- From: Zack Weinberg <zweinberg@mozilla.com>
- Date: Mon, 29 Dec 2008 19:31:24 -0800
- To: fantasai <fantasai.lists@inkedblade.net>
- Cc: Andrey Mikhalev <amikhal@abisoft.spb.ru>, www-style@w3.org
fantasai <fantasai.lists@inkedblade.net> wrote:
> Andrey Mikhalev wrote:
> > On Mon, 8 Dec 2008, fantasai wrote:
> >> Unless I'm missing something we need to replace
> >> : selector [ COMMA S* selector ]*
> >> with
> >> : selector [ S* COMMA S* selector ]*
> >
> > i think (in css2.1) tokens
> > {w}"," {return COMMA;}
> > ...etc.
> > intended for elimination parser state conflict between
> > selector S+ selector
> > and
> > selector S* COMMA S* selector
> > ...etc.
>
> Thanks, Andrey. I think you're right. I guess that solves
> the grammar problem, then. :)
Objection: the lexical analysis rules are already too complicated, and
it is already too hard to tell whether the lexical rules in section 4
are consistent with the rules in appendix G. We should not add any more
special token productions.
I see what Andrey is saying, that your original suggestion requires
two tokens of lookahead in the parser, but I much prefer that to an
even messier lexer. (CSS2.1 appendix G asserts that the grammar is
LALR(1) but I strongly suspect this is not true for a parser that
honors the section 4 conventions as well.)
zw
Received on Tuesday, 30 December 2008 03:32:08 UTC