Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
On Thu, Apr 26, 2007 at 10:29:47AM -0700, Mark Wong wrote: > I applied it to REL8_2_STABLE/pgsql and ran 'run_build.pl --test > --keepall REL8_2_STABLE'. Looks like it passed everything for me. :) Thanks for this test. I just committed the changes to CVS. Michael -- Michael Meskes Email: Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org) ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: [EMAIL PROTECTED] Go SF 49ers! Go Rhein Fire! Use Debian GNU/Linux! Use PostgreSQL! ---(end of broadcast)--- TIP 6: explain analyze is your friend
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
On Thu, Apr 26, 2007 at 01:35:42PM -0400, Tom Lane wrote: > AFAICS you do not need to inline create_statement. The risk factor > is where you call a routine that does something with a va_list, and > then you want to do something else (other than va_end) with that va_list > after it returns. The one use of create_statement doesn't do that, > hence no problem. (We know this works, because stringinfo.c does it.) Given all the problems I had so far with this code I'd prefer to keep it inlined. :-) > BTW, I think there is a small bug there: you fail to call va_end() in the > failure path. I'm not sure if there are any modern machines where > va_end() isn't a no-op, but nonetheless the code isn't meeting the spec. Fixed. Thanks. Michael -- Michael Meskes Email: Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org) ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: [EMAIL PROTECTED] Go SF 49ers! Go Rhein Fire! Use Debian GNU/Linux! Use PostgreSQL! ---(end of broadcast)--- TIP 3: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faq
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
Michael Meskes <[EMAIL PROTECTED]> writes: > Having spend countless hours debugging this stuff I fully agree with > you. It's not just ECPGget_variable though. I also had to inline > create_statement. AFAICS you do not need to inline create_statement. The risk factor is where you call a routine that does something with a va_list, and then you want to do something else (other than va_end) with that va_list after it returns. The one use of create_statement doesn't do that, hence no problem. (We know this works, because stringinfo.c does it.) BTW, I think there is a small bug there: you fail to call va_end() in the failure path. I'm not sure if there are any modern machines where va_end() isn't a no-op, but nonetheless the code isn't meeting the spec. regards, tom lane ---(end of broadcast)--- TIP 5: don't forget to increase your free space map settings
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
On 4/26/07, Michael Meskes <[EMAIL PROTECTED]> wrote: On Wed, Apr 25, 2007 at 04:38:30PM -0400, Tom Lane wrote: > My recommendation is to get rid of the APREF hack, deal only in > va_list not &va_list, and inline ECPGget_variable into the two > places it's used to avoid the question of passing va_lists around > after they've been modified. The routine's not that big (especially > seeing that only half of it is actually shared by the two callers) > and it's just not worth the notational effort, let alone any portability > risks, to factor it out. Having spend countless hours debugging this stuff I fully agree with you. It's not just ECPGget_variable though. I also had to inline create_statement. This is only called once, so no big deal, but the calling function gets bigger by quite a margin that way. It was definitely easier to read with these functions, but again all this hassle isn't worth it. Attached you'll find a patch that should inline both functions and remove the APREF stuff. This successfully runs the regression suite on my Linux box. Please test it on those archs that needed special treatment before I commit. I applied it to REL8_2_STABLE/pgsql and ran 'run_build.pl --test --keepall REL8_2_STABLE'. Looks like it passed everything for me. :) Regards, Mark ---(end of broadcast)--- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
On Thu, Apr 26, 2007 at 06:28:29AM -0500, Andrew Dunstan wrote: > If you commit to HEAD it will be automatically tested on the buildfarm. True. But it might also break a lot of other archs without helping on those troubled ones. I thought this way would be better. Michael -- Michael Meskes Email: Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org) ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: [EMAIL PROTECTED] Go SF 49ers! Go Rhein Fire! Use Debian GNU/Linux! Use PostgreSQL! ---(end of broadcast)--- TIP 3: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faq
