Re: Module names

[Maciej Stachowiak <mstachow@mit.edu>]

bothner@cygnus.com writes:
> > I think such a system would make separate compilation difficult,
> > because you could import arbitrary symbols at runtime.
> 
> Of course, but only if symbols are evaluated at run-time.
> And I am *not* saying that Guile should allow names to
> be evaluated at run-time.  I am just saying that it would be
> nice if Guile was *compatible* with a system where names can
> be looked up at run-time.
> 
> > I think supporting separate compilation is a goal for the future
> > Guile module system.
> 
> Of course.  I was only requesting a *syntactic* change:
> (MODULE 'NAME) instead of (MODULE NAME).  That has no direct
> implication as to *semantics*, except that it allows a
> more dynamic system as an extension or option.
> 

I think people would find it very confusing to see syntax that
requires a quoted symbol but will not accept other values that
evaluate to a symbol. I've never heard of anything in any Scheme
dialect that worked that way.

Incidentally, I don't think there should be a (MODULE ...) syntax to
directly access the symbols of a module, IMO modules should be
distinct from procedures, so compatibility is not an issue.

 - Maciej