I'm looking forward to MS just dropping their OS entirely and using linux as their back-end. Basically, Microsoft might be better off creating their own linux distribution running a "microsoft windows" windows manager.
It is technically doable, but I'm not sure the Windows NT kernel is worth discarding entirely. It has some fancy tech in there and if need be it can emulate POSIX through "personalities" aka subsystems (http://en.m.wikipedia.org/wiki/Windows_Services_for_UNIX). Subsystems actually make it more flexible than Linux.
But in general the NT Kernel is pretty good and Windows Server is already moving to be more "Linux-like" (e.g. Core mode, all the GUIs now generate Powershell commands, etc). So it might be a while before Microsoft actually dropped NT and moved to Linux, least of all because it would require a lot of code changes in other products.
There are many thousands of people who have done the same thing, and this gives us a "web of trust". And if you don't want to trust anyone, you can always compile it yourself.
That's the difference: I have the freedom to learn and study anything and everything. It might not be likely, and it might not be convenient, but it is possible.
https://wiki.debian.org/ReproducibleBuilds
(I gave a talk about this at CCC. The status quo is a reason why free software doesn't have some of the transparency benefits over proprietary software we might expect, if you fear some developer is attacking you or someone is attacking some developers' infrastructure.)
Of course, this doesn't do anything against a malicious install environment or a backdoored compiler, but it's a start.
Generally the code quality is considered good.
Oh wait, you were just trying to score cheap snark points, weren't you?
[1] - https://www.nsa.gov/ia/mitigation_guidance/security_configur...
Your comments is naive, I feel?
Let's not get into an OS war between us here, but looking at the changes over the last 15 years, it's clear that MS is warming up to Linux. This is something that was absolutely unthinkable during the Gates and Ballmer eras.
The linux community must be doing something right when developer demand and excitement is so high for ssh.
And no, epoll isn't enough.
I think kqueue is the best (well I guess I'm a BSD fan :D) but in general: app developers generally don't care about this, they just use libuv or libevent or something like that.
Does Windows really have "good stuff" for developers? .NET is excellent, but the lower level APIs look horrible. (And the CP/M / DOS era files stuff – drive letters, file extensions that actually matter – makes me very unhappy.)
Jails, ZFS, DTrace, FUSE, netmap, PAM, pf – that's good stuff!
It is, more or less, what it is. lower level APIs cannot change without breaking backwards compatiblity. Microsoft as an organization is all about backwards compatibility. So this is just not going away, like you say there is stuff in there from DOS that will just not ever go away.
If all a developer knows is Windows, it seems awesome.
https://speakerdeck.com/trent/pyparallel-how-we-removed-the-...
I also really like kqueue (and unsurprisingly FreeBSD is also my OS of choice) and it's better in general than epoll, but it has its limitations. For instance, directory and file change notifications are less than awesome as kqueue needs file handles. inotify on Linux works better for this stuff, IOCPs just work for this stuff.
My point was that dismissing everything in Windows like jadeddrag did is shortsighted and dumb. It's not as if Unix-like OSs don't have plenty of misfeatures: nobody in their right mind is going to say that sockets couldn't have been designed much better (see how Plan 9 implements them for a better designed API), or that audio on Linux isn't a godawful mess, or that X isn't a monstrosity.
You're right, all that stuff is great, but as they saying goes, "why worry about a speck in the other guy's eye when you've a log in your own?"