Make kedro-mlflow not require an experiment name with packaged projects?

[astrojuanlu]

I was running a .whl containing my Kedro project, and found this error coming from the MLflow hook:

...
  File "/usr/local/airflow/dags/kedro_spaceflights_mlops_data_science_dag.py", line 39, in execute
    session.run(self.pipeline_name, node_names=self.node_name)
...
  File "/usr/local/lib/python3.12/site-packages/pluggy/_callers.py", line 121, in _multicall
    res = hook_impl.function(*args)
          ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/kedro_mlflow/framework/hooks/mlflow_hook.py", line 129, in after_context_created
    metadata = _get_project_metadata(context.project_path)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/kedro/framework/startup.py", line 59, in _get_project_metadata
    raise RuntimeError(
RuntimeError: Could not find the project configuration file 'pyproject.toml' in ...

This is the offending code:

kedro-mlflow/kedro_mlflow/framework/hooks/mlflow_hook.py

Lines 120 to 130 in 7ec6610

	if (
	conf_mlflow_yml.get("tracking", {}).get("experiment", {}).get("name")
	is None
	):
	# the only default which is changed
	# is to use the package_name as the experiment name
	experiment_name = context._package_name
	if experiment_name is None:
	# context._package_name may be None if the session is created interactively
	metadata = _get_project_metadata(context.project_path)
	experiment_name = metadata.package_name

Looks like the project metadata is only used to derive the experiment name. I'm wondering if that could be lifted?

[Galileo-Galilei]

Thanks for the report, I'll definitely should use a default experiment in this case.