So for those who, like me, wonders why Apple keeps getting macOS Unix certified, it's to avoid a lawsuit. Apple misused the Unix trademark when they first launched MacOS, so to avoid legal trouble with The Open Group, Terry Lambert was put in charge of getting MacOS Unix compliant and certified: https://www.quora.com/What-goes-into-making-an-OS-to-be-Unix...
It's basically the only relevance the Unix trademark has these days. I can't imagine many companies choosing macOS because it's a real Unix, nor would anyone really opt out of z/OS, AIX og HPUX, if they where not certified.
> I can't imagine many companies choosing macOS because it's a real Unix, nor would anyone really opt out of z/OS, AIX og HPUX, if they where not certified.
While Unix compliancy isn't what's keeping me on macOS, the Unix tools it has under the hood still is. I've opted to use it over Linux because I still get everything that I need from a "Unix like" standpoint while having some serious enterprise level support and compatibility with work software that's often only available for windows or Mac.
If Apple stopped caring about being Unix compliant, I wouldn't be surprised to see the tools and infrastructure that make it Unix (and useful to me) slowly be removed. Then I'd stop using it.
I'd say that you care about it being UNIX-like, not UNIX®. You don't care that Linux isn't UNIX. You don't care that GNU versions of things like ed and awk are slightly off-spec.
In some ways, Apple's adherence to UNIX specifications probably makes macOS less useful for you. For example, I wish that grep on macOS was closer to GNU grep. When I look up commands online, I often find answers based on the GNU implementations. Those often work on macOS, but sometimes don't (or have subtly different behavior) because macOS is adhering to the UNIX specification rather than to what those utilities do on the vast majority of systems out there.
I don't think Apple would be removing UNIX-like tools from macOS even without certification. They know how valuable it is that most developers use their systems. Even Microsoft went so far as to implement the Windows Subsystem for Linux for developers. At this point, I think that UNIX certification makes macOS less compatible with the tools and help out there which generally targets Linux. Usually the differences are small, but they certainly can be meaningful.
While macOS only really gets Unix certified they design of the flavor of unix from FreeBSD. Homebrew is also the best port system and package manager I have ever used because it requires no thinking. I actually dislike using Linux because now I have to learn the replacement from ifconfig, the creation of launchd IMHO is way better than init.d and systemd, and the command line diskutil and other additions still feel like its more Unix like while Linux feels like its moving toward its own thing. Before I was using macOS I was using OpenBSD as my daily driver since high school. I still don't understand though why Ubuntu has the ability to break /boot because there isn't enough space to add another kernel to there...
The enterprise support is actually pretty bad. they offer some cool stuff like DEP but there's so many strings attached that for enterprise it's rarely actually possible in practice. Many things are boneheadedly designed.
For example, Apple federated accounts are a great idea. But, in your global directory the UPN and email must be the same. For us it's not, with good reason. We're not going to change our entire setup globally to suit Mac users that make up 0.5% of our systems. And there's never going to be more unless they become more accommodating. We even looked at JAMF but it's too much work to implement a whole separate management system. And the options in apple's configuration profiles are way too limited.
Another issue, every account that has already been created as a private Apple id on the corporate email must be manually resolved. Impossible with many tens of thousands of users.
AD binding while rudimentarily supported, causes so many issues. If someone's password expires there's no way to log in unless they're physically on the company network. The problem is that our security team demand we bind to AD. Not Azure AD. That's just reality in enterprise.
Having a managed local admin account is also a really big problem and there isn't really any tooling for that.
Maybe if you go all in on Apple like IBM did, then yeah you could adapt your environment to all their quirks. But it's a big blocker for small deployments. And really besides IBM nobody in enterprise did this. Apple meanwhile doesn't really care anyway. They only care about the customer market.
There is absolutely no way that happens. Apple cares about Unix compatibility for very good business reasons. Apple gets a lot of value from being even just Unix-ish. It's the same reasons Microsoft eventually knuckled under and made WSL -- the ubiquity of Unix conventions and tooling is just too cemented in the industry.
POSIX conformance is a cherry on top, and it helps get them certain sales that require a conforming Unix. But that's not the real value to the platform.
There was a time when OS X was new that Apple leaned pretty hard into the OS X is UNIX(tm) story. I was working in academia at the time where the Serious Work was being done on SunOS/Solaris and saw print ads aimed at scientists reminding them that OS X was a Real UNIX:
It's not for everyone, but at some point I got tired of the FreeBSD layer being there but not cared for, and WSL+git for windows kinda provides a decent compromise in that regard.
The thing I hated the most was spending time building installation scripts or running images for the prod environment, and then go fight macos to replicate the same setup locally. Especially having parralel installs for the system and my user account was a PITA.
One would argue I could just dev on the docker image as well, but then being on a somewhat unixy OS doesn't matter much anymore.
WSL let's the Windows side live it's life (shell level tools can still be injected for convenience), and the linux side be genuinely Linux, not some ersatz, and duplicable to one's heart content. It's still less pain and better perfs than straight docker, and just extremely well integrated in general.
