发布

  • [OPIK-3925] [BE] Optimize dataset items column type extraction with materialized column (#4816)

    frostbyte_neo 发布于 2026-01-20 09:38:17 +00:00

    • Add materialized column_types Map(String, Array(String)) to dataset_items and dataset_item_versions tables
    • Pre-compute column types at ingestion time instead of query time
    • Simplify queries using ARRAY JOIN instead of complex arrayFold operations
    • Use FINAL keyword for automatic deduplication
    • Update SELECT_DATASET_ITEMS_COLUMNS_BY_DATASET_ID, SELECT_DATASET_ITEMS_COUNT, and SELECT_COLUMNS_BY_VERSION queries
    下载附件