🔥 💥 Remove kedro_mlflow_line magic and mlflow_client global variable …

…in notebooks (#349)
Galileo-Galilei · Oct 9, 2022 · ec73e7e · ec73e7e
1 parent 4c90a98
commit ec73e7e
Show file tree

Hide file tree

Showing 9 changed files with 18 additions and 73 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -4,7 +4,11 @@
 
 ### Fixed
 
--   :bug: `MlflowArtifactDataSet.load()` now correctly loads the artifact when both `artifact_path` and `run_id` arguments are specified. Previous fix in ``0.11.4`` did not work because when the file already exist locally, mlflow did not download it again so tests were incorrectly passing.   ([#362](https://github.com/Galileo-Galilei/kedro-mlflow/issues/362))
+-   :bug: `MlflowArtifactDataSet.load()` now correctly loads the artifact when both `artifact_path` and `run_id` arguments are specified. Previous fix in ``0.11.4`` did not work because when the file already exist locally, mlflow did not download it again so tests were incorrectly passing ([#362](https://github.com/Galileo-Galilei/kedro-mlflow/issues/362))
+
+### Removed
+
+-   :fire: :boom: Remove ``reload_kedro_mlflow`` line magic for notebook because kedro will deprecate the entrypoint in 0.18.3. It is still possible to access the mlflow client associated to the configuration in a notebook with ``context.mlflow.server._mlflow_client`` ([#349](https://github.com/Galileo-Galilei/kedro-mlflow/issues/349)). This is not considered as a breaking change since apparently no one uses it according to a [discussion with kedro's team](https://github.com/kedro-org/kedro/issues/878#issuecomment-1226545251)
 
 ## [0.11.4] - 2022-10-04
 

diff --git a/docs/source/06_interactive_use/01_notebook_use.md b/docs/source/06_interactive_use/01_notebook_use.md
@@ -1,7 +1,7 @@
 # How to use `kedro-mlflow` in a notebook
 
 ```{important}
-You need to call ``pip install kedro_mlflow[extras]`` to access notebook functionalities.
+You need to install ``ipython`` to access notebook functionalities.
 ```
 
 ## Reminder on mlflow's limitations with interactive use
@@ -22,24 +22,26 @@ Open your notebook / ipython session with the Kedro CLI:
 kedro jupyter notebook
 ```
 
-Kedro [creates a bunch of global variables](https://kedro.readthedocs.io/en/latest/tools_integration/ipython.html#kedro-and-jupyter), including a `session` which are automatically accessible. It also registers some line_magic from plugins, including `%=reload_kedro_mlflow` from `kedro-mlflow`.
 
-In your first cell, launch:
-```
-%reload_kedro_mlflow
+Or if you are on JupyterLab,
+
+```notebook
+%load_ext kedro.ipython
 ```
 
-This automatically:
-- load and setup (create the tracking uri, export credentials...) the mlflow configuration of your `mlflow.yml`
+Kedro [creates a bunch of global variables](https://kedro.readthedocs.io/en/latest/tools_integration/ipython.html#kedro-and-jupyter), including a `session`, a ``context`` and a ``catalog`` which are automatically accessible.
+
+When the context was created, ``kedro-mlflow`` automatically:
+- loaded and setup (create the tracking uri, export credentials...) the mlflow configuration of your `mlflow.yml`
 - import ``mlflow`` which is now accessible in your notebook
-- Create a `mlflow_client` object with your mlflow server settings, which is now accessible in your notebook
 
-If you change your ``mlflow.yml``, re-execute this cell for the changes to take effect.
+If you change your ``mlflow.yml``, reload the kedro extension for the changes to take effect.
 
 ## Difference with running through the CLI
 
 - The DataSets `load` and `save` methods works as usual. You can call `catalog.save("my_artifact_dataset", data)` inside a cell, and your data will be logged in mlflow properly (assuming "my_artifact_dataset" is a `kedro_mlflow.io.MlflowArtifactDataSet`).
-- The `hooks` which setup configuration are only accessible if you run the session interactive, e.g.:
+- The `hooks` which automatically save all parameters/metrics/artifacts in mlflow will work if you run the session interactively, e.g.:
+
 ```python
 session.run(
     pipeline_name="my_ml_pipeline",
@@ -49,6 +51,7 @@ session.run(
 )
 ```
 but it is not very likely in a notebook.
+- if you need to interact manually with the mlflow server, you can use ``context.mlflow.server._mlflow_client``.
 
 ## Guidelines and best practices suggestions
 

diff --git a/kedro_mlflow/extras/__init__.py b/kedro_mlflow/extras/__init__.py
diff --git a/kedro_mlflow/extras/extensions/__init__.py b/kedro_mlflow/extras/extensions/__init__.py
diff --git a/kedro_mlflow/extras/extensions/ipython.py b/kedro_mlflow/extras/extensions/ipython.py
diff --git a/setup.py b/setup.py
@@ -59,22 +59,15 @@ def _parse_requirements(path, encoding="utf-8"):
             "pre-commit>=2.0.0,<3.0.0",
             "jupyter>=1.0.0,<2.0.0",
         ],
-        "extras": ["notebook>=6.0.0"],
     },
     author="Yolan Honoré-Rougé",
     entry_points={
         "kedro.project_commands": [
             "kedro_mlflow =  kedro_mlflow.framework.cli.cli:commands"
         ],
-        # "kedro.global_commands": [
-        #     "kedro_mlflow =  kedro_mlflow.framework.cli.cli:commands"
-        # ],
         "kedro.hooks": [
             "mlflow_hook = kedro_mlflow.framework.hooks.mlflow_hook:mlflow_hook",
         ],
-        "kedro.line_magic": [
-            "line_magic = kedro_mlflow.extras.extensions.ipython:reload_kedro_mlflow"
-        ],
     },
     zip_safe=False,
     keywords="kedro-plugin, mlflow, model versioning, model packaging, pipelines, machine learning, data pipelines, data science, data engineering",

diff --git a/tests/extras/__init__.py b/tests/extras/__init__.py
diff --git a/tests/extras/extensions/__init__.py b/tests/extras/extensions/__init__.py
diff --git a/tests/extras/extensions/test_ipython.py b/tests/extras/extensions/test_ipython.py