[Oneiric-Foundations-Topic]Python Goals

[Scott Kitterman]

On Friday, April 01, 2011 12:58:37 PM Barry Warsaw wrote:
> On Apr 01, 2011, at 12:50 AM, Scott Kitterman wrote:
> >For Python, I'd like to see us drop 2.6 and make 2.7 the default during
> >the toolchain setup (and before the first autosync run).  This will
> >(obviously) need to be agreed befor UDS.
> 
> +1
> 
> >For Python 3, we should drop python3.1 at the same time.
> 
> +1
> 
> >For a developmental goal during the cycle I think we should shoot for at
> >least experimental support for Python3 as the Python that is shipped on
> >the various Ubuntu ISOs.  There is a lot of upstream third party support
> >starting to appear and the packaging infrastructure and policy for
> >Python3 is maturing rapidly.  At the UDS for Lucid we set a nominal goal
> >of having Python3 desktops for the following LTS.  If we're going to have
> >a chance of accomplishing this at an acceptable level of risk for an LTS
> >cycle we need to be close for oneric.
> 
> Agreed.  Can you elaborate on what "experimental support for Python3 as the
> Python that is shipped on the various Ubuntu ISOs" means to you?  Does that
> mean no Python 2.7 on the ISO?  Also, by "experimental" do you mean having
> a process for creating alternative CDs that have only Python 3.2 but not
> on the standard daily CDs?
> 
> -Barry

There is a lot of Python code in the Ubuntu insfrastructure.  I'm not sure 
exactly what I meant by that, but here's an example:

Ubiquity is written in Python.  It's a reasonably complex program that is non-
trivial to maintain and improve.  It's also mission critical for Ubuntu.  I 
would be really suprised if it was fully ported with no regressions in one 
cycle.  In this case, I think "experimental support" would be a python3 branch 
that ~works, but may not be fully tested/have issues/or not be at feature 
parity so we wouldn't want to switch to it in the oneiric cycle.

The goal would be to have it be mature enough during oneiric that in the "P" 
cycle we could switch to it early and have it land ~smoothly for the LTS.

I know there are others.  

My impression is that most upstreams for core desktop packages support 
Python3.  Mostly what we lack is packaging changes to support it.  My 
expectation is that most of the challenge around a Python3 desktop in "P" will 
be around more peripheral modules/extensions and custom Ubuntu code.

That shouldn't preclude shipping some Python3 stuff in oneiric if it's ready 
and we've got room on the relevant image.

Does that help?

Scott K