[boost] Re: Interest in a Recursive Descent Parsing Library

From: Larry Evans (cppljevans_at_[hidden])
Date: 2004-12-24 08:42:22

On 12/24/2004 12:13 AM, christopher diggins wrote:
[snip]
> I would not say YARD is a better parser than Spirit, it is designed with 
> quite different goals. YARD is designed to be compact and flexible. 
> Concerning "first/follow set kind of things" I am not sure what you 
> mean. 

They're used to determine the next non-terminal to parse by looking
ahead 1 terminal symbol (for LL[1] parser).

Informally, they're defined for each non-terminal in the language.
For a non-terminal, N, FIRST(N) is the set of non-terminals which
can start a sentence derived from N.  Similarly, FOLLOW(N), is
the set that can follow that sentence.  For a formal definition,
see the paper:

   A Strong LL(k) Parser Generator That Accepts Non-LL Grammars
   and Generates LL(1) Tables: Extended Abstract

available at:

   http://www.iit.edu/~tc/llk.htm

for an implementation, see:

http://cvs.sourceforge.net/viewcvs.py/boost-sandbox/boost-sandbox/boost/grammar_pipeline/

actually, that's been replaced by one more in line with spirit (more
meta-programming).  I could upload it to the sandbox if you're
interested.