I ran the conformance test and it passed so that is not an issue. I've already tried heavily polling the driver as I said but I wonder about the retry code you mention. Is it possible that the retry is causing a problem? If you could tell me the details about how the retry works, I will see if I can cause a failure with a test program. What is it SGP detects from the driver that triggers a retry? Perhaps the trigger is not seeing -1 or seeing the old position? Is there a time delay before that check is made?
In the 1000 cycle test I ran, I was checking every 100ms for either the final position or -1. I have not seen any indication that the driver is returning a bad status at least after 100ms.
Let's not just focus only on the driver/SGP interaction. Why is SGP not timing out? Is that a clue as to what is happening?
Using ASCOM logging is a great idea but since the issue seems to be timing dependent, I can't set up a reliable failure to check it.