[MLton] more FFI revisions

Henry Cejtin henry.cejtin@sbcglobal.net
Sat, 06 Aug 2005 14:17:18 -0500


My argument isn't to make it easy to program in C from ML, but to make it
so that the syntax follows the `least surprise' principal.  Also the general
idea of being able to name things, which can't be done if it is more than one
type separated by commas.

Re doing C in ML, I think that the current method of requiring an annotation
really seems to do the job.