On Fri, 13 Apr 2007 09:30:07 +0900, Tim X <timx / nospam.dev.null> wrote:
> True, but I think thats due to all the additional add-ons. The basic
> underlying algorithm is/was fairly straight forward.

I was thinking in terms of the optimizations applied to make the backtracking algorithm faster, though that's a factor too...

> However, I think even the perl developers have realised that Perl's RE support
> was getting out of hand.
> Although I've not looked at it, I believe there is significant changes in
> perl 6 in this area. I seem to remember reading something from Larry, where he
> said that Perl's functionality in this area has extended to the point that in
> reality, you probably couldn't call it regexps anymore, but as that was
> the terminology people were use to, thats what it wold be known as. Perl has
> gone past what was traditionally defined as regexps.

The Perl 6 solution was to promote grammars and rules into first-class abstractions in the language, on a level with classes and methods.  So it's basically a general-purpose parsing framework now.  I'm not sure if that counts as less out-of-hand or not...

-mental