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
Michael Meskes wrote: > > Attached you'll find a patch that should inline both functions and > remove the APREF stuff. This successfully runs the regression suite on > my Linux box. Please test it on those archs that needed special > treatment before I commit. > If you commit to HEAD it will be automatically tested on the buildfarm. cheers andrew ---(end of broadcast)--- TIP 1: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
ARGH! This time with patch. Michael -- Michael Meskes Email: Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org) ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: [EMAIL PROTECTED] Go SF 49ers! Go Rhein Fire! Use Debian GNU/Linux! Use PostgreSQL! diff -ruN --exclude CVS /home/postgres/pgsql-ecpg/ecpglib/descriptor.c ecpglib/descriptor.c --- /home/postgres/pgsql-ecpg/ecpglib/descriptor.c 2006-10-04 09:38:18.0 +0200 +++ ecpglib/descriptor.c 2007-04-26 12:02:20.0 +0200 @@ -507,7 +507,6 @@ do { enum ECPGdtype itemtype; - enum ECPGttype type; const char *tobeinserted = NULL; bool malloced; @@ -516,13 +515,29 @@ if (itemtype == ECPGd_EODT) break; - type = va_arg(args, enum ECPGttype); -#if defined(__GNUC__) && (defined (__powerpc__) || defined(__amd64__) || defined(__x86_64__)) - ECPGget_variable(args, type, var, false); -#else - ECPGget_variable(&args, type, var, false); -#endif + var->type = va_arg(args, enum ECPGttype); + var->pointer = va_arg(args, char *); + var->varcharsize = va_arg(args, long); + var->arrsize = va_arg(args, long); + var->offset = va_arg(args, long); + + if (var->arrsize == 0 || var->varcharsize == 0) + var->value = *((char **) (var->pointer)); + else + var->value = var->pointer; + + /* + * negative values are used to indicate an array without given bounds + */ + /* reset to zero for us */ + if (var->arrsize < 0) + var->arrsize = 0; + if (var->varcharsize < 0) + var->varcharsize = 0; + + var->next = NULL; + switch (itemtype) { case ECPGd_data: diff -ruN --exclude CVS /home/postgres/pgsql-ecpg/ecpglib/execute.c ecpglib/execute.c --- /home/postgres/pgsql-ecpg/ecpglib/execute.c 2007-03-29 14:04:34.0 +0200 +++ ecpglib/execute.c 2007-04-26 12:01:33.0 +0200 @@ -79,136 +79,6 @@ } } -#if defined(__GNUC__) && (defined (__powerpc__) || defined(__amd64__) || defined(__x86_64__)) -#define APREF ap -#else -#define APREF *ap -#endif - -void -ECPGget_variable(va_list APREF, enum ECPGttype type, struct variable * var, bool indicator) -{ - var->type = type; - var->pointer = va_arg(APREF, char *); - - var->varcharsize = va_arg(APREF, long); - var->arrsize = va_arg(APREF, long); - var->offset = va_arg(APREF, long); - - if (var->arrsize == 0 || var->varcharsize == 0) - var->value = *((char **) (var->pointer)); - else - var->value = var->pointer; - - /* - * negative values are used to indicate an array without given bounds - */ - /* reset to zero for us */ - if (var->arrsize < 0) - var->arrsize = 0; - if (var->varcharsize < 0) - var->varcharsize = 0; - - var->next = NULL; - - if (indicator) - { - var->ind_type = va_arg(APREF, enum ECPGttype); - var->ind_pointer = va_arg(APREF, char *); - var->ind_varcharsize = va_arg(APREF, long); - var->ind_arrsize = va_arg(APREF, long); - var->ind_offset = va_arg(APREF, long); - - if (var->ind_type != ECPGt_NO_INDICATOR - && (var->ind_arrsize == 0 || var->ind_varcharsize == 0)) - var->ind_value = *((char **) (var->ind_pointer)); - else - var->ind_value = var->ind_pointer; - - /* - * negative values are used to indicate an array without given bounds - */ - /* reset to zero for us */ - if (var->ind_arrsize < 0) - var->ind_arrsize = 0; - if (var->ind_varcharsize < 0) - var->ind_varcharsize = 0; - } -} - -/* - * create a list of variables - * The variables are listed with input variables preceding outputvariables - * The end of each group is marked by an end marker. - * per variable we list: - * type - as defined in ecpgtype.h - * value - where to store the data - * varcharsize - length of string in case we have a stringvariable, else 0 - * arraysize - 0 for pointer (we don't know the size of the array), - * 1 for simple variable, size for arrays - * offset - offset between ith and (i+1)th entry in an array, - * normally that means sizeof(type) - * ind_type - type of indicator variable - * ind_value - pointer to indicator variable - * ind_varcharsize - empty - * ind_arraysize - arraysize of indicator array - * ind_offset - indicator offset - */ -static bool -create_statement(int lineno, int compat, int force_indicator, struct connection * connection, struct statement ** stmt, const char *query, va_list APREF) -{ - struct variable **list = &((*stmt)->inlist); - enum ECPGttype type; - - if (!