On Mon, Feb 20, 2017 at 6:50 PM, Martin Thomson <[email protected]> wrote:
> On 21 February 2017 at 00:31, Josh Soref <[email protected]> wrote: > > Richard Barnes wrote: > > > > What's the use case for this? Diagnosing clock issues? > > > > > > Sometimes the spec lists an expires time and says what to do if time is > > before/after. It's hard to tell if the time in a response is supposed to > be > > in the future when it's in fact in the past. > > HTTP provides a Date header field if you want to go crazy. I would > probably still NTP over that though. > The dates are all advisory anyway. The real comparisons are the ones the server does with its own clock; the "expires" attributes are just advice to the client on how quickly they should be getting things done. So I'm not tremendously worried about clock desync here. The Date header seems sufficient if clients want to try to figure out why their queries are failing. --Richard
_______________________________________________ Acme mailing list [email protected] https://www.ietf.org/mailman/listinfo/acme
