Repository navigation
test(agent-crdt): repro schema-gate latch leaving canvas at zero nodes - #18099
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 |
🎭 Playwright: ✅ 2114 passed, 0 failed · 3 flaky📊 Browser Reports
🎨 Storybook: ✅ Built — View Storybook📦 Bundle: 9.97 MB gzip 🔴 +18.9 kBDetailsSummary
Category Glance App Entry Points — 3.87 kB (baseline 3.87 kB) • ⚪ 0 BMain entry bundles and manifests
Status: 1 added / 1 removed Graph Workspace — 1.44 MB (baseline 1.46 MB) • 🟢 -19.6 kBGraph editor runtime, canvas, workflow orchestration
Status: 2 added / 2 removed / 1 unchanged Views & Navigation — 132 kB (baseline 132 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 — 37.9 kB (baseline 37.9 kB) • ⚪ 0 BAuthentication, profile, and account management bundles
Status: 8 added / 8 removed / 3 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 — 124 kB (baseline 123 kB) • 🔴 +190 BReusable component library chunks
Status: 9 added / 9 removed / 7 unchanged Data & Services — 3.93 MB (baseline 3.87 MB) • 🔴 +54.3 kBStores, services, APIs, and repositories
Status: 13 added / 13 removed / 3 unchanged Utilities & Hooks — 577 kB (baseline 577 kB) • 🔴 +38 BHelpers, composables, and utility bundles
Status: 23 added / 23 removed / 15 unchanged Vendor & Third-Party — 17.4 MB (baseline 17.4 MB) • 🔴 +1.92 kBExternal libraries and shared vendor chunks
Status: 3 added / 3 removed / 15 unchanged Other — 17.8 MB (baseline 17.8 MB) • 🔴 +24.9 kBBundles that do not match a named category
Status: 129 added / 129 removed / 173 unchanged ⚡ Performance Report
Show regressions
All metrics
Historical variance (last 15 runs)
Trend (last 15 commits on main)
Raw data{
"timestamp": "2026-09-23T04:35:13.508Z",
"gitSha": "3bba542c4ae9413da23a42be68a04d6cc67f5a92",
"branch": "test/agent-crdt-schema-gate-zero-nodes-repro",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 1991.314999999986,
"styleRecalcs": 12,
"styleRecalcDurationMs": 9.077000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 465.26300000000003,
"heapDeltaBytes": -6566208,
"heapUsedBytes": 61087008,
"domNodes": -259,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 10.4,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-idle",
"durationMs": 2005.2949999999328,
"styleRecalcs": 12,
"styleRecalcDurationMs": 9.663000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 467.45599999999996,
"heapDeltaBytes": 768544,
"heapUsedBytes": 68765716,
"domNodes": -258,
"jsHeapTotalBytes": 5496832,
"scriptDurationMs": 10.799000000000003,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1760.665999999901,
"styleRecalcs": 46,
"styleRecalcDurationMs": 14.840000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 702.8839999999999,
"heapDeltaBytes": 9905660,
"heapUsedBytes": 77417548,
"domNodes": -260,
"jsHeapTotalBytes": 6021120,
"scriptDurationMs": 146.654,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1775.7709999998497,
"styleRecalcs": 47,
"styleRecalcDurationMs": 16.664,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 777.6120000000001,
"heapDeltaBytes": -5104268,
"heapUsedBytes": 62570844,
"domNodes": -258,
"jsHeapTotalBytes": 6021120,
"scriptDurationMs": 136.63400000000001,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1726.6270000000077,
"styleRecalcs": 30,
"styleRecalcDurationMs": 14.977999999999998,
"layouts": 6,
"layoutDurationMs": 1.092,
"taskDurationMs": 343.263,
"heapDeltaBytes": 2766356,
"heapUsedBytes": 70337764,
"domNodes": 77,
"jsHeapTotalBytes": 5505024,
"scriptDurationMs": 13.444,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1717.0430000001033,
"styleRecalcs": 31,
"styleRecalcDurationMs": 14.961,
"layouts": 6,
"layoutDurationMs": 1.0550000000000002,
"taskDurationMs": 330.40200000000004,
"heapDeltaBytes": 3096224,
"heapUsedBytes": 70485016,
"domNodes": 80,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 12.223999999999998,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "dom-widget-clipping",
"durationMs": 633.6599999999635,
"styleRecalcs": 13,
"styleRecalcDurationMs": 8.767,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 433.10499999999996,
"heapDeltaBytes": 18930980,
"heapUsedBytes": 86421772,
"domNodes": 22,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 151.831,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "dom-widget-clipping",
"durationMs": 663.428999999951,
"styleRecalcs": 13,
"styleRecalcDurationMs": 8.665000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 444.43399999999997,
"heapDeltaBytes": 18931176,
"heapUsedBytes": 86510996,
"domNodes": 22,
"jsHeapTotalBytes": 5505024,
"scriptDurationMs": 157.879,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1190.6819999999243,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 882.429,
"heapDeltaBytes": -13229212,
"heapUsedBytes": 72971404,
"domNodes": -295,
"jsHeapTotalBytes": 19267584,
"scriptDurationMs": 59.237999999999985,
"eventListeners": -217,
"totalBlockingTimeMs": 532,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1119.3539999999302,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 775.4389999999999,
"heapDeltaBytes": 38392324,
"heapUsedBytes": 104027576,
"domNodes": 0,
"jsHeapTotalBytes": 19034112,
"scriptDurationMs": 57.73500000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 511,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 775.9479999999712,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 393.14600000000024,
"heapDeltaBytes": 14568700,
"heapUsedBytes": 91227708,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 127.817,
"eventListeners": 0,
"totalBlockingTimeMs": 159,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.80000000000109
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 778.0700000000706,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 397.75300000000027,
"heapDeltaBytes": 14331476,
"heapUsedBytes": 90238568,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 125.19699999999997,
"eventListeners": 0,
"totalBlockingTimeMs": 157,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "extension-clean-combined-draw",
"durationMs": 8759.031999999934,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7715.276000000001,
"heapDeltaBytes": -183068,
"heapUsedBytes": 81350500,
"domNodes": -14,
"jsHeapTotalBytes": 5767168,
"scriptDurationMs": 6158.721,
"eventListeners": 0,
"totalBlockingTimeMs": 190,
"frameDurationMs": 17.773333333333298,
