We were chatting at work recently and someone suggested that perhaps Microsoft should replace the Windows kernel with Linux.
That is, keep the "wrap" ... the parts seen by users ... but scrap the underlying kernel.
- Cost savings for Microsoft ... no more kernel to support
- Performance benefit for users ... the Linux kernel can run circles around Windows
It makes sense and it has me wondering ...
It's an interesting idea.
Not sure I see it happening anytime soon and not sure I would want to see it happen.
Wouldn't replacing the Windows NT kernel with a Linux one be a compatibility nightmare?
I think you're right about this. Drivers would have to be written.There are Windows drivers for most any consumer or business device. Coverage is not as complete on Linux.
Still ... if Microsoft isn't profiting adequately from Windows they'd be prudent to find alternative sources of profit.
Remember when mainframes disappeared ... and IBM had to re-invent itself?
Remember when we bought a servers when we needed them ... until AWS showed a better way?
Such re-invention is eventually a do-or-die situation for many firms.
Microsoft is already making headway in the cloud.
I wonder what they'll do in the desktop and mobile space?
If they ever did, I wonder how much it would vastly differ from the original Linux kernel.
"Will the Linux kernel replace the Windows kernel?"
I can't see that happening. This is one of those ideas that might seem attractive at first but fail badly confronted with the harsh facts of reality. Everything is simple if you don't consider the details.
History has shown that even the most insignificant changes in an OS can and do have compatibility issues. Windows components and third party drivers rely heavily on the current kernel. I can't see how changing the kernel in such a massive way could avoid serious compatibility issues. The compatibility layer would be a nightmare to implement and debug and probably cause more problems that it was intended to solve. And the overhead of the compatibility layer would likely erase any performance benefits.
My perspective is that of an amateur programming Windows for 15+ years. I know about compatibility problems.
That compatibility layer should not be a problem.
The reverse has already been done ... Linux (Unix/POSIX) functionality running on Windows has been around for nearly three decades.
There's no reason to expect that the reverse can't be done.
In any case ... Windows already has a Linux subsystem, so Linux programs can be run there.
The proposed notion is just a reversal of that ... Windows programs running on top of a Linux kernel.
If it makes financial sense, Microsoft could do it. And I would welcome it because Linux is a better kernel. It's on the UI side where Windows excels.
Hi there
as a P.O.C (Proof of concept) it really wouldn't make a lot of difference to the end users but could as the OP says (repp'ed you for interesting topic !!) make a lot of money for Ms - it would "Look, Feel, Taste, and Smell like Windows !!!. I think there's a saying - "If it walks like a Duck, looks like a Duck, squawks like a Duck etc .. it IS a Duck !!!!.
-- Since the GUI is configurable in Linux (you can choose almost an infinite variety of GUI's for a Linux desktop) you could present an almost identical "Windows" Interface to what the users currently get in Windows.
Hardware abstraction should make plugging in drivers probably do-able --that's where most of the effort would need to be spent -- currently Windows supports a load more hardware devices than Linux (Windows drivers often work straight out of the box -- Linux though can always with some effort be made to support those devices too).
Also to run windows .exe files Ms could take a leaf out of WINE where quite a few (not all of course) Windows.EXE files run under WINE on a Linux desktop.
This isn't a "Weekend project" for a few hobbyists" -- would take serious money and development -- but it would net Ms a whole slew of money -- companies like VMWare wouldn't like it as why would people who only use VM's to run say both Linux and Windows need it anymore if they were only running a "Window-ized" version of Linux. --There are loads of other reasons to run VM's but a lot simply do it to run a Linux Distro and Windows on the same machine - whichever OS they use as a Host.
Backward compatibilty would presumably be a problem with previous issues of Windows which depending on the importance of that scenario could though "scupper" the project. However even there Ms could possibly find some type of Work around as they did with W7 and "XP MODE" so people switching to the new OS could still run old stuff seamlessly without needing VM's etc.
Things like ZORIN OS currently make the Linux desktop more Windows looking than Windows !!!!--certainly for W7 users -- so fiddling around with the GUI IMO would be the best way that Ms should approach this if they wanted to go down this route.
@margrave55
have a look at this on running Windows executables on a Linux kernel.
WineHQ - Run Windows applications on Linux, BSD, Solaris and macOS
Not everything can be run but it works for a lot of things --with resources of Ms behind this project should be a "Shoe in" to get it working 100% for all Windows apps.
Thanks for providing a very (IMO) interesting topic !!!!!
Cheers
jimbo
Quote