UTC vs Interval Resolution

[imstimpy]

Under most circumstances, subtracting the log's prior row value from the current value will yield a millisecond delta that is consistent whether UTC or Interval. Occasionally, they do not match. Worse yet, UTC may go backwards.

I read that UTC pulled from GPS is guaranteed accurate within 97 nanoseconds, but obviously if it goes backwards something is amiss.

Can I depend upon the accuracy of the interval in Interval? Should I, instead, depend upon the interval of GPS but understand there are occasionally time shifts? Is there a bug in the GPS reading code that occasionally processes the wrong time?

This is all asked in the context of external data analysis software, FWIW.

[brentp]

Hi,

Interval is calculated from the internal system, and should not go backwards.

UTC is periodically corrected by GPS, and we have seen that it can fluctuate. We recommend relying on Interval.

Thanks!