SELECT\n+ \t\t\t\t\t&& rule->isInstead)\n+ \t\t{\n+ \t\t CheckViewPerms(relation, rule_action->rtable);\n+ \t\t viewAclOverride = TRUE;\n+ \t\t}\n \t}\n \telse\n \t{\n***************\n*** 284,290 ****\n \t\trte->inFromCl = false;\n \t}\n \trt_length = length(rtable);\n! \trtable = nconc(rtable, copyObject(rule_action->rtable));\n \tparsetree->rtable = rtable;\n \n \trule_action->rtable = rtable;\n--- 317,346 ----\n \t\trte->inFromCl = false;\n \t}\n \trt_length = length(rtable);\n! \n! \tif (viewAclOverride)\n! \t{\n! \t\tList\t\t*rule_rtable, *rule_rt;\n! \t\tRangeTblEntry\t*rte;\n! \n! \t\trule_rtable = copyObject(rule_action->rtable);\n! \t\tforeach(rule_rt, rule_rtable)\n! \t\t{\n! \t\t\trte = lfirst(rule_rt);\n! \n! \t\t\t/*\n! \t\t\t * tell the executor that the ACL check on this\n! \t\t\t * range table entry is already done\n! \t\t\t */\n! \t\t\trte->skipAcl = true;\n! \t\t}\n! \n! \t\trtable = nconc(rtable, rule_rtable);\n! \t}\n! \telse\n! \t{\n! \t\trtable = nconc(rtable, copyObject(rule_action->rtable));\n! \t}\n \tparsetree->rtable = rtable;\n \n \trule_action->rtable = rtable;\n***************\n*** 750,752 ****\n--- 806,850 ----\n \n \treturn rewritten;\n }\n+ \n+ \n+ static void\n+ CheckViewPerms(Relation view, List *rtable)\n+ {\n+ \tHeapTuple\tutup;\n+ \tNameData\tuname;\n+ \tList\t\t*rt;\n+ \tRangeTblEntry\t*rte;\n+ \tint32\t\taclcheck_res;\n+ \n+ \t/*\n+ \t * get the usename of the view's owner\n+ \t */\n+ \tutup = SearchSysCacheTuple(USESYSID, view->rd_rel->relowner, 0, 0, 0);\n+ \tif (!HeapTupleIsValid(utup))\n+ \t{\n+ \t\telog(ERROR, \"cache lookup for userid %d failed\",\n+ \t\t\t\t\t\tview->rd_rel->relowner);\n+ \t}\n+ \tStrNCpy(uname.data,\n+ \t\t\t((Form_pg_user) GETSTRUCT(utup))->usename.data,\n+ \t\t\tNAMEDATALEN);\n+ \n+ \t/*\n+ \t * check that we have read access to all the\n+ \t * classes in the range table of the view\n+ \t */\n+ \tforeach(rt, rtable)\n+ \t{\n+ \t\trte = (RangeTblEntry *)lfirst(rt);\n+ \n+ \t\taclcheck_res = pg_aclcheck(rte->relname, uname.data, ACL_RD);\n+ \t\tif (aclcheck_res != ACLCHECK_OK)\n+ \t\t{\n+ \t\t\telog(ERROR, \"%s: %s\", rte->relname, aclcheck_error_strings[aclcheck_res]);\n+ \t\t}\n+ \t}\n+ }\n+ \n+ \n+ \ndiff -r -c /usr/local/src/pgsql.orig/src/include/nodes/parsenodes.h ./src/include/nodes/parsenodes.h\n*** /usr/local/src/pgsql.orig/src/include/nodes/parsenodes.h\tWed Feb 11 15:25:44 1998\n--- ./src/include/nodes/parsenodes.h\tFri Feb 20 02:21:00 1998\n***************\n*** 864,869 ****\n--- 864,870 ----\n \tOid\t\t\trelid;\n \tbool\t\tinh;\t\t\t/* inheritance? */\n \tbool\t\tinFromCl;\t\t/* comes from From Clause */\n+ \tbool\t\tskipAcl;\t\t/* skip ACL check in executor */\n } RangeTblEntry;\n \n /*\n",
"msg_date": "Fri, 20 Feb 1998 21:50:38 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Here it is - view permissions"
},
{
"msg_contents": "> This flag is useful anyway. We change the world now that ALL\n> views are overriding the acl checks. If we have this flag we\n> can setup views that behave like before (user cannot see\n> anything through the view he cannot see without) and use the\n> overriding only in cases we need it (like for the pg_user\n> permissions problem). I'll work on it.\n\nOK, but why would anyone want the old behavior?\n\nI guess if you have a table that is not select-able by everyone, and you\ncreate a view on it, the default permits will allow select to others. \nYou would have to set the permit on that view. Is there more to that\npg_class flag you want to add?\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 17:11:44 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "> For now and as long as only superusers are permitted to\n> create rules or views due to the pg_rewrite permissions this\n> should be ok. But we should change it soon by adding the\n> flag to pg_class as I said earlier and have only the view\n> owner and superusers permitted to change this flag.\n> \n> This flag is useful anyway. We change the world now that ALL\n> views are overriding the acl checks. If we have this flag we\n> can setup views that behave like before (user cannot see\n> anything through the view he cannot see without) and use the\n> overriding only in cases we need it (like for the pg_user\n> permissions problem). I'll work on it.\n\nThis patch is missing changes to the other node handling functions for\nrangetableentry. Do you want me to add them?\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Fri, 20 Feb 1998 22:19:08 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Here it is - view permissions"
},
{
"msg_contents": "Bruce wrote:\n>\n> > This flag is useful anyway. We change the world now that ALL\n> > views are overriding the acl checks. If we have this flag we\n> > can setup views that behave like before (user cannot see\n> > anything through the view he cannot see without) and use the\n> > overriding only in cases we need it (like for the pg_user\n> > permissions problem). I'll work on it.\n>\n> OK, but why would anyone want the old behavior?\n>\n> I guess if you have a table that is not select-able by everyone, and you\n> create a view on it, the default permits will allow select to others.\n> You would have to set the permit on that view. Is there more to that\n> pg_class flag you want to add?\n\n No, there isn't more on that. It's just to be more backward\n compatible. Users out there might have already views and\n since it wasn't neccessary to set explicit permissions on a\n view up to now (views inherited the permissions from all\n tables they select), it's unlikely that anyone out there set\n them up.\n\n I think it's better to implement something that's a key to\n open the door instead of opening it by default and telling\n where's the key to lock it back.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Sat, 21 Feb 1998 12:46:10 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "> \n> This patch is missing changes to the other node handling functions for\n> rangetableentry. Do you want me to add them?\n\n If you are so kind.\n\n It's only a few lines to copy and modify in\n\n nodes/outfuncs.c\n\tnodes/print.c\n\tnodes/readfuncs.c\n\n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n\n\nThanks, Jan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Sat, 21 Feb 1998 12:53:02 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Re: [PATCHES] Here it is - view permissions"
},
{
"msg_contents": "> \n> No, there isn't more on that. It's just to be more backward\n> compatible. Users out there might have already views and\n> since it wasn't neccessary to set explicit permissions on a\n> view up to now (views inherited the permissions from all\n> tables they select), it's unlikely that anyone out there set\n> them up.\n> \n> I think it's better to implement something that's a key to\n> open the door instead of opening it by default and telling\n> where's the key to lock it back.\n\nThis code is big enough without adding system table columns to allow\nbackward compatability. Just tell them as part of the upgrade, they\nneed to set GRANT/REVOKE on views. Put it in the /migration files too.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sat, 21 Feb 1998 11:49:55 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "> \n> > \n> > This patch is missing changes to the other node handling functions for\n> > rangetableentry. Do you want me to add them?\n> \n> If you are so kind.\n> \n> It's only a few lines to copy and modify in\n> \n> nodes/outfuncs.c\n> \tnodes/print.c\n> \tnodes/readfuncs.c\n\nDone. Actually, several of the fields were missing. I don't know how I\nmissed fixing them previously.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sat, 21 Feb 1998 13:26:54 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Here it is - view permissions"
},
{
"msg_contents": "\nBruce wrote:\n>\n> >\n> > No, there isn't more on that. It's just to be more backward\n> > compatible. Users out there might have already views and\n> > since it wasn't neccessary to set explicit permissions on a\n> > view up to now (views inherited the permissions from all\n> > tables they select), it's unlikely that anyone out there set\n> > them up.\n> >\n> > I think it's better to implement something that's a key to\n> > open the door instead of opening it by default and telling\n> > where's the key to lock it back.\n>\n> This code is big enough without adding system table columns to allow\n> backward compatability. Just tell them as part of the upgrade, they\n> need to set GRANT/REVOKE on views. Put it in the /migration files too.\n\n If that's OK for all of us let's add the information to\n /migration and the view permissions stuff is done then.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 23 Feb 1998 09:28:29 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> OK, but why would anyone want the old behavior?\n> \n> I guess if you have a table that is not select-able by everyone, and you\n> create a view on it, the default permits will allow select to others.\n> You would have to set the permit on that view. Is there more to that\n> pg_class flag you want to add?\n\nWhy does views default to 'select' permission for 'public'?\nI think most people will never think of the possibility that others\nwill be able to SELECT their data through views.\nShould not 'create view' at least print a NOTICE about this?\n\n/* m */\n",
"msg_date": "Mon, 23 Feb 1998 11:32:04 +0100",
"msg_from": "Mattias Kregert <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "> \n> Bruce Momjian wrote:\n> > \n> > OK, but why would anyone want the old behavior?\n> > \n> > I guess if you have a table that is not select-able by everyone, and you\n> > create a view on it, the default permits will allow select to others.\n> > You would have to set the permit on that view. Is there more to that\n> > pg_class flag you want to add?\n> \n> Why does views default to 'select' permission for 'public'?\n> I think most people will never think of the possibility that others\n> will be able to SELECT their data through views.\n> Should not 'create view' at least print a NOTICE about this?\n\n Because the current ACL_WORLD_DEFAULT in include/utils/acl.h\n is ACL_RD.\n\n Anything not revoked explicitly is granted select to public.\n Not only views. Think of it as a umask of 022.\n\n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Mon, 23 Feb 1998 13:14:39 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "On Mon, 23 Feb 1998, Mattias Kregert wrote:\n\n> Bruce Momjian wrote:\n> > \n> > OK, but why would anyone want the old behavior?\n> > \n> > I guess if you have a table that is not select-able by everyone, and you\n> > create a view on it, the default permits will allow select to others.\n> > You would have to set the permit on that view. Is there more to that\n> > pg_class flag you want to add?\n> \n> Why does views default to 'select' permission for 'public'?\n> I think most people will never think of the possibility that others\n> will be able to SELECT their data through views.\n> Should not 'create view' at least print a NOTICE about this?\n\n\tConsidering how much security we are putting around everything\nelse, is it unreasonably to have both 'create view'/'create table' default\nto 'revoke all' to public, and 'grant all' to owner?\n\n\n",
"msg_date": "Mon, 23 Feb 1998 08:06:33 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "> > This code is big enough without adding system table columns to allow\n> > backward compatability. Just tell them as part of the upgrade, they\n> > need to set GRANT/REVOKE on views. Put it in the /migration files too.\n> \n> If that's OK for all of us let's add the information to\n> /migration and the view permissions stuff is done then.\n\nOk, the following has been added to migration 6.2.1_6.3, and a note has\nbeen added to the item on the TODO/HISTORY list, though you can only see\nit now on the www page.\n\n\n---------------------------------------------------------------------------\n\nIn addition, 6.3 has separate permissions for views, rather than relying\non the permissions set on the underlying tables. For this reason, you will\nhave to set permissions on your views if you want anything but the\ndefault permissions.\n\n\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 23 Feb 1998 08:07:26 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "> \n> Bruce Momjian wrote:\n> > \n> > OK, but why would anyone want the old behavior?\n> > \n> > I guess if you have a table that is not select-able by everyone, and you\n> > create a view on it, the default permits will allow select to others.\n> > You would have to set the permit on that view. Is there more to that\n> > pg_class flag you want to add?\n> \n> Why does views default to 'select' permission for 'public'?\n> I think most people will never think of the possibility that others\n> will be able to SELECT their data through views.\n> Should not 'create view' at least print a NOTICE about this?\n\nAll tables are created with default permissions for SELECT to PUBLIC, so\nviews are no different.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 23 Feb 1998 08:10:59 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "> > Why does views default to 'select' permission for 'public'?\n> > I think most people will never think of the possibility that others\n> > will be able to SELECT their data through views.\n> > Should not 'create view' at least print a NOTICE about this?\n> \n> \tConsidering how much security we are putting around everything\n> else, is it unreasonably to have both 'create view'/'create table' default\n> to 'revoke all' to public, and 'grant all' to owner?\n\nMost commercial databases don't do this.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 23 Feb 1998 08:15:17 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions]"