There are some infuriating issues though, I have wiped out a couple files on OSX using sed with the -i option to replace a text within a file only to realised OSX would wipe it out instead ....
It is also relevant, that as proven on FOSDEM corridors full of Apple laptops, most folks only care about some sort of POSIX experience, and couldn't care less about Linux/BSD religion.
So that target audience gets a cool modern experience, without fighting with driver issues and such.
It is also the reason why Microsoft ended up bringing Project Astoria from the ashes into WSL.
UNIX has won, but not as Richard Stallman would have liked to.
That explains why they got it UNIX certified back then, but couldn't they stop advertising macOS as UNIX and stop getting it certified? They even changed the name from Mac OS X to macOS since then.
That's my question too, why continue to bother? Apple doesn't even have any separate "Server" OS anymore. I can't find anything mentioning UNIX on any apple.com marketing pages.
I guess it's just, might as well keep it going, as an option for future marketing if ever needed. Maybe it helps the salespeople in some enterprise deals? I mean, if it doesn't really cost anything to keep it.
My (wholly unsupported) guess is that there are government or megacorp bids somewhere for Unix systems for employees, and this checks that box. The buyer could update their requirements, but why do that when you can just make your vendor jump through the hoop?
Not at Apple and don't have any knowledge here, but I'd imagine that the UNIX test suite, ever since it began passing, has been a useful set of additional regression tests even outside the certification context.
Does anyone want to be the person that removes regression tests from active use, only to be responsible when something breaks that would have been caught by that test? Far easier to just fix your code so the test passes.
(And for many years, OS X then macOS had a reputation for being rock-solid, capable of going much longer betwen restarts, going into BSOD much less frequently than Windows would. Having a set of third-party tests certainly didn't hurt this!)
I obviously don't know, but I could easily imagine that Apples legal team has flagged it as a potential risk and the cost of keep the certification up to date is minimal, compared to some imagined risk. Safer to pay the fee, and not having to worry about someone at Apple accidentally calling macOS a Unix system in public.
Also, Apple is a huge company, there's the question of who's going to make the call the not update a certification that's negligible within the scope of macOS development. Better to not be that person and just rubberstamp the invoice from The Open Group. If management disagree, they can make the call, but they won't because the cost is to small for them to deal with.
I think it’s a quiet but deliberate strategy to keep macOS the spiritual successor to NeXTSTEP. While many of Jobs principles are under pressure at current day Apple, his ghost lives on.
Used to work with UNIX servers in the early 2000s but out of that sector for coming up on two decades -- are z/OS, AIX and HPUX (and other old big iron enterprisey UNIXen) still around? I would have thought that Linux had killed them all of by now. Excuse my ignorance!
There’s an interesting story from the lead engineer to make OS X originally compliant:
> I was asked if I could lead a team to do #1. I said “Yes, under the condition that I could use the compliance project as a hammer to force other parts of the organization to make changes in their own code base, and that I could play it rather loose with commit rules regarding what it said in the bugs database for a given code change, and what the given code change actually did, in addition to what it said in the bugs database”.
…
> We were promised 1/10th of the $200 million, or $20 million in stock, on completion. $10 million to me, $5 million to Ed, and $5 million to Karen Crippes, who was looking for a home in Mac OS X development, I knew was an amazing engineer, and who could be roped into being technical liaison and periodically kicking off the tests and complaining to Ed and I about things not passing.
> Also, the tech lead has to fix anything no one else fixes, or no one else can fix, because they are the DRI (Directly Responsible Individual).
How many tech lead/project manager can say that they are capable for this in these days? It feels like based on my observations that other skills are taking priority on management/lead side.
Two general-purpose Linux distributions used to pay for Unix certification, although they don't do it anymore since hardly anyone is interested in it these days.
It's not just about paying for certification. You also have to replace a lot of things like ed, awk, grep, etc. with versions that are compatible with the UNIX specification. GNU utilities didn't target 100% UNIX compatibility and they have differences that mean that a command that works on UNIX might not work (or might not work the same) on a Linux distro using GNU utilities. glibc has slight differences from the spec too.
In order to get a Linux distro certified, you'd have to make changes which would make it less compatible with all the other Linux distros out there.
The reason why RedHat doesn't pay for UNIX certification is that their distros wouldn't be compliant. The reason why they don't make their distros compliant is that their customers would vastly prefer that RedHat use "standard Linux" tools than replace them with UNIX-compliant ones. Customers don't want a Linux distro that's subtly different/incompatible compared to what everyone expects in a Linux system. They'd rather it be not-UNIX.
Yes, you can modify a Linux distro to be UNIX. However, most Linux systems are not real UNIX - and you wouldn't want it to be real UNIX.
I'm not sure what you mean by "Unix specification". But if you mean the international standard POSIX, yes, people care. Red Hat routinely participates in POSIX spec revision.
There are a very few deviations where you have to enable "POSIXLY_CORRECT". If that's what you mean, then you can turn that on. But in every area that matters, Linux distros implement the POSIX spec by default, and you can even turn on the POSIXLY_CORRECT mode to exactly follow it. They extend beyond it, but that is allowed and expected.
