Re: Possible planner bug/regression introduced in 8.2.5 - Mailing list pgsql-bugs

From Greg Sabino Mullane
Subject Re: Possible planner bug/regression introduced in 8.2.5
Date
Msg-id 37b43c9a0baf9a954154d27aa6386668@biglumber.com
Whole thread Raw
In response to Re: Possible planner bug/regression introduced in 8.2.5  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: Possible planner bug/regression introduced in 8.2.5
List pgsql-bugs
-----BEGIN PGP SIGNED MESSAGE-----
Hash: RIPEMD160


> Tom Lane wrote:
>> Please try the attached patch (in addition to the one I sent earlier).

> This is biting us too, quite badly. Any chance this can get pushed into a
> 8.2.6?

> Those patches are certainly already in the 8.2 CVS branch, so your
> question seems to mean "are we going to push 8.2.6 immediately to fix
> this".  My vote would be no --- 8.2.5 is less than six weeks old and
> we don't have that many bugs against it.  Given the overhead involved
> in a release, both from our point of view as packagers and users' point
> of view in having to install it, a single bug has to be pretty darn
> catastrophic to force an update by itself.  This doesn't seem to me
> to reach that level...

I suppose catastophic is in the eye of the beholder, but this is very,
very severe to one of our clients. So much so that they may end up going
back to 8.1. The patches to 8.2.5 fix some of the queries, but not all;
some of the problems seem to exist on 8.2.4 as well. We'll try to
develop a self-contained test case that shows the problem, but until then
wanted to give a heads up that a problem may still exist.

- --
Greg Sabino Mullane greg@turnstep.com
End Point Corporation
PGP Key: 0x14964AC8 200711051303
http://biglumber.com/x/web?pk=2529DF6AB8F79407E94445B4BC9B906714964AC8
-----BEGIN PGP SIGNATURE-----

iD8DBQFHL1rpvJuQZxSWSsgRAzNuAJ9xkww4QwjoavHO9SkTf9Zm7Jl9PgCfQ77k
UXPer7AeI0xXe/f3XkMkUps=
=imfW
-----END PGP SIGNATURE-----

pgsql-bugs by date:

Previous
From: Bruce Momjian
Date:
Subject: Re: BUG #3681: fillers are NULL in pgbench
Next
From: Tom Lane
Date:
Subject: Re: Possible planner bug/regression introduced in 8.2.5