},
{
"msg_contents": "> \n> On Mon, 23 Feb 1998, Mattias Kregert wrote:\n> \n> > Bruce Momjian wrote:\n> > > \n> > > OK, but why would anyone want the old behavior?\n> > > \n> > > I guess if you have a table that is not select-able by everyone, and you\n> > > create a view on it, the default permits will allow select to others.\n> > > You would have to set the permit on that view. Is there more to that\n> > > pg_class flag you want to add?\n> > \n> > Why does views default to 'select' permission for 'public'?\n> > I think most people will never think of the possibility that others\n> > will be able to SELECT their data through views.\n> > Should not 'create view' at least print a NOTICE about this?\n> \n> \tConsidering how much security we are putting around everything\n> else, is it unreasonably to have both 'create view'/'create table' default\n> to 'revoke all' to public, and 'grant all' to owner?\n\n include/utils/acl.h line 65\n\n set ACL_WORLD_DEFAULT to ACL_NO\n\n Then tables and views default to what you wanted.\n\n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Mon, 23 Feb 1998 14:27:12 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "On Mon, 23 Feb 1998, Bruce Momjian wrote:\n\n> > > Why does views default to 'select' permission for 'public'?\n> > > I think most people will never think of the possibility that others\n> > > will be able to SELECT their data through views.\n> > > Should not 'create view' at least print a NOTICE about this?\n> > \n> > \tConsidering how much security we are putting around everything\n> > else, is it unreasonably to have both 'create view'/'create table' default\n> > to 'revoke all' to public, and 'grant all' to owner?\n> \n> Most commercial databases don't do this.\n\n\tWell, just checked with Wayne (My Oracle Guru) and in Oracle,\neverything is private by default, and you open it up as required/desired\nto other ppl...\n\n\n",
"msg_date": "Mon, 23 Feb 1998 08:39:39 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions]"
},
{
"msg_contents": "> \n> On Mon, 23 Feb 1998, Bruce Momjian wrote:\n> \n> > > > Why does views default to 'select' permission for 'public'?\n> > > > I think most people will never think of the possibility that others\n> > > > will be able to SELECT their data through views.\n> > > > Should not 'create view' at least print a NOTICE about this?\n> > > \n> > > \tConsidering how much security we are putting around everything\n> > > else, is it unreasonably to have both 'create view'/'create table' default\n> > > to 'revoke all' to public, and 'grant all' to owner?\n> > \n> > Most commercial databases don't do this.\n> \n> \tWell, just checked with Wayne (My Oracle Guru) and in Oracle,\n> everything is private by default, and you open it up as required/desired\n> to other ppl...\n\n Microsoft SQL server too defaults to private and requires\n explicit GRANT for public.\n\n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 23 Feb 1998 15:12:41 +0100 (MET)",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions]"
},
{
"msg_contents": "On Mon, 23 Feb 1998, Jan Wieck wrote:\n\n> > \n> > On Mon, 23 Feb 1998, Mattias Kregert wrote:\n> > \n> > > Bruce Momjian wrote:\n> > > > \n> > > > OK, but why would anyone want the old behavior?\n> > > > \n> > > > I guess if you have a table that is not select-able by everyone, and you\n> > > > create a view on it, the default permits will allow select to others.\n> > > > You would have to set the permit on that view. Is there more to that\n> > > > pg_class flag you want to add?\n> > > \n> > > Why does views default to 'select' permission for 'public'?\n> > > I think most people will never think of the possibility that others\n> > > will be able to SELECT their data through views.\n> > > Should not 'create view' at least print a NOTICE about this?\n> > \n> > \tConsidering how much security we are putting around everything\n> > else, is it unreasonably to have both 'create view'/'create table' default\n> > to 'revoke all' to public, and 'grant all' to owner?\n> \n> include/utils/acl.h line 65\n> \n> set ACL_WORLD_DEFAULT to ACL_NO\n> \n> Then tables and views default to what you wanted.\n\n\tHave you actually tried this? :) Does it break anything?\n\n\n",
"msg_date": "Mon, 23 Feb 1998 09:33:09 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "> \n> On Mon, 23 Feb 1998, Jan Wieck wrote:\n> \n> > > \n> > > On Mon, 23 Feb 1998, Mattias Kregert wrote:\n> > > \n> > > > Bruce Momjian wrote:\n> > > > > \n> > > > > OK, but why would anyone want the old behavior?\n> > > > > \n> > > > > I guess if you have a table that is not select-able by everyone, and you\n> > > > > create a view on it, the default permits will allow select to others.\n> > > > > You would have to set the permit on that view. Is there more to that\n> > > > > pg_class flag you want to add?\n> > > > \n> > > > Why does views default to 'select' permission for 'public'?\n> > > > I think most people will never think of the possibility that others\n> > > > will be able to SELECT their data through views.\n> > > > Should not 'create view' at least print a NOTICE about this?\n> > > \n> > > \tConsidering how much security we are putting around everything\n> > > else, is it unreasonably to have both 'create view'/'create table' default\n> > > to 'revoke all' to public, and 'grant all' to owner?\n> > \n> > include/utils/acl.h line 65\n> > \n> > set ACL_WORLD_DEFAULT to ACL_NO\n> > \n> > Then tables and views default to what you wanted.\n> \n> \tHave you actually tried this? :) Does it break anything?\n> \n> \n> \n\n No I didn't - but if I read your smiley correct it does - right?\n\n I'm close to fixing the backend crashes on REVOKE ALL ON pg_user\n and so I didn't wanted to loose any minute and check if the above\n works properly.\n\n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Mon, 23 Feb 1998 15:42:30 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "On Mon, 23 Feb 1998, Jan Wieck wrote:\n\n> > > include/utils/acl.h line 65\n> > > \n> > > set ACL_WORLD_DEFAULT to ACL_NO\n> > > \n> > > Then tables and views default to what you wanted.\n> > \n> > \tHave you actually tried this? :) Does it break anything?\n> > \n> > \n> > \n> \n> No I didn't - but if I read your smiley correct it does - right?\n\n\tJust rebuilding on my Solaris/i386 machine with this\nenabled...will let you know after I'm done :)\n\n\n",
"msg_date": "Mon, 23 Feb 1998 09:43:51 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "On Mon, 23 Feb 1998, Jan Wieck wrote:\n\n> > > include/utils/acl.h line 65\n> > > \n> > > set ACL_WORLD_DEFAULT to ACL_NO\n> > > \n> > > Then tables and views default to what you wanted.\n> > \n> > \tHave you actually tried this? :) Does it break anything?\n> > \n> > \n> > \n> \n> No I didn't - but if I read your smiley correct it does - right?\n\n\tOnly too well :)\n\n\tIf I enable this, a \\d no longer works as anyone but superuser,\nwith the first error being:\n\nERROR: pg_class: Permission denied.\n\n\tSo it hits all the system tables too...\n\n\tWhich wouldn't be too bad, just a matter of doing a 'GRANT SELECT'\nfrom inside of initdb to get around this...\n\n\t...except, doing 'GRANT SELECT on pg_class to public' causes a\n'PQexec() -- Request was sent to backend, but...' error message, and a\n'Failed Assertion' on file inval.c...\n\n\tI'm not going to get a chance to investigate this during the day\ntoday, so if anyone else wants to look into it? If not, I'll look at it\nfrom home tonight and see if I can figure it out...\n\n\n",
"msg_date": "Mon, 23 Feb 1998 10:30:19 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "On Fri, 20 Feb 1998, Jan Wieck wrote:\n\n> First step done,\n> \n> below is the patch to have views to override the permission\n> checks for the accessed tables. Now we can do the following:\n> \n> CREATE VIEW db_user AS SELECT\n> usename,\n> usesysid,\n> usecreatedb,\n> usetrace,\n> usecatupd,\n> '**********'::text as passwd,\n> valuntil\n> FROM pg_user;\n\t\n\tI used the rewrite RULE here, instead of the above VIEW, with the\nVIEW being merely a 'SELECT * FROM pg_user'...by using the RULE (unless\nI'm mistaken?), if we change pg_user (ie. add another field at some later\ndate), the view itself shouldn't require a change...\n\n\n",
"msg_date": "Mon, 23 Feb 1998 23:18:39 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Here it is - view permissions"
},
{
"msg_contents": "> \n> On Fri, 20 Feb 1998, Jan Wieck wrote:\n> \n> > First step done,\n> > \n> > below is the patch to have views to override the permission\n> > checks for the accessed tables. Now we can do the following:\n> > \n> > CREATE VIEW db_user AS SELECT\n> > usename,\n> > usesysid,\n> > usecreatedb,\n> > usetrace,\n> > usecatupd,\n> > '**********'::text as passwd,\n> > valuntil\n> > FROM pg_user;\n> \t\n> \tI used the rewrite RULE here, instead of the above VIEW, with the\n> VIEW being merely a 'SELECT * FROM pg_user'...by using the RULE (unless\n> I'm mistaken?), if we change pg_user (ie. add another field at some later\n> date), the view itself shouldn't require a change...\n\nI thought we couldn't use rules because COPY will dump out the password,\nno?\n\n\n-- \nBruce Momjian | 830 Blythe Avenue\[email protected] | Drexel Hill, Pennsylvania 19026\n + If your life is a hard drive, | (610) 353-9879(w)\n + Christ can be your backup. | (610) 853-3000(h)\n",
"msg_date": "Mon, 23 Feb 1998 22:50:12 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PATCHES] Re: [HACKERS] Here it is - view permissions"
}
] |
[
{
"msg_contents": "At 4:06 PM 98.2.19 +0000, Thomas G. Lockhart wrote:\n>Here is my current list of supported platforms. The outstanding issues\n>are:\n\n>mklinux/ppc - still core dumping on regression tests? why??\n\nI finally found the source of the problem. function \"_readConst\" in \nbackend/nodes/readfuncs.c there is a line:\n\nlocal_node->constlen = strtoul(token,NULL,10);\n\n(before 6.3beta, this was:\nlocal_node->constlen = atol(token);\n)\n\nFor text type constant, token would be \"-1.\" In this case\nstrtoul() of mklinux/ppc returns 0 not ULONG_MAX (same bit\npattern as -1). This is a problem. \nSo quick workaround might be surrounding the line by #ifdef like:\n\n#ifdef PPC\nlocal_node->constlen = atol(token);\n#else\nlocal_node->constlen = strtoul(token,NULL,10);\n#endif\n\nP.S.\nSure current code works except MkLinux/ppc. However calling with\nstrtoul() with minus value is not well-mannered IMHO\n(strtoul() returns error code ERANGE in this case).\nTo make matters worse, there are some codes that comparing\nconstlen and -1.\nWhat about changing constlen to signed int and using\nstrtol() instead of strtoul() here?\nTatsuo Ishii\[email protected]\n\n",
"msg_date": "Sat, 21 Feb 1998 13:38:02 +0900",
"msg_from": "[email protected] (Tatsuo Ishii)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Platform status"
},
{
"msg_contents": "I will fix this. Thanks for the report. Must have been tough to find.\n\n\n\n> \n> At 4:06 PM 98.2.19 +0000, Thomas G. Lockhart wrote:\n> >Here is my current list of supported platforms. The outstanding issues\n> >are:\n> \n> >mklinux/ppc - still core dumping on regression tests? why??\n> \n> I finally found the source of the problem. function \"_readConst\" in \n> backend/nodes/readfuncs.c there is a line:\n> \n> local_node->constlen = strtoul(token,NULL,10);\n> \n> (before 6.3beta, this was:\n> local_node->constlen = atol(token);\n> )\n> \n> For text type constant, token would be \"-1.\" In this case\n> strtoul() of mklinux/ppc returns 0 not ULONG_MAX (same bit\n> pattern as -1). This is a problem. \n> So quick workaround might be surrounding the line by #ifdef like:\n> \n> #ifdef PPC\n> local_node->constlen = atol(token);\n> #else\n> local_node->constlen = strtoul(token,NULL,10);\n> #endif\n> \n> P.S.\n> Sure current code works except MkLinux/ppc. However calling with\n> strtoul() with minus value is not well-mannered IMHO\n> (strtoul() returns error code ERANGE in this case).\n> To make matters worse, there are some codes that comparing\n> constlen and -1.\n> What about changing constlen to signed int and using\n> strtol() instead of strtoul() here?\n> Tatsuo Ishii\n> [email protected]\n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sat, 21 Feb 1998 00:18:04 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Platform status"
}
] |
[
{
"msg_contents": "\tI'm using a recent (20Feb98) snapshot and get \n\tERROR: SearchSysCache: Called while cache disabled\n\tI get this when I do any queries.\n\tAny ideas?\nJulie\n\n-- \n[ Julia Anne Case ] [ Ships are safe inside the harbor, ]\n[Programmer at large] [ but is that what ships are really for. ] \n[ Admining Linux ] [ To thine own self be true. ]\n[ Windows/WindowsNT ] [ Fair is where you take your cows to be judged. ]\n",
"msg_date": "Sat, 21 Feb 1998 06:35:49 +0000",
"msg_from": "\"Julia A.Case\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Sparc/Linux oddity..."
