Quote:Original post by mikeman
The thing is, while your project is admirable, that whenever we see someone claiming their implementation is faster, we always suspect that the implementation neglects several features and standards in order to "pass" some selected benchmarks. But SC++L is so much more than that. We must compare the 2 implementations in their entirety. Case in point: sorting a core::list takes more than 2x the time it takes to sort an std::list(in my test, sorting a std::list with 10000 integers takes on average 70ms, whereas doing the same with a core::list takes 180ms). With larger types it's even worse, std seems to be up to 4x faster.
That misses the point. The whole reason that someone would opt to write custom containers instead of STL is precisely because they can "neglect" some features which they find worthless and tailor their code to perform well in certain scenarios (which correlate to the benchmarks).