Repository navigation
test: repro agent-CRDT reconcile overwriting live node/title/widget state - #18075
Conversation
|
Important Review skippedBot user detected. To trigger a single review, invoke the ⚙️ Run configurationConfiguration used: Repository: Comfy-Org/ComfyUI_frontend/.coderabbit.yaml Review profile: ASSERTIVE Plan: Advanced Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Comment |
🎨 Storybook: ✅ Built — View Storybook🎭 Playwright: ✅ 2127 passed, 0 failed · 1 flaky📊 Browser Reports
📦 Bundle: 9.99 MB gzip 🔴 +24 BDetailsSummary
Category Glance App Entry Points — 3.77 kB (baseline 3.77 kB) • ⚪ 0 BMain entry bundles and manifests
Status: 1 added / 1 removed Graph Workspace — 1.45 MB (baseline 1.45 MB) • ⚪ 0 BGraph editor runtime, canvas, workflow orchestration
Status: 2 added / 2 removed / 1 unchanged Views & Navigation — 134 kB (baseline 134 kB) • ⚪ 0 BTop-level views, pages, and routed surfaces
Status: 14 added / 14 removed / 4 unchanged Panels & Settings — 639 kB (baseline 639 kB) • ⚪ 0 BConfiguration panels, inspectors, and settings screens
Status: 10 added / 10 removed / 18 unchanged User & Accounts — 39.4 kB (baseline 39.4 kB) • ⚪ 0 BAuthentication, profile, and account management bundles
Status: 7 added / 7 removed / 4 unchanged Editors & Dialogs — 126 kB (baseline 126 kB) • ⚪ 0 BModals, dialogs, drawers, and in-app editors
Status: 7 added / 7 removed / 1 unchanged UI Components — 125 kB (baseline 125 kB) • ⚪ 0 BReusable component library chunks
Status: 7 added / 7 removed / 9 unchanged Data & Services — 3.97 MB (baseline 3.97 MB) • 🔴 +115 BStores, services, APIs, and repositories
Status: 13 added / 13 removed / 3 unchanged Utilities & Hooks — 576 kB (baseline 576 kB) • 🔴 +1 BHelpers, composables, and utility bundles
Status: 22 added / 22 removed / 14 unchanged Vendor & Third-Party — 17.4 MB (baseline 17.4 MB) • ⚪ 0 BExternal libraries and shared vendor chunks Status: 18 unchanged Other — 17.9 MB (baseline 17.9 MB) • ⚪ 0 BBundles that do not match a named category
Status: 81 added / 81 removed / 220 unchanged ⚡ Performance Report
Show regressions
All metrics
Historical variance (last 15 runs)
Trend (last 15 commits on main)
Raw data{
"timestamp": "2026-09-24T21:25:00.516Z",
"gitSha": "3b8b1feb0d82524d03df43089f65225ae94c68aa",
"branch": "test/agent-crdt-reconcile-overwrite-repro",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 2022.4079999999844,
"styleRecalcs": 8,
"styleRecalcDurationMs": 4.308000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 526.4960000000001,
"heapDeltaBytes": 9691364,
"heapUsedBytes": 71775620,
"domNodes": -228,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 12.011000000000001,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-idle",
"durationMs": 2024.4089999999915,
"styleRecalcs": 10,
"styleRecalcDurationMs": 4.782000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 522.03,
"heapDeltaBytes": 3753636,
"heapUsedBytes": 66067624,
"domNodes": -228,
"jsHeapTotalBytes": 4448256,
"scriptDurationMs": 11.988,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1851.2479999999982,
"styleRecalcs": 68,
"styleRecalcDurationMs": 20.189,
"layouts": 12,
"layoutDurationMs": 4.849000000000001,
"taskDurationMs": 913.481,
"heapDeltaBytes": 13251592,
"heapUsedBytes": 75405500,
"domNodes": -228,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 178.333,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1838.367000000062,
"styleRecalcs": 69,
"styleRecalcDurationMs": 18.991,
"layouts": 12,
"layoutDurationMs": 4.324,
"taskDurationMs": 824.334,
"heapDeltaBytes": 10939176,
"heapUsedBytes": 73242964,
"domNodes": -228,
"jsHeapTotalBytes": 5496832,
"scriptDurationMs": 178.724,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66999999999998,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1717.9560000000151,
"styleRecalcs": 30,
"styleRecalcDurationMs": 10.715999999999998,
"layouts": 6,
"layoutDurationMs": 1.161,
"taskDurationMs": 385.5559999999999,
"heapDeltaBytes": 2625956,
"heapUsedBytes": 69764332,
"domNodes": 53,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 14.650999999999998,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1719.4419999999582,
"styleRecalcs": 33,
"styleRecalcDurationMs": 10.489,
"layouts": 6,
"layoutDurationMs": 1.226,
"taskDurationMs": 404.34899999999993,
"heapDeltaBytes": 2359872,
"heapUsedBytes": 64320424,
"domNodes": 53,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 16.270000000000003,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-clipping",
"durationMs": 699.7739999999908,
"styleRecalcs": 11,
"styleRecalcDurationMs": 3.6140000000000008,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 464.008,
"heapDeltaBytes": 18872608,
"heapUsedBytes": 80412628,
"domNodes": 0,
"jsHeapTotalBytes": 5767168,
"scriptDurationMs": 173.35,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 718.1850000000622,
"styleRecalcs": 11,
"styleRecalcDurationMs": 4.016000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 483.50499999999994,
"heapDeltaBytes": 19044968,
"heapUsedBytes": 80952404,
"domNodes": 0,
"jsHeapTotalBytes": 5505024,
"scriptDurationMs": 175.857,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1235.9480000000076,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 936.299,
"heapDeltaBytes": 9738440,
"heapUsedBytes": 95926564,
"domNodes": -259,
"jsHeapTotalBytes": 23744512,
"scriptDurationMs": 75.21499999999999,
"eventListeners": -185,
"totalBlockingTimeMs": 604,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1287.358999999924,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 999.678,
"heapDeltaBytes": -14155780,
"heapUsedBytes": 73616360,
"domNodes": -259,
"jsHeapTotalBytes": 20316160,
"scriptDurationMs": 75.75599999999999,
"eventListeners": -207,
"totalBlockingTimeMs": 609,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 855.777000000046,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 493.2249999999998,
"heapDeltaBytes": 14393844,
"heapUsedBytes": 91203788,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 175.42100000000005,
"eventListeners": 0,
"totalBlockingTimeMs": 230,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 856.9739999999229,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 483.6080000000003,
"heapDeltaBytes": 14534852,
"heapUsedBytes": 91358704,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 177.032,
"eventListeners": 0,
"totalBlockingTimeMs": 245,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "extension-clean-combined-draw",
"durationMs": 11979.614000000027,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10928.493,
"heapDeltaBytes": 2812728,
"heapUsedBytes": 85511300,
"domNodes": 0,
"jsHeapTotalBytes": 6291456,
"scriptDurationMs": 9023.101999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 1701,
"frameDurationMs": 20,
"p95FrameDurationMs": 50
},
{
"name": "extension-clean-combined-draw",
"durationMs": 10712.96899999993,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 9678.316,
"heapDeltaBytes": 5587400,
"heapUsedBytes": 81339448,
"domNodes": -228,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 7908.183,
"eventListeners": -164,
"totalBlockingTimeMs": 1286,
"frameDurationMs": 19.996666666666655,
"p95FrameDurationMs": 49.899999999999636
