I had high hopes purchasing this book, as its title combined with its chapter list promised a good read. I was pretty disappointed at the end, especially since the last chapter is by far the worst.
"The Final Story" chapter made some good points, but those could have been made in much less than thirty pages. The "Immutable Types" chapter should be skipped, as Josh Bloch describes the issue much better in "Effective Java". The "Data Modeling" chapter would make an excellent magazine article (or series even), but is completely out of place in this book.
But there are some good bits, too: the "Nested Classes", "Practical Reflection", "Proxies" and "References in Four Flavors" bring light to poorly understood topics.
The "Tiger: JDK 1.5" chapter is a disaster, because the author's arrogant attitude across strongly as he makes several false statements. It's puzzling, for example, how a reflection guru claims that all of the generic type information is stripped during compilation -- didn't he notice all of the new classes and methods in the java.lang.reflect package for 1.5? Didn't he realize that the compiler needed type signatures for the Collections classes his samples extended?
Then there's the whining, such as his claim that Sun rejected all proposals to implement runtime safety. This complaint ignores all of the other people on the JSR-14 expert group who participated in those decisions, including one of the original authors of the gjc compiler. The Java community has been reviewing the generics proposal since May, 1999, and perhaps the author could have better spent his time participating then.
So take this book with a big grain of salt (perhaps on the rim of a marguerita glass): it makes several good points, but you have to filter out the author's biases much more than other O'Reilly books.
|
