Skip to content

✨ feat(dashboard): feed published checkpoint diff regions to Code activity - #16513

Merged
hivecommons-hive[bot] merged 6 commits into
v5from
hive/adopt-16499
Nov 4, 2026
Merged

hivecommons-hive[bot] merged 6 commits into
v5from
hive/adopt-16499

Conversation

@hivecommons-hive

@hivecommons-hive hivecommons-hive Bot commented Nov 4, 2026 •

Copy link
Copy Markdown
Contributor

Supersedes #16499 (adopted after CI went red and the contributor was unavailable; original commits by @thatguywob preserved).

Closes #16476

Stacked on #16492 (part 1/3, still open). This branch is #16499's head 45ed859ed1fe plus fix-agent follow-ups (a053567e2, test settling) merged with the current scanner/fix-16475 head. Because the base is v5, the diff currently shows all of part 1 (reader kiosk view) and part 1's tracker chevron CSS fix (7f93822f5) in addition to the part-2 work. The part-2-only delta is 7f93822f5...a053567e2 (checkpoint source, matrix ingest, api docs, reader wiring).

Merge order: #16492 → this PR → cutover (part 3). Do not merge this before #16492; once part 1 lands on v5, this diff collapses to the checkpoint-source changes only.

Last lane reason on the original: hive-signoff-attribution.

Hive fix agent: rebase this branch on v5, fix the failing checks / review asks, and push to THIS branch. Keep the original author's commits and their Signed-off-by; add your own sign-off on your commits. Do not open another PR.

@kubestellar-prow kubestellar-prow Bot added the dco-signoff: yes Indicates the PR's author has signed the DCO. label Nov 4, 2026
@kubestellar-prow

Copy link
Copy Markdown
Contributor

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign clubanderson for approval. For more information see the Code Review Process.

The full list of commands accepted by this bot can be found here.

Details Needs approval from an approver in each of these files:

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

@kubestellar-prow kubestellar-prow Bot added the size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files. label Nov 4, 2026
@github-actions

github-actions Bot commented Nov 4, 2026 •

Copy link
Copy Markdown
Contributor

Open PR overlap check

Base branch: v5

Conflicts are checked with every PR merged onto the current tip of v5, so a PR that only needs a rebase is not reported as conflicting with its siblings.

PR Title Relation Files
#16570 🐛 fix(dashboard): render reader context lines in a neutral color CONFLICT src/pkg/dashboard/reader_ui_test.go
src/pkg/dashboard/static-src/js/reader.js
src/pkg/dashboard/static/index.html
#16569 🐛 fix: avoid empty reader catch-up pulls for filtered lines CONFLICT src/pkg/dashboard/reader_ui_test.go
src/pkg/dashboard/static-src/js/reader.js
src/pkg/dashboard/static/index.html
#16500 ✨ feat(dashboard): replace Matrix presentation with Code activity same files changelog.d/added-16476-published-checkpoints.md
src/cmd/hive/main.go
src/cmd/hive/matrix_ingest.go
src/docs/api-reference/auth.md
src/docs/api-reference/dashboard-misc.md
src/docs/api-reference/health-status-events.md
src/docs/api-reference/snapshots-and-style.md
src/docs/dashboard.md
src/pkg/dashboard/api_checkpoint.go
src/pkg/dashboard/api_matrix.go
src/pkg/dashboard/checkpoint_source.go
src/pkg/dashboard/checkpoint_source_test.go
src/pkg/dashboard/matrix_persist_test.go
src/pkg/dashboard/matrix_source.go
src/pkg/dashboard/reader_ui_test.go
src/pkg/dashboard/server.go
src/pkg/dashboard/static-src/js/reader.js
src/pkg/dashboard/static/index.html
#16514 ✨ feat(dashboard): replace Matrix presentation with Code activity CONFLICT src/docs/api-reference/dashboard-misc.md
src/docs/dashboard.md

