On Dec 27, 2014, at 08:21 AM, Andrew Stuart wrote:
If it is released to version 1.0 with support for Python 2 then it will need to support Python 2 for the forseeable future. It’ll be deployed all over the place with Python 2 - when would the Python 2 be removed ever after 1.0? David Murray has made substantial changes to email in Py3 - supporting those in both 2 and 3 could be an open question. A Python 3-only version 1.0 has the longer term gain of never having to write mixed 2/3 code after 1.0.
Exactly. I'm highly motivated to not release the *core* under Python 2 for the reasons you state above. Personally speaking, I've been so immersed in Python 3 for almost all my other projects (both work and play) that continuing to hack in Python 2 seems like I'm taking down my dinner with pointed sticks. ;)
After finishing the port, I think it's impractical to try to have the core support both Python 2 and 3. There's *a lot* of great new stuff in the Python 3 email package, and while I'm still mostly using the legacy API, I did have to use some of the new stuff to do Subject prefix munging correctly. I expect that as time goes on, we'll need to adopt more of the new email APIs in order to fix bugs, add features, and do things more efficiently. That's not even talking about so much of the other great new stdlib stuff available in Python 3.
Cheers, -Barry