pyspark dependency breaks Snowflake offline store

[beauchbum]

This commit introduced a pyspark dependency for the Snowflake offline store that is not reflected in setup.py. This causes feast actions to fail due to:

File "/usr/local/lib/python3.9/site-packages/feast/infra/offline_stores/snowflake.py", line 26, in <module>
  from pyspark.sql import DataFrame, SparkSession
ModuleNotFoundError: No module named 'pyspark'`.

Steps to reproduce

• Install feast = {extras = ["snowflake"], version = "^0.27.0"}
• Set up a Feature Repository with a Snowflake offline store
• Run feast plan

Specifications

• Version: 0.27.0
• Platform:
• Subsystem:

Possible Solution

Not sure - would prefer not needing pyspark for Snowflake offline store to work, but if that's the case then add to setup.py?

[achals]

cc @adchia @sfc-gh-madkins - should this dep on pyspark be optional (by moving the import inside the to_spark_df method, or should it be added as a proper dep to the snowflake extra?

[adchia]

yeah should be a try import probably

[beauchbum]

@adchia Thanks for weighing in on this. Any ideas on when this might be worked on (and eventually released)?