The Haskell translator might be able to work on separate files in parallel. We should investigate if that is the case, because on machines with a bunch of cores this could give you a nice (> 10 times) interactive speed up.

The main loop that goes through all the files is in python. Not clear to me if that lends itself to easy parallelisation. Probably not, tbh (big interpreter lock and all that). But then again, maybe one can fork many small pythons of the big python. Who knows..