On Fri, 13 Apr 2001, Eric Liao wrote:

> Robert,
>
Eric,

> Cool!  I anticipate the major release of this rockit Ruby parser
> generator very much.  
>
Great to hear that! It should be coming along shortly. I'll send you an
alpha release in the mean time.

> A couple of questions: does it resemble Perl's
> parse::recdescent in any way? (if you happen to know about it.)  Is it
> a top-down or bottom-up design?  Is the grammar LL(n) or non-LL?
> 
No. Bottom-up. Non-LL (LALR(1)).

But ANTLR and Parse::RecDescent looks like neat ideas so maybe we can add
them later (rockit is meant as a toolkit so shouldn't decide what
grammars/parse gens you'd like to use). Anyone working on translation of
ANTLR or Parse::RecDescent generators?

(My main concern with going the LL(k)-gen way is that people mostly seem
to prefer it since it gives "human-readable" parsers which is not a major
priority for me. IMHO, a parser should be generated as part of a
make/build process and not once and then hand-tuned. But maybe I'm wrong
here? And it seems for example ANTLR has other nice properties that
actually makes it stronger than LALR(1)? Comments?)

Regards,

Robert