Re: buildfarm member tapir failing PLCheck in 8.1 branch

Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> writes:
> I noticed that tapir is failing the plpython PLCheck test in the 8.1
> branch. I was worried for a jiffy that it could be related to my
> changes of yesterday, but on looking closer it seems it has been failing
> for quite a while ..

> Hmm, oh, and in looking even closer, the reason it's not failing in 8.2
> and HEAD seems to be that we added new alternative output to match
> Python 2.5.

> Should we add the plpython_error_3.out file to 8.1?

That's not the only change that would need to be made to support python
2.5 in 8.1. I suggested back-patching here:
http://archives.postgresql.org/pgsql-hackers/2007-10/msg00715.php
but there wasn't any enthusiasm for it.

What might make more sense is to encourage buildfarm owners not to
bother testing combinations that we have decided won't be supported.
For instance, I wouldn't mind seeing a mass decommissioning of pre-8.2
Windows testing; the intermittent failures we see on those machines
--- mostly stats tests, eg here
http://www.pgbuildfarm.org/cgi-bin/show_log.pl?nm=narwhal&dt=2007-11-23%2005:00:01
--- aren't ever going to be fixed, and they distract attention
from problems that we might like to fix.

regards, tom lane