CJKV Information Processing [Paperback]

Thank goodness he has. The author, Ken Lunde, has an encyclopedic knowledge of this material. In addition, he is one of those people to whom anything less than strict, literal correctness is intolerable. Authors of this sort usually write in a style reminiscent of the federal tax code. Lunde manages to avoid this, creating one of those rare and delightful computer books that serve as a lucid tutorial the first time through, and as a strict and comprehensive reference thereafter.

The principal reason I consider this book the bible for Asian language information processing is the extreme difficulty of getting most of this information via any other source. In fact, I'd be hard pressed to think of another computer book whose original source material is as scattered, poorly documented, and often unreliable as that Lunde had to gather to produce this book. His job of ferreting out the details, cross-checking, error correction, and organization into a single book makes this almost a work of journalism. If you do CJKV work, you'll need more than just this book, of course, but the book is full of references to other material, so this is the place to start.

Lunde also provides a lot of usable source code in the book. This is not unusual in a computer book, but this code is special in two ways. First, it's available in C, Java, and Perl, not just in C. This is refreshing, given the increasingly prominent roles played by Java and Perl on the Internet--the place where multilingual computing arguably matters most.

Second, his code serves as a great checklist for what has to be done by any similar code. This is one of those difficult types of programming where many bugs aren't easy to see, because of the large number of obscure "gotchas" and arcane details. Lunde doesn't miss much, and he revels in these arcane details. His code is not highly optimized, and he admits as much, but if his code does something, you need to do it, and if it doesn't, you (probably) don't need to, either. This fact alone justifies the cost of the book for any developer to whom bugs might have financial consequences.

If you're going to do CJKV work, this is the bible. As I said, it's not the only thing you'll need, but it's where you should start.

Lunde's explanation of the structure and history of Asian written languages is fascinating reading in its own right.

If developing on the Windows platform, I would also recommend "Developing International Software for Windows 95 and Windows NT" by Nadine Kano (Microsoft Press). Lunde's book contains crucial background information regarding Asian character sets, as well as some general algorithms; the Kano book focuses on implementation details specific to the Windows environment.

The book is intended primarily for software engineers, but the subject matter is treated so comprehensively that it is an essential desk reference for translators, information developers, project managers, production managers, and marketing executives.

Just get it, Ok?