On Fri, 25 Aug 2006 12:53:24 +0100 Chris Smith <chris@xxxxxxxxxxxxx> wrote: > I have sent a patch through to Andi which revamps the ft_snippet() > function. This function, or rather the preg_match_all() within it > seems to be the main factor in the search page execution time. Great, this is one of the things I mentioned at the helpneeded page. I just had a look at your code and I really like it. One question I'm not sure about: I noticed the use of some strlen calls there. Are they used in a UTF-8 safe way there or would it be possible that they split a multibyte char? If that could happen we should add a check to strip invalid UTF-8 chars from beginning and end of the snippet - this would be a nice addition to the utf-8 lib. BTW: I didn't see any differences in the snippets generated by opt1 and opt2. Do you have any ideas in which directions we should look to see if there are problems? Andi -- http://www.splitbrain.org -- DokuWiki mailing list - more info at http://wiki.splitbrain.org/wiki:mailinglist