C++ V4.1 Update and DriverStationLCD

[gjordan]

We've installed the C++ 4.1 update and made the necessary modifications for the camera changes. Our image builds without any errors. When booted on the cRio, we get an error:


Warning: module (t1) deleted: errno=1835009 (x1c0001) status=1
_ZN16DriverStationLCD11kLineLengthE.
(unloading partially loaded module FRC_UserProgram.out)
...FRC_UserProgram failed to load.

If we remove all references to print to the DriverStationLCD, then all is fine. Any one else upgrade and run into these issues?

Thanks,
_Greg (Team3126 FlashPoint Mentor)

[BotnEyedJoe]

Yes, we absorbed the 4.1 update, re-compiled and re-built with no errors (after changing a few names due to the update), and when we (try to) Run the code in debug mode, get the following dialog in Workbench, and the code does not run:

On downloading module '/blah/blah/Debug/robotcode.out' on target 'VxWorks6x_10.x.y.2_0-MPC5200', the module symbols could not be fully resolved.

Unresolved Symbols List:
DriverStationLCD::kLineLength

[pafwl]

Several of us are having the same problem...

<url>http://www.chiefdelphi.com/forums/showthread.php?t=81829</url>

[BradAMiller]

There seems to have been a compiler or linker bug that was causing a static const int variable to not initialize correctly that just started happening. I posed a new version here:

http://first.wpi.edu/FRC/frccupdates.html

That seems to fix the problem. That's the only change in this version so there's no need to update if you aren't using the DriverStationLCD class.

[gokul]

Originally Posted by BradAMiller

There seems to have been a compiler or linker bug that was causing a static const int variable to not initialize correctly that just started happening. I posed a new version here:

http://first.wpi.edu/FRC/frccupdates.html

That seems to fix the problem. That's the only change in this version so there's no need to update if you aren't using the DriverStationLCD class.

Thanks, Brad.
As the file sizes are quite close, I take it that the new update includes the changes from the original 4.1 also. Is this correct ? Or do we have to apply the 4.1 and then this new update ?

Thanks again

[stevethetinker]

Another question for Brad: is source available for the new update? I am having some nasty problems and need the source for a few classes.

My problems in case you might be able to help:

The gyro class won't initialize correctly. The constructor seems to run but calling GetAngle gives a memory access error. The debugger in Gyro.cpp shows that m_analog is NULL. BUT: I have Gyro.cpp from a previous version of the source.

Encoders don't work. I get no counts. A logic probe on the wires from the encoders shows there is a signal there. BTW, the encoders are powered from the DIO +5 pin. Is that OK?

Adding some bits of code cause obscure errors. I can add some loops to bring the return angle from the gyro (when I compute it myself rather than use the Gyro class) into the interval [-pi, pi) causes the joystick input not to go through to the wheels. I haven't isolated where the input goes away, but when I remove the angle loops the problem goes away.

All these suggest there may be a memory overwrite somewhere. I am in a panic this late for this problem. These things can take weeks to find.

[stevethetinker]

I found the source on Firstforge. Silly of me to ask, it had usually been posted on the frccupdates page.

With updated source the Gyro problems are the same. Still no counts for the encoder.

[gjordan]

Just updated to V4.2... We added back in the LCD code but our code failed to load. This time however, the symbol displayed was associated with the AxisCamera. If we remove our camera code, then our program will load and it can print to the LCD.

_Greg

[pafwl]

I got the same result. Love the LCD Display it works great.

Camera is now DOA, so I am going to work without the camera. Don't really need it. But that is another thread.

Thanks for getting the LCD fixed.

[stevethetinker]

I have done some more testing and with one of my fellow mentors we believe that our 9403 might be partly fried (we are using last year's computer for testing). We will find out more tomorrow when we meet next.

[Randy Forgaard]

This AxisCamera problem in WPILib 4.2 for C++ has been fixed. The new WPILib 4.3 release on 2/17/10 has resolved this AxisCamera unresolved symbol in our tests. The DriverStationLCD is still working correctly as well. Super thanks, WPI!