[PRL] values and expressions

[Richard Cobbe]

On Sun, Oct 23, 2005 at 12:02:57PM -0400, Dave Herman wrote:
> >I'd have to say no, think of any language where values include
> >references.  I've never seen a language that included them in
> >expressions.
> 
> Simpler example: closures.

That gets into tricky distinctions between the actual values and how you
represent them.  When I write a substitution-and-context-based
operational semantics, I just identify a closed lambda term with its
closure and move on.  (It has to be closed, because by the time I get
around to dealing with it during evaluation, I should have already
substituted for all its free variables.)

Clearly, though, this doesn't work at all for a CEKS machine, where you
do need a representation for closures that is distinct from a lambda
expression.

Richard