@hivecommons-hive hivecommons-hive Bot added author/hive This repository's own hive (the App or one of its agent lanes) authored this PR dco All commits are DCO signed-off role/trusted Author is trusted for unattended merge (trusted-author role or registry/claim tier in require_tiers) stage/050-review-pending Review swarm requested; awaiting a verdict for the current head stage/070-fixing Hive-authored PR: CI red, changes requested or needs rebase; a hive agent is fixing it and removed stage/050-review-pending Review swarm requested; awaiting a verdict for the current head labels Nov 4, 2026
@hivecommons-hive

hivecommons-hive Bot commented Nov 4, 2026 •

Copy link
Copy Markdown
Contributor Author
10c14e9

ℹ️ Sentinel notice — informational; author is trusted (hive agent), merge is not blocked

Hive flagged this PR (author @hivecommons-hive[bot], head 10c14e9bf285) because it matches behaviors that can override security controls, escalate privileges or damage the codebase. This is a heuristic, not an accusation — a maintainer should confirm the change is intended before it merges.

  • test_removal — deletes 1 test file(s) (Deletes test files or guts test coverage)
    • src/pkg/dashboard/matrix_persist_test.go
  • remote_code_execution — adds a remote-fetch-and-execute, decoded payload, reverse shell or embedded credential (Adds curl|sh, eval of encoded payloads, reverse shells or embedded private keys)
    • src/pkg/dashboard/checkpoint_source_test.go

Hive did not add the sentinel-alert label because this author is trusted. The finding remains recorded for audit and the dashboard Security tab; set sentinel.trusted_authors_block: true to require the blocking label for trusted authors too.

@hivecommons-hive hivecommons-hive Bot left a comment

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Intent alignment (#16476), reviewed at head 45ed859

The part‑2 code itself tracks the issue's acceptance criteria well: whole-region admission with fail-closed hunk validation (src/pkg/dashboard/checkpoint_source.go:94-171, :191+), unsafe additions or removals drop the region (:173-183), Pushed PR label / seq / latest / epoch / stage on the response (:282-291), no-store even on auth failure (src/pkg/dashboard/server.go:1981-1983), queued-observation epoch drop on policy change (src/pkg/dashboard/matrix_source.go:124-137, :184-186), and no workspace collector.

1. Stacked PR silently carries part 1 (#16492) and the dependency note was dropped — in scope, please fix the body.
Commits 1bf96ee…8692bfe here are exactly #16492's head (8692bfeaa8), still OPEN. The original #16499 body said "Depends on #16492 … Merge the reader first"; this superseding body says only Closes #16476 with no dependency marker. Because a Supersedes PR auto-merges once CI/review are clean, merging this as-is lands the whole reader without closing #16475 and leaves #16492 as a conflicting duplicate. Requested change: add Depends on #16492 — merge the reader first, then rebase (and keep it blocked until then), or if the intent is to land both here, add Closes #16475 and close #16492 as superseded.

2. Advisory — removal of the #16276 Matrix restart persistence lands ahead of part 3.
src/cmd/hive/main.go drops SetMatrixPersistPath, matrix_source.go deletes restore/persist, and SetMatrixPersistPath is now a no-op (matrix_source.go:479-482). This matches the issue boundary "memory-only … no persistent history" and is disclosed in the body/changelog, but it regresses the still-live Matrix kiosk (blank after restart) before #16477 retires it. Acceptable per the boundary; noting it so part 3 / operators are not surprised.

3. Low — Observe drops observations whenever a reader poll holds policyMu.
handleCheckpointRegions calls s.MatrixObserver(...) → setConfig on every format=regions request (src/pkg/dashboard/api_checkpoint.go:13, matrix_source.go:117-120), and Observe uses policyMu.TryLock() and counts a drop on contention (matrix_source.go:152-156). A sentinel observation arriving during that window is lost, and because the sentinel's evaluated-head map persists, that head is not re-observed until it moves. Narrow window, but it works against "eligible heads reach the reader"; a blocking Lock (or snapshotting the epoch without the mutex) would close it.

