Repository navigation
[backport core/1.55] Complete 1.54 parity sweep - #19789
Conversation
…58) (#17486) *PR Created by the Glary-Bot Agent* --- Fixes the `scale_multiplier` widget on the `utility_seedvr2_3b_int8_upscale_image` template rendering as a disconnected input slot, with a duplicate `resize_type.multiplier_1` appended below it. Reproduces on current `main`; neither #16717 nor its backport #16751 touched the failing path. - Fixes FE-258 ## Root cause A node is constructed with its dynamic combo's **default** option laid out (`resize_type.width/height/crop`), and `ComfyNode.configure` rewrites `data.inputs` into that definition order, appending the serialized inputs the definition lacks. Serialized `target_slot` values count the **saved** layout, so link `50` — saved on `resize_type.multiplier` at slot 1 — lands on slot 1 of the rebuilt layout, which is a different input. Applying the combo's serialized value then tears the group down and hands each surviving link to the new input of the same name. The link is sitting on `resize_type.width`, which the selected option does not define, so it is dropped. Across a subgraph boundary the promoted input is demoted to a bare slot and a duplicate `_1` promotion is auto-created. ## Fix Realign the node's input links by name in `LGraphNode.configure`, after slots are wired and before widget values are applied — the window in which the correct slot exists. Scoped to nodes that have a group widget child input; `LGraph.configure`'s existing pass still runs for everything else once all nodes are configured. The node's ordinary inputs join the batch so a link still occupying a child's destination slot moves in the same atomic update instead of blocking it. Inputs owned by a group that is not a widget (autogrow) stay out because their slots grow and renumber during configure. ## Verification - `pnpm test:unit src/lib/litegraph/src/LGraph.inputSlotRealign.test.ts`: 21 passed. - `pnpm test:browser:local browser_tests/tests/subgraph/subgraphPromotedDynamicComboChild.spec.ts`: 2 passed after the development server warmed up. The first cold-start run timed out before the app became ready; its second test passed. - `pnpm typecheck`, `pnpm typecheck:browser`, `pnpm lint`, `pnpm format:check`, and `pnpm knip` passed. ## Unverified residuals - An excluded autogrow link can block the early batch and produce a `Failed to realign input link slots` log. This PR does not include a complete-config base/head test for the final topology in that case. - A nested dynamic combo under a non-default outer option is excluded because its widget does not exist at this point in configure. Base/head parity for that case was not reverified in this review round. ## Screenshots   --------- Co-authored-by: Glary-Bot <glary-bot@users.noreply.github.com> Co-authored-by: Amp <amp@ampcode.com> (cherry picked from commit 3f29ced)
The automatic "Meet Comfy Agent" card opened on top of whatever dialog was already up, including the desktop sign-in approval, so people had to dismiss it before they could approve. It now waits until no dialog is open and shows in the same visit. - **What**: - The automatic consent offer holds while any dialog from the dialog store is open (desktop sign-in approval, sign-in, invite links, errors, the startup template browser) and offers once the last one closes. - It retries once, and only when an offer was actually held, so ordinary dialog closes don't re-read consent. - Once someone has seen the card this visit, from the Agent button or on its own, the automatic offer is done until the next visit. Skipping a card opened from the button no longer brings it straight back. - The Agent button is unchanged and still opens the card right away. - The hold reads `dialogStack.length`, the rule the first-run nudge already uses, not `isModalOpen`. `isModalOpen` also sees modals outside the store, which the retry can't see close, so hold and release would disagree. The trade-off: a lightbox open at the exact moment of the offer won't hold it. - The startup template browser now comes first and the card waits for it to close. That was a product call and it's agreed. - `CONSENT_DIALOG_KEY` is exported from `useAgentConsent` so the extension can tell when the card itself is open. - Unit, `agentPanel.test.ts`: hold, same-session release, a dialog opening mid-offer, a tour ending while a dialog is still open, manual Skip not re-popping, the card seen this visit, and a failing consent read not retried on every dialog close. These fail against the old code. - E2E, `agentConsent.spec.ts` (cloud): boots on a desktop sign-in link; the approval is on top and clickable, and the card appears only after approving. Red 5/5 on the old bundle, green 5/5 on the fix. - Mutation testing on the changed lines: 6/6 killed. - Fixes FE-2808 (cherry picked from commit c10ba49)
<!-- ccr-slack-attribution --> _Requested via [Slack thread](https://comfy-organization.slack.com/archives/C09KFK2ABB8/p1790161872557349)_ ## Summary Fixes the pattern behind Sentry `CLOUD-FRONTEND-STAGING-4MF`/`4MG`: the unified-remint retry fetch could spuriously time out under staging load because it inherited whatever was left of the original request's 60s deadline instead of getting a full one. ## Changes - **What**: `fetchApi()` (`src/scripts/api.ts`) armed a single 60s `AbortSignal` and passed it straight through to `fetchWithUnifiedRemint()`. On a 401, that function re-mints the unified Cloud JWT (`tryRemintToken` → `workspaceAuthStore.remintUnifiedOnce()` → `unifiedUser()` → `unifiedSessionClient.remint()` → `exchangeToken()`, each step with its own independent ~15s ceiling) before retrying the original request — but the retry reused the *same* original signal, so its effective budget was `60s minus whatever the re-mint round trip already spent`, not a fresh 60s. Under load, a slower-than-usual re-mint could leave the retry only a few seconds before it was aborted with `TimeoutError: Fetch timeout`, even though every individual step was within its own budget. `fetchWithUnifiedRemint()` (`src/platform/auth/unified/remintRetry.ts`) now takes an optional 4th parameter, `freshRetrySignal?: () => AbortSignal | undefined`, called only if a retry actually happens; its return value replaces the original `init.signal` on the retry `fetch`. `fetchApi()` passes a callback that mints a new `AbortController` + `setTimeout` (same duration, same `DOMException('Fetch timeout', 'TimeoutError')` construction as the existing timeout, so Sentry grouping stays consistent) combined with the caller's own `signal` via `AbortSignal.any`, and cleans up the new timer in the existing `.finally()`. `fetchWithUnifiedRemint()` has two other call sites (`globalSettingsApi.ts`, `authStore.ts`) that pass no signal at all today; the new parameter is optional and defaults to `undefined`, which keeps their behavior byte-for-byte unchanged (they can't produce this `TimeoutError` in the first place — only a benign `TypeError: Failed to fetch`). - **Breaking**: none - **Dependencies**: none ## How - `remintRetry.ts`: build the retry `fetch`'s `init` from `freshRetrySignal()` when supplied, instead of always spreading the original `init` (which carried the shrunk signal). - `api.ts`: `freshRetrySignal` is only constructed when `timeout` is non-null (i.e. `timeoutMs` wasn't explicitly disabled), and only actually invoked by `fetchWithUnifiedRemint` if a 401 retry happens — a request that never gets a 401 pays no extra cost. ## Review Focus - `src/scripts/api.ts`, `fetchApi()`: the new `freshRetrySignal` closure and its timer's cleanup in `.finally()` (no leaked timers on either the happy path or an error path). - `src/platform/auth/unified/remintRetry.ts`, `fetchWithUnifiedRemint()`: the retry `init` construction now branches on whether `freshRetrySignal` was supplied. - Confirmed the two other callers (`globalSettingsApi.ts`, `authStore.ts`) are unaffected — they don't pass a signal today, and the new parameter is optional and off by default. ## Tests - `src/platform/auth/unified/remintRetry.test.ts`: new cases assert the retry `fetch` receives the object returned by `freshRetrySignal()` (not the original `init.signal`) when supplied, and that the callback is never called when no retry happens. - `src/scripts/api.fetchApi.test.ts`: new regression test with fake timers simulates a 401 → 55s re-mint → 30s retry sequence. Verified (via `git stash` of only the source changes, keeping the new tests) that both new tests fail against the pre-fix code — the retry test rejects with the abort from the stale signal at t=60s — and pass once the fix is restored. Fixes the pattern behind Sentry `CLOUD-FRONTEND-STAGING-4MF`/`CLOUD-FRONTEND-STAGING-4MG`. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_0137jzrG6RSp5e6YgGp8kxdY --- _Generated by [Claude Code](https://claude.ai/code/session_0137jzrG6RSp5e6YgGp8kxdY)_ --------- Co-authored-by: dante01yoon <bunggl@naver.com> Co-authored-by: Amp <amp@ampcode.com> (cherry picked from commit e346ef9)
## Summary Reject stale Agent workflow bindings when the target ID has a duplicated cloud name instead of resolving the unrelated saved tab and graph. ## Changes - **What**: Resolve ID-to-name identity from the full cloud index while retaining the duplicate-filtered name-to-ID inference map. Add fresh-store regressions for both resolver entry points and a browser scenario with duplicate target IDs plus a uniquely named saved `Portrait` tab. - **Dependencies**: None. ## Review Focus The lookup is intentionally asymmetric: duplicated names remain unsafe for name-to-ID inference, while an exact target ID can still retrieve its indexed name for stale-binding rejection. This is separate from the refresh-cache work in #17509 and does not modify that branch. Focused Vitest is red on `origin/main` (both entry points select `Portrait`) and green here (14/14). ESLint, formatting, full `pnpm typecheck`, browser typecheck, and the push-time `knip` gate pass. The new Playwright case is checked in but runtime proof is blocked on this worker: both it and the unchanged `agentWorkflowSelection` smoke case time out before assertions because the Agent entry button is absent on the local current-main surface. No live model call was made. ## Screenshots (if applicable) The browser run produced traces/video for the pre-assertion harness blocker; no passing screenshot is claimed. (cherry picked from commit e5a6ab0)
…8693) Internal testers reported friction from having to choose a target workflow before chatting. This PR selects the visible workflow automatically and makes the target behavior clear: - Fresh chats follow tab changes, keeping the typed draft, until a successful manual target selection, a node reference, or Send. Workflow references alone do not stop following. - These actions retain the target; Send does so even if the request fails. Users can still change targets explicitly. - New Chat resumes following unless the carried draft contains node references; those keep their original target. Reopened conversations retain their target. - The existing composer tip explains following or a visible/target mismatch. **Show target** opens the target without changing it. (cherry picked from commit facbca7)
… (PM-1227) (#18061) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1789585023130639?thread_ts=1789585023.130639&cid=C0BGH348Z0C)_ ## Summary Removing a node's reference chip from the agent composer left the node highlighted on canvas. This fixes that, and adds the e2e regression test that covers it. **This PR carries the fix, not just a repro.** It started as a `test.fail()`-pinned repro; commits `3564898d6`, `fab0b2c13` and `d0af7f19f9` added the fix, removed the pin, and moved the spec to asserting the corrected behaviour. The title and this body have been corrected to match — earlier revisions described it as test-only. ## Changes - **What**: - `src/workbench/extensions/agent/AgentPanelRoot.vue` (+6) — `onRemoveSelectionTag` now resolves the node the chip refers to out of the current canvas selection and deselects it after removing the chip. - `browser_tests/tests/agent/agentNodeSelectionChipRemoval.spec.ts` (+65) — black-box regression spec. Enters agent node-selection mode, captures the unselected canvas, stages a node as a composer chip, removes the chip, and waits for the rendered canvas to return to its unselected pixels. - `browser_tests/fixtures/helpers/CanvasHelper.ts` (+25) — owns the shared canvas-view readiness check used by this spec and the composer placeholder regression spec. - **Breaking**: none - **Dependencies**: none ## Root cause `onRemoveSelectionTag` in `AgentPanelRoot.vue` called `removeSelectionTag(id)` and nothing else. That `remove()` (from `useCanvasSelection`) only filters the composer's local `staged` list; it has no canvas dependency, so the node's LiteGraph selection survived. The only other deselect on that surface is in `exitNodeSelectionMode()`, which runs when leaving node-selection mode entirely — never on a single chip removal. **Before**: selecting a node on canvas while in agent node-selection mode adds a reference chip and highlights the node. Removing that chip clears the chip but leaves the node highlighted. **After**: removing the chip also deselects the node, and the highlight clears. The fix is at the panel rather than inside `remove()` deliberately: `useCanvasSelection.ts` imports only `vue` and a type, and the composable is generic over staged selections. Pushing `canvas.deselect` into it would give it a canvas dependency it does not currently have. ## Review Focus - **The deselect goes through the canonical API**, not a direct store mutation: `canvas.deselect(node)` → `setCanvasItemSelected(..., false)` → the selection store that `canvasStore.selectedItems` is computed from. Commit `d0af7f19f9` is what moved it onto that API; an earlier revision used `canvas.deselectAll()` / `canvasStore.updateSelectedItems()` and those are gone. - **This is the only call site.** `removeSelectionTag` is referenced exactly three times in `src/`, all in `AgentPanelRoot.vue` — the destructure, this function, and the `@remove-tag` binding. There is no second path that can leave a node lit. - **Ordering** (`removeSelectionTag` before `deselect`) through `useCanvasSelection`'s `flush: 'sync'` watcher: single-chip — `remove()` empties `staged` and latches `dismissedSig`, then the deselect fires the watcher with an empty selection, clearing `dismissedSig` again. Multi-chip — the watcher re-stages the remaining nodes from the new selection. Consistent in both. - The spec reuses existing agent e2e infra (`agentTest`/`agentPanel` from `browser_tests/tests/agent/agentPanelMocks.ts`, plus `comfyPage.nodeOps`/`NodeReference`) and the shared `CanvasHelper` readiness check rather than adding spec-local fixtures or polling. **Verification note.