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

Re: Bug#439389: Changes in the xine-lib package require changes in xine-frontends



Josselin Mouette <joss@debian.org> writes:

> Hi,
>
> Le lundi 22 octobre 2007 à 15:30 +0200, Reinhard Tartler a écrit :
>> Josselin Mouette <joss@debian.org> writes:
>> 
>> > Le lundi 22 octobre 2007 à 13:58 +0200, Julien Cristau a écrit :
>> >> So if I have the frontend from etch, and upgrade to the new libxine1
>> >> without installing libxine1-x, the frontend is broken?
>> 
>> Err, I've never heared that partial upgrades were supported. 
>
> There's a difference between supporting them and knowingly breaking them
> - and knowingly breaking all upgrades in testing and unstable without
> adding proper Conflicts.
>
>> AFAIK, if you decide to stay on etch and want to upgrade libxine1
>> only, the procedure is called a 'backport'.
>
> No, it is called a partial upgrade and it should work, because working
> partial upgrades are a necessary condition for working full upgrades,
> and a necessary condition for a usable unstable distribution. If a
> specific partial upgrade doesn't work, you should mark it with a
> "Conflicts:" field (until the "Breaks:" field is implemented).

You are seriously suggesting libxine1 should be added versioned
conflicts against all X11-based frontends in etch? Please think a minute
about this.

> Otherwise, we could as well use Redhat or Ubuntu.

I don't really understand this argument, but I take this comment as
indication that you are not interested in a serious discussion. At least
until you clarify this.

Btw, Ubuntu has "Break"-support since a couple of releases.

>> > Yes. It sounds to me that the cure is worse than the disease.
>> 
>> Sorry? Can you please elaborate on this?
>> 
>> This change has not been uploaded yet, so we still have time to
>> reconsider. But please make sure that you have read and understood the
>> problem and the arguments from #439389 first.
>
> Moving the plugins to a separate package is definitely a good idea, but
> removing functionality in a library package is called breaking the ABI,
> and it requires a change in the package name.

Another indication that you haven't read the referenced bug. There is no
breakage in ABI, as already explained there.

Please do not make this discussion harder by contributing offensive or
unhelpful posts.


-- 
Gruesse/greetings,
Reinhard Tartler, KeyID 945348A4

Attachment: pgpoVcyg64k1k.pgp
Description: PGP signature


Reply to: