Email list hosting service & mailing list manager

SRFI 177, Portable keyword arguments: 270 days Arthur A. Gleckler (12 Jul 2020 18:34 UTC)
Finalizing SRFI 177 Lassi Kortela (15 Jul 2020 18:18 UTC)
Re: Finalizing SRFI 177 Lassi Kortela (15 Jul 2020 18:42 UTC)
Re: Finalizing SRFI 177 Lassi Kortela (15 Jul 2020 19:02 UTC)

Finalizing SRFI 177 Lassi Kortela 15 Jul 2020 18:18 UTC

> This is a reminder that SRFI 177 <>:
> Portable keyword arguments, has been under public discussion for 270
> days. (It was first published on 2019-10-16.) The discussion can go on,
> but in theory, the longest extension was supposed to have been to ninety
> days.

Apologies for the huge delay. I take full responsibility for it.

Let's decide on the final specification now.

I looked into identifier syntax as Marc suggested, and the standard
`syntax-case` in R6RS easily allows specifying a `foo` macro with
different behavior for the `(foo ...)` case (i.e. a list) and the `foo`
case (i.e. the bare identifier by itself, as if it were a variable).

(Chez Scheme's `identifier-syntax` is a different but related facility.
It specifies a `foo` macro with only the bare-identifier version; using
`(foo)` as a list will be an error. However, `identifier-syntax` lets
one define a custom setter for set!, similar to SRFI 17 (Generalized
set) or Common Lisp's "setf places".)

However, Kawa's syntax-case doesn't handle bare-identifier syntax the
way R6RS does. I sent a question about this to the Kawa mailing list.

I can't find any variant of `syntax-rules` supporting identifier syntax,
though perhaps it could be added in a backward-compatible way.
(define-syntax foo (syntax-rules () (_ 123))) is currently a syntax error.

It's doubtful that the various versions of the CL-style `define-macro`
can define macros with separate `(foo ...)` and `foo` expansions.