The people who build the tools in Linux distros care a lot. I know the implementors of dash and GNU make routinely refer to POSIX. The Linux distros don't have to as much with POSIX because that is generally a conpleted work and it's the maintainers of the tools who must address the updates to POSIX.
There was a time when “open standards” were treated as the definition of Unix. At least that's what we aspired to. POSIX, X/Open, and others competed to be the standard that mattered. Formal standards were hoped to be a sounder, fairer basis for compatibility and interoperability than the earlier era of “Unix is whatever this release says it is" for some subset of 7th Edition, System III, System V, BSD, or one's favorite commercial derivative (SunOS/Solaris, HP/UX, AIX, Xenix, UnixWare, ...).
That window window of optimism—roughly mid-1980s to mid-1990s—closed fast. Open source projects and _de facto_ standards proved far more powerful in deciding where applications would run, where investments would be made, and which variants survived. Today, the real baseline isn’t POSIX in a binder or some Open Group brand certificate, but Linux + GNU + the APIs everyone codes to. In some ways we've regressed—or more charitably, we shifted back to a more pragmatic form of standardization.
the big difference is that the Linux + GNU standard is not controlled by a corporation whose only motive is profit, and that the reference implementation is Free Software that everyone can potentially contribute to.
i would not call that a regression. compare that to the browser standard which is largely controlled by google.
is not controlled by a corporation whose only motive is profit
Instead it's a bunch of corporations with very similar motives, including profit, so the end result isn't all that different. Just look at who the majority of committers work for.
I had to work around this recently - it involved reimplementing my poll wrapper on top of select(), with the proprietary macOS extension that makes it support unlimited fds. (Of course -D_DARWIN_UNLIMITED_SELECT is completely unportable, so I still need poll too.)
Meanwhile poll() just works on Linux and the BSDs, certified or not.
> It's not simply that certification costs money. It's that a lot of modern UNIX-like operating systems don't adhere to the UNIX spec. For example, the OpenBSD man pages specify the ways in which they diverge from POSIX and UNIX in the Standards section: https://man.openbsd.org/sh.1#STANDARDS, https://man.openbsd.org/awk.1#STANDARDS. Often times these are small deviations that might not matter to most people, but it means that they aren't UNIX.
Except it seems like macOS diverges, too, yet it is certified. I wonder in what other ways it diverges.
I came here to mention this too, but saw you'd beaten me to it. It's remarkable how bad non-free OSes are at maintaining basic infrastructure in their kernel and userspace compared to Linux and the BSDs isn't it? Basic bugs get neglected for decades in favour of rearranging the cosmetics - in this case, it's been more than 20 years since I first reported this to Apple's /dev/null service, and I'm sure I wasn't first and I'm only one of many.
And here I just want pthread_condattr_setclock() on Darwin. Is there some other interface to set monotonic expiries? I tried to port netconsd to Darwin and that's the sole hang up (well and recvmmsg() but that's trivial...)
There's pthread_cond_timedwait_relative_np. Looking at the implementation, pthread_cond_timedwait starts by converting the deadline to a relative timeout [1], but this is skipped if pthread_cond_timedwait_relative_np is used. Then, in the kernel, the timeout is converted back to an absolute deadline by either [2] or [3], but both of those are using Mach absolute time rather than the wall clock.
Mach absolute time is monotonic. It pauses while the system is asleep, which may or may not be what you want.
If need the timer to keep incrementing while the system is asleep, there's no way to do it directly with a timed wait, but you can use kevent with EVFILT_TIMER + NOTE_MACHTIME + NOTE_MACH_CONTINUOUS_TIME.
Can there ever be a "living" standard to replace posix ...? Something that captures the good things about posix but allows all the related ecosystems to move everyone forward toward a better base specification (together rather than just increasing the number of bsd vs gnu special cases that with current plans will just have to be lived with forever)
Browser standards managed to do this in a lot of ways despite far more complex standards, more complex variations in behavior, and much more rapid continue evolution ...
No, and that's a good thing. Just look at how quickly features are added to the living browser standards. Instead, we get a new POSIX version every couple of years that includes the special cases that actually work the same on BSDs and Linux after they're already implemented.
I really don't care about the certification. I care that I have a terminal with bash (still) and I have ssh. This gets me through probably more than 50% of the work that I do. And that I can install packages and have an as close experience as if I were to be on a Linux machine (development frameworks, containers).
I also like the fact that I have a very polished desktop manager running on great hardware with a lot of battery life.
I may have mentioned on occasion, here or there, about how ludicrous it is that there appears to be no well-defined standard that user space shall have sqlite3 and git and gzip.
So, for all intents and purposes, nothing that would be relevant in any reasonable end-user way in 2025. It’s all just: here’s defaults and here’s scripts to set up your environment and here’s a dozen things to run brew with. But no standard.