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 11894.266000000016,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10851.752999999999,
"heapDeltaBytes": -9166984,
"heapUsedBytes": 73197900,
"domNodes": 0,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 8983.194000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 1632,
"frameDurationMs": 18.890000000000025,
"p95FrameDurationMs": 50
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 11880.220999999892,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10872.777,
"heapDeltaBytes": -6748180,
"heapUsedBytes": 75176372,
"domNodes": 0,
"jsHeapTotalBytes": 9175040,
"scriptDurationMs": 9050.323,
"eventListeners": 0,
"totalBlockingTimeMs": 1907,
"frameDurationMs": 18.886666666666617,
"p95FrameDurationMs": 49.89999999999782
},
{
"name": "large-graph-idle",
"durationMs": 2028.2310000000052,
"styleRecalcs": 6,
"styleRecalcDurationMs": 2.7170000000000005,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 673.456,
"heapDeltaBytes": -3835316,
"heapUsedBytes": 82680980,
"domNodes": -228,
"jsHeapTotalBytes": -495616,
"scriptDurationMs": 35.73499999999999,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-idle",
"durationMs": 2036.9889999999486,
"styleRecalcs": 7,
"styleRecalcDurationMs": 3.5640000000000014,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 694.2139999999999,
"heapDeltaBytes": 3395320,
"heapUsedBytes": 82878904,
"domNodes": -228,
"jsHeapTotalBytes": -1056768,
"scriptDurationMs": 42.515,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "large-graph-pan",
"durationMs": 3073.469999999986,
"styleRecalcs": 67,
"styleRecalcDurationMs": 10.425,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1830.434,
"heapDeltaBytes": 1971544,
"heapUsedBytes": 84438200,
"domNodes": -228,
"jsHeapTotalBytes": 516096,
"scriptDurationMs": 912.067,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.80000000000109
},
{
"name": "large-graph-pan",
"durationMs": 3088.1259999999884,
"styleRecalcs": 67,
"styleRecalcDurationMs": 10.826000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1742.5710000000001,
"heapDeltaBytes": -3545812,
"heapUsedBytes": 77908304,
"domNodes": -198,
"jsHeapTotalBytes": 2314240,
"scriptDurationMs": 976.8399999999999,
"eventListeners": -164,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-zoom",
"durationMs": 3897.648000000004,
"styleRecalcs": 63,
"styleRecalcDurationMs": 11.959000000000001,
"layouts": 60,
"layoutDurationMs": 14.991000000000001,
"taskDurationMs": 1816.95,
"heapDeltaBytes": 9852460,
"heapUsedBytes": 75286432,
"domNodes": 0,
"jsHeapTotalBytes": 524288,
"scriptDurationMs": 834.2539999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-zoom",
"durationMs": 3960.048000000029,
"styleRecalcs": 63,
"styleRecalcDurationMs": 11.612999999999998,
"layouts": 60,
"layoutDurationMs": 14.819,
"taskDurationMs": 1849.3369999999998,
"heapDeltaBytes": 8833080,
"heapUsedBytes": 74941976,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 855.408,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3281.34799999998,
"styleRecalcs": 45,
"styleRecalcDurationMs": 7.056999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 2224.281,
"heapDeltaBytes": -1814408,
"heapUsedBytes": 82755416,
"domNodes": -198,
"jsHeapTotalBytes": 1196032,
"scriptDurationMs": 1133.216,
"eventListeners": 16,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3227.7559999999994,
"styleRecalcs": 46,
"styleRecalcDurationMs": 7.716000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 2147.435,
"heapDeltaBytes": -722224,
"heapUsedBytes": 83865764,
"domNodes": -228,
"jsHeapTotalBytes": 5914624,
"scriptDurationMs": 1069.98,
"eventListeners": -161,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2036.6300000000024,
"styleRecalcs": 6,
"styleRecalcDurationMs": 2.685,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 698.457,
"heapDeltaBytes": -3850168,
"heapUsedBytes": 84310868,
"domNodes": -228,
"jsHeapTotalBytes": -1282048,
"scriptDurationMs": 38.453,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "minimap-idle",
"durationMs": 2023.3600000000251,
"styleRecalcs": 6,
"styleRecalcDurationMs": 2.529,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 502.432,
"heapDeltaBytes": -3890280,
"heapUsedBytes": 84240892,
"domNodes": -228,
"jsHeapTotalBytes": -495616,
"scriptDurationMs": 37.83500000000001,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 684.2909999999733,
"styleRecalcs": 46,
"styleRecalcDurationMs": 5.984000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 420.12899999999996,
"heapDeltaBytes": 13595572,
"heapUsedBytes": 77249176,
"domNodes": 0,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 151.382,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 654.7870000000557,
"styleRecalcs": 48,
"styleRecalcDurationMs": 7.0489999999999995,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 412.101,
"heapDeltaBytes": 13382332,
"heapUsedBytes": 77025240,
"domNodes": 0,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 147.178,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2024.0799999999695,
"styleRecalcs": 10,
"styleRecalcDurationMs": 4.974999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 529.184,
"heapDeltaBytes": -4108120,
"heapUsedBytes": 58997096,
"domNodes": -244,
"jsHeapTotalBytes": 4448256,
"scriptDurationMs": 11.479999999999997,
"eventListeners": -210,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "subgraph-idle",
"durationMs": 2020.490000000109,
"styleRecalcs": 10,
"styleRecalcDurationMs": 4.343000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 546.706,
"heapDeltaBytes": 362624,
"heapUsedBytes": 64107928,
"domNodes": -244,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 12.416999999999998,
"eventListeners": -210,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1716.304999999977,
"styleRecalcs": 73,
"styleRecalcDurationMs": 18.473,
"layouts": 16,
"layoutDurationMs": 5.864,
"taskDurationMs": 869.0620000000001,
"heapDeltaBytes": 11560920,
"heapUsedBytes": 75161744,
"domNodes": -244,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 124.08700000000002,
"eventListeners": -188,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1737.0429999999715,
"styleRecalcs": 74,
"styleRecalcDurationMs": 19.409,
"layouts": 16,
"layoutDurationMs": 6.01,
"taskDurationMs": 880.058,
"heapDeltaBytes": 11897356,
"heapUsedBytes": 75501480,
"domNodes": -244,
"jsHeapTotalBytes": 4710400,
"scriptDurationMs": 121.738,
"eventListeners": -188,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-transition-enter",
"durationMs": 1401.5490000000455,
"styleRecalcs": 18,
"styleRecalcDurationMs": 30.236,
"layouts": 5,
"layoutDurationMs": 13.231999999999998,
"taskDurationMs": 944.1179999999998,
"heapDeltaBytes": 19656128,
"heapUsedBytes": 97567740,
"domNodes": 13667,
"jsHeapTotalBytes": 11272192,
"scriptDurationMs": 31.194999999999993,
"eventListeners": 2383,
"totalBlockingTimeMs": 146,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "viewport-pan-sweep",
"durationMs": 10533.94800000001,
"styleRecalcs": 250,
"styleRecalcDurationMs": 35.833999999999996,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 6585.453,
"heapDeltaBytes": -12958624,
"heapUsedBytes": 67585356,
"domNodes": -228,
"jsHeapTotalBytes": 2613248,
"scriptDurationMs": 3088.7639999999997,
