Microsoft To Add Support for Linux GUI Apps on Windows 10 Later This Year (zdnet.com) 124
Microsoft is making the promised support for Linux graphical user interface (GUI) apps on Windows 10 available to customers as of the next Windows 10 release, officials said on May 25. Microsoft officials made the announcement on Day 1 of its virtual Build 2021 developers conference. From a report: During his Day 1 keynote, CEO Satya Nadella basically acknowledged there will be another event "soon" about the next Windows. He said: ""And soon we will share one of the most significant updates of Windows of the past decade." He said he has been self-hosting it over the past several months and called it "the next generation of Windows."
Microsoft released a preview of Linux GUI apps on the Windows Subsystem for Linux (WSL) in April, 2021. This capability is meant to allow developers to run their preferred Linux tools, utilities and apps directly on Windows 10. With GUI app support, users can now run GUI apps for testing, development and daily use without having to set up a virtual machine.
Microsoft released a preview of Linux GUI apps on the Windows Subsystem for Linux (WSL) in April, 2021. This capability is meant to allow developers to run their preferred Linux tools, utilities and apps directly on Windows 10. With GUI app support, users can now run GUI apps for testing, development and daily use without having to set up a virtual machine.
In other words (Score:2)
They'll port Xorg to Windows.
Re: (Score:2)
I seem to recall this being a big deal on MacOS X many years ago. It kind of lost its importance as the software library for macOS improved.
Why wouldn't people write their stuff in QT or wxWidgets and build GUIs for both Linux and Windows if that is their intended target market? I'm pretty sure Visual Studio supports CMake and Linux builds now, should make the transition between the platforms pretty painless for developers.
For those of you who want a real Xorg on Windows. MobaXterm (paid) includes it among
Re: (Score:2)
>CMake
Then it'll never work.
I've never met a project built with CMake that wasn't a pain in the arse and full of incomprehensible write-only configuration.
Re: (Score:2)
It used to be horrible and impossible to debug. Now it's only horribly inconvenient. The main advantage is that it's a meta-system and can transform your build configuration to GNU Makefiles, MS NMake, Ninja, and a few others. A brought range of platform support and it's quite a bit faster at detecting your system than old GNU autoconf/automake. The command-line options and UI for CMake is complete garbage though.
thinko (Score:2)
s/brought/broad/
Re: (Score:2)
Sounds like imake reincarnated with added bells and whistles if you ask me.
Re: (Score:2)
imake was easier and had nicer syntax than CMake (or autotools). But it still faced a lot of the scalability problems inherent in directed acyclic graphs as used in Make (top to bottom). But Xorg abandoned imake and I wouldn't recommend using it at this point.
GYP or Meson might be an alternative if CMake is not to your tastes. Off the beaten path but Premake [github.com] is cleaner looking than CMake and less likely to face the Python version struggles that Meson users have.
For a small project and full scratch (release
Re: (Score:2)
I tend to either make a vanilla Makefile or if I'm really lazy, build.sh which executes gcc a couple of times.
Then again, I mostly write command line programs that process buggerbytes of data for which a GUI or network stuff or fifteen layers of web abstraction would not be of much use (thankfully).
If I can't factor a problem down to something addressable that way, I'm doing it wrong. I'm glad I don't work in a huge design team.
Like this : https://github.com/dj-on-githu... [github.com]
or this : https://github.com/dj-on- [github.com]
Re: (Score:2)
Doesn't even check for header dependencies. And ignores differences between *nix platforms so no -lsocket for SunOS (main reason we've had GNU autoconf for the past 20-30 years). And no properly built shared libraries either, as ugly as libtool seems at first it's easier than figuring out how to correctly build shared libs on every platform.
What you've done is perfectly reasonable for trivial projects that don't need to be shared with others, packaged for a distribution, or ran through automated tests of a
Re: (Score:2)
It depends on your definition of trivial.
Entropy analysis is anything but trivial and is very serious when it comes to cryptographic security.
The algorithms within are the value in there. The wrapper is just there to call the algorithm.
I do have regression tests for djent and quickrdrand because they're used in silicon production testing, but I haven't shared those because they're rather tied to some megacorp internal infrastructure.
Re: (Score:2)
Heck, I was running X servers with Cygwin about 15 years ago. It's not that innovative.
Re: In other words (Score:2)
Hummingbird Xceed 25 years ago.
