kedro-mlflow is broken with kedro==0.18.1

[Galileo-Galilei]

Description

The plugin does not work with projects created with kedro==0.18.1

Context

Try to launch kedro run in a project with kedro==0.18.1 and kedro-mlflow installed.

Steps to Reproduce

conda create -n temp python=3.8 -y
conda activate temp
pip install kedro==0.18.1 kedro-mlflow==0.9.0
kedro new --starter=pandas-iris
cd pandas-iris
kedro mlflow init
kedro run

Expected Result

This should run the pipeleine and log the parameters.

Actual Result

This raises the following error:

AttributeError: module 'kedro.framework.session.session' has no attribute '_active_session'

Your Environment

Include as many relevant details about the environment in which you experienced the bug:

• kedro and kedro-mlflow version used (pip show kedro and pip show kedro-mlflow): kedro==0.18.1 and kedro-mlflow<=0.9.0
• Python version used (python -V): All
• Operating system and version: All

Does the bug also happen with the last version on master?

Yes

Solution

Currently, kedro-mlflow uses the private _active_session global variable to access the configuration inside a hook.

With kedro==0.18.1, this private attribute was removed and the new recommandation is to use the after_context_created hook.

Retrieving the configuration and set it up should be moved to this new hook:

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

Lines 108 to 109 in 963c338

	mlflow_config = get_mlflow_config()
	mlflow_config.setup()

[Galileo-Galilei]

Closed by #313