Repository navigation
fix: copy/paste beside the agent panel - #18062
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: ✅ 2073 passed, 0 failed · 2 flaky📊 Browser Reports
📦 Bundle: 9.96 MB gzip 🟢 -635 BDetailsSummary
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 — 1.46 MB (baseline 1.46 MB) • 🔴 +747 BGraph 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: 6 added / 6 removed / 5 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 — 123 kB (baseline 123 kB) • 🔴 +45 BReusable component library chunks
Status: 7 added / 7 removed / 9 unchanged Data & Services — 3.87 MB (baseline 3.88 MB) • 🟢 -2.5 kBStores, services, APIs, and repositories
Status: 13 added / 13 removed / 3 unchanged Utilities & Hooks — 577 kB (baseline 577 kB) • ⚪ 0 BHelpers, composables, and utility bundles
Status: 19 added / 19 removed / 19 unchanged Vendor & Third-Party — 17.4 MB (baseline 17.4 MB) • ⚪ 0 BExternal libraries and shared vendor chunks Status: 18 unchanged Other — 17.8 MB (baseline 17.8 MB) • 🔴 +39 BBundles that do not match a named category
Status: 73 added / 73 removed / 229 unchanged ⚡ Performance Report
Show regressions
All metrics
Historical variance (last 15 runs)
Trend (last 15 commits on main)
Raw data{
"timestamp": "2026-09-21T22:54:15.739Z",
"gitSha": "a76b2129a93262967a75ab5e3b0da1358919be77",
"branch": "test/copy-paste-agent-repro",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 1998.5080000000153,
"styleRecalcs": 9,
"styleRecalcDurationMs": 8.195,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 448.605,
"heapDeltaBytes": 12999832,
"heapUsedBytes": 74789644,
"domNodes": -261,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 8.076,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-idle",
"durationMs": 2004.3569999999136,
"styleRecalcs": 8,
"styleRecalcDurationMs": 6.957,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 439.81500000000005,
"heapDeltaBytes": -457112,
"heapUsedBytes": 61575260,
"domNodes": -264,
"jsHeapTotalBytes": 4710400,
"scriptDurationMs": 7.7799999999999985,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1792.917999999986,
"styleRecalcs": 73,
"styleRecalcDurationMs": 26.277,
"layouts": 14,
"layoutDurationMs": 4.28,
"taskDurationMs": 820.6020000000001,
"heapDeltaBytes": -5059024,
"heapUsedBytes": 56139256,
"domNodes": -264,
"jsHeapTotalBytes": 4448256,
"scriptDurationMs": 160.486,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1798.5589999999547,
"styleRecalcs": 74,
"styleRecalcDurationMs": 26.443,
"layouts": 14,
"layoutDurationMs": 4.101000000000001,
"taskDurationMs": 837.177,
"heapDeltaBytes": -3527452,
"heapUsedBytes": 63457260,
"domNodes": -261,
"jsHeapTotalBytes": 6021120,
"scriptDurationMs": 166.68800000000002,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1706.2429999999722,
"styleRecalcs": 31,
"styleRecalcDurationMs": 16.907999999999998,
"layouts": 6,
"layoutDurationMs": 1.255,
"taskDurationMs": 256.568,
"heapDeltaBytes": 2175844,
"heapUsedBytes": 64387660,
"domNodes": 75,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 10.654,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1707.2239999999965,
"styleRecalcs": 30,
"styleRecalcDurationMs": 14.456,
"layouts": 6,
"layoutDurationMs": 1.1139999999999999,
"taskDurationMs": 322.57800000000003,
"heapDeltaBytes": 2187344,
"heapUsedBytes": 64361636,
"domNodes": 74,
"jsHeapTotalBytes": 5505024,
"scriptDurationMs": 10.272999999999998,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-clipping",
"durationMs": 639.4379999999842,
"styleRecalcs": 10,
"styleRecalcDurationMs": 7.107000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 444.97099999999995,
"heapDeltaBytes": 18923916,
"heapUsedBytes": 80805664,
"domNodes": 16,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 169.30200000000002,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 643.5330000000477,
"styleRecalcs": 11,
"styleRecalcDurationMs": 7.886999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 447.96099999999996,
"heapDeltaBytes": 18254340,
"heapUsedBytes": 80440440,
"domNodes": 18,
"jsHeapTotalBytes": 6029312,
"scriptDurationMs": 164.471,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1182.657000000006,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 837.1419999999999,
"heapDeltaBytes": 34542560,
"heapUsedBytes": 104881112,
"domNodes": 0,
"jsHeapTotalBytes": 19034112,
"scriptDurationMs": 70.983,
"eventListeners": 0,
"totalBlockingTimeMs": 570,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1210.599000000002,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 893.857,
"heapDeltaBytes": -12854560,
"heapUsedBytes": 73108500,
"domNodes": -295,
"jsHeapTotalBytes": 20054016,
"scriptDurationMs": 67.17200000000001,
"eventListeners": -217,
"totalBlockingTimeMs": 542,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 831.2690000000202,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 429.3739999999997,
"heapDeltaBytes": 13636804,
"heapUsedBytes": 88808256,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 159.196,
"eventListeners": 0,
"totalBlockingTimeMs": 219,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.80000000000109
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 841.2299999999959,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 471.5619999999996,
"heapDeltaBytes": 14536508,
"heapUsedBytes": 91527144,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 161.39100000000002,
"eventListeners": 0,
"totalBlockingTimeMs": 225,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "extension-clean-combined-draw",
"durationMs": 10926.688000000013,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10012.710000000001,
"heapDeltaBytes": -11103584,
"heapUsedBytes": 71776644,
"domNodes": -16,
"jsHeapTotalBytes": 6815744,
"scriptDurationMs": 8266.874,
"eventListeners": 0,
"totalBlockingTimeMs": 1387,
"frameDurationMs": 20.00000000000006,
"p95FrameDurationMs": 50
},
{
"name": "extension-clean-combined-draw",
"durationMs": 10707.247000000052,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 9869.791000000001,
"heapDeltaBytes": 3343360,
"heapUsedBytes": 85557664,
"domNodes": -18,
"jsHeapTotalBytes": 6815744,
"scriptDurationMs": 8219.506000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 1341,
"frameDurationMs": 20.003333333333345,
"p95FrameDurationMs": 49.899999999999636
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 11117.404000000022,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10184.801,
"heapDeltaBytes": 1656444,
"heapUsedBytes": 83428732,
"domNodes": -12,
"jsHeapTotalBytes": 11272192,
"scriptDurationMs": 8511.765000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 1529,
"frameDurationMs": 18.886666666666617,
"p95FrameDurationMs": 50
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 10908.933000000048,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10042.275,
"heapDeltaBytes": 4345476,
"heapUsedBytes": 86327160,
"domNodes": -14,
"jsHeapTotalBytes": 11010048,
"scriptDurationMs": 8373.338,
"eventListeners": 0,
"totalBlockingTimeMs": 1387,
"frameDurationMs": 18.889999999999965,
"p95FrameDurationMs": 50
},
{
"name": "large-graph-idle",
"durationMs": 2023.2489999999643,
"styleRecalcs": 5,
