W3C home > Mailing lists > Public > www-style@w3.org > January 2012

Re: [css3-syntax] CSS escape sequences

From: Boris Zbarsky <bzbarsky@MIT.EDU>
Date: Thu, 12 Jan 2012 11:25:31 -0500
Message-ID: <4F0F097B.9030807@mit.edu>
To: Mathias Bynens <mathias@qiwi.be>
CC: www-style@w3.org
On 1/12/12 10:58 AM, Mathias Bynens wrote:
>> If you're proposing paving cowpaths... is this what UAs actually do?
>
> Except for Gecko, yes.

Except your next comment is all about how they don't.

>> How do they handle unpaired surrogates?  Surrogates in the wrong order?
>
> Let’s find out: http://jsfiddle.net/mathias/wvPdr/
>
> In Gecko, Presto, Trident and WebKit, unpaired surrogates like
> `\ud83d\uffff ` are displayed as `ud83fduffff`, i.e. the backslash
> just makes the `u` character literal.

CSS doesn't use a 'u' prefix on escapes, so the first part of this 
testcase is just wrong.

I can guarantee that Gecko converts surrogates, whether paired or not, 
to U+FFFD.  You can see this if you remove the 'u's there.

Looks like WebKit treats unpaired surrogates just like mis-ordered ones: 
it drops the whole declaration.

Opera 11.52 on Mac seems to use some sort of "half a surrogate" 
replacement characters or something.  Certainly it's not using U+FFFD 
for either the mis-ordered or unpaired surrogates over hre.  Does it do 
something different on different platforms?  Or did they change behavior 
from 11.52 to 12a?

-Boris
Received on Thursday, 12 January 2012 16:33:39 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:48 GMT