[plt-scheme] Abstraction in require statements

From: Matthew Flatt (mflatt at cs.utah.edu)
Date: Thu Sep 13 16:50:17 EDT 2007

At Tue, 11 Sep 2007 13:02:08 +0100, Dave Gurnell wrote:
> I would like to be able to do something equivalent to:
> 
>      (define-syntax (unlib stx)
>        (syntax-case stx ()
>          [(unlib file)
>           #'(planet file ("untyped" "unlib.plt" 2))]))
> 
>      (require (unlib "list.ss")
>               (unlib "string.ss")
>               (only (unlib "check.ss") pass fail warn))
> 
> Of course, this doesn't work because of the way require is  
> implemented, but you get the idea.
> 
> I have thought about implementing a require* form that expanded into  
> require:
> 
>      (define-require*-syntax unlib
>        ; blah blah blah
>        )
> 
>      (require* (unlib "list.ss")
>                (unlib "string.ss")
>                (only (unlib "check.ss") pass fail warn))
> 
> The idea would be to get define-require*-syntax to store its forms in  
> some kind of private namespace, and get require* to search that  
> namespace to retrieve the forms. That way, require* forms won't be  
> defined in module bodies. Because require* expands into require,  
> require* forms should only be able to expand into require forms.
> 
> Alas, my macro-fu is weak, and I don't know if this is a realistic  
> approach. Also, it's not worth trying to implement something like  
> this if there's an existing solution that does the job. Hence this  
> email!

Yes. Your `define-require*-syntax' will expand to

 (define-syntax (make-require*-macro ...))

and `require*' will use `syntax-local-value' to access require* macros.


Something like this was always the intent with the original `require',
except that we never got around to adding `define-require-syntax' ---
most due to namespace issues. The problem is that binding something
like `rename' for use in `provide' means that `rename' can't be used
anywhere else.

After considering various ways around this problem (including multiple
namespaces and implicit prefixes), I'm included to pick names for the
sub-forms in `require' 4.0 that won't collide anywhere else --- and
then finally add bindings for the built-in forms, as well as
`define-require-syntax' for adding new ones.

Matthew



Posted on the users mailing list.