Next week, Apple is expected to launch a new range of iPhones and iOS 11. In doing so, it is likely to regress earlier efforts to convince Mac users that it still takes them seriously. For those new iPhones and major release of iOS will quickly turn the later release of High Sierra into something of a disaster.
At the centre of these problems is Apple’s new file system, APFS. Under development for some years now, Apple announced it at WWDC in June 2016, and every iOS device which has been updated to run iOS 10.3 has been running APFS for the last (nearly) six months.
Introducing a new file system is a very major undertaking, not something to be breezed away in a couple of press releases and fatuous PR claims. It is comparable in scale and effect to introducing Mac OS X itself – something which Apple left in public beta for more than six months. APFS has instead had just two months in public beta, over a period when many people around the world take their major annual holiday.
The normalisation fiasco
Even before those public betas, some of Apple’s decisions about APFS have proved to be misjudgements. Most obvious was the design feature that the new file system would not perform any Unicode normalisation of file and folder names – which was still a feature of APFS when it was released to hundreds of millions of iOS devices back in March.
Apple has now recanted over normalisation, having to retrofit it to iOS (and pre-release to macOS too) to avoid catastrophes even with its own software products. Had it listened more to its developers and others – even I pointed out here what a mess was occurring – Apple could have saved itself time, cost, and corporate embarrassment.
Chain of events
The situation Apple now finds itself in is entirely of its own making. Releasing new iPhones this coming week triggers a chain of events which will result in further loss of corporate face. This is not because of the shortage of new iPhones which it will be able to ship – that will of course be blamed on its suppliers, and often strengthens market demand for fashionable consumer devices anyway – but because of the impact on Macs.
Being such a closed and secretive corporation, we don’t know how well beta-testing of APFS has gone on Macs, but there appear to have been a lot of reports that it has not worked well on Fusion Drives and traditional rotating hard disks. Apple seems to have been aware of such problems in June 2017, during WWDC, and warned that further work was still needed to get APFS optimised on hard disks.
Then at the end of August (and revised on 5 September), less a fortnight before the iPhone launch, Apple quietly released a note aimed at system administrators, who by then would be planning how to roll out High Sierra upgrades. In clear and dry language, it stated:
When you install macOS High Sierra on the built-in solid-state drive (SSD) of a Mac, that drive is automatically converted to APFS. Fusion Drives and hard disk drives (HDDs) aren’t converted. You can’t opt out of the transition to APFS.
It also dropped the bombshell that Sierra would never be able to access volumes formatted using High Sierra’s release version of APFS:
Devices formatted as APFS can be read from and written to by:
- Other devices formatted as APFS
- Devices formatted as Mac OS Extended, if using macOS High Sierra
For example, a USB storage device formatted as APFS can be read by a Mac using High Sierra, but not by a Mac using Sierra or earlier.
This is an issue which I have already discussed and warned about nearly two months ago, and once again, Apple seems to have made a serious misjudgement. Interestingly, my article was savaged by one commenter, who asserted:
Incorrect. Apple has on numerous occasions released an update for a prior OS, after the new OS has already been released. I expect to see 10.12.7 at some point, possibly after High Sierra is released.
Unless Apple reverses the announcement in this most recent note to sysadmins, those words need to be eaten, together with at least one substantial hat.
So the release of APFS for macOS is proving, once again, long experience that introducing a new file system is a very major undertaking, not something which can stick to a rigid marketing-driven timetable. In the face of that, Apple decided that it would release its new iPhones on 12 September, requiring the release of iOS 11, which supports the new models and their quest for further consumer ecstasy.
Now that there are many links between iOS and macOS, including common use of iCloud features which are built into iOS 11 and macOS 10.13, Apple then has to release High Sierra fairly shortly after iOS 11 – just as it has done over the last few years. High Sierra could (and almost certainly will) be released a week or two later, but it would be unconscionable to delay it for many weeks or months.
The result is that a small proportion of current Mac users will actually be upgraded to APFS when High Sierra is released. Those who are today paying up to $/€/£4,000 for brand new 2017 iMacs with Fusion Drives will be rightly incensed that they have little to gain from the trauma of upgrading to High Sierra until the problems in APFS are fixed.
It is widely accepted that Fusion Drives should benefit greatly from APFS, which will at last enable them to escape from reliance on CoreStorage, an additional layer interposed to make their two drives look like one. CoreStorage has served its time, enabled Apple to sell many millions of Macs with high-performance large storage capacity at relatively low prices, and needs to be replaced by APFS, just as HFS+ itself badly needs replacement.
As it stands, if you have just bought a brand new 2017 iMac with a 3 TB Fusion Drive, the only way that you will be able to get full benefit from High Sierra is to unfuse its Fusion Drive to separate SSD and hard disk, and format the SSD in APFS, the hard drive in HFS+, which defeats the whole purpose of having a Fusion Drive in the first place.
What Apple should have done was to have separated this autumn’s macOS release from that of APFS. Without the new file system, High Sierra surely would not merit being a major system release, and would have made a very attractive macOS Sierra 10.12.7. When APFS was ready to run on the majority of modern Macs, it could then have been released in 10.12.8, which would have ensured that Sierra systems would be able to use the finished version of APFS too.
High Sierra 10.13 initial release is likely to call into question again Apple’s commitment to the Mac and macOS. Many Mac users will rightly feel that they have been badly done by, and the apparent determination of Apple this year to make Macs and macOS a central part of its business will seem to have evaporated.
All because Apple made the oldest mistake in the industry, and let product marketing schedule its engineering development.