On Sun, Nov 6, 2011 at 10:38 AM, Meir Kriheli wrote:
> On Wed, Nov 2, 2011 at 5:48 PM, Ram Rachum wrote:
>> I think that your criticism applies only to print statements which have
>> comma in them, and I think these are the minority.
> For me the problem is that it alters the data passed to print, which is
> undesired (for example print can be overriden in 3, and that may lead to
> unexpected behavior).
> Plus the parenthesis is a minor one for me, and won't contribute much to
> the move to python 3, there are larger fish to fry (Unicode/Bytes,
> imports), and handling them is part of my preferred strategy for 2 to 3
> (see next section).
> As far as I know the `2to3` tool doesn't produce code that's
>> backwards-compatible with Python 2.
> And it doesn't have to. My preferred strategy for 2 and 3 is not making
> sure the same code base works with both branches - difficult and error
> prone and undesirable (later down the road one will be left with cruft to
> clean up).
> A better one is to make sure your code passes 2to3, and during build
> process, if it's python 3, run 2to3 with distribute:
Seems like a good idea.
BTW, anyone tried 3to2 <http://wiki.python.org/moin/3to2>
If that works fine, then at some point one could shift the main development
to 3, and keep 2.x support for a while using the reverse method.