Hacker News Reader

Comment by danhou | Hacker News Reader

Everything made sense to me except for the part about writing tests.

If you're confident you're product will have a long lifespan then tests make perfect sense. But from what I've seen most MVP's are experiments that may or may not stand up to real users.

If you're in the latter camp, writing test code is prematurely optimizing for stability.

View on HN

> If you're in the latter camp, writing test code is prematurely optimizing for stability.

Not sure test have something to do with stability. Writing test code will help document the code for the next developer who touches the hair ball and will force better definition of requisites in the MVP phase.

Totally agree, but I thought I wrote that ;) "If you know that you will start from scratch soon you can most likely skip the tests and focus on testing your hypotheses as quickly as possible."

Yeesh, just went back and re-read that. I fail at reading.

The "write tests" part could be rewritten as: don't let the freelancer dictate the rough technical direction. If you want a proper state-of-the-art software project, tests should be included.

Only if you write unnecessary tests: e.g. for CRUD code, or for front-end UI. But when you have complex procedures with clearly defined outcomes, writing tests is never premature.

It doesn't need to make sense. Author is obviously a devout monk in the church of TDD. The view is purely religion.

Reply on news.ycombinator.com