Against TDD

neopragma.com

For as much as I've heard about TDD over the years: in my education, articles, and mentions of it in industry -- I have yet to see anybody I know use it.

You can't design your entire interface up front. Too much handwaving. You'll start coding, and realize you aren't testing the right things, until after you write the code, which defeats the purpose of writing the tests first.

The article says many people don't use TDD correctly, or they try to use it in the wrong situations. If it was more broadly applicable, I think it's safe to say I would've encountered a situation where it was useful. If it's the case where there's a limited set of situations where it is applicable, the topic doesn't deserve as much mindshare as it receives.

I've used it for almost 2 decades now. The general rule is:

Don't write tests until you're sure about your design and interface. If you're doing exploratory coding, you write tests later. If you're adding features to something that already exists, you're more likely to write tests first.

The end target is roughly the same: 80-90% test coverage. But you'll waste a lot less time throwing test code away.

Reply on news.ycombinator.com