Re: doc rx regexps

[Tom Lord]

I'm not closely following this discussion but perhaps it 
would be helpful to point out that Rx is not currently 
distributed in an up-to-date form separately from other
projects.

Rx lives in something called `libhackerlab' which I maintain.
Up-to-date versions appear in projects such as GNU Arch.

I don't currently distribute `libhackerlab' separately.

So: if you have an Rx binding for Guile but aren't taking
the extra step of obtaining the latest Rx from my Arch archives
or from an Arch distribution, odds aren't bad that you'll wind
up with a known-bogus, older version of Rx.

I do happen to think that Rx is pretty great, if I am forgiven for
saying so myself.  It's a serious power tool: easy enough to 
hurt yourself with but also of very high utility.  My experiments
in Systas Scheme suggest that binding Rx to Scheme creates a 
functionally rich result and I strongly encourage people to explore
that combination.

I've lost track of where Guile is on shared substrings but I should
point out that Rx-in-Scheme is vastly more useful if at least
read-only shared substrings are provided.  (For example, the Posix API
to regexps returns integer offsets to indicate the positions of
matched subexpressions -- using shared substrings, strings can be
returned just as efficiently, sparing clients of the need to fuss with
integer offsets.)

But, one has to be cautious: be sure you are using an up-to-date Rx (and
that isn't *quite* trivial to do).

-t