"styleRecalcDurationMs": 4.511000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 529.2439999999999,
"heapDeltaBytes": 2041748,
"heapUsedBytes": 83140396,
"domNodes": -251,
"jsHeapTotalBytes": -1019904,
"scriptDurationMs": 29.500999999999998,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-idle",
"durationMs": 2026.7990000000964,
"styleRecalcs": 5,
"styleRecalcDurationMs": 4.536000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 561.938,
"heapDeltaBytes": 2987644,
"heapUsedBytes": 83717864,
"domNodes": -251,
"jsHeapTotalBytes": -1282048,
"scriptDurationMs": 34.331,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-pan",
"durationMs": 2288.7610000000222,
"styleRecalcs": 67,
"styleRecalcDurationMs": 12.606999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1647.1860000000001,
"heapDeltaBytes": 1411776,
"heapUsedBytes": 81512264,
"domNodes": -224,
"jsHeapTotalBytes": 1789952,
"scriptDurationMs": 846.1400000000001,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-pan",
"durationMs": 2273.038999999926,
"styleRecalcs": 66,
"styleRecalcDurationMs": 14.332,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1636.886,
"heapDeltaBytes": 16141760,
"heapUsedBytes": 80168624,
"domNodes": -251,
"jsHeapTotalBytes": 1798144,
"scriptDurationMs": 840.0609999999999,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-zoom",
"durationMs": 3502.7299999999855,
"styleRecalcs": 61,
"styleRecalcDurationMs": 10.018000000000002,
"layouts": 60,
"layoutDurationMs": 12.318,
"taskDurationMs": 1650.7369999999999,
"heapDeltaBytes": 7574708,
"heapUsedBytes": 73695416,
"domNodes": 2,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 775.925,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-zoom",
"durationMs": 3536.465999999905,
"styleRecalcs": 62,
"styleRecalcDurationMs": 11.769000000000002,
"layouts": 60,
"layoutDurationMs": 12.431000000000001,
"taskDurationMs": 1720.2360000000003,
"heapDeltaBytes": -15692824,
"heapUsedBytes": 73608512,
"domNodes": -263,
"jsHeapTotalBytes": -1019904,
"scriptDurationMs": 774.375,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3043.8990000000103,
"styleRecalcs": 40,
"styleRecalcDurationMs": 4.788,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1926.9700000000003,
"heapDeltaBytes": 25440276,
"heapUsedBytes": 91459244,
"domNodes": 0,
"jsHeapTotalBytes": 417792,
"scriptDurationMs": 992.5880000000001,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3047.113999999965,
"styleRecalcs": 42,
"styleRecalcDurationMs": 6.276000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1906.3680000000002,
"heapDeltaBytes": 22860232,
"heapUsedBytes": 89043648,
"domNodes": 4,
"jsHeapTotalBytes": 679936,
"scriptDurationMs": 966.6799999999998,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2010.4359999999701,
"styleRecalcs": 6,
"styleRecalcDurationMs": 4.957000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 384.57,
"heapDeltaBytes": 20216680,
"heapUsedBytes": 84163484,
"domNodes": -250,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 30.770999999999994,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "minimap-idle",
"durationMs": 2017.6360000000386,
"styleRecalcs": 5,
"styleRecalcDurationMs": 6.225999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 548.926,
"heapDeltaBytes": -4310608,
"heapUsedBytes": 83647720,
"domNodes": -252,
"jsHeapTotalBytes": -1544192,
"scriptDurationMs": 33.34599999999999,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 590.7849999999826,
"styleRecalcs": 47,
"styleRecalcDurationMs": 10.335,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 393.831,
"heapDeltaBytes": 14543092,
"heapUsedBytes": 78118512,
"domNodes": 20,
"jsHeapTotalBytes": 5505024,
"scriptDurationMs": 144.571,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 587.2150000000147,
"styleRecalcs": 49,
"styleRecalcDurationMs": 12.472,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 418.3859999999999,
"heapDeltaBytes": 14544260,
"heapUsedBytes": 78105056,
"domNodes": 24,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 147.54099999999997,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 1999.054000000001,
"styleRecalcs": 11,
"styleRecalcDurationMs": 10.579,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 441.3670000000001,
"heapDeltaBytes": -8280520,
"heapUsedBytes": 55129092,
"domNodes": -276,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 7.184000000000003,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2006.8360000000212,
"styleRecalcs": 9,
"styleRecalcDurationMs": 8.462000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 439.42400000000004,
"heapDeltaBytes": -6492536,
"heapUsedBytes": 56899284,
"domNodes": -277,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 7.227000000000001,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1719.0089999999714,
"styleRecalcs": 77,
"styleRecalcDurationMs": 26.304000000000002,
"layouts": 18,
"layoutDurationMs": 4.9750000000000005,
"taskDurationMs": 760.624,
"heapDeltaBytes": -6348672,
"heapUsedBytes": 56876196,
"domNodes": -275,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 104.51,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1705.157999999983,
"styleRecalcs": 78,
"styleRecalcDurationMs": 26.461000000000006,
"layouts": 18,
"layoutDurationMs": 4.756,
"taskDurationMs": 766.517,
"heapDeltaBytes": -6250780,
"heapUsedBytes": 57110664,
"domNodes": -276,
"jsHeapTotalBytes": 5496832,
"scriptDurationMs": 104.895,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-transition-enter",
"durationMs": 1400.1120000000356,
"styleRecalcs": 17,
"styleRecalcDurationMs": 29.245,
"layouts": 5,
"layoutDurationMs": 13.190999999999997,
"taskDurationMs": 943.584,
"heapDeltaBytes": -6304928,
"heapUsedBytes": 81469832,
"domNodes": 13667,
"jsHeapTotalBytes": 11010048,
"scriptDurationMs": 33.21300000000001,
"eventListeners": 2383,
"totalBlockingTimeMs": 143,
"frameDurationMs": 16.66333333333338,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "viewport-pan-sweep",
"durationMs": 8305.268000000013,
"styleRecalcs": 248,
"styleRecalcDurationMs": 35.433,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 5096.621000000001,
"heapDeltaBytes": 488080,
"heapUsedBytes": 80955296,
"domNodes": -248,
"jsHeapTotalBytes": -602112,
"scriptDurationMs": 2736.6800000000003,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "viewport-pan-sweep",
"durationMs": 8410.037999999986,
"styleRecalcs": 248,
"styleRecalcDurationMs": 34.120999999999995,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 5787.2390000000005,
"heapDeltaBytes": -574360,
"heapUsedBytes": 80577564,
"domNodes": -248,
"jsHeapTotalBytes": -270336,
"scriptDurationMs": 2783.762,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-idle",
"durationMs": 12616.888999999957,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 11942.79,
"heapDeltaBytes": -46695388,
"heapUsedBytes": 174488124,
"domNodes": -8061,
"jsHeapTotalBytes": -10133504,
"scriptDurationMs": 214.628,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.223333333333237,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-idle",
"durationMs": 12633.374000000003,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 11950.682999999997,
"heapDeltaBytes": -25911284,
