Thanks a lot guys. As Eli pointed out, the main objective here is to use the special identifiers as macros, so I'll take a look at his suggestion.
Basically, @anything is used as a macro for dsl-style variable assignments: ie, @x = something, @y = something Here '@' is the macro, and the rest of it is parsed out and stores 'x' or 'y' in local scope. Similarly, I'd like a macro $something, $something-else, where '$' is the macro for a kind of named action that has special properties. I could make these work as (@ x = 2) or ($ call-my-action), but would prefer the token to be self-contained as in (@x = 2) or ($do-something). Scott. On Sun, Jan 27, 2013 at 1:08 AM, Eli Barzilay <[email protected]> wrote: > On Wednesday, Scott Klarenbach wrote: > > Is it possible to have a macro bound to an identifier with wildcards > > in it? > > > > So, for example, a macro (@*) that would be used for (@x) and (@y), > > where the symbols x and y are available during expansion. > > > > Or is this something that would have to be done at the reader level? > > ie, (@ ...) is the macro, and the reader turns @x and @y into (@ x) > > and (@ y) respectively. > > This is possibly irrelevant, but since there were some pointers in a > direction that is not doing the above: > > 1. You should try to avoid doing this at the reader level as much as > possible. If you're trying to replace `@foo' by something else, > then you can easily run into a mess when these things are quoted > etc. In addition, since you're at the reader level, you cannot > know if you're looking at something which is any other kind of > non-expression, like argument names etc. > > (This is from experience: a previous attempt at the scribble reader > was reading @foo{blah} as (dispatch foo "blah"). It was > problematic for these reasons.) > > 2. Identifier macros don't seem so helpful either, since they're used > as bindings for known names. > > 3. Using `#%top' might work, if you want the bindings of the > wildcard-ed @foo identifiers to be values as in Danny's example, > but it won't work if you want them to be macros. > > So if you really want them to be macros, then I think that your best > bet is to hook into both the `#%top' and the `#%app' macros, the > former to identify cases where you want (.... @foo ....) to expand to > some (.... "at-foo" ....), and the latter to identify cases where you > want (@foo ....) to get expanded to something else (that is, not just > expand the @foo as in the identifier case). > > Here's an example that shows how both are used -- for fun, I made two > things magical, `foo%' identifiers, and `@foo' things that were read > by the scribble reader (which can be identified via a syntax > property). > > > > ------------------------------------------------------------------------------- > #lang at-exp racket/base > > (module magic racket/base > > (provide (rename-out [top #%top] [app #%app])) > > (require (for-syntax racket/base)) > > (define-for-syntax (magical? id) > (and (identifier? id) > (regexp-match? #rx"%$" (symbol->string (syntax-e id))))) > > (define-syntax (top stx) > (syntax-case stx () > [(_ . id) > (cond [(magical? #'id) #''id] > [(syntax-property stx 'scribble) #''id] > [else #'id])])) > > (define-syntax (app stx) > (syntax-case stx () > [(_ id x ...) > (cond [(magical? #'id) #'(vector 'id x ...)] > [(syntax-property stx 'scribble) #'(list 'id x ...)] > [else #'(id x ...)])])) > > ) > > (require 'magic) > > blah% > (blah% 1 2 3) > @blah > @blah{zzz} > > ------------------------------------------------------------------------------- > > > -- > ((lambda (x) (x x)) (lambda (x) (x x))) Eli Barzilay: > http://barzilay.org/ Maze is Life! > -- Talk to you soon, Scott Klarenbach PointyHat Software Corp. www.pointyhat.ca p 604-568-4280 e [email protected] #308 - 55 Water St. Vancouver, BC V6B1A1 _______________________________________ To iterate is human; to recur, divine
____________________ Racket Users list: http://lists.racket-lang.org/users
