I've fixed a couple problems in timing where an int was used to count
microseconds ... I sat down and figured it out before, but this gives
you somewhere in the neighborhood of 30 minutes depending on if you
are using a signed or unsigend int.  When this rolls over, you get a
huge delta t passed to routines.  This in the past has cuased things
like the instrument panel not updating it's text for 30 minutes, or
navaid lookups failing after 30 minutes.

What you describe sounds suspiciously like the same sort of thing.

We should probably take a look at what units the dt is passed in, or
if a time difference is calculated someplace, what limiting
intermediate structures might the calculations be passed through.
This is an insidious bug because you can test things for a few minutes
and they work fine, but try a long cross country flight and things
start to break or blow up.

3700 is about the number of usec an unsigned int can hold so that
sounds very suspicious.

Also look for some sort of uninitialized time stamp variable used to
calculate the dt passed to the subsystems.  Unless the mainloop is
hung up for an extraordinarily long time, the delta t passed to the
subsystems should never be that large.

Regards,

Curt.


Jim Wilson writes:
> Off and on I've been seeing some weird hangups using the 747 model at startup.  
> Currently I'm able to get it everytime at KBGR (but not at other airports).
> 
> The problem I am seeing is happening in FGInterface.  The delta time ends up
> being some ridiculous number, generally around 3700.000 in the
> "_calc_multiloop" which doesn't actually freeze the program, but it creates a
> very large number of iterations to be processed.
> 
> More interesting is that no matter what aircraft run (JSBSim's included) the
> delta time is always a few milliseconds different in the _calc_multiloop
> routine than it is in the calling routines.  I can't see anything that makes
> sense to explain this.
> 
> As for the 747 hanging at startup bug, I'm a bit baffled.  In this case the
> delta_time looks normal until it gets passed to _calc_multiloop.  The value is
> wildly deviant,  like the stack is getting corrupted somehow.
> 
> Is any one else able to observe this problem?
> 
> Best,
> 
> Jim
> 
> _______________________________________________
> Flightgear-devel mailing list
> [EMAIL PROTECTED]
> http://mail.flightgear.org/mailman/listinfo/flightgear-devel

-- 
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

Reply via email to