"eventListeners": -164,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "viewport-pan-sweep",
"durationMs": 9695.178000000054,
"styleRecalcs": 249,
"styleRecalcDurationMs": 36.422999999999995,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 6316.27,
"heapDeltaBytes": 53280,
"heapUsedBytes": 80441948,
"domNodes": -228,
"jsHeapTotalBytes": 1232896,
"scriptDurationMs": 2980.474,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.669999999999952,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-idle",
"durationMs": 17871.478000000025,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 17580.395,
"heapDeltaBytes": -50028936,
"heapUsedBytes": 175095412,
"domNodes": -8049,
"jsHeapTotalBytes": -17571840,
"scriptDurationMs": 324.586,
"eventListeners": -14923,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-idle",
"durationMs": 17709.466000000022,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 17419.277000000002,
"heapDeltaBytes": -56135744,
"heapUsedBytes": 183817472,
"domNodes": -8049,
"jsHeapTotalBytes": -17944576,
"scriptDurationMs": 282.95599999999996,
"eventListeners": -14923,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-pan",
"durationMs": 22834.347000000038,
"styleRecalcs": 171,
"styleRecalcDurationMs": 23.706000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 22714.148999999998,
"heapDeltaBytes": -72634804,
"heapUsedBytes": 167122560,
"domNodes": -8049,
"jsHeapTotalBytes": -19030016,
"scriptDurationMs": 973.37,
"eventListeners": -14923,
"totalBlockingTimeMs": 150,
"frameDurationMs": 17.780000000000047,
"p95FrameDurationMs": 16.700000000004366
},
{
"name": "vue-large-graph-pan",
"durationMs": 22579.473000000005,
"styleRecalcs": 164,
"styleRecalcDurationMs": 23.737999999999982,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 22442.121,
"heapDeltaBytes": -70495456,
"heapUsedBytes": 169272888,
"domNodes": -8049,
"jsHeapTotalBytes": -17195008,
"scriptDurationMs": 1014.2119999999999,
"eventListeners": -14923,
"totalBlockingTimeMs": 252,
"frameDurationMs": 18.333333333333332,
"p95FrameDurationMs": 16.700000000004366
},
{
"name": "vue-node-resize-workload",
"durationMs": 6447.406999999998,
"styleRecalcs": 20,
"styleRecalcDurationMs": 85.72200000000002,
"layouts": 20,
"layoutDurationMs": 30.199000000000005,
"taskDurationMs": 6420.137,
"heapDeltaBytes": -57269076,
"heapUsedBytes": 183192640,
"domNodes": -8049,
"jsHeapTotalBytes": 10887168,
"scriptDurationMs": 123.67699999999998,
"eventListeners": -14927,
"totalBlockingTimeMs": 608,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-node-resize-workload",
"durationMs": 6538.763000000017,
"styleRecalcs": 20,
"styleRecalcDurationMs": 81.31600000000003,
"layouts": 20,
"layoutDurationMs": 27.272000000000006,
"taskDurationMs": 6481.891,
"heapDeltaBytes": -33040384,
"heapUsedBytes": 185029336,
"domNodes": -8049,
"jsHeapTotalBytes": 16244736,
"scriptDurationMs": 131.22,
"eventListeners": -14923,
"totalBlockingTimeMs": 458,
"frameDurationMs": 18.333333333333332,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-zoom-out-idle",
"durationMs": 8877.343999999994,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8692.331,
"heapDeltaBytes": -28612260,
"heapUsedBytes": 162784956,
"domNodes": 0,
"jsHeapTotalBytes": -25952256,
"scriptDurationMs": 166.578,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.77333333333336,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-zoom-out-idle",
"durationMs": 8831.91899999997,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8594.317000000001,
"heapDeltaBytes": 13542472,
"heapUsedBytes": 176127240,
"domNodes": 0,
"jsHeapTotalBytes": -14155776,
"scriptDurationMs": 167.937,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.80000000000291
},
{
"name": "workflow-execution",
"durationMs": 136.8689999999333,
"styleRecalcs": 8,
"styleRecalcDurationMs": 17.752,
"layouts": 4,
"layoutDurationMs": 1.859,
"taskDurationMs": 109.33000000000001,
"heapDeltaBytes": 3259048,
"heapUsedBytes": 65179232,
"domNodes": 98,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 12.908,
"eventListeners": 33,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "workflow-execution",
"durationMs": 111.27399999998033,
"styleRecalcs": 7,
"styleRecalcDurationMs": 18.06,
"layouts": 5,
"layoutDurationMs": 3.2619999999999996,
"taskDurationMs": 95.223,
"heapDeltaBytes": 3137344,
"heapUsedBytes": 65230544,
"domNodes": 120,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 8.524000000000001,
"eventListeners": 33,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
}
]
}🔬 E2E Coverage
Top 10 files by uncovered lines
🌾 Fallow: ✅ PassedFallow 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 (14 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 3d64a9e35db6b35c3a1397795c9958f70176d387
Full docs: ADR-DEVEX-LINT-0015 🧩 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 |
a4a382f to
783bdf9
Compare
|
The Generated by Claude Code |
Playwright videos for new testsDownload the recorded videos and HTML report Recorded projects: cloud Recorded test titles:
|
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## main #18075 +/- ##
==========================================
+ Coverage 83.79% 84.78% +0.99%
==========================================
Files 3026 2302 -724
Lines 149499 131918 -17581
Branches 38645 36927 -1718
==========================================
- Hits 125276 111851 -13425
+ Misses 22918 19765 -3153
+ Partials 1305 302 -1003
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 746 files with indirect coverage changes 🚀 New features to boost your workflow:
|
783bdf9 to
e907ecf
Compare
Fix status: color/bgcolor loss fixed and pushed; title stomp + widget-overwrite need a design decisionPushed a fix for the node color/bgcolor loss and flipped its two proof tests (Playwright + Vitest) from Fixed: node color/bgcolor loss
Not trivial: title stomp
The incremental (non-full-reconcile) path in There is no existing intent marker anywhere in
A naive "prefer the incumbent title" fallback would be wrong: it would also block the agent's own legitimate ability to name/rename a node it just created or is actively renaming, since there's currently no way to tell that case apart from the stale-doc case. Not trivial: widget value overwritten mid-edit
A real fix needs new cross-layer plumbing: some system needs to know which widget (if any) is currently focused, and
None of that is a small, local change, so it needs a person to pick a direction before implementation. Generated by Claude Code |
e907ecf to
c01db32
Compare
…idget state Adds failing (test.fail / it.fails) proof coverage for the shared reconcile defect: a full reconcile (EcsFollowerAdapter reconcileNextFrame) always trusts the CRDT doc's snapshot over live state, with no local-edit-recency check anywhere in graphMutations.ts. - browser_tests/tests/agent/agentCrdtReconcileOverwrite.spec.ts: three e2e repros reusing the agentConversationFixture and the tab-switch rebind lever already proven in agentTabSwitchCatchUp.spec.ts - node color loss, a stomped title rename, and a widget overwritten mid-edit with no focus guard. - src/workbench/extensions/agent/crdt/graphMutations.test.ts: unit-level proof for the title and color cases directly against reconcileNode.
