Re: GNOME goal for 3.8: Python 3 -- impact for pygobject itself
- From: Colin Walters <walters verbum org>
- To: Martin Pitt <martin pitt ubuntu com>
- Cc: desktop-devel-list gnome org
- Subject: Re: GNOME goal for 3.8: Python 3 -- impact for pygobject itself
- Date: Mon, 05 Nov 2012 08:04:50 -0500
On Mon, 2012-11-05 at 10:28 +0100, Martin Pitt wrote:
> but if we now want to port GNOME to py3, I guess this should be
> rearranged.
>
> Should I just change the default to "python3", so that jhbuild will
> from now on build for py3 unless you explicitly configure for python2?
> That would go along with that goal, but would break modules which
> expect pygobject for py2.
Short answer: add a new pygobject-py2 module to jhbuild that points to
the same git repository, and just has a different --enable-python2
argument. (Neither jhbuild nor the gnome-ostree build system support
passing environment variables to control the build; the autoconf option
mechanism should be preferred here)
Longer answer: The Python3 porting should really be focused along
"runtime/devel" lines first. What I mean by this is that ideally for a
system without development tools, GNOME is only pulling in one version
of Python. I think Ubuntu achieved this in their most recent release?
But once you install devel tools, no one is going to care too much if
you have to install two versions of Python (e.g.
gobject-introspection-devel is on 2.x and caribou is on 3.x). We
already depend on Perl, and Ruby to build webkit, and just in general
a wide array of stuff. Keeping the runtime system lean though does
matter.
For pygobject specifically, it's highly likely that it has external
python2 consumers and will for a long time. So I'd just make a new
module in jhbuild. Distributions will package it and make it a
dependency of things like Anaconda. But if you want, we could demote
it to gnome-world or something.
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]