Hi there! On Wed, 23 Apr 2008 07:22:39 +0200, Adam D. Barratt wrote: > On Wed, 2008-04-23 at 09:30 +0800, Stefano Zacchiroli wrote: >> On Tue, Apr 22, 2008 at 09:51:42PM +0100, Adam D. Barratt wrote: > [...] >>> I'm not convinced this qualifies as "important", as the above is missing >>> a condition - "and the X-Vcs-* headers differ between the available >>> source versions".
I was unsure and then I considered the definition of "important": IMHO
the problem is not different Vcs-* headers, but the fact that a wrong
version (i.e. the one from testing) is choosen when we clearly state
that development is done on unstable. Anyway, I'm fine to downgrade the
severity.
>>> I agree the current behaviour isn't optimal in such a case though.
>>> Taking the header from the highest available version would be one
>>> relatively easy solution.
>>
>> Ack on this solution, just choose the "highest" (assuming we have a good
>> definition of "highest", I postulate we can only approximate it) among
>> the available entries.
>
> I've implemented this solution in SVN.
Can I disagree? We should not choose the "highest" version, because
this can be the one in experimental (which is wrong as well):
=====
[EMAIL PROTECTED]:~$ apt-cache policy ffmpeg
ffmpeg:
Installed: 0.svn20080206-3
Candidate: 0.svn20080206-3
Version table:
*** 0.svn20080206-3 0
1 http://cdn.debian.net experimental/main Packages
100 /var/lib/dpkg/status
0.cvs20070307-6 0
500 http://cdn.debian.net testing/main Packages
990 http://cdn.debian.net sid/main Packages
[EMAIL PROTECTED]:~$
=====
Either we need the possibility to choose which version we want to
retrieve (default to unstable), either let's go *always* for unstable.
However, I still think this is a bug in apt-cache, since I don't see the
rationale for showing firstly testing (and then all the rest). I'm
posting to d-d for wider audience.
Thx, bye,
Gismo / Luca
pgpbKYIOLAAwq.pgp
Description: PGP signature
