Use Shunting-Yard #22
Description
I believe rewriting the parser to use the shunting-yard algorithm is something we should look into.
The current way we parse input can do so reasonably well, but it's lacking in several areas (see #21, #17, #2, and possibly #4). Because of this, I think beefing up the parser and using something like shunting-yard would make things easier to follow and would provide us with many improvements pretty much immediately (e.g. unary operators).
I'd imagine this would warrant a major release and may result in a slight performance hit, but I think it's worth it.