I wish jq would be in the posix standard. JSON is EVERYWHERE nowadays. A system that can’t parse it is incomplete. Not having a standard way to write a script that does it and works across *nixes is a mistake.
Yes, there is "UNIX V7" in 2013... which apparently only IBM's AIX supports. This is ironic because the whole idea of UNIX is to create a common platform for interoperability, but only one platform actually supports. I really wonder why Apple just doesn't put a couple of FTEs on it and upgrade to V7. I'm sure it wouldn't take much. But it sort of reminds me of Java and HTML where there were standards to allow for independent implementations, but have collapsed to single implementations.
I don't know about the Unix certification process itself, but the Single Unix Specification explicitly mentions case-insensitivity among non-conforming file system behaviors that are allowed as extensions (in 2.1.1 item 4, third-to-last bullet):
So a conforming OS has to make case-sensitive file systems available (which MacOS does: you can create case-sensitive HFS or APFS volumes). But I'm not sure if a conforming OS instance (i.e., running system) has to have any case-sensitive mount points, and either way, AFAIK there's no practical and race-free way for a conforming application to detect whether any particular mount point behaves case-sensitively.
So I believe that as far as the standard goes, a conforming application might run on a conformingly-extended OS where no portion of the the file namespace behaves case-sensitively. IOW, a conforming application cannot rely on case-sensitivity for file names.
There’s a specific “Unix mode” you have to turn on to be in the compliant state, it’s not the default. Presumably among other changes this puts APFS into case-sensitive mode.
Think of it: it's a Unix system. Literally. A unix system with the usability that your grandma can use. It supports both commercial and open source applications. The year of the linux desktop folks have been trying this for decades.
EDIT: already downvoted to negatives. The Linux folks really don't like to be reminded of that.
Grandma doesn't use certified UNIX. She uses a UNIX-like system. Only the engineers who ran the conformance test use certified UNIX since it required what I would consider heavy modification to the shipped OS configuration.
the claim is that the linux community has been trying for decades to create a desktop with the usability that your grandma can use it. the implied claim is that they failed, which is of course nonsense. linux has caught up with the usability of mac os a decade ago, if not before that.
Sounds good, but grandma will need to disable SIP, run a few commands using sudo, and add a root account from the recovery system. It's Unix in the same way Windows 2000 was POSIX compliant, you just need to reconfigure the default system.
The Linux desktop is widely used all across the world in the form of ChromeOS and, if you count touch screen devices, Android.
So for those who, like me, wonders why Apple keeps getting macOS Unix certified, it's to avoid a lawsuit. Apple misused the Unix trademark when they first launched MacOS, so to avoid legal trouble with The Open Group, Terry Lambert was put in charge of getting MacOS Unix compliant and certified: https://www.quora.com/What-goes-into-making-an-OS-to-be-Unix...
It's basically the only relevance the Unix trademark has these days. I can't imagine many companies choosing macOS because it's a real Unix, nor would anyone really opt out of z/OS, AIX og HPUX, if they where not certified.
> I can't imagine many companies choosing macOS because it's a real Unix, nor would anyone really opt out of z/OS, AIX og HPUX, if they where not certified.
While Unix compliancy isn't what's keeping me on macOS, the Unix tools it has under the hood still is. I've opted to use it over Linux because I still get everything that I need from a "Unix like" standpoint while having some serious enterprise level support and compatibility with work software that's often only available for windows or Mac.
If Apple stopped caring about being Unix compliant, I wouldn't be surprised to see the tools and infrastructure that make it Unix (and useful to me) slowly be removed. Then I'd stop using it.
I'd say that you care about it being UNIX-like, not UNIX®. You don't care that Linux isn't UNIX. You don't care that GNU versions of things like ed and awk are slightly off-spec.
In some ways, Apple's adherence to UNIX specifications probably makes macOS less useful for you. For example, I wish that grep on macOS was closer to GNU grep. When I look up commands online, I often find answers based on the GNU implementations. Those often work on macOS, but sometimes don't (or have subtly different behavior) because macOS is adhering to the UNIX specification rather than to what those utilities do on the vast majority of systems out there.
I don't think Apple would be removing UNIX-like tools from macOS even without certification. They know how valuable it is that most developers use their systems. Even Microsoft went so far as to implement the Windows Subsystem for Linux for developers. At this point, I think that UNIX certification makes macOS less compatible with the tools and help out there which generally targets Linux. Usually the differences are small, but they certainly can be meaningful.
42 replies →
While macOS only really gets Unix certified they design of the flavor of unix from FreeBSD. Homebrew is also the best port system and package manager I have ever used because it requires no thinking. I actually dislike using Linux because now I have to learn the replacement from ifconfig, the creation of launchd IMHO is way better than init.d and systemd, and the command line diskutil and other additions still feel like its more Unix like while Linux feels like its moving toward its own thing. Before I was using macOS I was using OpenBSD as my daily driver since high school. I still don't understand though why Ubuntu has the ability to break /boot because there isn't enough space to add another kernel to there...
