Re: [Jsbsim-devel] Using FlightGear to visualize JSBSim state

[Stephan K. <sk...@ca...>]

Hey D,

By 'unhelpful win32 message' I meant the informative JIT debugger msg 'An 
unhandled win32 exception occured in JSBSim.exe [3744].' if you don't have 
VS then you simply get the option to send Microsoft an error report. (Maybe 
they'll fix it?)

JSBSim socket related messages are:
Winsock DLL loaded ...
Creating TCP socket on port 5500
Successfully connected to socket for output ...
  Output parameters read inline

Here's the last few lines of output from the VC++ output window:

... 'JSBSim.exe': Loaded 'C:\WINDOWS\SysWOW64\setupapi.dll', No symbols 
loaded. First-chance exception at 0x7d4e2366 in JSBSim.exe: 0x000006C5: The 
tag is invalid. 'JSBSim.exe': Loaded 'C:\WINDOWS\SysWOW64\rasadhlp.dll', No 
symbols loaded. 'JSBSim.exe': Loaded 'C:\WINDOWS\SysWOW64\hnetcfg.dll', No 
symbols loaded. 'JSBSim.exe': Loaded 'C:\WINDOWS\SysWOW64\wshtcpip.dll', No 
symbols loaded. The thread 'Win32 Thread' (0x6e8) has exited with code 0 
(0x0). The program '[1888] JSBSim.exe: Native' has exited with code -1 
(0xffffffff).

Note that I am compiling this on a 64bit OS, I have not had the chance to 
compile it on a 32bit one yet. I also have not yet found the exact part of 
the code where the exception is thrown in the debugger.

As for the net_fdm header change: Some time ago JSBSim refused to compile 
with VC++ because uint32 and int32 were not defined in net_fdm.hxx. To fix 
this the following typedefs were added:

#ifdef _MSC_VER 
    typedef unsigned long uint32_t; 
    typedef long int32_t; 
#endif 

(Shouldn't those be long ints? I tried changing them but it doesn't make a 
difference)

I am going to try and compile JSBSim with some other compiler and see if 
this error persists.

Stephan

On May 7 2008, D Wysong wrote:

>>  I was wondering if anyone has tested the below and gotten it to work?
>>
>>  I am using JSBSim compiled from the latest version on the CVS and the
>>  latest release of FlightGear (according to the FG CVS net_fdm.hxx has 
>> not
>>  changed so this should be okay).
>>
>>  FlightGear starts up without any problems given the parameters specified
>>  below. JSBSim manages to succesfully connect to the output socket. 
>> However
>>  as soon as the start up phase is complete JSBSim crashes with an 
>> unhelpful
>>  win32 error message. FlightGear is still running, and seems unaffected.
>>
>>  I will try to download and compile an older version of FlightGear to 
>> see if
>>  it works, however I don't think that's the problem. I remember some 
>> changes
>>  were made to net_fdm.hxx to make it compile with VC++, could that have
>>  affected it?
>
>Hello Stephan,
>
>Will you send the full text of the unhelpful win32 message you are
>seeing?  Also, what messages are you seeing from JSBSim during init
>(socket related)?
>
>I wrote/tested the original code in a Linux environment but the socket
>code is rather generic and should port without worries (... famous
>last words).  I have not tried it using the latest/greatest version of
>JSBSim and FG.  Probably oughta do that prior to Jon's v1.0 rollout.
>
>I am not aware of the changes to net_fdm.hxx for VC++ you mentioned.
>Were those discussed on the devel list?  If they are header related
>the changes shouldn't matter.  If the struct was toyed with or we're
>mixing/matching data types then we'll likely see odd things occurring
>on the FG end of things.  However, JSBSim should still pipe the data
>out even if the struct is no longer "valid" from an FG point-of-view.
>
>Let me know what you find.  I will attempt to bring up a system here
>with the latest/greatest JSBSim/FG versions.
>
>D
>