Re: Content Sniffing impact on HTTPbis - #155

OK. Please regenerate the diff with this change. It doesn't sound like  
people feel we need to close off the possibility of sniffing encoding,  
so I think the rest can stay the same...

On 06/06/2009, at 8:23 PM, Julian Reschke wrote:

> Mark Nottingham wrote:
>> ...
>>> How about:
>>> "Any HTTP/1.1 message containing an entity-body SHOULD include a  
>>> Content-Type header field defining the media type of that body,  
>>> unless that information is unknown."
>> I read that as effectively equivalent to the current text, but  
>> sure; it is more explicit about when the SHOULD can be violated.
>> ...
> I'm pretty sensitive to this kind of SHOULD requirements, because of  
> all the crap I've seen servers putting into WebDAV metadata, because  
> the implementors thought they had to (like providing content  
> language info defaulting to the server's locale, or time stamps  
> defaulting to Jan 01, 1970).
> BR, Julian

Mark Nottingham

Received on Sunday, 7 June 2009 01:02:34 UTC