Re: [External] Re: Time precision in gps_fix_t

[Greg Troxel]

"Gary E. Miller" <address@hidden> writes:

> Yo Greg!
>
> On Wed, 22 Jan 2020 12:54:12 -0500
> Greg Troxel <address@hidden> wrote:
>
>> "Gary E. Miller" <address@hidden> writes:
>> 
>> > Yo Greg!
>> >
>> > On Wed, 22 Jan 2020 12:16:02 -0500
>> > Greg Troxel <address@hidden> wrote:
>> >  
>> >> All true, but Marco asked a perfectly fair question about gpsd
>> >> losing the subsecond time-of-fix.  
>> >
>> > I see no evidence of that.  If that is true it is a serious bug.  
>> 
>> I don't have the logs handy, but I was doing captures of the json data
>> (gpspipe -w) from an M8030 set to 2 Hz.   Adjacent lines had the same
>> timesstamps, both with 0 subsecond fix times.
>
> Multiple TPV with the same time stamp is not unusual.  Care to share your
> raw dumps?

Sorry, I don't have raw dumps available to share.    I will if I can
generate some.

This was a clear pattern of two TPV lines with the same timestamp, for
every second.

> Regardless, getting time to 100 micro seconds from the serial data
> is unlikely by at least 2 orders of magnitude.

I am not talking about time sync.   If one has > 1 Hz positions, you
want to be able to have those time tagged correctly so you can
interpolate based on your (synchronized) system clock to get
within-second position estimates.   This is a very noticable issue with
lots of nav programs on android (probably more about not extrapolating
the pointer, vs multiple positions per s).