I´m actually developping a NLP appication (a morphological analyzer) I´ve implemented the tokenizer using boost regex++ engine, but the performance is poor in comparition with Perl´s one for the same task. I´ve read some messages about this questions in the mailing list, and then realized that boost regex++ gives really more than what i needed, and the price I have to pay is a lack of speed. But I wouldn´t like to mistake twice, so I´ve decided to make to all of you the question: Is there something I could change to enfinciently use boost regex in NLP programming? If not, what´s the better(at least better than Perl!) regex engine in C/C++ for NLP programming (and,of course... OPEN SOURCE) ? Thank you. (and sorry for my english) Bye!