René's URL Explorer Experiment

Title: fix: Use SELECT * when feature_name_columns is empty in pull_all_from_table_or_query by abhijeet-dhumal · Pull Request #6311 · feast-dev/feast · GitHub

Open Graph Title: fix: Use SELECT * when feature_name_columns is empty in pull_all_from_table_or_query by abhijeet-dhumal · Pull Request #6311 · feast-dev/feast

X Title: fix: Use SELECT * when feature_name_columns is empty in pull_all_from_table_or_query by abhijeet-dhumal · Pull Request #6311 · feast-dev/feast

Description: What pull_all_from_table_or_query always builds an explicit SELECT projection from join_key_columns + feature_name_columns + timestamp_fields. When feature_name_columns=[] — the "read all source columns" signal already used by FeatureBuilder.get_column_info for ray and pandas transformation modes — the generated SQL becomes: SELECT user_id, event_timestamp FROM s3a://bucket/reviews/ WHERE event_timestamp BETWEEN ... All raw feature columns (rating, text, helpful_vote, etc.) are silently dropped. The UDF receives a 2-column DataFrame and every aggregation returns null or fails. Why it happens # no guard for empty feature_name_columns (fields_with_aliases, aliases) = _get_fields_with_aliases( fields=join_key_columns + feature_name_columns + timestamp_fields, ... ) fields_with_alias_string = ", ".join(fields_with_aliases) # → "user_id, event_timestamp" when feature_name_columns=[]

Open Graph Description: What pull_all_from_table_or_query always builds an explicit SELECT projection from join_key_columns + feature_name_columns + timestamp_fields. When feature_name_columns=[] — the "read all sour...

X Description: What pull_all_from_table_or_query always builds an explicit SELECT projection from join_key_columns + feature_name_columns + timestamp_fields. When feature_name_columns=[] — the "read all ...

Opengraph URL: https://github.com/feast-dev/feast/pull/6311

X: @github

direct link

Domain: github.com

Links:

Viewport: width=device-width