Re: freezing python packages until python-2.3 becomes the default in testing

[Matthias Klose <doko@cs.tu-berlin.de>]

Josselin Mouette writes:
> Le ven 10/10/2003 =E0 14:02, Ron a =E9crit :
> > On Thu, Oct 09, 2003 at 08:03:32PM -0400, Derrick 'dman' Hudson wrote:
> > > The libwxgtk2.4-python in testing depends on python (2.2).
> >=20
> > Ahh, ok.  This is the piece of the vicious cycle I was overlooking.
> >=20
> > I've just sent some mail to Robin about getting wxPy to rewrite all
> > its bang paths at build time to pythonX.X instead of python.  If he
> > can do that I'll change the package to depend on pythonX.X instead
> > of python(X.X), and we can get one more package out of this nasty
> > bottleneck altogether.

note that this does not help in the current situation. If packages
depending on wxpython and on python, are waiting for the transition as
well, then the use of pythonX.Y in wxpython instead of python won't
help.

> This is a good idea for widely used python packages, but only if a dummy
> package relative to the default python version is made available
> simultaneously. Thus, such a solution doesn't help with testing
> transition.
> 
> What would really help for this kind of stuff would be to get rid of
> strict dependencies for architecture independent modules.

yes, which requires some more support ... let's target this for
sarge+1.

	Matthias