Repository navigation
chore(deps): update dependency bazel_lib to v3.8.0 - #4870
Conversation
|
Edited/Blocked NotificationRenovate will not automatically rebase this PR, because it does not recognize the last commit author and assumes somebody else may have edited the PR. You can manually request rebase by checking the rebase/retry box above. |
Record the existing enriched startup error before the capability composite continues with its partial surface. Preserve that policy and every startup and request budget. Enable the relevant analyzer trace and application debug targets in the production LSP acceptance fixture, and retain its full daemon log when standard capability negotiation fails.
Drive transient maintenance Retry outcomes until the retained page settles, then require its typed StoreRetention continuation before inspecting rows. Reuse the existing stronger fixture from the qualified history branch. Preserve the unrelated-writer interleaving, its two-second deadline, the ten-second settlement bound, and exact sixteen-row/frontier assertions.
Preserve the reviewed fixture contracts while integrating current master. Resolve only the overlapping schema or maintenance fixture edits.
There was a problem hiding this comment.
🔍 Devin Review: 2 flags
Not posted on this PR by your GitHub settings — view them in Devin Review. (Configure)
Retain bazel_lib 3.8 and its Windows manifest override while adopting rules_python 2.5.1 and the matching staged release archive. Preserve the shared runtime log fixes and all reviewed receipt and startup changes.
Why
The bazel_lib 3.8.0 Windows launcher can reach its sibling shell script without calling the runfile resolver. That path must still export the batch launcher's runfiles manifest for tools such as the esbuild lifecycle action.
Validation also exposed a session relation publication ownership problem: the batch retained an owned SQLite receipt snapshot across native graph work. The snapshot's reader lease is unnecessary after the expected receipt has been read.
Changes
Validation before this follow-up
Linux on
87ade1f7freshly passed the new reader-release regression in the temporal-store suite: 187 passed, with one pre-existing ignored test. The regression checks real registered-reader occupancy during native publication and verifies the resulting receipts and round trips.The runtime-acceptance suite freshly passed all 142 tests in 71.09 seconds.
Repository, Clippy, Dashboard, and Feature gates passed.
The overall Linux run remains 179/180 targets: the retained fake-daemon fixture timed out at its unchanged ten-second startup wait. The later measured run below passed this readiness case.
Windows on
87ade1f7completed successfully: 467 targets and 26,393 build actions. This was build/typecheck validation.The removed automation capture was qualified on chore(deps): reconcile Rsbuild update with runtime repairs #4862 head
753a4b5b: Linux passed 180/180 targets, including a fresh automation suite in 22.8 seconds; Windows build/typecheck passed 467 targets and 26,370 actions. The owner merged that head before its diagnostic cleanup was published. This PR carries the exact cleanup forward without changing its test budgets or production behavior.A later master run at
694f8e5cpassed 141/142 runtime-acceptance tests. The first initialize response on a newly opened cross-project LSP session timed out while receipt admission was outstanding. The blocking receipt callback had started, but the available log did not identify the remaining admission or I/O phase. The recorder and spool source are identical at this PR's integration base; no lock, scheduling, or disk-latency cause is claimed.Follow-up validation
Measured head
7c139c38passed 179/180 Linux targets, with 99 freshly executed. Runtime acceptance passed 142/142, core CLI passed 206/206, and automation passed freshly. The sole failure was global-db's existing 300-second target limit after 457/533 reported completions; the workflow cancellation and shape-refusal cases both passed. Repository, Clippy, Dashboard, and Feature gates passed. The four measured CLI startup roots took approximately 2.08–2.65 milliseconds, so that run did not reproduce the earlier startup delay. The successful runtime artifact contains no receipt-phase timings.Windows on that measured head completed successfully: 467 targets, 26,370 actions, and 272 remote executions in 6,363.811 seconds. The complete log was preserved before changing the branch. This is build/typecheck evidence, not native Windows test execution.
Current head
51f1fca5publishes the independently reviewed cleanup and workflow-fixture reuse. It removes both temporary measurements while retaining the permanent opt-in receipt phase events and all production fixes. Actual pinned formatting, commitlint, and diff checks passed before publication. Final Linux validation passed all 180 targets, with 92 freshly executed; global-db, automation, runtime acceptance, core CLI, temporal store, and MCP all passed freshly. Final Windows build/typecheck passed 467 targets and 26,370 actions, including 223 remote executions, in 4,312.161 seconds. The full Windows log was preserved before merging. All applicable repository, Clippy, Dashboard, and Feature gates passed; native Windows tests were not part of this run. The actual current-master merge preview is conflict-free. These results qualify the final candidate without assigning a cause to historical latency failures.This PR contains the following updates:
3.7.2→3.8.0Configuration
📅 Schedule: (UTC)
🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.
🔕 Ignore: Close this PR and you won't be reminded about this update again.
This PR was generated by Mend Renovate. View the repository job log.