@pnwamk http://drdr.racket-lang.org/39429/racket/share/pkgs/typed-racket-test/with-tr-contracts.rkt

@samth @stamourv is there a discussion somewhere about this issue: https://github.com/racket/typed-racket/issues/112

@pnwamk ugh

we should just relax the type of < (and all the other comparisons) on floats

we used to distinguish NaN in the types, but that was bad because too many things produce NaN

although that will probably break things like (if (> x 0.0) (sqrt (* x x)) 1.0)

sigh

should the type of sqrt be relaxed then?

no

that wouldn’t be sound

Sigh.

Yeah, relaxing the types of comparisons on floats is probably the right call.

@stamourv which parts of the types need changing? the type of < is pretty big :slightly_smiling_face:

I don’t know off the top of my head. Will look later today. Just about to bike to the office.