Den 2012-02-20 11:59:30 skrev Johan Bergström <bugs@...>:
> On Monday, 20 February 2012 at 6:44 PM, Sjujskij Nikolaj wrote:
>> Den 2012-02-20 09:04:45 skrev Johan Bergström <bugs@...
>> > Good day all,
>> > with Python 2.4 being removed and all (anyone seen complaints about
>> > this, btw?),
>> > I'd like to discuss the removal of Python 2.5.
>> > Although 2.5 was one of those versions that started to be useful, I
>> > really see no
>> > reason to just keep it around "just because".
>> > Did a quick glance in the tree and couldn't find a package that only
>> > depended
>> > on python:2.5 specifically. Please correct me if I'm wrong.
>> > The question is therefore: why keep python 2.5 in tree?
>> There're quite a few people developing for Python 2.5 (for other target
>> platforms) using Gentoo. Just as I know one guy who programs for RedHat
>> (with Python 2.4) using Gentoo ~amd64.
>> Though I'm not developer, I hold that there's no call to remove old
>> versions from tree: declare them unsupported, or mask, but don't remove
>> until it's too burdensome.
> This is one of the arguments also used for 2.4 (as you also state), which
> now is gone. I would rather put similar ebuilds in a python overlay.
That'd be another solution, but in that case our devuser would have to
deal with all the other Python-related packages in python-overlay, mostly
of bleeding-edge persuasion, of fiddle with symlinks.
And Python 2.4 did not make way into python overlay anyway, and is nowhere
to be found nowadays (except gentoo-x86 cvs).
> The way I see it, we have these "few people developing" vs us python
> dev's, testing and building packages on a daily basis. 2.4 was starting
> to be a real burden (I've seen 30+ package silently disregard 2.4) in
> 2011, and we'll most likely see the same thing happen for 2.5.
Wouldn't solution "declare them unsupported and mask" deal with that kind
of thing? toolchain-herd still keeps GCC 2.95 in tree and it was
hard-masked even before I started using Gentoo. I seriously doubt anybody
really *supports* it, and compiling anything recent with 2.95 is a tough
> It might not be time to punt it yet, but it doesn't hurt to discuss
> arguments until time's due.