},
{
"msg_contents": "In message <[email protected]>, \"Julia A.Case\" writes:\n> \tI'm using a recent (20Feb98) snapshot and get \n> \tERROR: SearchSysCache: Called while cache disabled\n> \tI get this when I do any queries.\n> \tAny ideas?\n> Julie\n> \n> -- \n> [ Julia Anne Case ] [ Ships are safe inside the harbor, ]\n> [Programmer at large] [ but is that what ships are really for. ] \n> [ Admining Linux ] [ To thine own self be true. ]\n> [ Windows/WindowsNT ] [ Fair is where you take your cows to be judged. ]\n> \n\nI just installed this yesterday. Aside from float8, geometry, and\nselect_views, all the regression tests were O.K.\n\nAs a data point, my Sparc/Linux machine is a sparc 20. I'm running a RedHat 4.1\nsystem. The kernel is 2.0.29, gcc version 2.7.2.3.\n\nTom Szybist\[email protected]\n",
"msg_date": "Sat, 21 Feb 1998 10:42:17 -0500",
"msg_from": "\"Thomas A. Szybist\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [PORTS] Sparc/Linux oddity... "
}
] |
[
{
"msg_contents": "\npts=> create table ttest ( id VARCHAR(10) NOT NULL );\nNOTICE: _outNode: don't know how to print type 715\nCREATE\npts=>\n\nIf I remove the 'NOT NULL', its created properly...this is with source\ncode as of 30minutes ago...and having wiped out my pgsql directory and\ndoing a gmake install from that...\n\nIt doesn't even seem to be restricted to VARCHAR():\n\npts=> create table test3 ( id text NOT NULL );\nNOTICE: _outNode: don't know how to print type 715\nCREATE\npts=> create index test3_i on test3 ( id );\nPQexec() -- Request was sent to backend, but backend closed the channel\nbefore responding.\n This probably means the backend terminated abnormally before or\nwhile processing the request.\npts=>\n\n\nDoing a little more searching, the core file shows as listed below. I'm\nstill investigating, but if anyone has an idea of what I might be missing,\nor should be looking at, please feel free to pop up :)\n\nScript started on Sat Feb 21 03:41:56 1998\n> gdb ~/pgsql/bin/postmaster postgres.core\nGDB is free software and you are welcome to distribute copies of it\n under certain conditions; type \"show copying\" to see the conditions.\nThere is absolutely no warranty for GDB; type \"show warranty\" for details.\nGDB 4.16 (i386-unknown-freebsd), \nCopyright 1996 Free Software Foundation, Inc...\nCore was generated by `postgres'.\nProgram terminated with signal 11, Segmentation fault.\nReading symbols from /usr/libexec/ld.so...done.\nReading symbols from /usr/lib/libcrypt.so.2.0...done.\nReading symbols from /usr/lib/libm.so.2.0...done.\nReading symbols from /usr/lib/libreadline.so.3.0...done.\nReading symbols from /usr/lib/libtermcap.so.2.1...done.\nReading symbols from /usr/lib/libcurses.so.2.0...done.\nReading symbols from /usr/lib/libc.so.3.1...done.\n#0 filepath (filename=0x0) at fd.c:473\n473\t\tif (*filename != SEP_CHAR)\n(gdb) where\n#0 filepath (filename=0x0) at fd.c:473\n#1 0xa6a15 in FileNameOpenFile (fileName=0x0, fileFlags=1538, fileMode=384)\n at fd.c:610\n#2 0x22920 in _bt_tapecreate (fname=0x0) at nbtsort.c:433\n#3 0x22c23 in _bt_spoolinit (index=0x1a5a10, ntapes=7, isunique=0)\n at nbtsort.c:584\n#4 0x1d9db in btbuild (heap=0x1a5710, index=0x1a5a10, natts=1, \n attnum=0x149510, istrat=0x1a6210, pcount=0, params=0x0, finfo=0x0, \n predInfo=0x149b10) at nbtree.c:164\n#5 0xdd2e5 in fmgr_c (finfo=0xefbfb39c, values=0xefbfb3ac, \n isNull=0xefbfb39b \"\") at fmgr.c:144\n#6 0xdd624 in fmgr (procedureId=338) at fmgr.c:285\n#7 0x2d3fa in index_build (heapRelation=0x1a5710, indexRelation=0x1a5a10, \n numberOfAttributes=1, attributeNumber=0x149510, parameterCount=0, \n parameter=0x0, funcInfo=0x0, predInfo=0x149b10) at index.c:1753\n#8 0x2ca7b in index_create (heapRelationName=0x149470 \"t\", \n indexRelationName=0x149450 \"t_idx\", funcInfo=0x0, attributeList=0x1494b0, \n accessMethodObjectId=403, numatts=1, attNums=0x149510, \n classObjectId=0x149530, parameterCount=0, parameter=0x0, predicate=0x0, \n islossy=0 '\\000', unique=0 '\\000') at index.c:1218\n#9 0x3628d in DefineIndex (heapRelationName=0x149470 \"t\", \n indexRelationName=0x149450 \"t_idx\", accessMethodName=0x84bd9 \"btree\", \n attributeList=0x1494b0, parameterList=0x0, unique=0 '\\000', predicate=0x0, \n rangetable=0x0) at defind.c:209\n#10 0xb47d6 in ProcessUtility (parsetree=0x14bd50, dest=Remote)\n at utility.c:396\n#11 0xb1f26 in pg_exec_query_dest (\n query_string=0xefbfb5f0 \"create index t_idx on t ( id );\", argv=0x0, \n typev=0x0, nargs=0, dest=Remote) at postgres.c:649\n#12 0xb1e40 in pg_exec_query (\n query_string=0xefbfb5f0 \"create index t_idx on t ( id );\", argv=0x0, \n typev=0x0, nargs=0) at postgres.c:597\n#13 0xb35e1 in PostgresMain (argc=9, argv=0xefbfd634) at postgres.c:1373\n#14 0x5a81a in main (argc=9, argv=0xefbfd634) at main.c:97\n(gdb) list\n468\t{\n469\t\tchar\t *buf;\n470\t\tint\t\t\tlen;\n471\t\n472\t\t/* Not an absolute path name? Then fill in with database path... */\n473\t\tif (*filename != SEP_CHAR)\n474\t\t{\n475\t\t\tlen = strlen(GetDatabasePath()) + strlen(filename) + 2;\n476\t\t\tbuf = (char *) palloc(len);\n477\t\t\tsprintf(buf, \"%s%c%s\", GetDatabasePath(), SEP_CHAR, filename);\n(gdb) quit\n> exit\n\nScript done on Sat Feb 21 03:42:36 1998\n\n\n",
"msg_date": "Sat, 21 Feb 1998 03:45:52 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Break down the problem..."
},
{
"msg_contents": "\nFurther to this...\n\nIf I follow this back, the problem appears to be in nbtsort.c, around line\n584, where:\n\n#3 0x22c23 in _bt_spoolinit (index=0x1a5a10, ntapes=7, isunique=0)\n at nbtsort.c:584\n584 btspool->bts_itape[i] =\n(gdb) list\n579 elog(ERROR, \"_bt_spoolinit: out of memory\");\n580 }\n581\n582 for (i = 0; i < ntapes; ++i)\n583 {\n584 btspool->bts_itape[i] =\n585 _bt_tapecreate(mktemp(strcpy(fname,\nTAPETEMP)));\n586 btspool->bts_otape[i] =\n587 _bt_tapecreate(mktemp(strcpy(fname,\nTAPETEMP)));\n5\n\nBasically, _bt_tapecreate() is being passed a temp name...this doesn't\nmake sense, does it? \n\nWhy are doing:\n\n btspool->bts_itape[i] =\n _bt_tapecreate(mktemp(strcpy(fname, TAPETEMP)));\n\nInstead of just doing:\n\n btspool->bts_itape[i] =\n _bt_tapecreate(mktemp(TAPETEMP));\n\nIs there a reason why we are copying TAPETEMP into fname first, and then\nfreeing it right after? \n\nRegardless of the above, GDB shows:\n\n\t(gdb) print fname\n\t$1 = 0x1a4ed0 \"pg_btsortX21438\"\n\nFor the above context, which is expected...\n\nBut, if I go back one, to inside teh _bt_tapecreate() function, GDB shows: \n\n#2 0x22920 in _bt_tapecreate (fname=0x0) at nbtsort.c:433\n433 tape->bttb_fd = FileNameOpenFile(fname, O_RDWR | O_CREAT |\nO_TRUNC, 0600);\n(gdb) print fname\n$2 = 0x0\n\nAm I missing something with GDB, or did fname lose its value here \nsomewhere? \n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sat, 21 Feb 1998 04:08:43 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Break down the problem..."
}
] |
[
{
"msg_contents": "Thanks\n\nMichael\n--\nDr. Michael Meskes, Projekt-Manager | topystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Use Debian GNU/Linux! | Tel: (+49) 2405/4670-44\n\n> ----------\n> From: \tThe Hermit Hacker[SMTP:[email protected]]\n> Sent: \tFreitag, 20. Februar 1998 14:03\n> To: \tMichael Meskes\n> Cc: \tBrook Milligan; PostgreSQL Hacker\n> Subject: \tRe: [HACKERS] libpgtcl undefined symbol error with\n> pgaccess-0.76\n> \n> On Fri, 20 Feb 1998, Michael Meskes wrote:\n> \n> > Brook Milligan writes:\n> > > Actually, the most recent version is\n> > > \n> > > \tftp://ftp.flex.ro/pub/pgaccess/pgaccess-0.76.tar.gz\n> > > \n> > > This, not the old version, really should ship with 6.3.\n> > \n> > Why isn't it in CVS?\n> \n> \tCause nobody had told me it was upgraded :( Done now...\n> \n> \n",
"msg_date": "Sat, 21 Feb 1998 16:27:16 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] libpgtcl undefined symbol error with pgaccess-0.76"
}
] |
[
{
"msg_contents": "I was afraid to hear something like that.\n\nI just tried starting posgres manually and enter the statement that\ncaused the seg fault manually but then it works!\n\nI will try to find some more time. \n\nMichael\n--\nDr. Michael Meskes, Projekt-Manager | topystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Use Debian GNU/Linux! | Tel: (+49) 2405/4670-44\n\n> ----------\n> From: \tThomas G. Lockhart[SMTP:[email protected]]\n> Sent: \tFreitag, 20. Februar 1998 14:26\n> To: \tThe Hermit Hacker\n> Cc: \tMeskes, Michael; [email protected];\n> [email protected]\n> Subject: \tRe: [HACKERS] RE: New ecgp code problem.\n> \n> > > Unfortunately createdb as user postgres doesn't work either, so I\n> have\n> > > to keep debugging without being able to run the binary.\n> >\n> > I'm curious here as to why nobody has looked into this\n> > problem...\n> \n> Because no one else sees the problem? I've forgotten what your symptom\n> actually is, but from the 980217 cvs tree:\n> \n> postgres=> \\q\n> golem$ createuser tgl\n> Enter user's postgres ID or RETURN to use unix user ID: 106 ->\n> Is user \"tgl\" allowed to create databases (y/n) y\n> Is user \"tgl\" allowed to add users? (y/n) y\n> createuser: tgl was successfully added\n> ...\n> golem> createdb\n> golem> psql\n> Welcome to the POSTGRESQL interactive sql monitor:\n> ...\n> You are currently connected to the database: tgl\n> \n> tgl=>\n> \n> Have you done a complete \"make clean; make install; initdb\" cycle??\n> \n> - Tom\n> \n",
"msg_date": "Sat, 21 Feb 1998 16:27:19 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] RE: New ecgp code problem."
}
] |
[
{
"msg_contents": "Let's say you have a parent relations as (parent, child). Now you want\nto compute all ancestors from this relation that is technically spoken\nyou need a transitive closure.\n\nIn logic this looks like:\n\nancestor(X,Y) :- parent(X,Y).\nancestor(X,Y) :- ancestor(X,Z), parent(Z,Y).\n\nX is ancestor of Y if X is parent of Y or X is ancestor of Z who is\nparent of Y.\n\nIn POSTQUEL this is written like this (note I'm not sure about correct\nPOSTQUEL syntax):\n\nretrieve into ancestor(p.x,p.y) from p in parent\n\nretrieve* into ancestor(a.x,p.y) from a in ancestor, p in parent where\na.y=p.x.\n\nYou see the second line generates new tuples every time it's executed\nuntil there is no more data deductable.\n\nMichael\n--\nDr. Michael Meskes, Projekt-Manager | topystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Use Debian GNU/Linux! | Tel: (+49) 2405/4670-44\n\n> ----------\n> From: \tThomas G. Lockhart[SMTP:[email protected]]\n> Sent: \tFreitag, 20. Februar 1998 17:44\n> To: \tMeskes, Michael\n> Cc: \t'Vadim B. Mikheev'; PostgreSQL Hacker\n> Subject: \tRe: [HACKERS] Recursive queries?\n> \n> > But such a statement is executed only once, isn't it?\n> >\n> > Postquel did execute a retrieve* into as long as it was able to\n> generate\n> > new data.\n> \n> OK, I'll bite :)\n> Can you give an example of this? How does new data become available in\n> the\n> middle of a transaction? Just curious...\n> \n",
"msg_date": "Sat, 21 Feb 1998 16:27:26 +0100",
"msg_from": "\"Meskes, Michael\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Recursive queries?"
