[gentoo-dev] Re: [typo] Re: Re: Multiple implementations shouldn't block Gentoo's progress.

["Steven J. Long" <slong@rathaus.eclipse.co.uk>]

Rich Freeman wrote:
> In general I'd avoid any requirement to use a non-base profile.
> Obviously using the right arch/prefix profile makes sense as those are
> fundamental config changes and they're all minimalist profiles anyway.
>  The issues come when you force users to use non-minimalist profiles
> where we start forcing packages/decisions the users don't agree with.
> (And yes, I realize the whole thread is about forcing a decision users
> don't agree with - but we don't have much choice there.)

That's the point though: given that certain decisions are forced if you want to
use gnome3 (ie you must use systemd, which in turn forces a whole set of decisions
about all the functionality you can no longer mix and match) it makes sense to
wrap those into a profile or some sort of configuration that ensures they get what
they need, by default, until such time as they want to tweak things, and pick up
the pieces.

It's also easier for developers to handle, similar to the KDE profiles. Though I'm
not sure why it's necessary to use a "non-base" profile. We have several
"non-minimalist" profiles already, and the suggestion seems to fit into the
existing framework well: what profiles (and sub-directories thereof) were designed
for, afaict.

Mixins appear to be akin to the eblit/eclass separation, and from what i've read so
far most other profiles don't appear to force so many decisions, just enable what
they need. However if a configuration set does require more, it would appear to be
even _more_ incumbent to use a profile, not less, since it has more consequences
for the end-user than simply adding a defined set of functionality that does not
conflict with anything else.

-- 
#friendly-coders -- We're friendly, but we're not /that/ friendly ;-)