1.1

> The 2008.0 stages will have whatever versions of packages
> are in the tree at  the time of our snapshot.

Right; that's also what I meant about rolling stage3's.  The only
difference being that much more careful attention is paid to the
official release tarballs, stripping out blockages and the like.


> not feasible when you start looking at all the architectures

Hm.

Portage runs Python bytecodes and bash scripts.  Those are portable.

The interpreters themselves need zero platform optimization for a
bootstrap ("stage3 install").  They can target generic x386 and ppc etc.
 Just like any commercial software vendor.

Following which users could recompile on their targets, if desired or
required.  Exotic targets could just do it the old way.


> That particular blocker isn't exactly hard to get around.

For you...but it cost me half a day, then defeat.  Please consider that
QA feedback.  The FAQs failed.  I don't even want to know how to fix
this.  To me, the fix is to embed the interpreters.


> The stage tarballs are *never* ~arch. We have talked about doing automated 
> builds, but we're not releasing them for public consumption, since there will be 
> absolutely *zero* QA done on them.

Of course not; I didn't mean you replace official releases with ~arch! 
I only meant that, like Debian et al, there be regular tarballs for
eager testers.  That seems more sensible that starting testing from a
year-old tarball.  Some others might test in kexec/vmware/chroot or
whatever.

> why exactly do PPC boxes need the newest everything?

Sigh.  I'd rather not go there.  Suffice to say ~arch is why we're here.
 The current ~arch is turning into 2008.0 anyway, so that's what where
we can help QA Gentoo.

Thanks again.
-- 
  
  davecode@...

-- 
http://www.fastmail.fm - A fast, anti-spam email service.

-- 
gentoo-releng@g.o mailing list