4 replies →
The enterprise support is actually pretty bad. they offer some cool stuff like DEP but there's so many strings attached that for enterprise it's rarely actually possible in practice. Many things are boneheadedly designed.
For example, Apple federated accounts are a great idea. But, in your global directory the UPN and email must be the same. For us it's not, with good reason. We're not going to change our entire setup globally to suit Mac users that make up 0.5% of our systems. And there's never going to be more unless they become more accommodating. We even looked at JAMF but it's too much work to implement a whole separate management system. And the options in apple's configuration profiles are way too limited.
Another issue, every account that has already been created as a private Apple id on the corporate email must be manually resolved. Impossible with many tens of thousands of users.
AD binding while rudimentarily supported, causes so many issues. If someone's password expires there's no way to log in unless they're physically on the company network. The problem is that our security team demand we bind to AD. Not Azure AD. That's just reality in enterprise.
Having a managed local admin account is also a really big problem and there isn't really any tooling for that.
Maybe if you go all in on Apple like IBM did, then yeah you could adapt your environment to all their quirks. But it's a big blocker for small deployments. And really besides IBM nobody in enterprise did this. Apple meanwhile doesn't really care anyway. They only care about the customer market.
There is absolutely no way that happens. Apple cares about Unix compatibility for very good business reasons. Apple gets a lot of value from being even just Unix-ish. It's the same reasons Microsoft eventually knuckled under and made WSL -- the ubiquity of Unix conventions and tooling is just too cemented in the industry.
POSIX conformance is a cherry on top, and it helps get them certain sales that require a conforming Unix. But that's not the real value to the platform.
There was a time when OS X was new that Apple leaned pretty hard into the OS X is UNIX(tm) story. I was working in academia at the time where the Serious Work was being done on SunOS/Solaris and saw print ads aimed at scientists reminding them that OS X was a Real UNIX:
http://xahlee.info/UnixResource_dir/apple_unix_ad.html
It's not for everyone, but at some point I got tired of the FreeBSD layer being there but not cared for, and WSL+git for windows kinda provides a decent compromise in that regard.
The thing I hated the most was spending time building installation scripts or running images for the prod environment, and then go fight macos to replicate the same setup locally. Especially having parralel installs for the system and my user account was a PITA.
One would argue I could just dev on the docker image as well, but then being on a somewhat unixy OS doesn't matter much anymore.
WSL let's the Windows side live it's life (shell level tools can still be injected for convenience), and the linux side be genuinely Linux, not some ersatz, and duplicable to one's heart content. It's still less pain and better perfs than straight docker, and just extremely well integrated in general.
1 reply →
Considering the core utils have even been ported to Windows, I don’t really see what you would lose.
The Unix don’t really share much between each other apart from a small core.
16 replies →
> I still get everything that I need from a "Unix like" standpoint
Everything except a package manager!
3 replies →
There are some infuriating issues though, I have wiped out a couple files on OSX using sed with the -i option to replace a text within a file only to realised OSX would wipe it out instead ....
It is also relevant, that as proven on FOSDEM corridors full of Apple laptops, most folks only care about some sort of POSIX experience, and couldn't care less about Linux/BSD religion.
So that target audience gets a cool modern experience, without fighting with driver issues and such.
It is also the reason why Microsoft ended up bringing Project Astoria from the ashes into WSL.
UNIX has won, but not as Richard Stallman would have liked to.
AFAIK Stallman didn't care about UNIX, considering he is more of a LISP guy. GNU targeted UNIX because it was the popular OS at the time.
1 reply →
Nailed it.
I don't think Richard Stallman much cares. The first thing every programmer does on a Mac is install GNU software.
5 replies →
That explains why they got it UNIX certified back then, but couldn't they stop advertising macOS as UNIX and stop getting it certified? They even changed the name from Mac OS X to macOS since then.
That's my question too, why continue to bother? Apple doesn't even have any separate "Server" OS anymore. I can't find anything mentioning UNIX on any apple.com marketing pages.
I guess it's just, might as well keep it going, as an option for future marketing if ever needed. Maybe it helps the salespeople in some enterprise deals? I mean, if it doesn't really cost anything to keep it.
6 replies →
My (wholly unsupported) guess is that there are government or megacorp bids somewhere for Unix systems for employees, and this checks that box. The buyer could update their requirements, but why do that when you can just make your vendor jump through the hoop?
1 reply →
Not at Apple and don't have any knowledge here, but I'd imagine that the UNIX test suite, ever since it began passing, has been a useful set of additional regression tests even outside the certification context.
Does anyone want to be the person that removes regression tests from active use, only to be responsible when something breaks that would have been caught by that test? Far easier to just fix your code so the test passes.
(And for many years, OS X then macOS had a reputation for being rock-solid, capable of going much longer betwen restarts, going into BSOD much less frequently than Windows would. Having a set of third-party tests certainly didn't hurt this!)
I obviously don't know, but I could easily imagine that Apples legal team has flagged it as a potential risk and the cost of keep the certification up to date is minimal, compared to some imagined risk. Safer to pay the fee, and not having to worry about someone at Apple accidentally calling macOS a Unix system in public.
