Re: SHOULD-level requirements in p6-caching

On 08.04.2011 12:11, Poul-Henning Kamp wrote:
> In message<4D9EDB5D.2020805@gmx.de>, Julian Reschke writes:
>> On 08.04.2011 11:04, Poul-Henning Kamp wrote:
>
>>> That web-apps expect caches to either use the cookies as part of their
>>> hash-key or not cache at all when cookies are present, or they break
>>> seven ways before breakfast.
>>
>> Ack.
>>
>> This is similar to the special case we have in P1 with respect to
>> handling multiple Set-Cookie header fields
>> (<http://greenbytes.de/tech/webdav/draft-ietf-httpbis-p1-messaging-13.html#rfc.section.3.2.p.7>).
>>
>> So we probably should mention it, but adding a requirement might break
>> existing conforming implementations, so I'd prefer not to.
>
> Each cookies should have an attribute that tells if they should
> affect the Vary/hash-key or not.

Well, if we were to change cookies-as-used-in-the-wild, then we just 
should tell producers to send the Vary header if they need to.

Best regards, Julian

Received on Friday, 8 April 2011 10:16:53 UTC