Since I wasn't having much luck yesterday, I chose to hang out in Doug's cube today and kibitz over his shoulder.
We found a LOT of stuff out, building on what I found yesterday and what he found yesterday morning, but of course no real resolution to the problems.
I'm compiling a test now, though, to see if a meddling monitor thread is foolishly and dangerously resetting something it shouldn't be.
This system was definitely not designed with debugging in mind. I think I see now why the designer of it all was having problems getting it to work right.
Oh, right. I doesn't work right.
We will be burning about two man-weeks of time on this, wasted time I add. It may be work, but it's not work that I feel good about.
But working with Douglas, who is cheerfully whacking away at this thing, does make it more pleasant. It's not just me and my brooding thoughts.