Eliot,

I apologize if this is retreaded ground.  I am still using an older, hacked, vm for most of my work.  The hacks provided diagnostic information in two ares:

(1) can't load an external library?  Then it helps to see the full path that was used; at least once, it ended up having an extra slash appended somewhere; various other times, I had not registered something correctly, and the mistake was obvious as soon as I saw what the vm was ultimately trying to load;

(2) Can't open a serial port?  Then it helps to see how the vm mapped it too.  I wish I remembered more of the details on this type of problem.  At least some of the trouble I have had is due to the realities of newer hardware forcing me to use USB to serial converters, which I find to be unreliable.  I am still sorting through the combination of Linux and some excellent hardware that I have used for years.

The short version is that when a library can't load or a port can't open, etc., and the vm did some translation of a name to a number (or vice versa), or manipulated a string, it would be nice to have syslog()/OutputDebugString() output to show what the vm tried to do.

Bill