[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gentoo-dev
Subject:    Re: [gentoo-dev] Let's populate IUSE_IMPLICIT in the base profile
From:       Zac Medico <zmedico () gentoo ! org>
Date:       2012-09-27 17:56:47
Message-ID: 5064935F.3050107 () gentoo ! org
[Download RAW message or body]

On 09/27/2012 10:45 AM, Ian Stakenvicius wrote:
> On 27/09/12 12:02 PM, Ian Stakenvicius wrote:
>> On 27/09/12 11:57 AM, Mike Gilbert wrote:
>>> On Thu, Sep 13, 2012 at 1:40 AM, Zac Medico <zmedico@gentoo.org> 
>>> wrote:
>>>> Hi,
>>>>
>>>> The council has approved [1] "Profile IUSE injection" [2] for 
>>>> inclusion in EAPI 5, and in latest Portage we have
>>>> experimental EAPI 5_pre2 [3] which implements all of the
>>>> approved features. So, now would be a good time to start
>>>> populating IUSE_IMPLICIT with whatever values may be
>>>> appropriate.
>>>>
>>>> What values belong there? Some of the flags that appear in 
>>>> profiles/base/use.mask might make good candidates, such as
>>>> prefix and selinux. How about other special flags like
>>>> bootstrap, build, and test?
>>>>
> 
>>> prefix and test make sense to me. I'm not so familiar with the 
>>> others.
> 
> 
>> build is specifically for catalyst and/or for building the stages, 
>> right?  If so, this one makes sense to me to add.
> 
>> bootstrap I would guess is similar?  Unsure how that one is used
>> at present.  If IUSE_IMPLICIT would still allow the boostrapping
>> tool to set the use flag, i see no issues having it in the list.
> 
> 
> For the purposes of EAPI5 testing (overlays etc), would it make sense
> to start with this list of flags within IUSE_IMPLICIT on
> base/make.defaults now, and then based on consensus that list can be
> trimmed or appended?

I would recommend to stay on the conservative side and only add ones
that we're sure we need for specific ebuilds. We can remove flags later,
but it's better if can avoid adding unneeded ones in the first place.

> floppym's already requested 'prefix' so that his chromium tests with
> EAPI5 don't fail or need an explicit 'prefix' in IUSE, for instance

I think it's pretty clear that the 'prefix' flag is special, so I would
go ahead and add it.
-- 
Thanks,
Zac

[prev in list] [next in list] [prev in thread] [next in thread]