Context
PR #5827 added dbt integration with single entity support.
Problem
Real-world feature engineering often requires multiple entities per FeatureView (e.g., a feature view keyed by both user_id and merchant_id). Currently only single entity column is supported.
Current Limitation
feast dbt import -m manifest.json -e driver_id # only one entity
Proposed Enhancement
Support multiple entity columns:
# Option 1: Multiple -e flags
feast dbt import -m manifest.json -e user_id -e merchant_id
# Option 2: Comma-separated
feast dbt import -m manifest.json -e user_id,merchant_id
Implementation Notes
- Update CLI to accept multiple entity columns
- Create separate Entity objects for each
- Pass list of entities to
create_feature_view
- Update code generation to handle multiple entities
- Update documentation with examples
Files to Update
sdk/python/feast/cli/dbt_import.pysdk/python/feast/dbt/mapper.pysdk/python/feast/dbt/codegen.py- Documentation
Related
Context
PR #5827 added dbt integration with single entity support.
Problem
Real-world feature engineering often requires multiple entities per FeatureView (e.g., a feature view keyed by both
user_idandmerchant_id). Currently only single entity column is supported.Current Limitation
feast dbt import -m manifest.json -e driver_id # only one entityProposed Enhancement
Support multiple entity columns:
Implementation Notes
create_feature_viewFiles to Update
sdk/python/feast/cli/dbt_import.pysdk/python/feast/dbt/mapper.pysdk/python/feast/dbt/codegen.pyRelated