Databricks

lawrence009 · ‎11-27-2022

I'm getting this message with the following code:

from databricks import feature_store
 
fs = feature_store.FeatureStoreClient()
 
fs.create_table(
    name='feature_store.user_login',
    primary_keys=['user_id'],
    df=df_x,
    description='user login features'
)

and I couldn't figure out why from the warning message below:

INFO databricks.feature_store._compute_client._compute_client: Created feature table 'hive_metastore.feature_store.user_login'.
 
/databricks/.python_edge_libs/databricks/feature_store/entities/_feature_store_object.py:8: FutureWarning: ``databricks.feature_store.entities.feature_table.FeatureTable.keys`` is deprecated since v0.3.6. This method will be removed in a future release. Use ``FeatureTable.primary_keys`` instead.
 
  yield prop, self.__getattribute__(prop)
 
Out[3]: <FeatureTable: keys=['user_id'], tags={}>

My runtime is 11.3 LTS ML (includes Apache Spark 3.3.0, Scala 2.12)

DavideAnghileri · ‎11-28-2022

Hi @Lawrence Chen,

It looks like a Databricks bug. As you can see the warning comes from a statement in the private `_feature_store_object.py` file. I don't think there is much you can do about it if not ignoring it and wait the fix. It seems like if they are still searching also for the old attribute.