|
|
|
|
|
by git-pull
3393 days ago
|
|
(One topic I see being brought up when it comes to runtimes, python 2 and 3 is C API stuff. It still bubbles up to the top now and then. This isn't related to grumpy specifically, but the FUD I've been witnessing) Maybe I'm far underestimating the amount of custom C API being used in production at places or been hanging out in the wrong places. I just doubt that intention is to kill off anything. Even by unintended side effect. A great deal of the python code I see with C extensions already is Python 3 compatible and even has wheels for them. I think it used to be numpy and libraries that pulled it in as a dependency. We're at the point where compatibility from 2 and 3 is so darn good in libraries we pull in I hardly notice it anymore. [1] I do write Python 2 + 3 compatible code. The differences in the syntax itself are trivial. A compat module [2] will do the trick assuming you have your own custom C extensions, which most don't. (Going off on a bit of a tangent) There has been some deliberation of what has to be done to get around GIL: The other thing is I haven't been convinced of is the idea being thrown around at conferences and on mailing lists that breaking the C API is this world-ending scenario. Yeah I understand the disruption, but what % of the code, assuming you're in the minority of python developers doing custom c extensions, that you can't update some API signatures? [1] https://python3wos.appspot.com/ [2] http://lucumr.pocoo.org/2013/5/21/porting-to-python-3-redux/... |
|