There's also the issue that gawk uses dfa.c etc., so any changes we make to grep should be done with gawk in mind.
Thanks, wow. I like the idea of splitting things apart, as dfa.c is
indeed too large and tricky. This'll have to wait until after the next
grep release, at least in terms of investment of time that I can make in
terms of reviewing. At this point we have six performance patches in
the queue (17136, 17203, 17204, 17229, 17230, 17240) and my hope was
that we can put out the next version once they're reviewed.
