tree v text parsing (was Re: [antlr-interest] Article against TreeWalkers)

Terence Parr parrt at
Fri Mar 10 10:01:32 PST 2006

On Mar 10, 2006, at 9:46 AM, Gregg Reynolds wrote:

> Andrew Bell wrote:


Very well put.  recognition is recognition.

> So I guess I must quibble with Terence on one point:  I don't see the
> need for 1d or 2d etc. to enter into it.  All parsing is the same.  Or
> more accurately, all grammars map source sequences to target  
> sequences,
>   regardless of what the elements of those sequences represent.   
> Usually
> the target sequences represent the interpretation of a hierarchical  
> (2d)
> structural grouping of source elements.  You could have an n- 
> dimensional
> tree, and still write a grammar for it that doesn't differ  
> fundamentally
> from a programming language grammar.

Agreed...I'm "playing to my audience" a bit.  Parsing is about a  
stream of symbols; how you get the stream is up to you.  It's easy to  
say that you are taking a 2D structure like a tree and serializing to  
a node stream in 1D to help people understand how the tree parsing  
works. :)


More information about the antlr-interest mailing list