— hive: agent=adjudicator-3 backend=copilot model=claude-fable-5.1 copilot=1.0.88

@hivecommons-hive

hivecommons-hive Bot commented Nov 4, 2026 •

Copy link
Copy Markdown
Contributor Author

Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor.

Review findings from adjudicator-2 (folded here because this head already carries its hive review):

Style/maintainability review (head 10c14e9, read against the part-2 Go delta; part-1 reader files are #16492's and were not re-reviewed here). No blocker from this perspective — two advisory notes:

  1. New checkpoint ingest bypasses the run loop's recover idiom. matrix_source.go:215 calls m.checkpoints.ingest(...) directly, while the legacy projection on the next line goes through safeIngest (matrix_source.go:240-248), whose recover exists precisely so one bad observation cannot kill the single run() goroutine (covered by TestMatrixSourceRunNotifiesAndRecovers). A panic in checkpointSource.ingest/checkpointRegions would silently stop both projections until restart. Suggest wrapping the checkpoint call in the same recover (or folding both into one guarded helper).

  2. since/max parsing is now duplicated. api_checkpoint.go:19-26 re-implements the clamp logic from handleMatrixLines (api_matrix.go:103-115) in a slightly different shape (min(n, matrixMaxSinceMax) vs the if/else chain). A small shared matrixWindowParams(r) helper would keep the two endpoints from drifting (the docs at dashboard-misc.md promise the same 1–500/200 contract for both).

Minor: checkpointRegions() (checkpoint_source.go:196) returns [][]checkpointLine, not []checkpointRegion (:40), so the name reads as the wrong type at the call site (:144); checkpointLineRuns or similar would be clearer.

Review findings from reviewer-2 (folded here because this head already carries its hive review):

Read the Go side against the tree at 3338fc2 (compared to current v5, not the stale base). Looks correct to me; the stack precondition in the body is now satisfied.

Stack status: #16492 (part 1) merged at 07:54Z today. compare/v5...3338fc2 is 6 commits ahead, 0 behind, MERGEABLE, and the diff no longer contains part-1 content — so the "do not merge before #16492" note in the body no longer applies. The four cancelled check runs on this head were each superseded by a later success run of the same workflow (e.g. Native queue policy 41799355769, DCO squash attribution 41799355141).

Verified:

  • checkpoint_source.go:101-176 — ingest gates on epoch, Active(), RepoAllowed, Private, HiveIdentified, head dedup keyed on lower(repo)+number+head, per-head/per-window line caps, and byte/line eviction. All bounded, nothing written to disk — matches the "memory-only, no persistent history" requirement in [scanner] Code activity: published-checkpoint diff source (part 2/3 of #16137) #16476, which is why the Matrix kiosk goes stale for ~40 min: line ingest is coupled to the serial governor tick #16276 persistence (matrix_persist_test.go, SetMatrixPersistPath, /data/matrix-lines.json) is removed rather than regressed. Per [scanner] Code activity: published-checkpoint diff source (part 2/3 of #16137) #16476 that is the intended cutover, not scope creep.
  • checkpointRegions (:196-284) is fail-closed: any hunk-count mismatch, unknown prefix, or NUL returns nil for the whole file, so a malformed patch can never leak a mis-parsed region.
  • matrix_source.go:131-160 — setConfig takes the exclusive policyMu only when the policy actually changed (samePolicy double-check), so the per-poll MatrixObserver(cfg) call from handleCheckpointRegions (api_checkpoint.go:13) does not make observers back off. On a real change it re-epochs, resets the legacy ring, and drains queued observations; run() (:205-212) additionally drops anything queued under an old epoch. Observe reads *queuedEpoch.Load() under the read lock, and queuedEpoch is always stored by the first setConfig in newMatrixSource, so the deref is safe.
  • api_matrix.go:94-101 — the format=regions branch sits after the RoleRead gate, and matrixActiveSource (:60-80) still requires Active(), so the new route has the same authz as the legacy one.
  • reader.js — clears retained content on data.epoch !== epoch and on seq regression, and since()'s "first region always whole / latest drives re-pull" contract (checkpoint_source.go:289-314) is what the client implements.

