1. 15
  1.  

  2. 5

    I was hoping to see a little more discussion of runaway failures. This old lobster thread is good, I think: https://lobste.rs/s/rgjfav/being_defensive_when_developing_automated_systems

    1. -4

      assert() has already been invented

      1. 6

        That’s absolutely not what the article is talking about; it’s actually closer to the AWS ELB health check (a failure causes sets the state of something to be an error state, continuous successes reset it back to an okay state, various actions defined for each state, etc., etc.)

        1. 1

          At the core the article is about failing early, which is what assert() is used for in other places. Being connected to the internet their error condition has to be a bit fuzzier, requiring multiple failures close to each other. Correct me if I’m wrong.