Some 14 and 16-inch MacBook Pro owners have been experiencing kernel crashes when watching HDR YouTube videos, according to a number of complaints on...
Apple and similar “hip” tech companies (like Tesla) loves reinventing the wheel for no real reason other than “ours is cooler.” It wouldn’t surprise me if they implemented it themselves just because, and didn’t"t have time to fully validate it before the launch deadline.
They didn’t exactly reinvent the wheel, they almost exclusively tacked together proven stuff. Regardless of that it’s not a bug that is impossible to occur but yeah it’s something that (probably) could have been caught in some kind of automated testing.
I feel like if they just used the prototypes for stuff that people actually use prior to launch, they would have noticed. It took all of two weeks for real users to not only notice, but realise it was a widespread problem. The prototypes would have debug tools hooked into them so they would have realised something was wrong even sooner.
Automated testing would have caught it even earlier. Especially since you think you’d put every format of video a video decoder is compatible with through the paces before you make your production chips.
since you think you’d put every format of video a video decoder is compatible with through the paces
That’s easier said than done, there is a reason people use fuzzy testing. When you have fairly complex software (like a soft/hard hybrid multi-format video decoder) the number of test cases explodes and trying to hit every single problematic one converges to trying to solve the halting problem. If you then account for hardware weirdness¹ it’s basically guaranteed that there will be bugs.
¹ Which is not an apple specific problem, just look up your CPU’s errata sheet
That implies they did at least some testing and validation.
Also (hardware accelerated) media decoding isn’t something that’s hard to accidentally do subtly wrong. On the other hand it’s “their own” hardware.
Apple and similar “hip” tech companies (like Tesla) loves reinventing the wheel for no real reason other than “ours is cooler.” It wouldn’t surprise me if they implemented it themselves just because, and didn’t"t have time to fully validate it before the launch deadline.
They didn’t exactly reinvent the wheel, they almost exclusively tacked together proven stuff. Regardless of that it’s not a bug that is impossible to occur but yeah it’s something that (probably) could have been caught in some kind of automated testing.
I feel like if they just used the prototypes for stuff that people actually use prior to launch, they would have noticed. It took all of two weeks for real users to not only notice, but realise it was a widespread problem. The prototypes would have debug tools hooked into them so they would have realised something was wrong even sooner.
Automated testing would have caught it even earlier. Especially since you think you’d put every format of video a video decoder is compatible with through the paces before you make your production chips.
That’s easier said than done, there is a reason people use fuzzy testing. When you have fairly complex software (like a soft/hard hybrid multi-format video decoder) the number of test cases explodes and trying to hit every single problematic one converges to trying to solve the halting problem. If you then account for hardware weirdness¹ it’s basically guaranteed that there will be bugs.
¹ Which is not an apple specific problem, just look up your CPU’s errata sheet