Re: Clarifying Content-Location (Issue 136)

On 06/10/2009, at 3:56 PM, Brian Smith wrote:

> Mark Nottingham wrote:
>> Sent: Monday, October 05, 2009 11:07 PM
>> To: Brian Smith
>> Cc: 'Julian Reschke'; 'Robert Brewer'; 'HTTP Working Group'
>> Subject: Re: Clarifying Content-Location (Issue 136)
>>
>> When did that happen? We removed its base-setting ability, not cache
>> invalidation, IIRC.
>
> http://trac.tools.ietf.org/wg/httpbis/trac/changeset/691
>
> The above change removes cache invalidation based on Content- 
> Location for
> GET/HEAD. The part of P6-cache that discussed cache invalidation  
> based on
> Content-Location for PUT, POST, and DELETE was retained. That  
> doesn't make
> sense to me; it seems like we should either always invalidation  
> based on
> Content-Location (for the same host) or we never should. I support  
> "we never
> should" as it makes it easier to get rid of Content-Location.

Yes, but PUT/POST/DELETE have side effects, and so invalidation makes  
sense in this case.

The text removed in changeset 691 was very specific to a situation  
where C-L matched the request URI and the ETag and Date said it was  
different; it's a very different thing than the invalidation specified  
for methods with side effects.

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

Received on Tuesday, 6 October 2009 05:18:07 UTC