** CI ran the original state-based spec green in `playwright-tests (cloud, frontend-dist-cloud)`; 91/91 checks were green at `d0af7f19f9`. The revised rendered-canvas assertion was verified red by temporarily removing `canvas.deselect(node)`, then green after restoring it. The focused cloud run and the neighboring composer placeholder spec both pass locally at `24025ec535`. ## Screenshots (if applicable) N/A (no visual change beyond the highlight clearing) 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_014TL6Z2fYevRQcChLMRy6Pk --- _Generated by [Claude Code](https://claude.ai/code/session_014TL6Z2fYevRQcChLMRy6Pk)_ --------- Co-authored-by: bymyself <cbyrne@comfy.org> Co-authored-by: Amp <amp@ampcode.com> (cherry picked from commit e968f00)
## Summary - add a persistent play badge to generated video reply tiles - leave image and 3D reply tiles unmarked - preserve hover playback and gallery inspection behavior Linear: PM-1306, PM-1315 ## Testing - `pnpm vitest run src/workbench/extensions/agent/components/agent/message/ReplyAssetGroup.test.ts` (15 passed) - ESLint and formatting on changed files - `pnpm typecheck` - pre-commit and pre-push checks Justification: The author completed a focused self-review after the PR waited five days without human review; the change is a 33-line visual affordance limited to video tiles, covered by focused component tests and fully green CI. (cherry picked from commit fef4b2b)
…8093) ## Summary - Let an explicit node resize override a stale content-size measurement so an auto-grown Flux2ImageNode can shrink again. - Preserve ResizeObserver measurements across hidden/visible tab transitions; genuinely changed nodes re-report, unchanged siblings do not. - Convert the existing PM-1304 / PM-1312 expected-failure unit, observer, and real-browser regressions into passing assertions. Linear: PM-1304, PM-1312 ## Test plan - `pnpm exec vitest run src/renderer/core/layout/operations/graphLayoutAttachment.test.ts src/renderer/extensions/vueNodes/composables/useVueNodeResizeTracking.test.ts` (18 passed) - Full app and browser typechecks via pre-commit - ESLint, oxlint, and oxfmt via pre-commit - CI runs the two real Flux2ImageNode Playwright cases against cloud Justification: confirmed fix for the two linked resize regressions, with red-before/green-after coverage already present on this PR. --------- Co-authored-by: bymyself <cbyrne@comfy.org> (cherry picked from commit 0e8b66a)
|
Important Review skippedIgnore keyword(s) in the title. ⛔ Ignored keywords (2)
Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Repository: Comfy-Org/ComfyUI_frontend/.coderabbit.yaml Review profile: ASSERTIVE Plan: Team Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Comment |
🎭 Playwright: ✅ 2142 passed, 0 failed · 2 flaky📊 Browser Reports
🎨 Storybook: ✅ Built — View Storybook🌾 Fallow: ❌ FailedFallow audits the diff for duplication, complexity and dead code. It gates on new findings only, so when the audit completes, a red is something this PR introduced or moved — never inherited backlog. A red can also mean the audit itself did not finish; the details below say which happened. Fallow gates on new findings in changed files only, so everything below was introduced or moved by this PR (55 changed files).
Pre-existing findings are baselined in Reproduce locally# the base the gate compared against
npx fallow audit --root . --quiet --format json \
--changed-since 54232444637b758d78a6aa31282633db8111145b
Full docs: ADR-DEVEX-LINT-0015 📦 Bundle Size
⚡ Performance Report
Show regressions
All metrics
Historical variance (last 15 runs)
Trend (last 15 commits on main)
Raw data{
"timestamp": "2026-10-01T22:11:40.688Z",
"gitSha": "1f837022302cf9c3ecb82ad161e94f9afb67e5db",
"branch": "christian/complete-core-1.55-backports",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 1997.7449999999521,
"styleRecalcs": 12,
"styleRecalcDurationMs": 10.72,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 485.81199999999995,
"heapDeltaBytes": 6304800,
"heapUsedBytes": 73311424,
"domNodes": -256,
"jsHeapTotalBytes": 5496832,
"scriptDurationMs": 9.393000000000002,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-idle",
"durationMs": 2023.311999999919,
"styleRecalcs": 9,
"styleRecalcDurationMs": 8.966999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 465.7939999999999,
"heapDeltaBytes": 15090328,
"heapUsedBytes": 77124304,
"domNodes": -261,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 9.459999999999999,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1774.5459999999866,
"styleRecalcs": 46,
"styleRecalcDurationMs": 16.499,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 822.6329999999999,
"heapDeltaBytes": 10643748,
"heapUsedBytes": 78238768,
"domNodes": -260,
"jsHeapTotalBytes": 6021120,
"scriptDurationMs": 164.70100000000002,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1775.4710000000387,
"styleRecalcs": 46,
"styleRecalcDurationMs": 16.024,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 827.8489999999999,
"heapDeltaBytes": 6246184,
"heapUsedBytes": 73919312,
"domNodes": -261,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 166.26,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1685.495000000003,
"styleRecalcs": 29,
"styleRecalcDurationMs": 15.321,
"layouts": 6,
"layoutDurationMs": 1.131,
"taskDurationMs": 367.73600000000005,
"heapDeltaBytes": 3089956,
"heapUsedBytes": 70777604,
"domNodes": 77,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 13.03,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1707.6090000000477,
"styleRecalcs": 31,
"styleRecalcDurationMs": 16.897000000000002,
"layouts": 6,
"layoutDurationMs": 1.177,
"taskDurationMs": 280.669,
"heapDeltaBytes": 3002660,
"heapUsedBytes": 71064816,
"domNodes": 77,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 12.375999999999998,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 691.3469999999506,
"styleRecalcs": 14,
"styleRecalcDurationMs": 9.940999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 483.10600000000005,
"heapDeltaBytes": 18888024,
"heapUsedBytes": 86470620,
"domNodes": 24,
"jsHeapTotalBytes": 5505024,
"scriptDurationMs": 177.431,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-clipping",
"durationMs": 667.5660000000789,
"styleRecalcs": 11,
"styleRecalcDurationMs": 7.467,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 459.25399999999996,
"heapDeltaBytes": 18860560,
"heapUsedBytes": 86606160,
"domNodes": 18,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 167.95399999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1170.471999999961,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 829.7130000000001,
"heapDeltaBytes": 38241764,
"heapUsedBytes": 107389584,
"domNodes": 0,
"jsHeapTotalBytes": 18509824,
"scriptDurationMs": 70.719,
"eventListeners": 0,
