- From: L. David Baron <dbaron@dbaron.org>
- Date: Sun, 17 Feb 2013 10:28:50 -0800
- To: Zack Weinberg <zackw@panix.com>
- Cc: www-style@w3.org
On Sunday 2013-02-17 13:08 -0500, Zack Weinberg wrote:
> First, white space's significance in selectors is a design error which
> should not be allowed to creep into other contexts. It would be good,
> therefore, to exclude whitespace tokens from appearing in any list
> other than the prelude of a qualified rule. This would also simplify
> many of the algorithms in this section, which would no longer need to
> mention that they discard whitespace.
There are two other cases where we've made whitespace significant
(both, perhaps, resulting from other bugs in the tokenization,
namely than - is an identifier character and that ident + "(" is a
totally different token):
* we require whitespace around + and - in calc()
* we require whitespace after 'not' and around 'and' and 'or' in
@supports
I prefer to require whitespace in these contexts than have the
bizarre behavior that would otherwise result.
> ยง5.2 (quirks mode): The quirks defined in this spec should be thought
> of as augmenting the set of acceptable values for particular property
> declarations; they do not affect the generic grammar's determination
> of where one declaration ends and another begins. Therefore, they
> belong with the definitions of <length> and <color>, or else with the
> definitions of the specific declarations to which they apply; they
> should not appear in this spec at all.
Absolutely; I've mentioned this multiple times.
-David
--
๐ L. David Baron http://dbaron.org/ ๐
๐ข Mozilla http://www.mozilla.org/ ๐
Received on Sunday, 17 February 2013 18:29:15 UTC