Jason (jcreed) wrote,

Trying to clean up code somewhat. I realize how much of the gritty detail of ML's module system I've forgotten since 212/312. I understand there are many issues to be had with Randy Pausch, but his praising of accidentally-learning-substantial-things-during-"play" in a speech of his I heard once is at least one thing I vigorously agree with.

There are some weird issues with how sml/nj vs. mlton (where mlton is merely being standard-conformant) deal with withtype substitutions. I can think of a few ways of getting around it. At the moment I am leaning towards parametrizing many of the types involved over the type of values of the underlying object language, and then "tying the knot" when defining exactly that (therefore recursive) type of values.

  • Post a new comment


    Anonymous comments are disabled in this journal

    default userpic

    Your reply will be screened

    Your IP address will be recorded