All the current tests are at the unit level. We need some more substantial tests before I can go ahead and start adding features/restructuring. I've begun working on this here.
These tests will run the main commands (learn, predict etc.), confirm the outputs exist and test against some precomputed outputs.
I'm currently doing this against the SirSam dataset - the test data lives here. It's around 100mb so if we can minimise this in future that's welcome, but otherwise I think it's important to have some small test data in the repo.
All the current tests are at the unit level. We need some more substantial tests before I can go ahead and start adding features/restructuring. I've begun working on this here.
These tests will run the main commands (learn, predict etc.), confirm the outputs exist and test against some precomputed outputs.
I'm currently doing this against the SirSam dataset - the test data lives here. It's around 100mb so if we can minimise this in future that's welcome, but otherwise I think it's important to have some small test data in the repo.