Re: questions on libraries, pattern matching etc

[Kjetil Matheussen <k.s.matheussen@gmail.com>]

I made this one a few years ago, which I think is pretty good:
https://github.com/kmatheussen/fedex2

I've used it a lot too, so it's well tested:
https://github.com/kmatheussen/radium/tree/master/bin/scheme
(might be some improvements in this repo)

It should only contain plain r4rs/r5rs scheme code (or thereabout),
except for the define-match macro, which looks like this:

(define-macro (define-match funcname . matchers)
  (create-matcher-func funcname matchers))

Example:

(define-match keep
  [        ] ____ :> '[]
  [A . Rest] Pred :> (cons A (keep Rest Pred)) :where (Pred A)
  [_ . Rest] Pred :> (keep Rest Pred))

(define-match quicksort
  []      :> '[]
  [A . R] :> (append (quicksort (keep R (lambda (B) (>= A B))))
                     (list A)
                     (quicksort (keep R (lambda (B) (< A B))))))

On Wed, Feb 19, 2020 at 10:39 PM Duncan Mak <duncanmak@gmail.com> wrote:
>
> Hello Ben,
>
> Did you figure out a way to get a pattern match library to work with Kawa?
> Which one did you go with?
>
> I came across this recently, and I'm gonna give it a try:
> http://www.j-paine.org/match.html
>
> I don't have particularly sophisticated needs, I'm hoping to find something
> that will let me match on lists with some literals and some open slots, and
> it'll be great if I can define optional elements also.
>
>
> Duncan.
>
> On Sun, Dec 15, 2019 at 7:59 PM Per Bothner <per@bothner.com> wrote:
>
> > On 12/15/19 3:46 PM, Ben wrote:
> > > hi
> > > I'd like to test how I can use pattern match in Kawa. First I did try to
> > use Kawas pattern matching function, but from what I saw it is a bit
> > limited, for example there is no matching of lists.
> >
> > Actually, there is matching of lists, but by matching them as general
> > sequences:
> >
> > #|kawa:1|# (! [a b c @rest] [1 2 3 4 5 6])
> > #|kawa:2|# (format "a: ~a b: ~a c: ~a rest: ~a~%" a b c rest)
> > a: 1 b: 2 c: 3 rest: #(4 5 6)
> >
> > Implementing more general matching is mainly an issue of design including
> > deciding on a syntax.
> > Fundamentally, should be syntax for matching a pair be:
> >
> > (! (pat_car . pat_cdr) value)
> >
> > or:
> >
> > (! (cons pat_car pat_cdr) value)
> >
> > or something else?
> > The latter is used in Racket, and is more flexible, I believe, but not as
> > elegant -
> > which ties back to fundamental awkwardness with the Scheme evaluation
> > model.
> >
> > Of course once we/I decide on a syntax, then it needs to be implemented,
> > but
> > should be fairly straight-forward, given the existing framework.
> > --
> >         --Per Bothner
> > per@bothner.com   http://per.bothner.com/
> >
>
>
> --
> Duncan.