Re: fedora core 2 postgresql regression tests fail

stig erikson wrote:
> John McBride wrote:
>
>>
>> As user postgres I try and run the tests:
>>
>> $ pwd
>> /usr/lib/pgsql/test/regress
>> $ time ./pg_regress.sh --schedule=parallel_schedule
>> ....
>> ====================================================
>> 6 of 93 tests failed, 1 of these failures ignored.
>> ====================================================
>
> might or might not help for your test, there is a 7.4.5 version for
> fedora core 2.
>
> you can find it on any of the mirrors, for example:
> ftp://ftp7.de.postgresql.org/pub/ftp.postgresql.org/binary/v7.4.5/rpms/fedora/fedora-core-2
>

I forgot to mention something. It looks like the tests are setup for
people who install from source, not rpm.

The README file says to run "gmake check" to run the regression tests;
if I do that I get:

$ gmake check
GNUmakefile:16: ../../../src/Makefile.global: No such file or directory
gmake: *** No rule to make target `../../../src/Makefile.global'. Stop.

So I look at the makefile and try to run the command myself (I have seen
from google that some other people do this, too) :

$ time ./pg_regress.sh --schedule=parallel_schedule
-bash: ./pg_regress.sh: Permission denied

real 0m0.006s
user 0m0.000s
sys 0m0.002s

$ chmod +x ./pg_regress.sh
$ time ./pg_regress.sh --schedule=parallel_schedule(using postmaster on
Unix socket, default port)
============== dropping database "regression" ==============
./pg_regress.sh: line 450: @bindir@/dropdb: No such file or directory
============== creating database "regression" ==============
./pg_regress.sh: line 490: @bindir@/createdb: No such file or directory
pg_regress.sh: createdb failed

real 0m1.404s
user 0m0.093s
sys 0m0.195s

Now I change the variable @bindir@ in the script to /usr/bin and try
again, getting the results of my previous email:

====================================================
6 of 93 tests failed, 1 of these failures ignored.
====================================================

(random sometimes fails too but I know about that from the docs).

It is now obvious to me that these failures are probably caused by the
remaining uninitialized variables in pg_regress.sh (if this was a source
build/install "gmake check" would successfully run and "sed" the
variables in the script for me).

I don't know how to set these:

pg_regress: pg_regress.sh GNUmakefile
sed -e 's,@bindir@,$(bindir),g' \
-e 's,@libdir@,$(libdir),g' \
-e 's,@pkglibdir@,$(pkglibdir),g' \
-e 's,@datadir@,$(datadir),g' \
-e 's/@VERSION@/$(VERSION)/g' \
-e 's/@host_tuple@/$(host_tuple)/g' \
-e 's,@GMAKE@,$(MAKE),g' \
-e 's/@enable_shared@/$(enable_shared)/g' \
-e 's/@GCC@/$(GCC)/g' \
$< >$@
chmod a+x $@

---
John