}
] |
[
{
"msg_contents": "\nFound it:\n\nChange, in backend/access/nbtree/nbtsort.c:\n\n#define TAPETEMP \"pg_btsortXXXXXX\"\n\nto:\n\n#define TAPETEMP \"pg_btsortXXXXXXX\"\n\nFor some reason, under FreeBSD, it appears that the mktemp() value needs\nthe extra 'X' to improve/ensure uniqueness\n\n\n\nMarc G. Fournier \nSystems Administrator @ hub.org \nprimary: [email protected] secondary: scrappy@{freebsd|postgresql}.org \n\n",
"msg_date": "Sat, 21 Feb 1998 15:23:45 -0400 (AST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": true,
"msg_subject": "FreeBSD/mktemp bug with create index...fixed"
}
] |
[
{
"msg_contents": "\nunsubscribe\n\n",
"msg_date": "Sat, 21 Feb 1998 16:57:30 -0500",
"msg_from": "[email protected] (Joao Leao)",
"msg_from_op": true,
"msg_subject": "[HACKERS] Unsubscribe"
}
] |
[
{
"msg_contents": "Vadim B. Mikheev <[email protected]> wrote:\n\n> create table a (a int, a1 char(8));\n> create table b (b int);\n> insert into a values (1, 'one');\n> insert into a values (NULL, 'null');\n\n> insert into b values (1);\n> insert into b values (NULL);\n\n> select * from a where a in (select * from b);\n> -- 1 row with a == 1 expected\n> select * from a where a not in (select * from b);\n> -- 0 row expected\n> select * from a where not a in (select * from b);\n> -- 0 row in Oracle & Informix, 1 row in Pg (with a == NULL), SyBase ???\n\nShould be 0 row(s) expected. Here's why: three-value logic.\n\na in (select...)\na == any (select...)\n\n\t(a=1) == (b==1) is 'true'\n\t(a=1) == (b==NULL) is 'unknown'\n\t\t'true' OR 'unknown' is 'true'.\n\n\t(a=NULL) == (b==1) is 'unknown'\n\t(a=NULL) == (b==NULL) is 'unknown'\n\t\t'unknown' OR 'unknown' is 'unknown'\n\t\tnot ('unknown') is 'unknown'\n\t\tshouldn't be in 'where not a in...' query\n\nHope this helps.\n\n--\nYurik\n\n",
"msg_date": "21 Feb 1998 22:31:48 GMT",
"msg_from": "[email protected] (Yurik V. Nazaroff)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Subselects and NOTs"
}
] |
[
{
"msg_contents": "Somebody mentioned they had a Tcl implementation of crypt() the other\nday but I've deleted the message. Could you send me a copy?\n\nThanks,\nPhil\n",
"msg_date": "Sat, 21 Feb 1998 23:26:52 +0000",
"msg_from": "Phil Thompson <[email protected]>",
"msg_from_op": true,
"msg_subject": "Tcl Implementation of crypt()"
},
{
"msg_contents": "Thus spake Phil Thompson\n> Somebody mentioned they had a Tcl implementation of crypt() the other\n> day but I've deleted the message. Could you send me a copy?\n\nWould this help? Just add includes and an init function to register it.\nI also have a Python function if anyone needs it.\n\n\nstatic const char *ach[] = {\n \"abcdefghijkamnbpqrstuvwxyzABCDEFGHcJKLMNdPQRSTUVWXYZ123456789eAB\",\n \"abcdefghjkamnbpqrstuvwxyzabcdefghcjkmndpqrstuvwxyzxyz23456789eab\",\n \"ABCDEFGHJKAMNBPQRSTUVWXYZABCDEFGHCJKMNDPQRSTUVWXYZXYZ23456789EAB\",\n}; \n\n/* \n usage:\n te_crypt <password>\n \n Encrypts a password\n \n Returns:\n The password in its encrypted form.\n*/\n \nint \nte_crypt(ClientData cData, Tcl_Interp *interp, int argc, char **argv)\n{ \n char salt[8];\n static int initflag = 1;\n \n if (argc != 2)\n {\n Tcl_AppendResult(interp, \"crypt: Invalid number of arguments\", 0);\n return TCL_ERROR;\n }\n \n if (initflag)\n srandom(time(NULL)); \n \n initflag = 0;\n salt[0] = ach[0][random() % 64];\n salt[1] = ach[0][random() % 64];\n salt[2] = 0;\n\n Tcl_AppendResult(interp, crypt(argv[1], salt), 0);\n return TCL_OK;\n}\n\n\n-- \nD'Arcy J.M. Cain <darcy@{druid|vex}.net> | Democracy is three wolves\nhttp://www.druid.net/darcy/ | and a sheep voting on\n+1 416 424 2871 (DoD#0082) (eNTP) | what's for dinner.\n",
"msg_date": "Sun, 22 Feb 1998 09:18:49 -0500 (EST)",
"msg_from": "[email protected] (D'Arcy J.M. Cain)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Tcl Implementation of crypt()"
}
] |
[
{
"msg_contents": "subscribe\n\n",
"msg_date": "Sat, 21 Feb 1998 19:10:00 -0500",
"msg_from": "Oliver Giller <[email protected]>",
"msg_from_op": true,
"msg_subject": "subscribe"
}
] |
[
{
"msg_contents": "Here they are. Please tell me if I can remove any of them. Unfixed\nitems get added to the TODO list after the final 6.3 release.\n\n\n---------------------------------------------------------------------------\n\nORDER BY NULLs problem?\ntest new view permission code, install new pg_user view for use with \\d\noptimizer memory exhaustion with many OR's\nmax tuple size patch(Darren)\nIs GROUP BY duplicates fixed?\nDo we have a self-join optimizer performance problem?\nGet interfaces fixed for new protocol\nProfiling improvements?\nDo we have a problem with GROUP BY without ORDER BY?\nProblem with tables >2Gb\nDo we want to add timestamps to elog messages?\nScanKeyData missing initializations\nCan we improve vacuum locking issues?\nAlpha/64-bit issues, mkoidname() problem\n'Can not write block blind' error on createdb and regression collision\nViews on aggregates fail\nlarge objects memory exhaustion\nsubselect issues, NOT IN (empty query)\nCommit sgml document sources(Thomas)\nCommit html documents(Thomas)\nCommit postscript documents(Thomas)\nFix isinf.c irix port problem reported by Andrew(Marc)\nCheck triggers regression test(Vadim)\nCheck select_views regression test\nTest ecpg new struct features(Michael)\nTest big-endian/little-endian operation(Tatsuo?)?\nREVOKE ALL ON pg_user FROM PUBLIC crashes if run many times\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sat, 21 Feb 1998 21:17:32 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Open 6.3 issues"
},
{
"msg_contents": "On Sat, 21 Feb 1998, Bruce Momjian wrote:\n\n> Here they are. Please tell me if I can remove any of them. Unfixed\n> items get added to the TODO list after the final 6.3 release.\n\n> Get interfaces fixed for new protocol\n\nJDBC is now using the new protocol.\n\n-- \nPeter T Mount [email protected] or [email protected]\nMain Homepage: http://www.demon.co.uk/finder\nWork Homepage: http://www.maidstone.gov.uk Work EMail: [email protected]\n\n",
"msg_date": "Sun, 22 Feb 1998 09:42:25 +0000 (GMT)",
"msg_from": "Peter T Mount <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Open 6.3 issues"
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> ORDER BY NULLs problem?\n\nHadn't time to look, yet...\n\n> optimizer memory exhaustion with many OR's\n\nI reported that it can't be fixed for 6.3 - please, move it into\nTODO now. Nevertheless, I could apply patch to prepqual.c to\nfree memory whenever it's possible...\n\n> Is GROUP BY duplicates fixed?\n\nI got two reports that current code is OK - remove it.\n\n> Do we have a self-join optimizer performance problem?\n\nYes. Not sure that I'll have time...\n\n> Do we have a problem with GROUP BY without ORDER BY?\n ^^^^^^^\nCould you remind what the problem is ?\n\n> Problem with tables >2Gb\n\nI still suggest to add elog(ERROR) to mdextend()...\n\n> Can we improve vacuum locking issues?\n\nI would leave vacuum as is. Improving of vacuum was on my 6.3 TODO\nlist - it's 6.4 TODO now :)\n\n> 'Can not write block blind' error on createdb and regression collision\n\nI'm fixing buffer manager now...\n\n> subselect issues, NOT IN (empty query)\n\nClosed: NOT IN (empty) and all others Op ALL (empty) now\nreturns TRUE.\n\n> Check triggers regression test(Vadim)\n\nFixed already.\n\n> Check select_views regression test\n\nResults can't be identical on all platforms => change second query with\n\nSELECT name, #thepath FROM iexit ORDER BY 1, 2;\n\nVadim\n",
"msg_date": "Sun, 22 Feb 1998 19:18:23 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Open 6.3 issues"
}
] |
[
{
"msg_contents": "-----Original Message-----\nFrom: Vadim B. Mikheev <[email protected]>\nTo: Maurice Gittens <[email protected]>\nCc: [email protected] <[email protected]>\nDate: zondag 22 februari 1998 17:47\nSubject: Re: [HACKERS] How To free resources used by large object Relations?\n\n\n>>\n>> Somehow I have to free the relation from the cache in the following\n>> situations:\n>> 1. In a transaction I must free the stuff when the transaction is\n>> commited/aborted.\n>\n>Backend does it, don't worry.\nI don't really understand all of the code so please bear with me.\nCould it be that large objects don't use the right memorycontext/portals so\nthat memory isn't freed automagically?\n>\n>> 2. Otherwise it must happen when lo_close is called.\n>\n>It seems that you can't remove relation from cache untill\n>commit/abort, currently: backend uses local cache to unlink\n>files of relations created in transaction if abort...\n>We could change relcache.c:RelationPurgeLocalRelation()\n>to read from pg_class directly...\nIs there a way to to tell the cache manager to free resources?\nThe relations concerned are know how to properly free them is not\nhowever.\n>\n>But how many LO do you create in single xact ?\nOnly one (in my real application).\n>Is memory allocated for cache so big ?\nNot really except that the leak accumulates as long as the connection\nwith the backend is not closed.\n>\n>Vadim\n\nI have a simple test program which goes like this:\n\n(this is C-like psuedo code)\n\nmain()\n{\n connection = createConnection();\n\n for(;;)\n {\n lo_create(connection,READ|WRITE);\n }\n\n destroyConnection(connection);\n}\n\nThis program will leak memory each time it goes through the for loop.\nIt doesn't matter if the statements in the for loop are in a transaction or\nnot.\n\nWhen I give each large object it's own memory context (so that memory\nis freed per large object) it seems to leak memory more slowly, but it leaks\nanyway.\n\nI've tried calling a number of the functions (like\nRelationPurgeLocalRelation)\nin relcache.c to try to free up the memory myself but the backend doesn't\nlike\nthis (== closed connection).\n\nIt looks like there is some assumption about which memorycontext/portal is\nused during transactions and that largeobjects don't obey this\nassumption.\n\nCan you make these assumptions explicite? Maybe I can then let\nlarge object respect these rules.\n\nNow I have the following understanding of these matters:\n1. In transactions\nAll memory should be freed automatically at commit/abort.\nHow do I tell the system to do it for me?\n\n2. In autocommit mode\nAll resources used by large object should be freed at lo_close.\nCan I have this delayed and done automatically in the CommitTransaction\nfunction?\n\n3. Somehow atomic functions like lo_create should not leak memory either.\nThis is the case however.\n\nThanks for any help,\nMaurice\n\n",
"msg_date": "Sun, 22 Feb 1998 12:51:55 +0100",
"msg_from": "\"Maurice Gittens\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] How To free resources used by large object Relations?"
},
{
"msg_contents": "I'm sorry, Maurice, but I haven't time to deal with this now.\nLet's 6.3 be out first...\n\nVadim\n",
"msg_date": "Tue, 24 Feb 1998 10:01:57 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] How To free resources used by large object Relations?"
}
] |
[
{
"msg_contents": "Hi All,\n\nI'm getting a SIGSEGV with the following subselect:-\n\nselect * from disks where diskid in \n ( select distinct diskid from tracks where artist = 'Led Zeppelin'); \n \nThe source tree is a cvsup from 22nd Feb (AM GMT)\nPlatform is Linux/SPARC\n \nI'm afraid my debugging skills are limited to generating a backtrace \nso here goes....\n\nKeith.\n\nProgram received signal SIGSEGV, Segmentation fault.\n0xe0104ee4 in strlen ()\n(gdb) bt\n#0 0xe0104ee4 in strlen ()\n#1 0xd1264 in pstrdup (string=0x0) at palloc.c:115\n#2 0x696c0 in _copyUnique (from=0x1b0b90) at copyfuncs.c:552\n#3 0x6abdc in copyObject (from=0x1b0b90) at copyfuncs.c:1687\n#4 0x69754 in _copySubPlan (from=0x1a83d0) at copyfuncs.c:594\n#5 0x6abfc in copyObject (from=0x1a83d0) at copyfuncs.c:1693\n#6 0x698b0 in _copyExpr (from=0x1b1f90) at copyfuncs.c:679\n#7 0x6ac2c in copyObject (from=0x1b1f90) at copyfuncs.c:1706\n#8 0x6a304 in _copyCInfo (from=0x1b1710) at copyfuncs.c:1334\n#9 0x6ad5c in copyObject (from=0x1b1710) at copyfuncs.c:1767\n#10 0x6ae54 in copyObject (from=0x1b2d10) at copyfuncs.c:1829\n#11 0x7e8b8 in create_seqscan_path (rel=0x1b0a90) at pathnode.c:228\n#12 0x737c4 in find_rel_paths (root=0x15d390, rels=0x1b2f70) at allpaths.c:118\n#13 0x73778 in find_paths (root=0x15d390, rels=0x1b2f70) at allpaths.c:72\n#14 0x78ae4 in subplanner (root=0x15d390, flat_tlist=0x1ae330, qual=0x1ae3f0) at \nplanmain.c:271\n#15 0x789fc in query_planner (root=0x15d390, command_type=1, tlist=0x1aa910, \nqual=0x1ae3f0) at planmain.c:181\n#16 0x79040 in union_planner (parse=0x15d390) at planner.c:138\n#17 0x78f2c in planner (parse=0x15d390) at planner.c:72\n#18 0xa80f4 in pg_parse_and_plan (\n query_string=0xefffdc38 \"select * from disks where diskid in ( select \ndistinct diskid from tracks where artist = 'Led Zeppelin');\", typev=0x0, \nnargs=0, queryListP=0xefffdb64, dest=Remote) at postgres.c:534\n#19 0xa81e0 in pg_exec_query_dest (\n query_string=0xefffdc38 \"select * from disks where diskid in ( select \ndistinct diskid from tracks where artist = 'Led Zeppelin');\", argv=0x0, \ntypev=0x0, nargs=0, dest=Remote) at postgres.c:615\n#20 0xa81bc in pg_exec_query (\n query_string=0xefffdc38 \"select * from disks where diskid in ( select \ndistinct diskid from tracks where artist = 'Led Zeppelin');\", argv=0x0, \ntypev=0x0, nargs=0) at postgres.c:597\n#21 0xa90c4 in PostgresMain (argc=1122304, argv=0xeffffc38) at postgres.c:1373\n#22 0x68788 in main (argc=11, argv=0xeffffd04) at main.c:97\n\n",
"msg_date": "Sun, 22 Feb 1998 14:50:28 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "SIGSEGV in sebselect."
