When Herb gave a talk at NWCPP not long ago about C++0x, Walter's questions and comments were priceless. Herb asks "would you expect this [messy-looking but seemingly optimal c++ loop] to be faster, or this [nice clean looking c++0x code that you might think has lots of overhead]". Walter raises his hand and says something like "the c++0x version should definitely be faster". Herb seems to think maybe he's just guessing what the punch line is, but Walter follows up with "Sure, because in the C++0x version you end up using using integer indexing, which is much easier for the compiler to optimize, while the 'optimal c++' version uses pointers." That was the right answer. Could be me misreading his reaction, but it almost seemed to me like Herb didn't know why his C++0x code was faster than his C++ code.
Not to put Herb down too much, he did put on a great talk. He's an excellent presenter and communicator. --bb On Wed, Nov 16, 2011 at 9:57 AM, Andrej Mitrovic <andrej.mitrov...@gmail.com > wrote: > There's also this other video of you guys discussing C++0x which might > be interesting (maybe not relevant to D but anywho): > > > http://channel9.msdn.com/Shows/Going+Deep/C-and-Beyond-2011-C11-Panel-Scott-Meyers-Andrei-Alexandrescu-and-Herb-Sutter >