Communicating Sequential Processes
The biggest problem in concurrency is that tasks can interfere with each other. There are certainly other problems, but this is the biggest. This interference generally appears in the form of two tasks attempting to read and write the same data storage. Because the tasks run independently, you can't know which one has modified the storage, so the data is effectively corrupt. This is the problem of shared-memory concurrency.
You will see later in this book that there are concurrency strategies which attempt to solve the problem by locking the storage while one task is using it so other tasks are unable to read or write that storage. Although there is not yet a conclusive proof, some people believe that this dance is so tricky and complicated that it's impossible to write a correct program of any complexity using shared-memory concurrency.
One solution to this problem is to altogether eliminate the possibility of shared storage. Each task is isolated, and the only way to communicate with other tasks is through controlled channels that safely pass data from one task to another. This is the general description of communicating sequential processes (CSP). The sequential term means that, within any process, you can effectively ignore the fact that you are working within a concurrent world and program as you normally do, sequentially from beginning to end. By defending you from shared-memory pitfalls, CSP allows you to think more simply about the problem you're solving.
This book explores a number of strategies that implement CSP, but the easiest
place to start is probably Python's built-in
multiprocessing module. It turns
multiprocessing is not a seamless implementation of CSP, but we
shall start by pretending that it is and ignoring the "leaks" in the abstraction.
This produces a fairly nice introduction to CSP, and later in the book we can
address the leaks as necessary.