David Barnett

Yep, behavior does get a little complicated if it's a core feature and I do suspect it's going to be just Vroom running into compatibility issues, so I was leaning...

That's not per-runner, but similar yes. I'd either need a user override to say "in this buffer, use this set of strategies instead" or a separate mapping to say "when...

Looks like https://code.google.com/p/vim/issues/detail?id=183. Adding `set t_u7=` at the top of my vimrc fixed a similar issue for me (https://code.google.com/p/vim/issues/detail?id=390).

Would more hands on this help, or too involved for that?

I ran into pretty annoying issues where I ironically lost *static* type information whenever I tried to enable *runtime* type checking with the `@pytypes.typechecked` decorator and things like that. Would...

Yeah, I think that matches what I had in mind. These would just be changes to add annotations in typechecker.py?

Okay, I figured out the gist of the changes, but pytype found several type issues in the same file, and I'm having trouble understanding from [PEP 561](https://www.python.org/dev/peps/pep-0561/) if adding a...

Yeah, uninitialized None errors are easily resolved with `assert`, just for some of them I didn't see a reason it was a safe assumption and will probably include a comment...

I was also wondering if we could change the module name case for `@typechecked` to not return anything. Passing a string for that will sometimes look up and return the...

Nah, I wasn't attached to the None idea. Actually I agree making it string→string / module→module unconditionally is intuitive and way simpler for the typing. Seems like you're on board...