This is the mail archive of the guile@cygnus.com mailing list for the guile project.
| Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
|---|---|---|
| Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
I should hang a sign on some wall, saying: "Don't assume that
people read your thoughts!".
Anyway.
Maciej Stachowiak <mstachow@mit.edu> writes:
> mike@olan.com writes:
> > I'd add:
> >
> > (RFI ...)
> > The submodules have the names (numbers?) of cataloged RFI's. For
> > instance, Olin Shivers' list lib will probably be (I'm making it up)
> > RFI #1, and any program that relies on it would be able to say:
> >
> > (use-modules (rfi 1)) ;use the common list functions.
> >
>
> I don't like this suggestion, especially referencing RFIs by number. I
> certainly expect them to also have nice symbolic names.
Me too, it was just an example.
> But beyond
> this, I think we should just put RFI-implementing modules wherever is
> convenient in the functional namespace, for instance I'd suggest
> putting the list lib in question under (util lists) or (util list).
My real point was to make it possible for a random person that has
written a portable Scheme program to declare the RFI's [s]he needs
without having to familiarize h{im,er}self with whatever Guile folks
decided to call them. The (RFI ...) namespace should exist only for
the sake of such uses, and things referenced there can (and should, in
fact) live in more logical places too.
> In many cases we will probably want to use something other than the
> sample implementation anyway, either to rewrite some or all in C for
> efficiency or because some of the relevant bindings might already be
> in the Guile core, or whatever other reasons.
It also might (theoretically) be possible that Guile module
"granularity" won't in some cases match the RFI's, and so some RFI's
will be implemented by several Guile modules, maybe with extra things
thrown in. The (RFI ...) modules will reference _just_ the RFI's, no
more, no less.
> Of course, if there is ever an officially blessed Scheme module system
> and RFIs get standardized package names we may want to reconsider
> this.
Doubtful...
> - Maciej
mike.