Berkeley Regex - was: fixincludes jumbo patch 2

[Harvey J. Stein]

Bruce Korb <korb@datadesign.com> writes:

 >   > The issue is whether or not the Berkeley regex processing
 >   > is okay for inclusion or not.  It will apparently cut
 >   > processing time by another 60% ...
 > 
 > Per Bothner writes:
 > > 
 > > > I am nearly certain that Henry Spencer has a newer version of
 > > > this code available but I was unable to find it.

I did some regex lib comparisons about a year ago (for guile).  At
that time Henry Spencer had a new version of the libs which were more
posix & had better character set support.  However, the library was
much slower (on the test cases I tried) than the version of Spencer's
library included with TCL.

The fastest was the regex library included with awk, which was snarfed
from some version of glibc.  The regex lib in glibc on my system
(linux, Redhat 5.x or 4.x at the time) was in between.  You can
probably find my timing tests in the guile mailing list archive.

-- 
Harvey Stein
Bloomberg LP
hjstein@bfr.co.il