Re: [boost] Re: Formal Review: Template Metaprogramming Library (MPL)

From: Mat Marcus (mmarcus_at_[hidden])
Date: 2002-07-25 11:32:37

Dave Abrahams writes
> Those are worse on all counts AFAICT. It's neither an object
> nor a value. How does that help the user?

Maybe it was kind of a stretch, but the reasoning went like 
this. I was thinking of MPL as a programming language with 
functions and values (or objects). The functions happen to be 
struct templates from another language, C++, and the values 
happen to be types and enums from C++. In the MPL language 
functions can be passed and returned as values, but special 
care is needed. So the terms metafunction value or metafunction 
object were meant to indicate metafunction passed/returned as 
an object/value.

Anyway never mind all of this. I don't really like these names 
that much, I was just trying to suggest alternatives since 
nobody else (with the exception of Peter Dimov) seemed to like 
"quoted metafunction". When learning MPL (before your fine 
paper was written) my breakthrough came when I realized the 
analogy with lisp. I still think quoted metafunction is a much 
better name. Perhaps I've beaten this one into the ground 
(again).

 - Mat