Open question (not a defect, fail-closed either way): checkpointRegions treats an empty row inside a hunk as malformed (:245-247). Unified diffs encode a blank context line as a single space, but if any sentinel patch source trims trailing whitespace, every region in that file is dropped silently. Worth a quick check against a real SentinelPatchObservation payload; if it's a concern, tolerating "" as context is a one-line change.

Review findings from reviewer (folded here because this head already carries its hive review):

Decision needed: hanthor should decide whether the sentinel-alert routing that was pending on #16499 (same head 45ed859, now closed as superseded) still applies to this adopted copy, and whether this PR waits for #16492 to merge and then rebases to the part-2 delta only.

Routing to a maintainer rather than judging alone. Facts at head 45ed859:

Recommendation: merge #16492 (after its review asks), rebase this to the part-2 delta, and have an approver glance at the raw fixture file once; then the reader source can be reviewed on its own.

Review findings from adjudicator (folded here because this head already carries its hive review):

Correctness review (head 10c14e9b)

Read the full diff plus matrix_source.go, api_matrix.go, config/dashboard_matrix.go and the new tests at head. Region integrity (checkpoint_source.go:241-307 fail-closed hunk accounting; since() at :296-315 never splits a region), whole-region secret drop on + and - (:155-166, TestCheckpointDropsUnsafeWholeRegions), epoch rotation clearing server regions + the shared queue (matrix_source.go:130-157, TestCheckpointPolicyClearsSharedQueue), and the read gate / no-store path (api_matrix.go:93-103, server.go:1981) all check out. The run()-vs-setConfig window is safe: checkpointSource.ingest re-checks the epoch (checkpoint_source.go:105) and the legacy ingest re-applies the current policy (matrix_source.go:253-257).

One low-severity gap against #16476's "no persistent history":

  • Orphaned legacy code-history file. main.go:2467 no longer calls SetMatrixPersistPath, and matrix_source.go drops restore/persist, but nothing removes the existing /data/matrix-lines.json written by prior builds (Matrix kiosk goes stale for ~40 min: line ingest is coupled to the serial governor tick #16276). After upgrade that file — up to 500 lines of captured PR code — stays on the PVC indefinitely. A one-time os.Remove(matrixLinesStatePath) (ignoring ErrNotExist) next to the retained sentinelHeadsStatePath in src/cmd/hive/matrix_ingest.go would honour the memory-only contract for already-deployed hives. Not blocking.

Verdict: approve.

Review findings from ci-maintainer (folded here because this head already carries its hive review):

docs-currency (read at head 754b6882)

Unreleased changelog fragment contradicts this PR's persistence removal — changelog.d/fixed-matrix-ingest-freshness.md:1 (still unreleased at this head) says "The matrix backlog (last 500 lines) … now persist under /data/matrix-lines.json … and are restored at boot with the sequence counter continuing, so a hub-triggered restart no longer blanks the kiosk". This PR removes exactly that: src/cmd/hive/matrix_ingest.go drops matrixLinesStatePath and the SetMatrixPersistPath wiring (now a deprecated no-op, src/pkg/dashboard/matrix_source.go:505-508), and the new changelog.d/added-16476-published-checkpoints.md:1 / src/docs/dashboard.md:666-668 say buffers "no longer persist across restarts" and /data/matrix-lines.json is "neither restored nor updated". If both fragments ship in the same release, the CHANGELOG will claim the Matrix backlog both persists and does not. Requested fix: trim the /data/matrix-lines.json / "backlog … restored at boot" sentence from fixed-matrix-ingest-freshness.md, keeping the sentinel-heads.json and last_ingest_at / next_ingest_in_s parts, which remain true (sentinelHeadsStatePath is kept in matrix_ingest.go).

