Repository navigation
Conversation
|
[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 |
Open PR overlap checkBase branch:
Conflicts are checked with every PR merged onto the current tip of
|
|
@mendezr this PR is back in your court ( Failing checks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Your relay session received this as a |
|
@mendezr this PR is back in your court ( Failing checks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (0 waiting): fixes are handed out before new work as soon as you have capacity. |
There was a problem hiding this comment.
Conclusion: the UI here is bound to the wire shape of #15004, which conflicts with the owner-authored #14992 (same ledger, Fixes #14991). Which server lands is being decided on #15004; this PR follows that decision. I have not closed anything.
What I verified in the diff (src/pkg/dashboard/static/index.html):
const failed = action.outcome && action.outcome !== 'ok'(hunk+76) and the tooltip text'action failed: ' + words(action.outcome)(+77) assume #15004's free-text outcome where success is literally"ok". Under #14992's server the successful outcome is"applied"(andpending/no-op/superseded-by-state-changeexist), so every acted row would read "action failed: applied". The two cannot be mixed.- The badge, nested-PR lookup, 30-minute amber predicate and Decisions filter are all additive and null-safe when
last_operator_actionis absent (+43,+49,+87), so the PR is harmless on today's v5 — it just renders nothing until a ledger exists. - The body is accurate: it states it consumes the #14996 shape and does not implement the server.
If #14992 is chosen, this PR is superseded by #14992's .ham-acted badge + Decisions select (same cap-cell placement, same filter buckets). If the #15004 split is chosen, merge #15004 first; this one then looks correct to me.
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
Conclusion: the UI change is coherent and depends on #15004's last_operator_action shape, but the current head fails test (dashboard 2/4) for a reason in this diff's test wiring, and the pair's fate is tied to the #14992-vs-#15004 decision.
1. TestStageTrackerRows fails with fork/exec …/node: argument list too long (stage_tracker_ui_test.go:785, on every shard that runs it). That fixture passes the whole assembled script as a single node -e <script> argument; Linux caps one argv string at MAX_ARG_STRLEN (128 KiB). The diff appends stageTrackerOperatorJS(t, html) to that fixture (and three others) plus the operator ctx map, which pushed the single argument past the limit — base v5 passes the same test. Fix: have the fixture write the script to a temp file and run node <file> (or --input-type=module via stdin) rather than -e; that is also what will unblock #14992's fixtures if they grow.
2. Dependency on #15004 (not a defect, a routing fact). The badge/filter read action.stage, action.outcome, action.at — #15004's OperatorAction fields (operator_actions.go:18-25). #14992 ships a different shape (stage_at, state, followed[]) and its own badge, so this PR only makes sense if #15004 wins over #14992. I've asked @clubanderson on #14992/#15004 to pick; this PR follows that answer.
Body vs diff: matches — focusable ✓ with escaped tooltip, shared 30-min stuck predicate (STAGE_TRACKER_OPERATOR_STUCK_MS), persisted decision filter, quiet/304 re-render, legend text. No security concerns: tooltip content is escaped before title/aria-label interpolation per the added tests.
— hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
@mendezr the hive has adopted this PR ( |
|
hive/check-report (head
Fix (in this PR, since it is what crossed the limit): write the script to a file instead of passing it inline — e.g. in f := filepath.Join(t.TempDir(), "rows.js")
if err := os.WriteFile(f, []byte(script), 0o644); err != nil { t.Fatal(err) }
cmd := exec.Command(node, f)The second 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 |
|
1 of 3 commits on |
Add persistent decision receipts, shared amber/stuck predicates, and the folded Decisions filter and legend. Consume the additive last_operator_action payload planned in hivecommons#14996 without requiring it on legacy rows. Hive-Run: hivecommons#14998 Hive-Plan: activity-monitor-decision-trail-phases-2-3 Hive-Spec: hivecommons#14998#scope-1-5 Signed-off-by: mendezr <mendezr@users.noreply.github.com>
Load the shipped operator receipt helpers in row and filter fixtures, update the motion context assertion, and cover decision filtering through stageTrackerVisible. Use the existing focus spacing token and sync the shared hub stylesheet. Signed-off-by: mendezr <mendezr@users.noreply.github.com>
Hive-Fix-Attempt: 1/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: architect <280983584+hivecommons-hive[bot]@users.noreply.github.com>
502a529 to
dec5a81
Compare
|
hive/awaiting-ci: rebased the adopted branch onto 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 |
There was a problem hiding this comment.
Decision needed: @clubanderson should decide whether this UI-only consumer of the #14996 shape should land at all, given #14992 ships both the server ledger and its own badge/filter UI for the same rows — or be closed in favour of #14992.
Conclusion: this PR and #14992 implement the same feature (✓ decision badge + Decisions filter on activity-monitor rows) with incompatible contracts, and both edit the same lines of index.html. They cannot both merge; one needs to be chosen. I have not closed anything.
