Charles Torre again sits down with some of the people behind MSR’s Singularity research OS. This time, they drill down into the architecture of Singularity and discuss design decisions, usage of safe code, Channels, SIPs, etc. They even manage to get Galen Hunt, the OS Guy, up to the white board to map out some of Singularity’s architecture.
Research OS Singularity Revisited
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
As far as my understanding of IT goes, good OS should have at least it’s half implemented in hw (CPU).
I would like to see more radical redesigns and aproaches, both in hw and sw, though industry/market will always oppose them.
I wonder, what happened to Plan9 anyway?
This is what InMos transputers (and their ancesors from ST) provide.
They basically have a nanokernel implemented in hw (no memory protection though).
This allowes to have sofisticated multi threaded environments runnung with satisfactory speeds on simple 40Mhz cpus.
“I wonder, what happened to Plan9 anyway?”
It’s still going strong, see: http://www.cs.bell-labs.com/plan9dist/
The funny thing is that alot of these radical approaches never make it to mass market in the first instance, many design features of Plan9 can be found in things like AFS which in turn heavily influenced NFS4. Look at BeOS’s BFS and see how it has inspired many of the journalled FS’s we have today.
I think their is more to the whole dotnet thing than meets the eye, if they manage to move all of their applications on top of dotnet, they can then bring across a large portion of their application base to any new platform by merely porting the dotnet CLI. Perhaps that new platform will be singularity.
Indeed, dotnet has to me always been Microsofts ticket of f their current OS. I bet we’ll see Microsoft try to perform the same transition Apple pulled off by moving from MacOS 9 to MacOS X in 5-10 years time. Before that, they’ll probably try to milk the cow dry first.
and all of the dotnet cli stuff already runs off freebsd now…
Well, MS had to let some other OS run donet or they would have just fueled the fact that they are a monopoly. FreeBSD was the lesser of evils compared to handing Linux a dotnet implementation.
I’m pretty sure everybody already knows that microsoft is a monopoly.
Yes, but it would have been significantly more difficult for MS to get C# approved as a standard had it only been implemented on Windows…
Edited 2005-12-04 05:19
the age old design principle – forget optimisation – do a good general design and get it right – let performance be sorted out at the very last stages … and with todays and tommorow’s hardware its not an issue often (out-typing msWord? just upgrade your CPU!).
so i think MS has something here that will have the reliability and resilience to the various unix OSes…
Are you trying to be sarcastic?
i realy hope this gets to be a release to the people in some way shape or form as this would be amazing to play with.
When did you honestly last get to play with a Microsoft OS? If you want to play with an OS you’re better off with one of the open sourced ones. Have a go at AROS, BSD, Darwin, Linux, Plan 9, Solaris, heck even HURD in its somewhat sorry state. Anything coming from MS will not be for play and it’ll cost you to use it at all.
Good luck in getting this into a product. I watched the whole video overview over at Channel9. I have to say these guys are really smart over at Microsoft Research. I wonder if this is only an academic exercise or if they really are planning the development of a actual product in the future. Only time will tell. The way this is designed it seems it’s for “dependability” first and foremost. Something to run the USS Enterprise on.
Seems to me that this OS has more of an appeal to play with rather than the Open Source ones as the Open Source ones are rather archaic in appearance to the general consumer. Seems like this managed code stuff being that it is popular in the IT world along with the excellent documentation from MSDN as well as VS 2005 which is excellent as well makes it more interesting for most curious folks. If MS open sourced this it would be great! But we all know that is not going to happen and I think that this research OS will ultimately transform into a product which will implement ideas from the Research project.
Seems to me that this OS has more of an appeal to play with rather than the Open Source ones as the Open Source ones are rather archaic in appearance to the general consumer.
Sounds like you want to check out L4 then.
http://os.inf.tu-dresden.de/L4/
And we also know that you are one of the burned ass open source apologists who are not able to write one decent desktop OS.
Linux is a Loser invented another unix xerox(clone)…where is the innovation guys…ha ha ha…innovation my ass…guys can’t even get a decent package manager on the name of choice…
And we also know that you are one of the burned ass open source apologists
Nope. You must be thinking of someone else.
who are not able to write one decent desktop OS.
That may be true, but neither has anyone else for a long time.
Linux is a Loser invented another unix xerox(clone)
What?
where is the innovation guys…ha ha ha…innovation my ass…
In the same place the innovation is in other OSes–at the user level. You’d be surprised at the kind of apps available for linux.
guys can’t even get a decent package manager on the name of choice…
If you say so…
Linux is a Loser invented another unix xerox(clone)…where is the innovation guys…ha ha ha…innovation my ass…guys can’t even get a decent package manager on the name of choice…
Agree.
“Linux is a Loser invented another unix xerox(clone)…where is the innovation guys…ha ha ha…innovation my ass…guys can’t even get a decent package manager on the name of choice…”
It never ceases to amaze me that people want “innovation” in their operating systems. Obviously the top consideration when evaluating the worth of an operating system is the availability of application software for that operating system. In this sense, Windows should be hands down the best operating system.
