Then your program would change semantics depending on your type annotations (and potentially be different than the untyped Racket version). I don't think that's a road we want to go down.
It would be better if `for/sum` let you supply the base case. On Thu, Mar 28, 2013 at 4:52 PM, Matthias Felleisen <[email protected]> wrote: > > Perhaps for/sum should adapt its base case answer to the return type? > > > > On Mar 28, 2013, at 4:03 PM, Sam Tobin-Hochstadt <[email protected]> wrote: > >> I'm not sure what you're referring to. This program doesn't typecheck, >> because if both vectors were empty, the result is exact 0, which isn't >> a `Float`. >> >> Sam >> >> On Thu, Mar 28, 2013 at 3:56 PM, Ray Racine <[email protected]> wrote: >>> Is this a bug? >>> >>> #lang typed/racket/base >>> >>> (define: βns : (Vectorof Float) '#(1.0 2.0 3.0)) >>> (define: χns : (Vectorof Float) '#(4.0 5.0 6.0)) >>> >>> (for/sum: : Float >>> ([βn : Float (in-vector βns)] >>> [χn : Float (in-vector χns)]) >>> (* βn χn)) >>> >>> ____________________ >>> Racket Users list: >>> http://lists.racket-lang.org/users >>> >> >> ____________________ >> Racket Users list: >> http://lists.racket-lang.org/users > ____________________ Racket Users list: http://lists.racket-lang.org/users