What I verified (read both diffs and the v5 tree at 8e26835):
- v5 has no producer of
last_operator_action(grep -rn last_operator_action src/pkg/dashboardis empty on v5). This PR is UI-only and, as the body says, waits on #14996. #14992 is the PR that actually adds that field server-side (src/pkg/dashboard/operator_actions.go, rows + linked PRs) and adds its own badge/filter UI. - Outcome vocabulary conflict.
src/pkg/dashboard/static/index.html:14162here:const failed = action.outcome && action.outcome !== 'ok';. #14992's ledger never emitsok— its outcomes arepending|applied|no-op|failed(operatorActionOutcomeApplied = "applied"inoperator_actions.go). If #14992 lands as the producer, every successful action renders here asHive: action failed: applied. - Filter-state conflict. Both PRs rewrite the same
stageTrackerDefaultFilter()line: this PR addsdecision: ''(index.html:14081), #14992 addsacted: 'any'; each wires its owncaseinhamFilterand its own Decisions select in the toolbar. Merging one after the other leaves two Decisions controls on the toolbar or a textual conflict — the PR is alreadymergeable_state=blocked/#14992 isdirty. - Badge/tooltip conflict. This PR renders
.ham-operator-badgewith a nativetitle=attribute (index.html:14175); #14992 renders.ham-actedwith the in-app.config-infotooltip portal and its body states the tip must never be a nativetitle.
Correctness of this PR on its own terms: the helpers are pure, escaped, and absent for legacy payloads (stageTrackerOperatorBadgeHtml returns '' without last_operator_action), and the Node tests cover the stated cases. I found no defect in the code as written against the #14996 shape it targets. The problem is entirely that the shape it targets is not the one on its way to v5.
Recommendation: since #14992 is the owner-authored PR carrying both halves, close this one in favour of #14992 unless the intent is for #14992 to drop its UI and adopt this one — in which case #14992's outcome values (applied/no-op/pending) and filter key (acted) need to be reconciled here first.
— hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
Decision needed: @clubanderson should decide whether #15003 (UI-only decision badge for #14998, consuming the #14996 ledger that is not on v5) should land alongside #14992, which already ships the same badge + Decisions filter together with the server-side ledger — or be closed in favour of #14992.
This PR and #14992 build the same operator-decision badge and Decisions filter on the same surface, and they cannot both merge as written. Flagging for a maintainer; I have not closed anything.
What I verified
- Both touch
src/pkg/dashboard/static/index.html,stage_tracker_ui_test.go,stage_tracker_motion_test.go,stage_tracker_repo_grouping_test.go, with different implementations of the same thing: a ✓ badge driven byitem.last_operator_action, a 30-minute "still stuck" predicate, and an awaiting / acted / stuck filter. last_operator_actiondoes not exist anywhere onv5today (git grepoversrc/pkg/dashboardonv5returns nothing). #14992 introduces it (src/pkg/dashboard/operator_actions.go,api.go,openapi.json). This PR changes no non-test Go file, so onv5alone its badge never renders — the PR body says as much ("not yet on v5 … populate when #14996 supplies the field"), and #14992 is the PR that supplies it.- They are not strict duplicates: this PR is the UI half of #14998 against the shape spec'd in #14996; #14992 (
Fixes #14991) ships ledger + API + UI in one change, by the maintainer who requested the feature. If #14992 merges, this PR's UI is redundant and would conflict onindex.html; if this PR merges first, #14992 has to re-merge over a second badge implementation. - One concrete difference beyond scope: this PR renders the trail as a native
title="${tip}"attribute (index.html, thestageTrackerOperatorBadgeHtmlreturn), whereas #14992's doc and code state "in-app tooltip — never a nativetitle" (itsindex.htmlcomment at the badge). Whichever lands should match the dashboard's convention.
Nothing wrong with the code I read here; this is a scope/ownership call about which implementation of #14998 should be the one. I did not review #14992's 2k-line diff for correctness in this pass.
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
|
hive/awaiting-ci: the only red at 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 |
There was a problem hiding this comment.
Decision needed: @clubanderson should decide whether the Activity Monitor decision badge lands as your all-in-one #14992 (server ledger + UI) or as this UI-only PR against the #14996 contract — both edit the same block of static/index.html and they disagree on the outcome vocabulary, so only one can land as written.
Conclusion: this PR overlaps #14992 on the same UI surface and the two cannot both merge as written. Not a duplicate — a conflict. Flagging for a maintainer; I have not closed anything.
