In Favour of Test Driven Development
joelhughes.co.uk
joelhughes.co.uk
Carefully consider the alternatives. No testing? Manual testing? Automated UI testing? Test second?
On the softer side (and more difficult to measure) I ask development teams how test first affects their thinking. Do you converge on users' requirements more quickly or slowly?
Finally, recognize that programming is a hard skill to master. TDD is another skill on top of that one.
Bottom line is that to deliver great software you need a solid conceptual understanding of what it's supposed to do (i.e. good requirements analysis) and smart people to design and implement the software. By its nature, software development will never be reduced to assembly-line work that any warm body can do.
I don't think of TDD and Agile as being fads, though, but the best we have to date in an evolving process. It'll go on for decades yet.
That last paragraph, though, is killer: Give great input to very smart people and it might result in a great result. Note the absence of ego.
And people, never forget that all software spends 99.999% of its life in maintenance. Be prepared.
1) TDD is not useful for interacting with existing APIs (probably the majority of programming today)
2) TDD works when it works and doesn't when it doesn't (which is rather tautological).
I'm not sure that's in favour of TDD.