Comment by mschuster91
Comment by mschuster91 3 days ago
It's a culture issue when such cases don't get caught in testing and support staff either doesn't know about it (=the scripts are bad), doesn't get told that there is a workaround (because clearly there is), or (the worst of the possible options) gets told to act like everything is fine.
It's a race condition which involves physical interaction with the product, which only occurs during a very rare operation, which looks pretty much identical to a genuine hardware defect.
This isn't something you can just trivially unit test. If you don't see this happen multiple times during initial hardware development, you are never catching it. A single failure of a prototype can easily be attributed to a manufacturing defect - especially if it was hand-soldered.
Once it's in the field replacing the 0.01% of units suffering from random issues under warranty is far cheaper than having an engineer spend weeks trying to diagnose every single weird failure mode. Unless it affects a number of units, it's just not worth the money. You have to consider that support doesn't get a "the LEDs stopped working when I unplugged the device immediately after flashing it" message, they just get "the LEDs don't work". Support scripts are made for horses, not zebras.