Advanced CORBA® Programming with C++ [Kindle Edition]

Most ORBs don't have the POA yet; they will by mid-2000. It was an important and correct decision by the authors to include it instead of the BOA, most of the details of which are vendor-specific anyway (that's why the POA exists). Note to BOA lovers: sorry, the BOA is no longer part of the spec. ORB vendors are free to keep it around, and many do, but it's only because they're nice. The POA is so vastly superior that the only reason you would use the BOA is if you have an existing code base to maintain.

This is not a book for beginners, grazers, or wanna-bes; it is a book for serious, working practitioners, and it works best as a reference (although you can read it cover to cover). It is also C++-specific; it is the best CORBA book available for any language, but programmers who don't know the CORBA interface language mappings in both C++ and another language will not know which parts are C++-specific and may be confused or frustrated. The CORBA C++ mapping is by far the most complex, so it makes sense to do this one, but be forewarned that the early chapters on the C++ language mapping will be of marginal use if you are not a C++ programmer. The POA section makes the POA sound harder than it is for most applications, and does not provide complete detail on POA policies and architecture, but it is close. Your alternative is to read the CORBA spec (a horrifying prospect for most people), or your ORB vendor's documentation (sometimes adequate, sometimes not).

Flaws? Yes. Alternatives? No. If you are a CORBA programmer and you don't like reading the CORBA spec, BUY THIS BOOK.

The content is very close to the current state of the art from the OMG. In some areas it even prempts proposals (eg some CORBA services) that have not yet been approved; this is both a potential trap for the unwary and useful information for those who need to be aware of what is probably coming over the horizon. The coverage of more established CORBA services (naming, trader, and event services) is a useful reference.

This book only considers the POA, and ignores the superseded BOA. That limits the value of this book for those charged with maintaining older systems, particularly those with servers implemented using the older BOA.

Organisation of the book could be improved. There were a number of instances of forward referencing in earlier chapters which makes it difficult to extract information. Some sections merged a number of bits of information together, and became a bit confusing. This disjoint information flow reflects the nature of the formal OMG offering, but it would have been nice to see the material organised more clearly than that! After all, most who buy this book will have been exposed to some material from the OMG, and found it tough going.

Despite the rather critical review above, I have found this book to be an extremely valuable resource on a complex topic. It is also, at this time, the best available in its class. My copy is already well thumbed, and will become more so.