Re: (Score:2)
As I recall I had both Xorg and an X server running on the Windows box. It wasn't the speediest thing in the world, but was fun to play xminesweeper on Windows. I actually used Cygwin for more useful things, like a Radius server.
Re: (Score:2)
killer app was GNU plot.
Re: (Score:2)
This is about Wayland on Windows, not Xorg. Somehow they are implementing Wayland in Windows proper and then connecting the Linux applications running in the Linux VM (that's what WSL2 is, a VM) through to this native Windows Wayland server.
Re:In other words (Score:5, Insightful)
Some notes:
https://github.com/microsoft/w... [github.com]
It's basically a Wayland implementation that is exported via RDP protocol.
Re: (Score:2)
Right! That makes the most sense. And it being open source, hopefully this of thing can be modified to eventually bring back application-level remoting to Wayland-based systems. I realize transparent X forwarding still works for GTK and Qt apps even when in a Wayland environment using Xwayland, but it's only a matter of time before that path bit rots in the toolkits. I'm not interested in remoting a whole desktop; I used application-level remoting regularly.
Re: (Score:2)
No one cares. The Wayland team literally published working RDP code to demonstrate to the detractors that remote per application hosting was possible, and said "There, now leave us alone and if you want this feature go implement it". And nothing happened.
Somehow despite the uproar, no one actually cares enough to do anything about it, which is kind of like saying the Open Source community has literally voted with their wal... code contributions.
Re:In other words (Score:4, Informative)
If they've locked non-x86, then shame on them, and I hope they suffer failure on that platform because of it...
I'm kidding- they did, and they did.
Secure Boot is a well-thought out PKI that (at least on x86 hardware) is completely user-manageable.
There is no reason it can't be implemented on non-x86 UEFI boot chains as well.
Note: Almost all ARM SOCs (worth mentioning in the personal computing space) require signed bootloaders.
To Microsoft's credit, x86- their stomping ground- is the only place where you have actual freedom in that regard, because unlike Random QC SOC, you can actually replace the platform key on an x86. That's literally baked into the specification.
Re: (Score:2)
Re: (Score:2)
They'll port Xorg to Windows.
At least someone will be maintaining the project. Maybe they can kill a few bugs in the process.
Re: (Score:2)
Re: (Score:2)
Whenever you're playing with a third-party windows X server, you're invariably playing with really fucking weird quirks. I've used a lot of them, and never found one that was "really good"
Maybe this way, all of my KDE titlebar graphics won't be doubled in size with half size icons.
Re: (Score:2)
Think you missed the obvious joke of my comment: A reflection of the state of X.org, and to be honest, it's no surprise that MS would chose to implement Wayland instead.
Re: (Score:2)
Well you can run PowerShell on Linux now, so I'm sure if ransomware isn't already there, it will be soon...
Re: (Score:2)
Well I installed it a couple of years ago. I was hoping that I could use PowerShell from Linux to easily talk to Active Directory servers. The existing options on Linux (aka various Perl and Python modules) are clunky to say the least.
Imagine my disappointment when I tried running "import-module ActiveDirectory" only to be told there was no module to import and all my hopes and dreams where flushed down the toilet again. Thanks a bunch Microsoft.
Re: (Score:3)
You can use it for that. The PowerShell ActiveDirectory module isn't available, but all the .NET ActiveDirectory functionality is there in Powershell on Linux, so you can just do something like
"[ System . DirectoryServices . ActiveDirectory . Domain ] : : GetComputerDomain ( ) . Name" (spaces added to bypass slashdot "ascii art" filter) instead of "Get-ADDomain". Is it more verbose? Yes. Does it still work? Also yes, and you could always wrap them in equivalent functions/cmdlets if you wanted to use them re
who is this ... (Score:2)
... linux guy they're talking about?
Re: (Score:2)
Rust (Score:2)
I'm personally waiting for the Windows kernel to be rewritten to a micro-kernel architecture, if this is even possible.
Re:Rust (Score:4, Insightful)
I'm waiting for Microsoft to fork Wine and then just abandon the Windows kernel entirely.
Re:Rust (Score:4, Insightful)
What is happening is Microsoft embracing (if not extending) Desktop Linux to enhance Windows. In Windows, you are now getting all the usual ecosystem of Windows apps (that Desktop Linux doesn't have), plus those Desktop Linux apps and shells as an extra.
Re: (Score:2)
Re: (Score:2)
It's no coincidence that Microsoft ported SQL Server to Linux. Having thousands of SQL Server instances running on Windows is not something to look forward to as a Windows administrator.
