[Tinyos-devel] improving FTSP interface

[Omprakash Gnawali]

On Fri, Aug 8, 2008 at 10:55 AM, Miklos Maroti <mmaroti at math.u-szeged.hu> wrote:
>> In that case, excluding wraparound, there is no possibility that 5
>> units difference between previous time and current time is different
>> from 5 units?
>
> Yes, it is possible because of skew (global time might run faster than
> the local time), or the time offset could change slightly over time
> (if the globla time and local time is not exactly linearly dependent
> because of crystal instability). However, the effect of these should
> be minimal (+/- 5%) compared to the local time difference. So if your
> application periodically wakes up using a Timer every 2^30 local time
> units, then it should not miss any wrap around.

When I plot the time values for the whole network, I have definitely
seen some numbers that are not even close to the globaltime. So,
perhaps is_sync logic is not correct? My understanding of is_sync is,
if it returns true then the time is valid. So, when you read time
through global time interface, you should either have a 0 or a valid
time. Brano said he would look at some of the issues later but some of
the time values there don't make any sense either. So, at least in the
current implementation, I think it is difficult to make that claim.

>
> By the way, you can query the root (leader) node id via the
> TimeSyncInfo interface. I think the overflow event could go into this
> interface as well.
>

Agree.

- om_p