Re: HTML5 vs content type sniffing

Frank Ellermann wrote:
> Julian Reschke wrote:
>  
>> So, please have a look at
>>    <http://www.hixie.ch/tests/adhoc/http/content-type/sniffing/>
>> and supply feedback on the expected results
> 
> I don't understand test case 11, I'd expect the same outcome as
> for cases 9 and 10.

That's because HTML5 tries to restrict content sniffing to precisely 
three content type header values, see 
<http://www.w3.org/TR/2008/WD-html5-20080122/#content-type-sniffing>.

> For test case 15 I don't get why it differs from test case 8.

Same reason.

> Similar 16 vs. 9, and 17 vs. 10.  What are "invalid text/plain
> characters" with respect to Latin-1, something about C0 or C1 ?

Not sure. I just checked 016, and it doesn't contain any characters 
beyond 127.

> Treating certain case sensitive (sic!) variations of text/plain
> or text/plain;charset=iso-8859-1 as erroneous depending on the
> actual content is something that should never show up in an RFC,
> it is only a hack (apparently designed for a popular Web server).

I personally agree with that.

> News from the UTF-7 front, there's a volunteering area director:
> <http://permalink.gmane.org/gmane.ietf.apps-discuss/953>

Sounds good.

BR, Julian