Re: [racket] Profiling help

2013-07-11 Thread Robby Findler
Nothing comes to mind. If you know where the runtime is supposed to start you might try adding printfs in a few places that print out current-process-milliseconds (put a printf right before the call to profile-thunk too) and see if you can kind of hone in on something. You might find something sur

Re: [racket] Profiling help

2013-07-11 Thread Joe Gibbs Politz
On Thu, Jul 11, 2013 at 5:22 PM, Robby Findler wrote: > Is it possible that the time is being spent in the expander or in other > C-based code that would be hidden from the profiler? I don't know quite what C-based code we might be touching, but I do know that our parser down to a racket/base pro

Re: [racket] Profiling help

2013-07-11 Thread Robby Findler
Is it possible that the time is being spent in the expander or in other C-based code that would be hidden from the profiler? Robby On Thu, Jul 11, 2013 at 4:07 PM, Joe Gibbs Politz wrote: > We're trying to profile some slow running programs in a #lang, and > having some trouble getting believa