},
{
"msg_contents": "This is a bug. I can reproduct it here with:\n\n\tselect * from pguser where usesysid in (select distinct usesysid\nfrom pg_user);\n\nThe Unique node copy has a NULL uniqueAttr value.\n\nVadim is this related to subselect distinct handling, or is there some\nlarger bug here?\n\n> \n> Hi All,\n> \n> I'm getting a SIGSEGV with the following subselect:-\n> \n> select * from disks where diskid in \n> ( select distinct diskid from tracks where artist = 'Led Zeppelin'); \n> \n> The source tree is a cvsup from 22nd Feb (AM GMT)\n> Platform is Linux/SPARC\n> \n> I'm afraid my debugging skills are limited to generating a backtrace \n> so here goes....\n> \n> Keith.\n> \n> Program received signal SIGSEGV, Segmentation fault.\n> 0xe0104ee4 in strlen ()\n> (gdb) bt\n> #0 0xe0104ee4 in strlen ()\n> #1 0xd1264 in pstrdup (string=0x0) at palloc.c:115\n> #2 0x696c0 in _copyUnique (from=0x1b0b90) at copyfuncs.c:552\n> #3 0x6abdc in copyObject (from=0x1b0b90) at copyfuncs.c:1687\n> #4 0x69754 in _copySubPlan (from=0x1a83d0) at copyfuncs.c:594\n> #5 0x6abfc in copyObject (from=0x1a83d0) at copyfuncs.c:1693\n> #6 0x698b0 in _copyExpr (from=0x1b1f90) at copyfuncs.c:679\n> #7 0x6ac2c in copyObject (from=0x1b1f90) at copyfuncs.c:1706\n> #8 0x6a304 in _copyCInfo (from=0x1b1710) at copyfuncs.c:1334\n> #9 0x6ad5c in copyObject (from=0x1b1710) at copyfuncs.c:1767\n> #10 0x6ae54 in copyObject (from=0x1b2d10) at copyfuncs.c:1829\n> #11 0x7e8b8 in create_seqscan_path (rel=0x1b0a90) at pathnode.c:228\n> #12 0x737c4 in find_rel_paths (root=0x15d390, rels=0x1b2f70) at allpaths.c:118\n> #13 0x73778 in find_paths (root=0x15d390, rels=0x1b2f70) at allpaths.c:72\n> #14 0x78ae4 in subplanner (root=0x15d390, flat_tlist=0x1ae330, qual=0x1ae3f0) at \n> planmain.c:271\n> #15 0x789fc in query_planner (root=0x15d390, command_type=1, tlist=0x1aa910, \n> qual=0x1ae3f0) at planmain.c:181\n> #16 0x79040 in union_planner (parse=0x15d390) at planner.c:138\n> #17 0x78f2c in planner (parse=0x15d390) at planner.c:72\n> #18 0xa80f4 in pg_parse_and_plan (\n> query_string=0xefffdc38 \"select * from disks where diskid in ( select \n> distinct diskid from tracks where artist = 'Led Zeppelin');\", typev=0x0, \n> nargs=0, queryListP=0xefffdb64, dest=Remote) at postgres.c:534\n> #19 0xa81e0 in pg_exec_query_dest (\n> query_string=0xefffdc38 \"select * from disks where diskid in ( select \n> distinct diskid from tracks where artist = 'Led Zeppelin');\", argv=0x0, \n> typev=0x0, nargs=0, dest=Remote) at postgres.c:615\n> #20 0xa81bc in pg_exec_query (\n> query_string=0xefffdc38 \"select * from disks where diskid in ( select \n> distinct diskid from tracks where artist = 'Led Zeppelin');\", argv=0x0, \n> typev=0x0, nargs=0) at postgres.c:597\n> #21 0xa90c4 in PostgresMain (argc=1122304, argv=0xeffffc38) at postgres.c:1373\n> #22 0x68788 in main (argc=11, argv=0xeffffd04) at main.c:97\n> \n> \n> \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sun, 22 Feb 1998 15:09:10 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] SIGSEGV in sebselect."
},
{
"msg_contents": "> \n> Hi All,\n> \n> I'm getting a SIGSEGV with the following subselect:-\n> \n> select * from disks where diskid in \n> ( select distinct diskid from tracks where artist = 'Led Zeppelin'); \n> \n> The source tree is a cvsup from 22nd Feb (AM GMT)\n> Platform is Linux/SPARC\n> \n> I'm afraid my debugging skills are limited to generating a backtrace \n> so here goes....\n\nThis is a partial fix. In make_unique(), if Query->uniqueAttr is '*',\nUnique->uniqueAttr gets null. However, with this fix, I get:\n\n\ttest=> select usesysid from pg_user where usesysid in (select distinct\n\tusesysid from pg_user);\n\tERROR: ExecReScan: node type 24 not supported\n\nI have not installed this patch yet until Vadim comments and perhaps has\nan idea on the ExecReScan message:\n\n\n---------------------------------------------------------------------------\n\n*** ./backend/nodes/copyfuncs.c.orig\tSun Feb 22 18:58:59 1998\n--- ./backend/nodes/copyfuncs.c\tSun Feb 22 19:03:37 1998\n***************\n*** 549,555 ****\n \t *\tcopy remainder of node\n \t * ----------------\n \t */\n! \tnewnode->uniqueAttr = pstrdup(from->uniqueAttr);\n \tnewnode->uniqueAttrNum = from->uniqueAttrNum;\n \tNode_Copy(from, newnode, uniquestate);\n \n--- 549,558 ----\n \t *\tcopy remainder of node\n \t * ----------------\n \t */\n! \tif (newnode->uniqueAttr)\n! \t\tnewnode->uniqueAttr = pstrdup(from->uniqueAttr);\n! \telse\n! \t\tnewnode->uniqueAttr = NULL;\n \tnewnode->uniqueAttrNum = from->uniqueAttrNum;\n \tNode_Copy(from, newnode, uniquestate);\n \n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sun, 22 Feb 1998 19:17:09 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] SIGSEGV in sebselect."
},
{
"msg_contents": "Bruce Momjian wrote:\n> \n> This is a partial fix. In make_unique(), if Query->uniqueAttr is '*',\n> Unique->uniqueAttr gets null. However, with this fix, I get:\n> \n> test=> select usesysid from pg_user where usesysid in (select distinct\n> usesysid from pg_user);\n> ERROR: ExecReScan: node type 24 not supported\n> \n> I have not installed this patch yet until Vadim comments and perhaps has\n> an idea on the ExecReScan message:\n\nPatch is Ok. DISTINCT is yet another node without re-scan support.\nHope to add it in 1-2 days.\n\nVadim\n",
"msg_date": "Mon, 23 Feb 1998 09:14:11 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] SIGSEGV in sebselect."
},
{
"msg_contents": "> \n> Bruce Momjian wrote:\n> > \n> > This is a partial fix. In make_unique(), if Query->uniqueAttr is '*',\n> > Unique->uniqueAttr gets null. However, with this fix, I get:\n> > \n> > test=> select usesysid from pg_user where usesysid in (select distinct\n> > usesysid from pg_user);\n> > ERROR: ExecReScan: node type 24 not supported\n> > \n> > I have not installed this patch yet until Vadim comments and perhaps has\n> > an idea on the ExecReScan message:\n> \n> Patch is Ok. DISTINCT is yet another node without re-scan support.\n> Hope to add it in 1-2 days.\n\nI have applied my patch.\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sun, 22 Feb 1998 21:52:18 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] SIGSEGV in sebselect."
}
] |
[
{
"msg_contents": "With Vadim's change, here it is:\n\n---------------------------------------------------------------------------\n\nORDER BY NULLs problem?\ntest new view permission code, install new pg_user view for use with \\d\nmax tuple size patch(Darren)\nSelf-join optimizer performance problem\nGet interfaces fixed for new protocol\nProfiling improvements?\nDo we have a problem with GROUP BY without ORDER BY?\nProblem with tables >2Gb\nDo we want to add timestamps to elog messages?\nScanKeyData missing initializations\nAlpha/64-bit issues, mkoidname() problem\n'Can not write block blind' error on createdb and regression collision(Vadim)\nViews on aggregates fail\nlarge objects memory exhaustion\nCommit sgml document sources(Thomas)\nCommit html documents(Thomas)\nCommit postscript documents(Thomas)\nFix isinf.c irix port problem reported by Andrew(Marc)\nCheck select_views regression test\nTest ecpg new struct features(Michael)\nTest big-endian/little-endian operation(Tatsuo?)?\nREVOKE ALL ON pg_user FROM PUBLIC crashes if run many times\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Sun, 22 Feb 1998 15:06:51 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Current 6.3 issues"
},
{
"msg_contents": ">Test big-endian/little-endian operation(Tatsuo?)?\n\nI have tested following platforms: Solairs2.6/Sparc, FreeBSD 2.2.1,\nLinux/x86, mklinux/ppc. For the testing, I used authentication method\n\"trust.\" Testing tool was psql and the tested command was \"\\dT.\" I\nhave comfirmed any combination of above servers/clients works fine.\n\nBTW I have found some problems with Feb 20 snapshot. Could you add\nfollowings to 6.3 issues?\n\n1. SunOS 4.1.4:\n\n(1) in include/config.h:\n\n#ifndef HAVE_INET_ATON\n# ifdef HAVE_ARPA_INET_H\n# ifdef HAVE_NETINET_IN_H\n# include <sys/types.h> <----- necessary\n# include <netinet/in.h>\n# endif\n# include <arpa/inet.h>\n# endif\n\n(2) 6.3 doesn't use -Dsunos4 anymore. However we have such codes as:\n\n#if defined(sunos4)\n#define memmove(d, s, l)\t\tbcopy(s, d, l)\n#include <unistd.h>\n#endif\n\nthis leads to compiling errors in various places. Should we use\n#if defined(sparc) && !defined(__svr4)\ninstead?\n\n2. Solaris 2.6:\n\nSeems almost ok except a minor error in ecpg:\n\ngmake[3]: Entering directory `/export/home/local/src/pgsql/pgsql/src/interfaces/ecpg/preproc'\ninstall -c -m 755 ecpg /usr/local/pgsql/bin\ngmake[3]: install: Command not found\ngmake[3]: *** [install] Error 127\n\n3. mklinux/ppc:\nWith Bruce's patches now regression tests are ok. Thanks Bruce!\n\nHowever ecpg seems to have small problem:\n\nmake[3]: Entering directory `/usr/local/src/pgsql/6.3/pgsql/src/interfaces/ecpg/preproc'\n/usr/bin/bison -y -d preproc.y\n\"preproc.y\", line 235: type redeclaration for S_SIGNED\nmake[3]: *** [y.tab.c] Error 1\n\n3. Linux/x86:\n\nconfigure doesn't use linux-elf for the default template even it is an ELF\nsystem. (I have tested on RedHat4.0)\n\nchanging line 105 in configure.in as follows would solve the problem.\n\n\t\tGUESS=`grep \"^$host_no_ver=\" template/.similar 2>/dev/null`\n--\nTatsuo Ishii\[email protected]\n",
"msg_date": "Mon, 23 Feb 1998 12:28:48 +0900",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Current 6.3 issues "
},
{
"msg_contents": "> Check select_views regression test\n\nSecond query is replaced -> there shouldn't be differences any more...\n\n> 'Can not write block blind' error on createdb and regression collision(Vadim)\n\nNew bug was introduced in mdblindwrt(): backend tried to use current\ndatabase path when writing data for another data base => path was invalid\n=> elog(ERROR). I fixed this: now backend uses GetRawDatabaseInfo()\nto get real path for given data base, BUT I don't like this way:\nGetRawDatabaseInfo() reads pg_database to get path and this is\nreally bad for performance!\n\nKeeping in mind implementation of TABLESPACEs in _near_ future\nwe have to either use symlinks to give backend simple and fast way\nto constract full path to a relation or store full path in\nbuffer header. Shared memory is critical resource but nevertheless\nI prefer the second way.\n\nNow about hanging another backends after 'blind' error. This is\nvery old bug in buffer manager: before doing blind write backend locks\nbuffer for IO using local spinlock of this buffer, but elog(ERROR)\ndoesn't releases spinlocks of this type! I tried to fix this too\nbut decided that there is not enough time to do this now: there are\nanother \"bad places\" in bufmgr. Re-visiting of bufmgr was in my\nplans after 6.0, but ... hope to do this for 6.4.\n\nVadim\n",
