Repository navigation
Fence the nightly relay flush and isolate read-only release planning - #124
Conversation
Wait for the committing delivery before queuing the sentinel so it cannot overtake the held edit flush. Classify the complete flush and next edit between socket fences, and probe both queue handovers under the state lock without a socket or clock racing the observation.
Keep version planning in a read-only job with no persisted checkout credentials. Gate the separate write-enabled release job on dry_run and pass the planned versions through job outputs.
There was a problem hiding this comment.
Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.
An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.
|
No actionable comments were generated in the recent review.
|
| Layer / File(s) | Summary |
|---|---|
Read-only release planning .github/workflows/release.yml |
The workflow defaults to no permissions and adds a read-only plan job that exposes version and current outputs. |
Gated release execution .github/workflows/release.yml |
The release job depends on the plan job, skips dry runs, and uses its outputs for release steps. |
Relay Frame Ordering
| Layer / File(s) | Summary |
|---|---|
Instance-based frame queuing crates/client/src/relay.rs |
Relay call sites use instance queue methods. The methods drain frames and send them through push. |
Lock-order handover tests crates/client/src/relay.rs |
Test probes check lock state during frame handover and require bounded worker completion. |
Delivery and edit-frame checks crates/client/src/relay.rs |
Test helpers classify sealed frames. Delivery tests check held-edit flushes and subsequent edit ordering. |
Priority:
Estimated code review effort: 3 (Moderate) | ~25 minutes
Change: Other
Sequence Diagram(s)
sequenceDiagram
participant ReleaseTrigger
participant PlanJob
participant ReleaseJob
ReleaseTrigger->>PlanJob: Start read-only planning
PlanJob-->>ReleaseJob: Provide version and current outputs
ReleaseJob->>ReleaseJob: Run only when dry_run is false
sequenceDiagram
participant RelaySession
participant Relay
participant RelaySocket
RelaySession->>Relay: Queue frames while state lock is held
Relay->>Relay: Drain frames and call push
Relay->>RelaySocket: Send each frame
Merge Risk: 
Minimal · up to 612c0
-
The change separates read-only release planning from the write-enabled release job and makes the relay ordering test deterministic. No concrete merge-blocking defect was identified. As with any workflow change, the split release flow is best confirmed with a hosted dry-run dispatch after merge.
Pre-merge checks: 
4 · 
1
-
Failed checks: 1 warning Passed checks: 4 passed-
Full details: Title check
-
Explanation
The title names the two changes in PR
#124: the nightly relay flush fencing in the relay test synchronization, and the split of read-only release planning into its own job. Both appear in the raw summary and objectives. The title is specific and short, though it has two clauses and uses the term "fence", which a reader may need context to interpret. It covers the main change without listing files.
-
Full details: Docstring Coverage
-
Explanation
Docstring coverage is 78.57% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff with cyclomatic complexity of at least 2. Documentation-only edits and deletions within surviving functions are included; unchanged diff context and fully deleted functions are excluded. Analyzed 14 functions across 1 files. (1 skipped: 1 unsupported.)
Resolution
Write docstrings for the functions missing them to satisfy the coverage threshold.
-
Finishing Touches: 💡 1
-
📝 Generate docstrings: 💡
-
- Commit to this branch
- Create a new PR
-
- Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.
Comment @coderabbitai help to get the list of available commands.
|
The retained local gates and their limits are in the description. Hosted checks (40839823151), container, amd64/arm64 smoke and publish-rehearsal (40839823361) are now green; publish/release are skipped as expected for a PR. Copilot reported insufficient remaining quota. The latest PR read has no actionable review or review threads; CodeRabbit remains in progress and is not a required check (gh pr checks --required reports none). Proceeding without a per-PR review wait under the owner policy. No nightly/release/image/deployment workflow is being manually dispatched, and no green main-nightly result is claimed. |
Changes
Two intentional commits on
fix/nightly-sentinel, head612c011bf3dcfe153b05a9af08e6090882652614:8802188 test(relay): fence the committed-state flush: repair the nightly relay regression by waiting for committing delivery before queuing the sentinel. Classify the held-edit flush and the next edit between socket fences, and test both queue handovers under the state lock without a socket or clock racing the observation.612c011 ci(release): isolate the read-only planner: move planning into a separate job withcontents: read, nonpersistent checkout credentials, and version/current outputs. Keep the existing write permissions only in the release job, which depends on planning and is skipped fordry_run. Preserve the current install-nix-action v31.11.3 pin.Both source changes received independent local review with no blocking findings.
Verification already executed
This publication does not rerun or replace the retained local evidence.
The complete relay-repair gate is retained in the worktree's
.tmp/full-checks.DZ39Ndgf. Deployment/release guards, package, licences, workflow lint and typos also passed. Normal commit hooks passed.After integrating the planner and preserving the current action pin, normal hooks and
scripts/ci-local.sh checksagain exited 0. That run reused the successful Rust derivations; it did not recompile or rerun 238 Rust tests. Its newly built dry-run guard reportedRan 23 tests ... OK, and links reported23 Total, 19 Unique, 23 OK, 0 Errors, 1 Redirect. Evidence:.tmp/planner-pin.HaTMLv72.Limits
the sentinel frame arrives; deny/audit were skipped. This PR repairs that regression but does not establish a green main nightly.Summary by CodeRabbit