August 2005 Archives

"Use Cases: Requirements in Context", by Daryl Kulak and Eamonn Guiney, professes to present a radical approach to gathering and defining software requirements, employing only Use Cases. Whilst many other writers embrace Use Cases as an important part of a requirements engineering process, here the authors take it all the way, and claim that not only is there no other workable alternative, but that any dilution by including other methods, such as narrative based approaches, is unacceptable. They take no prisoners, stomping all over Alistair Cockburn's notion of Use Case hierarchies (which I've always found a bit clumsy, to be honest), and generally being irreverent about everything else. It is quite entertaining, and presents an interesting thesis. Unfortunately, it is also appallingly edited, inconsistent, not terribly well written, and lacking in detail at crucial points.

For example in section 2.3.4, the text refers to a sample Use Case which has nothing to do with the scenario under discussion, nominally based on this Use Case. However, on the next page, the discussion abruptly switches to something which seems relevant to the Use Case, but is a total non-sequitur in the text. There are many such examples of bad editing throughout the book. A really serious snafu comes on page 85: "NOTE: In the first edition of this book, we advocated a tool called system context level use case … we formally disavow this technique in this new edition". Rather a pity, because in Section 4.4, Deliverables, we're told that the Facade iteration is complete when, amongst other things, a system context use case is documented. Whoops! In section 5.2, we're told that system context use case diagram shows the big picture, and each package's diagram will contain the detail for each set of use cases. Try "Find and Replace" in your word processor, chaps. It can be a surprisingly useful tool.

Another irritating habit found in several places is introducing an interesting sounding concept and then failing to explain it properly. For example, the idea that non-functional requirements can be captured as Use Case stereotypes is, well, interesting … but just one single example might help to convince the more sceptical reader. Equally, the much heralded "hierarchy killer" tool remains a complete mystery to me. It could be an interesting idea, but the author's writing skills do not reach a level where the odd example or diagram would not help a bit.

However, if you are interested in the sharp end of requirements engineering, I do recommend you read the book, with the proviso to not treat it as gospel. It is provocative, sometimes funny, quite often entertaining, but the various errors and inconsistencies are jarring and left me rather disappointed. I did not believe that Use Cases offer a 100% proof solution to requirements definition, and I still don't, but in at least 80% of situations they work well, and this book provides some very valuable insights into applying them effectively.

Publisher: Addison Wesley
Pub Date: July 25, 2003
ISBN: 0-321-15498-3

If I had been writing this stuff a decade or so ago, I suppose it would have had a much greater music related content. But these days what music I listen to is generally a matter between me and my iPod (note, just to make it clear that I am way ahead of the curve, I have a first generation iPod. A very early one. And the battery is just fine). Anyway. Perhaps I should rename this blog Recursive Digressions.

The point of all this is to express my enthusiasm for the new release (what do you call them these days anyway? Albums? CDs? Works? Platters? Downloads? Bitstreams?) from Brian Eno, "Another Day On Earth". I have been a big fan of Brian Eno back since he was just called Eno, and I like pretty much everything he's done. But this, his first truly solo vocal work since 1978, is exceptional. To me it came as a complete surprise, I had no idea it was on the way. Like much Eno stuff it takes a little while to engage, but when it does, it is addictive.

If I had to single out one highlight, it would have to be "Just Another Day". This has it all. Driving, catchy rythym, elegaic lyrics, wonderful ever so slightly detached singing, layer upon layer of sound sculpture, but essentially a classic pop song. This song shows Eno's ability to work at many levels simultaneously. The multi layered detail sets up an atmosphere that would fit happily on Before And After Science, but doesn't sound at all dated. It engages at all levels but never distracts from the song. The arrangements, whilst unique and intricately crafted, never become more important than the song, a trick few of his emulators have ever understood, let alone mastered.

Eno's penchant for playing with dark, disturbingly ambiguous lyrics has not gone away either. Since he does retain his talent for being suggestive rather than specific, interpretation has as much to do with the listener as the artist. This particularly true on the last song, "Bone Bomb", where a passively intoned lyric could be easily interpreted as a meditation by a suicide bomber. Not that terrorism would be a new subject for Eno - he was already there in 1978 with RAF.

Another Day on Earth is poetic, artistic, bursting with ideas, intellect, humour and above all humanity, all packaged up in fabulous songs.