SuperString

Actually, the UTF-8 decoding is broken even with proper input. This https://github.com/btwael/SuperString/blob/master/src/SuperS... will make `pointer` and `i` be out of sync, so everything after the first multibyte sequence is a coin toss. To the OP: you should use theses tests to validate your implementation https://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt (caveat: 5- and 6-byte sequences are no longer valid UTF-8 and should be rejected).

Just in case I guess. :) +// in case NULL is not defined +#ifndef NULL +#define NULL (0L) +#endif // NULL https://github.com/btwael/SuperString/commit/515f6896ee6a3e2...

Looking at the commit history on the project [0] (and inadvertently digging a bit deeper into the OP's StackOverflow reply history to notice that all are advertising his non-production ready projects) certainly gives a similar vibe. :) [0] https://github.com/btwael/SuperString/commits/master

Probably because his benchmark is broken. It always copies to the end of the file. https://github.com/btwael/SuperString/blob/master/test/withS... this line should be ` lines.push_back(string.substr(last, i - last));` (it also drops the very last line, but who's counting, eh?). If the 210kb file has a line every 80 bytes or so and is copied this way, it works out to around 490MB of data (caveat: my math may be wrong).