Troubleshooting

TypeError: inject_from_ctx… is not found in the Context

Cause: A step asks for a parameter that is not provided via treatments, datasource, or defaults.

Solution: Add the value to config.yaml or set a default in the step signature.

treatments:
  tuned:
    factor: 1.1

@pipeline_step()
def scale(data, ctx: FrozenContext, factor: float = 1.0):
    return [x * factor for x in data]

RuntimeError: An asyncio event loop is already running

Cause: Calling exp.run() inside Jupyter/async environments.

Solution: Use the async entrypoint:

import asyncio

if asyncio.get_running_loop():
    result = await exp.arun()
else:
    result = exp.run()

Warning: Using SeedPlugin with executor_type=‘thread’ is not reproducible

Cause: Thread pools share RNG state; seeds can interleave across threads.

Solution: Prefer processes for reproducible runs:

plugins=[SeedPlugin(), ParallelExecution(executor_type="process")]

ImportError: No module named ‘crystallize_extras…’

Cause: Extras plugins/steps are referenced but the package is not installed.

Solution: Install the extras bundle (or a specific extra):

pip install --upgrade --pre crystallize-extras[all]