"heapUsedBytes": 194713168,
"domNodes": -8061,
"jsHeapTotalBytes": -7012352,
"scriptDurationMs": 204.864,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.219999999999953,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-pan",
"durationMs": 20533.341000000008,
"styleRecalcs": 159,
"styleRecalcDurationMs": 17.30100000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 20229.368000000002,
"heapDeltaBytes": -35593440,
"heapUsedBytes": 180594656,
"domNodes": -8065,
"jsHeapTotalBytes": -18767872,
"scriptDurationMs": 879.871,
"eventListeners": -14928,
"totalBlockingTimeMs": 150,
"frameDurationMs": 18.333333333333332,
"p95FrameDurationMs": 16.80000000000291
},
{
"name": "vue-large-graph-pan",
"durationMs": 20160.38400000002,
"styleRecalcs": 160,
"styleRecalcDurationMs": 15.470999999999984,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 19833.607999999997,
"heapDeltaBytes": -67805280,
"heapUsedBytes": 168005436,
"domNodes": -8061,
"jsHeapTotalBytes": -16982016,
"scriptDurationMs": 887.7069999999999,
"eventListeners": -14928,
"totalBlockingTimeMs": 171,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-node-resize-workload",
"durationMs": 4676.082000000008,
"styleRecalcs": 20,
"styleRecalcDurationMs": 79.65500000000003,
"layouts": 20,
"layoutDurationMs": 23.906999999999996,
"taskDurationMs": 4500.238,
"heapDeltaBytes": -27394648,
"heapUsedBytes": 186439888,
"domNodes": -8061,
"jsHeapTotalBytes": 21037056,
"scriptDurationMs": 145.005,
"eventListeners": -14928,
"totalBlockingTimeMs": 64,
"frameDurationMs": 17.223333333333358,
"p95FrameDurationMs": 16.80000000000109
},
{
"name": "vue-node-resize-workload",
"durationMs": 6041.602000000012,
"styleRecalcs": 20,
"styleRecalcDurationMs": 76.69599999999998,
"layouts": 20,
"layoutDurationMs": 22.233000000000004,
"taskDurationMs": 5926.8189999999995,
"heapDeltaBytes": -49552952,
"heapUsedBytes": 181862332,
"domNodes": -8065,
"jsHeapTotalBytes": 18280448,
"scriptDurationMs": 107.522,
"eventListeners": -14928,
"totalBlockingTimeMs": 382,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-zoom-out-idle",
"durationMs": 8518.296999999962,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8319.921999999999,
"heapDeltaBytes": -1234352,
"heapUsedBytes": 188464904,
"domNodes": 0,
"jsHeapTotalBytes": -23330816,
"scriptDurationMs": 155.19299999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-zoom-out-idle",
"durationMs": 8451.41000000001,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8309.565999999999,
"heapDeltaBytes": 9203832,
"heapUsedBytes": 192748976,
"domNodes": 0,
"jsHeapTotalBytes": -26476544,
"scriptDurationMs": 127.05700000000003,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.780000000000047,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "workflow-execution",
"durationMs": 450.95200000002933,
"styleRecalcs": 14,
"styleRecalcDurationMs": 20.270000000000003,
"layouts": 2,
"layoutDurationMs": 0.581,
"taskDurationMs": 106.14400000000002,
"heapDeltaBytes": 4977212,
"heapUsedBytes": 66472364,
"domNodes": 122,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 10.186,
"eventListeners": 81,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "workflow-execution",
"durationMs": 454.95600000003833,
"styleRecalcs": 18,
"styleRecalcDurationMs": 24.659,
"layouts": 4,
"layoutDurationMs": 1.8719999999999999,
"taskDurationMs": 116.07499999999999,
"heapDeltaBytes": 4992248,
"heapUsedBytes": 66833944,
"domNodes": 144,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 9.548000000000002,
"eventListeners": 81,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
}
]
}🔬 E2E Coverage
Top 10 files by uncovered lines
🎨 Storybook: ✅ Built — View Storybook🌾 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. No new findings in the changed files. Reproduce locally# the base the gate compared against
npx fallow audit --root . --quiet --format json \
--changed-since f1e094f20781c4b0b6760defd3e6df4fba13dc67
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 |
0695b7b to
9a25111
Compare
Root-causes two compounding defects behind the agent CRDT canvas occasionally showing a disconnected LoadImage-shaped node at [10, 10]: - LGraphCanvas._deserializeItems / LGraph.configure call graph.add(node) (which synchronously mints a createNode layout op from node._pos) before node.configure(info) sets the real position, so a bulk add (paste, insert-workflow) snapshots LGraphNode's constructor default [10, 10] into the op instead of the real position. - opSender.abortIfUnbound() settles an in-flight, already-transmitted batch 'undeliverable' purely from a workflow-id mismatch on doc retarget, without checking whether the server actually committed it. Adds two Vitest it.fails() unit reproductions (each confirmed to fail for the documented reason, then marked as a known-bug repro) and one Playwright test.fail() end-to-end reproduction driving a real canvas paste + tab retarget through the agent-conversation-replay harness. The e2e spec could not be executed in this environment (no ComfyUI backend reachable), matching PR #18062's disclosed limitation for the same harness.
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## main #18062 +/- ##
==========================================
+ Coverage 83.15% 84.24% +1.09%
==========================================
Files 2972 2274 -698
Lines 146279 129235 -17044
Branches 39639 34939 -4700
==========================================
- Hits 121631 108880 -12751
+ Misses 23372 20059 -3313
+ Partials 1276 296 -980
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 730 files with indirect coverage changes 🚀 New features to boost your workflow:
|
Playwright videos for new testsDownload the recorded videos and HTML report Recorded projects: chromium cloud Recorded test titles:
|
|
Three things this PR looked at beyond the pointerdown fix. Item 2 is fixed here; item 1 is deferred until the recorder can produce its operations; item 3 stays open. 1. CRDT id collision (deferred; its case and the hand-authored
2. Stale
3. Paste gate for non-editable selections (open). When text is selected after the last click into the graph, Ctrl+C correctly copies the text, but Generated by Claude Code |
fb1049d to
6a49838
Compare
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: changes requested
I found two blockers:
- The three callback-form
test.fail(title, callback)declarations treat failures anywhere in fixture setup, replay, readiness, or lookup as expected. I reproduced the suite failing during boot while all three expected-failure cases were still reported as passing, so these tests can hide regressions unrelated to the deferred assertions. - The nominal passing copy/paste regression fails under the documented
pnpm dev:cloudworkflow: the Agent node is selected, but copy writes an empty node list and paste leaves the graph at 8 nodes instead of 9. The production-bundle CI arm passes, which makes this a development/production build-mode discrepancy rather than proof that the supported development path works.
I also left focused comments on a cross-boundary selection case, the hand-authored CRDT replay operations, missing primary-button coverage, and test maintainability issues.
Checks I ran at 6a49838a043ec932e97bc1259a405c00e78b4489 against merge base 3c5f47df3977b97428e9ec5a3ce315678ad98a5a:
- Focused Vitest: 2 files, 10 tests passed.
- Frontend and browser typechecks passed.
- Changed-file ESLint and formatting checks passed.
- Cloud production build passed.
- Fallow structural review completed and identified the new boundary suppression.
- Focused browser test under the documented cloud dev server failed with 8 nodes instead of 9; the broader local browser attempts also demonstrated that callback-form expected failures mask boot failures.
- Current CI reports 87 successful and 17 skipped check runs, with no failures.
| */ | ||
| export function collapseTextSelectionOutside(container: Element): void { | ||
| const selection = selectedText() | ||
| if (selection?.anchorNode && !container.contains(selection.anchorNode)) |
There was a problem hiding this comment.
P2 — Check the whole selection, not only its anchor. A range can be anchored in container while its focus extends into the Agent transcript. In that case this condition preserves the cross-boundary text (insideoutside in a focused DOM reproduction), so shouldIgnoreCopyPaste continues routing clipboard events away from the graph after the canvas click. Please collapse unless every range is wholly contained by the clicked element.
There was a problem hiding this comment.
Now collapses unless every range has both its start and end container inside the clicked element (rangesOf(selection).some(...) in collapseTextSelectionOutside), with a unit case for a range anchored inside the container that ends outside it. 3fa51be
Generated by Claude Code
| nodeId: number, | ||
| pos: [number, number] | ||
| ): RecordedGraphOperation { | ||
| return { |
There was a problem hiding this comment.
P2 — Please do not hand-author replay operations. The browser replay guide explicitly says graph_ops must come from the recorder because these payloads encode CRDT contracts that are easy to approximate incorrectly. Since this only serves the deferred ID-collision expected failure, please remove that case for now and restore it with recorder-produced operations when the scenario is ready to be covered.
There was a problem hiding this comment.
Removed the id-collision case, copyPasteOps.ts, its Vitest check and the fixture's applyGraphOps(); the PR body and the plan comment now list that scenario as deferred until recorder-produced operations exist. 3fa51be
Generated by Claude Code
| @@ -0,0 +1,109 @@ | |||
| import { beforeEach, describe, expect, it } from 'vitest' | |||
|
|
|||
| // eslint-disable-next-line import-x/no-restricted-paths | |||
There was a problem hiding this comment.
P3 — Avoid adding another architecture-boundary suppression just to retest unchanged behavior. The new linearMode case is not part of this change, and importing the renderer store from workbench solely for that assertion creates a second suppressed dependency. Please drop this change-detector case, or exercise the behavior through an existing owner-level boundary without a restricted import.
There was a problem hiding this comment.
Dropped the linearMode case and the restricted import; the remaining cases call shouldIgnoreCopyPaste() against the testing Pinia the Vitest setup installs, with no mock and no disable directive. 3fa51be
Generated by Claude Code
| }, | ||
| { id: nodeId, at: REVEAL_AT } | ||
| ) | ||
| const header = this.vueNodes |
There was a problem hiding this comment.
P3 — Reuse VueNodeHelpers.selectNode. After the pan and visibility poll, this repeats its node-header click recipe. Keeping the Agent-specific reveal/readiness here but delegating the actual selection to this.vueNodes.selectNode(nodeId) prevents the two interaction paths from drifting.
There was a problem hiding this comment.
Kept the pan and the visibility poll and delegated the click to this.vueNodes.selectNode(nodeId). 3fa51be
Generated by Claude Code
| .toHaveLength(before.length + 1) | ||
| }) | ||
|
|
||
| test.fail( |
There was a problem hiding this comment.
P1 — This expected-failure form masks failures before the intended assertion. I reproduced fixture/boot failures where all three callback-form test.fail cases still counted as passes. Move each deferred scenario into an ordinary test(...), complete setup and readiness first, then call test.fail() immediately before the specific assertion that documents the known defect. That preserves the boundary while allowing unrelated setup, replay, locator, and clipboard regressions to fail the suite.
There was a problem hiding this comment.
Converted both deferred cases to ordinary test(...): replay, node selection, clipboard keystrokes and readiness run as normal steps, and test.fail() is called immediately before the one assertion that documents the defect (the +1 node poll for the transcript-selection case, the pasted-type check for the stale-fallback case). The id-collision case is removed together with its hand-authored ops (see the copyPasteOps.ts thread). 3fa51be
Generated by Claude Code
| cleanupErrorHooks = null | ||
| }) | ||
| function collapseOutsideSelectionOnClick(e: PointerEvent) { | ||
| if (e.button === 0 && e.target instanceof Element) |
There was a problem hiding this comment.
P2 — Add coverage for the primary-button guard. The helper unit tests call the helper directly and every browser path uses a left click, so removing e.button === 0 and clearing selection on middle/right pointerdown would leave the new suite green. Please dispatch representative primary and non-primary pointerdowns through this listener and assert that only the primary click collapses the external selection.
There was a problem hiding this comment.
Extracted the listener body to collapseOutsideSelectionOnPrimaryPointerDown(event) in eventHelpers.ts; GraphCanvas.vue registers that same function for the canvas and the TransformPane pointerdown, and a table-driven unit test dispatches real PointerEvents for buttons 0, 1 and 2 through a listener bound to it (only 0 collapses). Mounting GraphCanvas.vue in happy-dom means booting the whole app, so the test binds the exported function rather than going through the component. 3fa51be
Generated by Claude Code
| const source = await agentConversation.nodeOfType(AGENT_NODE_TYPE) | ||
|
|
||
| await agentConversation.selectNode(source.id) | ||
| await page.keyboard.press('Control+c') |
There was a problem hiding this comment.
P3 — Route clipboard shortcuts through the existing helper. All 18 copy/paste keystrokes in this file use raw page.keyboard.press, bypassing ClipboardHelper.copy/paste, its platform-aware KeyboardHelper.ctrlSend, and its synchronization. Please expose/reuse that helper in this fixture so these interaction-heavy tests follow the repository's established clipboard path.
There was a problem hiding this comment.
The harness now exposes clipboard (a ClipboardHelper over a KeyboardHelper bound to #graph-canvas), and every copy and paste in the spec goes through clipboard.copy() / clipboard.paste(), passing the canvas locator where the test focuses the canvas first. 3fa51be
Generated by Claude Code
| await agentConversation.replayTurn(0) | ||
| }) | ||
|
|
||
| test('copying an agent-added node and pasting duplicates it', async ({ |
There was a problem hiding this comment.
P3 — Split these workflows into user-named test.step phases. Each of the seven tests performs several distinct actions (replay/select, copy, focus/selection change, paste, verify), but none marks those phases. Adding steps will make failures and traces identify which user action broke instead of reporting only the long scenario title.
There was a problem hiding this comment.
