Re: DOMHiResTimeStamp

On 02/07/2013 08:15 PM, Harald Alvestrand wrote:
> On 02/07/2013 05:50 PM, Dominique Hazael-Massieux wrote:
>> As discussed during the meeting on the way to define stats reports,
>> there is a spec that defines a high-resolution timestamp:
>> http://www.w3.org/TR/hr-time/  (and it's a recommendation since December!
>> )
>>
>> Dom
> Hm - I'm a bit confused about how to convert NTP timestamps into this 
> format:
>
> The|DOMHighResTimeStamp| 
> <http://www.w3.org/TR/hr-time/#sec-DOMHighResTimeStamp>type is used to 
> store a time value measured relative to the|navigationStart| 
> <http://www.w3.org/TR/navigation-timing/#dom-performancetiming-navigationstart>attribute 
> of the|PerformanceTiming| 
> <http://www.w3.org/TR/navigation-timing/#performancetiming>interface[NavigationTiming] 
> <http://www.w3.org/TR/hr-time/#NavigationTiming>, the start of 
> navigation of the document, or a time value that represents a duration 
> between two|DOMHighResTimeStamps| 
> <http://www.w3.org/TR/hr-time/#sec-DOMHighResTimeStamp>.
>
> I think we'll have to store the base somewhere and make explicit 
> reference to it; Performancetiming.navigationStart doesn't seem 
> terribly relevant.
>
> Having the base be different for different objects solves the "NTP 
> clock skew" problem (but does raise a few other issues).
>
(just for added confusion, hr-time defines navigationStart as an 
unsigned long long, and does not define what its unit or epoch time is.)