Teardown 01 - ArrayBlockingQueue

"Rather use java.util.concurrent than write your own"

Who has not heard this sage advice?

Of course we agree, and follow this advice ourselves. But what if the constructs in java.util.concurrent differ from what we need? Can we write our own? And what can we learn from studying how the java.util.concurrent classes work?

In our Teardown series we study the classes in java.util.concurrent line-by-line. We analyze every for-loop and if statement, asking why the authors wrote it this way. We dissect the concurrency controls, looking at how the authors apply it to the construct.

The first construct we chose to study is the ArrayBlockingQueue. The concurrency controls of this class are one of the easiest to understand. ArrayBlockingQueue has one ReentrantLock with two Conditions notEmpty and notFull. Unfortunately the weakly consistent iterators are somewhat complex. Furthermore, the authors overcomplicated their work by not keeping one space free, like we see in ArrayDeque. This means that when takeIndex==putIndex, the queue is either full or empty.

This mini-course consists of almost 4 hours of detailed instruction of how ArrayBlockingQueue works. Watch it either from start to finish, or jump around between sections as your curiosity is piqued. You will definitely learn something - or your money back!