Bug 10667 – http://dlang.org/cppstrings.html benchmark example doesn't really show off slices

While the wordcount benchmark programs do show what can be done with C++98 vs D2, this is not really a good demonstration of the performance benefits of slices. The D program stores word counts in a hash map. The C++ version given uses a map, which is required to be a balanced binary tree data structure. For sizable input, the algorithmic advantage will swamp the savings from using slices. This is also a small enough input that memory handling effects probably won't get noticed. I suspect the D program never needs to run a GC cleanup before the end. I highly suggest the C++ programs be converted to C++11 where unordered_map is available to remove the hidden algorithmic advantage. Secondly, repeat the input a number of times so the timing is more meaningful.

This page apparently is no longer included in the website build. Closing this for now. Please reopen if similar issues exist in currently-public website pages, thanks!