On Feb 28, 2014, at 21:53, ian o email@example.com wrote:
Agreed. But in the real world, an example of a major dependency requiring python2, web2py, states they will be motivated to python3 when the users of their framework move.
This is an interesting choice for an example.
web2py isn't just a library, it's a framework. Its main purpose is to run your code in the right situations. Which is exactly the kind of thing your bridge cannot handle, for the reasons you already gave. And I don't think this is a coincidence. The same thing is true of many of the most prominent packages that haven't been ported yet--Twisted, Paste, gevent, supervisor, etc.
On top of that, web2py is all about passing Internet strings around. The same thing that makes it difficult to port--the bytes/Unicode distinction--would make it just as hard to use through the bridge.
Anyway, if you're interesting in pursuing this, obviously nobody is going to stop you from doing it and putting it on PyPI. And if there's as much interest in the idea as you think, you should start getting outside contributions as soon as you have a workable beta (or you might even be able to get funded development before that). And if it became popular enough, you could come back to the core dev team and say, "Like it or not, there are tons of major projects relying on my bridge, and the consequences aren't nearly as gloomy as you suspected, so what about putting it in the stdlib?"