Anders Hejlsberg and Guy Steele: Concurrency and Language Design

Posted By: Charles | Oct 6th, 2008 @ 6:27 AM | 79,635 Views | 19 Comments
This is the second year I've been lucky enough to take part in the cross-platform software engineering conference JAOO. Like last year, I was very fortunate to get to sit down with a few key players in the programming languages design field and watch several technical presentations that span the industry and problems we face as software developers. One of the truly great things about JAOO is that it is not a product-focused conference: it's about programming first and foremost and enables the sharing of perspectives and ideas among the world's best and brightest programming minds. As you can imagine, I, like many technical types here at Microsoft, am a huge fan of JAOO. Thank you Trifork!!!

In this conversation Microsoft Technical Fellow and Chief Architect of C# Anders Hejlsberg sits down with programming language design legend and computer scientist Guy Steele (creator of Scheme and expert in several languages ranging from LISP to Java). I think Guy is one of the smartest people I've ever met.

The topic of conversation is the elephant in the modern general purpose programmer's living room: Concurrency. With today's widely-used general purpose languages like C++, Java, C#, VB, Ruby etc it's hard to express parallelism in productive ways. Anders et al are working on both language enhancements to C# and VB.NET and BCL support (Parallel Extensions to .NET for example). Today, Guy is working on a mathematical language (domain specific as opposed to general purpose) and runtime, Fortress, that is so concurrent it makes it hard for programmers to even write sequential code!

Listen in to two of the programming industry's most successful thinkers and get a sense of their perspectives on the future of general purpose programming languages now that Concurrency and Parallelism are entering the development status quo.

Enjoy. More JAOO coverage to come. You can watch Anders' keynote on language futures here.
Rating:
3
0
#Dec 7th, 2008 @ 11:33 PM
kasajian
kasajian
Assuming that someone writes software that super-concurrent, is there a way to verify that that is the case without actually trying it on a multi-processor computer?

For instance, let's say that I have a quad core computer, and I write software that scales amazingly to 4 threads.  However, my intent was to actually scale to 8 or 16 or 32 threads, but I don't have the machine to test it.

Is there some kind of execution path analysis tool (perhaps Intel can help with this), which analyzes the binary code (or may be source) and says, "yes, your code scales to 256 threads, with the following data set, but no more than that"

Is there any research done in this area that anyone knows about?
Microsoft Communities