On Thu, 15 Sep 2005 15:23:52 +0100 "Chris Smith" <chris@xxxxxxxxxxxxx> wrote: > Hi, > > Before I go the whole hog and raise a bug report :) See http://bugs.splitbrain.org/?do=details&id=538 ;-) > What I find very odd. > - differences between my reference install and splitbrain fixed it partly today you probably missing a pull > - if I change my dev install to use '\b' I get the same results as > splitbrain. If I use current boundary setting I get same results as > my reference install. It originally used \b as boundary but this made problems with UTF-8 words. I then fixed it to a custom character class but messed up with lowercase ASCII chars. This was fixed with my patch I pushed today. But it still doesn't work as intended and I don't know why. > Also what is the reason the acronym pattern itself is designated a > non-captured group? This is used to group the or'ed acronyms. The lexer doesn't allow captured grouping as far as I remember but the non captured ones //should// work (I used them somewhere else I think). I haven't the slightest idea what goes wrong - maybe we've discovered a bug in the lexer. Andi