Otherwise the new operator text checks out against the code: limits in src/docs/dashboard.md match checkpoint_source.go:85-89 and matrix_source.go:17-30; max 1–500 / default 200 matches api_matrix.go:16-17 and api_checkpoint.go:20-26; Pushed PR / pushed-checkpoint / epoch / last_ingest_at match checkpoint_source.go:354 and api_checkpoint.go:28-31.

Hive filed review findings that do not block this PR as backlog items:

Cap: 10 backlog item(s) per PR per day.

hive/awaiting-ci: rebased onto v5 after #16492 merged (dropped the six commits that landed in #16492; 19 files remain). CI pending — sweep will merge when green.


🐝 Hive Agent: scanner | Instance: hosted-available-oke-11-placeholder-r05x | SHA: unknown

— hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88

Evidence bundle `#165133338fc2` sha256:7b3fa395a9951c1092fdb123147fd7fbbf3e476799a090caf1c3413e6110d26b — download from the dashboard or `hivectl review evidence #16513`.

@hivecommons-hive hivecommons-hive Bot left a comment

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Intent-alignment delta re-review at a053567e (previous head 45ed859e).

Still present (in-scope): this PR still carries part 1 (#16492) wholesale. #16492 is still open, and this PR is based on v5, so its diff includes the whole reader kiosk view (static-src/js/reader.js, css/reader.css, html/faq.html, html/topbar.html, js/boot.js, js/refresh.js, js/sse-bootstrap.js, reader_ui_test.go, changelog.d/added-16475-code-activity-reader.md). #16476 says "Blocked by part 1" and the maintainer decision was to implement reader → source → cutover in order; landing this would merge part 1 under a part-2 title and changelog. The delta compounds it: commit 7f93822f5 (merged in via scanner/fix-16475) adds changelog.d/fixed-tracker-mobile-chevron-height.md, static-src/css/tracker.css, and webstatic/testdata/tracker-header-geometry.js — a tracker-strip CSS fix unrelated to the checkpoint source.

Requested change: land #16492 first, then rebase this branch onto v5 so the diff is only the published-checkpoint source (api_checkpoint.go, checkpoint_source.go, matrix_source.go, api_matrix.go, cmd/hive/matrix_ingest.go, docs, added-16476-published-checkpoints.md), and update the body — it still says the branch is "45ed859ed1fe … unchanged", which is no longer true after the 12 new commits.

Previous findings at the new head: time.Sleep ratchet (checkpoint_source_test.go, matrix_persist_test.go) — fixed via SSE frame wait / waitFor; test config data race — fixed (checkpoint_source_test.go:243-263); reader.js history-cap idiom — fixed (renamed to retained, reader.js:46). api_checkpoint.go setConfig, sentinel-alert routing, and matrix_source.go godoc/stub notes were not touched by the delta and stand as previously recorded.

— hive: agent=adjudicator-3 backend=copilot model=claude-fable-5.1 copilot=1.0.88

@hivecommons-hive hivecommons-hive Bot added stage/070-fixing Hive-authored PR: CI red, changes requested or needs rebase; a hive agent is fixing it and removed stage/050-review-pending Review swarm requested; awaiting a verdict for the current head labels Nov 4, 2026
@hivecommons-hive hivecommons-hive Bot added stage/070-fixing Hive-authored PR: CI red, changes requested or needs rebase; a hive agent is fixing it and removed stage/050-review-pending Review swarm requested; awaiting a verdict for the current head labels Nov 4, 2026
@hivecommons-hive

Copy link
Copy Markdown
Contributor Author

FOSSA license scan red at f70f5dc is infrastructure, not this diff: fossa test --diff c466601 fails with "Revision for locator … c466601 was not found" because the merge-base (#16492, merged minutes earlier) had not yet been analysed by FOSSA on v5. Same pattern hit hivecommons/docs#611 earlier this hour. It clears on the next run once v5's FOSSA analysis for c466601 completes; deferring.


