Re: Recursive define-syntax

[Philippe de Rochambeau <phiroc@free.fr>]

Hi Per,
Thank you for your feedback.
The code works on LispKit (cf. https://github.com/objecthub/swift-lispkit) which is based on R7RS. But if it doesn’t work in Kawa, it doesn’t really matter.

#|
Pamphlet, p. 52
|#
(import (lispkit match))
(import (srfi 48))

(define (dim m)
 (match m
   ((first . rest)
     `(,(length m) . ,(dim first)))
    (_
      '())))

Cheers,
Philippe

> Le 26 sept. 2021 à 23:04, Per Bothner <per@bothner.com> a écrit :
> 
> 
> 
>> On 9/26/21 05:08, Philippe de Rochambeau via Kawa wrote:
>> Hello,
>> the following procedure, excerpted from « A Pamphlet Against R »
>> (define (dim m)
>>  (match m
>>    ((first . rest)
>>      `(,(length m) . ,(dim first)))
>>     (_
>>       '())))
>> … causes Kawa to display the following error message:
>> /dev/tty:6:5: unrecognized pattern operator first
> 
> There is no standard Scheme 'match' syntax - there are a
> number of different and incompatible extensions.  « A Pamphlet Against R »
> uses the Guile language, which is quite different frm Kawa.
> 
> Specifically, the Kawa "pattern" syntax is quite limited - it
> is mainly a proof of concept - a test of the basic framework.
> 
> Thar said, this seems to work:
> 
> (define (dim m)
> (match m
>   ([first rest ...]
>     `(,(length m) . ,(dim first)))
>    (_
>     '())))
> 
> (dim m0) => (3 4)
> 
>> Are recursive « syntaxes »  not possible in Kawa?
> 
> Before you ask that question, you should try some code that
> works on at least one other (preferable two) Scheme implementations.
> If you have example code that works on other Scheme implementation(s)
> and does not depend on an implementation-specific extension, but
> not on Kawa - then we can discuss whether there is a Kawa bug or limitation.
> -- 
>    --Per Bothner
> per@bothner.com   http://per.bothner.com/