[pkg-discuss] Re: code review request: pkgdepend Python 3 support
- From: Danek Duvall <
- To: Tim Foster <
- Subject: [pkg-discuss] Re: code review request: pkgdepend Python 3 support
- Date: Thu, 8 Aug 2013 15:20:51 -0700
Tim Foster wrote:
> On 07/30/13 06:50 AM, Danek Duvall wrote:
> >There's really not much to see here -- Python 3 support is essentially
> >identical to 2.7 support, except for a few small syntax changes. I've
> >combined the multiple depthlimitedmfXY.py files back into depthlimitedmf.py
> >and it's not too too grotty, though there are some bits which need to be
> >copied. Overall, though, I think it's a win for now.
> > http://ips.java.net/webrev/dhduvall/pkgdepend3/
> This looks good, so long as you've remembered to actually 'hg rm' the
> After this change, given that depthlimitemf.py has to run with both Python 2
> and Python 3, are there any automated checks that we can run during the
> build to ensure we're not reintroducing Python2 syntax nits?
> [ 2to3 maybe? ]
Once Python 3 is available in the WOS, I think that'd be a fine idea,
though I think we should run the dependency tests with python 3, too, just
like we do some with 2.7, and that ought to take care of it.
> >Brian's been testing this on Python 3 code, and it seems to work as well
> >(and as poorly) as it does on Python 2 code, so that seems to be enough.
> That's a shame: I was hoping python3 was going to show some improvement on
> how modulefinder deals with relative imports, but Not This Bug (and I saw
> your update on that earlier)
Yeah; this doesn't seem to be a priority, and I can't blame 'em. I did see
recently, which handles *conditional* imports, but the associated code just
returns a 404, so I can't even check to see what license it's under.
Though at least it validates my idea that it's possible to do such a thing.