4 ms·
It's not surprising that OS upgrades can fail on some machines. No amount of QA can ever guarantee 100% success. What I do find surprising though is that there
by fauigerzigerk 6y ago
It's not surprising that OS upgrades can fail on some machines. No amount of QA can ever guarantee 100% success.
What I do find surprising though is that there is no bulletproof rollback mechanism.
- DennisP 6y agoAll true but given that Apple makes all the hardware too, I'm really curious whether they keep one of every version of the hardware to test OS installs.
- SwiftyBug 6y agoThey could even automate those tests
- fbelzile 6y agoApple's unique value proposition is that they control the hardware and software so that things "just work." It shouldn't be too far-fetched to imagine a room full of older hardware where it's the job of some poor intern to update each device during the alpha/beta phase. Bricking supported hardware on a software update goes directly against Apple's core business strategy and should have been top of mind for the QA team.
- danlugo92 6y agoThe low amount of possible configurations in their hardware should make this doable.
- totalZero 6y ago> No amount of QA can ever guarantee 100% success. You're too quick to let them off the hook. Compared to the ecosystem of Windows-supported devices, Apple only has to serve updates to relatively few SKUs. If they can afford to buy back a hundred million dollars' worth of stock every year, they can afford to test upgrades on affected older devices before rolling them out. When an upgrade is supported on an Apple machine, then there should be 100% success -- or at least, as you suggest, a painless way to catch the upgrade failure and roll the device back. They sell the hardware and develop the software, so they have to own the problem. No mumbo-jumbo. No excuses. No passing of the buck to the customer who trusted them at the moment of downloading/installing the upgrade. If they can't guarantee success for an OS upgrade on a particular machine, then they shouldn't offer it for that SKU -- and they should lower the retail price of the MBP because it doesn't "just work." People who own these machines paid a premium for them over every other comparable laptop. I have a late-2013 MBP 13 (one of the potentially affected machines), on which I am currently running Mojave because Apple has been guilty of this kind of garbage so many times in the past. Think of how horrible it would be to put your daily driver laptop in the shop for a few days when you have important work to handle. That's the reason why Apple can't just mail it in when rolling out OS upgrades. I have had similar situations in the past, and you know what the result has always been? I stop buying hardware from the company when it becomes clear that it doesn't give a damn about what happens to the customer. Bose rolled out firmware to the QC35 that nerfed the ANC, and then pretended that nothing happened -- essentially gaslighting the customer. I never bought another pair of Bose ANC headphones after that (I had purchased 5 pairs in the year prior to that, as gifts and for myself). Google rolled out an Android upgrade (Oreo) that immediately sent my Nexus 6P into a Bootloop of Death upon installation, and then passed the buck to Huawei. I swore never again to buy a Google hardware product as a result. I just bought a MBP 13 a few days ago, because despite all of the BS that Apple has done, they have always turned out to support the customer in the end. But if their defense sounds anything like what you are suggesting -- that there is always some collateral damage when an upgrade is rolled out because QA is not a perfect process -- then I would rethink my purchase decisions in the future.
- totalZero 6y ago> If they can afford to buy back a hundred million dollars' worth of stock every year billion* is what I meant to write.
- llamataboot 6y agoI don't know, there's just not that many configurations of mac hardware in the world. Bricking a hackintosh sure, bricking a MBP?
- m4rtink 6y agoIIRC while there are no that many official macbook models, there are actually multiple subtle variations released every year - Linux distros found that as users reported a distro working on MacBook 201x but not on a seemingly similar MacBook 201x (same model year). Why they do that is unclear to me (component EOL/cheaper components beclming available ?) but they paper it over in their firmware and OS code, so custo ers using MacOS don't notice. Well, until one of these sub models gets brucked.
- xenadu02 6y agoBulletproof is difficult to achieve thanks to hardware failures. Additionally there is a wide range of older hardware out there that may not have been engineered with the necessary support (eg enough space to hold two copies of the firmware). But putting all that aside: the new sealed system partition is a step in that direction (ROSP was partially about getting clean separation between the system and user files to make snapshot-based updates possible). With snapshots if the update fails you still have the previous snapshot to fall back on - hopefully that will make these kinds of failures less likely in the future.