IPv6 Network Programming [ペーパーバック]

I received this book today, and within three minutes of looking it over, had decided to return it. The book contains 361 pages, but 281 of those pages are RFCs and other documents freely available online. The actual original content of the book is all of 80 pages.

From a quick scan, the content looks like it's reasonably detailed and useful, and contains a fair bit of source code, which can certainly be helpful. The code was written (and presumably tested) using BSD, but there is some mention of other operating systems. However, I found no mention of AIX or HP-UX, and the numerous flavours of Linux are treated as a single operating system. Does IPv6 behave exactly the same on SuSE, RedHat, Ubuntu, Knoppix, Slackware, and all the rest? I don't know, but I would have expected this book to tell me, at least about the major ones.

I'm giving this book only 2 stars because almost 80% of the book was not written by the author, and is available for free online. I expect a lot more than 80 original pages from a $50 book.

IPv6 is taking far longer than expected to supplant IPv4. But this book can be taken as a useful sign of progress. Presumably you are familiar with socket programming under v4. This has remained mostly unchanged since the early 90s, and there are several worthy texts explaining these. (Like by Stevens and Rago, or Comer and Stevens.)

Hagino takes us on the next step. Showing what you need to know to enable sockets under v6. Yes, there are certainly complications compared to v4. But the book shows that the basic sockets approach of v4 still gives you useful experience that carries over.

I wholeheartedly agree with the other reviewer who was complaining about this book only having 80 pages of "new" information (the other 270 plus-or-minus pages being consumed by Appendices containing publicly available documents). I would also add that the first chapter is your basic introduction that most everyone who buys this book is already going to know, and the last Chapter is devoted to "A Practical Example" that I failed to really see a need for (though some might find it nice); so in reality, of the 80 pages of real text, only 46 are really worth anything. From that perspective, charging $50 for this thing is definitely a rip-off. By the way, has anyone else noticed that the pages are twice as thick as most other books? Normally I'd like that, but in this case I think it's just a meager attempt to "beef up" the book and make it look like it has more information than it actually does.

However, with all that said, the bottom line is that if you are doing professional work, trying to port your client-server applications into an IPv6 environment, this book is a necessity. It contains information that is not available in 'man' pages (or at least I was unable to find it, which is why I purchased the book); and it also contains information that is "hidden" in the 'man' pages, that you'd really have to search for in obscure sections to find (the voice of experience). If all the information I need can be squished into 46 pages instead of 300, that's just fine with me (time is precious)! I just wish the unnecessary bulk would have been omitted.

The book is very poorly written. I felt like I was reading a set of handwritten notes, or comments in source code, rather than a book. The sentences are very choppy, not flowing easily, etc. There are many typos in the source code examples, too. Clearly the proofreader(s) were asleep on the job.

And since the author seems to be somewhat critical of other programmers who take lazy shortcuts, with which I agree by the way, I feel it only fair to critisize him for not following his own advice. There is a difference between a "protocol family" and an "address family"; so even though, for example, PF_INET and AF_INET may translate to the same literal value, they are two different things and used in different situations. The author totally ignores the PF_xxx symbols in his book, substituting the corresponding AF_xxx symbol instead. He misuses gai_error(3) in some examples, and uses obsolete functions in others. So do _not_ follow his examples verbatum if you want to produce correct code; use his ideas, but think for yourself.

With the possible exception of one other book, this is the only book I could find on IPv6 _PROGRAMMING_. All the others seem to deal with the protocol itself (header formats, etc), or administration of an IPv6 network. If you actually need to program an application to run on IPv6, this is the book you need... well, for the most part. I'm still having a problem with an embedded system--and only on the server side--and whatever's causing it is obviously not addressed in this book. :-) For normal systems, though, this book contains all the socket API information you need.

This book contains no information on configuring an IPv6 network. For example, how to set up DNS services to translate hostnames into IPv6 addresses, how to use 'ifconfig' to enable an IPv6 network interface, etc. Nor does it talk about other ancillary items necessary for programs to work correctly. It is solely concerned with the socket API based on RFC's. I find this to be a serious drawback.