Java Concurrency in Practice

This is clearly one of the top notch Java books like Doug Lea's original "Concurrent Programming in Java". And like CPIJ it is scary. Can one ever get the concurrency aspects right enough? Are the books of other authors trustworthy (enough)? There is so little help from the language itself and the IDEs to get things concurrency correct. And the authors show in many examples, what all can go wrong and that our old "Von Neumann machine" intuition is plainly wrong and often highly misleading. With the inevitable concurrency, Java is in fact a language for advanced programmers. Things will in practice get worse due to the increasing ubiquity of multiprocessor computers even on the desktop. The book is written by leading Java experts. It cites and uses in an unusually extensive ways some of the other classics in our Java world: Arnold et al. "The Java Programming Language", Lea "Concurrent Programming in Java", Gosling et al. "Java Language Specification" and Bloch "Effective Java". It is helpful but not mandatory to know them. Better it is to have them handy to be able to quickly look things up. Most other referenced works are original articles. "Java Concurrency in Practice" is written in a readable style - though the material is presented in an unusual dense way for an Addison and Wesley Java book. Expect an information density more like an O'Reilly one, but a lot lower than a Springer one. Anyhow the book gets easier to read as you and your understanding progresses. The presented material relies on and explains the new concurrency features of Java 5 and even Java 6. But it is not a tutorial of the new concurrency classes. It is a high level introduction from the usage pattern point of view. It explains the new classes only as far as is necessary. One of my favorite chapters is on testing concurrent programs. Yes, it is possible to make unit tests for concurrent classes. No, it does not look like it is much fun though. But, you get a good head start. Besides peer review you also find some testing help in static analyzers like "findbugs". In summary I recommend this book as one of the core Java books. I still wish the world is - with respect to multithreading - easier and less intimidating.

Concurrency is hard and boring. Unfortunately, my favoured technique of ignoring it and hoping it will go away doesn't look like it's going to bear fruit. Fortunately, Java 5.0 introduced a new bunch of concurrency utilities, that work at a higher level of abstraction than marking blocks as synchronized and fields as volatile. Unfortunately, there haven't been that many books on the subject - even the good Java 5.0 books (e.g. Head First Java or Agile Java) make little mention of them - Thinking in Java being an honourable exception. Fortunately, JCIP is here, and it is authoritative stuff. And it's (mostly) very easy to understand. Plus, at 350 pages, it's not an enormous chore to slog through. It even covers changes to the upcoming Java 6. Before tackling this book, you should have at least some idea of pre-Java 5.0 concurrency. You don't need to be a threading master, though, as the first part of the book covers basics like deadlock, atomicity and liveness. This was my favourite part of the book, as it comes with lots of small code snippets, both right and (horribly) wrong, and pithy design guidelines. It's rather like Effective Java in that respect - although the material on threading was probably the weakest part of that book, so this is a definite improvement. The second part deals with thread pools, cancellation strategies, and GUIs. This is also excellent. Part three covers performance and testing. The last 75 pages are for advanced users and goes into a fair amount of low level detail (including a discussion of the Java Memory Model), which may be of interest to experts only. I would be lying if I said that reading this book will demystify concurrency completely. Who wrote which bit of the book is unclear (although readers of Effective Java will probably spot parts of the text that seem rather Joshua Blochish), but while it's mostly very clear, some parts of the text are a little murkier than other. Perhaps this is to be expected given the subject matter. But for the most part it's surprisingly easy reading, and very practical to boot. Let's face it, short of aliens landing and introducing a radically new way of computing, multicores are here for the medium term at least, so thread-dodging wimps such as myself will just have to get used to it. That being so, this book is going to be installed as one of the must-read books in the Java pantheon.

The book is by far the best one on Java concurrency. There is really nothing out there that has such comprehensive coverage of this topic. Doug Lee's book is a bit theoretical and somewhat dry, but would be a nice complement to this book if you want to think some more about concurrency. This book has a very strong practical vector. Coverage of Java 5 concurrency features is very thorough. The book is extremely well-written, relatively easy to read. The book stands on par with such established Java book jems as Josh Bloch's "Effective Java", Eckel's "Thinking in Java" and Rod Johnson's J2EE books. All in all, this is a definite must have for any Java specialist.

Concurrency, in the form of threads, has been present in the Java language from its beginning, and this book is all about concurrency in the current and future versions of Java with an emphasis on writing practical code. This book does for concurrent programming in Java what Geary's series of books did for graphical Java - it moves concurrent Java programming out of the realm of applets containing bouncing balls and into that of providing real solutions for professional programmers. This book is not meant to be an introduction to concurrency in Java. Its intention is to offer practical design rules to assist developers in the difficult process of creating safe, fast, and high-performance concurrent classes. While many of the general concepts in this book are applicable to versions of Java prior to Java 1.5, most of the code examples and all the statements about the Java Memory Model assume Java 1.5 or later. By "later" I mean that some of the code examples use library features added in the not-yet released Java 1.6. After the introduction, which consists of Chapter 1, the book is divided into four parts: Part one, "Fundamentals", (Chapters 2-5) are about the basic concepts of concurrency, thread safety, and composing thread-safe classes from those concurrent building blocks provided by the Java language. Chapter 2, "Thread Safety", and 3, "Sharing Objects", include nearly all of the rules on avoiding concurrency hazards, constructing thread-safe classes, and verifying thread safety. Thus, these chapters emphasize theory and have less code than other chapters in the book. Chapter 4 , "Composing Objects", covers techniques for composing large thread-safe classes from smaller thread-safe classes. Chapter 5, "Building Blocks", covers thread-safe collections and synchronizers, which are the the concurrent building blocks provided by Java. To conclude the section, the authors work through the steps of building an efficient, scalable result cache that could be used in a web server. A summary of the most important rules presented in Part one occur at the end of the section. Part two, "Structuring Concurrent Applications", describes how proper use of threading improves the throughput and responsiveness of concurrent applications. The topics covered in this section include identifying tasks that can be run in parallel and programming them as such, proper termination of tasks, using thread pools for greater efficiency in multi-threaded systems, and finally improving the responsiveness of single-threaded systems, GUI's being the most prominent example. Part 3, "Liveness, Performance, and Testing" is concerned with ensuring that concurrent programs actually do what is expected of them and do so with acceptable performance. The authors describe how to avoid situations where a thread waits forever, also known as a "liveness failure". Also included in this section is an excellent explanation of the use of the "ThreadLocal" class and how it makes it much ea

We've been reading a pre-release version of this book as part of a local study group and I'm very impressed with the treatment. Concurrency is perhaps one of the hardest issues to understand well and it's equally difficult to explain all of the issues, but Goetz et. al. do a very nice job of explaining clearly the different ways a multi-threaded process can fail and then providing concrete design philosophies that will help address those problems. In my experience, concurrency books often fall between providing too little detail ("just add synchronized to everything and all should be well") to providing too much ("details of how the Java Memory Model actually behaves on a multi-processor machine"). Java Concurrency in Practice seems to find a nice balance, e.g. thoroughly explaining why you need to worry about how values are "published" between threads but without swamping you with so much information about the details of how this is done by the VM that you're left gasping for air. For me, this is a focus on the practical aspects of building multi-threaded applications in real world situations. In such cases, you need to fully understand the potential pitfalls and then you want to have a list of specific design methodologies which help avoid those failings. Java Concurrency in Practice does a nice job of providing both.