assignNodeFields unconditionally reset color/bgcolor to undefined before applying a reconciled node's fields, so any agent-CRDT reconcile silently stripped a color the user had just applied from the properties panel. The CRDT doc never carries node color at all, so prepareNode only ever adds a color/bgcolor key to the replacement when the doc payload explicitly has one; leaving those two keys out of assignNodeFields's reset preserves the live value whenever the replacement doesn't own the key, without losing the ability to apply one the doc does provide. Flips the color repro tests (Playwright and Vitest) from test.fail/it.fails to real regression coverage. Title stomp and widget-overwrite-mid-edit stay as known-bug proof tests; see the PR discussion for why those two need a larger design decision rather than a local fix.
…cile A canvas rename only ever mutates the live node (useNodeEventHandlers.ts's handleNodeTitleUpdate); it never writes back into the CRDT doc, so a full reconcile kept replaying the doc's stale title over an unsynced local rename. prepareNode now resolves a node's title by comparing the incoming payload against the title recorded on the node's lastSerialization baseline: an unchanged doc title is a stale replay and keeps the live title, while a title that differs from that baseline is a genuine doc-side change (e.g. the agent naming or renaming the node) and still wins. Flips the title-stomp repro tests in graphMutations.test.ts and agentCrdtReconcileOverwrite.spec.ts from expected-fail to real regression coverage. Widget-overwrite-mid-edit remains an intentionally unfixed, separately-scoped repro.
c01db32 to
ca96a72
Compare
Title stomp: fixed and coveredImplemented the title fix following the same merge shape #18081 uses for node color/autogrow labels:
This avoids the naive "always prefer the incumbent" fallback flagged as unsafe in the RCA above (it would also have blocked the agent's own legitimate rename/name-on-create ability) by keying off whether the doc actually moved, not just whether a live/doc mismatch exists. Flipped both proof tests to real regression coverage:
Widget-overwrite-mid-edit: intentionally left unfixedConfirmed not tractable as a quick fix here. It needs the unbuilt Validation
Generated by Claude Code |
resolveNodeTitle only trusted a live rename when it had a lastSerialization baseline to compare the doc's title against. That baseline is written exclusively inside prepareNode, so a node whose live state was rebuilt outside this reconcile loop - e.g. a workflow tab serialized away and reloaded on tab switch - starts with no baseline at all, even if it went through a reconcile before the switch. With no baseline, the guard fell through and replayed the doc's stale title over the reload's live rename. Treat a missing baseline the same as an unchanged one: with no recorded doc title to compare against, there's no evidence the doc side changed, so the live title wins.
50b2c99 to
ddee679
Compare
…tion materialize() (agentNodeMaterializer.ts) turns a store-only NodeState record into a live LGraphNode by deleting the record and re-registering a freshly constructed node under the same id. That is a rendering-layer step, not a content change, but the fresh node's shell state starts with no `lastSerialization` at all, discarding the very baseline graphMutations.ts's resolveNodeTitle relies on to tell a stale doc replay from a genuine local edit. Every node an agent session renders - including a pre-existing workflow's nodes once they're projected into the CRDT doc - goes through this same materialization once. Losing the baseline there meant the very next full reconcile always saw a node with no baseline, so a local rename (or any other reconcile relying on the baseline) was replayed over by the doc's own value regardless of whether the doc actually changed. Carry the record's lastSerialization onto the newly constructed node so the baseline survives materialization, matching what the doc already held at that point instead of discarding it. Revert the earlier graphMutations.ts change from this branch: fixing the baseline at its source here restores the original resolveNodeTitle's "keep the live title unless the doc genuinely changed it" comparison to correct behavior without needing to touch it, and the wider comparison change regressed several existing reconcile scenarios (host resync, node retype rebuild) that legitimately rely on trusting the doc when a node is reconciled for the first time.
CI on this branch kept failing the title-reconcile e2e test even after the agentNodeMaterializer.ts fix (previous commit), which only covers a node re-materializing after a doc-only remote edit. Traced the remaining failure to a separate, deeper cause: returning to a workflow tab reloads it via LGraph.configure(), and LGraph.clear() tears each node down individually (teardownOwnedGraphs) before it resets the nodeDataStore bucket, so a node's reconcile baseline is gone before any store-level hook could preserve it - regardless of the materializer fix. A live rename does survive the reload itself (serialize() captured it), but the next reconcile has no baseline to compare the doc's title against and replays it over the rename regardless. A real fix needs the same kind of dedicated cross-layer plumbing as the widget-overwrite case already left unfixed in this spec - threading a "preserve this node's reconcile baseline" signal through LGraph.clear()'s per-node teardown - not a change local to graphMutations.ts, agentNodeMaterializer.ts, or nodeDataStore.ts alone (all three were tried and either didn't reach this path or regressed other reconcile scenarios). Mark the e2e test as a known, intentionally unfixed repro (matching the widget-overwrite test's own test.fail()), add the same repro at the unit level in agentNodeMaterializer.test.ts, and correct this spec's file-level comment to describe what's actually fixed (color, robustly; title, only within a session) versus what remains open.
