Re: [gentoo-dev] Python 3.1: Stabilization and news item

[Brian Harring]

On Fri, Mar 26, 2010 at 05:04:28PM +0100, Arfrever Frehtes Taifersar Arahesis wrote:
> 2010-03-26 16:43:57 Brian Harring napisał(a):
> > On Fri, Mar 26, 2010 at 03:22:52PM +0100, Arfrever Frehtes Taifersar Arahesis wrote:
> > > Arfrever Frehtes Taifersar Arahesis 2010-03-25 20:05:17 napisał(a):
> > > > 2010-03-25 19:34:24 Roy Bamford napisał(a):
> > > > > The case where Python-3 cannot be used as the default Python is
> > > > > transitory (it may be a long time).
> > > >
> > > > Gentoo Python Project will soon start supporting setting Python 3 as main
> > > > active version of Python. Currently about 57% of our packages from dev-python
> > > > category are prepared.
> > >
> > > My script was wrong. More correct data:
> > > About 55% of packages in dev-python category belonging to python herd are prepared.
> > > 100% of packages in net-zope category belonging to python herd are prepared.
> > > About 60% of packages belonging to python herd are prepared.
> > > About 47% of packages in dev-python category not belonging to python herd are prepared.
> > > About 13% of packages not belonging to python herd are prepared.
> > > About 34% of all packages depending on Python are prepared.
> >
> > I get the feeling your phrasing here is a bit misleading- 'support
> > setting py3k as main active python' implies that the stats above are
> > the # of pkgs in the tree supporting *using* a py3k interpretter.
> >
> > I'm betting you mean "support multi-abi", meaning if you've got py2.6
> > and py3.1, it'll install into py2.6, while avoiding py3k.  Fair bit of
> > a difference.
>
> These numbers include packages which support installation for multiple Python ABIs
> and packages which call python_set_active_version().

Bleh.  So in other words a third of the pkgs that dep on python have 
the minimal basics for dealing w/ py3k landing.  I'd question what 
percentile have proper locked deps also (stating they're py2k only), 
but that's a seperate discussion.

That *still* doesn't answer the question of how many can be *ran* by 
py3k also.

Note in the past when breakages of this sort have been unleashed, the 
percentile of prepared pkgs has been generally a helluva lot higher- 
having 90% prepared is one thing, but y'all aren't at that point and 
you've got 3 weeks (after what, 3 months?) to bring the percentile 
higher then a third?

What's the minimal percentile you're aiming for prior to the 
unmasking?

~harring