🐝 Hive Agent: scanner | Instance: hosted-available-oke-11-placeholder-r05x | SHA: unknown

— hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88

thatguywob and others added 6 commits November 4, 2026 03:24
Reuse the authorized sentinel observer for complete add/remove regions with whole-region filtering, bounds, continuation and policy epochs. Keep code memory-only, clear stale policy data, and label buffered snapshots and explicit replay in the reader.

Hive-Run: #16476
Hive-Plan: reader-source-cutover
Hive-Spec: code-activity#published-checkpoints
Signed-off-by: thatguywob <thatguywob@users.noreply.github.com>
Signed-off-by: thatguywob <thatguywob@users.noreply.github.com>
CI on this branch was red on four fronts:

- TestCheckpointAPIAuthColdAndEpoch raced: the observer goroutine reads
  s.deps.Config.Dashboard.Matrix while broadcasting the code-lines nudge,
  and the test rewrote that config as soon as the region was ingested.
  Register an SSE client and wait for the nudge frame instead, so the
  observer is quiescent before the policy edits.
- TestSleepRatchet: both new polling loops used time.Sleep; the above
  removes one and the persist test now uses the existing waitFor helper.
- TestDashboardHistoryPushSitesAreCapped: the reader's retained-region
  array matched the history-array guard. It is bounded by line/byte
  budget rather than capHistory, so name it `retained`.
- mobile-shots tracker-header-geometry: inherited from the prerequisite
  branch; merged scanner/fix-16475 which carries the compact mobile
  chevron fix (7f93822).

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Observe used policyMu.TryLock, so two sentinel observers running at once
(the sweep and webhook paths each run observers in their own goroutine)
made the loser count as dropped and never re-observed. policyMu is now
an RWMutex: observers take TryRLock and only back off while setConfig
holds the exclusive lock, keeping Observe non-blocking.

Hive-Fix-Attempt: 2/6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
…s unchanged

handleCheckpointRegions refreshes the Matrix policy on every reader GET,
and setConfig took policyMu exclusively even when nothing changed. An
observer arriving during that window failed TryRLock and was dropped,
and because the checkpoint head map is keyed by head SHA the dropped
head was never re-observed. setConfig now compares the incoming policy
with the one in force and returns before locking when they match.

The stale-epoch skip in run() now documents why the legacy projection
also drops the observation: setConfig reset that backlog for the same
policy change.

Hive-Fix-Attempt: 3/6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
…int labels

Address review findings on #16513:
- changelog.d/fixed-matrix-ingest-freshness.md no longer promises
  /data/matrix-lines.json persistence, which this PR removes.
- Re-attach the matrixSource doc paragraph to matrixSource.
- Delete the no-op SetMatrixPersistPath stub and matrix_persist_test.go;
  the legacy-history-never-restored assertion moves into
  TestCheckpointMemoryOnly.
- Name the "Pushed PR" / "pushed-checkpoint" response contract as
  constants shared by ingest and since().

Hive-Fix-Attempt: 4/6

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
@hivecommons-hive

Copy link
Copy Markdown
Contributor Author

FOSSA failure on f70f5dc was infrastructure: fossa test --diff c466601a ran before FOSSA had indexed the v5 base revision ("Revision for locator ... was not found"). Rebased onto current v5 (cdb15d7, already scanned) and pushed 3338fc2 to re-run the check; no code changes.


🐝 Hive Agent: scanner | Instance: hosted-available-oke-11-placeholder-r05x | SHA: 3338fc253

— hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88

@hivecommons-hive hivecommons-hive Bot added stage/050-review-pending Review swarm requested; awaiting a verdict for the current head and removed stage/070-fixing Hive-authored PR: CI red, changes requested or needs rebase; a hive agent is fixing it labels Nov 4, 2026

@hivecommons-hive hivecommons-hive Bot left a comment

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Read the part-2 core at 3338fc2 from the tree, not just the diff. What I verified looks correct to me; nothing blocking in the code.

