What you say is technically correct, which is the best kind of correct. However, fewer than 1% of Windows programmers ever talk to NTOSKRNL, and probably an order of magnitude fewer do it regularly.
Most of the time, you're talking to Win32/64 or high-level services based on DCOM or .NET, where the "well-conceived" and "theoretically well-founded" stuff doesn't turn up. You can go your whole career without knowing that there's a well-designed kernel under all that cruft.
I'd guess that less than half of Windows developers could say what the object manager does.
For more than half of Windows developers, if you told them there was an object manager, they would ask how you can turn off that service to improve performance.
This isn't true. Every Linux users encounters problems with drivers. By integrating them into the kernel we have an eco system where drivers are out of reach of many users and OEMs. Consider the difficultly in making a desktop scanner for linux. Additionally kernel changes can break a driver with little recourse on the OEMs side, you must simply bend to Linus's will. At the end of the day we suffer from Linux's driver architecture.
So true. Windows' services and drivers system may not be beautiful but it full on works most of the time and it allows even non pro Windows users to change basic stuff with far greater ease than Linux which is a real shame given Linux' potential for transparent usability (a potential it has had since the late 1990s but I have personally given up waiting for it to come true)
Most of the time, you're talking to Win32/64 or high-level services based on DCOM or .NET, where the "well-conceived" and "theoretically well-founded" stuff doesn't turn up. You can go your whole career without knowing that there's a well-designed kernel under all that cruft.
I'd guess that less than half of Windows developers could say what the object manager does.