12.07.2013, 20:09, "John Regehr" <[email protected]>: >> Btw, it might be possible to fix pass_peep::b by using possessive regexes, >> but since I don't completely understand what ids it supposed it do, I >> haven't >> tried to fix it. > > I think the regex-based approach is strictly worse than using a > tokenizer. I had a hateful time getting it to work even as well as it does. > > On the other hand, it does seem sort of backwards to delete string > manipulation code written in Perl and replace it with C :).
Rewriting poorly performing part of software in lower-level language is a well established practise. There's nothing really evil in pass_peep::b, it just should never be run on files larger than 30K. I have not measured performance of rm-toks-1 on such files, but I guess it scales much better. -- Regards, Konstantin
