Alex Perry writes:
> The current CVS hangs for me when ground started at KMYF, yet is fine at KSFO.
> Immediate crash. It's a long way to commute, could we fix that sometime?
Hmmm, I can verify this same problem here too. With fgfs
--airport-id=KMYF :
- Flightgear starts up.
- The JSBSim constructer is called, and the aircraft config file is
loaded. (But JSBSim is not init'ed until after the scenery loads
so we can pass in a proper ground elevation to the trimming
routine.)
- FlightGear continues to load/init various systems.
- FlightGear finally gets around to loading the local scenery tile and
a valid ground elevation appears on the bus.
- FlightGear determines that it is now "ok" to run the JSBSim init()
routine.
This triggers a ground trim and JSBSim prints touch down reports and
a bunch of other output and finally returns with:
Trim successful
JSBSim State
Trim complete
- This is immediately followed by a long sequence of:
0: GEAR_CONTACT 1
1: GEAR_CONTACT 1
2: GEAR_CONTACT 1
3: GEAR_CONTACT 1
4: GEAR_CONTACT 0
5: GEAR_CONTACT 0
6: GEAR_CONTACT 0
7: GEAR_CONTACT 0
8: GEAR_CONTACT 1
etc.
.
.
.
56: GEAR_CONTACT 1
- It appears that JSBSim has trimmed the aircraft to below ground so
FlightGear notices this and attempts to force JSBSim to use a higher
elevation. This triggers another ground trim, again which ends up
several meters underground, again causing FGFS to try to force
JSBSim back up above the ground.
- There seems to be a long exchange of angry words with subsequent
JSBSim trims failing.
- This is followed by JSBSim spewing out 985 successive GEAR_CONTACT
<bool> messages.
- Again FGFS tries to force/coax JSBSim up to ground level.
- More of the same with JSBSim triming under ground, the gear code
goes berzerk, spews 100's more GEAR_CONTACT messages.
- Eventually JSBSim flags a "CRASH".
- JSBSim is hosed at this point and copies a completely bogus lat/lon
onto the bus.
- The flightgear tile pager sees that we have moved to a completely
impossible location, flags a non-recoverable error and exits the
sim.
Wow, that was really ugly.
It happens very quickly, and most of those messages scroll right on by
before you can see them if you don't run the output through less or
more or something similar.
So it appears that at least for KMYF, JSBSim and FlightGear are having
a very strong disagreement over the elevation of the ground, and
neither is willing to budge. Sounds like two typical flightgear
developers. :-)
I haven't been able to reproduce this at any other airport I've tried,
only KMYF.
Ok, after all that, I tried the same thing with the YASim c172 at KMYF
and something is wierd at KMYF. YASim started just fine and took off
fine, but the entire time, FGFS was spewing message about being below
ground. I wonder if it is a problem with the scenery models right
there?
Hmmm, this appears at the moment to be some sort of flightgear
problem???
Ok, looking further, it appears that nothing, no place, no where is
setting the runway elevation any more ... it get's initialized at
startup, but then never is updated as the aircraft moves.
That is really odd ... what got sliced out of the code? It's been a
while so I don't remember where this was supposed to happen.
The view code has a bit of a hack to force the view height above the
ground so it's hard to notice. This is really weird. In theory,
you'd see very strange effects if you started at a lower elevation
airport and flew to a different elevation aiport and tried to land.
Ok, there is something really strange here, probably because things
were changed without a proper understanding of how everything worked
together. My mind is fryed at the moment looking at this stuff.
JSBSim seems to be doing the right thing *except* for at KMYF.
YASim assumes that the runway altitude will get set for it externally,
that makes a lot of sense, but apparently isn't the case.
This sucks.
Curt.
--
Curtis Olson IVLab / HumanFIRST Program FlightGear Project
Twin Cities [EMAIL PROTECTED] [EMAIL PROTECTED]
Minnesota http://www.menet.umn.edu/~curt http://www.flightgear.org
_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
