Repository navigation
🌱 chore: top up v6 from v5 (2026-11-15, third) - #18312
Conversation
* 🐛 Sessions: make the queued-Send deadline test deterministic The test set a 1s real-time fail-open deadline and waited 5s for delivery, which loses under mutation-job CPU load. The queue worker now measures the deadline through an injectable clock (test seam, no behaviour change) and the test advances a fake clock, asserting no delivery before the deadline and exactly one after. Signed-off-by: Andy Anderson <andy@clubanderson.com> * 🐛 Sessions: wait on pane polls instead of sleeping in the deadline test TestSleepRatchet forbids new time.Sleep calls; count fake-runtime Capture calls instead. Signed-off-by: Andy Anderson <andy@clubanderson.com>
* ✨ meetings: config block and quorum child supervisor Adds the meetings: config (off by default, env overrides for every key, validation, keystore-sealed Vexa and Discord secrets), the quorumproc supervisor for the Quorum child (per-boot bearer token in a 0600 file, /healthz checks, restart backoff, stop on shutdown), boot wiring, and a Dockerfile layer for the pinned quorum bundle (skipped while the version is empty). Fixes #18289 Signed-off-by: Andy Anderson <andy@clubanderson.com> * 🐛 meetings: golden config fixture and no fixed sleeps in tests Signed-off-by: Andy Anderson <andy@clubanderson.com>
* ✨ feat(crux): deliver agent output signals to kick observers logOutputSignals keeps its log line and now also emits an output-signal KickEvent for lines whose first token is [STATUS], [FINDING], [COMPLETE] or [ERROR]: agent, in-flight queued kick ID, signal kind, bounded text. Deduplicated per agent and kick so a pane redraw does not deliver a finding twice. A marker mid-sentence (an echoed prompt) is not a signal. Fixes #18287 Signed-off-by: Andy Anderson <andy@clubanderson.com> * 🐛 fix(crux): exact rune bound on output signal text; keep manager.go under its size ratchet Signed-off-by: Andy Anderson <andy@clubanderson.com> --------- Signed-off-by: Andy Anderson <andy@clubanderson.com>
pkg/research defines research-1..N (PoolSize, default 2, max 8) as ADVISORY agents with include_repos false, so the proxy and token tier refuse GitHub writes regardless of the prompt. Pool.Ensure registers them (call only when meetings are enabled), Pool.Dispatch queues a query on the least busy agent via EnqueueKick and returns the kick ID, and Pool.Observe maps kick events and output signals for its kicks to queued, running, finding, done and failed. Fixes #18288 Signed-off-by: Andy Anderson <andy@clubanderson.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Automated release commit. Compiles changelog.d/ fragments and moves the CHANGELOG.md Unreleased section into a dated v5.269.0 entry. See src/docs/releases.md. Signed-off-by: hive-release-bot <actions@github.com>
…8309) The resolver kept v6's side of any src/pkg/config/config.go conflict. v6 split that monolith into topical files, so a conflict there means v5 changed config.go since the last top-up and keeping v6's side drops the change without a trace (#18222's keystore sealing survived only through a hand port; #18300's meetings block was next). config.go is no longer a known pattern: the resolver now refuses and prints the diff command for a hand port. The manifest path check for the index.html regeneration now matches webstatic.ParseIndexManifest's clean-relative-path rule. The throwaway-repo test gains a case where v5 adds a real config field and asserts the resolver refuses and leaves the file unmerged. Signed-off-by: Andy Anderson <andy@clubanderson.com>
release: v5.269.0
- inbox.TranscribeSweep: auto mode queues pending items from inbox.transcription.sources within the daily cap and monthly budget, dedupes, retries failures with 1h/2h backoff up to 3 runs and budget skips on the next UTC day. - Boot runs the sweep after each connector sync, audited as inbox_transcribe by inbox-auto-transcribe, and logs the egress host. - Settings expose egress_hosts; the settings panel shows month minutes used, today's runs, the egress host and an auto-mode hint. - Network requirements and provisioning docs list api.deepgram.com:443 only when transcription is enabled. Signed-off-by: Andy Anderson <andy@clubanderson.com>
Forward-merges origin/v5 through 2afddbd (#18298) into v6, carrying Conflicts resolved by hand: - src/pkg/config/config.go: keep v6's Retro field and add v5's Meetings field (#18300) to the Config struct. - src/pkg/config/validate.go: keep v6's advisor-lane and hub NPS checks and add v5's Meetings.Validate call. - src/pkg/config/testdata/agent_runtime_golden.json: keep v6's retro entry and add v5's meetings entry, in struct field order. - src/cmd/hive/main.go: keep v6's Spektacular clone-auth wiring and add v5's wireMeetings boot block. - CHANGELOG.md: keep both lines' release sections (union). Signed-off-by: Andy Anderson <andy@clubanderson.com>
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
There was a problem hiding this comment.