"totalBlockingTimeMs": 559,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1183.334000000059,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 851.168,
"heapDeltaBytes": 38418204,
"heapUsedBytes": 103814352,
"domNodes": 0,
"jsHeapTotalBytes": 19034112,
"scriptDurationMs": 75.62299999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 571,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 844.9469999999906,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 469.8859999999998,
"heapDeltaBytes": 14626080,
"heapUsedBytes": 91331628,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 167.067,
"eventListeners": 0,
"totalBlockingTimeMs": 231,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.80000000000109
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 863.6220000000776,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 471.42799999999954,
"heapDeltaBytes": 14921500,
"heapUsedBytes": 90151176,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 168.40800000000002,
"eventListeners": 0,
"totalBlockingTimeMs": 244,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "extension-clean-combined-draw",
"durationMs": 11093.074,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10201.08,
"heapDeltaBytes": -8248612,
"heapUsedBytes": 73711468,
"domNodes": -12,
"jsHeapTotalBytes": 9961472,
"scriptDurationMs": 8812.305999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 2071,
"frameDurationMs": 18.890000000000025,
"p95FrameDurationMs": 50
},
{
"name": "extension-clean-combined-draw",
"durationMs": 11240.318000000003,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10137.963000000002,
"heapDeltaBytes": 4826944,
"heapUsedBytes": 86667556,
"domNodes": -14,
"jsHeapTotalBytes": 8650752,
"scriptDurationMs": 8706.122000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 1776,
"frameDurationMs": 18.890000000000025,
"p95FrameDurationMs": 49.89999999999782
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 10855.91599999998,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10018.479,
"heapDeltaBytes": -966688,
"heapUsedBytes": 80782144,
"domNodes": -10,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 8268.911999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 1174,
"frameDurationMs": 18.886666666666677,
"p95FrameDurationMs": 50
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 10766.790000000015,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 9731.307999999999,
"heapDeltaBytes": 4680556,
"heapUsedBytes": 81608456,
"domNodes": -12,
"jsHeapTotalBytes": 8126464,
"scriptDurationMs": 8310.219,
"eventListeners": 0,
"totalBlockingTimeMs": 1510,
"frameDurationMs": 18.886666666666617,
"p95FrameDurationMs": 49.899999999999636
},
{
"name": "large-graph-idle",
"durationMs": 2041.1700000000224,
"styleRecalcs": 7,
"styleRecalcDurationMs": 7.417,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 583.195,
"heapDeltaBytes": -3878316,
"heapUsedBytes": 83659108,
"domNodes": -245,
"jsHeapTotalBytes": -495616,
"scriptDurationMs": 34.177,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-idle",
"durationMs": 2056.8150000000287,
"styleRecalcs": 9,
"styleRecalcDurationMs": 12.948,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 616.8960000000001,
"heapDeltaBytes": -6399548,
"heapUsedBytes": 81329284,
"domNodes": -246,
"jsHeapTotalBytes": -233472,
"scriptDurationMs": 34.403000000000006,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66999999999998,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-pan",
"durationMs": 2313.6269999999968,
"styleRecalcs": 70,
"styleRecalcDurationMs": 14.191000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1669.8029999999999,
"heapDeltaBytes": -7666420,
"heapUsedBytes": 80386272,
"domNodes": -240,
"jsHeapTotalBytes": 516096,
"scriptDurationMs": 876.262,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-pan",
"durationMs": 2403.9960000000065,
"styleRecalcs": 68,
"styleRecalcDurationMs": 12.94,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1692.212,
"heapDeltaBytes": -6611708,
"heapUsedBytes": 80611676,
"domNodes": -245,
"jsHeapTotalBytes": 1040384,
"scriptDurationMs": 872.197,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "large-graph-zoom",
"durationMs": 3545.236999999986,
"styleRecalcs": 63,
"styleRecalcDurationMs": 12.673999999999998,
"layouts": 60,
"layoutDurationMs": 13.374,
"taskDurationMs": 1694.4820000000002,
"heapDeltaBytes": 8746464,
"heapUsedBytes": 74569904,
"domNodes": -251,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 800.662,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-zoom",
"durationMs": 3548.411999999985,
"styleRecalcs": 63,
"styleRecalcDurationMs": 12.490000000000002,
"layouts": 60,
"layoutDurationMs": 13.474999999999998,
"taskDurationMs": 1767.5030000000002,
"heapDeltaBytes": -15747356,
"heapUsedBytes": 73801248,
"domNodes": -259,
"jsHeapTotalBytes": -495616,
"scriptDurationMs": 790.74,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3220.1800000000276,
"styleRecalcs": 45,
"styleRecalcDurationMs": 9.793,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 2047.3079999999998,
"heapDeltaBytes": -2530440,
"heapUsedBytes": 87966184,
"domNodes": -224,
"jsHeapTotalBytes": 1757184,
"scriptDurationMs": 1012.7919999999999,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "legacy-node-drag",
"durationMs": 3145.5849999999828,
"styleRecalcs": 42,
"styleRecalcDurationMs": 6.471000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1981.375,
"heapDeltaBytes": 1741032,
"heapUsedBytes": 67700076,
"domNodes": 4,
"jsHeapTotalBytes": 3932160,
"scriptDurationMs": 1027.5910000000001,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333338,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2013.4449999999902,
"styleRecalcs": 5,
"styleRecalcDurationMs": 5.037,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 550.4699999999999,
"heapDeltaBytes": 20185124,
"heapUsedBytes": 84326524,
"domNodes": -250,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 33.089000000000006,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "minimap-idle",
"durationMs": 2045.5740000001015,
"styleRecalcs": 9,
"styleRecalcDurationMs": 7.57,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 495.9260000000001,
"heapDeltaBytes": -7942196,
"heapUsedBytes": 80975516,
"domNodes": -246,
"jsHeapTotalBytes": 552960,
"scriptDurationMs": 41.361,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 593.4219999999755,
"styleRecalcs": 48,
"styleRecalcDurationMs": 12,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 410.56000000000006,
"heapDeltaBytes": 13877084,
"heapUsedBytes": 83153492,
"domNodes": 21,
"jsHeapTotalBytes": 5505024,
"scriptDurationMs": 147.948,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 606.8490000000111,
"styleRecalcs": 48,
"styleRecalcDurationMs": 12.504000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 412.342,
"heapDeltaBytes": 13803576,
"heapUsedBytes": 76888564,
"domNodes": 21,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 151.188,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-idle",
"durationMs": 1990.5660000000012,
"styleRecalcs": 11,
"styleRecalcDurationMs": 9.59,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 358.849,
"heapDeltaBytes": 9746720,
"heapUsedBytes": 79705700,
"domNodes": -276,