(*stmt = (struct statement *) ECPGalloc(sizeof(struct statement), lineno))) - return false; - - (*stmt)->command = ECPGstrdup(query, lineno); - (*stmt)->connection = connection; - (*stmt)->lineno = lineno; - (*stmt)->compat = compat; - (*stmt)->force_indicator = force_indicator; - - list = &((*stmt)->inlist); - - type = va_arg(APREF, enum ECPGttype); - - while (type != ECPGt_EORT) - { - if (type == ECPGt_EOIT) - list = &((*stmt)->outlist); - else - { - struct variable *var, - *ptr; - - if (!(var = (struct variable *) ECPGalloc(sizeof(struct variable), lineno))) -return false; - - ECPGget_variable(ap, type, var, true);
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
On Wed, Apr 25, 2007 at 04:38:30PM -0400, Tom Lane wrote: > My recommendation is to get rid of the APREF hack, deal only in > va_list not &va_list, and inline ECPGget_variable into the two > places it's used to avoid the question of passing va_lists around > after they've been modified. The routine's not that big (especially > seeing that only half of it is actually shared by the two callers) > and it's just not worth the notational effort, let alone any portability > risks, to factor it out. Having spend countless hours debugging this stuff I fully agree with you. It's not just ECPGget_variable though. I also had to inline create_statement. This is only called once, so no big deal, but the calling function gets bigger by quite a margin that way. It was definitely easier to read with these functions, but again all this hassle isn't worth it. Attached you'll find a patch that should inline both functions and remove the APREF stuff. This successfully runs the regression suite on my Linux box. Please test it on those archs that needed special treatment before I commit. Michael -- Michael Meskes Email: Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org) ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: [EMAIL PROTECTED] Go SF 49ers! Go Rhein Fire! Use Debian GNU/Linux! Use PostgreSQL! ---(end of broadcast)--- TIP 3: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faq
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
On Wed, Apr 25, 2007 at 03:17:19PM -0400, Tom Lane wrote: > Why in the world is that like that? We don't have such a kluge > anyplace else we use va_list. stringinfo.c for instance has > never needed any such thing. I don't remember the exact details but this was added a long time ago before 8.0 because we had some problems with one of the archs. I was suprised about the differences back then too, but haven't checked since. Michael -- Michael Meskes Email: Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org) ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: [EMAIL PROTECTED] Go SF 49ers! Go Rhein Fire! Use Debian GNU/Linux! Use PostgreSQL! ---(end of broadcast)--- TIP 2: Don't 'kill -9' the postmaster
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
Andrew Dunstan <[EMAIL PROTECTED]> writes: > But I also see that my amd64/FC6 machine does pass these tests with gcc. Yeah, but the typedef represented by va_list can and probably does vary between amd64 and ppc64. I haven't an easy way to check, but I wonder whether it's not an array type on ppc. I'm of the opinion that ecpg is trying to do something here that's not portable. The C99 draft I have says [#3] The type declared is va_list which is an object type suitable for holding information needed by the macros va_start, va_arg, va_end, and va_copy. If access to the varying arguments is desired, the called function shall declare an object (referred to as ap in this subclause) having type va_list. The object ap may be passed as an argument to another function; if that function invokes the va_arg macro with parameter ap, the value of ap in the calling function is indeterminate and shall be passed to the va_end macro prior to any further reference to ap. (198) 198 It is permitted to create a pointer to a va_list and pass that pointer to another function, in which case the original function may make further use of the original