Alternative operating systems
techradar.com
techradar.com
In any case, two tongue-in-cheek:
``Systems Software Research is Irrelevant'' by Rob Pike http://doc.cat-v.org/bell_labs/utah2000/
``Why Some Dead OSes Still matter'' by Andrey Mirtchovski and Latchesar IonkoV http://doc.cat-v.org/plan_9/misc/dead_OSes_still_matter/
I discovered that at least one Plan 9 invention had success and is found today everywhere: UTF 8 encoding.
The OS is under development, and as more people jump in [1], the pace increases. A lot of software has been ported natively or to APE; you can get a pretty decent UNIX workstation with P9 alone those days.
On the other hand, if you want to try it the easy way, consider http://swtch.com/plan9port/ Runs on virtually any POSIX environment, but it's just a teaser of Plan 9. I'm using the Acme editor from it daily.
Linux adapted quite a bit, the clone() syscall, the --bind option for mount (only partly implemented, thou), the pervasive namespacing, the XRender API (modeled after BitBlt) and some more.
In a way Git's storage resembles Venti [2] but probably is just convergent evolution.
--
[1] http://9fans.net/archive/ search powered by grep :D
[2] http://en.wikipedia.org/wiki/Venti
EDIT Re: UTF-8
It's a funny story; Linux only imported half of the Unicode support from Plan 9. In P9, UTF-8 is used on-the-wire and on-the-storage. However, in case a process wants to deal with sizeable amount of text (say, a text editor, a grep etc.), it converts the UTF-8 into `Runes' -- strings of 32bit integers (till recently 16bit) directly denoting Unicode characters, for fast access by offset. And back to send over wire or write to storage.
Another nice story, you get a network connection with just one libcall: dial().
For example, int fd = dial("tcp!news.ycombinator.com!http", 0, 0, 0);
and that's it :D
No virtual machine, just a tarball (available for freebsd, linux and mac os x) that uses vx32 (http://swtch.com/9vx/) to dynamically execute plan9 code in a sandbox.
* tugs at Locke1689's neckbeard to see if it's fake
As soon as I read that someone was building an OS entirely from Java, I immediately wondered, "How did they solve the memory consumption problem?"
Evidently, they didn't. Why even start such a project without first attacking this obvious, fatal problem?
I actually played around with FreeVMS, but I am too young to have ever experienced one in the wild, so after I successfully booted it, I was then stuck. :-/
I have not dabbled with JNode in many, many years, but I find it especially amusing that one of their upcoming release goals is to "use less memory." :-)
Real operating systems have to run on a myriad of hardware and potentially interface with tons of devices and protocols, so you would need to allow programmers to go to the hardware level and read and write bytes by hand.
Historically untrue, and still not true for OSes like Windows NT (modern versions x86 and Itanium only), MVS (z Series only), and, of course, Mac OS X.
Historically, I think Unix was the first OS to run on multiple different kinds of hardware, as opposed to running on multiple variants of the same architecture.
Real operating systems
The article is talking about things that are not simply evolutions of existing technologies, but outright reinventions that show promise.
Using the ability to run Firefox as a metric, I think only Haiku, ReactOS and OpenBSD qualify as great.
Except for OpenBSD, which I suspect they made look as old and frumpy as possible for that article. I'm sure OpenBSD can run about as many attractive window managers as Linux can.
IMO, the most exciting OS stuff is far from visible in the UI, let alone screenshots...