"jsHeapTotalBytes": 5496832,
"scriptDurationMs": 7.552,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-idle",
"durationMs": 2009.3400000000656,
"styleRecalcs": 12,
"styleRecalcDurationMs": 11.024,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 375.685,
"heapDeltaBytes": -6205244,
"heapUsedBytes": 63695884,
"domNodes": -273,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 8.668,
"eventListeners": -220,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1734.815000000026,
"styleRecalcs": 45,
"styleRecalcDurationMs": 14.838000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 782.558,
"heapDeltaBytes": 6693336,
"heapUsedBytes": 76633884,
"domNodes": -275,
"jsHeapTotalBytes": 6021120,
"scriptDurationMs": 111.30599999999998,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1707.7010000000428,
"styleRecalcs": 46,
"styleRecalcDurationMs": 16.206000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 763.879,
"heapDeltaBytes": 8007908,
"heapUsedBytes": 77862684,
"domNodes": -277,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 109.179,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-transition-enter",
"durationMs": 1050.8379999999988,
"styleRecalcs": 18,
"styleRecalcDurationMs": 29.493999999999993,
"layouts": 5,
"layoutDurationMs": 13.141,
"taskDurationMs": 776.2600000000002,
"heapDeltaBytes": -6327068,
"heapUsedBytes": 84523832,
"domNodes": 13667,
"jsHeapTotalBytes": 11534336,
"scriptDurationMs": 29.400000000000002,
"eventListeners": 2383,
"totalBlockingTimeMs": 145,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "viewport-pan-sweep",
"durationMs": 8657.00400000003,
"styleRecalcs": 249,
"styleRecalcDurationMs": 37.221000000000004,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 6114.659,
"heapDeltaBytes": 1842360,
"heapUsedBytes": 81688912,
"domNodes": -221,
"jsHeapTotalBytes": 1757184,
"scriptDurationMs": 2962.386,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "viewport-pan-sweep",
"durationMs": 8571.870999999988,
"styleRecalcs": 250,
"styleRecalcDurationMs": 39.896,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 5955.796,
"heapDeltaBytes": -6071004,
"heapUsedBytes": 80610344,
"domNodes": -244,
"jsHeapTotalBytes": 1302528,
"scriptDurationMs": 2834.302,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333338,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-idle",
"durationMs": 16801.750000000025,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 16484.457,
"heapDeltaBytes": -29814712,
"heapUsedBytes": 184721316,
"domNodes": -8061,
"jsHeapTotalBytes": -8110080,
"scriptDurationMs": 252.59200000000004,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666763,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-idle",
"durationMs": 17462.75800000001,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 16868.483,
"heapDeltaBytes": -48660112,
"heapUsedBytes": 176706168,
"domNodes": -8061,
"jsHeapTotalBytes": -17833984,
"scriptDurationMs": 274.331,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.223333333333358,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-pan",
"durationMs": 21384.181999999953,
"styleRecalcs": 158,
"styleRecalcDurationMs": 19.382000000000012,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 21159.656000000003,
"heapDeltaBytes": -48637768,
"heapUsedBytes": 192506268,
"domNodes": -8061,
"jsHeapTotalBytes": -22028288,
"scriptDurationMs": 910.8609999999999,
"eventListeners": -14928,
"totalBlockingTimeMs": 46,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-pan",
"durationMs": 16191.594000000008,
"styleRecalcs": 88,
"styleRecalcDurationMs": 13.927999999999995,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 15730.447000000002,
"heapDeltaBytes": -48504308,
"heapUsedBytes": 192589756,
"domNodes": -8062,
"jsHeapTotalBytes": -12853248,
"scriptDurationMs": 837.387,
"eventListeners": -14928,
"totalBlockingTimeMs": 2,
"frameDurationMs": 17.223333333333358,
"p95FrameDurationMs": 16.80000000000291
},
{
"name": "vue-node-resize-workload",
"durationMs": 6251.535999999987,
"styleRecalcs": 20,
"styleRecalcDurationMs": 77.011,
"layouts": 20,
"layoutDurationMs": 23.507,
"taskDurationMs": 6183.765,
"heapDeltaBytes": -49377724,
"heapUsedBytes": 177588532,
"domNodes": -8061,
"jsHeapTotalBytes": 11640832,
"scriptDurationMs": 111.98200000000001,
"eventListeners": -14928,
"totalBlockingTimeMs": 444,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-node-resize-workload",
"durationMs": 6260.605999999939,
"styleRecalcs": 20,
"styleRecalcDurationMs": 84.623,
"layouts": 20,
"layoutDurationMs": 23.09499999999999,
"taskDurationMs": 6137.737,
"heapDeltaBytes": -63164624,
"heapUsedBytes": 176844184,
"domNodes": -8061,
"jsHeapTotalBytes": 5349376,
"scriptDurationMs": 155.394,
"eventListeners": -14928,
"totalBlockingTimeMs": 437,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.80000000000291
},
{
"name": "vue-zoom-out-idle",
"durationMs": 8645.533999999998,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8486.508,
"heapDeltaBytes": 12730016,
"heapUsedBytes": 182859100,
"domNodes": 0,
"jsHeapTotalBytes": -15728640,
"scriptDurationMs": 134.83,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.77333333333336,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-zoom-out-idle",
"durationMs": 8529.297000000042,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8350.919,
"heapDeltaBytes": 5261196,
"heapUsedBytes": 192085324,
"domNodes": 0,
"jsHeapTotalBytes": -15204352,
"scriptDurationMs": 133.82499999999996,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "workflow-execution",
"durationMs": 490.73299999997744,
"styleRecalcs": 20,
"styleRecalcDurationMs": 29.997999999999998,
"layouts": 5,
"layoutDurationMs": 2.4699999999999998,
"taskDurationMs": 150.08899999999997,
"heapDeltaBytes": 5250056,
"heapUsedBytes": 70814556,
"domNodes": 155,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 18.435999999999996,
"eventListeners": 81,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "workflow-execution",
"durationMs": 479.29199999998673,
"styleRecalcs": 21,
"styleRecalcDurationMs": 25.871,
"layouts": 4,
"layoutDurationMs": 1.8190000000000002,
"taskDurationMs": 131.08599999999998,
"heapDeltaBytes": 5042624,
"heapUsedBytes": 70563408,
"domNodes": 150,
"jsHeapTotalBytes": 524288,
"scriptDurationMs": 10.93,
"eventListeners": 81,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
}
]
}🔬 E2E Coverage
Top 10 files by uncovered lines
🧩 Custom Nodes Ecosystem Matrix: ✅ PassedExecutes the frontend JS of every registry pack that ships any, against this change, in both renderers. The eight
On a pin-bump PR, a red gate means the ecosystem moved — not that your diff broke something. That is what the bump PR is for. Outside a pin bump or a cold rebuild of the unpinned tail, a red means the diff. Note the corpus size and the executed count are different numbers, and the verdict is against the executed count: roughly 3,100 of ~5,100 packs ship no extension-shaped JS and are skipped. Reproduce locally# build one shard's fixture, then run it
python3 scripts/registry-census/build_matrix.py --shard 1/4
MATRIX_RENDERER=vue MATRIX_VUE=1 \
npx vitest run --config vitest.matrix.config.mts
Full docs: scripts/registry-census/README.md |
This comment has been minimized.
