>You will probably not be able to duplicate the problem on your development machine. Problems have a way of hiding on the development machine.

With all due respect, I don't think this is a very accurate approach to debugging. You've really got to try as hard as you can to duplicate the problem in your own office before you can try to fix it. Otherwise, you're just guessing, and a solution that works for one machine may not work for another. Only by duplicating the problem can you really determine why it existed.

I do grant that this is not always possible, but I try to avoid nebulous explanations like "problems have a way of hiding." And making the problem go away isn't the same as solving it.

That's just a philosophical thing, and I think your substantive suggestions are good.