On 14 May 2018 at 12:34, Chris Barker via Python-Dev <python-dev@python.org>

> On Sat, May 12, 2018 at 8:14 AM, Skip Montanaro <skip.montan...@gmail.com>
> wrote:
>> > I have found 2to3 conversion to be remarkably easy and painless.
>> > And the whole Unicode thing is much easier.
> Another point here:
> between 3.0 and 3.6 (.5?) -- py3 grew a lot of minor features that made it
> easier to write py2/py3 compatible code. u"string", b'bytes %i' %
> something -- and when where the various __future__ imports made available?
> If these had been in place in 3.0, the whole process would have been
> easier :-(

The __future__ imports were already there in 2.6/3.0.

The other ones weren't there initially because we didn't know which things
we were tempted to add back because they were actually useful, and which
ones we just thought we wanted because we were used to the way the Python 2
text model worked (or failed to work, as the case may be). (The build time
source code translation step was also far less effective than we hoped it
was going to be, since we completely failed to account for the problem of
mapping tracebacks for converted code back to the original pre-translation


Nick Coghlan   |   ncogh...@gmail.com   |   Brisbane, Australia
Python-Dev mailing list

Reply via email to