list after the other function returns. The footnote seems to say that what the code is doing is OK ... but there isn't any such footnote in the Single Unix Spec: http://www.opengroup.org/onlinepubs/007908799/xsh/stdarg.h.html which makes me wonder just how portable it really is. My recommendation is to get rid of the APREF hack, deal only in va_list not &va_list, and inline ECPGget_variable into the two places it's used to avoid the question of passing va_lists around after they've been modified. The routine's not that big (especially seeing that only half of it is actually shared by the two callers) and it's just not worth the notational effort, let alone any portability risks, to factor it out. regards, tom lane ---(end of broadcast)--- TIP 6: explain analyze is your friend
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
Tom Lane wrote: Hmm, and I don't have to look far to find a smoking gun: #if defined(__GNUC__) && (defined (__powerpc__) || defined(__amd64__) || defined(__x86_64__)) if (create_statement(lineno, compat, force_indicator, con, &stmt, query, args) == false) #else if (create_statement(lineno, compat, force_indicator, con, &stmt, query, &args) == false) #endif I also see: #if defined(__GNUC__) && (defined (__powerpc__) || defined(__amd64__) || defined(__x86_64__)) #define APREF ap #else #define APREF *ap #endif But I also see that my amd64/FC6 machine does pass these tests with gcc. I would certainly be nice if we could simplify all this. And if not, we should have a note about why it's needed. cheers andrew ---(end of broadcast)--- TIP 2: Don't 'kill -9' the postmaster
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
On 4/25/07, Tom Lane <[EMAIL PROTECTED]> wrote: "Mark Wong" <[EMAIL PROTECTED]> writes: > Does this help? > (gdb) p var->ind_pointer > $8 = (void *) 0x0 Well, that seems to be the reason why it's failing to indirect through ind_pointer ... but why is it only failing on your machine and not everyone else's? I think this indicates something unportable about ecpg's usage of va_list. Hmm, and I don't have to look far to find a smoking gun: #if defined(__GNUC__) && (defined (__powerpc__) || defined(__amd64__) || defined(__x86_64__)) if (create_statement(lineno, compat, force_indicator, con, &stmt, query, args) == false) #else if (create_statement(lineno, compat, force_indicator, con, &stmt, query, &args) == false) #endif Why in the world is that like that? We don't have such a kluge anyplace else we use va_list. stringinfo.c for instance has never needed any such thing. Mark, does your gcc define __powerpc__, or only __powerpc64__? $ touch foo.c; gcc -E -dM foo.c | grep __p ; rm foo.c #define __powerpc64__ 1 #define __powerpc__ 1 Regards, Mark ---(end of broadcast)--- TIP 4: Have you searched our list archives? http://archives.postgresql.org
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
"Mark Wong" <[EMAIL PROTECTED]> writes: > Does this help? > (gdb) p var->ind_pointer > $8 = (void *) 0x0 Well, that seems to be the reason why it's failing to indirect through ind_pointer ... but why is it only failing on your machine and not everyone else's? I think this indicates something unportable about ecpg's usage of va_list. Hmm, and I don't have to look far to find a smoking gun: #if defined(__GNUC__) && (defined (__powerpc__) || defined(__amd64__) || defined(__x86_64__)) if (create_statement(lineno, compat, force_indicator, con, &stmt, query, args) == false) #else if (create_statement(lineno, compat, force_indicator, con, &stmt, query, &args) == false) #endif Why in the world is that like that? We don't have such a kluge anyplace else we use va_list. stringinfo.c for instance has never needed any such thing. Mark, does your gcc define __powerpc__, or only __powerpc64__? regards, tom lane ---(end of broadcast)--- TIP 1: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
On 4/25/07, Tom Lane <[EMAIL PROTECTED]> wrote: Andrew Dunstan <[EMAIL PROTECTED]> writes: > I think you'll need to compile with optimisation turned off and then try > running the test under debugger control, putting a breakpoint in > ECPGget_variable() and then stepping through it. I wonder what value of > var->ind_pointer it is getting? You could probably inspect the contents of *var in that dump without having to recompile. Given that this is PPC64, I'm betting on a pointer size or alignment problem ... Does this help? (gdb) p var->type $1 = 4267828624 (gdb) p var->value $2 = (void *) 0x1 (gdb) p var->pointer $3 = (void *) 0x1 (gdb) p var->varcharsize $4 = 3 (gdb) p var->arrsize $5 = 4 (gdb) p var->offset $6 = 29 (gdb) p var->ind_type $7 = 0 (gdb) p var->ind_pointer $8 = (void *) 0x0 (gdb) p var->ind_varcharsize $9 = 0 (gdb) p var->ind_arrsize $10 = 0 (gdb) p var->ind_offset $11 = 5 (gdb) p var->next $12 = (struct variable *) 0x0 Regards, Mark ---(end of broadcast)--- TIP 6: explain analyze is your friend