I don't know how far Microsoft will take this (it's still an easy cash cow) but as far as the server Microso
Re: (Score:2)
Re: (Score:2)
Re: Rust (Score:2)
Re: (Score:2)
Why would they do that?
Because the code is easier to read.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Hope that will cure Linux's GUI/UX problem... (Score:2)
I sincerely hope that this development once effected, will tame or cure Linux's inherent Ux/GUI problem of sucking bigtime in the beauty category.
Re:Hope that will cure Linux's GUI/UX problem... (Score:5, Interesting)
Highly doubtful. Windows, if anything, promotes the kind of UI mess that Linux is famous for. MS started it all years ago when they started using a custom UI toolkit for MS Office. Every Office release since uses a slightly different widget set. Now we have a mess of apps that look and act differently, some skewmorphic, some not. It's particularly horrible with anti-virus and anti-malware tools. In short, only MacOS strives for any sort of UI visual consistency, and even then sometimes not.
Personally I don't find the current state of my Linux desktop to be ugly or unusable. For me quite the opposite. I of don't really care about mythical new users, though. I care most that the applications and desktop environment works for me and my use cases, which it does nicely. I get irritated when developers would rather alienate a user like me who has used, say Gnome for decades, to try to attract new users, which I'm not convinced actually exist.
Re: (Score:2)
I sincerely hope that this development once effected, will tame or
cure Linux's inherent Ux/GUI problem of sucking bigtime in the beauty
category.
If MS's "Linux GUI" support works like Windows Remote Desktop, then no. It
will be displaying the KDE/Gnome/whatever desktop inside a window on the
Windows desktop.
On the other hand, if they mapped the X-Windows protocol to WinAPI calls
AND, some how forward and map KDE/Gnome/whatever desktop management requests,
then, maybe, yes.
Re: (Score:2)
If MS's "Linux GUI" support works like Windows Remote Desktop, then no. It will be displaying the KDE/Gnome/whatever desktop inside a window on the Windows desktop.
You are a bit out of date. RDP has supported RemoteApp [microsoft.com] since 2008.
This allows you to connect to individual applications that appear on your remote system just like it is a local application. It has its own window and can have its own shortcut on your start menu. The remote program has access to your local resources like your locally installed printers.
Re: (Score:2)
Yes but as far as I know no Wayland implementation supports this idea of per-app remoting yet, so WSL2 won't either. Which is a real shame in general.
Re: (Score:2)
I replied to my own comment [slashdot.org] (because you can't edit on slashdot) with the video [youtube.com] showing that they do indeed show the applications as separate windows with their own start menu icons and taskbar entries.
Re: (Score:2)
Doesn't every Wayland app draw itself and its window decorations? I'm not sure why it would need a desktop.
I would assume you load all the libraries, but not the actual desktop. Windows handles the window placement, and the apps handle the way they're drawn. The Linux apps can drag between each other, but likely not to/from Windows apps.
I could be wrong here, but that's how I picture it working. I also suspect you'll be able to load a desktop into a window and do it that way if for some reason you prefer.
Re: (Score:2)
Further to my own comment, here is a video demonstrating WSLg [youtube.com]. The applications appear as separate windows, with their own entries on the taskbar.
The user interface for the applications is pure Linux, so no it doesn't fix the disparate user interfaces. To do anything else would introduce huge compatibility problems.
Re: (Score:2)
Linux's inherent Ux/GUI problem of sucking bigtime in the beauty category.
Which UI doesn't? Apple hasn't looked good ever since they became a company of product managers instead of artists.
Finally linux on desktop is here (Score:2)
Linux apps subject to Microsoft telemetry ... (Score:5, Insightful)
so Uncle Sam will get to see what a lot more people are doing.
Re: (Score:3, Informative)
so Uncle Sam will get to see what a lot more people are doing.
It's for this reason why I won't run Linux apps on Windows natively. I will continue to use my Linux Mint VirtualBox install. I have Mint running on VirtualBox in seamless mode. Mint's task bar is on the top, so I end up with a second task bar at the top of my screen with everything Linux. It's sitting on a 64GB virtual hard drive and is encrypted with LUKS inside it.
Wrapping Linux apps inside an emulator doesn't, per se, add any protection from Windows itself. Sure, the encrypted filesystem makes the
If they could get Wayland to work on WSL... (Score:4, Interesting)
Progress (Score:2)
Does a X connection on localhost not work? (Score:2)
Please explain to me...