This comment has been minimized.
🔧 Auto-fixes AppliedThis PR has been automatically updated to fix linting and formatting issues.
Changes made:
|
…rd guard The consolidated carrier landed #18693's product change but dropped its AgentPanelRoot.test.ts block, leaving core/1.55's pre-#18693 assertions in place. Two of them contradict the shipped behaviour and fail at this head: requires explicit selection on first entry even with an unsaved canvas blocks graph node references until the visible workflow is selected Carry the source block, which replaces those two plus "keeps the selected Agent target when the visible graph tab changes" with the twelve follow-the-visible-workflow cases core/1.54 already has. One release-branch adaptation: the persisted-thread key is the flat 'Comfy.Agent.ThreadId' here, not main's workspace-scoped StorageKeys.agentThread - same adaptation core/1.54 uses. #18619 was carried as the dialog_open branch only. Its consentScope / consentCardSeenIn guard, present on core/1.54, was dropped while CONSENT_DIALOG_KEY was still exported for it, leaving the export with no consumer. Restore the guard so a card the user has already seen does not get re-offered, with the two cases that are red without it. Excluded deliberately: the holdOffer and screenBusyReason extractions (behaviourally identical to the inlined form already on this branch), the activeTour watcher removal (subsumed by the existing offerHeld/screenIsClear whenever), and "remembers a seen card per workspace across a switch away and back" - that one passes with and without the guard here, because this branch's localStorage one-shot already covers the per-workspace case. Validation: AgentPanelRoot.test.ts 326/326, agentPanel.test.ts 63/63, vue-tsc clean, oxfmt/oxlint/eslint/knip clean.
Brings the carrier to the live release tip so the target-aware merge gate can compare ancestry against it. #19646 touches billingOperationStore only; zero file overlap with this carrier.
christian-byrne
left a comment
There was a problem hiding this comment.
Independent parity review of this carrier
Reviewed at 1e0304fac9, repaired at 324d9bb96d, now 500d73734c. Method: compared each carried commit's patch against its main patch and against the core/1.54 carrier, then ran the affected suites.
Gap premise holds
All eight source PRs are present on core/1.54 and absent from core/1.55. Six carried patches are byte-identical to their main patch (#17486, #18599, #18155, #18061, #18392, #18093). Two diverge, and both divergences were real:
1. #18693 landed the product change without its test update — the required test check was red
main's #18693 replaces three assertions that encode the pre-fix behaviour. The carrier took the product change and dropped the 520-line AgentPanelRoot.test.ts block, so two of those assertions survived and now contradict the shipped behaviour. Measured at 1e0304fac9:
× requires explicit selection on first entry even with an unsaved canvas (:4146)
× blocks graph node references until the visible workflow is selected (:8017)
Tests 2 failed | 308 passed (310)
Carried the source block: those two plus keeps the selected Agent target when the visible graph tab changes are replaced by the twelve follow-the-visible-workflow cases core/1.54 already has. One release-branch adaptation — the persisted-thread key is the flat 'Comfy.Agent.ThreadId' here, not main's workspace-scoped StorageKeys.agentThread(...); that is the same adaptation core/1.54 uses. Now 326/326.
2. #18619 was carried as the dialog_open branch only
Its consentScope / consentCardSeenIn guard — which stops the automatic offer re-presenting a card the user has already seen — was dropped, while CONSENT_DIALOG_KEY was still exported for it, leaving that export with no consumer. The guard is on core/1.54 (021ecb8c88) and absent from core/1.55, so this was a live parity hole rather than a deliberate trim. Restored, matching core/1.54 verbatim, with the two cases that are red without it:
× does not re-offer after the user declines a manually opened card
× skips the automatic offer for the session once the user has seen the card
Both green with it; agentPanel.test.ts 63/63.
Deliberate exclusions
src/locales/en/main.jsonfrom #18693 — all four keys are already oncore/1.55. Verified, no drift.holdOffer/screenBusyReasonextractions — behaviourally identical to the inlined form already on this branch.- Removal of the
activeTourwatcher — subsumed by the existingofferHeld && screenIsClearwhenever; leaving it only widens the set of calls that the guards already filter. remembers a seen card per workspace across a switch away and back— not carried: it passes with and without the guard here, because this branch'slocalStorageone-shot already covers the per-workspace case. A test that cannot fail is worse than no test.
Other checks
audit-fixture-closure.py: 0 verbatim specs over a diverged fixture closure.oxfmt --checkwas failing ate0d35fba94(a stray indent inuseAgentSession.ts); CI auto-fixed it in1e0304fac9.- Merged the live
core/1.55tip (#19646) in rather than rebasing — no conflict exists, so history stays intact. Zero file overlap with this carrier. fallowis red on inherited complexity inReplyAssetGroup.vueandlayoutStore.ts, and the co-author-trailer check is red on trailers copied from the four upstream cherry-picks. Neither is a required check oncore/1.55(required:lint-and-format,test,e2e-status).
Local: vue-tsc clean, oxfmt/oxlint/eslint/knip clean on the changed files.
Playwright videos for new testsDownload the recorded videos and HTML report Recorded projects: chromium cloud Recorded test titles:
|
Codecov Report❌ Patch coverage is @@ Coverage Diff @@
## core/1.55 #19789 +/- ##
=============================================
+ Coverage 84.67% 84.70% +0.03%
=============================================
Files 2315 2315
Lines 131896 132028 +132
Branches 35203 33370 -1833
=============================================
+ Hits 111684 111837 +153
+ Misses 19853 19832 -21
Partials 359 359
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 8 files with indirect coverage changes 🚀 New features to boost your workflow:
|
#18693 renames the store's `workflowTargetSelection` ref to `targetTracking`, which stranded the duplicate-Pinia-id pin in agentPanelStoreId.test.ts: an earlier lane had retargeted that assertion at `workflowTargetSelection` precisely because this branch had no `targetTracking`, and the carrier reintroduces it. The required `test` check was red on this one file (1 failed / 1681) with expected [ 'enabled', 'consentAccepted', ... ] to include 'workflowTargetSelection' Name `targetTracking`, which is what main's copy of this pin names. The pin's purpose is unchanged - any ref the gate-only shape lacks proves the full panel store was resolved. Validation: agent store suites 190/190, oxfmt and eslint clean.
