Jason (jcreed) wrote,
Jason
jcreed

By popular request, the haircut:





Not really all that exciting, is it?

Anyway. Got a lot more thesis hacking done today. Digging into the algorithmic equality stuff. I think it really is necessary to keep track of relevance labels in contexts, but I don't think I'm going to need to check algorithmic self-equality at the irrelevant application structural equality rule, for instance. Sadly, I may not have enough time to fit unification results into the thesis.
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 

  • 11 comments