Re: [gentoo-dev] EAPI 6 portage is out!

[Rich Freeman]

On Wed, Nov 18, 2015 at 7:06 AM, Alexander Berntsen <bernalex@g.o> wrote:
> We are talking about people who run Gentoo stable who need to
> keyword several specific packages because the lack of manpower
> leads to Gentoo stable by itself not being very usable for most
> people.
>

In this case, however, I don't really see that much impact on stable
users.  At most they need to accept a ~arch version of portage until
it becomes stable again.  It is a PITA because of how we tend to drop
versions of ~arch packages before they ever become stable, but any
stable user is already familiar with this pain and I don't really
think it is related to the EAPI6 introduction.

There really isn't a great alternative either.  It seems likely that
portage will end up having a bunch of little bumps with bugfixes until
things settle down, so it isn't a great time to try to stabilize EAPI6
versions of portage.  We'll get through the pain faster with the
widespread testing you get in ~arch.

>
> Whatever. I just wanted to raise my concern. It has been raised.
> You're all free to not care. Too bad for the user^Wthankless
> contributors.

Well, if you care that much, do more than post about it on a list.
This is actually a topic I care a lot about, but right now I don't
have a better solution to offer so it isn't productive to just hurl
abuse on those trying to actually improve things simply because they
aren't improving everything at once.

I don't really have a problem with politely pointing out the downsides
of the current state, but you need to be patient if you don't actually
have a solution for them as nothing is going to happen without one.

So, in an attempt to try to make this discussion more productive, feel
free to start a thread if you have any ideas of practical solutions
for making life better for mixed-keyword users?  My biggest suggestion
would be to avoid pruning older ~arch versions unless they have
serious problems, so that they can become potential stable targets
later, and that maintainers should always have a path to stable in
mind.  Another suggestion would be for maintainers to store some kind
of metadata that communicates their stabilization/versioning strategy
(which could be useful both to mixed-keyword users and to
co-maintainers or other random devs who need to touch ebuilds).  Some
package just can never go stable, and some version series might never
go stable due to upstream reasons, and it would be nice if that were
all captured in some way.

-- 
Rich