Archive for April 5, 2018

Thursday, April 5, 2018

New Mac Pro Won’t Arrive Until 2019

Matthew Panzarino (Hacker News, MacRumors):

“We want to be transparent and communicate openly with our pro community, so we want them to know that the Mac Pro is a 2019 product. It’s not something for this year.” In addition to transparency for pro customers, there’s also a larger fiscal reason behind it. […] This is why Apple wants to be as explicit as possible now, so that if institutional buyers or other large customers are waiting to spend budget on, say iMac Pros or other machines, they should pull the trigger without worry that a Mac Pro might appear late in the purchasing year.


To be blunt: Is this the original story arc of the Mac Pro’s development, or are we looking at a roadmap that has a fundamentally different timeline than one year ago.

“I don’t think that the timeline has fundamentally changed,” says Ternus. “I think this is very much a situation where we want to measure twice and cut once, and we want to make sure we’re building a really well thought-out platform for what our pro customers are doing today. But also with an eye towards what they’re going to be doing in the future, as well. And so to do that right, that’s what we’re focusing on.”

While there are no further details on the exact shape that the Mac Pro will take, Boger says they are still very much in the modular mindset.


My recent conversations with Apple (including the ones cited in this piece, but not those alone) lead me to believe that they know they kept going on a path with pro customers that they felt was working long after it had, in fact, begun to erode. I’m not exactly sure what the timeline was, but given the fact that the Mac Pro won’t arrive until 2019, I’m guessing just before the round-table discussion a year ago.

I don’t see a way to spin this as anything other than crushing news. Either pro customers are still a low priority at Apple or the company is just unable to execute. Yes, technically Apple did not promise last year that the Mac Pro would arrive in 2018, but that was the clear implication. If the timeline hasn’t fundamentally changed, that now seems like a deliberate misdirection. And, even now, Apple is not committing to “early 2019,” just sometime during that year.

How could the “path” have been working until 2017 when Apple knew in 2014 that the 2013 Mac Pro was a thermal dead-end? The 2013 design had already been met with skepticism when it was announced at WWDC six months before shipping. For many users, there hasn’t been a “real” Mac Pro update since 2011. The implication is that until a year ago Apple’s only plan was to build iMacs.

And as far as rectifying things, Ternus’s comments make it sound Apple is taking exactly the wrong approach. This was not the time to do fundamental market research, rethink what a professional computer should be, and come up with a revolutionary new design that will last. The way to show pro customers that Apple cares would have been to give them what they were asking for—a new cheese grater—as soon as possible and then get working on the future. (Foregoing a stopgap model might have made sense if Apple had been targeting, say, early 2018, but apparently that was never the case.) Making perfect the enemy of good just sends the message that Apple isn’t listening and you should switch to Windows.

Instead, everyone is stuck waiting, yet we still have no idea what the new Mac Pro will be, and thus whether it’s worth waiting for. Apple hasn’t said what it means by “modular.” If all it means is “no built-in display,” that’s not very helpful. Certainly, this is not “transparent” and “open” communication.

And, in the meantime, there hasn’t been so much as a CPU bump to the 2013 Mac Pro. This was the “Can’t innovate anymore, my ass” model that was supposed to show Apple hadn’t forgotten about the pro market.

Thomas Ricker:

The Mac Pro, announced in April 2017 is not coming until 2019. If Apple’s development lifecycle is generally two years, then the April announcment was pretty much an “oh shit, we need to respond to consumer outrage moment.” Bizarre.

Justin Flood:

I can’t imagine why sourcing a pretty space grey xeon motherboard from foxconn and CNC milling a slightly shrunken and thinned version of the Cheesegrater case should take this long. Unless of course, they are overthinking it. Which they are.

Steven Woolgar:

When Apple transitioned to Intel, they sent generic Intel motherboards in the cheesegrater G5. Suck up pride and just ship us a damned top of the line motherboard and GPU with drivers now (outsource it like the displays). 6+ years of waiting on a Pro machine is a bit much.

Steve Troughton-Smith:

Great that Apple is being open about its work with professionals. But yeah, with a six year gap between releases, be nice if they could give more detail about the plans for this product. Maybe find out ahead of time whether they’re building the right/wrong machine this time?

Paul Haddad:

Just stick a freaking Xeon in a large box. Literally should be the easiest product they release in any year.

Ryan Jones:

Bingo. Am I a broken record yet? STFU about new products TIM! AND, if you can’t (like Mac Pro) then don’t semi-lie. Set realistic goals, then add fluff, and over-deliver.

This unforced errors, and “delays” because you set your own deadlines you can’t hit is so idiotic.

Michael Love:

Interesting that now we know it’s 2019. But not much to offer hope of a pro-oriented MBP revamp - it feels to me like they’ve as with the Mac Pro they’ve painted themselves into a thermal corner making them so thin + can only partly mitigate that with eGPUs.


