Re: Guile and translators

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

>>>>> "Andrew" == Andrew Archibald <aarchiba@undergrad.math.uwaterloo.ca> writes:

    Andrew> - The rx bindings need to be fixed, and rx needs to
    Andrew>   compile properly.

    Andrew>  I have fixed this, although if rx is to be dynamically
    Andrew>  loaded, I don't know how to make it install to the
    Andrew>  correct place.  In fact, I'm not sure where the correct
    Andrew>  place is.

 What about `pcre', the Perl Compatible Regular Expression library?

 You can download the source from:

 http://cgi.debian.org/www-master/debian.org/Packages/unstable/libs/libpcre1.html

 I'd never seen code for a FA before this, and found it fairly easy to
 understand.  It implements perl compatible regular expressions,
 complete with lookahead assertions, negations, non-greedy
 quantifiers, and numeric quantifiers.  According to the author's
 benchmarks, it's faster than Henry Spencer's regexp engine also.

 The Perl regexp's are certainly more evolved than rx, aren't they?
 Wouldn't it be neat if Guile/CTaX had that?

 Perhaps there should be a regexp reader syntax as well, rather than
 over-toothpicked double quoted strings.  Perl does a good job at
 that; perhaps #?re? like perl uses would be good.

 I remember a year ago (???) on the scheme newsgroup seeing something
 from someone about how he was working on a new regexp syntax of some
 kind.  Most of it was over my head the day I read it; I lost the
 article.  I'm very cureious about that.