Following an initial (or subsequent) synchronization, what should the defined behavior be for periodic behavior? I think there are a couple cases that matter.
Unscheduled Periodic Events (sampling at a fixed frequency, local control loops, etc.)
In this case, the interval matters more than the absolute timestamp. In this case, I think synchronization should not affect the intervals of these events.
Scheduled Events that must happen at an exact timestamp
In this case, the absolute timestamp matters more than the interval. In this case, the event should happen at the synchronized global time.
Currently, this isn't defined, so it's left up to the device designer to decide whether a particular behavior's interval or absolute time matters.
It's worth noting that we don't have any devices that implement a "do-this-at-future-time-x" behavior right now. But this issue will surface if we take the "interval" approach everywhere and then use a device's Harp Reply timestamp to initialize subsequent periodic behavior on another device that will stay in sync with the previous device.
Following an initial (or subsequent) synchronization, what should the defined behavior be for periodic behavior? I think there are a couple cases that matter.
Currently, this isn't defined, so it's left up to the device designer to decide whether a particular behavior's interval or absolute time matters.
It's worth noting that we don't have any devices that implement a "do-this-at-future-time-x" behavior right now. But this issue will surface if we take the "interval" approach everywhere and then use a device's Harp Reply timestamp to initialize subsequent periodic behavior on another device that will stay in sync with the previous device.