feat(tests): Tests for utils #29
Conversation
|
I like the spirit of this, but ultimately I think most of the underlying code here is still undergoing such rapid change that this is unlikely to be worth it. I'm about to break like 20 of these tests doing the granular permissions stuff. It's also a little silly to be testing these utils when there is still a huge amount of non-determinism in the agents themselves, and a big range of performance across different agents for different tasks. When we make an effort towards better testing & stability, it will likely look very different from this—end-to-end agent testing and evals. But even that is a little down the road. I think a lot of this stuff (the pullfrog footer, etc) never needs to be tested at all. I'm also biased but I have an aversion to mock-heavy tests like this. Ultimately I think we should focus on other things right now (e.g. testing all the agents end-to-end and feature development). Since this is still a draft, I'm just gonna close for now. We can still use this later as a starting point for future testing work but it's not a high enough priority right now. |
Following #24