Curious what you use as a Terminal replacement in Windows? I tried cygwin a few years ago and it was painful. Windows Powershell isn't standard enough for me.
> Windows Powershell isn't standard enough for me.
Then Windows isn't standard enough for you. The traditional UNIX toolset does not work on Windows - something as simple as chmod, or the concept of files being owned by users, works entirely differently. Things as simple as mv have entirely different semantics - there isn't a guaranteed atomic file rename function (well, actually, looks like one was implemented in Windows Vista... which doesn't work on network drives, including redirected folders which are used to map your home directory transparently to a network drive in many companies).
In administering Windows systems, practically nothing is implemented as text mangling, so text mangling isn't a core part of the system's interface.
Once you've thrown out all the stuff that works differently or can't be implemented reliably, you don't have a terrible amount left. If you want some subset of the GNU toolset for development, there's MSYS and a few other projects.
Or you could learn how to use a system that doesn't buy into the UNIX monoculture, which has been a detriment to operating system research and development for the past couple of decades.
> Or you could learn how to use a system that doesn't buy into the UNIX monoculture
My first computer was a ZX Spectrum. It didn't run UNIX.
> which has been a detriment to operating system research and development for the past couple of decades.
You're claiming that having access to free Operating Systems, with free tools and access to the source for them all has stifled OS research? That claim appears nonsensical on the face of it. Care to expand?
Certainly. It's not access to free OSes that's the problem - it's that both the major one and most of the minor ones are derivatives of either UNIX or the POSIX specification. (Haiku is the only semi-popular one which attempts to do something different.) There's little chance of stepping outside that box in any major way.
For example, a slowly-moving research project of mine is to build a fully functional OS that uses capabilities, per-application sandboxing, a generic object store instead of a filesystem, a built-in intra- and inter-computer RPC mechanism, standardised configuration (from the network admin level downwards), and a powerful Smalltalk-based GUI for managing all of this, on the idea that today's information workers need better ability to manage and script their system, including building new applications out of existing ones. The Linux ecosystem doesn't provide a meaningful, standardised interface for doing so for many office-y applications or complex formats.
While you could theoretically layer all this on top of POSIX (and in fact, that's the first implementation of my system), you are at the same time throwing away 99% of POSIX, and you are highly expected to maintain compatibility with apps for the system you're building on top of - which likely don't fit in with your system's design, and doesn't encourage the creation of applications for, or modification of existing applications to fit, your new system design.
This sort of greenfield R&D is what's been missing. Even universities are primarily stuck getting POSIX running on various kernel architectures, avoiding the topic of OS design entirely. You could point towards Android as a counter-example, but that's not a desktop OS - its primary competitors were proprietary OSes made by the smartphone vendors, and it's still not really capable of being used as a primary OS.
> it's that both the major one and most of the minor ones are derivatives of either UNIX or the POSIX specification.
How is that the problem? How does that stop you from doing anything?
> you are at the same time throwing away 99% of POSIX, and you are highly expected to maintain compatibility with apps for the system you're building on top of
Expected by whom? Who has any expectations for your own personal research project? I'd fully expect an experimental OS not to have any apps at all, besides basic utilities.
I'm not going to point to any OS as a counter example, because you haven't made clear how the existence of any other UNIX-like OS prevents you from doing anything. If you want to do "greenfield R&D" then go ahead. No-one is stopping you.
It doesn't prevent anyone from doing anything, but the result is that non-POSIX OSes will never take off in any meaningful way, therefore research is close to worthless.
New POSIX OSes won't take off in a meaningful way either (whatever 'meaningful' means). It's nothing to do with it being POSIX or not, it's just difficult to get traction for a new OS. Linux faced exactly the same issue when it was released.
But why do you need it to take off in a meaningful way? Open Source software doesn't need a particular market share to keep going, it just needs a certain absolute number of people (which differs depending on the software). Is having a small committed core of users not enough for you? If your research is truly useful then that number will grow over time.
Unless you're talking about a commercial OS, in which case Microsoft is your obstacle, not POSIX. Good luck with that.
Maybe, maybe not. IMO, capabilities-based systems are all around us these days, just not in a completely pure form and not in the way most people expected (it's at a higher level, that is, in the way apps are being built on web APIs).
I also note that Microsoft built a capabilities-based OS (Singularity - it's open source, have you considered building on something like this?) and that one of the most important modern contributors to the space (J. Shapiro) now works at MS Research last I read. Given his lifetime interest and work in the area it's not a stretch to imagine what they may have hired him to work on.
