Subject: Re: the state of regex(3) (was: Policy questions)
To: Ben Harris <>
From: Greg A. Woods <>
List: tech-userlevel
Date: 01/01/2004 16:50:37
[ On Thursday, January 1, 2004 at 20:14:14 (+0000), Ben Harris wrote: ]
> Subject: Re: the state of regex(3) (was: Policy questions)
> PCRE's license is practically 4-clause BSDish, even if the words are
> different.  I suspect if there were anything that needed changing to make it
> acceptable to TNF, Phil would be happy to consider it.

OOPs!  I should have re-checked the latest copyright on PCRE instead of
assuming it had been made GPL as Exim has been.  Thanks for correcting me!

So, if that's the case then maybe that's the direction to head in.  PCRE
is certainly very well maintained, quite mature as such things go, and
as I said it's performance is quite astounding and it's very well
documented.  For all that it offers it's still quite small too, and it
has a reasonable memory footprint during execution as well.

One of the features I really like about PCRE's ERE syntax (which I
suppose is a generic feature first introduced by Perl's EREs) is the
ability to set or unset various options, even within a pattern, using
the "(?)" construct.  This is something I've wanted to be able to do
ever since the day I first learned of egrep's EREs oh so many years ago!

I guess the biggest question then is how the change in ERE syntax
support in NetBSD might be facilitated.  I can imagine that I'd be able
to "just do it" for my own systems, but given the ultra-conservative way
TNF seems to approach this kind change I'm guessing there might be some
hurdles to jump.  I haven't really considered yet whether Philip would
consider supporting a version of PCRE that could by default provide just
POSIX ERE syntax compatible with the existing library, however given
what he says in the pcreposix manual page I've been assuming he wouldn't.

						Greg A. Woods

+1 416 218-0098                  VE3TCP            RoboHack <>
Planix, Inc. <>          Secrets of the Weird <>