From: Eric Mahurin <eric.mahurin@...>
Date: 2008-02-06T23:46:18+09:00
Subject: Re: Parsing JSON (#155)

------=_Part_15838_468867.1202309181590
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

On Feb 6, 2008 2:05 AM, tho_mica_l <micathom@gmail.com> wrote:

> > I think the benefit of Regexp goes down the more of them you are using.
>
> They are a convenient solution for a wide range of problems.


You are taking my statement out of context.  I was referring to
performance.  If you are using a bunch of little regex's, you won't see much
performance benefit compared to matching a character at a time with a 1
character lookahead (LL(1) parsing).  I think you can write a pure ruby
LL(1) lexer/parser that usually meets or beats a regex version.

But yes, using regex's are are convenient for a wide range of problems.

With respect to extending ruby's regexp, I'd rather be interested in
> embedded code snippets that are evaluated when a clause matches. I
> didn't know perl (http://perldoc.perl.org/perlre.html) does this until
> I read about it in the ragel manual. It seems there are two kinds of
> embedded code: (1) evaluate the code and match with null width; (2)
> use the return value of the code as regexp. This sound quite
> interesting to me. It's currently listed under "A-3. Lacked features
> compare with perl 5.8.0" in the ruby RE manual.
>

Yep.  I used this in perl 5+ years ago (before coming to ruby) to write
recursive regex's (i.e. paren matching).

But, this doesn't address the issue I was referring to - applying a Regexp
to something other than a String.

------=_Part_15838_468867.1202309181590--