From: Clifford Heath <no@...>
Date: 2008-01-29T14:19:58+09:00
Subject: Re: Treetop Email Parser

Phrogz wrote:
> On Jan 28, 8:41 pm, Phrogz <phr...@mac.com> wrote:
> My apologies for second guessing you. Having read the PEG paper[1], I
> see now what you're saying.

Well, though the authors of the packrat paper described a parser that
behaves that way, I don't think it's fair to limit the possible other
behaviours of any PEG parser generator. IOW, PEG >= packrat. ANTLR is
also LL(*), but avoids this problem by analyzing the expressions and
either rejecting or re-writing them.

> So...how would you write a PEG rule to match "a string of any number
> of vowels, that ends with an 'e'"?

Assuming some EOF token (not provided free by Treetop) you'd write:

(!('e' EOF) [aeiou])* 'e'

Of course, in many cases, EOF would be replaced by a white-space or
other rule.

Clifford Heath.