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

Re: [CSS21] Please make the BOM take highest precedence for character encoding information for text/css (higher than HTTP)

From: Leif Arne Storset <lstorset@opera.com>
Date: Wed, 24 Oct 2012 17:42:18 +0200
To: www-style@w3.org, "Florian Rivoal" <florian@rivoal.net>
Message-ID: <op.wmowkssbtmo5g6@localhost.localdomain>
On Sun, 21 Oct 2012 10:40:01 +0200, Florian Rivoal <florian@rivoal.net>  
wrote:

> Henri Sivonen <hsivonen@iki.fi> wrote:
>
>> Regarding: http://www.w3.org/TR/CSS21/syndata.html#charset
>>
>> Letting HTTP-level character encoding information take precedence over
>> a BOM never leads to sensible results, because the bytes that a BOM
>> consists of never occurred for non-BOM purposes at the start of a
>> file. So far, it seems that making the BOM override HTTP for HTML and
>> JavaScript makes a Web engine more Web-compatible. The BOM overrides
>> HTTP for HTML in IE up to and including 9 (but not 10), in Chrome and
>> in Opera. The BOM overrides HTTP for JS in IE7 (but not in 8, 9 or
>> 10), in Chrome and in Opera.
>>
>> While I am unaware of Web compat evidence for CSS, it would make CSS
>> more consistent with the rest of the platform to make the BOM take the
>> highest precedence for CSS as well. Opera and Chrome already give the
>> BOM the highest precedence for CSS.
>>
>> I did not test Safari.
>>
>> Please change CSS to give the highest precedence to the BOM (UTF-8,
>> UTF-16 little endian or UTF-16 big endian).

[rethreaded:]

> On the face of it, it sounds like a good idea.
>
> Given that some browsers, as you said, already behave that way, I would  
> like to encourage the vendors of these browsers to check their bug  
> database to verify that this behavior isn't a source of problems.

It doesn't appear to have caused us problems since we released this change  
five months ago. We have been testing this internally since June last  
year, although neither of those timeframes give abundant time to catch  
issues.

-- Leif Arne Storset
Opera Software
Oslo, Norway

>> Gecko bug for the CSS case:
>> https://bugzilla.mozilla.org/show_bug.cgi?id=796882
>>
>> Gecko bug for the JavaScript case:
>> https://bugzilla.mozilla.org/show_bug.cgi?id=687859
>>
>> Gecko bug for the HTML and XML case:
>> https://bugzilla.mozilla.org/show_bug.cgi?id=716579
>>
>> Spec bug for the JavaScript case:
>> https://www.w3.org/Bugs/Public/show_bug.cgi?id=19623
>>
>> Spec bug for the HTML case:
>> https://www.w3.org/Bugs/Public/show_bug.cgi?id=17810
>>
>> Test case for HTML:
>> http://hsivonen.iki.fi/test/moz/bom/html.html1251
>>
>> Test case for CSS and JS:
>> http://hsivonen.iki.fi/test/moz/bom/no-charset-attribute.html1251
>> (Check baseline for :after support:
>> http://hsivonen.iki.fi/test/moz/bom/baseline.html )
Received on Wednesday, 24 October 2012 15:43:17 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:21:01 GMT