"Mark Haney" <mha...@ercbroadband.org> posted
4a2fbda2.4030...@ercbroadband.org, excerpted below, on  Wed, 10 Jun 2009
10:05:22 -0400:

> Okay, I'm at a loss here, I've been using the ~amd64 version of portage,
> which supports sets, however I'm not able to do an 'emerge -pve @system'
> as I get:
> 
> octavian ~ # emerge -pve @system
> !!! '@system' is not a valid package ato
> 
> So, what's up with that?  Do I need to download the sets files again?

Markos is correct.  Here's a bit more detail.

Back when portage-2.1.4 was stable, 2.2 was the development series.  It 
got stable enough to release to ~arch, so it was.  However, some of the 
features included in 2.2 matured faster than others.  It turned out the 
sets and preserve-libs features needed a bit more work while Zac and 
others wanted a stable portage with the other new features (USE-deps I 
think, parallel merges) as soon as possible.

So Zac split off the 2.1.5 series without sets and preserve-libs but with 
the other features so it could be stabilized as soon as possible.  The 
trouble was, they needed testers, and 2.2 was already ~arch with sets, 
etc.  So they masked 2.2 again, thus forcing most ~arch users back to 
2.1.5 (and now 2.1.6) to encourage more testing of that branch, thus 
losing the sets feature that the temporarily ~arch 2.2 series had.

However, Gentoo/kde had been waiting on 2.2 with sets to go to ~arch so 
they could put kde-4.2 in the tree, also ~arch, with sets as the 
recommended installation method.  The kde4 upgrade guide in fact 
recommends the use of sets, so anyone that had followed it while merging 
kde 4.2 was likely using sets (tho there were also the old-style meta-
packages available, and that's what a lot of folks who didn't follow the 
upgrade guide used).

So anyone using sets, including those who had followed the kde4 upgrade 
guide for 4.2, ended up having to unmask the portage 2.2 series when it 
was masked, so they could continue using the sets they'd based their KDE 
4.2 (or other) installation on.

I was in this group, so I had to unmask the portage 2.2 series and have 
continued to be able to play with sets. =:^)

Meanwhile, after FEATURES=preserve-libs screwed up my system a bit I 
decided it was causing me way more problems than it solved and turned it 
off, here.  That feature obviously needed a LOT more work.  Personally, 
it reminds me a lot of confcache, another FEATURE that was great in 
theory but not so great in fact, and I think it might meet the same fate 
-- eventual disabling and removal, if it's ever enabled in a stable 
portage in the first place.  We'll see.  Meanwhile, I'm staying FAR away 
from it and have no plans to reenable it any time soon.  So yes, 
definitely, it was a good thing to hold that feature back for more work.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman


Reply via email to