Since the expansion of (bar x) depends on the identifiers
in the surrounding usage of foo, one solution is to use
syntax-parameters that adjust the meaning of (bar x)
according to the context in which it appears.
#lang racket
(require (for-syntax syntax/parse racket/stxparam) racket/stxparam
(for-syntax (for-syntax racket syntax/parse)))
(define-syntax-parameter bar
(λ (stx)
(raise-syntax-error 'bar "bar keyword used outside foo" stx)))
(begin-for-syntax
(define-syntax-class bar-expr
(pattern ((~datum bar) x:id))))
(define-syntax (my-foos-of-bars stx)
(syntax-parse stx
[(_ ((~datum foo) foo-id:id ... e:bar-expr ...))
#'(syntax-parameterize
([bar (λ (stx)
(syntax-parse stx
[(_bar x:id)
(if (memq (syntax-e #'x) '(foo-id ...))
#'(list 'x)
#''x)]))])
(list 'foo e ...))]))
(my-foos-of-bars (foo x y (bar a) (bar x)))
2015-02-12 17:16 GMT+01:00 Konrad Hinsen <konrad.hin...@fastmail.net>:
> Greg Hendershott writes:
>
> > Preface: Whenever I attempt to answer a question about macros, my secret
> > agenda is that I'll be the main beneficiary... from someone else jumping
> > in to correct my answer and teach me. So here goes. :)
>
> Fair enough ;-)
>
> > Does `bar` need to be a syntax-class? Instead, what if it's just another
> > macro -- that `foo` expands into, passing the set?
>
> Introducing another macro lets me factor out the expansion of "bar",
> but not the pattern itself. In my case, it would already yield some
> benefits, so I'll keep this solution in mind if nothing better comes up.
>
> But ideally, I would like to factor out the pattern itself, which in
> my real application is more complex and consists of several cases.
>
> Thanks,
> Konrad.
> ____________________
> Racket Users list:
> http://lists.racket-lang.org/users
--
--
Jens Axel Søgaard
____________________
Racket Users list:
http://lists.racket-lang.org/users