"p95FrameDurationMs": 33.29999999999927
},
{
"name": "extension-clean-combined-draw",
"durationMs": 8930.850999999848,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7832.155000000001,
"heapDeltaBytes": 12171116,
"heapUsedBytes": 94348392,
"domNodes": -18,
"jsHeapTotalBytes": 4718592,
"scriptDurationMs": 6113.660999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 68,
"frameDurationMs": 17.779999999999987,
"p95FrameDurationMs": 33.29999999999927
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 9090.135000000146,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7918.981,
"heapDeltaBytes": -2132824,
"heapUsedBytes": 79965696,
"domNodes": -12,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 6176.643,
"eventListeners": 0,
"totalBlockingTimeMs": 202,
"frameDurationMs": 17.776666666666703,
"p95FrameDurationMs": 33.400000000001455
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 9202.089000000115,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8260.491999999998,
"heapDeltaBytes": 13099392,
"heapUsedBytes": 95372656,
"domNodes": -16,
"jsHeapTotalBytes": 7864320,
"scriptDurationMs": 6524.352000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 115,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 33.30000000000109
},
{
"name": "large-graph-idle",
"durationMs": 2038.5490000001028,
"styleRecalcs": 8,
"styleRecalcDurationMs": 6.312999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 574.676,
"heapDeltaBytes": -4205512,
"heapUsedBytes": 82800164,
"domNodes": -240,
"jsHeapTotalBytes": -1544192,
"scriptDurationMs": 37.45799999999999,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "large-graph-idle",
"durationMs": 2033.9160000000902,
"styleRecalcs": 7,
"styleRecalcDurationMs": 8.161000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 550.504,
"heapDeltaBytes": -3459988,
"heapUsedBytes": 83624780,
"domNodes": -244,
"jsHeapTotalBytes": -1019904,
"scriptDurationMs": 32.163,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-pan",
"durationMs": 2280.714999999873,
"styleRecalcs": 65,
"styleRecalcDurationMs": 11.810999999999995,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1533.8020000000001,
"heapDeltaBytes": 18934140,
"heapUsedBytes": 83693724,
"domNodes": 6,
"jsHeapTotalBytes": 487424,
"scriptDurationMs": 766.6139999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-pan",
"durationMs": 2254.361000000017,
"styleRecalcs": 68,
"styleRecalcDurationMs": 12.603,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1550.988,
"heapDeltaBytes": 15865420,
"heapUsedBytes": 80554664,
"domNodes": -243,
"jsHeapTotalBytes": 1011712,
"scriptDurationMs": 781.4169999999999,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-zoom",
"durationMs": 3713.7740000000576,
"styleRecalcs": 65,
"styleRecalcDurationMs": 16.024,
"layouts": 60,
"layoutDurationMs": 15.425999999999998,
"taskDurationMs": 1742.242,
"heapDeltaBytes": -15952812,
"heapUsedBytes": 73677016,
"domNodes": -251,
"jsHeapTotalBytes": -495616,
"scriptDurationMs": 740.8249999999999,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-zoom",
"durationMs": 3551.7449999999826,
"styleRecalcs": 64,
"styleRecalcDurationMs": 15.149,
"layouts": 60,
"layoutDurationMs": 15.660999999999998,
"taskDurationMs": 1633.775,
"heapDeltaBytes": 9525508,
"heapUsedBytes": 74955956,
"domNodes": -248,
"jsHeapTotalBytes": 786432,
"scriptDurationMs": 732.2059999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66999999999998,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3102.8519999999844,
"styleRecalcs": 42,
"styleRecalcDurationMs": 7.162999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1857.475,
"heapDeltaBytes": 25809492,
"heapUsedBytes": 91599304,
"domNodes": 4,
"jsHeapTotalBytes": 679936,
"scriptDurationMs": 907.9,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3076.489000000038,
"styleRecalcs": 41,
"styleRecalcDurationMs": 6.486000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1833.4189999999999,
"heapDeltaBytes": 21030720,
"heapUsedBytes": 87176268,
"domNodes": 2,
"jsHeapTotalBytes": 1466368,
"scriptDurationMs": 930.519,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666636,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "minimap-idle",
"durationMs": 2048.475000000053,
"styleRecalcs": 8,
"styleRecalcDurationMs": 6.884000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 503.1349999999999,
"heapDeltaBytes": -5686032,
"heapUsedBytes": 82747756,
"domNodes": -248,
"jsHeapTotalBytes": -233472,
"scriptDurationMs": 34.382,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2020.9970000000794,
"styleRecalcs": 8,
"styleRecalcDurationMs": 6.03,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 544.344,
"heapDeltaBytes": 20639140,
"heapUsedBytes": 84530824,
"domNodes": -243,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 35.82000000000001,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 574.2390000000341,
"styleRecalcs": 48,
"styleRecalcDurationMs": 10.540000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 382.61,
"heapDeltaBytes": 14074240,
"heapUsedBytes": 83851840,
"domNodes": 22,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 134.071,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 605.1589999999578,
"styleRecalcs": 47,
"styleRecalcDurationMs": 10.738999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 392.34399999999994,
"heapDeltaBytes": 13772384,
"heapUsedBytes": 83390448,
"domNodes": 20,
"jsHeapTotalBytes": 5505024,
"scriptDurationMs": 136.14399999999998,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2001.1599999997998,
"styleRecalcs": 10,
"styleRecalcDurationMs": 6.760999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 454.9050000000001,
"heapDeltaBytes": 10303712,
"heapUsedBytes": 80219712,
"domNodes": -276,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 9.405,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2022.0420000000559,
"styleRecalcs": 11,
"styleRecalcDurationMs": 8.559000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 463.05699999999996,
"heapDeltaBytes": -14727884,
"heapUsedBytes": 55136736,
"domNodes": -275,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 9.454,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1721.1469999999736,
"styleRecalcs": 44,
"styleRecalcDurationMs": 15.091000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 742.598,
