Predictability may seem boring to some, but it’s essential when co-ordinating multinational projects involving many thousands of different organisations – such as releasing a major new version of an operating system.
Last week’s release of iOS 12 followed the established pattern. Prior to June, in-house alpha or development versions had been worked up to the point where they were ready to demonstrate, then release as a first beta to third-party developers around the world. Once the beta releases were sufficiently functional and stable to offer more widely, Apple opened its public beta scheme.
During the next couple of months, a succession of betas moved ever closer to the final release, then about a week before that was due, Apple released to those testers what it considered was going to be the final release: the Golden Master. This gives commercial developers, system administrators, and everyone else who needed to hit the ground running a chance to test, and report back any serious issues which had to be fixed before the final release.
When everything goes right, what is put on the upgrade servers is exactly the same as that Golden Master, and there are no nasty shocks or surprises when it all goes live for millions of users.
Last year, that didn’t happen with High Sierra. There were some major changes made at the last minute which turned what should have been the peak of the cycle into a debacle.
Most significantly, late problems seem to have been discovered in the macOS implementation of APFS which rendered it unfit for use on anything other than SSDs, something which Apple is only now, a year later, ready to fix. Apple also made the baffling decision to use an essentially untested two-part installer, which had to be replaced almost as soon as High Sierra was released. Then users stumbled into several security gaffes which Apple had to patch in a seemingly relentless series of ‘supplemental updates’.
These and many other problems came down to rushing an immature product to market, something which Apple cannot afford to repeat with Mojave.
All the signs with Mojave so far have been very promising. Yes, it has features, notably its privacy protection, which have been causing developers grief. To some extent those are inevitable: for apps delivered independently of the App Store, there has been nothing quite like its restrictions and behaviours before. For some users these may prove rough edges, but over the next few months should mature and become less obtrusive.
More critical are fundamentals which High Sierra got badly wrong: the APFS file system and security. APFS has had another year to improve, and judging from changes in version numbers, its engineers have been working intensively on Mojave’s implementation for the last six months. Whether this will address issues such as the inability to make an APFS AppleRAID bootable remains to be seen. So long as it delivers reasonable performance and reliability on Fusion Drives, I think it will win a great many friends.
Where APFS has been badly lacking is documentation. In the last few days, Apple has addressed that in the release of a reference manual detailing the internals of APFS running unencrypted on disks other than Fusion Drives. Although a big and very important step forward, for the great majority of the changes in Mojave the best available documentation remains the presentations from WWDC in June. Changes made since then, which should at least have been described in release notes, usually remain unrecorded.
Apple is steadily changing for the better. With a bit of luck, High Sierra will oxymoronically turn out to be its nadir, and the vast majority of us who are making our final preparations to upgrade to Mojave won’t look back.
Those who are about to upgrade salute you, Apple.