type inference - terminology

• To: rwh@cs.cmu.edu
• Subject: type inference - terminology
• From: meyer@theory.LCS.MIT.EDU (Albert R. Meyer)
• Date: Fri, 10 Nov 89 17:51:34 EST
• Cc: types Reply-To: Robert Harper <rwh@cs.cmu.edu> Date: Fri, 10 Nov 89 17:33:58 EST From: Robert Harper <Robert.Harper@proof.ergo.cs.cmu.edu>
• In-Reply-To: Robert Harper's message of Fri, 10 Nov 89 17:33:58 EST <26755.626740438@PROOF.ERGO.CS.CMU.EDU>

   Yes, there has been some confusion about the two.  The accepted terminology in
   the lambda calculus literature for what you have dubbed "type inference" is
   "type assignment".  Some work in the literature has tended to suggest that ML
   ought to be regarded as a system of type assignment.  But as Mitchell and I
   have argued, we really ought to think of it as reconstructing information
   omitted from what are "officially" explicitly-typed terms.  However, rather
   than introduce a new phrase, I'd prefer to retain "type inference" for what ML
   does, but to stress the reconstructive point of view.

   Bob Harper

-------
But ``type assignment'' is also used as a synonym for ``type
environment'', namely to designate the thing which assigns
type-expressions to identifiers.  I'll stick to ``type reconstruction.''

Regards, A.

• Prev: type inference - terminology
• Next: Call-by-value (eager) vs. Call-by-name (lazy)
• Index(es):
  • Main
  • Thread