Note it looks like it's for developing drivers to run on embedded ARM devices in silicon validation. I'm at least as uninformed as anybody here, but that has lots of non-earth-shaking interpretations.
If you have, say, a farm of little ARM-driven devices that you use as test harnesses for chips or other hardware, it isn't necessarily cheaper to port whatever variation of xnu to this particular hardware than to use an existing port of Linux and hire some folks who can help with the low level I/O bits. There might even be Linux in your world because, e.g. an acquired company built something on it that hasn't been worth porting.
Or any of a ton of other possibilities.
Having an OS doesn't mean you have to use (variations of) it for everything, especially if the new use you're thinking about is far from your usual.
I've seen multiple people stop all their open source contributions and communicating online after being hired by them. I assume they have very draconian NDAs.
From what I hear, at Apple you're not allowed to contribute to open source projects (unless it's part of your day job) even in your own time - in contrast to Google (from what I hear) and Facebook (personal experience) where you can but you must clear it with legal and it can't be related to your day job.
And how does it work in CA? AFAIU, there're laws which prevent such limitations in most cases. I.e. everything which is done on the person's time is their own choice.
It is hard to find something that can't somehow be considered to compete with Apple. And in CA you can be let go for no reason, so...
For example, I couldn't get approval to contribute patches to a gallery plugin for Wordpress that I was using for my personal blog, because you know, the Photos app is a gallery of photos.
I stayed a few years for a few reasons, first because it is still a good company with a “real” product, and also because 99% of my time there was spent working on open-source (clang/llvm).
When I decided to quit, it was in part because of their open source policy.
There are tons of developers who not only don't care about writing open source software on the side, actually go home after working 8 hours a days on computers and decide not to do any code at all and spend time on other hobbies, with their family, etc.
Nobody is going to confirm it in any way you’ll find acceptable (like verifiable list of names/contributions), for one obvious reason: Apple employees tend not to post identifiably in public because articles (like this one) try to extrapolate public knowledge with hilariously wrong guesses at the roadmap, and it creates risk that isn’t worth the trade offs for the employee.
Most Apple SOCs boot Linux inside apple HQ to run tests. There are drivers for everything and it all works. Linux turns out to be a more convenient way to do validation and testing than iOS. Obviously this code is never released outside of apple.
My response was just to counter the common misapprehension that macOS itself is built on a FreeBSD userland. It's NetBSD, and devices like the AirPort Extreme/Time Capsule use NetBSD directly. And, back when the iTunes Store was a WebObjects service, it was hosted on Xserves running a weird reverse-Darwin (i.e. a Darwin-shimmed userland atop a NetBSD kernel, because macOS itself wasn't yet performant enough to be a good server but they still needed WebObjects to run.)
These days, Apple's datacenters are running all sorts of things, including Darwin, Linux, Solaris, and AIX[1]. But the "Darwin" part is still essentially NetBSD, not FreeBSD.
> My response was just to counter the common misapprehension that macOS itself is built on a FreeBSD userland.
I don't think most people say that macOS is built on top of a FreeBSD userland. Most people rightly point out that there is a lot of FreeBSD code inside the macOS code base specifically the XNU kernel[1].
> But the "Darwin" part is still essentially NetBSD, not FreeBSD.
That's incorrect while there is NetBSD code in MacOS. Dismissing the FreeBSD code, the Mach code, the code Apple wrote and saying that Darwin is essentially NetBSD is extremely disingenuous[2].
Bingo. Apple has a good amount of engineers working on Linux and QNX embedded devices. I think it's pretty clear at this point that they are still working on cars and that they are also integrating those cars with Apple maps. I think they want to kill two birds with one stone with Project Titan: get more map data and develop software for self-driving cars.
Texas and California; I wonder if that's a clue too. If so, it'd be very interesting, especially at a time when Waymo is now the only one left with a clean sheet. I wonder how far away Apple is from a product?
I would expect that the wholesale change from Darwin to Linux would introduce an amazing number of new and awful bugs. It's not like their development processes will suddenly be fixed when it's been touched by Linux.
I believe Linux (the kernel) has one of the most productive, QA production lines out there. If Apple were to adopt the "Linux way", we may indeed see a change ..
On the other hand, it may work against us, having Apple involved in Linux things. I'm not sold either way - but I sit here, re-installing MacOS on two Macbook Pro's, both of which absolutely refuse to make a usable/safe disk image out of their own disks. This is such a frustrating conundrum, it feels like Solaris in the 90's. I can only hope something changes at Apple, or else I'm just going to wipe MacOS and put Linux on these machines. That is quite a watershed, imho.
Yes, I've done it and Disk Utility doesn't think there's anything wrong. But yet, I've tried over and over to make a disk image - of different disks, even - and every single time I get the error "no usable disk image found". At this point, after having scoured the web for clues, I can only conclude: this is a bug in MacOS.
I'm now trying to do a full update to the latest release, and see if it works beyond that point - but if it doesn't, the good ol' tarball approach is all I've got left.
yes i noticed that job posting and it is actually very close to where I live, I even thought about applying, then I'm not an apple fan and I don't appreciate the way Apple deals with OSS, so I stopped.
it's about chip validation and has nothing to do with MacOS etc, because it will be pretty unwise to use MacOS to check silicons.
Although maybe it's too hard to find "osx kernel developers" outside of Apple so they look for the next best thing. It may not be a sign of them switching to Linux.
A very far fetched possibility would be that Macs will run iOS - no UNIX stuff exposed, everything signed and secure. An embedded SoC will run in secure space and expose Linux in a iOS terminal application so you can do all your development work - ssh, vim and whatnot there.
Kind of like WSL but instead main OS is completely locked down and instead of software emulation you get isolated hardware to run Linux.
Secure lock down is not possible with hypervisor that allows you to run any OS. The idea would be to only run what Apple allows - a tailored Linux distro - and only on the coprocessor which is isolated from the fully locked down main OS.
Oh and by the way "perfectly good" is way too generous for the hypervisor macOS ships with, unless there have been recent drastic improvements..
What's wrong with exposing UNIX stuff? When I jailbroke my iPhone and ssh-ed into it, there were UNIX tools there. I'm not sure whether they are from iOS or from jailbreak.
Can’t see the point. I need my Unix tools, like git, ssh and wget etc to have full file system and network access. Having them run in a hypervisor adds overhead and complexity for no real benefit.
To enable iOS apps on the Mac, they only need to activate the iOS developer API with their existing OS - Darwin. AFAIK, iOS and macos share the same kernel, more or less
Just to be clearer, when you run iOS apps under the simulator, the code is compiled as x86 assembly and linked against an x86 version of the iOS framework. There's no emulation involved.
If you have, say, a farm of little ARM-driven devices that you use as test harnesses for chips or other hardware, it isn't necessarily cheaper to port whatever variation of xnu to this particular hardware than to use an existing port of Linux and hire some folks who can help with the low level I/O bits. There might even be Linux in your world because, e.g. an acquired company built something on it that hasn't been worth porting.
Or any of a ton of other possibilities.
Having an OS doesn't mean you have to use (variations of) it for everything, especially if the new use you're thinking about is far from your usual.