Gregory Maxwell wrote: > I'm interested in knowing the nature of the error (understanding is > the key to avoidance in the future!)
It was my fault, and it was pretty much identical to the error I made in 2007, where certain kinds of edits were double-counted and so the effective edit count threshold was lower than it should have been. Avoiding it in the future remains basically the same as it was in 2007, either: * Stop changing the voting rules every year so that I don't have to keep rewriting the scripts. Obviously I can change the numbers and dates, but the CentralAuth integration this year required a whole new architecture. * Assign someone to do this who doesn't have a hundred other responsibilities and can afford the time to do rigorous testing of every critical component. * Have someone review critical parts of SecurePoll instead of just trusting me to write perfect code. -- Tim Starling _______________________________________________ foundation-l mailing list [email protected] Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
