|
From: Julian S. <js...@ac...> - 2011-03-04 23:51:27
|
> Is there a reason to have the sophisticated version ? Er .. I can't think of one :-) > The below looks simpler and I believe more efficient. The order will > be different than of the above, but I have not seen where the specific > order implemented by the above cmp is significant. Looks good to me. I'll commit it in the morning. I think the general inefficiency of the laog mechanism (I am amazed that it takes 1.4GB in your tests) is partly because you are really the first person to seriously try to tune/optimize it. So you are seeing a lot of stupidity in my initial implementation. J |