Ah! Thanks for the pointer. Should have looked in the issues. Yes, 2 filters certainly work fine.
John > On Oct 5, 2021, at 4:45 PM, Ben Greenman <[email protected]> wrote: > > There's an issue open about partition: > https://github.com/racket/typed-racket/issues/138 > > Does `my-partition` work like you'd want? Based on Alex's last comment > on the issue, it seems hard to give a predicate that matches the type. > > > (Whenever I've wanted `partition` in typed code, I was always able to > use 2 filters instead.) > > On 10/5/21, 'John Clements' via Racket Users > <[email protected]> wrote: >> I was somewhat surprised to see today that I can’t use a predicate with both >> positive and negative propositions in the way I would expect with >> partition: >> >>> (:print-type partition) >> (All (a b) >> (case-> >> (-> (-> b Any : #:+ a) (Listof b) (values (Listof a) (Listof b))) >> (-> (-> a Any) (Listof a) (values (Listof a) (Listof a))))) >> >> >> Specifically, I would have expected the type to be something like this: >> >> (All (a b c) >> (case-> >> (-> (-> b Any : #:+ a) (Listof b) (values (Listof a) (Listof b))) >> ;; the second list must consist of 'c's: >> (-> (-> b Any : #:+ a #:- c) (Listof b) (values (Listof a) (Listof c))) >> (-> (-> a Any) (Listof a) (values (Listof a) (Listof a))))) >> >> … so that if, say, I had a list of Elephants and Emus, that I could use >> elephant? to split it into two lists: one of type (Listof Elephant) and one >> of type (Listof Emu). >> >> I tried to roll my own, and got pretty close: >> >> (: my-partition >> (All (a b c) >> (case-> >> ;; the second list must consist of 'c's: >> (-> (-> b Any : #:+ a #:- c) (Listof b) (values (Listof a) (Listof >> c))) >> ))) >> >> (define (my-partition my-pred elts) >> (cond [(empty? elts) >> (values '() '())] >> [else >> (define-values (stacks non-stacks) >> (my-partition my-pred (rest elts))) >> (define f (first elts)) >> (cond [(my-pred f) >> (values (cons f stacks) >> non-stacks)] >> [else >> (values stacks (cons f non-stacks))])])) >> >> That is, I can do it for the case-> clause that I care about. Putting the >> other two back in there causes it to fail type-checking. Is that the >> problem, that TR can’t accommodate both flavors in the same type? >> >> John >> >> >> -- >> You received this message because you are subscribed to the Google Groups >> "Racket Users" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/racket-users/aa1e77a8-9cc4-4f99-b413-1304daeec12b%40mtasv.net. >> -- You received this message because you are subscribed to the Google Groups "Racket Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/racket-users/724d3012-b8e3-4be2-9ce1-fbc15cbd0bd6%40mtasv.net.
