[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Let's make a plan for python3.0 in Fedora 10+

Toshio Kuratomi wrote:
Matej Cepl wrote:
On 2008-06-09, 16:56 GMT, Toshio Kuratomi wrote:
1) Fedora 10 ships with the latest python-2.5.x.
2) The Fedora 11 development tree switches to python-2.6.x soon after=20
F10 is released.

Well, aren't we early enough in the F10 Rawhide to do it now instead of waiting six months?

It's Geppetto's call but one drawback of switching now is it could mean shipping a non-release python-2.6 in F10 if the python release date slips. Here's python.org's schedule:

    Aug 06 2008: Python 2.6rc1 and 3.0rc1 planned
    Aug 20 2008: Python 2.6rc2 and 3.0rc2 planned
    Sep 03 2008: Python 2.6 and 3.0 final

Here's our tentative schedule:
    Aug 19 2008: Feature Freeze
    Sep 30 2008: Final devel freeze
    Oct 10 2008: Preview release
    Oct 28 2008: Final release

Another drawback is that since we have so many programs that we code which are dependent on python, this would mean we'd have less time to test that yum or anaconda, for instance, worked with all the new python code before the release. ie: More moving parts later in the cycle.

The pro sides would be:
1) We could include code that is written to run on 3.x/2.6's 3.x compat layer. (I haven't run across any code that only targets 3.x yet.)

2) We and our users can get started on ports of code and experimenting with 3.x/2.6 features sooner. (Not sure how helpful this is for us as we're not likely to want to port until, at least, the py-2.6 release candidate (otherwise something could still change on us.) For users, it does mean that they'll be able to use F10 as a base for porting their apps.)


As a python programmer, i would love to play with new python features, and hopefully it will make life easier for the current unicode nightmare in python when working with i18n, but as a yum contributor, i also see the potential upgrade nightmare a switch to a non backwards compatibility version of python and for anaconda there will also be a lot of problems and all the other python stuff, it will need a lot of work to make the code work with python 3.0, without any benefit than running the latest & greatest, over time there will befits (i hpoe ). The only safe way i can see is to have to 2 python stacks 2.5 & 3.0 for a while, i know it sucks, but in this case, it will make the transition much smoother without breaking every python program out there.

SO my proposal will be:

python 2.5 (primary)
python 3.0 (secondary)

python 3.0 (primary)
python 2.5 (secondary)

python 3.0 (primary)

Python 2.6 will only introduce an extra step in transformation 2.5 -> 2.6 -> 3.0, so it only fit if we what a very slow transformation into 3.0 and want to run 2.6 for a 2-4 fedora releases, before switching to 3.0.

Let me now know if it sound totally insane :)



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]