Back in school in the late '90s, I ran apps on a Unix box from a rootless X server on Windows NT.
Is that not possible with WSL today for some reason? What is different?
Re: (Score:2)
It's possible if you're willing to go through the motions of setting up an X server and doing all the things necessary to make it work.
As far as I can tell, the difference is that this works out of the box.
Re:Does a X connection on localhost not work? (Score:4, Informative)
It's possible if you're willing to go through the motions of setting up an X server and doing all the things necessary to make it work.
Like: Downloading Xming, installing it, starting it, setting up a standard configuration (just click through the wizard), export the display on your Linux terminal and run the Linux GUI program?
Yeah, that stuff takes all of 60 seconds. (Speaking from experience.)
Re: (Score:2)
Oh, I know. I've done it before, too.
Like most things in life, it's easy once you've learned how to do it.
Re: (Score:2)
Assuming you mean WSL2, you'd better add ten minutes for getting your clipboard to work. If you don't already know how, call it a couple hours: search for it, try it, debug it, satisfy yourself that you haven't just let everyone on your network read your clipboard.
Re: (Score:2)
This also handles audio input and output. It has hardware acceleration for mesa. It has hardware acceleration that is gpu agnostic for gpgpu. Part of what is coming with this is dx12 directcompute for things like tensorflow both inside and outside wsl.
Re: (Score:2)
Can you even run a daemon reliably under WSL 2? I had nothing but problems running a LAMP stack on WSL2, with Mysql and Apache only starting up intermittently. Went out and bought a Raspberry Pi for the server. Worked a lot more reliably. So far I just don't get the point of WSL.
Re: (Score:2)
I don't think that was ever a use case for WSL. It's mostly for supporting Linux-based local development environments, which don't usually require long-running server processes. I guess they figure that you need a long-running server process, that's what HyperV (or your VM system of choice) is for.
Re: (Score:2)
At the end of the day I prefer the Pi. It's configured as an actual server, no games to play, it fires up, is fast enough for development purposes.
Just what I have always wanted! (Score:2)
Now in addition to the massive bloat of Windows install, I can enjoy the disk choking cruft of a full desktop stack linux install at the same time!
Re: (Score:3)
They should call it "Beer" (Score:2)
Re: (Score:2)
EEE (Score:4, Insightful)
Re: (Score:2)
That always comes to mind, but they could make Windows into the best Linux in the world, and it still would cost money. I suppose this could bite into the non-free Linux market someday though.
Re: (Score:3)
Ah yes, that old FUD. The problem is that this isn't all Microsoft code - they are using distros made by other people to do the heavy lifting. Here is Ubuntu [microsoft.com], which you can see was made by Canonical Group Limited. Here is the page on Ubuntu.com about it [ubuntu.com]. If you search on the Microsoft Store for "linux" then you will find the other distros available like Debian, SUSE, Kali etc.
It is all open source, so here is the source code for the kernel [github.com]. Here is the source code for the GUI parts [github.com].
So if they do extend anyt
Re: (Score:2)
Yes, it is FUD. Microsoft did coin the term - LAST CENTURY.
IEâ(TM)s JavaScript was not because of EEE, but was just a product of an extremely simple imprecise language standard in a time when all browsers raced to add new features that were nit part of any standard. For example, Netscape Navigator added layers while IE added frames. In that case,IEâ(TM)s addition was the one to be eventually codified into the language.
Borland was not killed with EEE. In fact, Borland had their own non-standard add
Re: (Score:2, Insightful)
And the stupid EEE comment is at it again. I've said it before, and I'll say it again. Those people who think what MS is doing here is EEE don't have a clue how EEE works, don't have a clue the market conditions needed for it to happen, don't have a clue about MS leadership and don't have a clue about how MS makes money. In summary basically thinking this is EEE is a declaration of not having a clue about anything.
Re: (Score:2)
OS/2 -- a better Windows than Windows!
Exactly the same thing. Windows somehow survived...
Reintroduction of Aero (Score:3)
Wouldn't it be easier to build a Windows userland? (Score:2)
I've been using Windows on and off for the last couple years, and I consistently benchmark it between 30 - 40% slower then Linux, for the same tasks, and I know other developers who agree with that figure, but I'm not sure how well tracked it is in their workflow.