Talk is irrelevant when the action reveals disdain for pro user. Had a lengthy discussion this AM about Pro direction, the widespread attitude is find a new platform for work. And try to convince purchasing managers that 2015 hardware is needed for ports, escape key, etc 🙃🤔🙄

Steve Troughton-Smith:

As somebody who desperately needs a new headless desktop Mac, it’s incredibly frustrating that Apple can’t ship one. Apple’s determination to create the ‘perfect’ Mac Pro instead of just shipping a damn box leaves us with no options. At this point I’d take a damn G5/DTK chassis

Previously: The Mac Pro Lives.

Update (2018-04-05): Tim Ekl:

Taking this thought one step further: what if the Mac Pro were the first ARM-based Mac? What if it shipped with a couple dozen ARM cores, happily parallelizing the kinds of workloads the Mac Pro is known for?

That would be interesting—and yet another reason to ship a stopgap cheese grater.

John Gruber:

Sure, I wish the new Mac Pro were coming sooner. But overall this story is fantastic news for pro users — it shows Apple not only cares about the pro market, but that they’ve changed course and decided that the best way to serve pros is to work with them hand in hand.

Josh Centers:

Apple should just reopen its clone program at this point.


This gigantic delay between announcement and delivery? That tells some of us that there’s no point in waiting, because if they were going to give us what we wanted they could have done it pretty much instantly: give me a plain old tower — you can make it all fancy and aluminum and pretty, who cares, I’ll even shrug and pay extra for it — with a plain old motherboard and some plain old PCIE slots.


Based on all that, when I hear them say it’ll be modular, I shudder to think what their interpretation of that will be[…]

Update (2018-04-06): Nolan O’Brien:

Mac Pro and Mac Mini for years were the easiest products for Apple. Bump the damn specs and ship it. Maybe adopt modern I.D. improvements over time, but holy hell this is abysmal.

The complaint of the Mac Pro in 2012 was not that the I.D. was stale, but that the specs were.

Andy Lee:

I wonder if (a) the decision makers at Apple are struggling with cognitive dissonance -- if they’re honestly incapable of grasping how easy it would be, in this particular case, to delight us with a new thing that isn’t revolutionary; or (b) more cynically, they know perfectly well how easy it would be, but are terrified of hurting the mystique around Apple design by proving their detractors are right, that there’s nothing uniquely magical about Apple’s hardware; or (c) we’re all wrong about the obvious thing to do now, and in the long run, what they’re doing will actually be best for all concerned, even if it’s hard for us outside Apple to see it. There was a time I could have believed (c), but I don’t now.

Bob Burrough:

I don’t care if you put 32 cores in it and 256GB RAM. If it’s a sealed, all-in-one, non-expandable product, it isn’t Pro.


Workflow teams from StarWars were the same narrow group that liked the trashcan. I’m getting worried about this upcoming “modular” Mac. The trashcan Mac was “modular” with a octopus nest of wires and boxes snaking from it. I don’t want a “modular” Mac. I want an expandable Mac.


I don’t think Apple considers developers Pros. They are building these machines for video and animation studios that buy dozens at a time. Why would this Apple care what a single developer thinks if they’re only going to buy one pro machine every 5-10 years? I don’t agree, but...

Sebastiaan de With:

I have to say, though: this makes no sense to me. Pros need new Mac hardware more than anything, and it’s needed yesterday. You can’t research for all current (and future) workflows. The Mac Pro is half a decade old. Release. a. new. one.

John Martellaro:

I’d like to add something, however, that hasn’t been widely discussed. And that is the notion of technical admiration and aspiration versus mere aesthetic design.

Update (2018-04-09): Benjamin Mayo:

It’s certainly interesting that rather than questionnaires or soliciting phone calls, Apple is actively hiring these creative professionals in-house. […] I guess the danger here is that you tune too heavily towards the workflow team’s base of talent, which is currently composed of video artists, animators and music technicians. Requirements from other fields — of which the ‘pro’ market is very vast — may be de-prioritised or ignored entirely. Software development comes to mind.

Update (2018-04-14): See also: Accidental Tech Podcast, The Talk Show.

Update (2018-08-23): Steve Troughton-Smith:

Nvidia’s new RTX 2080 Ti is a mind-bogglingly-powerful GPU; that very likely means Thunderbolt 3 isn’t gonna come anywhere close to maximum performance as eGPU. that new Mac Pro better be something, because otherwise Mac users are gonna be left behind in a graphics revolution

It does suck that if you want a GTX 2080 Ti-like performance in a Mac, the only option you have is “wait 5 years and maybe AMD will have something as powerful that Apple’s comfortable shipping”. This is precisely why the Mac Pro needs to exist, and why it needs to have PCIe slots

Update (2019-01-01): Marco Arment:

When I made that $5 bet with @caseyliss in @atpfm #213 that the 2013 Mac Pro would still be for sale on January 1, 2018, I wouldn’t have predicted that it would also still be for sale on January 1, 2019.