"msg_date": "Mon, 23 Feb 1998 21:04:55 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Current 6.3 issues"
},
{
"msg_contents": "On Mon, Feb 23, 1998 at 12:28:48PM +0900, [email protected] wrote:\n\n> (2) 6.3 doesn't use -Dsunos4 anymore. However we have such codes as:\n> \n> #if defined(sunos4)\n> #define memmove(d, s, l)\t\tbcopy(s, d, l)\n> #include <unistd.h>\n> #endif\n> \n> this leads to compiling errors in various places. Should we use\n> #if defined(sparc) && !defined(__svr4)\n> instead?\n\nThe above test does not necessarily mean that the system runs SunOS 4!\nIt would be true on the machine I'm sending this from as well, an SS2\nrunning NetBSD/sparc...\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n",
"msg_date": "Mon, 23 Feb 1998 21:47:34 +0100",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Current 6.3 issues"
},
{
"msg_contents": ">> (2) 6.3 doesn't use -Dsunos4 anymore. However we have such codes as:\n>> \n>> #if defined(sunos4)\n>> #define memmove(d, s, l)\t\tbcopy(s, d, l)\n>> #include <unistd.h>\n>> #endif\n>> \n>> this leads to compiling errors in various places. Should we use\n>> #if defined(sparc) && !defined(__svr4)\n>> instead?\n>\n>The above test does not necessarily mean that the system runs SunOS 4!\n>It would be true on the machine I'm sending this from as well, an SS2\n>running NetBSD/sparc...\n\nGood point. What are predefined #define directives for NetBSD/sparc?\nFor SunOS we have -Dsparc -Dsun -Dunix -D__GCC_NEW_VARARGS__ ...\n--\nTatsuo Ishii\[email protected]\n",
"msg_date": "Tue, 24 Feb 1998 09:57:10 +0900",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Current 6.3 issues "
},
{
"msg_contents": "On Tue, Feb 24, 1998 at 09:57:10AM +0900, [email protected] wrote:\n> Good point. What are predefined #define directives for NetBSD/sparc?\n> For SunOS we have -Dsparc -Dsun -Dunix -D__GCC_NEW_VARARGS__ ...\n\nbarsoom% tools/ccsym \n__GNUC__=2\n__GNUC_MINOR__=7\nunix\nsparc\n__NetBSD__\n__KPRINTF_ATTRIBUTE__\n__unix__\n__sparc__\n__NetBSD__\n__KPRINTF_ATTRIBUTE__\n__unix\n__sparc\nsystem=unix\nsystem=NetBSD\ncpu=sparc\nmachine=sparc\n\n-tih\n-- \nPopularity is the hallmark of mediocrity. --Niles Crane, \"Frasier\"\n",
"msg_date": "Tue, 24 Feb 1998 06:44:28 +0100",
"msg_from": "Tom I Helbekkmo <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Current 6.3 issues"
},
{
"msg_contents": "[email protected] writes:\n> 2. Solaris 2.6:\n> \n> Seems almost ok except a minor error in ecpg:\n> \n> gmake[3]: Entering directory `/export/home/local/src/pgsql/pgsql/src/interfaces/ecpg/preproc'\n> install -c -m 755 ecpg /usr/local/pgsql/bin\n> gmake[3]: install: Command not found\n> gmake[3]: *** [install] Error 127\n> \n> 3. mklinux/ppc:\n> With Bruce's patches now regression tests are ok. Thanks Bruce!\n> \n> However ecpg seems to have small problem:\n> \n> make[3]: Entering directory `/usr/local/src/pgsql/6.3/pgsql/src/interfaces/ecpg/preproc'\n> /usr/bin/bison -y -d preproc.y\n> \"preproc.y\", line 235: type redeclaration for S_SIGNED\n> make[3]: *** [y.tab.c] Error 1\n\nBoth will be solved with my next patch. Thanks.\n\nMichael\n\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Tue, 24 Feb 1998 11:55:04 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Current 6.3 issues"
},
{
"msg_contents": "I tested Feb 24 snapshot on SunOS and encountered a problem. any\nsuggestion?\n\nflex -t pgc.l > pgc.c\ngcc -I../../../include -I../include -Wall -DMAJOR_VERSION=1 -DMINOR_VERSION=0 -DPATCHLEVEL=0 -c pgc.c -o pgc.o\nIn file included from pgc.l:6:\nextern.h:14: warning: parameter names (without types) in function declaration\nextern.h:14: parse error before `size_t'\npgc.l: In function `yylex':\npgc.l:49: warning: implicit declaration of function `fprintf'\npgc.l:144: warning: implicit declaration of function `fwrite'\npgc.l:108: warning: label `find_rule' defined but not used\npgc.l: In function `yy_get_next_buffer':\npgc.l:386: warning: implicit declaration of function `_filbuf'\npgc.l:387: warning: implicit declaration of function `fread'\npgc.l: At top level:\npgc.l:1005: warning: `yy_flex_realloc' defined but not used\npgc.l:488: warning: `yyunput' defined but not used\ngmake[3]: *** [pgc.o] Error 1\ngmake[3]: Leaving directory `/auto/srafsb/export5/prj/psarch/src/6.3/pgsql/src/interfaces/ecpg/preproc'\ngmake[2]: *** [all] Error 2\ngmake[2]: Leaving directory `/auto/srafsb/export5/prj/psarch/src/6.3/pgsql/src/interfaces/ecpg'\ngmake[1]: *** [all] Error 2\ngmake[1]: Leaving directory `/auto/srafsb/export5/prj/psarch/src/6.3/pgsql/src/interfaces'\ngmake: *** [all] Error 2\n",
"msg_date": "Wed, 25 Feb 1998 10:49:31 +0900",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Current 6.3 issues "
}
] |
[
{
"msg_contents": "Hi,\n\nYes, it's definitely related to the DISTINCT clause.\n\nIf I drop the distinct, which isn't needed anyway, the query works fine.\n\nKeith.\n\n\nBruce Momjian <[email protected]>\n> \n> This is a bug. I can reproduct it here with:\n> \n> \tselect * from pguser where usesysid in (select distinct usesysid\n> from pg_user);\n> \n> The Unique node copy has a NULL uniqueAttr value.\n> \n> Vadim is this related to subselect distinct handling, or is there some\n> larger bug here?\n> \n> > \n> > Hi All,\n> > \n> > I'm getting a SIGSEGV with the following subselect:-\n> > \n> > select * from disks where diskid in \n> > ( select distinct diskid from tracks where artist = 'Led Zeppelin'); \n> > \n> > The source tree is a cvsup from 22nd Feb (AM GMT)\n> > Platform is Linux/SPARC\n> > \n> > I'm afraid my debugging skills are limited to generating a backtrace \n> > so here goes....\n> > \n> > Keith.\n> > \n> > Program received signal SIGSEGV, Segmentation fault.\n> > 0xe0104ee4 in strlen ()\n> > (gdb) bt\n> > #0 0xe0104ee4 in strlen ()\n> > #1 0xd1264 in pstrdup (string=0x0) at palloc.c:115\n> > #2 0x696c0 in _copyUnique (from=0x1b0b90) at copyfuncs.c:552\n> > #3 0x6abdc in copyObject (from=0x1b0b90) at copyfuncs.c:1687\n> > #4 0x69754 in _copySubPlan (from=0x1a83d0) at copyfuncs.c:594\n> > #5 0x6abfc in copyObject (from=0x1a83d0) at copyfuncs.c:1693\n> > #6 0x698b0 in _copyExpr (from=0x1b1f90) at copyfuncs.c:679\n> > #7 0x6ac2c in copyObject (from=0x1b1f90) at copyfuncs.c:1706\n> > #8 0x6a304 in _copyCInfo (from=0x1b1710) at copyfuncs.c:1334\n> > #9 0x6ad5c in copyObject (from=0x1b1710) at copyfuncs.c:1767\n> > #10 0x6ae54 in copyObject (from=0x1b2d10) at copyfuncs.c:1829\n> > #11 0x7e8b8 in create_seqscan_path (rel=0x1b0a90) at pathnode.c:228\n> > #12 0x737c4 in find_rel_paths (root=0x15d390, rels=0x1b2f70) at \nallpaths.c:118\n> > #13 0x73778 in find_paths (root=0x15d390, rels=0x1b2f70) at allpaths.c:72\n> > #14 0x78ae4 in subplanner (root=0x15d390, flat_tlist=0x1ae330, \nqual=0x1ae3f0) at \n> > planmain.c:271\n> > #15 0x789fc in query_planner (root=0x15d390, command_type=1, tlist=0x1aa910, \n> > qual=0x1ae3f0) at planmain.c:181\n> > #16 0x79040 in union_planner (parse=0x15d390) at planner.c:138\n> > #17 0x78f2c in planner (parse=0x15d390) at planner.c:72\n> > #18 0xa80f4 in pg_parse_and_plan (\n> > query_string=0xefffdc38 \"select * from disks where diskid in ( select \n> > distinct diskid from tracks where artist = 'Led Zeppelin');\", typev=0x0, \n> > nargs=0, queryListP=0xefffdb64, dest=Remote) at postgres.c:534\n> > #19 0xa81e0 in pg_exec_query_dest (\n> > query_string=0xefffdc38 \"select * from disks where diskid in ( select \n> > distinct diskid from tracks where artist = 'Led Zeppelin');\", argv=0x0, \n> > typev=0x0, nargs=0, dest=Remote) at postgres.c:615\n> > #20 0xa81bc in pg_exec_query (\n> > query_string=0xefffdc38 \"select * from disks where diskid in ( select \n> > distinct diskid from tracks where artist = 'Led Zeppelin');\", argv=0x0, \n> > typev=0x0, nargs=0) at postgres.c:597\n> > #21 0xa90c4 in PostgresMain (argc=1122304, argv=0xeffffc38) at \npostgres.c:1373\n> > #22 0x68788 in main (argc=11, argv=0xeffffd04) at main.c:97\n> > \n> > \n> > \n> \n> \n> -- \n> Bruce Momjian\n> [email protected]\n> \n\n",
"msg_date": "Sun, 22 Feb 1998 22:22:32 +0000 (GMT)",
"msg_from": "Keith Parks <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] SIGSEGV in sebselect."
}
] |
[
{
"msg_contents": ">Date: Sat, 21 Feb 1998 21:17:32 -0500 (EST)\n>From: Bruce Momjian <[email protected]>\n>Subject: Open 6.3 issues\n>\n>Here they are. Please tell me if I can remove any of them. Unfixed\n>items get added to the TODO list after the final 6.3 release.\n>\n>\n>- ---------------------------------------------------------------------------\n>\n>ORDER BY NULLs problem?\n\nI'd say this has been solved by my patch of late December.\n\n[snip]\n\nCheers,\n\nJeroen van Vianen\n",
"msg_date": "Mon, 23 Feb 1998 09:12:15 +0100",
"msg_from": "Jeroen van Vianen <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: Open issues - ORDER BY NULLs problem"
}
] |
[
{
"msg_contents": "> In POSTQUEL this is written like this (note I'm not sure about correct\n> POSTQUEL syntax):\n> \n> retrieve into ancestor(p.x,p.y) from p in parent\n> \n> retrieve* into ancestor(a.x,p.y) from a in ancestor, p in parent where\n> a.y=p.x.\n> \n> You see the second line generates new tuples every time it's executed\n> until there is no more data deductable.\n\nThis sounds really interesting, I have been puzzeling on how to resolve the\nCNAME points\nto another CNAME ... which finally points to an IP address for our name\nresolution\nadmin database. This would be it !\n\nAndreas\n",
"msg_date": "Mon, 23 Feb 1998 09:43:59 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "RE: [HACKERS] Recursive queries?"