…cile-overwrite-repro # Conflicts: # src/workbench/extensions/agent/crdt/graphMutations.ts
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: changes requested
I found one pre-merge correctness defect:
- [P1] A type-changing reconcile can preserve the previous node incarnation's local title. I reproduced this through
reconcileNode: replacingType1withReplacementwhile the incoming title still matched the old serialization baseline producedLocal titleinstead of the replacement payload'sNode 1. Title preservation needs the same same-type boundary already used for slots and colors.
Findings without an inline anchor
- [P2] The body attributes the widget-focus fix to ADR-CRDT-FOLLOWER-0025 and specifically prescribes
pendingOpLedger.tstoprocessedWidgetRenderModel.tswiring. The ADR contains neither symbol nor a focus arbitration contract. Please state the verified requirement, a conflict policy for remote writes during a local edit, without presenting an undecided design as accepted architecture. - [P2] The overlap section omits #18065. That PR shares this merge base, changes
graphMutations.tsandagentNodeMaterializer.ts, and implements a competing title-authority model with an explicit customization store and a same-type guard. Please document whether these PRs stack, supersede one another, or will be reconciled. - [P2] "Color fixed for every case" is broader than the code's intentional semantics. The existing type-change test sets
colorandbgcolorand expects both to clear when a different node type replaces the old incarnation. Please narrow the claim to same-type reconciliation.
Merged candidates
- The free-standing tab helper, raw
.p-togglebuttonselector, and duplicated tab recipe all merge into the inline readiness finding. - The repeated baseline/reload comments merge into the inline factual correction about the widget case's separate root cause.
- The unit expected-failure masking concern merges into the inline non-null-assertion finding.
- Missing
bgcolorcoverage merges into the duplicated color-policy finding. Removing the redundant store-level branch removes that new test obligation.
Checked and cleared
- I confirmed that colors intentionally reset on a type-changing replacement. That is new-incarnation behavior, not the same-type omission bug.
- Playwright's two runtime
test.fail()calls do not hide setup failures before the marker. My local setup timeout was reported as an ordinary failure for both tests. - The title baseline is attached to graph-scoped
NodeState, so a process-global node-ID leak is excluded by the representation. The real uncovered identity boundary is type replacement, reported above. - Normal
LGraph.configure()does not repopulatelastSerialization; only missing-node fallback does. The body's statement that a normal reload loses this baseline is accurate. - Exact-head CI reports the broad test and lint-format checks green. My focused local logs intentionally cover fewer tests, and ESLint's successful run produced no output before oxfmt.
Coverage and verification
Standard-depth correctness, maintainability, idiomaticity, tests, and documentation lanes read all seven changed files. The tests lane inventoried 536 fixture methods across 80 files, traced 41 browser interactions, applied checks A through H and all 23 Playwright rulebook items, and reviewed the unit-test boundaries.
At 33bea4dc320de99d6160f5adfd4f492ceb7750e0, focused Vitest passed 99 tests with 2 expected failures. Main and browser typechecks passed. ESLint and oxfmt passed on all changed files. CI checks at this head are green. A focused production-path red test reproduced the type-replacement title defect.
The focused local Playwright run could not validate app behavior because this orb's local cloud setup never exposed the Agent button. All three tests timed out in fixture setup; importantly, the two runtime expected-failure markers did not convert those setup errors into expected failures. Exact-head CI's cloud Playwright and video checks are green.
PR #18065 directly overlaps the title policy and materialization files. I did not duplicate review work from its dedicated reviewer.
…or reset - resolveNodeTitle now takes prepareNode's type-scoped incumbent instead of the unrestricted existing node, so a type-changing reconcile can no longer keep a stale local title from the old incarnation - assignNodeFields (nodeDataStore.ts) restores its unconditional bgcolor/color reset; graphMutations.ts's resolveNodeColors already resolves the same-type-preserved value onto the replacement before it reaches the store, so the store-level exclusion was dead weight - adds a materialize()-then-reconcile composition test proving local-vs-remote title authority end to end, not just that lastSerialization is copied - hardens agentCrdtReconcileOverwrite.spec.ts's tab-switch helper to wait for the follower replay boundary instead of only the tab's checked class, via a new Topbar.openBlankTabAndReturn() helper - replaces non-null assertions on getNode()/getNodeById() lookups in tests with assert.exists narrowing - separates the widget-overwrite case's explanation from the reconcile/ follower-rebind root cause throughout the spec file
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: changes requested
The previous production blocker is fixed: title preservation is now limited to same-type incumbents, and the new regression test is red against the prior implementation and green here.
One test blocker remains. The updated browser helper waits for canvas state that already matched after runTurns(), so it can pass before the tab-return subscription and reconcile occur. The neighboring tab-switch spec exposes the required causal boundary by recording subscribeCount() and polling for before + 1 before asserting canvas state.
The PR body also needs these factual corrections:
- It still says
assignNodeFieldsno longer resetscolor/bgcolor; current code deliberately resets them, whileprepareNodepreserves same-type colors by copying them into the replacement. - It says the widget overwrite remains as a Vitest
it.failsand identifies it as one of the focused run's two expected failures. The two unit expected failures are the title reload and an unrelated store-only serialization mismatch; widget overwrite is Playwright-only. - #18065 is no longer an active competing merge choice: it was closed unmerged in favor of this PR. Its explicit-intent model remains relevant design history, but this PR is the selected implementation.
FE-2504 does not require this PR to implement the semantic remote-apply architecture. It does make the baseline/materialization logic temporary safety machinery that should have a deletion path when store-first reconciliation is removed. The behavioral color/title regression coverage remains worth retaining through that migration.
I reran the focused tests (101 passed, 2 expected failures), full and browser typechecks, changed-file ESLint, type-aware Oxlint, formatting, Knip, Playwright discovery, and a red/green proof for the type-change regression. All executed gates passed; exact-head CI was still in progress when reviewed.
…atch The tab-return helper in agentCrdtReconcileOverwrite.spec.ts checked canvas state that could already match before the follower's new subscription and reconcile ran. Mirror agentTabSwitchCatchUp.spec.ts: capture subscribeCount() before switching tabs, poll for it to rise by one, then assert the replayed canvas. Also corrects a graphMutations.test.ts comment that described a discarded store-level color-preservation policy; assignNodeFields unconditionally resets color/bgcolor, and preservation actually comes from prepareNode copying the same-type incumbent's color onto the replacement before assignNodeFields runs.
…cile-overwrite-repro
This comment has been minimized.
This comment has been minimized.
|
The "Check for AI agent co-author trailers" check (https://github.com/Comfy-Org/ComfyUI_frontend/actions/runs/36052942371/job/107812769986) is failing on commit The fix is a commit-message-only amend (strip both trailer lines, no code/diff change) followed by a force-push to 🤖 Generated with Claude Code |
4135c45 to
efb2c59
Compare
Feedback was addressed on later commits; DrJKL subsequently recorded “ready to approve”. The latest rewrite changed commit metadata only and preserved the exact tree.
christian-byrne
left a comment
There was a problem hiding this comment.
