Microsoft is going to ship a full Linux kernel in Windows 10
theverge.com
theverge.com
Perhaps it is a bold claim, I haven’t seen it first hand. And it certainly has me hoping it goes much further than experimentation. I’d love to have a Windows environment on top of ext4 or zfs and the Linux kernel.
Don’t underestimate what new leadership who have no nostalgia towards Windows, but want to save money to pay themselves bonuses might do.
Just how far back are you talking? Because around 1988 most people were suspecting that apple would migrate to the 88000 at some point in the near future. And by the time it was obvious to apple users and (3rd party) developers that the 88000 was going nowhere fast, AIM had happened.
...doesn't quite have the same ring.
They are not putting in a Linux kernel for Windows. Huge difference.
He’s not saying it’s happening right now.
So technically WSL was not intended to be used for porting android apps.
But they did use Project Astoria as at starting ground to create WSL.
This is nothing more than a way for Microsoft to provide a *NIX userspace competitive with Apple, you know, so developers actually consider using their products instead of Apple's.
Edit:
We're actually in the midst of quite the opposite happening with the Linux kernel. After Android pivots to Fuchsia/Zircon, it'll only be a matter of time before Linux running on the bare-metal outside of datacenters will be exclusively the niche domain of Raspberry Pi style tinkerer devices and the tiny population of Linux enthusiasts running them on increasingly poorly supported PCs.
PC owners will be increasingly less compelled to bother even trying a Linux installation when their OEM Windows install already contains it as a compatibility shim.
This all translates into Linux's native support for modern desktops and laptops deteriorating for lack of users, I fear it's not going to be a growing market.
I am surprised that you could seriously suggest that the market for Linux on the desktops and laptops was even commercially viable (you say "I feal it's not growing to be a growing market"). When I started using Linux, it was basically people writing stuff for themself and releasing it to everyone. That actually seems like a fundamentally good goal. So what else could you want?
It is true that desktop Linux users are a niche case. Hell, I would not be surprised if that niche got smaller, even; but I have some pretty intense doubts that it’s dying or even going in that direction. In fact I would argue that the desktop platform as a whole has been on a decline, whereas desktop Linux has merely stagnated. The average user who is either gaming or just Facebooking on Windows PCs are finding increasingly less use out of their desktops. The gaming market on PC is healthy, but it’s cracking: the fragmentation of PC game stores and the increasing power and versatility of game consoles probably poses a threat in the long term. For all other use cases, mobile operating systems like iOS and Android are rapidly becoming not only usable, but preferable. Banking on my iPhone feels like the future, compared to the shoddy web UIs full of age old cruft.
So Microsoft has made wonderful strides for developers using Windows. No sarcasm here; I’ve been developing software on Windows since Visual C++6 and it’s been nothing but improvements. But I develop software for servers and browsers. The Microsoft ecosystem does not appeal to me. .NET Core is great but there’s still a lot of work to go and it’s reaching maturity in a world where we have amazing languages like Go and Rust. I have still dabbled a bit in .NET Core and I’m impressed, but they would’ve probably had more success if it had kicked off years earlier than it did.
So what’s going on with Linux? Frankly, quite a lot!
- Wayland is finally reaching maturity, slowly. For now, this is causing loads of problems, but as those issues clear up it’s looking like this might finally work out!
- Valve continues to improve Steam, with integrated Windows emulation. It’ll never be a substitute for real ports, but it’s still cool.
- The reproducibility and immutability bug has bit a few hands. New, advanced distributions like GuixSD and NixOS offer a pretty unique, extremely predictable experience (even if that predictable experience is not exactly compatible with the full range of existing software.) Fedora has rolled out OStree-based system upgrades too, which is awesome. This is one of those things where desktop Linux was really awful that is becoming a strength.
- There is apparently genuine demand for OEM Linux. We see not one, but two vendors that appear to be thriving: Purism, and System76. I want one of those darn Helios boxes just to look nice in the living room (and yes, I run Linux in the living room, and it’s been way less horrible than I would’ve imagined.)
(There’s more; like btrfs and other innovations that benefit desktop AND server use cases)
There’s more. And it goes without saying that there have also been negatives. Like graphics drivers from our green logoed friends. But by and large, I’ve had a great past decade with Linux. Aside from a small audio issue, my desktops have been free of any hardware compatibility issues throughout upgrades. Intel offers GPU passthrough that you can use in your KVM boxes for better virtualization. AMD IOMMU is reliable for nearly any Ryzen setup (got a GPU passthrough setup going great over here, and it didn’t even take a whole weekend to set up.)
Is Linux dying? Hell no. Is it the year of the Linux desktop? Also no. These claims both come up constantly and neither ever materialize. Desktop Linux has its niche and it probably will for the foreseeable future. Don’t extrapolate the ebb and flow too far.
There is still a plenty of reasons to avoid Windows, from privacy concerns (telemetry anyone?) to cost (sure, OEM is cheap but there are laptops that can be purchased OS-free for less. Also, there's virtually no such thing as OEM Windows in custom PC builds).
Why wouldn't they? Windows does billions of dollars of sales a year.
