This is a collection of papers, which were presented two years ago at the first Component Users' Conference. Such collections rarely offer much in the way of coherence or structure and being the first such conference, this one is worse than most. The subject is too new for there to be any agreement about what the best approach should be, or what technology offers the best hope of success.
More on the mark is the first of two papers by the pairing of Cuno Pfister and Clemens Szyperski, which actually deals with the crux of Component-Based SE and what differentiates it from a simple OO approach. The second paper is more of a technical sell of their product, Oberon/F. (Szyperski's 'Component Software: Beyond Object-Oriented Programming' is reckoned worth looking at, but I haven't got my hands on that yet). Generally, the papers are too short to have any depth; there are 18 papers in 152 pages. Since there is some introductory material in each one, such as what CORBA is or what OO offers, there is actually very little meat at all. Depending on your interests, you are likely to find no more than a third of it worth reading, perhaps 50 pages at most, for the thirty pounds it costs. Given that it is all two years old as well (and consequently some of the URLs quoted are broken), not a very attractive package.
However, there is a very worthwhile paper in it; the first one, by David Smyth of NASA Jet Propulsion Laboratory. He manages to be both witty and insightful in a short (5-page) paper. It is a shame this one paper isn't available in some cheaper form. In summary, don't bother with this book.