[Prev][Next][Index][Thread]
Re: Students/techniques
-
To: info-dylan@ai.mit.edu
-
Subject: Re: Students/techniques
-
From: ejr@lotus.cs.berkeley.edu (Edward Jason Riedy)
-
Date: Fri, 9 Jun 2000 17:00:02 -0400 (EDT)
-
Organization: University of California, Berkeley
-
References: <200006060812.EAA03005@life.ai.mit.edu>
-
Xref: traf.lcs.mit.edu comp.lang.dylan:12148
And Gabor Greif writes:
-
- The only area where C++ really shines are templates, since the template
- selection rules contain type based lookup and incorporate a prolog-like
- sublanguage that can be used for metaprogramming.
A month or so of debugging code that uses these techniques will
cure you of any admiration. They are evil.
This use of C++ templates is just another pseudo-macro system,
although it is wrapped in fair syntactic sugar for its intended
use. I admit that resolving based of known, compile-time types
is interesting, but I prefer that to be hidden in a compiler for
a dynamically typed language...
I know more of Scheme macros than of Dylan macros, so this may
be trivial: Can the Dylan macro system access any type
information? In other words, can a macro expand different ways
depending on how much type information is known during compilation?
Jason
Follow-Ups: