|
From: Roman Y. <rom...@gm...> - 2006-08-27 19:22:13
|
On 8/27/06, Allen Bierbaum <al...@vr...> wrote: > Roman: > > It looks like about 1/3 of the time is now being spent inside > proceed_single in make_flatten. > > Is there any reason I shouldn't be able to use _make_flatten_generator > instead of make_flatten so I can remove the performance hit of > allocating the memory for such a large list every time make_flatten is > called? Yes, performance reason :-). Last time I tested it, it was very slow. But you can try and report the numbers. P.S. I think that few testers and may be some real code will fail here and there but we can fix it. -- Roman Yakovenko C++ Python language binding http://www.language-binding.net/ |
