Software Transactional Memory

[imported_Dhaval00]

Folks in Redmond have been busy: http://msdn.microsoft.com/en-us/devlabs/ee334183.aspx

I installed it on a VM running Server 2008 and VS 2008 - no issues so far. They are planning on including this in Framework 4.0.

There is a white paper (PDF) which should provide ample starting info. Enjoy at your own risk .

 

[Markbnj]

I'm looking forward to checking this out. Need to make sure I can install it alongside 3.5 without screwing anything up, though. Haven't found a plain statement on that yet. The blog linked from the page Dhaval posted has a good discussion of contract semantics in the first post.

 

[brandonbull]

Surprised it is only for C#.

 

[ObscureCaucasian]

Originally posted by: brandonbull
Surprised it is only for C#.

It is like.... prerelease.

 

[Markbnj]

This blog post by Yossi Levanoni gives a good introduction to the Bartok CLR model on which the STM release was based. The debate in the comments with Dmitriy V'jukov of Intel is also pretty informative.

http://blogs.msdn.com/stmteam/...rogramming-models.aspx

 

[Crusty]

Originally posted by: Markbnj
This blog post by Yossi Levanoni gives a good introduction to the Bartok CLR model on which the STM release was based. The debate in the comments with Dmitriy V'jukov of Intel is also pretty informative.

http://blogs.msdn.com/stmteam/...rogramming-models.aspx

Thanks for the link, that's definitely a good read

 

[degibson]

Originally posted by: brandonbull
Surprised it is only for C#.

There's a lot of subtlety getting an STM to be intuitive, useful, and fast in an unmanaged environment. Strong atomicity especially is quite hard.

 

[brandonbull]

Originally posted by: ObscureCaucasian

Originally posted by: brandonbull
Surprised it is only for C#.

It is like.... prerelease.

I say that because I saw an article speaking to VB and C# having a co-evolutionary approach.