The looming Python 3(000) monster

[James Antill]

On Fri, 2008-12-05 at 21:21 +0100, Kevin Kofler wrote:
> James Antill wrote:
> >  We also have _much less_ resources dedicated to python than we do GCC.
> >  My guess is that if a second python is proposed for py3k, it'll get
> > voted down again.
> 
> I think migrating without shipping parallel versions of Python at any point
> is a completely unrealistic dream.

 And yet everyone who's been involved in trying to do that says it's a
giant mess, which they have no wish to repeat. And it has been voted
down when we had only a single significant application that needed an
old version, which would be _much easier_ to manage than your vision of
the future where random amounts of applications have migrated and random
amounts haven't.
 Of course, you could be right ... I would just not bet on that
happening.

>  Just look at how many packages still
> depend on qt3 and kdelibs3, and how there are still packages depending on
> gtk+-1.2 even when plans for GTK+ 3 are already ongoing. We have no reason
> to believe Python will be any different.

 My hope is that as future iterations of both the 2.* and 3.* versions
come out (and the applications using them), the porting burden dwindles
to something "easy enough" for a flag day.

 And as I said before, my current guess is that it'll be much more like
Apache-httpd's 2.* migration and thus. RHEL-7 (yes, seven) will have a
single python that is from the 2.* version stream.
 Yes, py3k was released yesterday. No, that doesn't mean we need to move
to it by next week, month or year.

 Also as has been said before, a lot of the timeline depends on the
various upstreams ... we may be forced to make certain less desirable
choices depending on what happens.

-- 
James Antill <james at fedoraproject.org>
Fedora