Last Year on My Mac: Accident-prone

The year just passing was never going to cruise easily by.

Apple needed to reaffirm its commitment to the Mac with a broad range of new models, and over the first half at least, was doing quite well. Sierra matured steadily – this time last year we were still running 10.12.2 – but by 10.12.6 a lot of unfixed bugs remained: silly things to keep reminding us that Apple couldn’t be bothered to fix them, like Bluetooth spontaneous disconnect-reconnect cycles, stuttering wireless keyboards, and having to restart every week or two because backups stopped running.

Instead of having a last big cleanup, in the summer Apple abandoned Sierra as a work in progress, only without any more progress. Most seriously this resulted in Sierra’s support for APFS being left in late alpha, and broken the moment that High Sierra was launched in the autumn.

Apple kept its promise to launch new Macs, and the summer saw one MacBook, three MacBook Pro, and three iMac models, as well as two more iPad Pro models in addition to the Fifth Generation launched in March. No further progress has been seen in the evolution of the Touch Bar, though, and this year’s designs have remained conservative, apart from the iPhone X and its beguiling animoji.

I branded High Sierra a lemon, and for a great many Mac users it remains so: an upgrade without any tangible benefits, but some significant risks. Apple fluffed the release badly too. Substantial changes were made after the last Golden Master, in particular to APFS, and the initial two-step installer had to be pulled very quickly because it simply didn’t work.

That chaos was nothing compared to the ensuing weeks. The brand new, shiny, all-singing macOS couldn’t actually use its major feature (APFS) on the majority of desktop Macs, as it was only safe and reliable on SSDs. As the new file system had already lost most of its clothes, we realised that support for new image and movie compression (HEIF/HEVC) was all but absent too. Early adopters were reduced to enthusing over a few new features and bug-fixes in Photos and Mail.

If Apple failed to give most users good reasons to upgrade, it gave everyone good cause to stick at Sierra, with two major security flaws. The first was a simple programming error which, in some circumstances, showed the password instead of the password hint for encryption. As if that wasn’t good enough evidence of High Sierra being rushed to market weeks too soon, the root user vulnerability certainly was. Apple found itself a major story in national news media around the world, for what can only be termed corporate stupidity.

Response to those gaping vulnerabilities was swift and effective, but in the latter case had to be progressively bodged with a succession of additional updates.

Having started the year with such promise – new Mac hardware, new macOS, new iPhones, new iPads, and new file system – Apple found itself in the headlines again over its handling of iPhone battery ageing problems. It was another case of how Apple may have made the right technical decision, but because it deliberately didn’t discuss or even explain what it was doing, it fell short in its customer care. A perfectly reasonable engineering solution was so readily interpreted as an underhand marketing ploy.

There was also a great deal that Apple needed to do, but hasn’t. AppleScript is still slowly dying of starvation and neglect, but Apple has failed to come up with a suitable alternative. Swift has moved on, its pace of change living up to its name, but lacks an appropriately lightweight development environment, and remains unsuitable for scripting other apps.

If Apple’s operating systems have had a stormy year, its developer tools have been basking in sunshine. Apple introduced Swift 4.0 with an excellent tool for code migration, in Xcode 9. It most importantly reinstated runtime support for earlier versions of OS X, making it possible once again to build apps which run seamlessly across all recent releases.

As predicted, APFS caused havoc on iOS devices in countries like Korea, where its intended lack of normalisation even broke Apple’s tools. Policy was hastily revised, and normalisation retrofitted to the file system, but some users are still encountering issues with some non-Roman characters in certain situations, such as using git.

The new unified log system in macOS, iOS, and the rest, presses on in High Sierra and iOS 11, despite its appalling lack of support in Apple’s own tools. The new version of Console, an app which in El Capitan and earlier was valuable to most power users, has changed little over the last eighteen months, and remains thoroughly inadequate as the standard log browser. One of the most valuable diagnostic features of any Unix-like operating system has consequently been rendered inaccessible to a great many users.

macOS has had another charmed year with respect to security. A year ago I warned that “anyone using macOS or iOS in an environment in which spying or intelligence gathering is possible should be particularly watchful for spearphishing and more sophisticated targetted attacks.” That has sadly proved true, but the threat of waves of ransomware or more general malware has not yet materialised.

While Apple has still not recognised the importance of good, up-to-date documentation to users and developers, just before Christmas, Jonathan Levin published a mine of information about macOS and iOS which goes a long way to filling the gap.

Apple’s end-of-year report thus conflicts with its record of achievement. It has delivered a lot of new products, with a lot of new problems. Far from High Sierra ‘fixing bugs’, it has come with a raft of fresh ones, some which were so deeply embarrassing that everyone has been asking how they could ever have made it into beta, let alone final releases.

Next year, Apple needs to set itself marketing targets which it can achieve in comfort and, above all, quality. It needs to be far more open about problems and what it does to address them. Each update of macOS is accompanied by details of all the vulnerabilities it addresses, and a handful of offhand remarks about all the other things which have changed. Apple must do better than that, and must become more open about what it pushes to our Macs for silent installation.

Accidents don’t just happen. They follow trajectories through human and corporate weaknesses, which Apple must now strengthen.