Re: NEW ISSUE: Transfer-Encoding in 1.0 messages

We've also briefly discussed folding RFC2145 into 2616bis, or at  
least expanding upon the reference to make it more prominent and give  
more context.


On 23/11/2007, at 2:17 PM, Henrik Nordstrom wrote:

> On fre, 2007-11-23 at 18:23 +0100, Bjoern Hoehrmann wrote:
>
>> Your reading of the specification requires to assume that it does not
>> explicitly discuss this case for a reason. However, it might just not
>> discuss it because the case has been overlooked, and some would see
>> the conflicting implementation behavior as security problem. I  
>> disagree
>> that doing nothing about this in the specification is the best  
>> course of
>> action.
>
> I am fine with adding yet another comment mentioning that for
> interoperbility reasons HTTP/1.1 implementations SHOULD follow the
> HTTP/1.1 specifications even if seeing HTTP/1.0 messages except where
> specifically noted in the HTTP/1.1 specifications.  HTTP/1.1  
> clients or
> servers SHOULD NOT downgrade their implementation to the informal
> HTTP/1.0 specification when seeing an HTTP/1.0 message but continue
> processing that message according to the rules defined in the HTTP/1.1
> specifications.
>
> Would that solve your concerns in terms of specifications?
>
> Regards
> Henrik


--
Mark Nottingham     http://www.mnot.net/

Received on Monday, 26 November 2007 23:39:11 UTC