Managed OS and contracts
November 08, 2005 11:46Today I had some time to flush my blog queue, and I found two
interesting posts: the last episode of Mark Russinovich "war" against
Sony and its buggy rootkit used as a DRM system (surely something
agaist trused computing.. I'd like to know if anyone knows Microsoft's
official position) and Adi Oltean's Sunday post on the Singularity OS at http://research.microsoft.com/os/singularity/
I think the first one is surely worth discussion, but many (too many, since it reached the mainstream media) words have already been spent.
The second one came unexpected, and with a little sense of envy.. for not being there. It's since my conversion to the CLR that I'm dreaming and speculating on how to build an OS in "managed" code (or using as much managed code as possible). And now they have done it!
In the technical report there are many interesting ideas: from contracts (not so different from the one's we'll be using in Indingo, it seems) to the use of MSIL to produce verifialble code, the global optimizing compiler (better than a JIT) or the global garbage collector. And versioning, trusted components, and all the benefits the CLR brought to us... OS wide!
As my MSC thesis was on concurrency contracts for the CLR, I think a very interesting part of the project is the Sing# programming language. I had experience of the Spec# programming language (a good project from another MSR team). I actually had the luck to speak with Schulte and Tillmann from the Spec# team at SAPS last year. Their work inspired mine on the CxC language. Now I really want to go and see how Sing# works with concurrency, and if it is possible to fit it with my idea of concurrency contracts. I will do further reading on the subject in the following days, time permitting as always. For those interested, you can find the paper here, and my thesis here.
I think the first one is surely worth discussion, but many (too many, since it reached the mainstream media) words have already been spent.
The second one came unexpected, and with a little sense of envy.. for not being there. It's since my conversion to the CLR that I'm dreaming and speculating on how to build an OS in "managed" code (or using as much managed code as possible). And now they have done it!
In the technical report there are many interesting ideas: from contracts (not so different from the one's we'll be using in Indingo, it seems) to the use of MSIL to produce verifialble code, the global optimizing compiler (better than a JIT) or the global garbage collector. And versioning, trusted components, and all the benefits the CLR brought to us... OS wide!
As my MSC thesis was on concurrency contracts for the CLR, I think a very interesting part of the project is the Sing# programming language. I had experience of the Spec# programming language (a good project from another MSR team). I actually had the luck to speak with Schulte and Tillmann from the Spec# team at SAPS last year. Their work inspired mine on the CxC language. Now I really want to go and see how Sing# works with concurrency, and if it is possible to fit it with my idea of concurrency contracts. I will do further reading on the subject in the following days, time permitting as always. For those interested, you can find the paper here, and my thesis here.