2009-03-29 20:02:25 Greg Stein napisał(a):
> On Sun, Mar 29, 2009 at 19:23, Stefan Sperling <stsp_at_elego.de> wrote:
> > On Sun, Mar 29, 2009 at 05:49:43PM +0100, Greg Stein wrote:
> >> I say it complicated things. More methods instead of fewer. Lots of
> >> additional comparisons instead of just using cmp(). Additional logic.
> >> And "Python 3.0 compatibility" for scripts intended for 2.4 doesn't
> >> add any benefit.
> > Won't we eventually have to depend on Python 3.x, when Python 2.x is
> > being phased out in a few years or whenever they're planning to do it?
> > At some point, we'll have to do what Arfrever is doing anyway.
> > If we can do some of the necessary conversion work now, and have
> > someone how is eager to do it, even in a way that preserves compatibility
> > with Python 2.x, I don't think it's bad. Especially because converting
> > to Python 3.x does not seem to be trivial.
> > I'm not talking about particular scripts that might well not be useful
> > anymore in a few years (think change-wc-format.py). We don't need to
> > update every line of our Python code. But I don't think there is anything
> > wrong with the general idea of trying to be Python 3.x-compatible as much
> > as possible, even at the cost of temporary extra complexity in the
> > scripts. Eventually, we can drop the Python 2.x code and things will
> > become simpler again.
> I agree in principle, and Arfrever's work has been a great help for
> our code. He's been making it more consistent, throwing out the
> *really* old crap, and cleaning up some weirdnesses.
> There are some changes (like these past two revs) which just make it
> more complex than a plain 2.x needs to be.
__eq__() and __ne__() were already used in subversion/tests/cmdline/svntest/wc.py
before these changes, so for consistency they can also be used in
subversion/tests/cmdline/svntest/tree.py and subversion/tests/cmdline/svntest/verify.py.
> Note that 2.x is going to be around for *years*. Five? Ten? There is
> not going to be a string impetus to move to 3.x for a long while.
Please note that the 'python-3-compatibility' branch still supports (and will
support in foreseeable future) Python 2.6.
> AndPython 3 is going to need to be installed on a lot more platforms as a
> base dependency for it to be worthwhile, especially since 2.x will
> contine to be installed. Even the recent "Unladen Swallow" project is
> based off of 2.x. So five years of "little pain" adds up to "lots of
> pain". When we decide to switch, then we can do it in a month and be
> done with it.
Without previous work on supporting Python 3, one month would be rather
Arfrever Frehtes Taifersar Arahesis
Received on 2009-03-29 20:44:52 CEST