Re: [gentoo-dev] package.use.mask / package.use.stable.mask priority

[Bernard Cafarelli]

Le 11/01/2017 8:30, Ulrich Mueller a écrit :
>>>>>> On Tue, 10 Jan 2017, Zac Medico wrote:
>
>> On 01/10/2017 01:56 PM, Bernard Cafarelli wrote:
>>> gnustep-base/gnustep-make has a USE flag (libobjc2) masked globally
>>> in
>>> base/package.use.mask, and unmasked on specific arches in
>>> arch/{amd64,x86}/package.use.mask
>>>
>>> To get a stabilization (#579232) bug finally moving on, I wanted to
>>> leave this flag out, adding a corresponding line in
>>> base/package.use.stable.mask
>
> So do I understand this correctly, there is:
> flag in base/package.use.mask,
> -flag in arch/{amd64,x86}/package.use.mask, and
> flag in base/package.use.stable.mask?

Exactly :)

>>> But repoman replied with a batch of dependency.bad errors...
>>> Does package.use.mask (stable and ~arch) have a higher priority on
>>> package.use.stable.mask (stable only)? Bug or intended behavior?
>
>> If I understand you correctly, then it's the intended behavior. If the
>> flag is masked in both package.use.mask and package.use.stable.mask,
>> then the package.use.stable.mask setting is irrelevant because both
>> package.use.mask and package.use.stable.mask are considered when
>> calculating use.mask settings for any given package.
>
> I believe this is not correct. package.use.stable.mask should take
> precedence within the same profile:
> https://projects.gentoo.org/pms/6/pms.html#x1-58002r1
>
> The problem here is rather that the base profile is processed as a
> whole before the arch specific profile, so you end up with -flag from
> the arch profile.
>
> Putting flag in arch/{amd64,x86}/package.use.stable.mask should solve
> it.
Indeed, I tested (and committed) this yesterday and repoman was happy 
again

-- 
Bernard Cafarelli (Voyageur)
Gentoo developer