Summary
While documenting flexible/databases.rs::DatabaseBackupConfig field-by-field for the #80 rustdoc pass (#96), spotted three pairs of fields that look like wire-shape duplicates surfacing the same value under two different JSON keys:
| Field A |
Field B |
interval |
backup_interval |
time_utc |
database_backup_time_utc |
storage_type |
backup_storage_type |
In the #96 PR these were documented as "server-returned alias for ..." to unblock the doc lint, but the underlying question remains: which form does the real API return on which path? Is one of them obsolete or only present on a specific endpoint variant (e.g. the embedded view inside a Database response vs a standalone backup-config response)?
Why this matters
If both fields can be populated simultaneously the doc framing is fine; if only one is populated at a time, callers reading the "wrong" field will silently see None even when the API supplied a value (same class of bug as #76 activatedOn).
Suggested investigation
- Live validation against a real Cloud account: trigger a backup-config retrieval and inspect the raw response. Confirm which key the API actually emits.
- If only one form is real, mark the other
#[deprecated] (or drop it under a major bump) and update the docs to point at the canonical field.
- If both can appear, document the conditions under which each is populated.
References
Summary
While documenting
flexible/databases.rs::DatabaseBackupConfigfield-by-field for the #80 rustdoc pass (#96), spotted three pairs of fields that look like wire-shape duplicates surfacing the same value under two different JSON keys:intervalbackup_intervaltime_utcdatabase_backup_time_utcstorage_typebackup_storage_typeIn the #96 PR these were documented as "server-returned alias for ..." to unblock the doc lint, but the underlying question remains: which form does the real API return on which path? Is one of them obsolete or only present on a specific endpoint variant (e.g. the embedded view inside a
Databaseresponse vs a standalone backup-config response)?Why this matters
If both fields can be populated simultaneously the doc framing is fine; if only one is populated at a time, callers reading the "wrong" field will silently see
Noneeven when the API supplied a value (same class of bug as #76activatedOn).Suggested investigation
#[deprecated](or drop it under a major bump) and update the docs to point at the canonical field.References
activatedOnrename) and docs: OpenAPI spec deltas - fields missing from official spec #40 (spec deltas tracking)