Re: Proposal for #486: Requiring proxies to process warn-date

On 03/09/2013, at 4:46 PM, Julian Reschke <julian.reschke@gmx.de> wrote:

> RFC 2616 was the first spec to define "Warning" -- what is the backwards compat story here? After all, an 1.0 intermediary will not even know what "Warning" is.
> 
> What am I missing here?

The actions of a HTTP/1.0 cache can affect the validity of a warning. For example, if a response has the warning "This response is stale" and a HTTP/1.0 cache revalidates it, the response is no longer stale, but the 1.0 cache will not have removed the warning. 

Warn-date is a mechanism to detect this situation; if warn-date is different to Date, we know that the warning is no longer valid.

However, 2616 places the requirement for detecting it upon *every* sender, including intermediaries, whether or not they cache, whether or not they do anything with Warning. This is pretty clearly Bad Protocol Design, because of the aforementioned misalignment of incentives.

Cheers,

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

Received on Tuesday, 3 September 2013 07:24:11 UTC