Last Week on My Mac: mid-life mistakes

Last week we crossed the hump and entered the second half of Mojave’s public release cycle, with the arrival of 10.14.3. By now all its features should be securely in place, and most of its major bugs fixed. It’s from here on that Apple may add elements which prepare the ground for 10.15, and its development teams start focussing more on that next major release.

Internally, the signs are encouraging. The version number of APFS has stopped rocketing upwards, now advancing at a more pedestrian pace. Some features already seem static rather than just stable: support for full linguistic parsing hasn’t spread beyond the seven or eight languages which it reached soon after the release of 10.14, and still doesn’t include key languages such as Japanese or Chinese, or well-researched European languages such as Dutch or Polish.

There’s ample room for improvement in some of the newer apps, in particular the App Store, but like Photos before it, Apple seems content to let it evolve its way to acceptability rather than squaring up to its many problems and fixing them.

In the release of 10.14.3, Apple made two howling errors, though.

The first was in the Combo version of the standalone updater, whose scripts prevented it being installed on many Macs. Building such installer scripts isn’t easy, but it shows yet again that code checking and good quality management aren’t Apple’s strong suits, a year after Apple was driven to admit that some of the gaffes in High Sierra should never have seen the light of day.

The other was in the total lack of release notes, other than one remark which would only have been of interest to enterprise users, and some brief security notes. Compare the release notes between 10.14.3 and 10.12.3, which are only two years apart. Not only that, but Apple even refuses to provide information about 10.14.3 to its developers: on this page it states simply “There are no release notes for this software update.”

Where Apple has, with great reluctance, bowed to pressure from developers and published some documentation, notably in its limited account of APFS, progress has been pitifully slow. The first version, published last September, has at last been revised to provide information about software encryption in APFS – a task requiring an additional ten pages that has apparently taken Apple four months to complete.

This brings one fascinating piece of information, though. Software encryption (FileVault on APFS disks other than internal storage in Macs with T2 chips) relies on the UUID of the container or volume. It is thus extremely quick and simple to render a whole APFS volume inaccessible by deleting or changing its UUID: it then can’t be decrypted at all, and can only be deleted.

Then came the rather strange story of Legacy Software – the section in System Information which has behaved most bizzarely since we first got Mojave. Lots of people keep telling me how Apple isn’t investing in macOS development unless the code is common to iOS. Yet Apple appears to have invested significantly in a carefully-designed nudge campaign to get us to upgrade all our apps to 64-bit before macOS switches 32-bit support off, presumably also in 10.15 later this year.

The idea is simple: start by alerting the user to the occasional newly-installed 32-bit app. Then rack the pressure up, displaying the warning dialog when all existing 32-bit apps are opened. Meanwhile, accumulate a blacklist in the Legacy Software section of System Information. As it steadily grows, the user will feel more pressured into taking action.

Whether or not Apple has got its psychology right (I don’t think it has), this isn’t a quick hack on a shoestring. Apple has extended its SystemPolicy service with a new sub-system, SPLegacySoftwareReporter. As far as I’m aware there’s nothing like it on iOS, and I don’t recall Apple using any similar technique when iOS went 64-bit only.

Apple may not be investing in what we consider its priorities should be – fixing annoying bugs like the Finder hanging when trying to display the contents of a broken folder alias, or documenting macOS for developers. But it is certainly investing in what it considers are its priorities, which includes this final step in migration to 64-bit.

Over the next six months and more, I think there are going to be plenty of Apple’s engineers working on macOS 10.15. I wish them success.