Search in Book Reviews

Some people find that they learn most easily by example. You know who they are, they are the ones who when reading a book on programming want to dissect every source code example line by line. They spend time trying to determine exactly why the author did it that way instead of some other way. Do I sound disparaging? Well yes because proponents of this style of learning too often forget that sometimes a decision has been made simply because, like which side of the corridor you walk on, there was more than one equally good way of doing it. Some authors really get up the noses of such readers by doing the same thing differently each time they need to. I have seen people tearing their hair out trying to decide why the author did it differently the second time.

All this is a bit beside the point but is by way of setting the scene for this book. Most books on analysis and design methodologies start off with highly theoretical introductions to the subject. If you have tried this approach and you are still struggling (or you have never tried it because you could never get past the first page) then this is the book for you. The book is substantially five case studies presented in a laid back style that should be just what you need if you are one who learns by example.

Now let me pick a few holes (well you expect me to don't you). I am unhappy with the inclusion of 'Patterns' in the title because while some are there to be abstracted the authors only do so in and appendix where the information is presented in a, to me, unhelpful format. All the quality of the earlier text goes out the window and we are left with a sub-minimalist presentation. His names for patterns are frequently at variance with those provided by others. I am reminded that I once sat through an extremely boring presentation from Coad on patterns. Not only was it confusing but his interaction with his audience was superficial. The only relief was that I was sitting with Andy Koenig and Jim Coplien. The latter seeing my confusion told me not to worry because he was confused as well. (For those that do not know, Jim Coplien is one of the founders of the Pattern movement.)

This is only a criticism of the title and the appendix because I think that the main text is well suited to its target readership but please do not buy this book if you want one describing conventionally named pattern abstractions.

The next caveat is the cover description 'results in COAD, OMT and Unified.' The book fundamentally uses COAD diagrams while providing the other two as alternatives for summaries etc. It's as if the lead author cannot quite abandon his methodology and its diagrams. I sympathise but at the same time he has done his readers no favours by prevaricating.

If you like to abstract a methodology from case studies with guidance then this book is for you.