Also, Apple is a huge company, there's the question of who's going to make the call the not update a certification that's negligible within the scope of macOS development. Better to not be that person and just rubberstamp the invoice from The Open Group. If management disagree, they can make the call, but they won't because the cost is to small for them to deal with.
1 reply →
there’s no downside as far as i’m aware.
11 replies →
I think it’s a quiet but deliberate strategy to keep macOS the spiritual successor to NeXTSTEP. While many of Jobs principles are under pressure at current day Apple, his ghost lives on.
5 replies →
Literally the only reason that kept me on the platform until recently despite becoming increasingly hostile to developers...
Used to work with UNIX servers in the early 2000s but out of that sector for coming up on two decades -- are z/OS, AIX and HPUX (and other old big iron enterprisey UNIXen) still around? I would have thought that Linux had killed them all of by now. Excuse my ignorance!
They definitely are, not all though, HP-UX is on life support.
However Aix, Solaris (and Open Solaris derivatives), z/OS, IBM i (AS/400), ClearPath MCP, OS 2200, are still being updated and sold.
That list is not only UNIX systems.
3 replies →
You forgot solarius, well, so did oracle.
I don't think Solaris is Unix certified anymore. It was previously, but the current supported versions are not.
There’s an interesting story from the lead engineer to make OS X originally compliant:
> I was asked if I could lead a team to do #1. I said “Yes, under the condition that I could use the compliance project as a hammer to force other parts of the organization to make changes in their own code base, and that I could play it rather loose with commit rules regarding what it said in the bugs database for a given code change, and what the given code change actually did, in addition to what it said in the bugs database”.
…
> We were promised 1/10th of the $200 million, or $20 million in stock, on completion. $10 million to me, $5 million to Ed, and $5 million to Karen Crippes, who was looking for a home in Mac OS X development, I knew was an amazing engineer, and who could be roped into being technical liaison and periodically kicking off the tests and complaining to Ed and I about things not passing.
—-
Source: https://news.ycombinator.com/item?id=29984016
And if you read further in the comments, he never got the stock. The executive who promised it to him “took it for himself.”
And left his wife for an HR person, can't forget that. Lol sounds like a shitshow all around when it comes to execs but wouldn't be surprised.
Guess it shows that when it comes to compensation promises always get it in writing.
1 reply →
Damn. Don't ever agree to something like this without getting it in writing.
If they balk, it's precisely because they want to be able to be free to cheat you out of it once the work is done.
Where do you see that?
Sorry, I’m probably missing the obvious.
1 reply →
Further below:
> Also, the tech lead has to fix anything no one else fixes, or no one else can fix, because they are the DRI (Directly Responsible Individual).
How many tech lead/project manager can say that they are capable for this in these days? It feels like based on my observations that other skills are taking priority on management/lead side.
Thank you for sharing, what an interesting read.
Two general-purpose Linux distributions used to pay for Unix certification, although they don't do it anymore since hardly anyone is interested in it these days.
https://www.opengroup.org/openbrand/register/brand3617.htm
https://www.opengroup.org/openbrand/register/brand3622.htm
Save these links for the next time someone moans that Linux "is not a real Unix".
It's not just about paying for certification. You also have to replace a lot of things like ed, awk, grep, etc. with versions that are compatible with the UNIX specification. GNU utilities didn't target 100% UNIX compatibility and they have differences that mean that a command that works on UNIX might not work (or might not work the same) on a Linux distro using GNU utilities. glibc has slight differences from the spec too.
In order to get a Linux distro certified, you'd have to make changes which would make it less compatible with all the other Linux distros out there.
The reason why RedHat doesn't pay for UNIX certification is that their distros wouldn't be compliant. The reason why they don't make their distros compliant is that their customers would vastly prefer that RedHat use "standard Linux" tools than replace them with UNIX-compliant ones. Customers don't want a Linux distro that's subtly different/incompatible compared to what everyone expects in a Linux system. They'd rather it be not-UNIX.
Yes, you can modify a Linux distro to be UNIX. However, most Linux systems are not real UNIX - and you wouldn't want it to be real UNIX.
this talk of GNU being "standard" is toxic, as if anything that doesn't use it is weird or off spec.
the GNU userland might be common for user facing systems, but it's nowhere close to standard.
8 replies →
Citation needed.
I'm not sure what you mean by "Unix specification". But if you mean the international standard POSIX, yes, people care. Red Hat routinely participates in POSIX spec revision.
There are a very few deviations where you have to enable "POSIXLY_CORRECT". If that's what you mean, then you can turn that on. But in every area that matters, Linux distros implement the POSIX spec by default, and you can even turn on the POSIXLY_CORRECT mode to exactly follow it. They extend beyond it, but that is allowed and expected.
The people who build the tools in Linux distros care a lot. I know the implementors of dash and GNU make routinely refer to POSIX. The Linux distros don't have to as much with POSIX because that is generally a conpleted work and it's the maintainers of the tools who must address the updates to POSIX.