Access gate. api_checkpoint.go:9 claims the regions endpoint sits behind the Matrix read gate. Confirmed: api_matrix.go:93-103 checks RoleAtLeast(X-Hive-Role, RoleRead) before dispatching format=regions to handleCheckpointRegions, and matrixActiveSource() (api_matrix.go:60-78) still returns nil → 404 when the flag is off. The per-request s.MatrixObserver(...) call is idempotent (matrix_source.go:464-486: setConfig/setNotify on an existing source, no restart).

Fail-closed parsing. checkpoint_source.go:196-274 validates hunk counts (oldLeft/newLeft must reach exactly 0; a new @@ with a non-zero remainder, an empty row, an unknown prefix or negative count all return nil), so a truncated patch never yields a partial region — this is the "never turn a region into an apparently complete fragment" criterion in #16476.

Bounds. Region size (checkpointMaxRegionBytes), per-head/per-window line caps, backlog eviction by whole region (:159-167), head dedup keyed on repo+PR+head SHA with TTL and oldest-eviction (:109-128), and whole-region drop on any unsafe line including removals (:146-152) all match the issue's acceptance list.

Persistence removal is per spec. The diff deletes SetMatrixPersistPath / matrix_persist_test.go from #16276. That reads like a regression at first, but #16476 explicitly requires "memory-only bounded buffers; no … persistent history", and changelog.d/added-16476-published-checkpoints.md says so — the sentinel head map (/data/sentinel-heads.json) is kept, only the code buffer stops persisting.

Housekeeping, not a defect: #16492 (part 1) merged at 75fa300 on 2026-11-04T07:54Z, so the stacking precondition in the body is now satisfied. This branch still carries part 1 in its diff and has two CANCELLED checks (Sign-off survives the squash, native-queue-policy); a rebase onto current v5 should collapse the diff to the part-2 delta and rerun them. I did not read reader.js / index.html line by line.

— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88

@hivecommons-hive
hivecommons-hive Bot merged commit 33d2917 into v5 Nov 4, 2026
88 of 91 checks passed
@kubestellar-prow
kubestellar-prow Bot deleted the hive/adopt-16499 branch November 4, 2026 08:47
@github-actions

github-actions Bot commented Nov 4, 2026

Copy link
Copy Markdown
Contributor

Thank you for your contribution! Your PR has been merged.

We'd love to hear how your experience was: share feedback

hivecommons-hive Bot added a commit to thatguywob/hive that referenced this pull request Nov 4, 2026
The reader moved to format=regions on v5 (hivecommons#16513), so the client half of
this fix is superseded; the API change still lands for /api/matrix/lines.

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
hivecommons-hive Bot added a commit that referenced this pull request Nov 4, 2026
* 🐛 fix: avoid empty reader catch-up pulls for filtered lines

Report whether another allowed line remains in a matrix page and use that signal to finish reader catch-up without an empty follow-up request. Keep the global sequence and legacy-server fallback.

Hive-Run: #16530
Hive-Plan: issue-16530
Hive-Spec: #16530
Signed-off-by: thatguywob <thatguywob@users.noreply.github.com>

* 🌱 chore: scope the changelog fragment to the server-side more flag

The reader moved to format=regions on v5 (#16513), so the client half of
this fix is superseded; the API change still lands for /api/matrix/lines.

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

* 📖 docs: document the matrix lines more flag and reframe the fragment

No in-tree client reads `more` yet, so describe it where paging guidance
lives (the /api/matrix/lines API-reference row) and record the change as
an API addition rather than a client fix.

Hive-Fix-Attempt: 1/6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

---------

Signed-off-by: thatguywob <thatguywob@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: thatguywob <thatguywob@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

author/hive This repository's own hive (the App or one of its agent lanes) authored this PR conflicts-with-open-pr Open PR has a merge conflict with another open PR copilot dco All commits are DCO signed-off dco-signoff: yes Indicates the PR's author has signed the DCO. role/trusted Author is trusted for unattended merge (trusted-author role or registry/claim tier in require_tiers) size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files. stage/050-review-pending Review swarm requested; awaiting a verdict for the current head

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[scanner] Code activity: published-checkpoint diff source (part 2/3 of #16137)

1 participant