Comment by sr-white

Good question! Currently FLATTEN does a simple translation to DuckDB's UNNEST, which works for basic array expansion but won't give you the full Snowflake output columns (SEQ, KEY, PATH, INDEX, VALUE, THIS) or the named parameter syntax (input => ...). That said, a more complete implementation is definitely doable — DuckDB has the primitives (unnest, generate_subscripts, json_each) to build this properly. If this is something you'd find useful, feel free to open an issue with your use case. Happy to prioritize it if there's demand!