# Result identifiers and download errors

statement_id comes from saved SQL query results for ScanSavedResult and DownloadSavedResultCSV; it is not an IAM policy statement ID. Retain chart execution IDs, query IDs, and saved result IDs separately. Result reading scans at most 1000 saved rows. Ordinary downloads use saved-result CSV, while point drill-down downloads rerun the current detail query.

An unready ordinary chart_result may omit result. Empty protobuf subobjects/zero values do not establish an actual empty result. Public SDK examples now require the execution ID to match the submitted ID, status=completed, and a present result before printing rows/columns, with bounded polling. Complete historical-selection, epoch, and scan-limit behavior remains on the existing task list.

| Failure stage | HTTP/error format |
| --- | --- |
| Missing Workspace or failed time-parameter parsing | 400; string code and msg |
| Saved result not ready | 409；ErrDataDashboardResultNotReady、message |
| Point drill-down cannot be rewritten | 400；ErrDataDashboardDrillThroughUnsupported、message |
| Refresh paused/configuration missing | 409; dedicated string code and msg |
| Scheduling/refresh service unavailable | 503; integer code and message |
| Ordinary SQL download error | Integer HTTP status code and message, including 403, 404, 413, 429, and others |

Fallbacks, including error-localization failures, may still return string code/msg. Read the actual response; one unexpected field shape does not establish that the service failed to respond.
