[oe] Supported Python version for OE?

[Joshua Lock]

On Tue, 2010-05-11 at 09:23 +0100, Martyn Welch wrote:
> Joshua Lock wrote:
> > Hi,
> >
> > A question, (perhaps for the TSC?):
> >
> > 	"What's the minimum Python version we want to support in OE?"
> >
> > According to the wiki we support Python 2.4 and above but I wonder if
> > people have any thoughts with regards to bumping it?
> >   
> 
> I'd suggest that the better question to ask is:
> 
>      "Which versions of which distros do we currently intend OE to work on?"

That's a good way to re-phrase it.

> 
> Given that the revisions of Python for the following distributions are
> as follows:
> 
> Ubuntu 10.04 LTS - 2.6.5
> Ubuntu 8.04 LTS - 2.5.2
> 
> Debian lenny (stable) - 2.5.2
> Debian squeeze (testing) - 2.5.3
> Debian sid (unstable) - 2.5.4
> Debian etch (oldstable) - 2.4.4
> 
> Fedora 12 - 2.6.2
> Fedora 11 - 2.6
> Fedora 10 - 2.5.2
> Fedora 9 - 2.5.1
> Fedora 8 - 2.5.1
> Fedora 7 - 2.5
> Fedora 6 - 2.4.3
> 
> RHEL6 (beta) - 2.6.2
> RHEL5 - 2.4.3
> 
> OpenSUSE 11.2 - 2.6.2
> OpenSUSE 11.1 - 2.6.0
> OpenSUSE 11.0 - 2.5.2
> 
> This would suggest that using 2.5 features should be ok for the majority
> of people. My one area of concern would be those using RHEL. RHEL 6
> isn't out yet and v.5 uses 2.4.3 - this wouldn't impact me, so I'm not
> overly fussed.

Hmm, thanks for the data. I'm not overly fussed either but suspect that
supporting RHEL5 is a desire for at least a while after RHEL6 comes out?

> 
> > The reason I ask is because I had a user contact me about using Python
> > 2.5 features (str.partition) in relocatable.bbclass, I hadn't even
> > noticed this and seems like not many others have but it's clearly
> > affecting at least one person.
> >
> > I have a pretty trivial (if ugly) patch to work around this, but it
> > raised an interesting question so I thought I'd ask that before sending
> > the patch.
> >   
> 
> The only other question I can think of is "is there an advantage to
> using the Python 2.5 features?".

Only that it's less code and that the code is more thoroughly tested.
The patch was just what prompted me to question a Python 2.4 dependency.

Thanks,
Joshua
-- 
Joshua Lock
        Intel Open Source Technology Centre