- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Wed, 15 Jan 2014 15:34:24 -0800
- To: John Cowan <cowan@mercury.ccil.org>
- Cc: Richard Ishida <ishida@w3.org>, Simon Sapin <simon.sapin@exyr.org>, www-style <www-style@w3.org>, WWW International <www-international@w3.org>
On Wed, Jan 15, 2014 at 2:24 PM, John Cowan <cowan@mercury.ccil.org> wrote:
> Tab Atkins Jr. scripsit:
>> @charset 'iso-8859-1';
>> .foo© { color: red; }
>>
>> Totally valid CSSCharsetRule shows up in the stylesheet. Assuming
>> that the file is actually saved in iso-8859-1, though, the @charset
>> has zero effect, and the © symbol is corrupted. Remove the extra
>> whitespace, change to double-quote chars, and it'll work.
>
> Feh. Do we really suck so much that this result is unavoidable? I'd like
> to see the syntax definition of the @charset rule narrowed to match the
> @charset marker. Is that impossible?
Doing so is more trouble than its worth. The CSS tokenizer doesn't
care what string-quoting character was used, or how many whitespace
characters are in a run, so that information isn't preserved into the
parsing stage. Let's not add layering violations just to "fix"
something that still wouldn't work anyway (since it *still* doesn't
actually do anything, despite being mutable/constructable).
I still have an action from months ago to investigate actual usage of
CSSCharsetRule, to see if we can just drop it entirely from the CSSOM.
I'll bet we can.
~TJ
Received on Wednesday, 15 January 2014 23:35:13 UTC