Each test now runs its phases as named test.steps (select, copy, selection or focus change, paste, verify). 3fa51be
Generated by Claude Code
| // transcript or focus in the composer used to swallow or misroute the graph | ||
| // clipboard. The recording adds three unwired core nodes to the seed graph; | ||
| // the reported types (a Seedance API node, SaveVideo) are not in its catalog, | ||
| // and none of the paths under test read the type, so the agent's |
There was a problem hiding this comment.
P2 — This substitution rationale is factually wrong. The tests repeatedly locate nodes by type and then assert the copied or colliding type, so the paths do read this value. Please describe the narrower reason these available core types are valid stand-ins, or use a fixture that preserves the reported types.
There was a problem hiding this comment.
Rewrote it to the narrower reason: the mechanisms under test (selection collapse, clipboard gating) do not depend on the node type, and the recording holds core types only. 3fa51be
Generated by Claude Code
|
|
||
| await expect | ||
| .poll(() => agentConversation.graphNodes()) | ||
| .toHaveLength(before.length + 1) |
There was a problem hiding this comment.
P1 — This claimed passing regression fails in the documented cloud development workflow. With pnpm dev:cloud and this focused test, the graph stays at 8 nodes instead of reaching 9. Focused inspection showed the Agent node present in canvas.selectedItems, but copy stored {"nodes":[],"groups":[],"reroutes":[],"links":[],"subgraphs":[]} in litegrapheditor_clipboard. The production-bundle CI arm passes, so I have not assumed an exact root cause, but the supported development path needs to produce the same copy/paste result before this test can establish the regression.
There was a problem hiding this comment.
I could not run pnpm dev:cloud from this environment (testcloud.comfy.org and the Playwright browser download are both blocked here), so I have not reproduced it and am not claiming a fix. What the symptom pins down: _serializeItems (LGraphCanvas.ts, the item.clone()?.serialize() branch) drops a selected node when LGraphNode.clone() returns null, and clone() returns null exactly when LiteGraph.createNode(this.type) finds the type unregistered. The follower's materializer then falls back to a missingNode placeholder that still carries type (agentNodeMaterializer.ts materialize / missingNode), so graphNodes() and nodeOfType() find it, the header click puts it in selectedItems, and copy writes {"nodes":[]}: a page whose /api/object_info registered no EmptyLatentImage.
The fixture opts into the served catalog (objectInfo: 'server'). Under pnpm dev:cloud that request is proxied to testcloud.comfy.org with the fixture's mocked Firebase user, which that backend cannot honour, while CI's cloud arm and the documented replay command (DISTRIBUTION=cloud DEV_SERVER_COMFYUI_URL=http://127.0.0.1:8188 pnpm dev, browser_tests/README.md "Replay coverage for agent bug fixes") serve it from a local ComfyUI. Change in 3fa51be: boot() now fails with a named error when the served catalog lacks a type the recording uses, so this shows up at boot instead of as 8 vs 9 nodes. The only import.meta.env gates in this area (agentPanel flag forcing, crdtDebugGate logging) do not touch selection or the clipboard. If the empty copy still happens with a local ComfyUI behind the dev server, the /api/object_info status from that run is what I would need next.
Generated by Claude Code
6a49838 to
3fa51be
Compare
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: non-blocking follow-up remains
The updated head addresses all ten findings from my blocking review. In particular, the expected failures no longer mask setup errors, cross-boundary selections are covered by a red/green regression, hand-authored graph operations are gone, clipboard actions use the shared helper, and missing node definitions now fail during fixture boot instead of producing an empty copy payload.
I left four inline findings. The material gap is that the browser coverage boots only Vue Nodes, so it does not prove the separate legacy-canvas pointerdown registration. The remaining items are Playwright conventions: add the required @vue-nodes tag, derive the live-node projection from the application type, and move free-standing scenario constants out of the spec.
FE-2504 does not supersede this PR. This change touches none of its store-first apply or reconciliation targets. The Agent fixture drives the existing projection path and asserts live clipboard behavior, so that coverage remains useful when semantic remote apply replaces the current projection internals. The conversation REST mocks are separate from graph projection reconciliation.
Checks at 3fa51bea979ba8e775ac514a532eb1dcd0456656 against merge base 82c12d82a5421f1256193b97e45d438507365e75:
- Focused Vitest: 12 tests passed.
- Frontend and browser typechecks passed.
- Changed-file ESLint, type-aware Oxlint, and formatting checks passed.
- Knip passed.
- Playwright discovery found all six cloud tests.
- The cross-boundary selection test failed on the pre-fix tree with
insideoutsideand passed on this head. - Fallow's only scoped decision was the changed clipboard contract; I traced both consumers,
useCopyandusePaste. - Playwright execution was left to CI as required for this follow-up round; the cloud suite and new-test video job passed. The final check snapshot had 84 successes, 26 skips, and one unrelated job still running.
Prior findings checked and cleared
- Every selection range now checks both endpoints.
- The hand-authored CRDT operation fixture and collision case were removed.
- The added restricted renderer import and unchanged linear-mode test were removed.
- Agent node selection delegates to
VueNodeHelpers.selectNode. - Deferred cases run setup normally and call
test.fail()only before the expected assertion. - Real pointer events cover primary, middle, and secondary buttons.
- Clipboard shortcuts use
ClipboardHelper. - Multi-action tests use named
test.stepphases. - The node-type substitution rationale now states mechanism independence accurately.
- The fixture serves recorded node definitions, validates the catalog, and waits for projected node readiness before clipboard actions.
|
|
||
| test.describe( | ||
| 'Copy and paste beside the agent panel', | ||
| { tag: ['@cloud', '@agent'] }, |
There was a problem hiding this comment.
P2 — Add the required @vue-nodes tag. This suite selects .lg-node-header elements and the fixture forces the Vue renderer, but the browser-test guide says every Vue-node test must carry this load-bearing tag and must not own renderer setup manually. The custom Agent fixture may need an adapter for the tag-owned setup, but its special boot path does not replace the tag's routing and discoverability metadata.
There was a problem hiding this comment.
Added @vue-nodes to the suite, and the conversation fixture now reads the tag the way comfyPageFixture does: Comfy.VueNodes.Enabled comes from testInfo.tags, and an untagged spec fails by name before boot instead of the fixture forcing the renderer. The other specs on this fixture (agentConversationReplay, agentTabSwitchCatchUp, agentTabSwitchRunSubmission) carry the tag now too; agentGraphActivity already did. 438fe8e.
Generated by Claude Code
| } | ||
|
|
||
| /** One live graph node as the canvas holds it. */ | ||
| interface LiveGraphNode { |
There was a problem hiding this comment.
P3 — Derive this projection from the authoritative graph-node type. Browser-test guidance forbids copied node/store shapes because they can drift. Please base type on LGraphNode while representing the deliberate string normalization of id, rather than declaring both fields independently. The docblock also restates the interface; the same redundant prose can be removed from graphNodes, nodeOfType, and nodesAddedSince.
There was a problem hiding this comment.
LiveGraphNode is now Pick<LGraphNode, 'type'> & { id: string }, the id half being the un-branded string the harness compares by. The docblock on it and the comments on graphNodes, nodeOfType, and nodesAddedSince are removed. 438fe8e.
