Re: find (and friends) bug?

[Tom Lord]

       > me:
       > Also, am I alone in thinking that accurately implementing Posix
       > regexps is a desirable goal for the GNU tools?

       > eggert:
       No.  It's just that nobody has gotten around to doing it.  


I don't think that's historically accurate.

GNU regex uses an NFA algorithm.  To give is Posix semantics is to
make it too slow.  Many years ago, this was briefly done, then undone.
Maybe it never propogated to all packages that include regex.c, but it
was done in Emacs, for example.

That's part of why I started Rx.  For many years, Rx was inaccurate
(didn't really implement Posix).  The version currently distributed by
the FSF is still inaccurate (more "process problems" there).  The
latest versions (at www.regexps.com) seem to be the most accurate
implementation of Posix regexps available and to have decent
performance.  For those packages that combine dfa.c from grep with
regex.c, Rx is an interesting alternative: it provides a dfa, when
applicable, via a posix regex interface (plus extensions, if you want
them).  Of course, if Posix has recently changed out from under Rx,
well then, there's a waste of a few years....

-t