There is a huge and profitable class of applications that require the features of a PC. The Internet and wireless LAN/WAN and ARM architecture haven't changed that.
I suspect their lawyers would gently remind them that the Department of Justice and the European Commission were not best pleased when they pulled that crap in the past.
Edit: NT kernel & hypervisor.
In any case, the WSL experience on windows is still pretty sub-par. Until you can at least support the functionality offered by PodMan directly, including SystemD etc, it's not there yet. Even then, it's been kind of nasty to work with in general the handful of times I've tried.
Currently using Docker for Windows, and the msys bash that comes with Git, which though annoying for some things is still better experience wise than WSL is. Next desktop (Zen 2) will be on Linux.
Constantly people are bringing up that MS should switch to the Linux kernel. It makes zero sense.
You lose the ABI to run all the old Windows binaries out there when you can keep that AND run Linux binaries.
Also people seem to forget that the NT kernel runs on multiple platforms, on my XBox One with x86-64 CPU I can run old XBox 360 RISC CPU games. Windows 10 can run x86 binaries on ARM CPU:s. How is switching to a Linux kernel ever going to help with that?
Consequence of that argument is that Microsoft needs to support TWO flavors of Windows, one with the NT kernel & one with the Linux kernel. On top of that you get all the editions as a multiplier.
I did also like the Ubuntu UX a lot as well, not as much as Windows. Mac's is really usable but the MacOS-like shells are not quite there for the most part, at least from those I've tried.
Given Microsoft's history, I can't help but wonder if this wholehearted embrace of Linux is part of a classic Microsoft embrace-extend-extinguish strategy.[a]
Is the endgame a Windows-only userspace layered on top of the Linux kernel?
[a] https://en.wikipedia.org/wiki/Embrace,_extend,_and_extinguis...
If they could manage Linux desktops with group policy settings they wouldn’t need Windows at all.
Some apps still wouldn't work, like anything that calls NT APIs, but they could make that work with a compatibility layer until they're ported.
I agree that this seems a bit unlikely, though. But who knows where their priorities may lie in the future?
For example, MS could create a whole new graphical alternative to X and Wayland that works better than them both. Their challenge for capitalizing on that would be to create a software that works only somewhat well on X or Wayland, but works perfectly on MS Linux. I predict the Linux crowd will not use it. At best, they'll end up providing an alternative for the benefit of MS aficionados.
Android was an embracing of Linux, they extended it with Google Play Services, and now they're moving toward extinguishing it.
The classic EEE strategy aims to eliminate a competitor entirely. If that was Google's plan (which I doubt), it was doomed to fail from the start.
In contrast, Windows and macOS both have stable ABIs for third-party drivers, with the result that you can upgrade your OS to the latest version and install the same drivers and they’ll probably work. One notable exception was the Windows XP to Vista transition, which made a number of breaking changes that required new drivers… and just look how that turned out.
If Microsoft switched Windows to the Linux kernel, well… To start off, of course, there‘d be a one-time break of all existing drivers; popular devices already have equivalents in or for mainline Linux, but there’s a very long “long tail” of devices that don’t. But even more importantly, they’d be opting into the Android model of periodic breakage for perpetuity. In theory, they might be able to avoid that problem by forking the kernel and adding their own stable APIs on top. But that would require such deep changes that the resulting fork would either (a) have an “abstraction layer” big and complex enough to be a kernel of its own, (b) be unable to keep up with upstream refactorings and thus diverge permanently and rapidly from mainline, negating the benefit of adopting Linux in the first place… or, most likely, both of those. Never mind the amount of engineering effort required to build that, compared to maintaining the kernel they have. Or the licensing issues they’d encounter in an ecosystem that’s currently even more proprietary-oriented than Android. Or the control they’d be giving up over future development directions (although to be fair, you could have said the same about Edge).
So yes, Microsoft could embrace Linux as their new kernel. But they’d be crazy if they did.
Single-source licensing for enterprise-grade Linux and Windows, with a sweetheart deal. And they wouldn't have to do much, just take RHEL and offer certified builds and support. (Which is what Oracle does with their Unbreakable Linux.)
C:\Users\Ryan>wsl cowsay hello
_______
< hello >
-------
\ ^__^
\ (oo)\_______
(__)\ )\/\
||----w |
|| ||
C:\Users\Ryan>wsl
ryan@DESKTOP:/mnt/c/Users/Ryan$ notepad.exe test.js
ryan@DESKTOP:/mnt/c/Users/Ryan$ node.exe test.js
hello, world!But if you are ever curious, you can use Get-Verb[1] which will tell you what is going on.
[1] https://docs.microsoft.com/en-us/powershell/module/microsoft...
They can still be useful, but just mostly while navigating in a command window.
Also, unlike others here, I'm not a fan of a computing monoculture where everything is UNIX going forward.
Checkout the lightweight VMs (start around 38:00)
Those were always subpar with WSL.
Also, output of `which code`:
> zeusk@DESKTOP-F776PPQ:/mnt/c/Users/zeusk/Desktop$ which code
> /mnt/c/Users/zeusk/AppData/Local/Programs/Microsoft VS Code/bin/code