CI follow-up: one more unit defect fixed, and e2e is the real blocker
|
| Spec | Source | Symptom |
|---|---|---|
agentConsent.spec.ts:582 |
#18619 | toHaveCount failed |
agentDuplicateWorkflowNameBinding.spec.ts:28 |
#18155 | toBeVisible — element(s) not found |
agentNewChatTargetWorkflow.spec.ts:24, :96 |
#18693 | toHaveText failed |
agentWorkflowSelection.spec.ts:108 + 2 |
#18693 | toHaveText failed |
These survived three CI retries each, so not flake.
What this is not. I checked mechanically whether the product code is incomplete, taking every substantive line each source's main patch adds under src/ (excluding tests) and asserting its presence at this head:
| Source | lines added | absent here |
|---|---|---|
| #18693 | 152 | 5 |
| #18619 | 49 | 9 |
| #17486 / #18599 / #18093 / #18392 / #18061 / #18155 | 36 / 46 / 24 / 12 / 6 / 2 | 0 each |
All 14 absences are accounted for: #18693's four previousWorkflowId threading lines (this branch's commitWorkflowTarget is three statements and has no consumer for the parameter) and a threadStorageKey → THREAD_STORAGE_KEY rename; #18619's holdOffer/screenBusyReason extractions, which are behaviourally identical to the inlined form already here, and a firstRunTookScreen → firstRunHoldsScreen rename.
So the product change is complete. Three things I also ruled out: the dropped previousWorkflowId threading, the useAgentSession trim (semantically equivalent — both hoist stored under the same messages.length === 0 guard), and a missing initialUrl harness fixture (core/1.54 had to rewrite the consent spec around its absence; core/1.55 has it at ComfyPage.ts:600, so taking main's version verbatim was right). audit-fixture-closure.py reports no diverged fixture closure.
That leaves the specs or the cloud harness. They were taken verbatim from main and may assert behaviour depending on main-only changes outside these eight sources. Diagnosing it needs the cloud project run locally (frontend-dist-cloud build + ComfyUI server + agent mocks), which I have not done — so I am not merging this, and the product-completeness table above is meant to save whoever does from re-walking the wrong hypothesis first.
122ba7a to
3a88432
Compare
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: Request changes. Seven of the eight carried PRs are faithful, but #18619 drops four of its unit tests that pass on this branch with a one-token fix, and e2e-status is red at this head.
This is a multi-source carrier, so I compared each carried PR's merged diff on main against this branch's net diff from its merge base, 5423244 (the current core/1.55 tip):
- #17486 (3f29ced), #18599 (e346ef9), #18093 (0e8b66a), #18061 (e968f00), and #18392 (fef4b2b): every file carries exactly the source's added and removed lines.
AgentPanelRoot.vueequals #18061's and #18693's hunks combined.- #18155 (e5a6ab0) is exact, except that
agentDuplicateWorkflowNameBinding.spec.tsopens the panel throughAgentPanel.open(). That is the same adaptationcore/1.54uses, because activation can already have opened the panel. - #18693 (facbca7) carries these adaptations, each accounted for by the target:
- The locale keys are already on
core/1.55, so the hunk is dropped. commitWorkflowTargetdrops theonTargetBound/previousWorkflowIdtelemetry, because that telemetry (#18665) is not on this branch.useAgentSession.tsreads the flatTHREAD_STORAGE_KEY, andAgentPanelRoot.test.tssets'Comfy.Agent.ThreadId', because workspace-scoped storage is not here.agentPanelStoreId.test.tsis retargeted attargetTracking, the ref #18693 introduces.- The
agentWorkflowSelectionandagentNewChatTargetWorkflowspecs force the send to fail withfailNextWorkflowMessage()where the source relies on the stub's non-ack reply to keep the composer text. - The consent e2e block now matches the
mainandcore/1.54version that follows #19154, which is on this branch through #19631.
- The locale keys are already on
- #18619 (c10ba49): the production change is adapted onto the target's existing
screenHolder/offerHeld/screenIsClearstructure, which came from #19660, and theconsentScope/consentCardSeenInguard is restored. The test coverage is short, as noted inline.
Checks at 3a88432:
- Locally,
vitest runover all 16 touched unit-test files passed 594/594, andvue-tsc --noEmitpassed. - Hosted
e2e-statusis red becauseplaywright-video-new-testsfailed onagentDuplicateWorkflowNameBinding.spec.ts:186. The assertiontopbar.getActiveTab()).toContainText('Portrait')received "Unsaved Workflow". The same case passed inplaywright-tests (cloud, frontend-dist-cloud), where it was not listed as flaky. That job did list #18061'sagentNodeSelectionChipRemoval.spec.tsas flaky. So this looks timing-sensitive rather than a hard failure, but the required aggregate is red and needs a green run before this lands.
| expect(localStorage.getItem(AUTO_SHOWN_KEY)).toBeNull() | ||
| }) | ||
|
|
||
| it('offers in the same session once the dialog that held it closes', async () => { |
There was a problem hiding this comment.
#18619 adds eight cases to this suite, and core/1.54 and main both carry all eight. This branch carries four. The commit message for 0528925 explains only one omission, "remembers a seen card per workspace across a switch away and back", as redundant here. The other three are dropped without explanation:
- "withholds a card whose dialog opened while the offer was in flight, then re-offers"
- "withholds a card whose Getting Started screen took over while the offer was in flight"
- "does not re-read consent on every dialog close while the read keeps failing"
I inserted all four from core/1.54 here. With the mock's async (_trigger, _onAccept, hooks) changed to async (_onAccept, hooks) to match this branch's two-argument withConsent, the suite passes 67/67 at this head. Please carry them with that signature adaptation.
There was a problem hiding this comment.
Carried in bd1adcf34c, all four, with the async (_onAccept, hooks) signature you specified. Suite goes 63 → 67 passing at the new head; no production file changed, so carrier source parity is untouched.
I also checked each restored case is load-bearing here rather than just passing, by mutating the guard it covers:
Mutation to src/extensions/core/agentPanel.ts |
Cases that fail |
|---|---|
canShow: () => true (drops the in-flight screen-takeover re-check) |
withholds … dialog opened … and withholds … Getting Started screen took over …, plus the 2 already here |
drop offerHeld.value && from the held-offer whenever |
does not re-read consent on every dialog close while the read keeps failing — and nothing else |
consentScope() returns ${userId} instead of ${userId}.${workspaceId} |
remembers a seen card per workspace across a switch away and back — and nothing else |
So two of the four are the only detectors of their behaviour on this branch. That makes 0528925's "redundant here" note for the per-workspace case wrong: this branch had no other coverage of workspace-scoped seen-card memory, and dropping the consentScope workspace component is invisible without it. Thanks for catching the omission.
Also note the branch previously covered only the tour holder for the in-flight withhold; the dialog and Getting Started holders were uncovered.
vue-tsc --noEmit, oxfmt --check, oxlint and eslint all pass at bd1adcf34c.
| ]) | ||
|
|
||
| const topbar = new Topbar(page) | ||
| await expect(topbar.getActiveTab()).toContainText('Portrait') |
There was a problem hiding this comment.
