David Van Horn caught the bug. The fast module has an incorrect bind function that when fixed makes performance terrible. Nevermind this thread. -Ian ----- Original Message ----- From: "J. Ian Johnson" <[email protected]> To: "Sam Tobin-Hochstadt" <[email protected]> Cc: "J. Ian Johnson" <[email protected]>, "users" <[email protected]> Sent: Thursday, October 4, 2012 9:47:52 PM GMT -05:00 US/Canada Eastern Subject: Re: [racket] Vast performance differences with minute syntactic differences
I wrote two set-adds? Ugh, I didn't mean to. That code wouldn't even work due to arity mismatch. (set-add res (let ([v v]) v)) is more likely. -Ian ----- Original Message ----- From: Sam Tobin-Hochstadt <[email protected]> To: J. Ian Johnson <[email protected]> Cc: users <[email protected]> Sent: Thu, 4 Oct 2012 19:46:36 -0400 (EDT) Subject: Re: [racket] Vast performance differences with minute syntactic differences On Thu, Oct 4, 2012 at 7:01 PM, J. Ian Johnson <[email protected]> wrote: > > By almost exactly I mean > (let () (set-add res v)) > > versus > > (let ([res res]) > (set-add (let ([v v]) > (set-add res (let () v))))) These seem importantly different, in that: (set-add v (set-add res v)) and (set-add res v) are totally different values, and the former only works if both `v` and `res` are sets. If you want to determine the behavior of the compiler, which is what matters for differences like extra `let` bindings, 'raco decompile' is the best tool. -- sam th [email protected] ____________________ Racket Users list: http://lists.racket-lang.org/users
