Brian Ward

We need to work on this issue to make CI less of a hassle, but I’m not convinced fixing a seed is the approach we should take

Yes, the tests for e.g. HMC have tolerances that mean they still fail on some runs. Our current test suite runs in ~12 seconds, so we have a lot head...

I recently stumbled on https://github.com/pytest-dev/pytest-rerunfailures, which could be another approach we use (maybe only as a stopgap)

We could try to separate those errors at the python level (e.g., we could have it so `IndexError` and others are not caught, but `ArithmeticError` and `RuntimeError` are). There is...

For test coverage, I would look here: https://app.codecov.io/gh/flatironinstitute/bayes-kit/blob/affine-invariant-walker-v1/bayes_kit/ensemble.py That will get updated every time CI is run on this branch. Agreed it’s good to finish things. I’m a bit hesitant...

I’m not very picky about fixing the type errors _before_ anything is merged - in particular, we already have versions merged to `main` which are both wrong and have type...

Would this be resolved by providing an environment variable to change the location, a-la `BISECT_FILE` for ppx_bisect?

BISECT_FILE is actually a file prefix, I believe for similar reasons of avoiding overwriting itself. I may take a pass at doing something like this in my free time and...

Moved existing work to https://github.com/WardBrian/scikit-stan/tree/feature/8-wiener

Moved existing work to https://github.com/WardBrian/scikit-stan/tree/feature/5-gamm