Re: [gpsd-users] STATUS_DGPS_FIX mystery

[Eric S. Raymond]

Roger Oberholtzer <address@hidden>:
> > From: Eric S. Raymond address@hidden
> >> Gary E. Miller <address@hidden>:
> 
> >> Eric, any reason for this commit?  Looks like a step backwards.  I
> >> would like STATUS_DGPS_FIX restored.
> 
> > Why? What you really want to know in practuce is the XYZ %95 error , no?
> 
> We use the gpsd shared memory segment to access current gps information. 
> Previously we were parsing the NMEA strings ourselves. In our measurement 
> systems, the operators are instructed to monitor the status. We use receivers 
> that should be in DGPS mode during data collection. If they are not, the 
> operator needs to figure out what is wrong. And decide if not being in DGPS 
> is constant or intermittent. When measuring way up in northern Sweden in tree 
> canyons, this is relevant information. While this may not be the most robust 
> test one can do, our years of experience have led us to think that it is a 
> useful metric of quality.
> 
> It seems odd to me to hide information that the GPS is providing.
> 
> If I connect a GPS receiver that does not indicate DGPS mode, then I should 
> not be surprised if gpsd does not report that mode. OTOH, if I hook up a 
> receiver that goes to all the trouble to report DGPS mode and gpsd does not 
> report it, I think it would be correct to be confused. How is it possible, 
> with such a receiver, to determine that the GPS is using DGPS?

You make a reasonable case, but fully exposing the information you are
requesting would take an API break.  That can't happen soon; when I do
those too often, distro people come and scream at me.

Based on the UERE, I suggest you set an alarm threshold for spherical
error at, hm, I'd say 21 meters or so.  If epe is more than that you
have pretty certainly lost DGPS.

#define H_UERE_NO_DGPS          15.0    /* meters, 95% confidence */
#define H_UERE_WITH_DGPS        3.75    /* meters, 95% confidence */
#define V_UERE_NO_DGPS          23.0    /* meters, 95% confidence */
#define V_UERE_WITH_DGPS        5.75    /* meters, 95% confidence */
#define P_UERE_NO_DGPS          19.0    /* meters, 95% confidence */
#define P_UERE_WITH_DGPS        4.75    /* meters, 95% confidence */
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>