NOTE: I'm replying manually since I can't deal with Savannah. > From: Paolo Bonzini <[email protected]> > Date: Mon, 08 Mar 2010 18:22:01 +0000 > Subject: [patch #6899] Speed-up for searching in multibyte and ignore-icase. > > Especially now that there is a > good DFA-based matcher in glibc,
Has anybody proven this? There was a period of time where I removed dfa from gawk because "it wasn't needed anymore now that regex has a dfa matcher" but I had to reinstate it - the regex matcher was intolerably slower. I added the GAWK_NO_DFA environment variable so that it'd be possible to test using just the regex matcher and I think grep now has a similar variable. Paolo - the enthusiam and progress in dfa is great - but please bear previous experience in mind so that we don't make any changes we'll regret later. Thanks, Arnold