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
Andrew Dunstan <[EMAIL PROTECTED]> writes: > I think you'll need to compile with optimisation turned off and then try > running the test under debugger control, putting a breakpoint in > ECPGget_variable() and then stepping through it. I wonder what value of > var->ind_pointer it is getting? You could probably inspect the contents of *var in that dump without having to recompile. Given that this is PPC64, I'm betting on a pointer size or alignment problem ... regards, tom lane ---(end of broadcast)--- TIP 2: Don't 'kill -9' the postmaster
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
Mark Wong wrote: On 4/25/07, Michael Meskes <[EMAIL PROTECTED]> wrote: I also saw that wombat is segfaulting in ecpg tests but not only with CVS HEAD but also trying to test 8.2. Any idea what's going on with this machine? I generated a stack trace for REL8_2_STABLE, but I'm not sure how helpful it is. Let me know what other information I can provide... Looks like I don't have symbols for libc. Core was generated by `sql/update '. Program terminated with signal 11, Segmentation fault. #0 ECPGget_variable (ap=, type=out>, var=0x10028730, indicator=1 '\001') at execute.c:131 131 var->ind_value = *((char **) (var->ind_pointer)); (gdb) bt #0 ECPGget_variable (ap=, type=out>, var=0x10028730, indicator=1 '\001') at execute.c:131 #1 0x04048948 in ECPGdo (lineno=28, compat=out>, force_indicator=, connection_name=, query=) at execute.c:195 #2 0x1d20 in main (argc=, argv=) at update.pgc:28 #3 0x0463ce4c in .generic_start_main () from /lib/libc.so.6 #4 0x0463d0f8 in .__libc_start_main () from /lib/libc.so.6 #5 0x in ?? () I think you'll need to compile with optimisation turned off and then try running the test under debugger control, putting a breakpoint in ECPGget_variable() and then stepping through it. I wonder what value of var->ind_pointer it is getting? cheers andrew ---(end of broadcast)--- TIP 5: don't forget to increase your free space map settings
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
On 4/25/07, Michael Meskes <[EMAIL PROTECTED]> wrote: I also saw that wombat is segfaulting in ecpg tests but not only with CVS HEAD but also trying to test 8.2. Any idea what's going on with this machine? I generated a stack trace for REL8_2_STABLE, but I'm not sure how helpful it is. Let me know what other information I can provide... Looks like I don't have symbols for libc. Core was generated by `sql/update '. Program terminated with signal 11, Segmentation fault. #0 ECPGget_variable (ap=, type=, var=0x10028730, indicator=1 '\001') at execute.c:131 131 var->ind_value = *((char **) (var->ind_pointer)); (gdb) bt #0 ECPGget_variable (ap=, type=, var=0x10028730, indicator=1 '\001') at execute.c:131 #1 0x04048948 in ECPGdo (lineno=28, compat=, force_indicator=, connection_name=, query=) at execute.c:195 #2 0x1d20 in main (argc=, argv=) at update.pgc:28 #3 0x0463ce4c in .generic_start_main () from /lib/libc.so.6 #4 0x0463d0f8 in .__libc_start_main () from /lib/libc.so.6 #5 0x in ?? () Regards, Mark ---(end of broadcast)--- TIP 1: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
Dave Page wrote: Andrew Dunstan wrote: Please don't do that on your buildfarm repo copy (if that's what you did). You should not touch *anything* inside it. If need to you do this, make a copy (see later) and alter that. If you did do this to the buildfarm repo copy, please blow it away so that buildfarm will get a fresh clean copy next time it runs. No, I reset the permissions on /msys. Thats entirely necessary on Vista where the permissions are so locked down by default that you can't even create it without some measure of pain. Yes, my Vista experience so far has been very unpleasant indeed. It strikes me as the OS equivalent of jumping the shark. I haven't even thought about Msys ... cheers andrew ---(end of broadcast)--- TIP 5: don't forget to increase your free space map settings
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
