[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Is Sid for broken stuff? Is it too much to ask for testing the packages?



On Thu, Dec 12, 2002 at 04:33:30AM +0000, Oliver Elphick scribbled:
> On Thu, 2002-12-12 at 00:08, Marek Habersack wrote:
> > Another case of an untested package. Going back to postgres - apparently the
> > maintainer "has no time" to test upgrades from earlier formats of the
> > database to the latest format. 
> 
> Nearly correct.
> 
> I can test from the last version to the current one, and I do.  I have
> not uploaded packages that I haven't installed, as you seem to think.
I take your word on that, although I have my doubts. It's not a coincidence
that postgresql, postgresql-contrib and postgresql-dev all had problems in
any of the 7.3-X releases. And the data upgrade that failed for me was from
7.2 -> 7.3 (and, as I said, it seems it's an upstream problem - a package
with such a bug should never hit unstable. I use an unstable machine for
development which includes postgresql - the data isn't important, but it's
needed to do my work. A failed upgrade in this case means more wasted time
to get it working.)

> But a full test would test 6.5.3 in multiple Debian versions to 7.3
> And 7.0release-x to 7.3
> And 7.1
> And 7.2.1 (current stable)
> And 7.2.2
> And 7.2.3
> And all the imtermediate releases I haven't mentioned
No, all you need to test are upgrades from the current versions in
current stable, testing and unstable and perhaps potato. You can whip up a
script to do that in no time, I'm sure.

> And different combinations of binary packages
> And all of the above in multiple locales and encodings and using tape
> backup in postgresql-dump (I don't have a working tapedrive - I wonder
> if anyone has ever used that facility?) and multi-gigabyte databases and
> in all possible combinations of circumstances that the config and
> postinsts test for.
You are exaggerating. Especially that in this case a test whether
7.2 (database format, that is) upgrades to 7.3 would show you the problem.
And the postgresq.env code was untested from March - that is, never tested.
If I didn't happen to have removed that file, the bug might've hit a guy
upgrading their machine to Sarge in 6 months from now. And it would be
because you never tested an obvious, straightforwad and easy case.

> So no, I don't have the time.  If there's anyone who does, please
> contact me.  In the meantime, I have to settle for installing the new
> packages and seeing if they work for me.
Fine, that's fair. But please - issue a formal plead for help on
debian-devel and debian-mentors perhaps. I sincerely hope you will find
helpers as I do desire your packages to be of top notch quality.

> Incidentally, why did you delete postgresql.env? or how did it get
> deleted?
It must've happend in the course of the previous 7.3 misfortunes. I had to
purge postgresql and postgresql-contrib at some point (I think it was the
-3 release, I'm not sure) and the purge didn't remove postgresql.env, so I
removed it by hand. And then the accident happened.

regards,

marek

Attachment: pgpdG1tBUSNQ7.pgp
Description: PGP signature


Reply to: