Jason (jcreed) wrote,
Jason
jcreed

Woo, my type-checker implementation is actually sort of working. At least it does one example correctly with only the barest, stupidest unification heuristics. I know unification for this problem is decidable, but the complete algorithm is nasty. I bet the stupid heuristics will handle most of the cases I care about, and also they're probably more efficient to try first than going to full linear diophantine solving.

Next down the road is type reconstruction and coverage checking...
Tags: work
Subscribe
  • Post a new comment

    Error

    Anonymous comments are disabled in this journal

    default userpic

    Your reply will be screened

    Your IP address will be recorded 

  • 2 comments