regex status report #10

[Gabor Kovesdan]

Hi,

I reworked the fixed string matching code quite a bit and now it seems 
to run correctly without segfaults. I also made some cleanups, added 
support for REG_ICASE, which was missing so far. In this current state, 
now it runs significantly faster for fixed string pattern than the 
unpatched TRE. It still uses the quick search algorithm, now I am 
experimenting with Boyer-Moore to get even more out of it. It is quite 
important to do this well because this will also be the foundation of 
the heuristical matching, which is the next major step. I'm testing the 
performance with BSD grep but it may have some other bottlenecks so it 
may be necessary (and useful since the ultimate goal is to get rid of 
the GNU bits) to look at it, as well.

I arrived back to Hungary yesterday from my Portuguese internship so 
probably I'll make a bit less progress during the next week until I 
settle down here again but I'll try my best.

Gabor