Have a single place to define all RPCs, and if requested to run as a Celery task, fall back to a regular RPC if Celery isn't available. To do in conjunction with Klein/Joblib/Dask in case this solves the parallelization problem.
Have a single place to define all RPCs, and if requested to run as a Celery task, fall back to a regular RPC if Celery isn't available.
To do in conjunction with Klein/Joblib/Dask in case this solves the parallelization problem.