Re: [gentoo-dev] Sets in the tree

[Michał Górny]

Dnia 2013-08-14, o godz. 16:56:09
Ciaran McCreesh <ciaran.mccreesh@××××××××××.com> napisał(a):

> On Wed, 14 Aug 2013 11:50:56 -0400
> "Anthony G. Basile" <blueness@g.o> wrote:
> > On 08/14/2013 11:41 AM, Patrick Lauer wrote:
> > > On 08/14/2013 10:17 PM, Ciaran McCreesh wrote:
> > >> On Wed, 14 Aug 2013 17:07:32 +0400
> > >> Sergey Popov <pinkbyte@g.o> wrote:
> > >>> I am all for the standarts, but as we did not brought sets to PMS
> > >>> yet(when we updated it for EAPI changes), my question is: 'why?'.
> > >>> It is one of the long-standing feature of quite experimental
> > >>> 2.2_alpha branch, that should finally come to release(Thanks to
> > >>> portage team, by the way :-)).
> > >>>
> > >>> Why it was not added as a part of the PMS? Some implementation
> > >>> flaws? Or maybe, architecture problems?
> > >> Because the Portage format involves executing arbitrary Python code
> > >> that can depend in arbitrary ways upon undocumented Portage
> > >> internals that can change between versions.
> > >>
> > > You keep repeating that.
> > >
> > > That doesn't make it more true.
> > >
> >
> > Even if it were true, this does not stop pms from providing an
> > abstraction layer which provides the needed support despite the
> > details of the underlying implementation.  The argument that
> > implementation details limit such possibilities is spurious and
> > should be ignored.
>
> Why would we design a spec around "arbitrary list of class names that
> happen to be present in some particular version of Portage"?

Well, I'm pretty sure I *asked* at some point to have the thing
formalized, and therefore replacing portage class names with some
official abstract package set classes. As far as I remember, it ended
up like 'we don't want anything except plain simple package lists'.

-- 
Best regards,
Michał Górny