Re: [CSS21] out of range unicode escapes

At 01:30 07/06/12, fantasai wrote:

>As Mark pointed out, replacing the invalid Unicode escape with the replacement
>character would have that effect anyway, unless it were in a string value such
>as for the 'content' property. I think that solution has the advantage of being
>much simpler and also doing what we want (making the property declaration etc
>invalid) in most cases.

Looking at it this way, this definitely seems to be a good way forward.

As an example, such code points wouldn't match in selectors, because there
is nothing they could match to. But the problem with this that yo're relying
on the document language to not pull the same convenient trick. As an example,
if HTML 5.0, or some other new markup language, decided to do the same, you
would end up both with matching corresponding errors (same out-of-range number)
and with matching unrelated errors (different out-of-range number).

It's not clear to me that it's worth worrying about such a case, but the odds of
it definitely should be considered.

For properties and other stuff mentioned at
http://www.w3.org/TR/REC-CSS2/syndata.html#parsing-errors,
things definitely should work out fine.

For string values (as well as for other cases), I'd like it to be possible
for a validator or some such to point out the error. So the final wording
that the CSS WG adopts should allow that, rather than giving ammunition
to somebody who might come and say "CSS says you convert this to a replacement
character, so it's not an error, so this validator is in error flagging it
as an error".

Regards,    Martin.


#-#-#  Martin J. Du"rst, Assoc. Professor, Aoyama Gakuin University
#-#-#  http://www.sw.it.aoyama.ac.jp       mailto:duerst@it.aoyama.ac.jp

Received on Tuesday, 12 June 2007 04:49:13 UTC