Generated by Claude Code
| // Selection collapse and clipboard gating never depend on the node type, and | ||
| // the recording holds core types only, so EmptyLatentImage stands in for the | ||
| // reported Seedance node and the seed's SaveImage for SaveVideo. | ||
| const CASE = 'agent-rec-three-sequential-adds' |
There was a problem hiding this comment.
P3 — Keep scenario data out of the spec's top level. The Playwright structure guide requires spec files to contain imports, test blocks, and Playwright hooks/configuration rather than free-standing constants. Move these conversation IDs and surrogate node types into the fixture/data owner, or keep narrowly used values inside the scenario arrangement.
There was a problem hiding this comment.
Moved them to browser_tests/fixtures/data/agent/agentCopyPasteScenario.ts (COPY_PASTE_SCENARIO: conversation id, the agent-added surrogate type, and the earlier node's id and type). The spec now holds imports, test.use/setTimeout, hooks, and tests only. 438fe8e.
Generated by Claude Code
7de9c05 to
cd4026d
Compare
|
Rebased onto main to clear a conflict with #18070 in Generated by Claude Code |
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: one non-blocking Playwright convention remains
The follow-up addresses the prior renderer-ownership and classic-canvas coverage gaps, and the rebase onto the current base preserves the reviewed behavior. The @vue-nodes tag now owns replay rendering, the live-node projection derives from LGraphNode, the scenario data has a fixture owner, and a classic-renderer test exercises the separate canvas pointerdown binding. I left one inline P3 about the new spec’s location.
The implementation behavior looks sound: primary graph pointerdown collapses selections that extend outside the clicked graph element, contenteditable keeps native clipboard ownership, and the deferred stale-selection and stale-clipboard cases remain explicit expected failures. FE-2504 remains orthogonal; this patch adds durable behavioral coverage without adding the store-first reconciliation machinery that issue proposes removing.
Checks on the live head: focused Vitest passed 12/12; frontend and browser typechecks passed; changed-file ESLint, type-aware Oxlint, formatting, Knip, and git diff --check passed; Playwright discovery found all seven targeted copy/paste tests. I used CI for browser execution as required: all Chromium shards, cloud Playwright, and the new-test video job succeeded. The head check set has 84 successes, 20 skips, and one cancelled generic test job.
| @@ -0,0 +1,28 @@ | |||
| import { | |||
There was a problem hiding this comment.
P3 — Place this new spec in a feature folder. browser_tests/README.md requires new specs to live in a feature subfolder rather than at the root of browser_tests/tests/. Please move this regression into the matching canvas/clipboard feature folder so new coverage follows the current layout contract.
There was a problem hiding this comment.
Moved to browser_tests/tests/canvas/copyPasteAfterTextSelection.spec.ts in 3405499 (imports are @e2e/* aliases, so nothing else changed; Playwright lists it at the new path).
Generated by Claude Code
|
Holding the requested main merge for now. christian-byrne added this PR to the merge queue at 00:13 UTC (see the Slack thread), and a new push would remove it from the queue and dismiss all three approvals under the branch ruleset. The eight commits main gained since the recorded base were compared against this branch: the only overlap is the selectionStore refactor in #17048, whose hunks in Generated by Claude Code |
|
|
|
Manual backports opened after the automated run hit cherry-pick conflicts: #18336 (cloud/1.54) and #18337 (core/1.54). Adaptations: dropped the agentConversationFixture/agent replay spec hunks and the AgentCopyPasteDriver + agentCopyPaste.spec.ts e2e (the recorded-conversation replay fixture does not exist on the release branches); the canvas regression spec and unit tests carry the coverage. Generated by Claude Code |
…#18336) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1789772243794029?thread_ts=1789772243.794029&cid=C0BGH348Z0C)_ ## Summary Source: #18062, merged as `ec5da77d0ff9a45d637bcfe128772ac8557e8a06`. Target: `cloud/1.54` at `19a4f18b799f228a0faf39d951c22a4fdb53daf8`. **Before:** With the agent panel open, copying a node after text in the transcript had been selected pasted an older node or nothing: the canvas cancels the default pointerdown, so the transcript selection survived the click on the node and `shouldIgnoreCopyPaste()` swallowed both the Ctrl+C and the Ctrl+V, leaving the next paste to fall back to the stale `localStorage` node clipboard. Pasting into the composer (a contenteditable editor) could likewise drop a stale node onto the canvas. **After:** Clicking into the graph collapses a text selection made outside the clicked element, so the copy and paste that follow reach the graph. Copy respects text selection: with non-editable text selected, Ctrl+C copies the text and canvas copy/paste stays out of the way, and a later text copy clears the canvas clipboard slot when it still holds the node copy it superseded. Ownership is tracked by a per-copy id (`localStorage['litegrapheditor_clipboard_id']`) stamped by every `LGraphCanvas.copyToClipboard()`, so `useCopy` only clears the slot while its own Ctrl+C copy is still current. Contenteditable editors count as text inputs, and selections inside a textarea or text/search input are detected through `selectionStart`/`selectionEnd`. ## Changes - **What**: - `src/workbench/eventHelpers.ts`: `collapseTextSelectionOutside()`, `collapseOutsideSelectionOnPrimaryPointerDown()`, `hasTextSelection()` (textarea/input selections, whitespace-only selections), `shouldIgnoreCopyPaste()` treats contenteditable as a text input. - `src/components/graph/GraphCanvas.vue`: collapses an outside text selection on primary pointerdown into the graph. - `src/composables/useCopy.ts`: remembers the id of its own Ctrl+C node copy and clears the canvas clipboard slot when a later text copy supersedes it. - `src/lib/litegraph/src/LGraphCanvas.ts`: `copyToClipboard()` stamps a fresh `litegrapheditor_clipboard_id`. - `browser_tests/fixtures/components/Topbar.ts`: `getWorkflowTabLabel()` helper. - Tests: `src/composables/useCopy.test.ts`, `src/workbench/eventHelpers.test.ts` (new), `src/lib/litegraph/src/LGraphCanvas.clipboard.test.ts`, `browser_tests/tests/canvas/copyPasteAfterTextSelection.spec.ts` (new