Andrew Dunstan wrote: > Please don't do that on your buildfarm repo copy (if that's what you > did). You should not touch *anything* inside it. If need to you do this, > make a copy (see later) and alter that. > > If you did do this to the buildfarm repo copy, please blow it away so > that buildfarm will get a fresh clean copy next time it runs. No, I reset the permissions on /msys. Thats entirely necessary on Vista where the permissions are so locked down by default that you can't even create it without some measure of pain. Regards, Dave. ---(end of broadcast)--- TIP 4: Have you searched our list archives? http://archives.postgresql.org
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
Dave Page wrote: Michael Meskes wrote: On Wed, Apr 25, 2007 at 10:47:57AM +0100, Dave Page wrote: I'm seeing an ECPG-Check failure on Windows Vista - any ideas what might be causing this? Hmm, first glance suggests some permission problems. Yes, that was my thought as well, however I ran cacls down the entire build tree, granting full control to Everyone on everything, but the problem persists. Please don't do that on your buildfarm repo copy (if that's what you did). You should not touch *anything* inside it. If need to you do this, make a copy (see later) and alter that. If you did do this to the buildfarm repo copy, please blow it away so that buildfarm will get a fresh clean copy next time it runs. Additionally, all the other tests pass, including make check and make installcheck, so unless the ECPG tests are trying to create something outside of the buildtree, I don't think it is a permissions issue. Is there anything I can try building/running manually to help debug this? To recreate the failing buildfarm environment (including making an alterable repo copy), run the following (with suitable local mods): run_build.pl --test --keepall At the end of that you should have the build tree that was actually used, and the install tree too if it got that far. Then you can start playing manually. Remember to move or remove the kept trees before your next buildfarm run. cheers andrew ---(end of broadcast)--- TIP 1: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
Michael Meskes wrote: > On Wed, Apr 25, 2007 at 10:47:57AM +0100, Dave Page wrote: >> I'm seeing an ECPG-Check failure on Windows Vista - any ideas what might >> be causing this? > > Hmm, first glance suggests some permission problems. Yes, that was my thought as well, however I ran cacls down the entire build tree, granting full control to Everyone on everything, but the problem persists. Additionally, all the other tests pass, including make check and make installcheck, so unless the ECPG tests are trying to create something outside of the buildtree, I don't think it is a permissions issue. Is there anything I can try building/running manually to help debug this? > I also saw that wombat is segfaulting in ecpg tests but not only with > CVS HEAD but also trying to test 8.2. Any idea what's going on with this > machine? > > Michael Can't help with that one I'm afraid. Regards, Dave ---(end of broadcast)--- TIP 1: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
Re: [HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
On Wed, Apr 25, 2007 at 10:47:57AM +0100, Dave Page wrote: > I'm seeing an ECPG-Check failure on Windows Vista - any ideas what might > be causing this? Hmm, first glance suggests some permission problems. I never touched a Vista system so far, so I'm at a loss as far as details are concerned. I also saw that wombat is segfaulting in ecpg tests but not only with CVS HEAD but also trying to test 8.2. Any idea what's going on with this machine? Michael -- Michael Meskes Email: Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org) ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: [EMAIL PROTECTED] Go SF 49ers! Go Rhein Fire! Use Debian GNU/Linux! Use PostgreSQL! ---(end of broadcast)--- TIP 1: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
[HACKERS] ECPG failure on BF member Vaquita (Windows Vista)
I'm seeing an ECPG-Check failure on Windows Vista - any ideas what might be causing this? http://www.pgbuildfarm.org/cgi-bin/show_log.pl?nm=vaquita&dt=2007-04-24%2020:00:05 The only other Vista buildfarm member (baiji, on the same physical box) is running MSVC builds which don't yet test ECPG from what I can see. Regards, Dave ---(end of broadcast)--- TIP 2: Don't 'kill -9' the postmaster