Re: [gpsd-dev] ✘cgps

[Fred Wright]

On Thu, 31 Mar 2016, Gary E. Miller wrote:
> On Thu, 31 Mar 2016 13:10:32 -0700 (PDT)
> Fred Wright <address@hidden> wrote:
>
> > > Gack, much worse than that.  Not limited to window size at all.
> > > Just a check to keep frmo running off the end of the list...
> >
> > Aside from implementing the limit correctly, there's the question of
> > whether it's time to increase the limit, due to the growth of the
> > overall GNSS constellation.  Anything that puts lists of satellites
> > in a window form is affected; this includes cgps, xgps, and gpsmon.
>
> Yeah, sounds like a good project for someone handy with curses or
> Python Gdk.
>
> > Even just using my GLONASS-capable Android phone as a source
> > *indoors* can exceed xgps's 20-satellite limit.
>
> What do you use on your Android?  I use 'GPS Status', but it is only
> graphical.

For looking on the phone itself I use 'GPS Test', which I decided I liked
better than 'GPS Status' (though I haven't compared them recently).  It
has a similar issue where the bar-graph display only accommodates 12
satellites, and it's not smart enough to prioritize the active ones, so
some get left out.  The skyview display is complete though, and uses
different icons for GPS vs. GLONASS.

For the testing I referred to, I found an app called 'GPS over BT', which
exports data over a Bluetooth serial connection.  It has a number of bugs
such that it couldn't really be used "for real" (and as usual, email
to the developer via Google Play seems to redirect to /dev/null), but it's
useful for certain types of testing (e.g., gpscat, which needs a direct
serial connection).  From the behavior, I don't think it's passing through
data from the GPS engine at all, but rather synthesizing NMEA sentences
from data in Android OS form.

My understanding is that recent versions of Android use GPSD internally,
but when sshing into the device I don't see anything relevant in ps or
netstat.  I'd hoped I could just pass the internal GPSD output through an
ssh tunnel.  It's possible that it's really there but hidden for "security
reasons".

> Do you see 20 active?  cgps, gpsmon and xgps can get by showing just
> active if there are too many.

No, I didn't see more than 20 active, but this is indoors.  And it's
useful to see SNR values even from "inactive" satellites (note that it has
to have acquired a satellite in order to report SNR).

With my SxBlueIII-GNSS, I've noticed a certain reluctance to incorporate
GLONASS satellites into the fix, which is I suspect due to the less
tightly controlled clock offsets between systems.  Claims of "it's like
having twice as many satellites" are oversimplified.  At the least, I'd
expect to have to add an extra state variable to the Kalman filter to
represent the inter-system clock offset.

BTW, the SxBlue receivers also have an option to allow including SBAS
satellites in the fix, which typically adds another two or three in the
US.  And they show up in the satellite list whether that option is used or
not.

Currently, GPSD chokes on data from the SxBlueIII-GNSS, which I suspect is
due to not liking the mixture of NMEA and binary data.  I can change the
configuration to get rid of that, but for test purposes it's easier just
to use a receiver that doesn't confuse it.

> Right now I have 13 GP/BD sats in view, 14 can show in cgps.  I hae a
> GLONASS receiver coming next week, that should do more.

By the time Beidou and Galileo are fully deployed (in addition to the
currently operational GPS and GLONASS), there will be around 150
navigation satellites globally (not counting the regional IRNSS or QZSS),
though only about half of the non-geostationary ones are visible at any
given time.

Fred Wright