Jason (jcreed) wrote,
Jason
jcreed
For some reason I found myself meditating on a rather PTS-y seeming representation of canonical-forms LF, where types and kinds are somewhat blurred together. I know one can handle rather general predicative polymorphism by eta-expanding on the fly during type substitution, but things are just stunningly pretty if type variables are only allowed to stand for base types or else type operators that eventually return base types. Then no eta-expansion is required at all, and there are various pleasing collapses between concepts at the term and type levels, and between the type and kind levels.
Tags: lf, 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 

    Help
  • 0 comments