instead keep it as optional params in get_historical_features

add docstring for this

import this from from feast.infra.offline_stores.offline_utils import DEFAULT_ENTITY_DF_EVENT_TIMESTAMP_COL

Line 229 - Line 244 is common across the offline store to figure out the start_date & end_date. If its not too much, is it possible to create utility function which can be re-used?

Yes, I’ll refactor this into a reusable utility function and apply it to similar date range calculations across the codebase in a follow-up PR

Missing docstring for this function. Given the non-trivial logic around join_key_map handling, field_mapping, and timestamp filtering, a brief docstring explaining the purpose ("Build a synthetic entity DataFrame from feature view sources for non-entity retrieval mode") and the return value would help future maintainers.

This start_date/end_date derivation logic (defaulting end_date to now, computing start_date from max TTL, falling back to 30 days) is duplicated across the DuckDB, Postgres, and File offline stores. The Postgres PR (#6057) has the same pattern inlined.

Please extract this into a shared utility (e.g. in offline_utils.py or reuse the existing compute_non_entity_date_range from feast.utils). This would keep the stores consistent and avoid divergence over time.

This importlib.import_module call runs at module load time, which transitively imports ibis (via duckdb_repo_configuration → feast.infra.offline_stores.duckdb). If ibis is not installed, this breaks all tests that import from repo_configuration.py — not just DuckDB-specific tests.

Other non-core offline stores (BigQuery, Redshift, Snowflake at lines 116-137) are added conditionally behind environment variable checks. Please wrap this in a try/except ImportError:

try:
    from tests.universal.feature_repos.duckdb_repo_configuration import DuckDBDataSourceCreator
    OFFLINE_STORE_TO_PROVIDER_CONFIG["duckdb"] = ("local", DuckDBDataSourceCreator)
except ImportError:
    pass