When I used to work for a large mobile network, we would spend ages testing the firmware of new devices. Even giants (at the time) like Nokia would release phones which couldn't dial the emergency services, or put out more than the legal limit of radiation, or were in other ways defective. We'd test, send a report back, wait a few weeks, get a new firmware, test again, repeat until everything worked. It took months.…
How would you have tested 911 dialling? Did you go all the way to a real operator?
When bringing up a new tower, we had a bazillion tests to run, and one of the very last was 911. When that day arrived, my supervisor would call the non-emergency line at the PSAP and check in; sometimes I was in the room for this call. Basically we'd make sure they weren't short-staffed, that they had capacity for the calls, etc. Just a formality.
When testing actually began, I'd simply tell the operator "This is not an emergency, I'm a tech with Verizon making a test call, do you have time to take this call?", and if they were bouncing off their limit, they'd say no, and I was done for the night.
If they said yes, I'd ask them to read back my E911 data. I'd transcribe what they said for later comparison against what was expected, and ask again if they figured they had time to take a few more in the coming minutes. If so, I'd move to the next sector on the tower, lather, rinse, repeat.
It was really quite straightforward, and sometimes I'd get the same operator over and over, they'd recognize my DN and just start rattling off coordinates. At the last call of the night we always made sure to thank them, but briefly, and that was that. My report would go back to the datafill engineer, and that was the last step before unlocking the tower for customer traffic.