Evidence-backed facts
- Iceberg documents typed table schemas with persistent field IDs that are never reused within a table and supports explicit schema evolution. informative
- A compute engine reads table metadata and data files through its catalog and integration, and field IDs rather than names alone preserve column identity across evolution. informative
- Schema, partition, metadata, transaction, branch, and tag updates require explicit catalog and storage authority with concurrency handling. informative
- A table-format specification provides no catalog, object-store credential, engine compatibility, access control, committed snapshot, data quality, or successful write evidence. informative