It turns out that though enabling debug allowed me to build, it wrecked my deployment. I'd build, then deploy. LabVIEW says all is OK. But after deploying the code would not run on the cRIO. The only way to run was from LabVIEW.
So I went back to the most recent version that built without enabling debug. (Lesson: I was saved a lot of grief because I make daily or more often backups of our project.) That version also deployed so the two problems are linked.
Since LabVIEW gave no clues, I took the current project and went back to the Teleop.vi that worked. Using that fixed my problem. Then I retraced my steps bit by bit until I found the step that broke it. After a couple of hours I found it was caused by a new sub vi. That vi referenced the drive motors (specifically arcade drive) which is a logical conflict with the reference in Teleop.vi. Cleared that up and everything builds OK and deploys OK. WHEW!
So my problem is fixed and my code is better. It's strange that the problem was caused by a logic error. Furthermore, adding the same logic error to a pristine FRC robot project does not produce any error. Combine that with the comments about cases/loops controlled by constants and constants setting structures and I suspect there are some complex, seemingly random things going on when LabVIEW is building and saving the cRIO image.
Glad I got rid of the problem. Thanks for all the helpful comments.
