On Fri, Mar 2, 2012 at 2:08 PM, Éric Araujo <merwok@netwok.org> wrote:
I can’t read Guido’s mind, but I think that here he pronounced somewhat quickly because he was convinced by the arguments in the PEP, while choosing to ignore the problems therein, knowing that they could be fixed later.
It's also the case that this particular point has been the subject of debate for a *long* time. When the decision was first made to offer the unicode_literals future import, one of the other contenders was just to allow the u/U prefix in Python 3 and not worry about it, and while the "purity" side carried the day back then, it was a close run thing. While that approach turned out to work pretty well for many users that didn't use unicode literals all that much, the web community understandably feel like they're being actively *punished* for doing Unicode right in Python 2. Consider: an application that uses 8-bit strings everywhere and blows up on non-ASCII data in Python 2 has at least a fighting chance to run unmodified *and* handle Unicode properly on Python 3. Because unicode literals are gone, a Unicode-aware Python 2 application currently has *no* chance to run unmodified on Python 3. So even though the PEP doesn't currently do a good job of *presenting* that history to new readers, it's still very much a factor in the decision making process. Accordingly, I'd like to ask folks not to stress too much about the precise wording until I get a chance to update it over the weekend :) Cheers, Nick. -- Nick Coghlan | ncoghlan@gmail.com | Brisbane, Australia