Re: [CSS21] [css3-syntax] Noncharacters

Ernest Cline wrote to <mailto:www-style@w3.org> on 28 November 2003 in 
"RE: [CSS21] [css3-syntax] Noncharacters" 
(<mid:410-22003116290014218@mindspring.com>):

>> From: Etan Wexler <ewexler@stickdog.com>
>>
>> How should CSS processors react to the presence of
>> a noncharacter code in a style sheet?
>> How do CSS processors react to the presence of
>> a noncharacter code in a style sheet?
>> How should CSS processors react to a reference to
>> a noncharacter code  in a style sheet?
>> How do CSS processors react to a reference to
>> a noncharacter code in a style sheet?
>>
>> By "noncharacter" I mean codes like unpaired surrogates or U+FFFF.
>
> [... The] following
> sentence from CSS3-Syntax (Section 4.2 Tokenization) is I believe
> on target:  "If at some point it is not possible to continue tokenizing
> an incoming style sheet, the remainder of the style sheet should be
> ignored and only the largest initial segment of the style sheet that
> can be tokenized according  to the above rules."
>
> Your first question thus becomes should Unicode non-characters
> be tokenizable?  Common sense would say no, but as far as I can
> tell the normative  grammar in CSS 2.1 Appendix G would say yes.

I'd like to see actual noncharacters handled at a low level, in a 
process that is, at least in concept, separate from and prior to the 
parsing. I want noncharacters dropped. Silent dropping is fine, as is a 
warning.

After dropping the noncharacters, parsing can begin. The style sheet at 
this point consists only of characters, although it may contain 
references to noncharacters. What to do with those references is a 
trickier question.

-- 
Michel and Martin: I send you this message because the matter wants 
attention by Unicode experts. You two came to mind as Unicode experts 
who are involved in CSS. Please let us know if you'd prefer not to 
participate in this discussion.

Etan Wexler.

Received on Friday, 28 November 2003 19:43:44 UTC