My Thoughts on Desktop Operating System Software Architecture

Desktop computer operating system experiences have changed a lot lately. The real difference has been that all companies are providing less customer support and producing products with little or no input from their customer base. Both Apple and Microsoft find themselves with less of a desktop market share mostly due to the popularity of mobile devices.

Cloud based SaaS has proven to be where the profit is made. Microsoft has never had control over the hardware desktop architecture and so they have tried to be the one OS to run on a plethora of hardware. This has changed due to both the cost of maintaining such a broad range of compatibility and computing security.

As a result, Microsoft is adopting something closer to an Apple model of support where the supported hardware variance is minimal and where all software is distributed through a channel which they control just like Apple (Microsoft Store). This has even evolved to the point that CD/DVD support and even virtual mounting of ISO’s is going away.

The traditional WISE installation packages may even be approaching end of life as the Microsoft Store and the new Microsoft Package Manager (winget) are finally maturing.

Several years ago Apple left the PowerPC architecture behind and moved to an Intel based desktop. At that time they adopted OS internals based upon FreeBSD. Since the new architecture was not binary compatible, Apple users were forced to new binaries. The same change of binaries will occur as Apple migrates from Intel to Arm.

The essential desktop experience of MacOS has not changed as the internals have. People associate the Apple experience with the interface. The same is true of Windows. People associate the Windows experience with the Windows desktop appearance and interface.

For that reason, I have always hoped that the 1990’s internals of Windows NT that is the essential underlying infrastructure of even Windows 11 today would be scrapped in favor of a new Windows based upon a Linux kernel.

This would finally provide the stability and security that Windows has lacked for so long. In addition, key essential capabilities like X-11/Wayland could become a core part of the OS. In addition, there would be no need to constantly enforce new requirements for the underlying hardware since Linux runs on a much broader hardware range.

Probably the biggest advantage is that a Windows with a Linux kernel could embrace OS containers with a shared host kernel (something like LXD or just LXD). This would expand the utility of hardware running Windows at the desktop and more efficiently utilize the hardware as well as isolate applications into their own partitions.

The best case utility achieved in all of this is that Windows could become an immutable operating system that would be literally immune from viruses and poorly executed updates and software installations performed by users. The subject of immutable OS is extensive, but even Ubuntu is planning an immutable release option in 2024.

I realize that this solution would require that everyone throw out all of their existing Windows binaries. However, a lot of Windows software development is done in Linux and once this move is accomplished, future maintenance costs would be radically reduced. In addition, a change to the Linux kernel for Windows would open access to the full world of Open Source.

Honestly, Microsoft’s cash cow is SaaS and not Windows desktop. Isn’t it time to give Windows users a much better experience?

1 Like