3 ms·

I'm kind of amused by the number of comments on this thread that amount to "all the devices that meet the standard are still required to be tested in situ? lol

I'm kind of amused by the number of comments on this thread that amount to "all the devices that meet the standard are still required to be tested in situ? lol wat!?". Yet, this is the same place that advocates heavy test driven development practices, which include testing your code against libraries that meet standards, testing that the libraries actually do the right thing when you use them, and personally verifying that sites work on all browsers, even if automated compliance checkers approve them. Isn't the FAA essentially doing the same thing here?

mgkimsal 15y ago
I wouldn't compare this to TDD - the devices would have been built to 'test' specs the FAA puts out, testing via an FAA process would be a regular, repeatable things manufacturers could perform regularly, for a fee perhaps. And... clients/bosses complain when testing takes a long time, which is what we're doing here to the FAA. Given how many flights happen, with a variety of devices, you might think that with some extra devices and controlled test procedures, the FAA would be able to gather 'real world' test data (their devices, their tests, real flights, etc.) and make conclusions over the past 3-5 years. MAYBE... then issue guidelines (or even paid-for device approval processes). Would Apple pay a few million to get an 'FAA-approved' sticker for iPads and iPhones? If there were specified guidelines and a testing procedure, more companies might work in that direction. I'd imagine there are guidelines, but not being in that industry, I don't know how easy to access they are. If there were public knowledge, I'd imagine device manufacturers already promoting "meets FAA guidelines for xyz", but since we don't see that, I have to assume they're not easily accessible.
mkopinsky 15y ago
Testing code is important because we know, from previous experience, that code written by humans tends to have bugs. In this case, we also know from experience that small devices like iPads do not cause high levels of EMI. Once we have tested a dozen models on a dozen planes and had successful results on all 144 tests, there is no need to test further models. This is unlike software, where of 144 unit tests, a good percentage are almost guaranteed to fail at some point in time.

mgkimsal 15y ago

I wouldn't compare this to TDD - the devices would have been built to 'test' specs the FAA puts out, testing via an FAA process would be a regular, repeatable things manufacturers could perform regularly, for a fee perhaps. And... clients/bosses complain when testing takes a long time, which is what we're doing here to the FAA. Given how many flights happen, with a variety of devices, you might think that with some extra devices and controlled test procedures, the FAA would be able to gather 'real world' test data (their devices, their tests, real flights, etc.) and make conclusions over the past 3-5 years. MAYBE... then issue guidelines (or even paid-for device approval processes). Would Apple pay a few million to get an 'FAA-approved' sticker for iPads and iPhones? If there were specified guidelines and a testing procedure, more companies might work in that direction. I'd imagine there are guidelines, but not being in that industry, I don't know how easy to access they are. If there were public knowledge, I'd imagine device manufacturers already promoting "meets FAA guidelines for xyz", but since we don't see that, I have to assume they're not easily accessible.

mkopinsky 15y ago

Testing code is important because we know, from previous experience, that code written by humans tends to have bugs. In this case, we also know from experience that small devices like iPads do not cause high levels of EMI. Once we have tested a dozen models on a dozen planes and had successful results on all 144 tests, there is no need to test further models. This is unlike software, where of 144 unit tests, a good percentage are almost guaranteed to fail at some point in time.