}
] |
[
{
"msg_contents": "Jan, I think you describe the correct picture of what is needed for\nPL/pgSQL.\n\nMy comments:\n> No actual development - just have something in mind how I\n> would implement it. I'll get into details after 6.3 release.\n> PL/pgSQL will have at least the following capabilities:\n> \n> - local variable\n\t\tlocal to the procedure (in a per call context)\n\tI think it will also need:\n - global variable\n\t\tin a session context (standard mentions these also)\n> - local records\n> - access to the database over SPI\n> - control structures (if/else/while/loop)\n> - elog messages\n> - triggers can modify new tuple\n> - triggers can skip operation\n> \n\n> If perl doesn't have such a restricted interpreter facility,\n> then perl might never become a TRUSTED procedural language\n> like Tcl is.\n\nThere is taintperl.\nI don't see anything that speaks against 2 variants of pl/perl. One trusted,\nusing taintperl\nand one untrusted opening the internals to superusers, like in C.\nBTW.: How do you write an input or output function in pl/tcl if all you get\nis the external representation ?\n\nAndreas\n\nP.S.: there is no need to email me directly, unless you need something very\nfast,\n\tsince I do read all pgsql-hackers mail in the digest. Thanx all.\n",
"msg_date": "Mon, 23 Feb 1998 12:11:52 +0100",
"msg_from": "Zeugswetter Andreas SARZ <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: pl/{perl,pgsql} (was Re: AW: [HACKERS] triggers, views and ru\n\tles (not instead))"
},
{
"msg_contents": "> \n> Jan, I think you describe the correct picture of what is needed for\n> PL/pgSQL.\n> \n> My comments:\n> > No actual development - just have something in mind how I\n> > would implement it. I'll get into details after 6.3 release.\n> > PL/pgSQL will have at least the following capabilities:\n> > \n> > - local variable\n> \t\tlocal to the procedure (in a per call context)\n\n Of course.\n\n> \tI think it will also need:\n> - global variable\n> \t\tin a session context (standard mentions these also)\n\n Might be good either. Thanks.\n\n> > - local records\n> > - access to the database over SPI\n> > - control structures (if/else/while/loop)\n> > - elog messages\n> > - triggers can modify new tuple\n> > - triggers can skip operation\n> > \n> \n> > If perl doesn't have such a restricted interpreter facility,\n> > then perl might never become a TRUSTED procedural language\n> > like Tcl is.\n> \n> There is taintperl.\n> I don't see anything that speaks against 2 variants of pl/perl. One trusted,\n> using taintperl\n> and one untrusted opening the internals to superusers, like in C.\n\n PL/Tcl could be enhanced to do this too. Will make it in the\n future.\n\n> BTW.: How do you write an input or output function in pl/tcl if all you get\n> is the external representation ?\n\n Impossible. \n\n> \n> Andreas\n> \n> P.S.: there is no need to email me directly, unless you need something very\n> fast,\n> \tsince I do read all pgsql-hackers mail in the digest. Thanx all.\n> \n> \n\n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Mon, 23 Feb 1998 13:24:25 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: pl/{perl,pgsql} (was Re: AW: [HACKERS] triggers, views and ru"
},
{
"msg_contents": "\nA plus for PL/perl -- you can write input/output functions with\npack/unpack.. \n\nOn Mon, 23 February 1998, at 13:24:25, Jan Wieck wrote:\n\n> > BTW.: How do you write an input or output function in pl/tcl if all you get\n> > is the external representation ?\n> \n> Impossible. \n> \n> > \n> > Andreas\n> > \n> > P.S.: there is no need to email me directly, unless you need something very\n> > fast,\n> > \tsince I do read all pgsql-hackers mail in the digest. Thanx all.\n> > \n> > \n",
"msg_date": "Mon, 23 Feb 1998 05:00:34 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: pl/{perl,pgsql} (was Re: AW: [HACKERS] triggers, views and ru"
},
{
"msg_contents": "Jan Wieck wrote:\n> > BTW.: How do you write an input or output function in pl/tcl if all you get\n> > is the external representation ?\n> \n> Impossible. \n\nWe could have the backend check file permissions, open the file and\npass the file pointer to the pl/tcl procedure. This sounds a bit\ncomplicated I know. It seems to me that we would want at least a\n\"stdout\" available to the procedure so that it can \"talk to\" the user.\nIf we provide stdin, stdout and stderr, then we could use redirection\nto pipe input to and from a procedure.\n\nOcie Mitchell\n\n",
"msg_date": "Mon, 23 Feb 1998 12:15:11 -0800 (PST)",
"msg_from": "[email protected]",
"msg_from_op": false,
"msg_subject": "Re: pl/{perl,pgsql} (was Re: AW: [HACKERS] triggers, views and ru"
},
{
"msg_contents": "\nOcie Mitchell wrote:\n>\n> Jan Wieck wrote:\n> > > BTW.: How do you write an input or output function in pl/tcl if all you get\n> > > is the external representation ?\n> >\n> > Impossible.\n>\n> We could have the backend check file permissions, open the file and\n> pass the file pointer to the pl/tcl procedure. This sounds a bit\n> complicated I know. It seems to me that we would want at least a\n> \"stdout\" available to the procedure so that it can \"talk to\" the user.\n> If we provide stdin, stdout and stderr, then we could use redirection\n> to pipe input to and from a procedure.\n\n That's not the point. With input/output functions, the\n postgres data types input/output registered procedures where\n ment. That are the functions, that convert one Datum from\n it's external string representation into and back from it's\n internal binary value used in the database tuples.\n\n Direct file access from inside a backends function is IMHO\n never a good thing. This is the job of an application\n program, because the actually accessed file might be on an\n NFS filesystem, this can hang and the database backend will\n hang inside of a transaction where maybe even the DBA or root\n cannot kill them (process is blocked inside a system call).\n\n This is IMHO a NONO.\n\n\nJan\n\n--\n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n\n",
"msg_date": "Mon, 23 Feb 1998 21:49:48 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: pl/{perl,pgsql} (was Re: AW: [HACKERS] triggers, views and ru"
},
{
"msg_contents": "\nI must chime in and agree strongly, but I also have a query: do the\npointers for HeapStruct (and related form struct) ever have to be\nre-initialized using the cache lookup functions?\n\nOn Mon, 23 February 1998, at 21:49:48, Jan Wieck wrote:\n\n> Direct file access from inside a backends function is IMHO\n> never a good thing. This is the job of an application\n> program, because the actually accessed file might be on an\n> NFS filesystem, this can hang and the database backend will\n> hang inside of a transaction where maybe even the DBA or root\n> cannot kill them (process is blocked inside a system call).\n> \n> This is IMHO a NONO.\n",
"msg_date": "Mon, 23 Feb 1998 13:22:30 -0800",
"msg_from": "Brett McCormick <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: pl/{perl,pgsql} (was Re: AW: [HACKERS] triggers, views and ru"
}
] |
[
{
"msg_contents": "> \n> max tuple size patch(Darren)\n> \n\nCan move this to the TODO list. I'd like to get the\nvariable block size stuff done first, then this would\nmake much more sense.\n\ndarrenk\n",
"msg_date": "Mon, 23 Feb 1998 08:08:11 -0500",
"msg_from": "[email protected] (Darren King)",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Current 6.3 issues"
}
] |
[
{
"msg_contents": "> \n> On Fri, 20 Feb 1998, Michael Meskes wrote:\n> \n> > Brook Milligan writes:\n> > > Actually, the most recent version is\n> > > \n> > > \tftp://ftp.flex.ro/pub/pgaccess/pgaccess-0.76.tar.gz\n> > > \n> > > This, not the old version, really should ship with 6.3.\n> > \n> > Why isn't it in CVS?\n> \n> \tCause nobody had told me it was upgraded :( Done now...\n> \nMight I suggest that any external support software like pgaccess\nbe mirrored on ftp.postgresql.org. It then becomes easy to ensure \nthat the latest versions get into a PGSQL release and all such\ns/w is available from ftp.postgresql.org\n\n\nBest wishes,\n\nAndrew\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Mon, 23 Feb 1998 13:09:43 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] libpgtcl undefined symbol error with pgaccess-0.76"
}
] |
[
{
"msg_contents": "Here is the newest list:\n\n---------------------------------------------------------------------------\n\ntest new view permission code, install new pg_user view for use with \\d\nmax tuple size patch(Darren)\nSelf-join optimizer performance problem\nGet interfaces fixed for new protocol\nProfiling improvements?\nProblem with tables >2Gb, add elog()?\nDo we want to add timestamps to elog messages?\nScanKeyData missing initializations\nAlpha/64-bit issues, mkoidname() problem\n'Can not write block blind' error on createdb and regression collision(Vadim)\nViews on aggregates fail\nlarge objects memory exhaustion\nCommit sgml document sources(Thomas)\nCommit html documents(Thomas)\nCommit postscript documents(Thomas)\nFix isinf.c irix port problem reported by Andrew(Marc)\nCheck select_views regression test\nREVOKE ALL ON pg_user FROM PUBLIC crashes if run many times, cache issue(Jan)\n\n\n-- \nBruce Momjian\[email protected]\n",
"msg_date": "Mon, 23 Feb 1998 08:39:08 -0500 (EST)",
"msg_from": "Bruce Momjian <[email protected]>",
"msg_from_op": true,
"msg_subject": "Open 6.3 issues"
},
{
"msg_contents": "> \n> Here is the newest list:\n> \n> ---------------------------------------------------------------------------\n> \n> Views on aggregates fail\n\n Uh oh - Aggregates require a group by clause in the select.\n This is correctly copied into the query action in the views\n pg_rewrite entry. But the rewrite handler doesn't insert it\n into the parsetree. Thus a wrong execution plan in built.\n\n I'll try some things, but expect to get into trouble if the\n select from the view contains it's own grouping clauses.\n\n Let's see.\n\n> REVOKE ALL ON pg_user FROM PUBLIC crashes if run many times, cache issue(Jan)\n\n Done - sent to Marc - not cvs committed yet!\n\n\nJan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Mon, 23 Feb 1998 18:08:14 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Open 6.3 issues"
},
{
"msg_contents": "On Mon, 23 Feb 1998, Jan Wieck wrote:\n\n> > REVOKE ALL ON pg_user FROM PUBLIC crashes if run many times, cache issue(Jan)\n> \n> Done - sent to Marc - not cvs committed yet!\n\n\tIs now...\n\n\n",
"msg_date": "Mon, 23 Feb 1998 12:43:41 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Open 6.3 issues"
},
{
"msg_contents": "> \n> On Mon, 23 Feb 1998, Jan Wieck wrote:\n> \n> > > REVOKE ALL ON pg_user FROM PUBLIC crashes if run many times, cache issue(Jan)\n> > \n> > Done - sent to Marc - not cvs committed yet!\n> \n> \tIs now...\n\n\tArrrg\n\n\tPlease add the line \n\n\t\tif (!bootstrap)\n\n\tabove the \n\t\t\n\t\tInitLocalInvalidateData();\n\n\tin postinit.c manually. Calling it in bootstrap mode causes\n\tpostgres to dump core. With that test everything is fine.\n\n\nThanks, Jan\n\n-- \n\n#======================================================================#\n# It's easier to get forgiveness for being wrong than for being right. #\n# Let's break this rule - forgive me. #\n#======================================== [email protected] (Jan Wieck) #\n\n",
"msg_date": "Mon, 23 Feb 1998 19:16:44 +0100 (MET)",
"msg_from": "[email protected] (Jan Wieck)",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Open 6.3 issues"
},
{
"msg_contents": "Jan Wieck wrote:\n> \n> > Views on aggregates fail\n> \n> Uh oh - Aggregates require a group by clause in the select.\n> This is correctly copied into the query action in the views\n> pg_rewrite entry. But the rewrite handler doesn't insert it\n> into the parsetree. Thus a wrong execution plan in built.\n> \n> I'll try some things, but expect to get into trouble if the\n> select from the view contains it's own grouping clauses.\n\nThere is another way of VIEW implementation: put VIEW' query\nas subselect into FROM clause. The last feature is not implemented\nyet. It's just a thought...\n\nVadim\n",
"msg_date": "Tue, 24 Feb 1998 08:19:52 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Open 6.3 issues"
}
] |
[
{
"msg_contents": "> I have been working with many of you for over a hear, and would like to\n> know more details about you.\n> \n\nI am 34 and have been working with Postgres since the first release of\nPostgres95. My contributions to the project have been fairly small\n(the Linux and Irix-specific FAQs and a few small patches). I live\nwith my partner and our nearly-3 year old daughter (see\nhttp://www.biochem.ucl.ac.uk/~martin/other/) in Ashtead, Surrey which\nis about 35min on the train from London. Like almost everyone else\nhere, I have 2 cats (Bruce, I think this should be mentioned in\nthe developers' FAQ :-)\n\nI work in the Biochemistry Department at University College London on\nanalysis and prediction of protein structures. I uses databases to\nstore and analyse data on antibody sequence and structure, structural\nclassification of protein domains, mutation data in p53 (a protein\nwhich is mutated in nearly 50% of human cancers) and in analysing\nthe results of a protein modelling competition.\n\n\nBest wishes,\n\nAndrew\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Mon, 23 Feb 1998 13:39:12 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Who is everyone?"
}
] |
[
{
"msg_contents": "\n",
"msg_date": "Mon, 23 Feb 1998 20:53:32 +0700",
"msg_from": "\"Vadim B. Mikheev\" <[email protected]>",
"msg_from_op": true,
"msg_subject": "ExecReScan implemented for Sort & Unique (DISTINCT) nodes (empty)."
}
] |
[
{
"msg_contents": "\nOK, here are my experiences trying to install the latest snapshot (23Feb98)\nunder Irix 5.3.\n\nIn summary, installation is currently BROKEN!!!!!\n\nI shall wait till installation works properly before I try another regression\ntest :-)\n\n\nAndrew\n\n\n\nInstallation of Snapshot 23Feb98 under Irix 5.3\n===============================================\n\n1. .../src/template/.similar: Should say irix5 not irix\n-------------------------------------------------------\n\n*** .similar.old Mon Feb 23 14:18:37 1998\n--- .similar Mon Feb 23 14:13:49 1998\n***************\n*** 13,19 ****\n i586-pc-linux-gnu=linux-elf\n i586-pc-sco3.2v=sco\n i686-pc-linux-gnu=linux-elf\n! mips-sgi-irix=irix\n powerpc-ibm-aix3.2.5=aix_325\n powerpc-ibm-aix4.1.4.0=aix_41\n powerpc-ibm-aix4.1.5.0=aix_41\n--- 13,19 ----\n i586-pc-linux-gnu=linux-elf\n i586-pc-sco3.2v=sco\n i686-pc-linux-gnu=linux-elf\n! mips-sgi-irix=irix5\n powerpc-ibm-aix3.2.5=aix_325\n powerpc-ibm-aix4.1.4.0=aix_41\n powerpc-ibm-aix4.1.5.0=aix_41\n\n\n\n2. configure dies (!)\n---------------------\nRunning configure, dies with the following:\n\nchecking whether time.h and sys/time.h may both be included... yes\nchecking whether struct tm is in sys/time.h or time.h... time.h\nchecking for int timezone... yes\nchecking for union semun... yes\nArgument expected\n\n\n\nThis occurs during the check for whether gcc requires -traditional\nwhen I have specified --with-compiler=cc in order to use the native\nIRIX cc compiler and not gcc. (If I opt to use gcc, then I don't\nhave this problem...)\n\nNOTE this bug has been introduced SINCE THE LAST SNAPSHOT I tried\non the 11th Feb\n\n\n\n3. My expected output files for Irix have not been incorporated in \n.../src/test/regress/expected/ :-(\n------------------------------------------------------------------\n\n\n\n4. gmake install fails for ecpg files\n-------------------------------------\n\nThe ecpg files are using \n install -c -m 755 \nto do the installation rather than\n $(INSTALL) $(INSTLOPTS)\n\nThis applies to .../src/interfaces/ecpg/include/Makefile\nand .../src/interfaces/ecpg/preproc/Makefile\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Mon, 23 Feb 1998 15:18:15 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Snapshot 23Feb98 under Irix5 --- INSTALLATION BROKEN!!!"