The WSL is just a sad attempt for
Re: (Score:2)
I don't think the WSL has any effect on POSIX compliance and government contracts. I wouldn't be surprised if some government requirements preclude running the WSL at all.
No this is fundamentally about developers who are comfortable with Windows and its tools (Visual Studio) but need or want to target Linux. I think it's part of their Azure cloud strategy as well. Develop on Windows, test in the WSL2, deploy to Azure. MS knows as well as you and I that Windows itself has no place in the cloud. This k
Re: (Score:2)
Re: (Score:2)
> At this point why not build a Windows User land that sits on top of Linux, so if you want to use Windows apps you can, without have to loose the stability, security, usability and performance benefits of Linux?
s/loose/lose/
Re: (Score:2)
Re: Wouldn't it be easier to build a Windows userl (Score:2)
Most of that is already done by WINE.
All they would have to do, is contribute to the lowest level of Windows libraries in WINE, and release the rest to run on top of that, as is already possible and done.
In other words: Complete WINE.
Re: (Score:2)
Re: (Score:2)
For a C++ simulator I use I don't see any performance difference between Windows 10 and Linux using optimized binaries for both. With previous versions of windows the linux version was usually about 10% faster or so especially if more than 4 cores are used but I don't see any of that anymore with Windows 10.
I like WSL and WSL2. I find that Visual Studio is a better IDE for C++ and Python and having it support Windows and Linux transparently is really nice.
See, Nadella is mostly like you and me (Score:2)
The main difference being - the basement he lives in is worth several million dollars.
Keeping people off of the linux desktop (Score:5, Insightful)
Re: Keeping people off of the linux desktop (Score:2)
Re: (Score:2)
Or rather: make it harder for us to say to IT we need a real Linux. My experience is, that many companies have a Microsoft only IT department. Linux is barely accepted. They will do anything to get you to use their standard Windows solution. Now they can just point at WSL2 and say you can just use that.
And for good reason - complexity costs, and supporting more complexity costs more money. At some point the complexity costs more than the productivity gained.
Next think about where the costs are born (IT) and productivity gained (not-IT) and you'll understand more why standardising on something mediocre that pretty much works for everyone is a lot cheaper than best of breed for every solution.
Where I am *nix goes on a lot of servers, but for workstations only our managed Windows PCs are allowed to touch th
Embrace and Extend? (Score:5, Interesting)
I just have to wonder... Is this Linux being accepted as a valid desktop operating system, or is this an attempt to prevent it from being the next big Windows competitor?
I'm not saying it is either, just that this does feel like the old "Embrace and Extend".
Re: (Score:2)
Re: (Score:2)
I'm not saying it is either, just that this does feel like the old "Embrace and Extend".
And if you think it is it shows you don't know how EEE works, the market conditions needed for it to happen, MS's strategy, nor how MS is actually making money.
Pop quiz: Why would MS EEE something that makes a profit for them and costs them essentially nothing? And if you want to look at how much profit I suggest scrolling down to the Azure section in their last financial reports.
Not only is it not possible to apply EEE to Linux, it also makes zero sense for them to do so.
Re: (Score:2)
You're likely correct, but the one case I could see where "Embrace, Extend, Extinguish" for Linux does makes sense is if Microsoft was trying to protect their Windows desktop marketshare. They did make about $22 Billion on windows sales last year.
Without *setting up* a virtual machine? (Score:3)
Re: (Score:2)
There's a big difference to spinning up a VM and just clicking a button on a windows store and then clicking an icon. Just because it uses virtualisation underneath doesn't mean anyone is "setting up" a VM.
If you're a Linux user (Score:2)
I tried WSL2 (Score:3)
Meanwhile, Cygwin has it (Score:2)
And has had it for a long time.
Year of the Linux Desktop (Score:2)
Its the year of the Linux Desktop, on Windows.
Why ? (Score:2)
I don't get the running linux under windows . if I develop in linux I want a pure linux experience. I work on some of the most locked down desktops on the planet. However we have Vmware running Linux VM's for development. I find that far preferable than the reverse wine crap of Linux under windows. Embrace and extend anyone ? If you are working on Windows in a SCIF type environment as I do. Then precious little works correctly anyway so why try to run Linux under that mess. At home I dual boo
This is a good thing for slow ass X11 forwarding (Score:2)
Windows Subsystem for Linux *IS* a virtual machine (Score:2)
Key to this technology is the fact that Windows Subsystem for Linux *IS* a virtual machine. Not sure where the poster gets the idea that it isn't a VM.