Designing software is a tricky thing. It’s tempting to front-load it on a project. That won’t work because the start of a project is when you know the least about it. So some folks try to do as little design as possible. I’m guilty of this. However, that can lead to software that doesn’t adequately express the problem it’s trying to solve. Further, there is often a temptation to over-design software with lots of ceremony and architecture. Contrary to this is the temptation to not design it at all, which again leads to software that doesn’t express itself.
There’s a book that draws a reasonable compromise between these forces. I’ve been meaning to read Eric Evans’ _Domain Driven Design_ for a while now. The emphasis of the book is in collaborating with domain experts and other developers to find the essence of the problem space and then express that in software (as objects). I’ve often pointed out the utility of building applications from the language up and the problem domain down. DDD focuses precisely on the latter.
One of the core concepts in the book is the _ubiquitous language_ that is used to describe the problem at hand. This language is used by the domain experts (customers) *and* the developers. The language is then woven into the design of the system. This leads to software that is more likely to succeed, both in business terms and in terms of development effort. Evans spends the first part of the book describing the particulars of this language.
He then moves on to describing the technical side of the software. Entities, value objects, services, factories, modules and repositories are terms I was already somewhat familiar with that Evans gave a more crisp and satisfying definition to. For most people, this is probably the tasty meat of the book, illuminating the way from a competent developer to an outstanding developer.
The last part of the book focuses on the larger scale issues of deep design. I was particularly pleased that he covers how software design is affected by various good and bad social issues. It also gives a strategic view of the forest, where most books on software development focus on a more tactical view of each tree.
I’m fond of pointing out books that are inflection points in my way of thinking about software development. Code Complete, The Pragmatic Programmer, The Dragon Book and My Job Went To India all fall under this category. Domain Driven Design is certainly the latest edition. It makes sense of trends I see in great software and illuminates a path to make software like it myself.
If reading this review didn’t make you want to vomit, you should probably read the book posthaste.