Re: Last Call: <draft-snell-http-prefer-14.txt> (Prefer Header for HTTP) to Proposed Standard

[James M Snell <jasnell@xxxxxxxxx>]

I am certainly open to alternatives on this particular point. The wait preference has proven to be quite useful in environments where the latency is low and predicable and there is good clock synchronization between the client and server. Such conditions can be easily achieved when the deployment environment is well managed. It does not work so well, obviously, for arbitrary web clients running on mobile devices talking to arbitrary servers. I never really intended wait to be used within such environments, however.

I could drop the Date header recommendation altogether and stress in the text that good clock synchronization and predictable latency is required for the wait preference to be used effectively. 

- James

On Fri, Oct 5, 2012 at 10:22 AM, Martin Thomson <martin.thomson@xxxxxxxxx> wrote:

On 5 October 2012 08:12, Julian Reschke <julian.reschke@xxxxxx> wrote:
> 3.4.  The "wait" Preference

> I'm not totally convinced that taking the Date request header field into
> account is necessary given the additional complexity; what do others think?

I said as much in my GenART review:
http://www.ietf.org/mail-archive/web/gen-art/current/msg07790.html

Having had experience implementing this exact feature, relying on Date
isn't going to work.