Bug#774737: unblock: libjpeg9/1:9a-2

[Niels Thykier <niels@thykier.net>]

On 2015-02-02 00:04, Bill Allombert wrote:
> On Fri, Jan 23, 2015 at 08:02:38AM +0100, Niels Thykier wrote:
>> I am afraid I do not see how removing libjpeg9 from testing is
>> inconsistent with the tech-ctte decision. 
> 
> You need to reread the full decision in context.
> 
>> The very first item of their
>> resolution text states that:
>>
>>  1. [...] The release team does not want to have more than one libjpeg
>>     implementation.
> 
> This is in the Whereas part, not in the Therefore. Thus this is what the
> release team want, but not necessarily what the TC has decided.
> 
>> Then further down, they follow up with:
>>
>> 10. The Technical Committee resolves that libjpeg-turbo should
>>     become the libjpeg implementation in Debian, [...]
> 
> And this is the case now. However the TC did not say all other libjpeg
> implementations need to be removed from testing.

The release team decided that there would be only one libjpeg
implementation in Jessie.  The tech-ctte made the choice of which
implementation.

> Indeed wheezy includes both libjpeg6b and libjpeg8 so there is a 
> precedent for that. At the very least it is customary to provide old
> libraries in the next release as part of the oldlibs section.
> 

I see no precedent for or against that.  We got plenty of libraries
Wheezy that will be upgraded without the old version of the library
being available in Jessie.  In fact, I /suspect/ that for most libraries
it is more common to *not* do this.

> Then the TC gives a detailed view fo what should happens:
> 
> 12. [...]
> 
> This is an unambiguous statement that they only intent the "Provides:
> libjpeg-dev" of libjpeg8 to be removed and not the whole package, otherwise
> they would have stated it directly (in particular since removing libjpeg8
> automatically remove the Provides making it a non-issue).
> The text shows they anticipate the existence of multiple 'real libjpeg*-dev
> packages' but only one providing libjpeg-dev.
> 

I believe this makes sense, as decision of the release only affects
testing.  Accordingly, there is nothing preventing multiple
implementations of libjpeg in testing or unstable.

> And in any case, the release team never communicated to me their intent to
> remove libjpeg6b, libjpeg8 and libjpeg9 from jessie. I only learned about
> it in January from the archive notification. And so far no rationale has been
> given.
> 
> Cheers,
> 

I am sorry you feel that way.  I assumed that you were aware, given that
you replied to the tech-ctte decision in which our statement was
included (in fact, you retained it in quoted part of your reply).

~Niels