[rmcdaniel]

Part of #980, first slice (start-visibility carryover, Core workflow#985).

Native refused to redrive any failed run that had a memo or search attribute (unsupported_visibility_metadata), or either in its WorkflowStarted (unsupported_source_metadata). Core redrives start-time and inherited values.

Change (rust/src/runtime/redrive.rs)

• The successor keeps each source memo and search-attribute row as it is, with its upsert sequence and inherited_from_parent, as Core's copyForRedrive does. This uses new per-dialect MEMO_REDRIVE_COPY / SEARCH_ATTRIBUTE_REDRIVE_COPY. Its WorkflowStarted already copies the source's payload, so it keeps the memo and search attributes there too.
• Only rows that workflow code upserted (upserted_at_sequence > 0 and not inherited) still refuse with unsupported_visibility_metadata. Such a run's MemoUpserted/SearchAttributesUpserted history also refuses as unsupported_history until native reuses upsert steps (a later Match Core's failed-run redrive: visibility carryover, eligibility in describe, reused markers and timers #980 slice).
• parent_workflow_run_id in the start still refuses as unsupported_source_metadata.

Not in this slice: describe's can_redrive/redrive_blocked_reason (Core workflow#990). That changes a describe response the parity corpus pins to published PHP, so it comes next with reviewed native expectations. Reused markers, timers, signals, updates and upserts follow Core's later slices.

Tests: new redrive_carries_start_time_memo_and_search_attributes. It covers a failed run started with a memo and a registered keyword attribute that redrives; the successor's describe, WorkflowStarted and visibility list carry both. A run whose code upserted a memo is refused before anything is written. It passes on SQLite, PostgreSQL 16 and MySQL 8.4, along with the existing redrive and replay-clock redrive tests.

🤖 Generated with Claude Code

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud