I have re-run that test, and the issue indeed exists.
There is a race condition inside the GUI program where if you quit the program too quickly, there is a chance that the update function will be unable to shut down cleanly due to its need to show a dialogue window to the user. This will cause the program to crash.
It was just that I was lucky, and the runner was busy at the time, so the script was delayed and the program did not crash.
I have re-run that test, and the issue indeed exists.
There is a race condition inside the GUI program where if you quit the program too quickly, there is a chance that the update function will be unable to shut down cleanly due to its need to show a dialogue window to the user. This will cause the program to crash.
It was just that I was lucky, and the runner was busy at the time, so the script was delayed and the program did not crash.