On Fri, May 7, 2021 at 5:35 PM Marc Nieper-Wißkirchen
<xxxxxx@nieper-wisskirchen.de> wrote:
>
> Am Fr., 7. Mai 2021 um 09:42 Uhr schrieb Alex Shinn <xxxxxx@gmail.com>:
>>
>> If you compare the (chibi disasm) bytecode for:
>>
>>   (case-lambda ((x) (+ x 1)) ((x y) (+ x y)))
>>
>> vs
>>
>>   (opt-lambda (x (y 1)) (+ x y))
>
>
> I wasn't comparing case-lambda against opt-lambda or some other specialized construct but case-lambda against (lambda args (match args ...)). That's why I wrote "general match code".

You cut out the following comment, which states that Chibi currently
optimizes the (lambda args ...) case even faster than opt-lambda.

> Yes; eventually we need an optimizing implementation of match as in Racket.

This could be done by the compiler, without any need to optimize the
match macro itself.

--
Alex