Reviewed the current tree after the metadata-only tip rewrite. The prior substantive findings are addressed, the rewritten head preserves the exact tree, and the remaining expected-failure cases stay explicitly scoped as unfixed gaps.
|
|
…e node/title/widget state (#18075) Manual backport of #18075 to core/1.54. Automated backport failed: #18075 (comment) Cherry-picked from 96821ac. Dependency check: the bot flagged a possible stacking conflict with #18727, #18722, #18683, #17872 (all still open against core/1.54). None of those touch graphMutations.ts, graphMutations.test.ts or agentConversationFixture.ts, so they are not the cause. The real causes: - graphMutations.ts: core/1.54 predates #18191 ("keep the ghost placement flag out of the document"), which hoisted node flags into a local `flags` variable (`cloneNodeFlags`) reused by `lastSerialization`. #18191 has no needs-backport label and was never backported here. Resolved by keeping core/1.54's existing flags handling (`cloneRecord(payload.flags)` / `structuredClone(payload)`) and porting only #18075's actual change: `resolveNodeTitle` plus the new `titleReconcileBaseline` field. - graphMutations.test.ts: pure insertion, no real conflict — accepted the new test cases as-is. - browser_tests/fixtures/agentConversationFixture.ts and the new browser_tests/tests/agent/agentCrdtReconcileOverwrite.spec.ts: this fixture (and the wider main-only agent Playwright test scaffolding it depends on, e.g. agentTabSwitchCatchUp.spec.ts) does not exist on core/1.54 at all — introduced by main-only work, not by any of the PRs above. Dropped from this backport along with the Topbar.ts#openBlankTabAndReturn() helper added only for that spec, so the branch keeps compiling. The runtime fix ships with its Vitest regression coverage (graphMutations.test.ts, agentNodeMaterializer.test.ts); the e2e visual-proof spec is left for a follow-up once the agent Playwright fixture infrastructure itself is backported. Validation on this branch: - pnpm exec vitest run src/workbench/extensions/agent - pnpm exec eslint / pnpm exec oxfmt --check on touched files - pnpm typecheck
…ve node/title/widget state (#18075) (#18759) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-org.slack.com/archives/C0BGH348Z0C/p1789514768597649)_ Manual backport of #18075 to `cloud/1.54`. The automated backport failed with merge conflicts: #18075 (comment) Cherry-picked from `96821ace03bc81d7a29965751a401fcf1818cae3` (`git cherry-pick -x -m 1`). ## Dependency check The bot flagged a possible stacking conflict with #18728, #18723 (both still open against `cloud/1.54`). Neither touches `graphMutations.ts`, `graphMutations.test.ts`, or `agentConversationFixture.ts` — they are not the cause of this conflict. The real causes (same underlying gap as the `core/1.54` backport, #18758): - **`graphMutations.ts`** — `cloud/1.54` predates #18191 ("keep the ghost placement flag out of the document"), which hoisted node flags into a local `flags` variable (`cloneNodeFlags`) that `lastSerialization` also started reusing. #18191 is merged to `main` but carries no `needs-backport` label and was never backported to `cloud/1.54`. Resolved by keeping `cloud/1.54`'s existing flags handling (`flags: cloneRecord(payload.flags)` / `lastSerialization: structuredClone(payload)`) and porting only #18075's actual change: `resolveNodeTitle(payload, incumbent)` for the title, plus the new `titleReconcileBaseline` field feeding it. - **`graphMutations.test.ts`** — pure insertion (no code on the `HEAD` side of the conflict), accepted the new test cases as-is. - **`browser_tests/fixtures/agentConversationFixture.ts`** and the new **`browser_tests/tests/agent/agentCrdtReconcileOverwrite.spec.ts`** — this Playwright fixture (and the wider main-only agent-test scaffolding it depends on, e.g. `agentTabSwitchCatchUp.spec.ts`) does not exist on `cloud/1.54` at all; it isn't introduced by either PR above. Dropped from this backport, along with the `Topbar.ts#openBlankTabAndReturn()` helper added only for that spec, so the branch keeps compiling. The runtime fix ships with its Vitest regression coverage (`graphMutations.test.ts`, `agentNodeMaterializer.test.ts`); the e2e visual-proof spec is left for a follow-up once the agent Playwright fixture infrastructure itself is backported. ## Validation - `pnpm exec vitest run src/workbench/extensions/agent/crdt` — 827 passed, 5 expected-fail (the intentional `it.fails` cases this PR introduces/carries). - `pnpm exec eslint` / `pnpm exec oxfmt --check` clean on all touched files. - `pnpm typecheck` clean. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01L4WMaCPdk7KSva7vwKfz19 --- _Generated by [Claude Code](https://claude.ai/code/session_01L4WMaCPdk7KSva7vwKfz19)_ Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
…e node/title/widget state (#18075) (#18758) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-org.slack.com/archives/C0BGH348Z0C/p1789514768597649)_ Manual backport of #18075 to `core/1.54`. The automated backport failed with merge conflicts: #18075 (comment) Cherry-picked from `96821ace03bc81d7a29965751a401fcf1818cae3` (`git cherry-pick -x -m 1`). ## Dependency check The bot flagged a possible stacking conflict with #18727, #18722, #18683, #17872 (all still open against `core/1.54`). None of them touch `graphMutations.ts`, `graphMutations.test.ts`, or `agentConversationFixture.ts` — they are not the cause of this conflict. The real causes: - **`graphMutations.ts`** — `core/1.54` predates #18191 ("keep the ghost placement flag out of the document"), which hoisted node flags into a local `flags` variable (`cloneNodeFlags`) that `lastSerialization` also started reusing. #18191 is merged to `main` but carries no `needs-backport` label and was never backported to `core/1.54`. Resolved by keeping `core/1.54`'s existing flags handling (`flags: cloneRecord(payload.flags)` / `lastSerialization: structuredClone(payload)`) and porting only #18075's actual change: `resolveNodeTitle(payload, incumbent)` for the title, plus the new `titleReconcileBaseline` field feeding it. - **`graphMutations.test.ts`** — pure insertion (no code on the `HEAD` side of the conflict), accepted the new test cases as-is. - **`browser_tests/fixtures/agentConversationFixture.ts`** and the new **`browser_tests/tests/agent/agentCrdtReconcileOverwrite.spec.ts`** — this Playwright fixture (and the wider main-only agent-test scaffolding it depends on, e.g. `agentTabSwitchCatchUp.spec.ts`) does not exist on `core/1.54` at all; it isn't introduced by any of the PRs above. Dropped from this backport, along with the `Topbar.ts#openBlankTabAndReturn()` helper added only for that spec, so the branch keeps compiling. The runtime fix ships with its Vitest regression coverage (`graphMutations.test.ts`, `agentNodeMaterializer.test.ts`); the e2e visual-proof spec is left for a follow-up once the agent Playwright fixture infrastructure itself is backported. ## Validation - `pnpm exec vitest run src/workbench/extensions/agent` — 1 pre-existing, unrelated failure in `AgentPanelRoot.test.ts` (confirmed present on `core/1.54` before this backport, untouched by this diff); everything else passes, including the 2 intentional `it.fails` cases this PR introduces. - `pnpm exec eslint` / `pnpm exec oxfmt --check` clean on all touched files. - `pnpm typecheck` clean. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01L4WMaCPdk7KSva7vwKfz19 --- _Generated by [Claude Code](https://claude.ai/code/session_01L4WMaCPdk7KSva7vwKfz19)_ Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
|
|
## Summary Pins [PM-1717](https://linear.app/comfyorg/issue/PM-1717) as the known full workflow-tab reload title-loss gap, distinct from the within-session reconcile behavior fixed by Comfy-Org#18075. ## Changes - **What**: Renames the existing expected-failure browser scenario and its reload step to state the user-visible lifecycle boundary precisely. - **Breaking**: None; test-only. - **Dependencies**: None. ## Review Focus The case remains `test.fail()` and red on current `main`: a manual title is visible before leaving the tab and is stomped after returning fully reloads it. This PR changes no production behavior. Source: https://comfy-organization.slack.com/archives/C0A8Z4U7Y1K/p1790329354333089 Follow-up from: Comfy-Org#18075
Requested by Memie Osuga, Christian Byrne · Slack thread
Before / After
Before: two reported bugs (node color loss after reconcile, title stomp after reconcile) share one root cause with zero regression coverage: whenever the agent CRDT follower does a full reconcile (
EcsFollowerAdapter'sreconcileNextFrame, confirmed triggered by returning to a workflow tab),graphMutations.tsalways trusts the CRDT doc's own snapshot over live/local state, with no "was this edited more recently locally" check anywhere in the reconcile path. A third reported bug — a widget value overwritten mid-edit, plus related linked reports — is a separate, directset_widgetcollision:applyWidgetValues/setWidgetValuewrite straight intowidgetValueStorewith no focus/in-progress-edit guard, and it never goes through the reconcile/follower-rebind path at all. All three bugs happen silently — nothing fails, nothing is asserted, so they ship and regress again freely.After: node color loss on a same-type reconcile is fixed, robustly. Title stomp is fixed within a session (a follower rebind, or a node re-materializing after a doc-only remote edit) but not across a full workflow-tab reload — that's a second, deeper gap, confirmed and left as a known repro alongside the pre-existing widget-overwrite-mid-edit one. See "Fix status" below for exactly what's fixed and why the remaining two are out of scope here.
What's covered
browser_tests/tests/agent/agentCrdtReconcileOverwrite.spec.ts(Playwright e2e, using the existingagentConversationFixtureand the tab-switch rebind lever already proven inagentTabSwitchCatchUp.spec.ts):test.fail()) — the reload itself is the problem; see "Fix status".set_widgetlands on aCLIPTextEncodetextwidget while the user is actively typing into it, viasetWidgetValue), not a reconcile-path symptom — see "Fix status". Known, intentionally unfixed repro (test.fail()).reconcileByReturningToTab) captures the follower'ssubscribeCount()before switching tabs and polls for it to rise by one — proof the mock host's request/response round-trip happened — then captures and polls a browser-sideappliedFrameCount()(status.outcomes.appliedfromuseAgentCrdtFollower's CRDT debug instrument) to rise by one before asserting canvas state.subscribeCount()alone only proves the mock host sent the catch-up frame at send time;appliedFrameCount()only rises once the browser has actually merged that frame into the stores and finished reconciling the live graph for it, so the assertions genuinely wait for the reconcile to complete, not just for the tab click to land or the frame to be sent.src/workbench/extensions/agent/crdt/graphMutations.test.ts(Vitest): unit-level regression coverage for the title and color fixes directly againstreconcileNode/prepareNode, isolating the fix from any DOM/CRDT-transport involved, including a type-changing reconcile that must not carry a stale local title onto the new-type replacement. Oneit.failshere covers an unrelated, pre-existing store-only serialization mismatch (keeps a store-only node in LGraph.serialize() without reporting a mismatch).src/workbench/extensions/agent/crdt/agentNodeMaterializer.test.ts(Vitest): regression coverage for the title fix's in-session materialization case — including a composition test that renames the live node, reconciles an unchanged payload, then a changed one, and checks local-then-remote title authority end to end — plus anit.failscase proving the workflow-tab-reload title gap at the unit level (no Playwright needed to reproduce it).test.fail()inagentCrdtReconcileOverwrite.spec.ts), since it's a DOM-focus/in-progress-edit collision rather than something the store-level unit tests can observe.Root cause (confirmed, shared by color and title; the widget case is distinct — see below)
graphMutations.ts'sreconcileNode/reconcileNodeFields, feedingnodeDataStore.ts'supdateNode/updateNodeFields(assignNodeFields), always fully replaces scalar node fields from the CRDT payload:nodePayload.ts'snodeTitle()had no "keep the incumbent title" fallback, so a canvas rename (which only ever touches the live node, peruseNodeEventHandlers.ts'shandleNodeTitleUpdate) was stomped back to the doc's stale title on the next reconcile.The widget-overwrite case is a separate mechanism, not a third symptom of the above:
applyWidgetValues/setWidgetValuewrite straight intowidgetValueStorewith no focus/in-progress-edit guard, so an agentset_widgeton a widget the user is actively typing into silently overwrites it — this doesn't go through the tab-return reconcile helper at all, it's a local-vs-remote edit collision on a single widget write.Fix status: color fixed (same-type reconciliation); title fixed within a session only; two known repros left out of scope
Fixed: node color/bgcolor loss on same-type reconciliation.