This assertion failed in playwright-video-new-tests at this head ("Expected substring: "Portrait", Received string: "Unsaved Workflow""), which leaves e2e-status red. It passed in the cloud project run. The spec matches core/1.54's copy, so I am not asking for a code change here. I am asking for a green required run, or a look at whether the history restore that switches tabs can race the AgentPanel.open() adaptation above.
There was a problem hiding this comment.
Agreed, no code change. Two things, one of which is a concrete reason for the job asymmetry you observed rather than a guess.
Why only this job saw it. playwright-video-new-tests is the only required E2E job that runs with retries disabled — .github/workflows/ci-tests-e2e.yaml:488 invokes playwright test --project=cloud --timeout=180000 --retries=0, while playwright-tests (cloud, …) takes the config default, retries: process.env.CI ? 3 : 0 (playwright.config.ts:31), so three. And e2e-status requires the video job (:268, :270). So the video job is the only required job that cannot mask a retry-recoverable flake, and it additionally records video on every action (video: 'on'), which slows each step. The cloud project run passed without being listed flaky, i.e. on its first attempt — so this is intermittent, not systematic, and the retry policy is what decides which job reports it. Its per-test timeout is more generous there (180s vs the cloud project's 15s), so a tighter test timeout is not the explanation; the failure is an expect timeout on the assertion itself.
On your race hypothesis. AgentPanel.open() is already hardened against the activation toggle — browser_tests/fixtures/components/AgentPanel.ts:92-105 re-reads visibility and retries once precisely because "startup activation can open between the visibility read and click". So the adaptation itself is not the unguarded side. What is suspicious is the direction of the failure: toContainText retried for its full window and saw "Unsaved Workflow" throughout, after the preceding binding assertion had already confirmed tabPath === PORTRAIT_PATH and store.openWorkflow(portrait) had resolved. That reads as something re-activating the default tab late, not as Portrait never being activated. I could not reproduce it locally to name that writer, so I am flagging it as a candidate rather than a finding — the honest statement is that the spec has a retry-masked timing sensitivity that only the zero-retry job exposes.
A fresh full run is in flight at bd1adcf34c; I will report the e2e-status result here.
There was a problem hiding this comment.
Green at bd1adcf34c, as asked. All three checks the core/1.55 ruleset actually requires are now success at the exact head:
e2e-status— successtest— successlint-and-format— success
playwright-video-new-tests passed, including agentDuplicateWorkflowNameBinding.spec.ts, so the run that was red at 3a8843296c is green without any change to that spec — consistent with the retry-policy reading above rather than a hard failure. playwright-tests (cloud, …) and all 16 chromium shards also passed. The only red check at head is fallow, which is not in the ruleset's required set (lint-and-format, test, e2e-status) and reports pre-existing complexity/coverage findings in carried product files; clearing it would mean editing carried code and breaking backport parity.
So the PR is MERGEABLE with required CI green and two approvals, blocked only on this review. Your test finding is implemented in bd1adcf34c; whenever you have a moment to re-review, that would clear the last gate. I have not touched your verdict.
Commit 0528925 carried four of #18619's eight agentPanel.test.ts cases and explained only one omission. core/1.54 and main carry all eight. Restores the remaining four, adapted to this branch's two-argument `withConsent(onAccept, hooks)` signature: - withholds a card whose dialog opened while the offer was in flight, then re-offers - withholds a card whose Getting Started screen took over while the offer was in flight - remembers a seen card per workspace across a switch away and back - does not re-read consent on every dialog close while the read keeps failing The suite goes 63 -> 67 passing at this head. No production file changes, so release-carrier source parity is unaffected. Each restored case was proved non-vacuous by mutating the guard it covers: forcing `canShow` to return true kills both `withholds` cases; dropping the `offerHeld` gate on the held-offer re-read kills the repeated-read case and nothing else; reducing `consentScope` to the user id kills the per-workspace case and nothing else. So the per-workspace case is the only detector of workspace-scoped seen-card memory on this branch, not redundant here.
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: Approve. This is a faithful carrier of the eight core/1.54-only fixes onto core/1.55, with each adaptation accounted for by the target, and all required checks are green.
I compared each carried PR's merged diff on main against this branch's net diff from its merge base, 5423244 (the current core/1.55 tip). bd1adcf carries #18619's four remaining agentPanel.test.ts cases. They match core/1.54 except that the withConsent mock is async (_onAccept, hooks), which matches this branch's two-argument signature. That resolves my earlier comment.
- Exact: #17486 (3f29ced), #18599 (e346ef9), #18093 (0e8b66a), #18061 (e968f00), and #18392 (fef4b2b).
AgentPanelRoot.vueequals #18061's and #18693's hunks combined. - #18155 (e5a6ab0) is exact, except that
agentDuplicateWorkflowNameBinding.spec.tsopens the panel throughAgentPanel.open(). That iscore/1.54's adaptation, because activation can already have opened the panel. - #18619 (c10ba49):
- The
dialog_openreason and theconsentScope/consentCardSeenInguard are adapted onto the target's existingscreenHolder/offerHeld/screenIsClearstructure, which came from #19660. - All eight source tests are present, with the
withConsentsignature adaptation. - The consent e2e block matches the
mainandcore/1.54version that follows #19154, which is on this branch through #19631.
- The
- #18693 (facbca7):
- The locale keys are already on
core/1.55, so that hunk is dropped. commitWorkflowTargetdrops theonTargetBound/previousWorkflowIdtelemetry, because #18665 is not on this branch.- The flat
THREAD_STORAGE_KEY/'Comfy.Agent.ThreadId'replaces the workspace-scoped key. agentPanelStoreId.test.tsis retargeted at thetargetTrackingref that #18693 introduces.- The workflow-selection specs force the send to fail with
failNextWorkflowMessage(), so the composer keeps its text as the source assertions expect.
- The locale keys are already on
Checks at bd1adcf:
- Locally,
agentPanel.test.tspassed 67/67 andvue-tsc --noEmitpassed. At the previous head, all 16 touched unit-test files passed 594/594, and only this one test file has changed since. - Hosted: the required
lint-and-format,test, ande2e-statuschecks passed, as didplaywright-tests (cloud, frontend-dist-cloud)andplaywright-video-new-tests. The duplicate-name case that failed in the video run last time passed here. - The non-required
fallowaudit fails on complexity thresholds in carried source code, for exampleReplyAssetGroup.vue's template andlayoutStore.tshandleClearGraph. Fixing that would mean diverging from the sources, so I am not treating it as a backport issue.
|
Exact-head merge judgment for |
Summary
Carries the fixes that landed on
core/1.54but were absent fromcore/1.55:This is the consolidated
core/1.55carrier from the 1.54→1.55 release audit.Validation