"heapDeltaBytes": 7756588,
"heapUsedBytes": 77885760,
"domNodes": -279,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 100.15199999999999,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1726.2479999999414,
"styleRecalcs": 44,
"styleRecalcDurationMs": 14.033999999999997,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 752.922,
"heapDeltaBytes": -4525924,
"heapUsedBytes": 65751964,
"domNodes": -277,
"jsHeapTotalBytes": 6807552,
"scriptDurationMs": 97.897,
"eventListeners": -220,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-transition-enter",
"durationMs": 1367.776000000049,
"styleRecalcs": 18,
"styleRecalcDurationMs": 25.491999999999994,
"layouts": 5,
"layoutDurationMs": 10.138000000000002,
"taskDurationMs": 861.831,
"heapDeltaBytes": 19460616,
"heapUsedBytes": 106585952,
"domNodes": 13667,
"jsHeapTotalBytes": 11010048,
"scriptDurationMs": 28.451999999999998,
"eventListeners": 2383,
"totalBlockingTimeMs": 132,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "viewport-pan-sweep",
"durationMs": 8368.520999999873,
"styleRecalcs": 250,
"styleRecalcDurationMs": 45.11600000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 5530.662,
"heapDeltaBytes": -21069676,
"heapUsedBytes": 66329844,
"domNodes": -252,
"jsHeapTotalBytes": 2912256,
"scriptDurationMs": 2556.314,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "viewport-pan-sweep",
"durationMs": 8278.80899999991,
"styleRecalcs": 250,
"styleRecalcDurationMs": 43.273,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 5591.057,
"heapDeltaBytes": -6583296,
"heapUsedBytes": 80496420,
"domNodes": -241,
"jsHeapTotalBytes": 1040384,
"scriptDurationMs": 2568.3959999999997,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-idle",
"durationMs": 14476.15299999984,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 14191.784,
"heapDeltaBytes": -63417296,
"heapUsedBytes": 169177132,
"domNodes": -8061,
"jsHeapTotalBytes": -20582400,
"scriptDurationMs": 225.723,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.220000000000073,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-idle",
"durationMs": 14597.887000000128,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 14298.452,
"heapDeltaBytes": -37635072,
"heapUsedBytes": 187710712,
"domNodes": -8061,
"jsHeapTotalBytes": -7348224,
"scriptDurationMs": 229.817,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.779999999999927,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-pan",
"durationMs": 14126.696999999922,
"styleRecalcs": 68,
"styleRecalcDurationMs": 15.217000000000008,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 13725.867999999999,
"heapDeltaBytes": -47692192,
"heapUsedBytes": 166889236,
"domNodes": -8062,
"jsHeapTotalBytes": -13438976,
"scriptDurationMs": 801.4650000000001,
"eventListeners": -14928,
"totalBlockingTimeMs": 119,
"frameDurationMs": 17.219999999999953,
"p95FrameDurationMs": 16.80000000000291
},
{
"name": "vue-large-graph-pan",
"durationMs": 17360.83899999994,
"styleRecalcs": 129,
"styleRecalcDurationMs": 18.46600000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 17170.155,
"heapDeltaBytes": -68241020,
"heapUsedBytes": 166973032,
"domNodes": -8061,
"jsHeapTotalBytes": -10608640,
"scriptDurationMs": 846.264,
"eventListeners": -14928,
"totalBlockingTimeMs": 10,
"frameDurationMs": 17.780000000000047,
"p95FrameDurationMs": 16.80000000000291
},
{
"name": "vue-node-resize-workload",
"durationMs": 5331.176000000141,
"styleRecalcs": 20,
"styleRecalcDurationMs": 68.56199999999998,
"layouts": 20,
"layoutDurationMs": 25.438000000000002,
"taskDurationMs": 5287.420999999999,
"heapDeltaBytes": -58617840,
"heapUsedBytes": 181314708,
"domNodes": -8061,
"jsHeapTotalBytes": 11939840,
"scriptDurationMs": 95.166,
"eventListeners": -14928,
"totalBlockingTimeMs": 230,
"frameDurationMs": 17.219999999999953,
"p95FrameDurationMs": 16.80000000000109
},
{
"name": "vue-node-resize-workload",
"durationMs": 5330.81100000004,
"styleRecalcs": 20,
"styleRecalcDurationMs": 72.28700000000002,
"layouts": 20,
"layoutDurationMs": 26.167999999999996,
"taskDurationMs": 5270.563999999999,
"heapDeltaBytes": -60768548,
"heapUsedBytes": 180451544,
"domNodes": -8061,
"jsHeapTotalBytes": 8450048,
"scriptDurationMs": 98.33200000000001,
"eventListeners": -14928,
"totalBlockingTimeMs": 320,
"frameDurationMs": 17.223333333333358,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-zoom-out-idle",
"durationMs": 7352.489999999989,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7168.311000000001,
"heapDeltaBytes": 8604108,
"heapUsedBytes": 182724980,
"domNodes": 0,
"jsHeapTotalBytes": -8388608,
"scriptDurationMs": 116.58099999999993,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.220000000000073,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-zoom-out-idle",
"durationMs": 7222.4369999999,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7132.675000000001,
"heapDeltaBytes": 20176860,
"heapUsedBytes": 193404296,
"domNodes": 0,
"jsHeapTotalBytes": -28573696,
"scriptDurationMs": 118.40099999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.219999999999953,
"p95FrameDurationMs": 16.80000000000291
},
{
"name": "workflow-execution",
"durationMs": 457.31900000009773,
"styleRecalcs": 15,
"styleRecalcDurationMs": 20.565,
"layouts": 2,
"layoutDurationMs": 0.603,
"taskDurationMs": 115.92700000000004,
"heapDeltaBytes": 4828892,
"heapUsedBytes": 70349092,
"domNodes": 124,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 10.115,
"eventListeners": 65,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "workflow-execution",
"durationMs": 444.6330000000671,
"styleRecalcs": 14,
"styleRecalcDurationMs": 22.9,
"layouts": 4,
"layoutDurationMs": 2.186,
"taskDurationMs": 112.098,
"heapDeltaBytes": 4912060,
"heapUsedBytes": 70790980,
"domNodes": 123,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 9.534,
"eventListeners": 81,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.800000000000182
}
]
}🌾 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 (8 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 6edcb6b69fa5d13b4ceb9abc362c619425efe24d
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 |
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## main #18099 +/- ##
==========================================
+ Coverage 83.49% 84.49% +0.99%
==========================================
Files 2988 2285 -703
Lines 147957 130680 -17277
