Files

How to pull upstream changes into the vendored cloudpickle

Right now, we will need to update the vendored cloudpickle manually. Here are the steps to do so:

  1. Overwrite the cloudpickle.py and cloudpickle_fast.py from the upstream
  2. Update the version in python/ray/cloudpickle/__init__.py
  3. Run performance benchmarks to validate no regression. (e.g. microbenchmark, scalability tests, etc.)