Re: gpsd | libgps_dbus: Fix for platforms without clock_gettime(). (404217a9)

[Fred Wright]

On Sat, 14 Aug 2021, Gary E. Miller (@garyedmundsmiller) wrote:

> Does osX not have CLOCK_MONOTONIC?

It does whent it has clock_gettime() at all, which didn't happen until 
10.12.  But the current fallback implementation in GPSD only has 
CLOCK_REALTIME, since that's all that's been used in the past.

On Sat, 14 Aug 2021, Gary E. Miller (@garyedmundsmiller) wrote:

> Oh, this on the man page: "This clock does not count time that the system is 
> suspended. "
>
> So not good for timeouts...

Well, yes and no.  When an activity involves the system itself, then one 
usually wants to suspend the timeout while the system is suspended, since 
otherwise one gets false timeouts due to the suspension.  But that's 
primarily an issue for timeouts within the system itself.

The biggest problem with CLOCK_MONOTONIC is that it's this bastardized 
timescale that's no good for timestamping and inaccurate for intervals.

Fred Wright