Chromatic, All, Fair enough. What with the recent merges, and maybe a problem with PMC** in the pmc compiler, I've not had a chance to recut the patch.
Since about 1/3 time seemed to be spend in the GC (and some more in memory allocation), this is one area where huge wins might be possible. (Some tuning might be all that's required) Nick On Tue, Oct 14, 2008 at 5:34 AM, chromatic <[EMAIL PROTECTED]> wrote: > On Thursday 02 October 2008 07:35:09 NotFound wrote: > >> I think will be better the other way, using the return value to flag >> existence, and passing a pointer to store the result. This will allow >> shorter and cleaner code. > > A 5% improvement in some subsystems is nice, but nothing else uses out > parameters in this way. I suggest we leave the idea as a potential > improvement (and look for other ways to speed up all code).
