On Sep 14, Jed Davis wrote:
> On Fri, Sep 11, 2009 at 01:10:01AM -0400, Eli Barzilay wrote:
> >
> > But if you consider macros with a module system: it allows you to
> > keep the abstraction barrier by letting you export a macro but not
> > utility functions and values that the macro uses, and it allows
> > you to define your own language (for example, a language with a
> > new kind of `if') and still be able to import macros that are
> > written in the usual Scheme language.
>
> Both of these use cases work just fine with Common Lisp's defmacro,
> because CL symbols are interned with respect to a package. A macro (or
> anything else) whose definition contains a given symbol will continue to
> refer to that same symbol, regardless of what might happen to different
> symbols that happen to share its name.
Yes, they "work" for some *very* weak definition of "work".
Specifically about the two cases:
1. In this case CL just throws its hands in the air and provides
absolutely no abstraction barriers. Any function in any package is
*always* available to *any* code. (And yes, I know how CL
advocates defend this, and I know the "thieves" quote. It's still
broken.) A very simple example in PLT:
#lang scheme
(provide foo)
(define-syntax-rule (foo) (bar))
(define (bar) 1)
Actually, PLT takes this another step further and if you try to:
> (require "the-above-file.ss")
> (eval (cadr (syntax->list (expand '(foo)))))
the-above-file.ss:3:32: compile: access from an uncertified
context to unexported variable from module:
"/tmp/the-above-file.ss" in: bar
2. As for defining new languages based on given ones, like a new
definition of `if' -- this is also "possible" with CL, but you're
left with doing the name management yourself. In this case it
would roughly be declaring that you're shadowing `if', then
referring to the CL version with a ::-qualified name when you need
it. Your new language is similarly suffering from all the usual CL
constructs being available, so people can still break your
carefully-laid abstractions. Again, a quick PLT example:
#lang scheme
;; provides a language like `scheme' except for `if'
(provide (except-out (all-from-out scheme) if)
(rename-out [if* if]))
(define-syntax-rule (if* cond then else)
(if (equal? cond 0) then else))
> One might almost be tempted to declare that defmacro is "perfect",
Yes. Almost.
IMO, the conceptual jump from `defmacro' to hygienic macros is similar
in magnitude to the jump from CPP to `defmacro'.
--
((lambda (x) (x x)) (lambda (x) (x x))) Eli Barzilay:
http://barzilay.org/ Maze is Life!
_______________________________________________
r6rs-discuss mailing list
[email protected]
http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss
