From: Clifford Heath <no.spam@...>
Date: 2011-05-18T11:00:28+09:00
Subject: Re: LL(k) grammar for an interesting *subset* of Ruby ?

On 05/18/11 01:46, Jean-Christophe Le Lann wrote:
> Google gave me this : ruby-tp-dw-gram
> http://files.rubyforge.vm.bytemark.co.uk/ruby-tp-dw-gram/ruby-tp-dw-gram-0.1.tar.gz

Four years old (so no 1.9 syntax), a 0.1 version number, written in Lisp,
and claims *not* to parse Ruby correctly. Good luck with that! (really)

>> Are you aware of any kind of LL(k) grammar for an interesting *subset*
>> of Ruby ?
>> Anyone tried that with Citrus or Treetop ?

Nathan Sobo (author of Treetop, which I now maintain) intended to, but got
side-tracked into adding incremental parser facilities into Treetop so that
he could implement a syntax-directed editor for Ruby... I don't think the
actual Ruby grammar ever got properly started.

It's definitely possible, and now, there's a Ruby test suite to use as well.

The advantage of using a PEG is that PEG grammars are composable, so you
could incorporate grammar rule definition into the core language, allowing
Ruby to extend itself into any language. I hoped this might eventuate, but
now, I think that Javascript and V8 is the right implementation target.

Clifford Heath.