Branches 38210 33370 -4840
==========================================
- Hits 123540 110420 -13120
+ Misses 23132 19960 -3172
+ Partials 1285 300 -985
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 704 files with indirect coverage changes 🚀 New features to boost your workflow:
|
Playwright videos for new testsDownload the recorded videos and HTML report Recorded projects: cloud Recorded test titles:
|
|
Fix assessment: not trivial. Traced every doc-creation path in The fail-closed behavior in Suggested next steps:
Generated by Claude Code |
…mid-turn doc reset (PM-1355-shaped) Intentionally-failing repro for the nightly bug report (Jo Zhang, 2026-09-20): asking the in-app agent to run the canvas workflow blanks the canvas the moment the agent starts working, for the whole run, with every node reappearing at once when the run finishes. No fix is attempted here. Two independently-confirmed frontend mechanisms combine to produce exactly that timeline: - A `doc_reset`/`follower_replaced` frame synchronously sweeps every live materialized node off the canvas (`AgentCrdtProjection.clearForReset`) -- the only path that actively removes already-rendered nodes. - The resubscribe that follows can be acknowledged (`doc_subscribed:ok`) without its catch-up `doc_update`. This exact anomaly is visible in the bug report's own attached debug log, on a different (later, smaller) workflow in the same session, self-healing only via the passive 30-second `STALE_AFTER_MS` stale-probe heartbeat (`agentCrdtDocLifecycle.ts`) -- there is no faster active recovery. Together, whatever triggers the reset, the canvas can sit blank for up to 30 real seconds with zero active recovery -- comfortably spanning one generation run (this report's own run + wait_for_job took 15386ms). Step 0 (existing PR check): searched open PRs across ComfyUI_frontend/cloud for "canvas blank"/"run"/"agent run"/"crdt-authority" during this run. #18099 ("repro schema-gate latch leaving canvas at zero nodes") is the closest sibling but covers a PERMANENT latch (until an explicit doc_reset) triggered by an unreadable schema_version -- not this transient, self-healing, run-duration-scoped blank. #18064 (new-chat stale canvas) is a different trigger (thread binding, not a mid-turn reset). No existing PR covers this. Open half of the RCA, stated plainly: the backend condition that would emit a doc_reset for a plain, edit-free "run the workflow" turn was NOT found. `run` / `wait_for_job` / `get_output` (cloud/services/agent/internal/loop/ tools.go) make no calls into the CRDT doc store at all, and the turn-start reconciliation that can emit a reset (`ensureDoc`/`prepareWorkflow` in cloud/services/agent/internal/loop/crdt.go) is heavily unit-tested against firing on an unmodified canvas. Which backend path actually emitted the reset in production is unconfirmed; this repro pins what the user sees once one is emitted, for whatever reason, and gives backend engineers a concrete signature (a `doc_reset` frame, or a `doc_subscribed:ok` with no following `doc_update`) to search production logs for around the incident. Verification: - pnpm exec eslint / oxlint / oxfmt --check on the new file -- clean - pnpm exec vue-tsc --project browser_tests/tsconfig.json --noEmit -- clean - pnpm exec playwright test --list -- collects the spec correctly under the cloud project with the right tags - pnpm exec vitest run src/workbench/extensions/agent/crdt/ useAgentCrdtFollower.test.ts -- 61/61 passed, no regressions - Known limitation: this sandbox's egress policy blocks cdn.playwright.dev (403), so the pinned Chromium build could not be downloaded and the spec could not be executed against a real browser here (same limitation noted on sibling repro PRs #18099/#18096 in this program). CI's Playwright run on this PR is the first real execution; treat its result, not this description, as the authoritative pass/fail signal. Related Slack thread: #comfy-agent-user-feedback, Jo Zhang's nightly report, 2026-09-20 03:24 UTC.
…mid-turn doc reset (Comfy-Org#18179) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C) (#comfy-agent-user-feedback, Jo Zhang's nightly report, 2026-09-20 03:24 UTC)_ ## Summary Intentionally-failing repro for Jo Zhang's nightly bug report: asking the in-app agent to run the canvas workflow (ask mode) makes the canvas go fully blank the moment the agent starts working, blank for the whole run, with every node reappearing at once once the run finishes. Ask-mode confirmation UI and the run itself both worked correctly (`print_workflow`/`validate`/`run`/`wait_for_job`/`get_output` all `ok`, `wait_for_job` 15386ms) — only the canvas rendering is affected. **No fix is attempted here.** The spec is now two tests sharing one arrange helper: the first pins the known defect (`test.fail()`, canvas still blank the instant the run reports done), the second pins that the passive stale-probe genuinely recovers once given the chance — the "all nodes came back" half of the report is real, intended behavior, not part of the bug. ## Mechanism (confirmed) Two independently-confirmed frontend behaviors combine to produce exactly this timeline: 1. A `doc_reset`/`follower_replaced` frame makes `AgentCrdtProjection.clearForReset` (`src/workbench/extensions/agent/crdt/agentCrdtProjection.ts`) synchronously sweep every live materialized node off the canvas, before any replacement content exists. This is the only path that actively removes already-rendered nodes — already pinned by an existing test (`useAgentCrdtFollower.test.ts`: "reconciles after a doc_reset clear, without waiting for another frame"). 