Am Sa., 25. Apr. 2020 um 18:04 Uhr schrieb John Cowan <xxxxxx@ccil.org>:

>> SRFI 177 whose purpose is to abstract
>> all native keyword systems out there to some common denominator.
>
>
> That's incomplete.  Its purpose is to provide a universal keyword system that is compatible with all known native keyword systems, but also works on systems without native keywords.

It depends on what you call "native" keywords. I would count the plist
approach that SRFI 177 uses in its portable implementation also
amongst such native keyword systems.

>> This let-keywords-style could, however, be the basis for SRFIs that
>> aim for inclusion into R7RS-large because it is readily supported by
>> any system, no less performant than CL-style keywords, it does work
>> with apply and does not need any new (lexical) syntax.
>
>
> SBCL is smart enough to compile function calls with literal keywords (as opposed to expressions that evaluate to keywords)  as efficiently as positional arguments.  Scheme compilers could in principle do the same.

Literal keywords would mean Racket/Kawa style, right?

I don't see why expressions evaluation to keywords (or symbols in the
simplest approach) would lead to less performant code on an optimizing
Scheme compiler when the callee is known at the call site.