Responsibilities

• Execute SQL queries against DuckLake datasets via DuckDB
• Track query execution state and timing
• Handle query cancellation
• Emit completion/failure events with results or error messages

Aggregate: QuerySession

The QuerySession aggregate models query execution as a state machine.

State Machine

Idle → Running(queryId, datasetRef, sql) → Completed(queryId, results) | Failed(queryId, error)
                      ↓ (cancel)
                    Idle

Key Value Objects

• QueryId: Unique identifier for query execution
• DatasetRef: Reference to catalog + dataset (e.g., hf://datasets/sciexp/fixtures/experiment_results)
• SqlQuery: The SQL string to execute
• QueryResults: Row count, column names (results cached separately)
• ErrorMessage: Execution error details