Future of Singularity?

Apr 12, 2011 at 11:03 AM

Hi! Great job! I've built changeset 45126 last night, not tested it yet though.

I'd like to know what is the future of Singularity project? Are there any plans to develop it into a real system? Is it under development now or is it abandoned? Neither project page on the ms research site nor project page here on codeplex conatin any up-to-date information :(

Apr 13, 2011 at 6:11 AM

I'd like to know it too.

Once I attempted to write to somebody developer, but none of them didn't reply.

Apr 13, 2011 at 7:05 AM

I don't work for MS or anything so take this with a grain of salt but there's a rumor floating around that there's a commercial incubation project based at least partly on the Singularity research project.

It's supposedly codenamed Midori and sometimes referred to as Windows.Next.


Sounds a lot like Singularity but with added support for sandboxed/VM native code for backwards compatibility.  And I'm assuming it would have a much more sophisticated set of drivers and APIs if it is intended as a commercial OS.

I do kind of wish they'd license Singularity at least for embedded projects, though.  But that would compete with the .NET Compact Framework and I expect they wouldn't want to muddle things for developers by making two very different platforms for similar uses.

If you want to do more than what you can do with Singularity at the moment (due to licensing) you'd have to look at similar open source projects like Cosmos or SharpOS.


SharpOS, I think, has been rolled together with another project as MOSA:


But I'd be very surprised if these don't step on a variety of MS patents which could get sticky for anything commercial.

Plus once you've done much work with Singularity, those projects would prefer you don't contribute to them because it potentially pollutes their intellectual property.

Apr 13, 2011 at 7:24 AM


Well, these news about Midori and its similar to Singularity are all from 2008. They appeared the same time Singularity RDK 2.0 was released AFAIU.

As I can see, Singularity is better designed than SharpOS or CosmOS. If it is going to be a commercial (or at least public) OS I would really like to write some stuff for it - simple GUI framework, for example, or some CLI utils. Just for fun, u know :)

Apr 13, 2011 at 7:53 AM

I agree Singularity is far and away better but the licensing is too restricted to be anything more than a toy unless you just happen to be working on a degree.  There's significant disincentive for anyone to develop or contribute improvements, drivers or shared services for it -- and without those there's not much point in developing applications for it.  I'm sure the student OS developer crowd (which I'm guessing is rather small) probably love it for experimenting but that's about all it can be used for.  

It's fun to tinker with it but even for robotics and embedded network device experimentation (which require far less sophistication than a desktop OS) most people are probably better off sticking with .NET Compact Framework or an embedded build of Linux with Mono just so you don't have to work so hard to write drivers for everything and you at least have the option of commercializing or open-sourcing.

I have no doubt Singularity would quickly become a much better platform with a wide variety of contributed drivers, services and applications if they'd just allow people to license it for FOSS or commercial users.

I'm not going to hold my breath, though.  Like I said, it'd compete with existing MS offerings.

When I first started tinkering with the first release I kind of saw it as the modern equivalent of XINU -- which I extended quite a bit back in the day... but XINU is GPL which at least allows you to both do FOSS work on it and distribute commercial applications that run on it. 

FOSS only works because the community has incentive to share.   The academic community is small and doesn't have a huge incentive to share outside of published papers and most of the rest of us don't want to build practically everything from scratch just for a toy we can't distribute or sell.







Apr 13, 2011 at 8:01 AM

Sure. U're right. In the first post I was asking what is the future. They could change license, for example.

Apr 13, 2011 at 8:21 AM

My guess, they wait till their internal team gets something like Midori to the point where it's seriously feature-competitive with CE/Compact Framework/WP7 at the very least before they put out something that can be used outside of academic research.  And with the rumored size of the team... I'd say at least 2-5 years from when we first heard about Midori.  And that's if they don't wait till it could compete fully on the desktop.  And of course that OS, if and when it is released, will probably be shared source not open source.  And it'll likely look quite a bit different from an application developer's perspective.

I've heard it suggested that the next version of windows will run on ARM as well... That suggests a possibility of merging the codebases and doing a sort of scalable distro like linux uses to go from devices to mainframes and clusters with the same codebase.  If that's their plan then Midori probably won't see the light of day until it's capable of replacing Win7/Win2008R2 as well as CE/Compact Framework/WP7.

I would think we'd see more signs and rumors pointing to a significant ramp up in the size of the project if that sort of thing were going to happen for the next version of windows so it could be another 5-6 years easy in that case.



Apr 13, 2011 at 9:28 AM
Edited Apr 13, 2011 at 9:30 AM

Hmm, but XINU, as I saw, isn't truly managed object-oriented OS as JNode and Singularity. But, maybe, I'm erroring, however.

And it is almost nothing about Midori in official sources, excepting, AFAIR, only one  presentation where slip the word.

Sorry for my English.

Apr 13, 2011 at 9:33 AM

Still hope to hear smth from developers :)

Apr 13, 2011 at 9:39 AM

I'm too hope. :)

But, I'm afraid, we  couldn't see they for a long time... :(

Apr 13, 2011 at 9:46 AM

No, XINU was a "how to build an OS" example (from a book used for instructional purposes at Purdue University, originally) from a different era... Linux is to XINU like what one might expect Midori to be to Singularity.  It was an academic exercise in building a simple yet state of the art OS in the 80s -- at the time that meant a Unix-like OS in C.  The similarity really ends with the academic focus and minimalistic initial functionality.  It's since been ported to about every platform there is and runs most anything that runs on unix/bsd/linux and was even used as an embedded OS in some commercial products.

I was just pointing out the difference between what happens when you release something like that GPL vs academic licensing.  XINU is what got me away from DOS and into Unix and Linux starting in the late 80's as I ported my multi-threaded multi-line BBS system from DOS to XINU -- rebuilding on Linux was trival when that OS took off.

But I can't distribute anything for Singularity now and I don't have much hope that any commercial OS based on the project would be similar enough for a trivial port.  It's a non-starter for anything big as far as I'm concerned.

Apr 13, 2011 at 9:53 AM

From a business perspective they'd be crazy to do anything other than possibly paid licensing for embedded projects because they'd be potentially jump starting a competitor for their own future product line.  And there's not much of a business case for competing with their existing feature-superior offerings for embedded development just yet, either.

Really, I wouldn't hold my breath for any movement on the licensing.

Apr 14, 2011 at 5:48 PM

If you read the papers published on Singularity's website at Microsoft Research (http://research.microsoft.com/en-us/projects/singularity), you will find out that Singularity was never meant to be a commercial OS. It is just a prototype and it's purpose is to find out the advantages and problems with building operating systems in high-level languages, such as C#.

As to the future of the project, the source code has been publicly released so that academic institutions, such as universities, can use singularity for their projects. For example, I am currently writing my Master thesis about Singularity. However, AFAIK Microsoft Research is no longer working on the project.

Apr 15, 2011 at 1:18 AM
MpDzik wrote:

... you will find out that Singularity was never meant to be a commercial OS.

It is also very easy to tell by just looking at the source. There is no anti-piracy / DRM related code to be found... Normally Microsoft would start with this and then build the rest of the operating system as a secondary requirement. ;)

Then again, it would be quite impractical to activate the thing every time you go to debug it!