On Jul 29, 2014, at 10:13 AM, Aurelien Bompard wrote:

I'm interested in helping port MM3 to SQLAlchemy, I've had some experience with it in the past. But first I'll add some tests to my patch, then I'll port KittyStore to SQLAlchemy, so I'll make less mistakes porting MM3, which is a significantly bigger piece.

Sounds great. Help is greatly appreciated! I have an unfinished branch somewhere which I'll try to resurrect if there's anything useful there. I think this will be a deep change, especially since there are ties to the testing infrastructure, so it might be best to collaborate on it. I'm quite happy of course to do that with you!

Do you think we can release MM3 with Storm? Should we wait for the SQLAlchemy port? It's not in the TODO list on http://wiki.list.org/display/DEV/Mailman+3.0

I'm becoming increasingly concerned about our dependence on Storm, and not just because of the lack of Python 3 support[*]. While I like the simple and transparent model, I think the problems we're having with both SQLite and PostgreSQL point to the need to switch. If we don't have a reliable database layer, how can we possibly release?

Opinions very much welcome. If there's consensus that the switch is needed, I'll add it to the wiki page above and start spending time on it.

Cheers, -Barry

[*] Don't worry, porting Mailman to Python 3 is *not* on the table for 3.0.