[R6RS] Re: modules

[Michael Sperber]

>>>>> "RK" == Richard Kelsey <kelsey at s48.org> writes:

RK>    Would you have a separate way to associate names with pathnames, or do
RK>    you prefer that modules be explicitly loaded?

RK> I am not sure that any single mechanism is going to be adequate.
RK> Programs are executed in so many different ways.  Explicit loading
RK> seems the most flexible, as it leaves implementations free to add
RK> whatever implicit loading mechanism they want.

Well, but to be practical, you're still going to have to offer some
kind of standard mechanism so you can ship sets of files with module
declarations in them to some user type.

Having extensively worked with, and shipped applications based on, the
module systems of Scheme 48 (explicit loading) and MzScheme (implicit
loading), I can't really say that I ever felt a particular advantage
for one or the other approach.  The only complaints I had directly
reflect the fact that each system isn't the other:

Scheme 48: The module name space is basically unstructured, and thus
  too small.

MzScheme: The module name syntax is tied to the file-name syntax, and
  I can only have one module declaration per file.

Both problems are easily fixable (at least design-wise).  The
follow-up proposal to Matthew's I posted does, for example.

-- 
Cheers =8-} Mike
Friede, V?lkerverst?ndigung und ?berhaupt blabla