Add ‘match?’, a pattern-matching predicate

[leafac]

FEATURE PROPOSAL

Add match? to racket/match. It is given a pattern and a subject expression, if the expression evaluates to a value that matches the pattern, then match? returns #t, otherwise it returns #f. Also, as an alternative form, match? could be given just a pattern, in which case it returns a predicate function.

Here’s a proposed implementation:

#lang racket
(require syntax/parse/define)

(define-syntax-parser match?
  [(match? pattern:expr subject:expr)
   #'(match subject
       [pattern #t]
       [_ #f])]
  [(match? pattern:expr)
   #'(match-lambda
       [pattern #t]
       [_ #f])])

(match? `(,c ,d) `(a b)) ; => #t

((match? `(,c ,d)) `(a b)) ; => #t

This design follows redex-match?.

---

If you agree with this proposal, please let me know and I’ll send a pull request.

Thanks.

[samth]

This seems like a good idea, but I'm not sure about the name. What about matches?

…

On Fri, May 12, 2017, 11:46 AM Leandro Facchinetti ***@***.***> wrote: *FEATURE PROPOSAL* Add match? to racket/match. It is given a subject expression and a pattern, if the expression evaluates to a value that matches the pattern, then match? returns #t, otherwise it returns #f. Here’s a proposed implementation: #lang racket (require syntax/parse/define) (define-simple-macro (match? subject:expr pattern:expr) (match subject [pattern #t] [_ #f])) (match? `(a b) `(,c ,d)) # => #t ------------------------------ If you agree with this proposal, please let me know and I’ll send a pull request. Thanks. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1690>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAO78y51FGzkCcAtE7m1qAc2TFE6YEpIks5r5H7RgaJpZM4NZdcT> .

[lexi-lambda]

I thought matches? at first, too, but the match? name is actually more consistent with regexp-match? and redex-match?.

[leafac]

We need to consider potential conflicts. Particularly interesting is that it seems match? was a thing in the past.

[samth]

Well, that wasn't any more of thing in the past than now, but it suggests we should use that name. But we do need to make sure there aren't conflicts.

[winny-]

Any other conflicts? This would be a very handy syntax :)

[rfindler]

It is exported by redex: http://docs.racket-lang.org/search/index.html?q=match%3F

[samth]

match? is provided already by unstable/match, and could presumably be in a smaller package. Is there a reason this needs to be provided by racket/match?

[leafac]

1. Convenience.
2. unstable/match “is deprecated; use racket/match, instead.”

[samth]

Right, I think we should have a new package that provides the remaining components of unstable/match.

[stamourv]

Having to install and require a separate package would cancel out the small convenience that match? would give, IMO.

[samth]

We likely can't add it to racket/match itself, as @rfindler points out, redex already exports the same name.

I think we should get to a point with the pkg ecosystem where adding a new dependency for something this small is a reasonable choice. Maybe we're not there yet, though.

[rfindler]

I agree with Sam's sentiment about small packages (both that it would be nice and that perhaps we aren't there).

Is there another name to consider?

[stamourv]

matches? was suggested, and sounds fine to me, given the constraints.

[leafac]

I’d be happy if matches? was available when I (require racket/match) the same way that hash-union is available when I (require racket/hash)—meaning it wouldn’t be part of racket/base or racket. I’d be happier if it were part of racket/base and racket as well.