[Chicken-hackers] Re: cond-expand and macros

[felix]

Thomas Bushnell BSG <tb <at> becket.net> writes:

> 
> I thought (erroneously) that the problem I reported previously was that
> somehow the feature set available to cond-expand inside a syntax-rules
> template was the evaluation one and not the compilation one.  No, the
> situation is far worse.
> 
> The following code does not work:
> 
> (define-syntax foo
>   (syntax-rules ()
>     ((_)
>      (cond-expand
>       (chicken (display "all good\n"))))))
> (foo)
> 
> It does not work whether compiled or evaluated; the issue of
> compile-time is a complete red herring, and my previous email was
> incorrect in implying that compile vs. eval was relevant.  AFAICT, it's
> not.
> 
> I submit that the failure of the above code is a clear bug,
> unambiguously.
> 

Absolutely correct. The feature-identifier is not stripped of syntactical
context information. This patch will help:


diff --git a/expand.scm b/expand.scm
index 053e278..cca83fc 100644
--- a/expand.scm
+++ b/expand.scm
@@ -1195,7 +1195,7 @@
                     x
                     (cons 'cond-expand clauses)) )
       (define (test fx)
-       (cond ((symbol? fx) (##sys#feature? fx))
+       (cond ((symbol? fx) (##sys#feature? (##sys#strip-syntax fx)))
              ((not (pair? fx)) (err fx))
              (else
               (let ((head (car fx))


cheers,
felix