On Saturday, October 11, 2014 9:44:39 AM UTC-5, David Smith wrote: > > Thanks! I still don't understand why something this simple, even in the > global scope, can't be optimized. >
Because the type can be changed out from under the optimizer. If you make the variables const, that should also work. > Also, any idea why the regression in run time from 0.3 to 0.4? > Hard to say; v0.4 is in significant flux right now. You can file an issue, but I wouldn't expect anything to be done until much later in the dev cycle since there's more stuff Jeff wants to break first.