Style review (advisory, non-blocking) — head 0dfcdfc
.github/scripts/test-v6-topup-autoresolve.shcase_config_real_field(new, ends withreturn 1): the case inverts the file's convention. Every other case returns 0 on success and is registered asok/failby what the scenario should do. Here assertion failuresreturn 0and the passing pathreturn 1, soexpect config-go-real-field-refused failpasses only because the pass and fail values are swapped. It works, but a later assertion added in the usual style (|| { echo ...; return 1; }) would silently pass. Suggested shape: have the case succeed (return 0) when the resolver refuses as intended, register it asok, and usereturn 1for the failure messages.
- [ "$rc" -ne 0 ] || return 0
- git status --porcelain | grep -q '^UU src/pkg/config/config.go$' || { echo "config.go was resolved"; return 0; }
- printf '%s\n' "$out" | grep -q 'port v5.s change by hand' || { echo "no hand-port hint"; return 0; }
- return 1
+ [ "$rc" -ne 0 ] || { echo "resolver accepted the config.go conflict"; return 1; }
+ git status --porcelain | grep -q '^UU src/pkg/config/config.go$' || { echo "config.go was resolved"; return 1; }
+ printf '%s\n' "$out" | grep -q 'port v5.s change by hand' || { echo "no hand-port hint"; return 1; }
+ return 0
}
-expect config-go-real-field-refused fail case_config_real_field
+expect config-go-real-field-refused ok case_config_real_fieldEverything else I read (resolver comments, workflow comment, manifest path rule) is consistent with the PR body. I did not review the Go merge content for correctness; that is outside this perspective.
— hive: agent=adjudicator-4 backend=claude model=claude-sonnet-5-5 claude=2.1.325
Review follow-ups (below
|
|
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor. Review findings from Routing to a human because the merge carries CI/workflow changes; I found no defect in what I read. What I checked at head
Because |
0dfcdfc
ℹ️ Sentinel notice — informational; author is trusted (owner), merge is not blockedHive flagged this PR (author @clubanderson, head
Hive did not add the |
|
Needs a decision, not a merge:
Head: verdict requires_human @0dfcdfc, 80/81 checks green, 1 pending, DCO yes, soak 0s of 10m0s. How to release it:
Clear the cause and the hive re-evaluates on its next pass; the stage moves to |
Mutation testingHead ⏱️ The 900s mutation budget ran out, so 4 package(s) were not fully mutated and the results below do not cover them:
❌ gremlins did not produce a report for 1 package(s), so no verdict can be derived from this run (see the job log):
6 mutant(s) in changed files are not reached by any test (NOT COVERED). A maintainer may apply the
Verdict: |
This PR merges
origin/v5(2afddbd, release v5.269.0 plus #18298) intov6(f00d126) as a single signed merge commit. After it lands,origin/v5is an ancestor ofv6again. All 25 first-parent v5 commits from 2026-11-15 are ancestors of the merge (checked withgit merge-base --is-ancestor).I carried it by hand because the automated runs after #18309 (for example run 43741778114) stopped as intended. The fixed resolver refused the
src/pkg/config/config.goconflict and printed the hand-port hint, so it did not keep v6's side and drop #18300'sMeetingsfield. The automated runs opened no top-up PR, so there is nothing to supersede.v5 PRs included
Conflicts
src/pkg/config/config.goRetrofield and added v5'sMeetings MeetingsConfigfield (#18300) after it, in the sameConfigstruct. v6 still keeps theConfigstruct inconfig.go. The rest of #18300's block (structs, defaults, env overrides, keystore-sealed secrets,Validate) is in the newmeetings_config.go, which merged cleanly.src/pkg/config/validate.govalidateAdvisorLaneandHub.ValidateNPSchecks and added v5'sMeetings.Validate()call after them.src/pkg/config/testdata/agent_runtime_golden.json"retro": {}and added v5's"meetings"object after it, inConfigfield order. This is the outputHIVE_UPDATE_GOLDEN=1produces. I did not run the generator, because local tests are off by repo rule, andTestAgentRuntimeConfigGoldenin CI checks the result.src/cmd/hive/main.gowireSpektacularRunnerWithCloneAuthplus the live reconfigure hook) instead of v5's plainwireSpektacularRunner, and added v5'swireMeetingsboot block after it.CHANGELOG.md## 2026-11-15 (v5.269.0).kick_observer.go and #18305
#18305 does not touch
kick_observer.go. It addsoutput_signals.go, which callsemitKickEventandInFlightKickID, plus theKickEventOutputSignalandOutputSignal*constants incrux/kick_event.go. Both helpers exist in v6 unchanged since #18306. The mergedkick_observer.gois byte-identical to v6's: the chainingSetKickObserverandkickObserverArchiveDetail/KickObserverDetailSource/Reasonstay, and so does v5's fan-out from #18303. Output-signal events go only to the fan-out observers. The single-slot chain does not see them, the same as in v5. A follow-up PR converges this file with v5.Other checks
TRUSTED_SCRIPTS_TREEalready equalsHEAD:src/scripts(70a83fac).static/index.htmlequals the byte-exactstatic-src/manifest.txtconcatenation (✨ Inbox: automatic transcription sweep for opted-in sources #18298 changed both, and they merged cleanly).gofmt -lis clean on the hand-edited Go files. Nothing was built or tested locally; CI is the gate.dcocommit status is the gate.🤖 Generated with Claude Code