2. The resubscribe that follows can be acknowledged (`doc_subscribed: {ok: true}`) **without** its catch-up `doc_update`. This exact anomaly is visible in the bug report's own attached debug log, on a different (later, smaller) workflow in the same session: a content-less `ok` ack, followed exactly 30 seconds later — `STALE_AFTER_MS` in `agentCrdtDocLifecycle.ts` — by the passive stale-probe's forced resubscribe, which is the one that actually delivers content and materializes the nodes. There is no faster active recovery (already pinned by "a confirmed subscribe clears the retry timer", which asserts nothing fires before `STALE_AFTER_MS`). Together: whatever triggers the reset, the canvas can sit fully blank for up to 30 real seconds afterward with zero active recovery — comfortably spanning one generation run (this report's own run + wait_for_job took 15386ms) — sufficient by itself to produce the reported timeline, independent of whether the run itself actually finished at that moment. ## Open half of the RCA (stated plainly) The backend condition that would emit a `doc_reset` for a plain, edit-free "run the workflow" turn was **not found**: - `run` / `validate` / `wait_for_job` / `get_output` (`cloud/services/agent/internal/loop/tools.go`) make no calls into the CRDT doc store at all (`docDeps()`, `replaceDocFromClient`, `.pub.Reset` — zero hits). - The turn-start reconciliation that *can* emit a reset (`ensureDoc`/`prepareWorkflow` in `cloud/services/agent/internal/loop/crdt.go`) is heavily unit-tested against firing on an unmodified canvas (`TestHumanDraftWithoutAcceptanceSnapshotResyncsFromDocument`, `TestClientSeedDoesNotOverwriteANewerClientSeed`) — both the doc projection and the posted draft pass through the same `draft.NormalizeGraph`. - Ask-mode's `answerAsk` (approving the run) posts to a separate REST endpoint (`/agent/threads/{id}/asks/{askId}/answer`) and does not repost `draft`, ruling out a confirmation-time draft/CAS race. - The PM-852/PM-1152 "`workflow_docs` authoritative at turn start" family doesn't cleanly fit either: its promotion PR (cloud#9718) merged to `main` on 2026-09-19T20:10:38Z, but the commit nightly.engcomfy.com actually ran at the incident (`ad2efe2`, committed 2026-09-20T02:59:08Z) is confirmed not a descendant of that merge, yet already carries mature pre-Comfy-Org#8447 CAS/`HumanAuthored` reconciliation guards. Which backend path actually emitted the reset in production is unconfirmed. This repro pins what the user sees once one is emitted, for whatever reason, and gives backend engineers a concrete signature to search production logs for: a `doc_reset` frame, or a `doc_subscribed:ok` with no following `doc_update`, around a `run` tool call. ## Step 0 — checked for an existing PR first - **Comfy-Org#18099** ("repro schema-gate latch leaving canvas at zero nodes") is the closest sibling but covers a **permanent** latch (until an explicit `doc_reset`) triggered by an unreadable `schema_version` — not this transient, self-healing, run-duration-scoped blank. - **Comfy-Org#18064** (new-chat inherits stale canvas, ADR-CRDT-AUTHORITY-0035) is a different trigger (thread/tab binding at new-chat time), not a mid-turn run. - No open PR on either repo covers this exact symptom. Built fresh, targeting `main`. ## Test structure fix CI failed with "expected test to fail, but it passed". Reading the spec found the real defect: the `test.fail()`-guarded assertion was followed by more code (a further `page.clock.fastForward` and two recovery `.toBeVisible()` assertions) that could never actually run — if the guarded assertion throws as intended, execution stops there; if it doesn't throw, Playwright already reports the expected-fail-but-passed failure before reaching the rest. Either way the recovery half was dead code and never verified anything. Fixed by extracting the shared arrange (drive the run turn to the blanked-canvas point) into a `driveRunTurnUntilCanvasIsBlank` helper and splitting into two tests in the same `describe`, matching the pattern already used by `agentZeroNodesOnUnreadableSchema.spec.ts` (Comfy-Org#18099) and `agentTurnSurvivesSocketDrop.spec.ts`: `test.fail()` is now the last thing the first test does, and the recovery check (not a bug — genuinely intended, if slow, behavior) is its own second test with no `test.fail()`. ## Production-fix experiment (added at Christian Byrne's request) A follow-up commit (`2b0b4a9`) adds a real, scoped production fix for the RCA's "no active recovery" gap: `AgentCrdtDocLifecycle` now also arms a short (2s) active catch-up probe right after a subscribe is confirmed, instead of relying solely on the passive 30s stale-probe heartbeat, so a missing catch-up `doc_update` is retried well before the full budget. The repro spec itself (`test.fail()`/assertions) is untouched — this exists to see whether it changes this PR's e2e outcome, not to make the test pass by construction. ## Verification - `pnpm exec oxfmt --check`, `pnpm exec oxlint`, `pnpm exec eslint` on the changed file — clean - `pnpm typecheck` (full `vue-tsc` + workspace packages) — clean - `pnpm exec playwright test --project=cloud --list browser_tests/tests/agent/agentCanvasBlanksDuringRunTurn.spec.ts` — now collects **two** tests under the `cloud` project with `@cloud @agent` tags - **Known limitation, unchanged:** this sandbox cannot install Playwright's browser binaries (blocked egress to the CDN that serves them), so neither test could be executed against a real browser here. CI's Playwright run on this PR remains the first real execution and the authoritative pass/fail signal for the e2e half of this repro. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_012sDXswotMsabTTHot1LTTu --- _Generated by [Claude Code](https://claude.ai/code/session_012sDXswotMsabTTHot1LTTu)_ --------- Co-authored-by: Claude <noreply@anthropic.com>
dante01yoon
left a comment
There was a problem hiding this comment.
Reviewed at 11111c0b. The Vitest half is solid and I'd merge it as-is. The e2e half does not reproduce the mechanism it claims to, and because it's wrapped in test.fail() that defect is invisible in CI — so I'm requesting changes on that one point.
The blocking problem
agentZeroNodesOnUnreadableSchema.spec.ts:239 calls host.apply([...]) and throws away the HostFrame it returns. HostDoc holds no socket — it is a pure frame factory, and apply()'s only effect on the browser is via the doc_update frame it hands back. The established call site does this.hostSocket.send(this.host.apply(entry.ops)) (agentConversationFixture.ts:283); this spec is the only host.apply() call in the repo that discards the result.
So the add_node never goes on the wire. The node is absent from the canvas because the browser was never told about it — not because the schema-gate latch dropped it. I confirmed there is no second delivery path: resubscribe() is only reached from onDocReset (layoutFollowerBridge.ts:342), a schema_error does not re-subscribe, and the spec sends no doc frame after the catch-up. The latch at line 256 is never consulted for this op.
