Re: [CSS21] DELIM? in core syntax

Anne van Kesteren wrote:
> "Parts of style sheets that can be parsed according to this grammar but 
> not according to the grammar in Appendix G are among the parts that will 
> be ignored according to the rules for handling parsing errors."
> "are among" is essential here, I think.

I see. I assumed that was referring to the fact that illegal properties 
(say "font-variant") /could/ be parsed with the CSS 2.1 grammar, but 
would still be handled according to the rules for handling parsing 
errors. It's just not very natural for me to read that as leaving open 
the possibility that parts of style sheets that cannot be parsed 
according to this grammar will be ignored according to the rules for 
handling parsing errors, especially when it adds: "The meaning of input 
that cannot be tokenized or parsed is undefined in CSS 2.1."

> Given 
> that implementors have understood this part of the specification I 
> suggest we not change is as that will only increase the chance of 
> introducing errors at this point.

Some have; some perhaps haven't. As I mentioned, the SACParser the Lobo 
browser uses skips whole blocks on encountering *property rather than 
skipping to the next declaration.

I can appreciate the incentive of resisting substantive changes at such 
a late stage. Changing the wording to reflect intention can always 
introduce new errors into the spec. On the other hand, changing the 
wording could also reduce errors in new implementations, which might be 
more important. It does seem to me there's a balance to be struck there, 
and looking at the most important implementations may not be the way to 
strike it.

Still, it's great to get confirmation that the removal of DELIM? doesn't 
mean *property has undefined (error) handling; thanks for clearing that 
up for me. :)

Benjamin Hawkes-Lewis

Received on Saturday, 5 April 2008 14:36:46 UTC