},
{
"msg_contents": "On Mon, 23 Feb 1998, Andrew Martin wrote:\n\n> ===============================================\n> \n> 1. .../src/template/.similar: Should say irix5 not irix\n> -------------------------------------------------------\n> \n> *** .similar.old Mon Feb 23 14:18:37 1998\n> --- .similar Mon Feb 23 14:13:49 1998\n> ***************\n> *** 13,19 ****\n> i586-pc-linux-gnu=linux-elf\n> i586-pc-sco3.2v=sco\n> i686-pc-linux-gnu=linux-elf\n> ! mips-sgi-irix=irix\n> powerpc-ibm-aix3.2.5=aix_325\n> powerpc-ibm-aix4.1.4.0=aix_41\n> powerpc-ibm-aix4.1.5.0=aix_41\n> --- 13,19 ----\n> i586-pc-linux-gnu=linux-elf\n> i586-pc-sco3.2v=sco\n> i686-pc-linux-gnu=linux-elf\n> ! mips-sgi-irix=irix5\n> powerpc-ibm-aix3.2.5=aix_325\n> powerpc-ibm-aix4.1.4.0=aix_41\n> powerpc-ibm-aix4.1.5.0=aix_41\n\n\tFixed...\n\n> 2. configure dies (!)\n> ---------------------\n> Running configure, dies with the following:\n> \n> checking whether time.h and sys/time.h may both be included... yes\n> checking whether struct tm is in sys/time.h or time.h... time.h\n> checking for int timezone... yes\n> checking for union semun... yes\n> Argument expected\n\n\tPlease investigate this one further and let me know what is\nrequired to fix it?\n\n> 3. My expected output files for Irix have not been incorporated in \n> .../src/test/regress/expected/ :-(\n> ------------------------------------------------------------------\n\n\tErk...resend it to me privately?\n\n\n> 4. gmake install fails for ecpg files\n> -------------------------------------\n> \n> The ecpg files are using \n> install -c -m 755 \n> to do the installation rather than\n> $(INSTALL) $(INSTLOPTS)\n> \n> This applies to .../src/interfaces/ecpg/include/Makefile\n> and .../src/interfaces/ecpg/preproc/Makefile\n\n\tDamn, thought I had got all of these :( Fixed now...\n\n\n\n",
"msg_date": "Mon, 23 Feb 1998 10:40:27 -0500 (EST)",
"msg_from": "The Hermit Hacker <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] Snapshot 23Feb98 under Irix5 --- INSTALLATION BROKEN!!!"
}
] |
[
{
"msg_contents": "> Here is the newest list:\n> \n> ---------------------------------------------------------------------------\n> \n> Fix isinf.c irix port problem reported by Andrew(Marc)\n\nThis is now fixed, but the latest snapshot has introduced 2 new bugs\ninto installation under Irix and the template/.similar file hasn't\nbeen fixed.\n\nThe current situation is detailed in a separate post.\n\n\nAndrew\n\n----------------------------------------------------------------------------\nDr. Andrew C.R. Martin University College London\nEMAIL: (Work) [email protected] (Home) [email protected]\nURL: http://www.biochem.ucl.ac.uk/~martin\nTel: (Work) +44(0)171 419 3890 (Home) +44(0)1372 275775\n",
"msg_date": "Mon, 23 Feb 1998 15:20:41 GMT",
"msg_from": "Andrew Martin <[email protected]>",
"msg_from_op": true,
"msg_subject": "Re: [HACKERS] Open 6.3 issues"
}
] |
[
{
"msg_contents": "\n1. ecpg Makefiles have hard-coded gcc-only options hard coded.\n\n2. ecpg/preproc/Makefile trashes CFLAGS instead of appending to it.\n\n3. ecpg/preproc/preproc.y has duplicate S_SIGNED\n\nPatch appended to fix #1-#3 for Michael's review. Don't appear to\nbreak anything else, but I'm not the one working on this.\n\n4. ecpg/preproc/preproc.y has an illegal assignment between \"void*\" \n and \"const char*\". Log of the make error is...\n \ngmake -C preproc all\ngmake[3]: Entering directory `/ceo/darrenk/pgsql63d/src/interfaces/ecpg/preproc'\nxlc -I../../../include -I../../../backend -I/gnu -qchars=signed -qmaxmem=8192 -qhalt=w -qsrcmsg -I../include -DMAJOR_VERSION=1 -DMINOR_VERSION=0 -DPATCHLEVEL=0 -c y.tab.c -o y.tab.o\n 289 | free(yyvsp[-2].type.name);\n ...........................a......\na - 1506-280 (E) Function argument assignment between types \"void*\" and \"const char*\" is not allowed.\ngmake[3]: *** [y.tab.o] Error 1\n\nThis one I don't know how to fix. I just commented ecpg out of the\ninterfaces make and all else went a-ok.\n\ndarrenk\n\n*** interfaces/ecpg/lib/Makefile.in.orig\tMon Feb 23 09:00:20 1998\n--- interfaces/ecpg/lib/Makefile.in\tMon Feb 23 09:00:51 1998\n***************\n*** 61,66 ****\n libecpg.a : libecpg.a(ecpglib.o) libecpg.a(typename.o)\n \n ecpglib.o : ecpglib.c ../include/ecpglib.h ../include/ecpgtype.h\n! \t$(CC) -O2 -g -Wall -I../include $(PQ_INCLUDE) -c ecpglib.c\n typename.o : typename.c ../include/ecpgtype.h\n! \t$(CC) -g -O2 -Wall -I../include $(PQ_INCLUDE) -c typename.c\n--- 61,66 ----\n libecpg.a : libecpg.a(ecpglib.o) libecpg.a(typename.o)\n \n ecpglib.o : ecpglib.c ../include/ecpglib.h ../include/ecpgtype.h\n! \t$(CC) -I../include $(PQ_INCLUDE) -c ecpglib.c\n typename.o : typename.c ../include/ecpgtype.h\n! \t$(CC) -I../include $(PQ_INCLUDE) -c typename.c\n*** interfaces/ecpg/preproc/Makefile.orig\tMon Feb 23 09:01:35 1998\n--- interfaces/ecpg/preproc/Makefile\tMon Feb 23 09:09:02 1998\n***************\n*** 6,12 ****\n MINOR_VERSION=0\n PATCHLEVEL=0\n \n! CFLAGS=-I../include -O2 -g -Wall -DMAJOR_VERSION=$(MAJOR_VERSION) -DMINOR_VERSION=$(MINOR_VERSION) -DPATCHLEVEL=$(PATCHLEVEL)\n \n all:: ecpg\n \n--- 6,12 ----\n MINOR_VERSION=0\n PATCHLEVEL=0\n \n! CFLAGS+=-I../include -DMAJOR_VERSION=$(MAJOR_VERSION) -DMINOR_VERSION=$(MINOR_VERSION) -DPATCHLEVEL=$(PATCHLEVEL)\n \n all:: ecpg\n \n***************\n*** 21,27 ****\n \n # Rule that really do something.\n ecpg: y.tab.o pgc.o type.o ecpg.o ../lib/typename.o\n! \t$(CC) -g -O2 -Wall -o ecpg y.tab.o pgc.o type.o ecpg.o ../lib/typename.o $(LEXLIB)\n \n y.tab.h y.tab.c: preproc.y\n \t$(YACC) $(YFLAGS) $<\n--- 21,27 ----\n \n # Rule that really do something.\n ecpg: y.tab.o pgc.o type.o ecpg.o ../lib/typename.o\n! \t$(CC) -o ecpg y.tab.o pgc.o type.o ecpg.o ../lib/typename.o $(LEXLIB)\n \n y.tab.h y.tab.c: preproc.y\n \t$(YACC) $(YFLAGS) $<\n*** interfaces/ecpg/preproc/preproc.y.orig\tMon Feb 23 08:38:33 1998\n--- interfaces/ecpg/preproc/preproc.y\tMon Feb 23 08:38:39 1998\n***************\n*** 231,237 ****\n \n %token <tagname> S_SYMBOL S_LENGTH S_ANYTHING S_LABEL\n %token <tagname> S_VARCHAR S_VARCHAR2\n! %token <tagname> S_EXTERN S_STATIC S_AUTO S_CONST S_REGISTER S_STRUCT S_SIGNED\n %token <tagname> S_UNSIGNED S_SIGNED\n %token <tagname> S_LONG S_SHORT S_INT S_CHAR S_FLOAT S_DOUBLE S_BOOL\n %token <tagname> '[' ']' ';' ',' '{' '}' '=' '*' '(' ')'\n--- 231,237 ----\n \n %token <tagname> S_SYMBOL S_LENGTH S_ANYTHING S_LABEL\n %token <tagname> S_VARCHAR S_VARCHAR2\n! %token <tagname> S_EXTERN S_STATIC S_AUTO S_CONST S_REGISTER S_STRUCT\n %token <tagname> S_UNSIGNED S_SIGNED\n %token <tagname> S_LONG S_SHORT S_INT S_CHAR S_FLOAT S_DOUBLE S_BOOL\n %token <tagname> '[' ']' ';' ',' '{' '}' '=' '*' '(' ')'\n\n",
"msg_date": "Mon, 23 Feb 1998 10:31:04 -0500",
"msg_from": "[email protected] (Darren King)",
"msg_from_op": true,
"msg_subject": "ecpg glitches in Feb 23rd beta."
},
{
"msg_contents": "Darren King writes:\n> \n> 1. ecpg Makefiles have hard-coded gcc-only options hard coded.\n> \n> 2. ecpg/preproc/Makefile trashes CFLAGS instead of appending to it.\n> \n> 3. ecpg/preproc/preproc.y has duplicate S_SIGNED\n> \n> Patch appended to fix #1-#3 for Michael's review. Don't appear to\n> break anything else, but I'm not the one working on this.\n\nThanks. I've included them in my source tree and will submit them with my\nnext patch.\n\n> 4. ecpg/preproc/preproc.y has an illegal assignment between \"void*\" \n> and \"const char*\". Log of the make error is...\n\nHopefully I've fixed this one, too.\n\nMichael\n-- \nDr. Michael Meskes, Project-Manager | topsystem Systemhaus GmbH\[email protected] | Europark A2, Adenauerstr. 20\[email protected] | 52146 Wuerselen\nGo SF49ers! Go Rhein Fire! | Tel: (+49) 2405/4670-44\nUse Debian GNU/Linux! | Fax: (+49) 2405/4670-10\n",
"msg_date": "Tue, 24 Feb 1998 12:17:47 +0100 (CET)",
"msg_from": "Michael Meskes <[email protected]>",
"msg_from_op": false,
"msg_subject": "Re: [HACKERS] ecpg glitches in Feb 23rd beta."
}
] |
[
{
"msg_contents": "Marc,\n\nDid you get this message. New version to go into 6.3.\n\nCheers,\nBrook\n\n------- Start of forwarded message -------\nDate: Mon, 23 Feb 1998 17:18:34 +0200\nFrom: Constantin Teodorescu <[email protected]>\nOrganization: FLEX Consulting Braila\nTo: PostgreSQL Interfaces <[email protected]>\nCC: PostgreSQL Questions <[email protected]>\nSubject: [QUESTIONS] New version of PgAccess released - 0.80\nContent-Type: text/plain; charset=us-ascii\nSender: [email protected]\nPrecedence: bulk\n\nHello all,\n\nis good to be back again :-) because last week I've got a flue , I had\nto stay at home but fortunatelly I succeded getting a notebook with me\nand worked a little bit at PgAccess again.\n\nWhat's new in this 0.80 version of PgAccess ?\n\nThe most important thing is that I introduced the form design module.\nIt's in an early alpha stage but I intend to make it full functional\nuntil the next PostgreSQL 6.3 official release so it could be included\nin the CD distribution. BTW, when should this happen ? Also, I would\nlike to release it a week or two before in order to get some feed-back\nfrom users.\n\nI didn't succeed trying to make Visual Tcl to work as a module of\nPgAccess so I decided to reinvent the wheel and writing a tiny and less\nperformant form designer of my own. It wouldn't be certainly better\nthan Vtcl but it would be quite good for database forms designing. Also,\nI would like to improve it in that direction, because it deals with\ndatabases, tables, etc.\n\nMore information on http://www.flex.ro/pgaccess/forms.html\n\nYou can download the last version (as always) from\nhttp://www.flex.ro/pgaccess or by ftp from ftp.flex.ro/pub/pgaccess\n\nI am burning getting some reply's from any of you :-)\n\nAll the best,\n- -- \nConstantin Teodorescu\nFLEX Consulting Braila, ROMANIA\n------- End of forwarded message -------\n",
"msg_date": "Mon, 23 Feb 1998 08:31:39 -0700 (MST)",
"msg_from": "Brook Milligan <[email protected]>",
"msg_from_op": true,
"msg_subject": "[[email protected]: [QUESTIONS] New version of PgAccess released - 0.80]"
}
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.