Beyond Java [ペーパーバック]

There is a lot to like in this book, and a lot not to like. I'll talk about the negatives first and then discuss the positives.

The style of the book is extremely annoying. The author is a kayaker and a mountain biker, and he introduces each chapter with a kayaking/mountain biking story meant to serve as some kind of analogy to the programming topics he'll cover in the chapter. I found this unnecessary and distracting (I don't need sugar coating on my technical reading), and it felt like the real purpose of the stories was for the author to demonstrate how cool he is. In addition, the author uses the phrase "a perfect storm" over and over and over to describe the factors that led to Java's position of dominance in the programming world. I hated that expression even before I read the book; it has to be the most abused expression of the last few years.

Despite the fact that the book is ostensibly about programming languages, the author is by no means an expert on the subject. To his credit, he admits this freely, but he also makes numerous small and not-so-small mistakes when describing programming language features which may lead more knowledgeable readers to wonder if he's really qualified to write this book. For instance, in several places he describes the advantages of static typing as being mainly for early error detection, without also pointing out another big advantage of static typing: faster code (there are other advantages as well). In another place he makes the blatantly false statement that "Smalltalk is where all the continuation research is happening", ignoring the fact that Scheme (a Lisp dialect) has had continuations since 1986, and that there has been and continues to be active research on continuations in Scheme ever since. Then he says that "In Lisp, everything is a list." Wrong again. He also seems to assume that statically-typed programs will always be verbose, indicating that he is unfamiliar with statically-typed languages like Objective Caml which have type inference. The bottom line is that he often doesn't know what he's talking about when he compares the features of different programming languages.

The author's programming expertise appears to be primarily in the domain of small to medium-sized web applications. This is an important domain, but it's far from the only programming domain. So the book's subtitle "A Glimpse At the Future of Programming Languages" should really be "A Glimpse At the Future of Programming Languages for People Who Exclusively Write Small to Medium-Sized Web Applications and Nothing Else". If you want a broader picture, or are primarily interested in another domain, look elsewhere.

The author is clearly aiming the book at programmers who primarily program for money. Not that there's anything wrong with that, but given the author's expressed fondness for open source software (which I share) it might have been nice to acknowledge that some people also write programs for the joy of it in addition to (or in lieu of) the financial rewards.

Now that I've covered the bad, let's look at the good. The author seems to know his domain well, and he makes a good case that Java frameworks for writing web applications have gotten so baroque and tedious that they are simply miserable to use and lead to very slow development cycles. This might be just the fault of the frameworks and not the language, but the author makes a good case that specific language features (or the lack thereof) are what make Java so clumsy for writing web applications. He contrasts Java primarily with the Ruby language, and specifically with the Ruby on Rails web framework, and shows that several features of Ruby make for a much more productive programmer experience. For instance, Ruby's ability to express configuration information inside the language obviates the need for much of the XML that clutters up Java web applications. The author spends a lot of time discussing Ruby's metaprogramming capabilities and how they make it possible to write much more concise code with much less repetition. An example is a class which reads from a database and populates itself on-the-fly with methods to access specific database fields. (Small gripe: the term "metaprogramming" has a variety of semi-related meanings, all coming down to the notion of code being automatically generated instead of being written by hand, but the way that e.g. Lisp or Scheme macros or C++ templates represent metaprogramming is very different from Ruby's metaprogramming, and the author doesn't discuss that.)

The most interesting chapter by far is the chapter on continuation servers. The author gives a clear (though short) description of continuations, which is no mean feat given that they are one of the most difficult-to-explain concepts in the entire universe of programming languages. More importantly, he discusses how they turn out to be extremely useful in web servers, allowing web-based programs to be written in a much more natural and concise style than is possible without them. He uses the Smalltalk language as an example, because there is a continuation server written in Smalltalk. (Another gripe: while I'm happy to see him acknowledge that Smalltalk is a good language, I might have liked it better if he also mentioned that the Scheme dialect of Lisp also has continuations (as I mentioned above), and that there is also a web server in PLT Scheme that is also continuation-based).

The take-home message of the book is that writing small to medium-sized web apps in Java is rapidly becoming more trouble than it's worth, that more dynamic languages like Ruby have features that make it much easier, and that therefore there is a good chance that these languages will take over much of the programmer mindshare. I agree with all of these points, and think that today's Java web programmers are well advised to look at other programming languages. However, I also feel on general principles that all programmers would do well to learn another programming language every year or two, so for me this book just reinforces my pre-existing view.

On the other hand, the author mentions, but doesn't emphasize the fact that the JVM has many capabilities which far outstrip those available in Ruby or Python or indeed most other languages. He does point out that many advanced features of Java frameworks are only useful for very large scale web
applications (the domain that J2EE targets explicitly), and are unnecessary for the majority of (smaller scale) web applications. He leaves open the extremely important question of whether frameworks like Ruby on Rails will scale well to larger applications, or to smaller applications when they need to grow larger, though he seems to believe that they will (if not now, then later when the language and its implementation or implementations mature).

In summary, I think this book is a decent read with a lot of food for thought, if you can handle the irritating style, the mistakes and the omissions. This book is far from the last word on this subject (the reader should be aware that there are lots of interesting computer languages that are not even mentioned in this book, with Objective Caml and E being two obvious candidates), but it's interesting nonetheless.

While probably not the best analogy (on my part), the author's message seems to be similar to the book, The Innovator's Dilemma, that is on the required reading list for MBA students. The "Java Developer's" dilemma: How do you take time away from your comfort zone and what is paying the bills to focus on what isn't ready to pay the bills (the "disruptive" technology/language). I.e., there are more employers looking for Java experience than for, say, Ruby experience. So, when do you make that jump, what language should you jump to, or better yet, how do you prepare yourself to get ready to make that jump without getting left behind.

This last statement is what I took away from the first half of the book. The author gives his opinions and suggestions as to how to prepare yourself for that jump. For that, I think this book should probably be required reading for software developers in general, not just Java developers. I know a handful of ex-mainframe developers that wish they had taken time to learn something other than COBOL. :)

As noted in the other reviews, the book also discusses things such as Java's pros and cons, as well as what the characteristics of the "next big language" are likely to be. While interesting, this aspect of the book was ancillary to the underlying message I took away from the book. What I took away from the book was; when Java is replaced as the programming language of choice, are you going to be ready?

Overall a persuasive argument. If you believe Java is on the way out, this book contains your talking points. If you believe Java is king and your job is safe, this book contains the arguments you will most likely have to fend off.

Bruce Tate's Beyond Java is not up to O'Reilly standards. It's poorly organized, repetitive, and error-ridden. As of the present writing (Jan. 2, 2006), the online example code promised in the front of the book is not available on the O'Reilly web site. A long list of errata is available there, however, but it's incomplete.

Expect much frustration in trying to type in the main Ruby on Rails example and getting it to run successfully. And expect much hair-tearing at all the repetition and the author's attempt to be artsy by beginning each chapter by comparing white-water kayaking to the perils of picking a new programming language.

This would have made a good book or special journal issue at one-quarter it's present size, with better editing, and with better testing of the example code -- and deep-sixing the tedious kayak stories.