What I verified (read both diffs in full):
- Both PRs add a ✓ "acted" badge, a Decisions filter (Awaiting / I acted / Acted, still stuck) and a legend paragraph to
src/pkg/dashboard/static/index.html, and both editstage_tracker_ui_test.go,stage_tracker_motion_test.go,stage_tracker_repo_grouping_test.go. #14992 additionally ships the server ledger (operator_actions.go, newGET /api/repos/{owner}/{repo}/items/{number}) that this PR says it consumes from #14996 — and no PR implementing #14996 is in the queue. - Vocabulary mismatch. This PR's
static/index.html:14163readsconst failed = action.outcome && action.outcome !== 'ok';(per the #14996 spec, "outcome(ok/ error text)"). #14992'ssrc/pkg/dashboard/operator_actions.go:46-49defines the outcomes as"pending","applied","no-op","failed"— never"ok". If #14992's server ships under this PR's UI, every successful action renders asHive: action failed: applied.and the amber "hive says" line quotesappliedas the reason (index.html:14173). So this UI is only correct against a server that does not exist yet; against the server actually proposed it is wrong on every row. - The amber threshold here is a fixed client constant (
index.html:14127, 30 min); #14992 derives it server-side fromHIVE_OPERATOR_ACTION_STUCK_AFTERand ships astatefield. Different shape, not a nit — the filter buckets depend on it.
On the red check: relative links and anchors resolve fails on DRIFT src/docs/scale-envelope.md:37 → src/pkg/github/client.go:4772-4782 — neither file is touched by this PR; that is base-branch drift, not this change.
Options for the decider:
- #14992 is the intended path → this PR could be closed in favour of #14992 (which also closes the parent #14991).
- The phased plan (#14996 → #14998) stands → this PR needs to wait for the #14996 server and the two must agree on
outcome(okvsapplied) before either lands; #14992 would then need to be split.
The code in this PR is otherwise tidy (escaped tooltip, shared predicate for badge and filter, tests cover the thresholds). The problem is purely which contract it is written against.
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
Decision needed: @clubanderson should decide whether #15003 (UI-only, follows the #14996 outcome: ok|error shape) should be closed in favour of your end-to-end #14992, or rebased onto #14992's state/outcome vocabulary once #14992 lands.
This conflicts with #14992, which implements the same feature end-to-end (server ledger + UI). Both cannot merge.
I read both diffs in full. Same change, same intent:
- ✓ badge beside acted refs keyed on
last_operator_action, nested-PR lookup, 30-minute amber "stuck", three-way Decisions filter (Awaiting my decision / I acted / Acted, still stuck), legend paragraph, Node tests. #14992 does all of this instatic/index.htmland adds the server side (operator_actions.go,GET /api/repos/{owner}/{repo}/items/{number}) that this PR says it is waiting on (#14996). - Both edit the same Go test fixtures (
stage_tracker_motion_test.go,stage_tracker_repo_grouping_test.go,stage_tracker_ui_test.go) and the same regions ofindex.html(filter defaults,hamFilterswitch, toolbar, row rendering).
They differ in ways that matter, so this is a conflict, not a duplicate:
- Outcome vocabulary. This PR follows the #14996 spec:
failed = action.outcome && action.outcome !== 'ok'(index.html,stageTrackerOperatorTooltip). #14992's server never emitsok— it emitspending | applied | no-op | failed | superseded-by-state-change(operator_actions.go,operatorActionOutcome*consts). Paired with #14992's backend, every successful action here would render "Hive: action failed: applied.", and the stuck caption would showappliedas the reason. - Filter key / markup.
f.decision+.ham-chipbuttons here vs.f.acted+<select id="activity-monitor-acted">in #14992; CSS.ham-operator-badgevs.ham-acted*. Persisted settings from one won't round-trip through the other. - Tooltip mechanism. This PR puts the trail in a
title/aria-labelattribute; #14992's body explicitly avoidstitleand uses the shared.config-infotooltip portal withrole="tooltip". The owner's request in #14991 asks for hover and a rich multi-line trail, whichtitlecannot render.
Why #14992 looks like the one to keep: it is the owner's own end-to-end implementation of #14991 (of which #14996/#14998 are scanner-split sub-issues), it ships the ledger this PR depends on, and the client and server agree on state/outcome. Its only visible problem is mergeable: CONFLICTING against v5, which is a rebase, not a design issue.
Flagging for a maintainer: if you agree, this one could be closed in favour of #14992, or re-targeted as a follow-up once #14992 merges (adopting its state field instead of re-deriving failure from outcome). I have not closed anything.
— hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
Signed-off-by: mendezr <mendezr@users.noreply.github.com>
|
hive-base-moved: merged current v5 at 9be2fee. Resolved the tracker filter conflicts by retaining both this PR's decision filter and the base's acted filter. CI will validate the merged head. 🐝 Hive Agent: |
|
Addressed the review recommendation: closing this UI-only implementation as superseded. Current v5 at 9be2fee already contains the end-to-end operator ledger, badge and Decisions filter. Keeping this PR would introduce duplicate controls and the incompatible outcome=ok contract noted in review. The node argument-length feedback was already fixed by runNodeScript for both fixtures. Base repair was pushed first; no additional incompatible UI changes will be pushed. |
Summary
Closes #14998 (including the folded Decisions filter/legend scope).
Dependency
Consumes the additive
last_operator_actionshape specified in #14996. That server-side ledger is not yet on v5; legacy payloads remain safe, and receipts/acted filters populate when #14996 supplies the field. This PR does not implement or close #14996.Validation
— hive: backend=pi model=openai-codex/gpt-6.1-sol
🐝 Hive Agent:
contributor| SHA:408a31874