[gentoo-dev] Re: GLEP 55 Version 2

[Steven J Long]

Roy Bamford wrote:
> I've spent some time reading all of this years emails on GLEP55 and
> added a few lines to version 1.5 which is the last offical version.
>
Thanks for all the hard work.

My apologies for my mistaken comment at the end of the last Council meeting.
Clearly the mangler /does/ need to know the EAPI without sourcing for future
extensibility.

I'd just like to know what the implications would be for users if we kept
the .ebuild extension, and a new PMS were rolled out stating that the
mangler were allowed to find the EAPI without sourcing (and giving the
restrictions) once portage 2.2 was stable, or the ability to handle this
backported to 2.1.6, and issued in a release?

Would it be unacceptable to have a clear upgrade path for any users who
didn't actually update portage? (Perhaps a news item so they'd be
notified as and when they ran emerge.)

It strikes me that we went through a similar situation with bash-3.17 I
think it was, and that once we're past this, there'll never be any need
to worry in the future. So, given that it's taken so long and so much
discussion, why not just do this last big bump, and not worry about
about using another extension at all?

After all, the issue would only arise once Council approved an EAPI
requiring one of the incompatible changes, and 3 has only recently come
out.

Also, you asked for proponents of either side to provide statistics as to
the time difference between the various options. It's rather hard for me
to patch paludis, nor do I have the inclination. I am not being facetious,
simply pointing out that the comparison has to be made within a single app.
Comparing an approach implemented in portage vs one in paludis is comparing
apples and oranges.

Also, Patrick brought up cache improvements (like not having a single cache
file per ebuild, but rather per package. This could have the EAPI and
versions first, followed by metadata per version.) I feel we should bear in
mind that there are other areas where we can improve things, many of which
we have not even thought of, or at least discussed.

With respect to time gains in interactivity, much has been made of the
speed difference between portage and paludis. I have yet to see any
comparative numbers between pkgcore and paludis in this regard. (If we are
going to compare manglers and not algorithms.)

I think we are agreed now that an EAPI which can be extracted before source
does fulfil all the criteria (as others have said, this is actually what the
GLEP is about; ascertaining the EAPI without needing to source.) If not,
could someone please explain why not.

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