Details inline. Everything below that is non-blocking.
Verified (not taken from the description)
- The latch itself is real and the Vitest case proves it causally:
onDocUpdatedoesif (this.schemaError !== null) returnatlayoutFollowerBridge.ts:256, beforeapplyRemoteUpdate, so a later same-lineage frame is not even merged.assertReadableSchemaruns post-merge at line 303. The new test's asserts line up with that exactly —unreadable-seedpresent (merged before the gate tripped),added-after-latchabsent (never merged). Ran it: 36/36 passed. test.fail()placement is right, and I want to flag that as a positive since it's easy to get wrong. Putting it at line 303 rather than at the top means the setup checkpoints at lines 215/279/283 are genuine failures if they break, instead of being swallowed as expected-failures.corruptSchemaVersion()/repairSchemaVersion()are additive and don't disturb existingHostDocconsumers. The ordering reasoning holds:catchUp()encodes at call time, so repairing at spec line 141 can't retroactively alter the frame sent at line 140.- CI is fully green (45 pass / 0 fail), including the
cloudPlaywright project this spec routes to. That is precisely the problem — withtest.fail(), green cannot distinguish "failed for the intended reason" from "failed because no frame was sent."
I appreciate the Step 0 prior-art check and the Step 2.3 write-up; reading layoutFollowerBridge.ts on main rather than trusting the RCA was the right instinct, and the unit test is the part of this PR that actually discharges it.
Review on #18099 found the repro's host.apply() call discarded the returned HostFrame instead of sending it, so the add_node delta never reached the browser — the final assertion failed because nothing was transmitted, not because the schema-gate latch dropped it, and the defect was invisible under test.fail(). Wrap it in send(), matching the host.apply() -> send() pattern in agentConversationFixture.ts. Fixing that exposed a second gap: repairSchemaVersion() mutated the host's doc but was never sent, and even if it had been, apply()'s own before/after delta is captured fresh at call time, so the repair would never ride along with the add_node frame. Once send() was added the repair's effect still would not reach the browser, which would leave this repro permanently red even after a fix that stops the latch from outliving the one bad frame (rather than tolerating an undefined version outright). Give repairSchemaVersion() the same before/after delta shape as apply()/catchUp() so it returns its own sendable frame, and send it as its own later-seq frame ahead of add_node. That isolates the latch itself as what this repro pins: the schema-version repair alone must be enough for the gate to pass again before add_node arrives. Also reword the spec's header comment and the corrupt-schema inline comment: they stated the missing schema_version as an established root cause, but tracing every doc-creation path in cloud found none that mints a doc without one — the trigger is unproven. State the conditional the test actually pins instead.
|
The
Likely a transient render/animation-timing issue in the topbar-button locator under this job's slow-mo/video-recording conditions. Re-running once. Update — second occurrence,
Generated by Claude Code |
This comment has been minimized.
This comment has been minimized.
christian-byrne
left a comment
There was a problem hiding this comment.
Attempt a fix that swaps test result
|
Re: @christian-byrne's review "Attempt a fix that swaps test result" (top-level review, no inline anchor): Read as: attempt the real product fix for the schema-gate latch so this repro's final assertion flips from failing ( Implemented in
Also fixed an unrelated CI failure on this same spec (found by a parallel investigation): the agent-open button locator was unscoped and hit a strict-mode violation, plus the branch had fallen behind Merged Full verification (vitest for Generated by Claude Code |
Playwright e2e (test.fail) drives the real agent panel against a fake doc host whose catch-up frame has no readable meta.schema_version: the agent reports a successful add_node tool call, the op lands on the host's own document, but the canvas the user is looking at stays at 0 nodes because LayoutFollowerBridge's KA-11 read gate latches for the whole document lineage once tripped, not just the offending frame. Adds a focused Vitest case in followerSubscription.test.ts isolating that exact latch for the undefined-schema_version shape (existing coverage only exercised the "too new" trigger): a schema-valid follow-up frame for the same lineage is still dropped until an explicit doc_reset. HostDoc gains a test-only corruptSchemaVersion() helper so a repro can simulate a doc's meta.schema_version being missing on the wire.
corruptSchemaVersion() removed meta.schema_version from HostDoc's shared Y.Doc permanently, so the spec's own pre-test.fail() sanity checkpoint (host.graph() after host.apply()) started throwing SchemaVersionError before test.fail() ever ran, and CI reported a genuine unexpected failure instead of the intended expected failure. Add HostDoc.repairSchemaVersion() and call it right after the corrupted catch-up frame has been sent, once Y.encodeStateAsUpdate has already captured the corrupted bytes. This only unblocks the host's own later apply()/graph() calls; the browser-side read gate still trips exactly as intended, so the repro still proves the same bug.
agentZeroNodesOnUnreadableSchema.spec.ts asserted a blanket vueNodes.nodes toHaveCount(0) before test.fail(), which is only true when the canvas starts empty. Under the cloud Playwright project (VITE_USE_LEGACY_DEFAULT_GRAPH: 'true'), the app boots with 7 pre-existing default-graph nodes, so that assertion failed for real before test.fail() ever ran, turning the intended expected-failure into a genuine CI failure. Scope the pre-test.fail() visual-proof check to the one node the agent claims to have added (ADDED_NODE_ID), matching the pattern in sibling repro agentGraphOpLostOnDisconnect.spec.ts, which only ever asserts on that node's own locator.
Review on #18099 found the repro's host.apply() call discarded the returned HostFrame instead of sending it, so the add_node delta never reached the browser — the final assertion failed because nothing was transmitted, not because the schema-gate latch dropped it, and the defect was invisible under test.fail(). Wrap it in send(), matching the host.apply() -> send() pattern in agentConversationFixture.ts. Fixing that exposed a second gap: repairSchemaVersion() mutated the host's doc but was never sent, and even if it had been, apply()'s own before/after delta is captured fresh at call time, so the repair would never ride along with the add_node frame. Once send() was added the repair's effect still would not reach the browser, which would leave this repro permanently red even after a fix that stops the latch from outliving the one bad frame (rather than tolerating an undefined version outright). Give repairSchemaVersion() the same before/after delta shape as apply()/catchUp() so it returns its own sendable frame, and send it as its own later-seq frame ahead of add_node. That isolates the latch itself as what this repro pins: the schema-version repair alone must be enough for the gate to pass again before add_node arrives. Also reword the spec's header comment and the corrupt-schema inline comment: they stated the missing schema_version as an established root cause, but tracing every doc-creation path in cloud found none that mints a doc without one — the trigger is unproven. State the conditional the test actually pins instead.
