Re: CVS script

[karlheg@inetarena.com (Karl M. Hegbloom)]

>>>>> "Jim" == Jim Blandy <jimb@red-bean.com> writes:

    Jim> [Greg sent the script below to me in personal E-mail,

    >> For comparison (performance testing, perhaps?), this was my
    >> (very very) quick hack to do the same thing a couple of week
    >> ago.  It uses Zsh and Perl (I'm sure it'd be faster if I just
    >> used one perl script, but, like I said, I was interested in a
    >> mostly one-shot script).  I'm a little surprised that things
    >> were so long in the guile version--- perhaps some of the
    >> wildcard matching functionality of Zsh would be nice to port to
    >> guile.
    >>
    >> Greg

 Even better, perhaps, the globbing that Zsh does ought to become part
 of the `glibc' globbing function(s), and guile should use that...  I
 guess you'd need to have it available in a compat lib for folks who
 aren't using our shared C libraries.  As usual, these days I still
 don't have the skills to actually code that.  I've quit wishing for
 them and have been reading...

    Jim> - Perl's -i construct is really winning for this particular
    Jim>   case.
    Jim> - Perl's -p construct is quite winning, too.

    Jim> Now, if Guile had equivalents of these, but with nice names
    Jim> and good semantics, I'd be happy.  The script still wouldn't
    Jim> be as short as yours, but it'd be more readable, and still in
    Jim> the range of plausibility.

 Have yous seen the `pcre' library?  It's a perl compatible regular
 expression library.  Its sources are very easy to understand.  It is
 supposed to be faster than Henry Spencer's regular expression codes.
 There's a Debian package of `pcre' available, so please look through
 <URL:http://www.debian.org> for the source.  (You can search for it,
 click "Debian Packages" under "Distribution" on the left.)

 It has occured to me that this would be a neat addtion to guile,
 perhaps with a reader syntax designed for it.  I'm not skillful
 enough to code it...  I'd love to learn how it's done and help debug
 it by looking at a finished beginnings of it.

 `pcre' supports negated regular expressions, non-capturing (and not
 counted) groups, non-greedy quantifiers, zero width lookahead
 assertions, and all of the other fancy things that Perl can do.

 I think we need this.  Anyone else?