2 replies →
Reminded of the origin story of making Mac OS X Unix certified: https://www.quora.com/What-goes-into-making-an-OS-to-be-Unix...
Thanks for the link, that's a good read.
Related:
https://www.osnews.com/story/141633/apples-macos-unix-certif...
Identical changes were made for Tahoe's certification. This includes disabling SIP.
https://www.opengroup.org/csq/repository/noreferences=1&RID=...
So Tahoe is certified, but what ships with every Mac isn't certified UNIX. Not that being certified makes any difference what-so-ever.
There was a time when “open standards” were treated as the definition of Unix. At least that's what we aspired to. POSIX, X/Open, and others competed to be the standard that mattered. Formal standards were hoped to be a sounder, fairer basis for compatibility and interoperability than the earlier era of “Unix is whatever this release says it is" for some subset of 7th Edition, System III, System V, BSD, or one's favorite commercial derivative (SunOS/Solaris, HP/UX, AIX, Xenix, UnixWare, ...).
That window window of optimism—roughly mid-1980s to mid-1990s—closed fast. Open source projects and _de facto_ standards proved far more powerful in deciding where applications would run, where investments would be made, and which variants survived. Today, the real baseline isn’t POSIX in a binder or some Open Group brand certificate, but Linux + GNU + the APIs everyone codes to. In some ways we've regressed—or more charitably, we shifted back to a more pragmatic form of standardization.
the big difference is that the Linux + GNU standard is not controlled by a corporation whose only motive is profit, and that the reference implementation is Free Software that everyone can potentially contribute to.
i would not call that a regression. compare that to the browser standard which is largely controlled by google.
is not controlled by a corporation whose only motive is profit
Instead it's a bunch of corporations with very similar motives, including profit, so the end result isn't all that different. Just look at who the majority of committers work for.
The best “standards” just ratify what people are already doing.
OK, great.
Can I call poll(2) on a terminal device's file descriptor?
Requirement for certification: https://news.ycombinator.com/item?id=41822308
I had to work around this recently - it involved reimplementing my poll wrapper on top of select(), with the proprietary macOS extension that makes it support unlimited fds. (Of course -D_DARWIN_UNLIMITED_SELECT is completely unportable, so I still need poll too.)
Meanwhile poll() just works on Linux and the BSDs, certified or not.
And someone said:
> It's not simply that certification costs money. It's that a lot of modern UNIX-like operating systems don't adhere to the UNIX spec. For example, the OpenBSD man pages specify the ways in which they diverge from POSIX and UNIX in the Standards section: https://man.openbsd.org/sh.1#STANDARDS, https://man.openbsd.org/awk.1#STANDARDS. Often times these are small deviations that might not matter to most people, but it means that they aren't UNIX.
Except it seems like macOS diverges, too, yet it is certified. I wonder in what other ways it diverges.
Note: while the link above points to a newer standard, the quote has already been present in the one for UNIX 03 it seems: < https://pubs.opengroup.org/onlinepubs/009696699/functions/po...>
I don't think it's that surprising that the Open Group would cut corners certifying Unix compatibility.
Also, the weird `fsync` behaviors is not part of it?
I came here to mention this too, but saw you'd beaten me to it. It's remarkable how bad non-free OSes are at maintaining basic infrastructure in their kernel and userspace compared to Linux and the BSDs isn't it? Basic bugs get neglected for decades in favour of rearranging the cosmetics - in this case, it's been more than 20 years since I first reported this to Apple's /dev/null service, and I'm sure I wasn't first and I'm only one of many.
And here I just want pthread_condattr_setclock() on Darwin. Is there some other interface to set monotonic expiries? I tried to port netconsd to Darwin and that's the sole hang up (well and recvmmsg() but that's trivial...)
There's pthread_cond_timedwait_relative_np. Looking at the implementation, pthread_cond_timedwait starts by converting the deadline to a relative timeout [1], but this is skipped if pthread_cond_timedwait_relative_np is used. Then, in the kernel, the timeout is converted back to an absolute deadline by either [2] or [3], but both of those are using Mach absolute time rather than the wall clock.
Mach absolute time is monotonic. It pauses while the system is asleep, which may or may not be what you want.
If need the timer to keep incrementing while the system is asleep, there's no way to do it directly with a timed wait, but you can use kevent with EVFILT_TIMER + NOTE_MACHTIME + NOTE_MACH_CONTINUOUS_TIME.
[1] https://github.com/apple-oss-distributions/libpthread/blob/1...
[2] https://github.com/apple-oss-distributions/libpthread/blob/1...
[3] https://github.com/apple-oss-distributions/xnu/blob/e3723e1f...
Interesting, so it behaves more or less like pthread_cond_reltimedwait_np() in Solaris? I wonder which came first.
The context is a UDP network receiver which infers packet loss after a timeout:
https://github.com/facebook/netconsd/blob/main/threads.c#L15... https://github.com/facebook/netconsd/blob/main/worker.c#L526
Time while sleeping doesn't matter here.
