[rmcdaniel]

Part of #325.

Native's repair command and repair pass now create the missing task for the wait cases #887 still refused. Each case matches published Server 2.11.7 (TaskRepair::createMissingTask with WorkflowTaskPayload::forMissingWorkflowTask):

Case	Task created
Received signal	workflow task with forSignal's payload
Accepted update	workflow task with forUpdate's payload; the update's open wait moves to it
Resolved child	workflow task {workflow_wait_kind, open_wait_id, resume_source_kind: child_workflow_run, resume_source_id, child_call_id, child_workflow_run_id}
Pending condition or signal timeout	the timer task again, available at the deadline; the wait moves to it
Fired condition or signal timeout	the workflow task that applies the timeout, available when the timer fired

Every created task is ready with repair_count 1.

• A run with a received signal and no open task is now projected as PHP projects it: a signal wait ("Waiting to apply signal …", signal-application:<id>) with liveness repair_needed. Before, native left it without a summary, so the repair command answered repair_not_needed.
• No case is refused any more, so the development_workflow_task_repair_unqualified refusal and the pass's unported-recovery path are gone.

PHP findings. I recorded seven cases against the published image with each run's open task row deleted and maintenance stopped:

• every case answered repair_dispatched, and the run was described as before;
• native on this branch matches on task type, payload fields and the description afterwards;
• native main refused the update, child and four timeout cases (422) and answered repair_not_needed for the signal.

The new checkpoint is docs/rust-server-port-checkpoints/2026-11-01-repair-missing-wait-tasks.md. It records two small native differences: the repaired timeout timer task keeps a condition occurrence id, and native's own accepted-update task payload is narrower than forUpdate.

Checks run locally

• cargo test (all suites, SQLite): passed (368 lib, 649 execution, 2 upgrade_inspect).
• workflow_repair tests on PostgreSQL 17 and MySQL: 4/4 passed on each.
• Full execution suite on PostgreSQL: 644 passed, 5 failed. The failures are in completion_budget and worker_poll_fence, with stale_worker_registration. They ran while the host was overloaded and the MySQL suite ran alongside, and this change doesn't touch those paths.
• Full MySQL execution suite: not completed locally (the host stalled), so CI's MySQL and MariaDB shards cover it.
• Node unit suite, excluding ServerParityUpdateValidationTimingTest.mjs: 6459 passed, 1 failed. The failure is PhpImageContextTest, which needs a docker CLI the test container doesn't have.

🤖 Generated with Claude Code

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
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

[rmcdaniel]

Sonar's one new issue is rust:S3776, cognitive complexity 39 against 30, on create_missing_timeout_task (task_repair.rs:346). It's a style finding, not a defect, and the quality gate passed. The function mirrors PHP's condition and signal timeout branches in one place. I'm accepting it for this slice; a later refactor can split the fired and pending paths. Enqueueing at 74c4ee7 with all 58 checks green and no review objection.