…e schema_version LayoutFollowerBridge.onDocUpdate latched schemaError for the whole document lineage after the first unreadable meta.schema_version, before the frame was even merged into the follower's doc. That made a later same-lineage repair permanently unreachable, so the canvas stayed empty even after a valid schema_version and subsequent ops arrived. Merge every same-lineage, in-order frame regardless of latch state, and re-check the KA-11 read gate after each merge instead of only the first. A later frame that restores a readable schema_version now clears schemaError and resumes projecting. Updates the two followerSubscription.test.ts cases that pinned the old latch-forever behavior to instead prove the un-latch, and flips agentZeroNodesOnUnreadableSchema.spec.ts from test.fail() to a normal passing assertion (PM-1343).
page.getByRole('button', { name: OPEN_AGENT_LABEL }) with no scope was a
strict-mode violation in CI (playwright-video-new-tests and
playwright-tests (cloud) both failed at this line, matching 63-81
buttons including unrelated topbar/tab buttons). Scope to
[data-testid=integrated-tab-bar-actions] and wait for
data-agent-gate-settled first, matching the pattern already used in
agentChatRefreshPersistence.spec.ts and agentPanelLifecycle.spec.ts for
this same button.
468f6c1 to
981bd86
Compare
christian-byrne
left a comment
There was a problem hiding this comment.
Reviewed the rebased implementation and regression coverage at 981bd86. The original dropped-frame and unsent-repair blockers are fixed; the gate now remains fail-closed for unreadable merged state and resumes only after a later frame restores a readable schema. Verified locally: 58 CRDT test files / 901 cases, focused regression suite, root and browser typechecks, lint, format, Playwright discovery, and pre-push knip.
The sole blocking issue was addressed: the repair and add_node frames are now both transmitted, the E2E is a normal passing assertion, and all four review threads are resolved. Re-reviewed and approved at 981bd86 after rebase and local validation.
|
|
Backport the production fix and focused follower unit coverage. The main-branch browser spec is omitted because the 1.54 browser harness lacks its required APIs.
Backport the production fix and focused follower unit coverage. The main-branch browser spec is omitted because the 1.54 browser harness lacks its required APIs.
…g canvas at zero nodes (#18541) ## Summary Manual backport of #18099 to `core/1.54` after the label-driven cherry-pick conflicted in `layoutFollowerBridge.ts`. ## Resolution - Preserves the release branch inline stale/gap/read-gate implementation. - Applies the production behavior and focused follower unit coverage. - Omits the main-only browser spec and its fixture helpers because the 1.54 browser harness lacks the APIs required to typecheck it. ## Verification - `pnpm typecheck` - `pnpm typecheck:browser` - `pnpm test:unit src/workbench/extensions/agent/crdt/followerSubscription.test.ts` — 34 passed - pre-push `knip` Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
…ng canvas at zero nodes (#18542) ## Summary Manual backport of #18099 to `cloud/1.54` after the label-driven cherry-pick conflicted in `layoutFollowerBridge.ts`. ## Resolution - Preserves the release branch inline stale/gap/read-gate implementation. - Applies the production behavior and focused follower unit coverage. - Omits the main-only browser spec and its fixture helpers because the 1.54 browser harness lacks the APIs required to typecheck it. ## Verification - `pnpm typecheck` - `pnpm typecheck:browser` - `pnpm test:unit src/workbench/extensions/agent/crdt/followerSubscription.test.ts` — 41 passed - pre-push `knip` Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
|
|
<!-- ccr-slack-attribution -->
Requested by Christian Byrne · Slack thread
Summary
Started as an intentionally-failing repro for the in-app agent reporting a turn as done while the canvas ends up with zero nodes (agent CRDT schema-version fail-closed with no signal to backend). Per review feedback (see below), a real fix for the underlying schema-gate latch bug has now landed alongside the repro:
test.fail()is removed and the repro's final assertion passes normally.Changes
browser_tests/tests/agent/agentZeroNodesOnUnreadableSchema.spec.tsdrives the real agent panel against a fake doc host (HostDoc, the same fixture the sibling PM-1260 repro (fix(agent): retry an unanswered CRDT doc subscribe after transport reconnect #18073) andagentConversationFixtureuse). The host's catch-up frame is corrupted so its merged doc'smeta.schema_versionis missing/undefined — an unreadable-schema shapeassertReadableSchema(KA-11,src/workbench/extensions/agent/crdt/schemaGuard.ts) refuses; see the spec's header comment for why this trigger is simulated rather than a proven real-world path. A schema-version repair and then a successfuladd_nodetool call (reported over chat) are each sent to the browser as their own frames on the host's own authoritative document, and now both actually reach the canvas.followerSubscription.test.ts›FE-KA11-1, isolating the latch mechanism directly againstLayoutFollowerBridge(no browser, no panel).browser_tests/fixtures/agentConversationHostDoc.ts's sharedHostDoctest fixture gains two additive, test-only methods:corruptSchemaVersion()andrepairSchemaVersion(), which restoresmeta.schema_versionand returns the delta frame carrying just that repair (mirroringapply()'s before/after shape).src/workbench/extensions/agent/crdt/layoutFollowerBridge.ts'sonDocUpdateused to latchschemaErrorfor the entire document lineage the moment one merged frame'smeta.schema_versionwas unreadable (if (this.schemaError !== null) return, checked before the frame was even merged). That made a later same-lineage repair permanently unreachable — the repair frame was silently dropped along with everything after it, forever, until an explicitdoc_reset. Now every same-lineage, in-order frame is merged regardless of latch state, and the KA-11 read gate (assertReadableSchema, now behind theisReadableUpdate()helper) is re-checked after every merge instead of only the first. A later frame that restores a readablemeta.schema_versionclearsschemaErrorand resumes projecting normally — the gate now un-latches on repair instead of only on a lineage reset.Before / After
