| Previous slide | Next slide | Back to first slide | View graphic version |
Notes:
Original I said due by christmas, but once this system was working the other players discovered serious flaws in their code that had never been tested, thus an 8 week test cycle so we pushed off into Feb.
In comparison, I'd run out of fingers in attempting to count the IMS and SQL production system unplanned outages, but to be fair they have 3 or 4 more orders of magnitude of code to worry about.
The support staff has mostly forgotten that the application exists...which was of course the goal! This could on the other hand be a bad thing, since any nasty failures would require much remembering/relearning on part of support staff. To address this isssue, yet more documentation was written. I suspect the client will just call a Smalltalk expert if they have problems. Maybe someone here will be the lucky fellow... I left my phone number in the class comments.
1999 There were no calls. Perhaps I should ask to look at the logs to ensure nothing bad is going on?
We did discover a problem in the VM where a millisecond counter overflowed after 24 days. This caused a VM crash. To solve it as we waited (years!) we restarted the system once a week.