On 13 April 2016 at 05:40, Andrew Beyer <beyer andrew gmail com> wrote:
> I don't think 2to3 is going to get you terribly close to anything that
> actually works.
This is true. I'd go so far as to say that for a code base like Meld,
2to3 isn't really helpful at all. While it might 'work', it will do a
*lot* of unnecessary changes and won't actually fix any of the
problems.
> AFAIK, meld code itself is largely going to work ok w/ 2to3, but the
> interactions w/ gobject & gtk bindings probably require quite a bit of
> updating to work properly in py3. (I haven't actually tried it w/
> meld, but do know that another codebase I worked w/ had similar
> issues.) I know Kai has mentioned that this is something he's looked
> at before, but not sure the current status of it...you may want to
> wait for him to chime in before you go too far w/ this.
I have a local branch with basic py3k support, but it's not really in
a pushable state. For a start, I'm confident that a lot of our
filename handling code is just plain wrong, and I haven't had the time
to figure out how to rework things yet. I *will* try to get that
tidied up to a point where I can at least push the branch so that
people can play, but it will still take a while.
For the next release series (I'm hoping to release 3.16 in the coming
weeks) the python 3 port is probably my major goal, along with merging
the OSX build support.
cheers,
Kai