SkyOS does not fall into any of this categories, because it’s not compatible with proprietary software and I doubt that it will get much support from the open-source community.

So as an end-user buying SkyOS would be pretty much pointless.

The way I think of it, they all have their place..

1. OS X for the end-users, mobiles developers (Android or iPhone), design teams etc..
2. FreeBSD (and such-like) servers for large scale solutions (or for web sites with high throughput)
3. Windows – in the toilet

FreeBSD is an awsome OS, it just doesn’t have a GUI.. deal with it?! ;-)

I would have liked to get further into it, but I had to decide between being more thorough and getting the freakin article up before I went home! Also, I’m not familiar with a lot of these so I would be out of my depth. I was going to include Haiku… and I believe Syllable is Amiga-based? And I listed Plan 9 separately because it is pretty different, UI and otherwise.

I fully agree with you that major changes aren’t really going to be found in just people tweaking the usual formula – but I don’t know enough about the history you mentioned to include what might be scraped from those options. But you raise good points about the nature of static code… I think the “why” ends up being pretty much that it’s “good enough.”

Anyway, thanks for your insightful comments.

For PDA use it will be useful..

I think if the list was going to include Linux, something really different like Gobo would have been a better one. Makes for a dull screenshot, though.

I can’t see ReactOS ever being significant. As a Windows clone, [1] MS could kill it easily and [2] it can never really exceed Windows’ capabilities. I suspect you are showing your bias here. Haiku has a lot more of to offer than ReactOS, I reckon.

I think you’d have been better off giving some family tree type groupings – Minix, Plan 9 and Linux all belong together as Unix spinoffs, and Gobo would have been a better Linux, because it’s weirder, but it doesn’t *look* special.

As others have said, Haiku deserved a mention, and therefore probably Zeta as well.

And if you go beyond x86, there’s MorphOS and AmigaOS 4.1 and RISC OS.

But to echo and amplify what some grumbled about innovation, apart from the nutters trying to build a GUI OS in assembler, these are all monolithic multitaskers built in C or C++ taking inspiration from existing products.

For my money, there was some really interesting stuff back in the 1970s, in the form of Lisp machines, for instance, and in the 1980s, in the form of Taos, that never even gets remembered today. Things that throw out assumptions about writing in C and compiling for a particular processor, stuff that is just taken for granted now.

Why should we compile? Why should we target a native CPU? Why should there be “binaries” that are tied to one system? Why should we have static code that cannot change or be changed? Why are there build processes? Are C-family programming languages actually an outdated legacy technology that is tying us down?

These are the areas where /real/ innovation could come from, I think…

Of the OSs listed, several started from a blank slate and contain unique, ground-up kernels and userlands that are more or less totally unlike anything else. This includes SkyOS, Syllable and Menuet.

Others of which the same could be said are Haiku, RISC OS (on ARM) and MorphOS (on PowerPC). AROS is a new PC OS, but is an open-source re-implementation of AmigaOS, as ReactOS is an open-source reimplementation of Windows.

If you want exciting new kernels or security systems, then you have to look to things like Singularity (MS Research) or Coyotos/KeyOS/EROS, which are academic research OSs. The thing is, they are tiny and incomplete, have no GUId and nothing to show screengrabs of. This makes them very dull to most readers.

I don’t think your objection is a fair one.

QNX is also interesting and different, as is Hurd, especially L4 Hurd.

VisOpSys is worthy of mention, too.

Singularity shows some interesting new directions.

GoboLinux might have been a better Linux, as it is far and away the most “different” Linux out there, completely discarding the traditional Unix directory layout, which even Mac OS X keeps (but hides).

[1] PC-BSD is a flavour of FreeBSD.
[2] FreeBSD, OpenBSD and NetBSD are so alike, one needs to be an expert to understand the differences.
[3] For non-experts, they are all just Linux but less friendly and with fewer drivers and fewer apps.

And to experts who see the bigger picture, *all* the even marginally-successful rivals to Windows today are Unix: Linux, Mac OS X, *BSD, Solaris/AIX/HPUX etc., they are all Unix.

This list contained a bunch of things people will never have heard of which are not even slightly related to Unix. Sure, Minix is, and Plan 9 is distantly, but even they are less like traditional Unix than Linux is.