[racket] non-terminating type check for typed racket

From: Vincent St-Amour (stamourv at ccs.neu.edu)
Date: Mon Apr 18 13:01:24 EDT 2011

At Mon, 18 Apr 2011 12:26:26 -0400,
Eli Barzilay wrote:
> But in the TR case the transformation seems fine because
> `filter-equal?' is defined as a simple
>   (define (filter-equal? a b) (= (Rep-seq a) (Rep-seq b)))
> Sam/Vincent: that's probably a good thing to do, and even better --
> get rid of `filter-equal?' so it's easier to use the seq directly.
> (IIUC, it's the thing that has efficiency as its whole point.)  A
> quick grep finds this loop:
>   (for/or ([f (in-list result)]) (or (filter-equal? f t) (implied-atomic? t f)))
> which could also be improved if the -seq of `t' is taken only once.

Fixed. Thanks.

I'll push that later.


Posted on the users mailing list.