Re: p6: Vary and effects on future requests

[Mark Nottingham <[e-mail]>]

These don't seem controversial, so I'll mark for -23.


On 29/04/2013, at 10:39 PM, Mark Nottingham <[e-mail]> wrote:

> On a related note, the text in p6 4.3 doesn't explicitly account for the case when there are stored responses both with and without Vary that might match.
> 
> I think this can be addressed by changing:
> 
> "If multiple selected responses are available..."
> 
> to
> 
> "If multiple selected responses (potentially including responses without a Vary header) are available..."
> 
> 
> On 29/04/2013, at 3:13 PM, Mark Nottingham <[e-mail]> wrote:
> 
>> p6 4.3 says:
>> 
>> """A Vary header field value of "*" always fails to match, and subsequent requests to that resource can only be properly interpreted by the origin server."""
>> 
>> Someone reading this could take away that *any* stored response with a Vary: * has an effect on ALL other stored responses that share that primary key (and a couple of the issues we discussed reflected this confusion, from memory).
>> 
>> As I recall, the goal for 4.3 Using Negotiated Responses was to assure that stored responses were independent and self-describing; i.e., that a cache didn't have to rationalise the value of the Vary header among multiple stored responses.
>> 
>> So, I'd suggest changing this to just:
>> 
>> "A Vary header field value of "*" always fails to match."
>> 
>> Make sense?
>> 
>> 
>> --
>> Mark Nottingham   http://www.mnot.net/
>> 
>> 
>> 
>> 
> 
> --
> Mark Nottingham   http://www.mnot.net/
> 
> 
> 
> 

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