Michael Sperber <sperber <at> informatik.uni-tuebingen.de> writes:
> Emilio Lopes <eclig <at> gmx.net> writes:
>
> > I tracked the problem down to this change in scsh/rx/parse.scm, in the
> > procedure `char-set-ein-pair':
> >
> > <at> <at> -709,7 +714,7 <at> <at>
> > . ,ranges))))
> > (values loose ranges)))))
> >
> > - (let lp ((i 127) (from #f) (to #f) (loose '()) (ranges '()))
> > + (let lp ((i 255) (from #f) (to #f) (loose '()) (ranges '()))
> > (if (< i 0)
> > (add-range from to loose ranges)
> >
> > I've not found any evil effects after reverting this one change some
> > months ago (dreimal klopfen auf den Kopf!).
>
> Except you get unwanted effects with non-ASCII characters.
Michael is right. The right thing to do is to link Scsh against a
POSIX compliant regexp library. One can use the GNU regexp library,
as described e.g. in http://www.openldap.org/faq/data/cache/148.html.
It's really simple and worked for me on our old Cygwin version. I
guess it would also work on OpenBSD.
I'll upload Cygwin binaries for Scsh-0.6.7 to my web site in the next
days.
--
Emílio C. Lopes
Munich, Germany
|