Re: Proposed Modules, My Take

[Mark McLoughlin <markmc redhat com>]

On Wed, 2005-01-19 at 21:43 +0100, Murray Cumming wrote:
> On Wed, 2005-01-19 at 15:14 -0500, Sean Middleditch wrote:
> > On Wed, 2005-01-19 at 21:07 +0100, Murray Cumming wrote:
> > > On Wed, 2005-01-19 at 14:58 -0500, Sean Middleditch wrote:
> > > > New versions must not break ABI.  While PyGTK itself does not break, it
> > > > does heavily depend on the ABI of Python and that *does* break.  It's
> > > > like saying it's alright if GTK+ is stable but glib can be unstable.
> > > > Unless each release of GNOME has a standard Python version that must be
> > > > available with that release, PyGTK would be breaking the spirit of the
> > > > rules, if not the exact letter of the rules.
> > > 
> > > The application must specify the version of python. If not then the
> > > application is choosing not to use a stable ABI.
> > 
> > Right.  The problem is that the application has no idea _which_ version
> > of Python to specify.
> 
> The application developer knows by 
> a) Reading the pygtk documentation, which says what versions of Python
> work with pygtk. 
> b) Seeing what works.

	So, I'm using pygtk for the first time at the moment and I have to
admit that this whole conversation is just a bit bewildering to me...

	You want me to put 

  #!/usr/bin/python2.3

	in my scripts rather than

  #!/usr/bin/env python

	But what about the python built as part of a jhbuild bootstrap that I
really want to use? What we really want then each apps configure.in to
find the path to the correct version of python and stick that at the top
of each script?

	Hmm. If I think really hard or you explain it with little words, I can
probably understand the rationale for this and change my scripts
accordingly ... but surely that's going to be a huge uphill struggle to
convince all developers using pygtk? Something must have gone wrong here
somewhere, right?

Cheers,
Mark.