Subject: Re: tranformations and node visiting
From: Jonathan Aldrich (jonal@cs.washington.edu)
Date: Sat Mar 03 2001 - 22:39:30 PST
> Maybe I'm misunderstanding you, but I don't see why this would be
> especially ugly. All you have to do is wrap your visit method in a
> closure:
Beauty is in the eye of the beholder, but this beholder thinks it ain't
beautiful. First of all, what you have for visit methods is probably not
> method my_visit_order(@foo):collection[foo] ...
but rather an internal iterator method that takes a closure:
method my_visit(root@:foo, &(:foo):void):void { ... }
This makes the usage scenario much more hairy:
caching_visit(root, &(x:&(:foo):void) {
my_visit(root, x);
}, &(y:foo) {
-- do something
});
Compare this to the incredibly clean scenario:
> caching_visit(root, my_visit, &(x:foo){
> -- do something
> });
The point of language design is not what you can do (everything's
turing-complete), but how clean and error-free it is to express things,
and the style of programming the language encourages. In the first case,
the closure of a closure thing is very confusing and in my opinion
error-prone (I tried implementing the closure version, for example, and it
made my head spin).
Second, because this is such a pain, it discourages curried-style
functions that take other functions as arguments. In Cecil, we use
user-defined control structures a lot, but I don't think we typically use
higher-order functions that take method arguments like caching_visit. If
we really had first-class multimethods, maybe we'd all program in a more
higher-order way.
Self disclosure: ugly language defects are a pet peeve of mine, so take my
complaints with the appropriate grain of salt. Fortunately Cecil has very
few ugly defects (the other one that comes to mind is the && for escaping
closures). I also realize that avoiding such defects takes effort, and
it's not always productive research to chase them all down.
Jonathan :-)
_______________________________________________
Cecil mailing list
Cecil@cs.washington.edu
http://majordomo.cs.washington.edu/mailman/listinfo/cecil
This archive was generated by hypermail 2b25 : Sat Mar 03 2001 - 22:40:04 PST