Microsoft Singularity Project {OS Research}

[BehindEnemyLines]

I came across Microsoft's research on an Operating System applications, device drivers, and kernel run in "managed code" (Wikipedia) started in 2003. Will MS bring this new resarch into their next OS considering it's already three years into the research. It seems promising, but we all know how these concepts sometimes aren't realized...

More info: http://research.microsoft.com/os/singularity/
Video: http://channel9.msdn.com/Showpost.aspx?postid=227260

 

[drag]

Maybe it will bring some improvements in terms of applications using managed code or maybe something using secure drivers or whatnot.. but not in terms of real operating systems.

A operating system as complex as Linux or Windows would require computers much more powerfull then they are now if they were rewritten in managed code.

I beleive Microsoft originally wanted to rewrite the explorer shell in Windows Longhorn using managed code, but that was scrapped after a few years of development due to the performance penalty. There is a very good reason why operating systems are still written in C.

Actually I would expect that this project is about as much about improving the performance and fesability of the programming language rather as it is seeking ways to design operating systems.

One of the classic 'proofs' of a programming language is the ability to do things like writing it's own compiler. This is very difficult thing to do write and if your unable to do it then it pretty much proves that your new language is not nearly as capable as the language you originally used to write the new language. It's a regression in functionality which is not how things should be going in programming. (as a modern example see the folks working on writing Pypy interpretor for python to eventually (hopefully) relace the current C one.)

I think that this using singularity would be used to expose problems in .NET and would lead to eventual improvements in later releases.

Also writting in a higher level language is very much more productive. So even if you end up with something that is slow it may provide new insights to operating system design and developments much quicker then if they were forced to program in C for a production environment. That way if they find anything that is cool then they refactor it to C and incorporate it's design back into NT or NT's replacement.