MlflowDataSet fails to log on remote storage when underlying dataset filepath is converted as a PurePosixPath

[Galileo-Galilei]

When I register a dataset in the catalog.yml

my_dataset:
  type : kedro_mlflow.io.MlflowDataSet 
  data_set : 
    type: pickle.PickleDataSet
    filepath: data/02_intermediate/my_dataset.pkl

and I run kedro run I got a expected string or bytes-like object when the local path is linux AND the mlflow_tracking_uri is an Azure blob storage (it works locally). I don't know really why this append, but it can be fied by replacing self._filepath by self._filepath.as_posix() in these 2 locations:

kedro-mlflow/kedro_mlflow/io/mlflow_dataset.py

Line 51 in 94bae3d

local_path=local_path,

kedro-mlflow/kedro_mlflow/io/mlflow_dataset.py

Line 55 in 94bae3d

mlflow.log_artifact(local_path, self.artifact_path)

@kaemo @akruszewski did you experience some issues with S3 too?

EDIT: @akruszewski it is the very same issue you encountered here