I fear this is the trap that most "new" developers will fall into over the next few years. I'm also worried the "great programmer" will cease to exist as the current greats retire, and the potential greats will never reach that level due to their reliance on LLMs.
Its a great time to be a non-AI user, and even better to have never been one, because its easier now than ever to differentiate oneself from those who are reliant on it and, over the long run, much less effective because of it.
MoreQARespect•1w ago
No, not "some function". The tests should mimic the expected behavior of a scenario using the app.
Only if the function is acting as a close approximation of the app itself do you get any benefits out of TDDing it.
90% of the shitty tests I've ever seen in my life - whether written with TDD or without - have been written as a result of a mid level developer thinking that the test should "surround some function" (they dont seem to think it matters which ones you pick) rather than reflecting an actual user story.
This blog post is a reflection of that same attitude.