[concurrency-interest] samples for parallel programming with Java 7

Gregg Wonderly gregg at cytetech.com
Fri Sep 16 13:30:27 EDT 2011


On 9/16/2011 10:21 AM, Danny Dig wrote:
> I am frustrated by the slow adoption of ForkJoinTask, ParallelArray,
> and other goodies for improving throughput on Java programs.
>
> One reason could be the lack of small (but realistic) projects that
> illustrate how to use parallelism to improve throughput. I think
> Microsoft is doing a good job to show samples of such programs:
> http://code.msdn.microsoft.com/Samples-for-Parallel-b4b76364/
>
> Does anyone know where to find such sample programs for Java?

I feel that current, relevant documentation is the single largest issue that 
Java has to deal with.  You might find some examples at ibm.com and on some 
other commercial vendor sites who actually do provide support for a Java VM, or 
Java related technologies.

My viewpoint is that Java has been shepherded by a largely technical and "on 
college campus" community which doesn't seem to "document" solutions, just APIs. 
  I guess that this would be because it takes time which they would have to 
volunteer, away from their real job, and in many cases, it could document how 
they are "competing" with others, and thus close the gap on any advantage they 
get from Java.

Sun used to create a lot of good examples in the early days, of basic stuff. 
But so much of that stopped when everyone and their dog were publishing books on 
Java, or so it seems to me.

The book written by Brian Goetz, and others, "Java Concurrency in Practice" 
(http://www.amazon.com/Java-Concurrency-Practice-Brian-Goetz/dp/0321349601) 
provides a great deal of resources for what existed when it was written.

But, ultimately, that is why technical books, in the end are pretty worthless 
after a few months.  Since things continue to change in the j.u.c APIs, it would 
be much better for there to be some type of online version of that book that 
could be purchased with continued updates happening.  Maybe, someday, the 
publishing industry will understand the need and implement something useful. 
The Kindle version, for example, could just be continually updated and I'd be 
extremely excited by that, particularly if there were new examples being added 
regularly as "patterns" were proven or tuned up.

Gregg Wonderly


More information about the Concurrency-interest mailing list