Re: How to enable testing migration for packages Architecture: all but depending from Architecture: any packages [and 2 more messages]

[Niels Thykier <niels@thykier.net>]

Mattia Rizzolo:
> On Thu, Mar 29, 2018 at 03:37:10PM +0100, Ian Jackson wrote:
>> But, Andreas linked to the excuses page (his [1], above) which mention
>> a lot of other architectures, where installability of the dependencies
>> is not relevant.  Eg
>>   paleomix/s390x unsatisfiable Depends: python-pysam
>>
>> I can see why Andreas thought the way he did.  Would it be possible
>> for the excuses to be made more precise ?  Can I file a bug somewhere
>> to request that ?
> 
> That would be a Britney bug, therefore file it against relase.debian.org
> and usertag it with 'britney'.
> 
> In that case, I'd rather keep listing all the broken architectures, but
> explicitly mark those allowed to break as such.
> 

For the record; It is a known issue documented on
https://www.debian.org/devel/testing.en.html

Quote:
> "Why is it sometimes hard to get Architecture: all packages into "testing"?"
> 
> If the Architecture: all package is to be installed, it must be possible to satisfy its dependencies on all architectures. If it depends on certain packages which only compile on a limited set of Debian's architectures, then it can't do that.
> 
> However, for the time being, "testing" will ignore Architecture: all packages' installability on non-i386 architectures. ("It's a gross hack and I'm not really happy to have made it, but there you go." —aj)

(Out of dateness note: We now use i386 AND amd64 as minimum requirement
instead of only i386).

The problem with undoing this gross hack is replacing it with something
that "just works" without (too much) manual hand-holding and does not
allow "obvious regressions" to migrate.  For those who are interested in
working on such a patch, my notes on the area are the following:

"""
> Britney's current arch:all has work ok for a long while, but it is a hack and should be replaced by a proper solution (that does not require force-hints when Britney is wrong).
> 
> What is expected from Britney's arch:all handling?
> 
>  * arch:all packages must be installable on at least 1 architecture.
>  * arch:all packages may be uninstallable on a strict subset of the architectures assuming they have no arch:any rdeps on that architecture.
>  * arch:all that used to be installable must not regress without a hint/manual approval except:
>      - If the source previously provided arch:any binaries but no longer does for that architecture. (Assumption, if arch:any + arch:all are provided, the arch:all are assumed only to be relevant on architectures with arch:any binaries)
"""

(caveat emptor: My notes might not be the canonical source of truth for
this problem)

Thanks,
~Niels