“ I still hope to see a true "Windows Subsystem for Linux" by Microsoft or a windows becoming a linux distribution itself and dropping the NT kernel to legacy.
Windows is currently overloaded with features and does lack a package manager to only get what you need...”
People that comment things like this probably have their heart in the right place, but they do not understand just how aggressive Microsoft is about backwards compatibility.
The only way to get this compatibility in Linux would be to port those features all over to Linux and if that happened the entire planet would implode because everyone would say “I knew it! Embrace Extend Extinguish!” At the same time.
I agree. For years I supported some bespoke manufacturing software that was written in the 80s and abandoned in the late 90s. In the installer, there were checks to see what version of DOS was running. Shit ran just fine on XP through W10 and server 2016. We had to rig up some dummy COM ports, but beyond that, it just fuckin worked.
NT is a better consumer kernel that Linux. It can survive many driver crashes that Linux cannot. Why should Microsoft drop a better kernel for a worse one?
Is this a Wayland issue? This works fine for me on X. But yes, progress goes backwards in Linux. I had hope for the Linux desktop around 2005-2010, since then it only got worse.
If your $DISPLAY managed by Xorg server goes away your X apps will also crash. Wayland combines the server with the parts that draw your window decoration into the same process.
Under Windows everything including the GPU driver can crash. As long as it didn't take the kernel with it, causing a BSOD. Your applications can keep running.
I can restart window manager and compositor just fine in X. Also it is not generally true that X apps crash when the server goes away. This is a limitation of some client libraries, but I wrote X apps myself that could survive this (or even move their display to a new server). It is of course sad that popular client libraries never got this functionality under Linux, but this is a problem of having wrong development priorities.
Can you expand on this? I've used Windows 10 for 2-3 years when it came out and I remember BSODs being hell.
Now I only experienced something close to that when I set up multiseat on single PC with AMD and Nvidia GPUs and one of them decided to fall asleep. Or when I undervolt GPU too much.
Of course that depends on the component and the access level. RAM chip broken? Tough luck. A deep kernel driver accessing random memory like CrowdStrike; you'll still crash. One needs an almost microkernel-like separation for preventing such issues.
IBM marketed "OS/2 for Windows" which made it sound like a compatibility layer to make Windows behave like OS/2. In truth it was the OS/2 operating system with drivers and conversion tools that made it easier for people who were used to Windows.
Untrue. OS/2 for windows leveraged the user’s existing copy of windows for os/2’s compatibility function instead of relying on a bundled copy of windows, like the “full”
Os/2 version.
Os/2 basically ran a copy of windows (either the existing one or bundled one) to then execute windows programs side by side with os/2 (and DOS) software.
It was previously called the Windows Subsystem for Android before it pivoted. It had a spiritual predecessor called Windows Services for UNIX. I doubt the name had been chosen for the reasons you say, considering the history.
That said, to address the grandparent comment’s point, it probably should be read as “Windows Subsystem for Linux (Applications)”.
That's not what I say, that's what the former PM Lead of WSL said. To be fair, Windows Services for UNIX was just Unix services for Windows. Probably the same logic applied there back then: they couldn't name it with a leading trademark (Unix), so they went with what was available.
It was called Project Astoria previously. Microsoft releasing the Windows Subsystem for Android for Windows 11 is news to me. I thought that they had killed that in 2016.
Astoria and WSA are different things. Sort of. WSL and WSA both use the approach that was proven by Astoria. That approach was possible since the NT kernel was created, but no one within Microsoft had ever used that feature outside of tiny pieces of experimentation prior to Astoria. Dave Cutler built in subsystem support from the beginning, and the Windows NT kernel itself is a subsystem of the root kernel, if I am remembering a video from Dave Plummer correctly.
Anyway, Astoria was an internal product which management ultimately killed, and some of the technology behind it later became WSL and much later, WSA. WSA's inital supported OS was Windows 11.
Microsoft being Microsoft, they artificially handicapped WSA at the outset by limiting the Android apps it could run to the Amazon App Store, because that's obviously the most popular Android app store where most apps are published. [rolls eyes] I don't think sideloading was possible. [rolls eyes again]
I don't work for Microsoft and I never have; I learned all of this from watching Windows Weekly back when it was happening, and from a few videos by Dave Plummer on YouTube.
I believe that both Windows Services for UNIX (Interix) and OS/2 application support were NT subsystems too. I am under the impression that Windows Services for UNIX was the foundation for Astoria.
Source: https://x.com/richturn_ms/status/1245481405947076610?s=19