Re: [Xcsoar-devel] UDP Port and Silent Wings support.

[Max K. <ma...@du...>]

Damiano, please keep the mailing list on Cc!

On 2012/06/20 13:58, Damiano Bortolato <da...@da...> wrote:
> On Wed, 2012-06-20 at 11:49 +0200, Max Kellermann wrote:
> > Damiano, please post to the mailing list, so others can participate,
> > too.
> > 
> > On 2012/06/20 11:39, Damiano Bortolato <da...@da...> wrote:
> > > I've a little problem: when I produce fake NMEA sentences it works
> > > properly (switch between circling and cruise mode), when I populate
> > > directly the NMEAInfo it detects the GPS, (altitude location and speed
> > > are ok and the glider is moving on the map) but it is not able to detect
> > > the fly mode (circling/cruise).
> > > 
> > > Is some crucial data missing ?
> > > 
> > > 
> > > bool
> > > SilentWingsDevice::DataReceived(const void *data, size_t length,
> > >                             struct NMEAInfo &info)
> > > {
> > >   if(length % sizeof(SilentWingsBinary)){
> > >     return false; // malformed data?
> > >   }
> > 
> > Can there be more than one packet in a datagram?  If no, why check for
> > modulus?  If yes, why not parse all of them?
> 
> May be, I don't know, in that case I take the first and discard the
> others (or I may take the last a discard the previous).
> 
> > 
> > >   SilentWingsBinary &binary = *(SilentWingsBinary *)data;
> > 
> > This cast is invalid, because it removes the "const".
> > 
> Ok
> 
> > >   info.location.latitude = Angle::Degrees(binary.position_latitude);
> > >   info.location.longitude = Angle::Degrees(binary.position_longitude);
> > >   info.gps_altitude = binary.altitude_msl;
> > >   info.track = Angle::Degrees(binary.heading);
> > 
> > All of the above don't work with FIXED=y.  You must cast to "fixed"
> > explicitly.  (More such errors below)
> > 
> Ok
> 
> > >   info.gps.fix_quality = FixQuality::GPS;
> > 
> > This should better be FixQuality::SIMULATION.
> > 
> 
> I tried also that but nothing changes.

This does not affect your problem, but still the code line above is
wrong.

> > >   info.gps.satellites_used = -1;
> > 
> > Why bother filling this attribute when you don't have any data?
> > 
> > >   info.gps.real = true;
> > >   info.gps.simulator = false;
> > 
> > This is both obviously wrong.
> > 
> 
> Yes I know, they were just trials.
> 
> > >   info.ground_speed = sqrt(binary.vx*binary.vx + binary.vy*binary.vy);
> > 
> > Use hypot().
> > 
> 
> Ok
> 
> > >   info.ground_speed_available.Update(info.clock);
> > >   info.location_available.Update(info.clock);
> > >   info.gps_altitude_available.Update(info.clock);
> > >   info.gps.fix_quality_available.Update(info.clock);
> > >   info.gps.satellites_used_available.Update(info.clock);
> > >   info.track_available.Update(info.clock);
> > >   info.alive.Update(info.clock);
> > 
> > The Validity objects should be updated right after the according
> > attribute was set.  It's now difficult to see if you missed one.
> > 
> 
> Ok
> 
> > >   info.gps.Expire(binary.timestamp);
> > 
> > What's this?
> > 
> 
> Another trial. It was not very clear to me what Expire() is supposed to
> do (check expiration and clear?)

Did you read its API documentation?  Anyway, it's not supposed to be
called by device drivers.  Remove that.

> > What is binary.timestamp?  And why don't you use the time stamp to
> > update NMEAInfo::time?  My guess is that there is no time, and XCSoar
> > needs a clock for many of its algorithms.
> 
> This is also my opinion (that is why i tried with Expire()), but
> info.clock is running, I can see it from the debug print...
> 
> ...
> 183097.12   45°24'41"N 11°50'52"E (45.411492, 11.847639)
> 183097.32   45°24'41"N 11°50'52"E (45.411515, 11.847683)
> 183097.52   45°24'42"N 11°50'52"E (45.411537, 11.847728)
> 183097.72   45°24'42"N 11°50'52"E (45.411557, 11.847774)
> 183097.92   45°24'42"N 11°50'52"E (45.411576, 11.847820)
> 183098.12   45°24'42"N 11°50'52"E (45.411593, 11.847867)
> 183098.32   45°24'42"N 11°50'52"E (45.411609, 11.847915)
> 183098.52   45°24'42"N 11°50'53"E (45.411622, 11.847963)
> ...
> 
> binary timestamp is a timestamp provided by the simulator, it is the
> same field used to generate the UTC time on the $GPGGA NMEA sentence.
> I'll check everything again.

Consider that info.clock and info.time are two different things.

> 
> I forgot to say that it's only a draft version on the code. ;)
> 
> 
> Damiano