main): once a doc's schema becomes unreadable (e.g. a catch-up frame with nometa.schema_version), the CRDT follower dropped every later update for that document's lineage forever, with nodoc_resetfollowing an ordinary tool call and no signal to the backend. The agent kept reporting tool-call and turn success because nothing told it the browser stopped projecting. The user saw the agent claim success while the canvas stayed empty.meta.schema_versionun-latches the gate, so ops sent afterward (e.g.add_node) merge and project normally. The repro's final assertion (expect(vueNodes.getNodeLocator(...)).toBeVisible()) passes withouttest.fail().Step 0 — checked for an existing PR first
layoutFollowerBridge.tsanduseAgentCrdtFollower.ts, but its whole diff is about a different drop family:FollowerApplyError(malformed Yjs updates) andEcsFollowerAdapter.applyFramereturningfalsefor no bound session. It does not cover this bug.layoutFollowerBridge.ts's schema handling, but only for the subscribe-time refusal case. It does not cover the post-mergeassertReadableSchemalatch either.schemaGuard.ts, theschemaError/KA-11 latch, or this specific case was found.Verification
NODE_OPTIONS="--no-experimental-webstorage" pnpm exec vitest run src/workbench/extensions/agent/crdt/followerSubscription.test.ts— 43 passed, 2 expected fail (pre-existing, unrelated, frommain's ownit.failscases), including the twoFE-KA11-1cases now proving the un-latch-on-repair behavior.NODE_OPTIONS="--no-experimental-webstorage" pnpm exec vitest run src/workbench/extensions/agent/crdt/— full directory post-merge, 859 passed, 15 expected fail (pre-existing, unrelated) — no regressions.pnpm typecheck— passed (clean across the whole workspace).pnpm run typecheck:browser(vue-tsc --project browser_tests/tsconfig.json) — passed.pnpm exec eslint/pnpm exec oxfmt --check/pnpm exec oxlinton all changed files — passed.pnpm exec playwright test --list browser_tests/tests/agent/agentZeroNodesOnUnreadableSchema.spec.ts— collects successfully.cdn.playwright.dev(403), so the Chromium binary could not be downloaded and the e2e spec could not be executed against a real browser in this session — only collected (--list), typechecked, and linted. CI's actual Playwright run on this PR is the authoritative pass/fail signal for the e2e half.Step 2.3 — did the RCA's latch claim hold up?
onDocUpdate's early return (before the sequence/apply logic, so a later frame was not even merged into the follower'sY.Doc) was present, unmodified, onmainat the commit this branch was cut from.doc_reset.Response to review (dante01yoon)
host.apply(...)'s returnedHostFramewas discarded instead of being sent, soadd_nodenever reached the browser. Fixed: wrapped insend(...), matchingagentConversationFixture.ts'sthis.hostSocket.send(this.host.apply(entry.ops))pattern.repairSchemaVersion()'s effect never reached the browser becauseapply()computes its own before/after delta fresh at call time. Restructured:repairSchemaVersion()now returns its own before/after delta frame, exactly likeapply()/catchUp(), sent as its own later-seq frame beforeadd_node.corruptSchemaVersion()andrepairSchemaVersion()in this description.Response to review (christian-byrne) — "Attempt a fix that swaps test result"
Read as: attempt the real product fix for the schema-gate latch so the repro's final assertion flips from failing (
test.fail()) to passing — matching this PR's own "After (once a fix lands)" note above. Implemented inlayoutFollowerBridge.ts(see "Production fix" above): the KA-11 gate now re-checks on every merged frame and un-latches when a later frame restores a readableschema_version, instead of latching forever on the first failure.test.fail()is removed from the e2e spec and its final assertion (toBeVisible()) now passes; the twofollowerSubscription.test.tscases that previously pinned the "latches forever" behavior now prove the un-latch-on-repair behavior instead. Fix scoped tolayoutFollowerBridge.tsonly (noschemaGuard.tschange needed).Also folded in a second, unrelated fix for the CI re-run failures a parallel investigation found on this same spec:
page.getByRole('button', { name: OPEN_AGENT_LABEL })was an unscoped locator that hit a strict-mode violation (matching 63-81 buttons — unrelated topbar/tab/feedback/user buttons) and was failing both theplaywright-video-new-testsandplaywright-tests (cloud)jobs (107028559537, 107022232650) for real, not flake. Root cause:mainhad already renamed the entry button's i18n key fromagent.askComfyAgent("Ask Comfy Agent") toagent.entryButton("Agent", now rendered byAgentEntryButton.vue) before this branch merged, soOPEN_AGENT_LABELresolved toundefinedat runtime — an unnamedgetByRole('button', ...)matches every button on the page, which is exactly the 63-81-match symptom (including the "Agent" button itself). Updated the constant toenMessages.agent.entryButton(matching every other agent spec already onmain) and additionally scoped/gated the locator to[data-testid=integrated-tab-bar-actions]+data-agent-gate-settled, the pattern already used inagentChatRefreshPersistence.spec.tsandagentPanelLifecycle.spec.ts.Merge with main
Merged
maininto this branch (merge commit, no rebase/force-push) to pick up the base's moved-on history. One real conflict, inlayoutFollowerBridge.ts:mainhad independently refactoredonDocUpdateintorejectStaleUpdate/rejectSequenceGap/isReadableUpdatehelpers and added aClassifiedDocUpdatedispatch shape. Resolved by keepingmain's refactored structure and adding this PR's fix inside it:isReadableUpdate()now clearsthis.schemaError = nullon a successful re-check, preserving the same un-latch-on-repair behavior undermain's current method layout.followerSubscription.test.tsandagentConversationHostDoc.tsauto-merged cleanly. Re-ran the full verification pass (vitest, typecheck, typecheck:browser, eslint, oxfmt, playwright --list) after the merge; all green.🤖 Generated with Claude Code
https://claude.ai/code/session_01UZ9MAkZ9grYSDqqbmMdXvE
Generated by Claude Code