Post #1722057
2026-04-27 02:01 UTC
@phillmv @jneen Personally, not using LLMs in my experience, I prefer writing the implementation, then writing the automated tests to pass, and then check what I need to do to the automated test to cause it to fail. That way, I can confirm it passes for the right reasons, rather than I managed to make the code pass the test.
Replies (1)
-
@brib@bribstodon.xyz 2026-04-27 07:28
@AT1ST @phillmv @jneen That's my approach too. I don't know if it's the best approach; for simple stuff I can write the tests first but for more significant programs I don't necessarily know how to structure the tests without the program in front of me. Having said that, maybe the test-first approach will help keep my software not becoming a ball of confusion.....