Can there ever be a "living" standard to replace posix ...? Something that captures the good things about posix but allows all the related ecosystems to move everyone forward toward a better base specification (together rather than just increasing the number of bsd vs gnu special cases that with current plans will just have to be lived with forever)
Browser standards managed to do this in a lot of ways despite far more complex standards, more complex variations in behavior, and much more rapid continue evolution ...
No, and that's a good thing. Just look at how quickly features are added to the living browser standards. Instead, we get a new POSIX version every couple of years that includes the special cases that actually work the same on BSDs and Linux after they're already implemented.
I really don't care about the certification. I care that I have a terminal with bash (still) and I have ssh. This gets me through probably more than 50% of the work that I do. And that I can install packages and have an as close experience as if I were to be on a Linux machine (development frameworks, containers).
I also like the fact that I have a very polished desktop manager running on great hardware with a lot of battery life.
I wouldn't trade this for anything.
Has there been any work for something post Unix 03?
I may have mentioned on occasion, here or there, about how ludicrous it is that there appears to be no well-defined standard that user space shall have sqlite3 and git and gzip.
So, for all intents and purposes, nothing that would be relevant in any reasonable end-user way in 2025. It’s all just: here’s defaults and here’s scripts to set up your environment and here’s a dozen things to run brew with. But no standard.
I wish jq would be in the posix standard. JSON is EVERYWHERE nowadays. A system that can’t parse it is incomplete. Not having a standard way to write a script that does it and works across *nixes is a mistake.
C17 support is required in latest standard.
2 replies →
Yes, there is "UNIX V7" in 2013... which apparently only IBM's AIX supports. This is ironic because the whole idea of UNIX is to create a common platform for interoperability, but only one platform actually supports. I really wonder why Apple just doesn't put a couple of FTEs on it and upgrade to V7. I'm sure it wouldn't take much. But it sort of reminds me of Java and HTML where there were standards to allow for independent implementations, but have collapsed to single implementations.
https://en.wikipedia.org/wiki/Single_UNIX_Specification#Comp...
Solaris did support it, I guess it was more relevant when there were multiple vendors
[dead]
I recently learned that macOS has a (by default) case insensitive filesystem. How does this line up with the certification?
I don't know about the Unix certification process itself, but the Single Unix Specification explicitly mentions case-insensitivity among non-conforming file system behaviors that are allowed as extensions (in 2.1.1 item 4, third-to-last bullet):
https://pubs.opengroup.org/onlinepubs/9799919799/
So a conforming OS has to make case-sensitive file systems available (which MacOS does: you can create case-sensitive HFS or APFS volumes). But I'm not sure if a conforming OS instance (i.e., running system) has to have any case-sensitive mount points, and either way, AFAIK there's no practical and race-free way for a conforming application to detect whether any particular mount point behaves case-sensitively.
So I believe that as far as the standard goes, a conforming application might run on a conformingly-extended OS where no portion of the the file namespace behaves case-sensitively. IOW, a conforming application cannot rely on case-sensitivity for file names.
Apple calls out case-sensitivity as a change they made to pass certification. [0]
> 3. APFS file systems can be formatted as either case-sensitive or case-insensitive. Always format as case-sensitive for UNIX Conformant behavior.
[0] https://www.opengroup.org/csq/repository/noreferences=1&RID=...
If they standardize mktemp ( https://www.austingroupbugs.net/view.php?id=1616 ) then I think you could just
There’s a specific “Unix mode” you have to turn on to be in the compliant state, it’s not the default. Presumably among other changes this puts APFS into case-sensitive mode.
How much stuff breaks when you do that?
1 reply →
same for sequoia as well, discussed here: https://news.ycombinator.com/item?id=41698775
macOS 26? I'm still on 15.6 -_-
apparently sequential version numbers is not in the spec.
Think of it: it's a Unix system. Literally. A unix system with the usability that your grandma can use. It supports both commercial and open source applications. The year of the linux desktop folks have been trying this for decades.
EDIT: already downvoted to negatives. The Linux folks really don't like to be reminded of that.
Grandma doesn't use certified UNIX. She uses a UNIX-like system. Only the engineers who ran the conformance test use certified UNIX since it required what I would consider heavy modification to the shipped OS configuration.
> The year of the linux desktop folks have been trying this for decades.
Have they? I'm not aware that any of the desktop linux projects particularly care about POSIX or being a certified Unix™.
the claim is that the linux community has been trying for decades to create a desktop with the usability that your grandma can use it. the implied claim is that they failed, which is of course nonsense. linux has caught up with the usability of mac os a decade ago, if not before that.
Sounds good, but grandma will need to disable SIP, run a few commands using sudo, and add a root account from the recovery system. It's Unix in the same way Windows 2000 was POSIX compliant, you just need to reconfigure the default system.
The Linux desktop is widely used all across the world in the form of ChromeOS and, if you count touch screen devices, Android.
Windows 2000/NT POSIX support was the bare minimum, it isn't comparable.
It only got usable when SUA came to be.
my grandma used linux. in 1995!
Good thing to know when I go back to 1980s with a macbook!