[racket] Lifting submodules

From: Greg Hendershott (greghendershott at gmail.com)
Date: Wed Sep 18 16:54:41 EDT 2013

Thank you for showing how to make it work.

Now as to why it works... sadly I'm still confused.

> For this example, if I understand the docs correctly, the scope of the
> require-spec has to match the scope of the use.
> http://docs.racket-lang.org/reference/syntax-model.html?#(part._macro-introduced-bindings)
> "In require for other require-specs, the generator of the require-spec
> determines the scope of the bindings"

Although I wasn't familiar with that part of the docs (thanks for
pointing it out!), it did occur to me that the module name would need
to have the same lexical scope.

In my attempt...

    (define-syntax (define-modularize stx)
      (syntax-case stx ()
        ([_ thing-name mod-name]
             (module mod-name racket
               (define thing-name #t)
               (provide thing-name))
             (require 'mod-name))))) ;; or, (require (submod "." mod-name))
    (define-modularize foo foo-mod)
    foo ;; => foo: unbound identifier in module

...I reasoned that "mod-name is a pattern variable, and pattern
variables automatically carry the lexical scope of the syntax given to
syntax-case". Right?

So in my attempt, how does the usage scope of `foo-mod` get "sliced
off" in the pattern variable `mod-name`? Is it a result of needing to
quote the module name, `(require 'mod-name)`?  But `(require (submod
"." mod-name)` doesn't quote it and that doesn't work either.

Posted on the users mailing list.