assignNodeFields(src/stores/nodeDataStore.ts) unconditionally resetscolor/bgcolortoundefinedbefore applying a reconciled replacement's own fields — that store-level reset is unchanged and stays generic. Preservation instead lives entirely at the CRDT boundary:prepareNode(graphMutations.ts) callsresolveNodeColors(payload, incumbent), which copies the same-type incumbent's livecolor/bgcoloronto the prepared replacement object whenever the doc payload doesn't carry its own value (the CRDT doc never carries node color at all). That replacement, already carrying the live color, is what reachesassignNodeFields, so the store's unconditional reset is immediately overwritten by the replacement's own (copied-forward) value in the same object-literal spread. A type-changing replacement is a different case and intentionally keeps resetting color/bgcolor to the new incarnation's own values (see the existing "replaces a field-reconciled node when its type changes" test) — that's correct, documented behavior, not something this PR changes.Fixed within a session: title stomp. A canvas rename only ever mutates the live node (
useNodeEventHandlers.ts'shandleNodeTitleUpdate); it never writes back into the CRDT doc.prepareNode(graphMutations.ts) resolves a node's title by comparing the incoming payload's title against the title recorded on the node'slastSerializationbaseline (the doc snapshot as of the last reconcile), scoped to the same type boundaryprepareNodealready uses for colors and slots (incumbent, i.e.existing?.type === payload.type): an unchanged doc title on a same-type incumbent means nothing about the doc's title actually changed, so the live title (which may reflect an unsynced local rename) is kept; a title that differs from that baseline, or a type change, is treated as a genuine doc-side change and the payload's title still wins.agentNodeMaterializer.ts'smaterialize()also carries that baseline onto a newly materialized node instead of discarding it — otherwise the very first reconcile after a node re-materializes (a normal, frequent event for agent-created nodes) would always see no baseline and blindly trust the doc.Which doc-silent cases the baseline actually covers. The underlying reason a client-side baseline is needed at all:
comfy-multi-player'sFROZEN_OPShas noset_title/update_nodeop, so a rename performed afteradd_nodeproduces no CRDT op, and the doc's title is simply silent on it. Concretely, against the three cases where the doc stays silent: an empty/undefined doc title is covered the same way any other title is —resolveNodeTitle'spayload.title === existing.lastSerialization?.titlecheck does no special-casing for empty values, so it preserves a live rename whenever the doc's title is consistently empty/undefined across reconciles (the same code path as the covered non-empty case, though no dedicated test here exercises an empty baseline specifically). A subgraph node instance's own renamed title is not covered by a dedicated test:prepareNoderuns on subgraph instances the same as any other node (the incumbent'stypeis just the subgraph's UUID), so the mechanism should extend to them, but the only subgraph-title regression test in this PR (agentNodeMaterializer.test.ts's promoted-set_widgettest) pins the registered-class-name fallback fromnodeTitle(), not a local rename of the instance surviving a reconcile. A serialization round-trip (workflow-tab reload) is not covered at all — see the confirmed gap directly below:LGraph.clear()'steardownOwnedGraphsempties the node'slastSerializationbaseline before any store-level preservation could run, so there's no baseline left to compare against.Not fixed (confirmed, needs dedicated cross-layer work): title stomp survives a workflow-tab reload. Returning to a workflow tab reloads it via
LGraph.configure(), andLGraph.clear()tears each node down individually (teardownOwnedGraphs) before it resets thenodeDataStorebucket — so a node'slastSerializationbaseline is already gone by the time any store-level hook could try to preserve it, regardless of the fix above. A live rename does survive the reload itself (the tab's own saved JSON already carries it), but the next reconcile has no baseline to compare the doc's title against and replays it over the rename regardless. Tried and rejected: a broader "no baseline means trust the live node" change ingraphMutations.ts(regressed several other reconcile scenarios — host resync, node-retype rebuild — that legitimately rely on trusting the doc the first time a node reconciles); anodeDataStore.ts-level baseline-preserving cache keyed offclearGraph/registerNode(dead code in practice, sinceteardownOwnedGraphsempties the bucket first). A real fix needs the reconcile baseline to surviveLGraph.clear()'s per-node teardown itself — the same kind of dedicated plumbing as the widget-overwrite case below, not a change local to one module. Left astest.fail()/it.fails.Review also flagged that
lastSerializationis currently overloaded — it serves both the missing-node serialization fallback and this CRDT title-freshness baseline, with different lifetimes. That's a legitimate point; giving the reconcile baseline its own narrowly-scoped, CRDT-owned field/lifecycle is a bigger design change than fits this PR's blast radius and is tracked as a follow-up rather than done here.Not fixed (by design, out of scope for this PR): widget value overwritten mid-edit.
applyWidgetValues/setWidgetValue(graphMutations.ts) write straight intowidgetValueStorewith no focus/in-progress-edit check, andwidgetValueStorehas no concept of DOM focus at all (it's pure state, by design). A real fix needs new cross-layer plumbing so some system knows which widget is currently focused beforeapplyWidgetValues/setWidgetValuewrite to it — a verified requirement (a conflict policy for remote writes landing during a local edit), not yet an accepted design. Left unchanged astest.fail()(Playwright only; there is no Vitest coverage of this gap).Existing PRs checked
#16964and#17499(both still open/draft) touch nearby code ingraphMutations.ts/nodePayload.ts, but are scoped to a different bug (subgraph-instance title loss after a promotedset_widget) and do not address the regular-node title/color/widget defects here.#16309/#16302/#16288(optimistic-overlay pending-op shadow) are a presentation-only overlay for the agent's own in-flight ops, not a guard for user-initiated local edits, so they don't fix this either.#18065(closed, unmerged) shared this PR's merge base and touched the same two files (graphMutations.ts,agentNodeMaterializer.ts) plus added a newsrc/stores/nodeTitleCustomizationStore.tsanddocs/adr/CRDT-TITLE-0035-session-local-node-title-customization-flag.md. It proposed a different, more explicit solution to the same title-stomp problem this PR's "Fixed within a session" section covers: instead of comparing the incoming title against alastSerializationbaseline, it haduseNodeEventHandlers.tsexplicitly mark a node as human-customized on rename, andprepareNode/resolveNodeTitleingraphMutations.tscheck that explicit flag (scoped to the sameexisting.type === payload.typeboundary this PR's fix also uses) instead of inferring "did the doc change" from a stale snapshot.#18065was closed unmerged in favor of this PR: this PR (#18075) is the selected implementation going forward, and#18065's explicit-intent model is retained as relevant design history for the follow-up work called out above (giving the reconcile baseline its own CRDT-owned field/lifecycle, and eventually the workflow-tab-reload gap) rather than as an open decision.Verification
pnpm exec vitest run src/workbench/extensions/agent→ 1653 passed, 2 expected-fail: the workflow-tab-reload title gap (agentNodeMaterializer.test.ts) and the unrelated store-only serialization mismatch (graphMutations.test.ts). The widget-overwrite gap has no Vitest coverage — it's a Playwright-only repro (test.fail()).pnpm exec oxfmt --check,pnpm exec eslintclean on all touched files.pnpm typecheck(vue-tsc across all sub-projects) clean.playwright-tests (cloud),playwright-video-new-tests, thee2e-statusgate, andtest(Vitest) all green on this branch's head, exercising the color fix and the two remaining repros astest.fail()/it.fails(expected failures, not blocking).🤖 Generated with Claude Code
https://claude.ai/code/session_01L4WMaCPdk7KSva7vwKfz19