canvas regression spec). ## Backport adaptations - Dropped the hunks touching `browser_tests/fixtures/agentConversationFixture.ts` and the agent replay specs (`agentConversationReplay`, `agentHumanAddTabSwitch`, `agentHumanDeleteTabSwitch`, `agentLayoutQuality`, `agentNewChatClearedCanvas`, `agentPastedNodeOrphanedOnRetarget`, `agentTabSwitchCatchUp`, `agentTabSwitchRunSubmission`, `cmpDocumentAcceptance`): none of these files exist on `cloud/1.54` (modify/delete conflicts), so they stay absent. - Dropped `browser_tests/fixtures/AgentCopyPasteDriver.ts` and `browser_tests/tests/agent/agentCopyPaste.spec.ts`: they build on `agentConversationTest` from the recorded-conversation replay fixture, which has no equivalent on this branch (no `agentConversationFixture`, no recorded conversation data). The canvas regression spec `copyPasteAfterTextSelection.spec.ts` and the unit tests carry the coverage here. - `src/composables/useCopy.test.ts`: content conflict resolved by taking the #18062 version wholesale; its imports (`fromPartial`, `useCanvasStore`, `eventHelpers`) all resolve on the branch. - `src/lib/litegraph/src/LGraphCanvas.clipboard.test.ts`: the auto-merge dropped the source commit's import hunk, so `createTestNode` (`@/lib/litegraph/src/__fixtures__/nodeHelpers`) and `createTestRootGraph` (`@/lib/litegraph/src/subgraph/__fixtures__/subgraphHelpers`) were re-added; both helpers already exist on the branch. ## Review Focus Straight cherry-pick of the source files; only e2e files tied to the absent agent replay harness were dropped. Unit tests and the canvas e2e spec remain. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01M9ZKkgcvdGt7LHjkvJUq63 Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
…18337) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1789772243794029?thread_ts=1789772243.794029&cid=C0BGH348Z0C)_ ## Summary Source: #18062, merged as `ec5da77d0ff9a45d637bcfe128772ac8557e8a06`. Target: `core/1.54` at `6d2daca629b4e21155ae8270e0fce503da4ee482`. **Before:** With the agent panel open, copying a node after text in the transcript had been selected pasted an older node or nothing: the canvas cancels the default pointerdown, so the transcript selection survived the click on the node and `shouldIgnoreCopyPaste()` swallowed both the Ctrl+C and the Ctrl+V, leaving the next paste to fall back to the stale `localStorage` node clipboard. Pasting into the composer (a contenteditable editor) could likewise drop a stale node onto the canvas. **After:** Clicking into the graph collapses a text selection made outside the clicked element, so the copy and paste that follow reach the graph. Copy respects text selection: with non-editable text selected, Ctrl+C copies the text and canvas copy/paste stays out of the way, and a later text copy clears the canvas clipboard slot when it still holds the node copy it superseded. Ownership is tracked by a per-copy id (`localStorage['litegrapheditor_clipboard_id']`) stamped by every `LGraphCanvas.copyToClipboard()`, so `useCopy` only clears the slot while its own Ctrl+C copy is still current. Contenteditable editors count as text inputs, and selections inside a textarea or text/search input are detected through `selectionStart`/`selectionEnd`. ## Changes - **What**: - `src/workbench/eventHelpers.ts`: `collapseTextSelectionOutside()`, `collapseOutsideSelectionOnPrimaryPointerDown()`, `hasTextSelection()` (textarea/input selections, whitespace-only selections), `shouldIgnoreCopyPaste()` treats contenteditable as a text input. - `src/components/graph/GraphCanvas.vue`: collapses an outside text selection on primary pointerdown into the graph. - `src/composables/useCopy.ts`: remembers the id of its own Ctrl+C node copy and clears the canvas clipboard slot when a later text copy supersedes it. - `src/lib/litegraph/src/LGraphCanvas.ts`: `copyToClipboard()` stamps a fresh `litegrapheditor_clipboard_id`. - `browser_tests/fixtures/components/Topbar.ts`: `getWorkflowTabLabel()` helper. - Tests: `src/composables/useCopy.test.ts`, `src/workbench/eventHelpers.test.ts` (new), `src/lib/litegraph/src/LGraphCanvas.clipboard.test.ts`, `browser_tests/tests/canvas/copyPasteAfterTextSelection.spec.ts` (new canvas regression spec). ## Backport adaptations - Dropped the hunks touching `browser_tests/fixtures/agentConversationFixture.ts` and the agent replay specs (`agentConversationReplay`, `agentHumanAddTabSwitch`, `agentHumanDeleteTabSwitch`, `agentLayoutQuality`, `agentNewChatClearedCanvas`, `agentPastedNodeOrphanedOnRetarget`, `agentTabSwitchCatchUp`, `agentTabSwitchRunSubmission`, `cmpDocumentAcceptance`): none of these files exist on `core/1.54` (modify/delete conflicts), so they stay absent. - Dropped `browser_tests/fixtures/AgentCopyPasteDriver.ts` and `browser_tests/tests/agent/agentCopyPaste.spec.ts`: they build on `agentConversationTest` from the recorded-conversation replay fixture, which has no equivalent on this branch (no `agentConversationFixture`, no recorded conversation data). The canvas regression spec `copyPasteAfterTextSelection.spec.ts` and the unit tests carry the coverage here. - `src/composables/useCopy.test.ts`: content conflict resolved by taking the #18062 version wholesale; its imports (`fromPartial`, `useCanvasStore`, `eventHelpers`) all resolve on the branch. - `src/lib/litegraph/src/LGraphCanvas.clipboard.test.ts`: the auto-merge dropped the source commit's import hunk, so `createTestNode` (`@/lib/litegraph/src/__fixtures__/nodeHelpers`) and `createTestRootGraph` (`@/lib/litegraph/src/subgraph/__fixtures__/subgraphHelpers`) were re-added; both helpers already exist on the branch. ## Review Focus Straight cherry-pick of the source files; only e2e files tied to the absent agent replay harness were dropped. Unit tests and the canvas e2e spec remain. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01M9ZKkgcvdGt7LHjkvJUq63 Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
|
|
|
|
|
|
The guard's tests dispatched on document with shouldIgnoreCopyPaste mocked out, so they only showed that a cancelled event is skipped. Run the real shouldIgnoreCopyPaste instead: a paste into a contenteditable is left alone by target (#18062), and a paste claimed from inside an uneditable chip, whose target is not editable, is left alone only by the guard. The guard's comment now gives that reason.