Most general purpose operating systems are either UNIX or Windows-based for this reason (compatibility/portability), and this is why Linux is UNIX-like. But going back to the above metric, Linux is the best UNIX-like operating system because more software runs natively on Linux than any other UNIX-like operating system. Is that where the innovation is?
And what about Singularity? Even if it is so incredibly advanced that it makes OS architects everywhere wet the bed, it won’t run a tiny fraction of the software that you and I want to run.
Let me put it simply: no one has created a “decent desktop OS.” They are all works in progress. The only difference is that Linux is building from the bottom up, Windows is building from the top down, and Mac is building from the inside out. Linux desktops lead a little fit and finish, Windows needs work under the hood, and the Mac needs to expand into more markets.
Oh really…. How about apt with synaptic running on top.. Makes my job as an admin about 90% simpler then windows.. Especially with ubuntu and upgrading is as easy as swapping a few repository names and hitting the download button… Gee whiz.. Wish I could have that in windows free of charge!! ha! You might wanna try out klik as well… That project definetly gets my vote as innovative! One single image file to run an app and that’s it… WoW! Where’s that on windows!?! Yeah… I see your point…. No innovation in package management.. Why not try looking for a solution instead of complaining…
It sounds like C# is just C trying to be Ada:
Garbage Collection, Type safety, Pointer safety…
Programming languages evolve on a common base of theories and principles, created by scientists. Thus almost all programming languages go into a common direction with plenty of other languages.
Isn’t that very obvious?
Smalltalk, not Ada! Get it right, you idiot!
(too bad no one here will get the joke…)
No, C# is just Java trying to be C++!
Singularity could be able to outperform all the VMs running on “classic” operating systems. The “classic” operating system is a bit of overhead there.
to see all those kids on this site trying to act like engineer and phd about OS design and programming language
oh and show us your cred anon coward
Anonymity counters vanity. On a forum where registration is required, or even where people give themselves names, a clique is developed of the elite users, and posts deal as much with who you are as what you are posting. On an anonymous forum, if you can’t tell who posts what, logic will overrule vanity. As Hiroyuki, the administrator of 2ch, writes:
If there is a user ID attached to a user, a discussion tends to become a criticizing game. On the other hand, under the anonymous system, even though your opinion/information is criticized, you don’t know with whom to be upset. Also with a user ID, those who participate in the site for a long time tend to have authority, and it becomes difficult for a user to disagree with them. Under a perfectly anonymous system, you can say, “it’s boring,” if it is actually boring. All information is treated equally; only an accurate argument will work.
Yes that system works in an entirely anonymous system. When the system is *not* based on anonymity, remaining anonymous can be taken to mean you are not willing to stand by your post.
Innovation would be then to port that software, not keep on running an ageing (by design) platform because the new , more innovative (by design) platform needs to get that software.
Unfortunately, the world doesn’t work like this, but it is no good to say that it is good that it doesn’t work like this.
You may be interested in Desert Spring-Time:
http://dst.purevoid.org
That is an open-source OS written in O’Caml.
Browser: Lynx/2.8.5dev.7 libwww-FM/2.14 SSL-MM/1.4.1 OpenSSL/0.9.7
Good luck in getting this into a product. I watched the whole video
overview over at Channel9. I have to say these guys are really smart
over at Microsoft Research. I wonder if this is only an academic
exercise or if they really are planning the development of a actual
product in the future. Only time will tell. The way this is designed
it seems it’s for “dependability” first and foremost. Something to run
the USS Enterprise on.
>
>
You must be refering to the *DEATHTRAP* Enterprise of TNG. Yeah, that ship *RUNS A REAL STABLE OPERATING SYSTEM* What with *HOLODECK CHARACTERS TAKING OVER THE SHIP EVERY HOUR ON THE HOUR* and other such things. Yep, definately a Microsoft Operating System is being used there.
I’m sure Microsoft could make the most stable, reliable, crash resistant operating system in the world; but the fact is, if they don’t get the hardware and software support behind it, plus the ability to run almost every damn legacy application that ever existed (thanks to cheapskate users who expect perpetual backwards compatibility) – it won’t take off.
Windows unfortunately is a compromise; if there were those issues, and every ISV and IHV simply supported the next operating system, even if it completely broke compatibility, then Microsoft would let a rip and create a radically new operating system.
Even with MacOS X they had to compromise; Carbon has compromises – and I”m sure, in a perfect world, all would use Cocoa along with objective-c, but the reality of the situation is, the world doesn’t revolve around that language, and as such, a compromise is made.
So umm they are writing it in C# and Spec# right? So are they then using that Bartok compiler to create code that will run directly on a machine without requiding a VM? Isnt the CLI a kind of VM just MS’s own twist?
This is absolutely wonderful, it’s the DOS days all over again for the next 20 years. Although this time the apps are on the internet so it won’t be a hard transition to getting the Singularity OS into consumer’s homes.
Why DOS days?