On Thu, 12 Oct 2023 09:13:14 +1300, Peter Reutemann quoted:
'>From a report: ...'
The linked article <https://www.theregister.com/2023/10/11/microsoft_documents_installing_linux/> itself has some useful further references. For example, the original Windows NT architecture had provision for alternative userland “personalities”: Win32 itself was one of these, and there was also a POSIX subsystem, though it seemed more of a box-ticking exercise than anything really worth using <https://www.youtube.com/watch?v=BOeku3hDzrM>. However, the original version of Windows Services for Linux, WSL1, was *not* implemented as such a “personality”. Nor was its precursor, “Project Astoria”, which was intended as a way to run Android apps on Windows. Why not? And why did it have to be abandoned, in favour of WSL2, which brought an actual Linux kernel into Windows? Seems the reason is that the core Windows kernel simply isn’t versatile enough to emulate all the nuances of the Linux kernel API. This in spite of the fact that the API is fully documented (along with all the relevant error codes, unlike, say, Win32), and you even have source code you can refer to! This rather quaint item <https://jmmv.dev/2020/11/wsl-lost-potential.html> mourns the “lost opportunity” that might have been, had WSL1 progressed to the point where Windows and Linux processes could run side by side and be managed by the same tools, and where you could have used Linux admin tools to manage Windows services. But really, that was never possible.