Re: Why Prolog and CBV?

Date: Mon, 18 Nov 91 09:18:52 PST
To: Robert Harper <rwh@PROOF.ERGO.CS.CMU.EDU>

   I'm not sure I understand the relevance of whether CBV or CBN is a complete
   strategy for finding normal forms since no programming language that I know of
   takes as values the normal forms.

For basic types (int, nat, bool, ...), the normal forms and values
coincide. Given a closed term of PCF of type nat (for example),
leftmost evaluation will produce a numeral if any evaluation order
will, CBV may not.