Re: [timingobject] Timestamps and reference clocks

Hi again Francois

Here's a concrete suggestion that I think address your concern for 
simplicity on the client provider side.

The timing provider simply exposes (whenever it does change)
.vector 
.skew - fresh estimate on diff between provider timeline and user 
agent timeline (HR-TIME)

(I'm assuming that HR-TIME is available to all parties involved)

That's all. This pushes the responsibility onto the timingobject for 
smoothing the sudden effects of changes to skew to ensure monotonic 
(it even opens up for a possible onskewchange event )

The timing object then implements software clock that is simply  by 
using vector[_init], skew and some linearly decreasing smoothing 
effect. As a result it can always resolve queries without consulting 
with the timing provider object.

Is this good, or did I miss something now?

-- 
GitHub Notif of comment by ingararntzen
See 
https://github.com/webtiming/timingobject/issues/16#issuecomment-134579392

Received on Tuesday, 25 August 2015 13:08:37 UTC