I’m curious what stance others have taken? Or what they’ve experienced with this? I’m toying with adding a rule to our values.md that the LLM can only significantly generate one or the other. I’m reviewing a merge request at the moment where it’s clear the LLM mostly did both, and they’re highly tautological. And I can see how it over engineered the code, just to be able to write more tests almost.
You must log in or register to comment.
My opinion is that tests too are code. I don’t see a problem letting agents write or modify code. But there must be a sufficient set of tests verified correct to ensure conformance to all essential requirements. If an agent writes an essential test that cannot be understood and confirmed correct by the reviewer, with reasonable effort, then the test should be revised until it can be.
If you can’t be bothered to write code yourself, don’t bother fooling yourself into thinking you’re writing code when it’s actually just slop. That goes for tests too.