Re: NEW ISSUE: Methods and Caching (proposal)

Henrik Nordstrom wrote:
> On ons, 2008-11-19 at 12:15 +0000, Jamie Lokier wrote:
> 
>> There is no harm in other method responses being cachable if the
>> server says they are, especially if an Etag is used so the origin
>> server is always consulted for accuracy.
> 
> etag values in requests refer to the "selected representation", not the
> method specific response.

Yes, and we will need to define what the selected representation is.

> Basing the selected representation on the method used does not make much
> sense imho. 

Sure. That would break for instance conditional headers on WebDAV 
requests, such as LOCK.

> ETag headers in responses also refer to the selected representation the
> response either contains or relates to, and is meant to be able to use
> as validator cross methods for the same URI.

I may agree with this, but we certainly need to do better stating it.

BR, Julian

Received on Wednesday, 19 November 2008 14:18:51 UTC