docs(coordination): clarify operation retry and no-change receipt semantics - #4130
Conversation
…p replay Signed-off-by: huangruiteng <huangrt01@163.com>
|
Self-review of exact head Changed surfaces are the English/Chinese shared-authority RFC and one existing real FileAuthorityStore update regression. There is no production-code or authority-policy change. The test now distinguishes an old terminal no-change replay after an intervening edit from a genuinely independent attempt with identical parameters. The RFC explicitly preserves the current recovery limitations rather than presenting documentation as their implementation. Risk-based premerge evidence: 30 focused TS tests and 33 public-caller Python tests passed on the current baseline; documentation governance, repository hygiene, whitespace and public/private scans passed. Five selected transport/checkpoint tests also passed during the investigation on unchanged runtime. No local failures or manual holds. Full runtime/PostgreSQL suites were not rerun because no such implementation changed. Hosted checks are still running; their completion is not claimed as evidence. Typed-state, authority, domain-neutrality, behavior-change, guidance/obligation and default-off review: no new state rule, permission, runtime default, heuristic classifier or execution obligation. The bounded future-facing pass is the focused counterexample and explicit retry contract, not a new framework. This small docs/test clarification is covered by the owner's explicit self-merge authorization. Proceed with an exact-head admin squash merge using the local validation above, without changing repository protection or persistent local gate configuration. |
* origin/main: fix(control-plane): type oversized runtime requests (loopx-project#4112) refactor(todos): retire Python lifecycle field rules into TypeScript (loopx-project#4118) perf(state): bound limited rollout log reads (loopx-project#4100) refactor(planning): close canonical Todo planning consumer paths (loopx-project#4117) fix: align frontier projection role typing fix: keep frontier index off diagnostic payloads test: retire stale frontstage strategy smoke fix: preserve complete todo frontier revisions docs(coordination): clarify retry identity and preserve terminal no-op replay (loopx-project#4130) docs(contributing): structure public-safe PR validation disclosure Signed-off-by: Huashuai Qu <256761+huashuai@users.noreply.github.com>
Summary
Follow up the review on #4053 without changing runtime authority semantics.
Validation
342e02bf9(based on current main91baa0f96).node --no-warnings --experimental-strip-types --test tests/control_plane_ts/local_authority_runtime.test.ts tests/control_plane_ts/todo_update.test.ts: 30 passed, zero skipped; real disposable FileAuthorityStore, identity conflicts and lost-response coverage.python -m pytest -q tests/control_plane/test_local_coordination_authority.py: 33 passed; public caller/provider/projection regressions.python examples/docs-governance-smoke.py,python examples/repository-hygiene-smoke.py, diff whitespace and public/private scans.Coverage/gaps: only documentation and one existing regression test change; no production implementation, permissions, provider schema, default behavior, UI or installation changes. PostgreSQL/full runtime suites are not required for this bounded non-runtime diff and were not rerun. No active goal or private data was used. The negative replay/new-intent contrast exercises the relevant real backend; this is not a migration or a new recovery framework.
Review and merge scope
Product/architecture judgment: preserve existing typed identity-consumption semantics while preventing a misleading blanket UUID/no-op fix. The future-facing pass documents the actual recovery boundary and adds a narrow regression instead of introducing speculative tracking infrastructure. No typed-state, domain-neutrality, guidance/obligation or default-off behavior is changed.
Owner authorized self-merge after validation and a temporary local Git time-gate bypass for this PR only. Bypass uses per-command configuration; no permanent hook or repository policy change is included. The diff is limited to two RFC files and a focused test; DCO sign-off is present.