At this point, it’s remarkable that they can even get the parts for it.

Update (2019-05-14): Malcolm Owen:

The first “leak” for the modular Mac Pro has surfaced in the form of a supposed internal Apple document for the device, alleged evidence for the highly-anticipated Mac that has many questionable elements making it very hard to believe is genuine at all.

App Store Shrank for First Time in 2017

Sarah Perez (Hacker News, MacRumors):

The App Store shrank for the first time in 2017, according to a new report from Appfigures. The report found the App Store lost 5 percent of its total apps over the course of the year, dropping from 2.2 million published iOS apps in the beginning of the year to 2.1 million by year-end.

Google Play, meanwhile, grew in 2017 — it was up 30 percent to more than 3.6 million apps.

Appfigures speculated the changes had to do with a combination of factors, including stricter enforcement of Apple’s review guidelines, along with a technical change requiring app developers to update their apps to the 64-bit architecture.


Android developers kept busy, and in the last year alone released more than 1.5 million new apps. This is an increase of about 17% year over year, the largest jump since 2014. iOS developers on the other hand, took their time releasing just 755k new apps in 2017. That’s a big drop! 29% to be precise, the first drop since the App Store launched in 2008.


More than twice as many iOS apps came to Android in 2017 than Android apps came into the App Store.

Instagram Disables APIs

Juli Clover (Hacker News):

Along with multiple privacy-related API changes being made to Facebook, Facebook, which owns Instagram, announced that it has disabled several Instagram Platform APIs as of today, disabling certain Instagram features that are available in third-party Instagram apps.

Third-party Instagram apps will no longer be able to use APIs that provide access to follower lists, likes, relationships, and public comments.

The actual changelog is not very clear, saying only that the APIs are “deprecated,” which normally means that they will be removed at some time in the future. In fact, they seem to have been turned off yesterday.


The important thing to note here is that these endpoints were scheduled for retirement at future dates; some in July 2018, some in Dec 2018. Today, with no prior announcement, these endpoints were abruptly retired.

Instagram had previously published a deprecation schedule for these endpoints, and threw that schedule out the window without notifying even Instagram Partners.

Making the Touch Bar Useful by Abandoning Apple Guidelines

Vasily Zubarev (via Hacker News):

The Touch Bar should always be predictable. I don’t want to remember button configuration for every app. I want it to be useful and informative. Just give me an additional screen, like in the early fan renders.


The grown-up me started to track my everyday habits and search Reddit for ideas. The only proper app for Touch Bar customization I’ve found, is BetterTouchTool. It’s still buggy, user-hostile and drains the battery, but it only costs $5. Reasonable price, I recommend it.


Standard Touch Bar will never support the most useful feature — to display the name of track playing. Instead, Apple provides you with huge and useless timeline slider. I don’t know what’s playing but know the moment in time.


A lot of cool guys wanted to share their presets too, so we created a GitHub-repo for that. Feel free to add your own:[…]

Previously: Making Better Use of the Touch Bar.

Update (2018-04-09): hrbrmstr:

Combining @boastr_net’s BetterTouchTool & @holman’s spark to add auto-updating sparklines to the macOS Touch Bar

Quiet By Design: Naomi Campbell Interviews Jony Ive

Jonathan Ive (via 9to5 Mac):

One of the defining things about the nature of ideas is just how fragile they are: when you’re not sure whether some-thing is going to work, the idea is vulnerable. Part of protecting the idea is to be careful about who you show it to; premature criticism can shut something down that perhaps deserves more of a chance.


Ultimately, Steve’s legacy is a set of values and, I think, the belief in trying. Often the quietest voices are the easiest to overlook, but he was brilliant at listening as well as leading and speaking. A lot of communication is listening – not just listening to figure out what you want to say in response.

BlackBag on APFS Encryption

Joe Sylve:

Whether or not you “unlock” an APFS volume during acquisition, physical APFS images will still contain encrypted data. Analysis tools such as BlackLight will then need the proper user password or recovery key to decrypt this data on demand, as you are conducting an examination. It is not technically possible to create a forensically sound physical image of an encrypted APFS volume that contains logically decrypted data.


APFS encrypts data blocks using the XTS-AES-128 cipher. In order to decrypt data, several pieces of information are required:

  • The encrypted data
  • 128-bit Volume Encryption Key
  • 128-bit Secondary Encryption Key
  • The original “block number” of the file

Each volume in an APFS container uses unique volume and secondary encryption keys. When a file is deleted, its data blocks are released to the container pool. At that point it is not possible to map an unallocated data block back to its original volume. We therefore do not know which encryption keys to use to correctly decrypt the data. This process is further complicated due to the fact that encrypted blocks can be relocated on disk. When this happens, the data is not re-encrypted, and the original block number must be known to decrypt the data. This information is generally lost when a file is deleted.

Update (2018-05-01): See also: APFS Internals for Forensic Analysis (PDF, via Enno Rey).