W3C home > Mailing lists > Public > ietf-http-wg@w3.org > April to June 2009

Re: Proposal: 205 Bodies [#88]

From: Adrien de Croy <adrien@qbik.com>
Date: Mon, 08 Jun 2009 23:19:38 +1200
Message-ID: <4A2CF3CA.5030004@qbik.com>
To: Mark Nottingham <mnot@mnot.net>
CC: "Roy T. Fielding" <fielding@gbiv.com>, Julian Reschke <julian.reschke@gmx.de>, HTTP Working Group <ietf-http-wg@w3.org>

from a proxy's POV it's easier to allow an entity body, even if 0 
length.  Otherwise you have to cater to the case of a message that must 
have an empty entity (and what to do if it's not).  Are there any other 
responses than 205 which fall into this category?

Do we also then need to cater for all those server writers who probably 
mis-read the spec and don't send Content-Length on 205?  Do we know if 
there are any?  Having just run into IIS 5 sending Content-Length: 0 on 
all 304 responses, I wouldn't be holding my breath that people are 
sending empty entities on 205.

I guess it's going to need special case handling in a proxy for that 
reason anyway, but it would be nicer not to have to code for a 
must-be-empty entity.  The proxy shouldn't need to care if there's an 
entity or not there anyway?  Or is it expected to clean it up?

Regards

Adrien


Mark Nottingham wrote:
> I.e., allow entity bodes on 205?
>
> *shrug* In the scheme of things, it's not that important, I suppose, 
> since they're not widely used. However, it seems you'd have the same 
> philosophical debates either way -- "what is the meaning of an entity 
> on a 205 response" vs. "what is the meaning of the entity headers on a 
> 205 response"?
>
>
>
> On 08/06/2009, at 9:03 PM, Roy T. Fielding wrote:
>
>> On Jun 8, 2009, at 12:43 PM, Julian Reschke wrote:
>>
>>> Roy T. Fielding wrote:
>>>> ...
>>>> Big objection.  205 was added late in the process of 2068 and
>>>> could not be grandfathered into the message parsing algorithm
>>>> as yet another (bad) exception.  The requirement that 205 not
>>>> include an entity means that the message-body MUST be of zero size
>>>> (i.e., Content-Length must be supplied with a value of 0
>>>> or Transfer-Encoding chunked is used with a zero-length chunk).
>>>> Hence, it is correct as specified, albeit confusing.  It will
>>>> be less confusing when the terminology is cleaned up.
>>>> ...
>>>
>>> Yes, I was wondering about that (and duplicated language about 
>>> special cases in Part 1 & 2).
>>>
>>> So, shouldn't we change part of the description for status 205 from
>>>
>>>     "The response MUST NOT include an entity."
>>>
>>> to
>>>
>>>     "The response MUST include a zero-length entity."
>>>
>>> ?
>>
>> I think that would lead to more philosophical arguments than simply
>> removing the sentence (it is a stupid requirement).
>>
>> ....Roy
>>
>
>
> -- 
> Mark Nottingham     http://www.mnot.net/
>
>

-- 
Adrien de Croy - WinGate Proxy Server - http://www.wingate.com
Received on Monday, 8 June 2009 11:16:53 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:03 GMT