News for hackers. Stuff that matters.
Re:New York Mayor’s Race in Chaos After Elections Board Coun... by mr-wendel on Wednesday June 30, 2021 @01:24PM (#27691147)
Lots of reasons probably:

  - Scalability: how long does it take to process ballets to create the electronic records?

  - Error handling: there may be lots of permutations being tested (dark marks, light marks, `x` vs `/`, marks going out of bounds, etc)

  - Realism: being able to roughly represent what will actually happen in "live" mode and think through scenarios that limited test data won't highlight
Re:New York Mayor’s Race in Chaos After Elections Board Coun... by LorenPechtel on Wednesday June 30, 2021 @03:58PM (#27692806)
Yup. One part of any such test has to involve a large load--I wouldn't want to certify something like that without testing it against a load at least as large as anything it's likely to encounter in a production environment.

Long ago I spent *weeks* narrowing down a load-caused fault, then figuring out how to reproduce it with nothing but the operating system. (Cisco, it was completely irrelevant what software was throwing the load at your switch! When I reproduced it from the command line they accepted it was their problem and at that point it was between management and Cisco, I wasn't involved.)