Re: [HACKERS] query crashes backend - cvs (fwd)

Bruce,

after sending you email I made an experiment to verify if locale
settings effect to failures of some regression tests.
I did a clean recompilation of latest cvs with --enable-locale option.
First time I run regression test strictly follow instructions in
INSTALL document. Almost all tests run ok except geometry and random.
Second time I start postmster as I usually run on production machine -
i.e. I specified LC_CTYPE and LC_COLLATE env. variables ( koi8-r).
Again geometry fails but random ok (strange) and
select_implicit .. failed
select_having .. failed
select_views .. failed

So, certainly something either in locale code or in my system locale is broken.
But as far as I remember there were no problem with 6.3.2 and there is
nothing locale specific ( I mean national characters) in regression test.
So, most probable there is a bug in a code.
Differences are only in the order of rows returned from query,
for example:
dv:~/cvs/pgsql/src/test/regress/results$ diff select_having.out ../expected/select_having.out
33d32
< bbbb    |  5
34a34
> bbbb    |  5
I run all these tests on regular Linux box, nothing bleeding edge -
Linux dv 2.0.34 #7 Wed Aug 26 14:05:30 MSD 1998 i586 unknown
Kernel modules         found
Gnu C                  2.7.2.3
Binutils               2.8.1.0.23
Linux C Library        5.4.44
Dynamic Linker (ld.so) 1.9.7
Linux C++ Library      27.2.8
Procps                 1.2.6
Procinfo               12
Mount                  2.7l
Net-tools              1.432.
Kbd                    0.94
Sh-utils               1.16


It would be great if other linux users could verify my results.

    Best regards,

        Oleg



_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83