font-lock complex format questions

[David Abrahams]

If I have a format in which different subexps of a single matcher can
be used to recognize multiple elements, is there a provision for
font-lock to run the matcher only once?

Thinking more ambitiously, I have a complex grammar that font-lock
doesn't handle very well by itself.  One of the problems is that it
can be hard to decide how far back in the buffer to go before starting
to identify/refontify.  That might be easier if I could leave clues
around as text properties.  Another problem is that I really need to
parse nested constructs.  Once you've recognized the outer ones, you
have recognized the inner ones as well, so you might as well do it all
at once.  Are there any examples out there for me to follow that could
help with this?

Thanks,
Dave

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com