…fy-Org#18248) *PR Created by the Glary-Bot Agent* --- Ctrl+V an image into the agent chat input and it attaches to the in-progress message, the same way a dropped file does. ## Root cause `InlinePromptEditor`'s `handlePaste` claimed every paste unconditionally and inserted `clipboardData.getData('text/plain')`. For an image that string is `''`, so it dispatched an empty `insertText` (a no-op) and returned `true` — telling ProseMirror the paste was handled. The image was dropped on the floor. That is the reported "nothing happens at all". ## Change - `attachableClipboardFiles(clipboard)` filters `clipboard.files` through the existing `isAgentAttachable`, so paste accepts exactly the file types drag-and-drop does. - `handlePaste` emits `attachFiles` when the clipboard carries attachable files; `AgentPanelRoot` routes them into `attachment.addFiles`, the same upload path as drag-and-drop. - The emit is forwarded `InlinePromptEditor` → `Composer` → `AgentPanel` → `AgentPanelRoot`. - A paste behaves like a drop in the panel: it keeps node selection mode active (only the + menu's attach, which opens the OS picker, ends it), and a paste whose upload lands records `trackAgentAttachButtonClicked({ method: 'paste' })`, gated on `addFiles` succeeding the same way `drag_drop` is (Comfy-Org#18665). Two ordering constraints worth knowing when reading the diff: - **Content is inserted before attaching.** Staging an attachment rewrites the prompt through the store, while the editor's `watch(model)` is `flush: 'post'`. Emitting first lets the editor write its pre-attachment document back over the staged asset. There is a test that fails if these two lines are swapped. - **A files-only clipboard skips the insert entirely**, because `insertText('')` would delete whatever the user had selected. This holds whether or not the files are attachable, so pasting only a `.zip` over selected text leaves the text alone. A clipboard carrying both text and an attachable file keeps both: the text is inserted and the file attaches. ## `usePaste` defers to a paste the editor already claimed `usePaste` returns early when the paste was already cancelled (`e.defaultPrevented`). Since Comfy-Org#18062, `shouldIgnoreCopyPaste` treats contenteditable targets as text inputs, so an ordinary composer paste is ignored by target. The guard covers the case the target check cannot: a collapsed caret inside an uneditable reference chip (`contenteditable=false`, e.g. after pressing on a chip and moving a few pixels before release) makes the chip's label the paste target. ProseMirror still handles and cancels that paste, but `shouldIgnoreCopyPaste` sees a non-editable target and would hand it to the canvas — an image would also become a LoadImage node, and workflow JSON would replace the graph via `loadGraphData`. The `usePaste` tests now run the real `shouldIgnoreCopyPaste` (`vi.mock(..., { spy: true })`) and dispatch from realistic targets: pastes into a contenteditable editor or paragraph leave the graph alone by target, and a claimed paste from inside a chip is left alone only by the guard (those tests fail with the guard removed). Canvas-focused paste is untouched. ## Verification Exercised against a real Chromium clipboard (`navigator.clipboard.write` + a real Ctrl+V keypress), with the ComfyUI backend serving uploads: | Case | Result | |---|---| | Screenshot only | attaches, `POST /api/upload/image` → 200, canvas unchanged | | Image + text (spreadsheet-shaped clipboard) | attaches **and** keeps the text | | Plain text | inserts as before, nothing attaches | | Image pasted on the canvas | still creates a `LoadImage` node | The feature was re-checked by hand on a local dev server against staging after the review follow-ups. Unit coverage: the agent, paste, and telemetry suites pass (180 files, 3,008 tests); `pnpm typecheck`, `pnpm lint`, `pnpm format:check`, `pnpm knip` clean. Each new regression test was confirmed to fail without its fix — including the node-selection, paste-telemetry (both the success and rejected-upload cases), unattachable-file selection, and chip-guard tests. ## Known gaps - Only Chromium was exercised; Safari and Firefox clipboard shapes are unverified. Chromium hands a pasted screenshot over as `image.png`; an engine that hands over an unnamed blob would not attach. - Ctrl+Shift+V (paste as plain text) of an image over selected text deletes the selection, because the browser strips the file and the empty-text insert runs. Pre-existing on `main`; not changed here. - Copying and pasting an asset or node reference chip pastes its readable text rather than a chip (only workflow chips round-trip since Comfy-Org#17443). Out of scope for this PR. - Fixes PM-1438 - Fixes PM-528 ## Screenshots    --------- Co-authored-by: Glary-Bot <glary-bot@users.noreply.github.com> Co-authored-by: ryan-di <ryan.weihao.di@gmail.com>
…#20147) ## Summary Pin the reported symptom of [PM-1261](https://linear.app/comfyorg/issue/PM-1261/fix-pasting-text-into-the-agent-box-also-pastes-the-last-copied-node) — pasting text into the agent composer also pastes the last copied canvas node — which had no black-box case even though the behaviour is fixed. ## Changes - **What**: two cases in `browser_tests/tests/agent/agentCopyPaste.spec.ts`. - `pasting text into the composer leaves the graph untouched` — copy a node on the canvas, put plain text on the clipboard, paste into the composer: the text lands in the composer and no node is added to the graph. - `a composer paste keeps the node clipboard usable on the canvas` — after that composer paste, an ordinary canvas Ctrl+V still pastes the copied node. This is the half that stops the first case from being satisfiable by simply disabling the node clipboard. - No production code changes. ## Review Focus **Why these were missing.** #18062 fixed this family and added four cases, but all four exercise the *copy* side or a *canvas* paste: a node clicked after a transcript selection, Ctrl+C in the composer preserving the node clipboard, a transcript text copy clearing it, and copy-id ownership. None pastes into the composer, so the exact user-visible symptom was unpinned. **Why the precondition is real.** The text goes on the clipboard with `ClipboardHelper.writeText`, which is a programmatic write, not a `copy` event — so `useCopy` never runs and `localStorage['litegrapheditor_clipboard']` still holds the Ctrl+C node copy. That is the state the reports describe ("the *last copied* node"). With `text/html` carrying no node payload, `pasteClipboardItems` declines it, the workflow parse fails, and the only remaining route to the graph is the `canvas.pasteFromClipboard()` fallback in `usePaste` — the fallback the report is about. **Red-on-parent evidence.** The spec cannot run on #18062's parent because `AgentCopyPasteDriver.ts` was added by that PR, so the equivalent is removing the guards that cover this path today. There are now **two**, which is worth knowing: 1. `isContentEditable` in `shouldIgnoreCopyPaste` (#18062), and 2. the `if (e.defaultPrevented) return` early return in `usePaste` (added later by #18248, 2026-09-28) — ProseMirror cancels the paste it handles, so this covers the composer independently. Removing only (1) leaves both new cases **green**, because (2) still holds. Removing both makes both cases fail with a `SaveImage` node landing on the graph — the reported symptom exactly. So the cases bind the *behaviour* rather than either mechanism, and would catch the regression if either guard were removed in isolation only once both went. **Measured locally** against `DISTRIBUTION=cloud` at head `314cfe9d62`, which contains current `main`: all 6 cases in the spec pass (19.2s, `--project=cloud --workers=1`). Both new cases fail with both guards removed — now at the strengthened whole-graph assertion rather than the single-frame diff. ## Review round: the four Cursor findings are addressed All four were about the same risk, and it is the one that matters most for a PR whose only job is to be a pin: each case could stay green while binding nothing. They are fixed in `ce3c7a7d5e`, and every change is a strengthening — no assertion was weakened and no case was retired. | Finding | Fix | | --- | --- | | Nothing proved the node clipboard held a node, so the negative assertion could pass vacuously | Both cases now assert `localStorage['litegrapheditor_clipboard']` holds exactly the copied node before asserting anything about the graph | | A single `nextFrame` samples once, but `usePaste`'s handler is `async` | `expectGraphUnchanged` re-asserts across a 10-frame settle window | | `nodesAddedSince` is a one-way diff, blind to a removal | The whole node set is compared, sorted by id | | `before` spans both pastes, so the canvas-paste addition is not attributable | The canvas-paste assertion re-baselines on a post-composer-paste snapshot, taken only after the composer paste is proved to have added nothing | **Each finding was verified real rather than accepted.** For all four, the failure mode it names was injected and both halves were checked: the original assertion stayed green under it — so the hole was genuine — and the new guard rejects. Clearing the clipboard key after the copy left the old negative assertion passing; a node inserted on a 120 ms timer slipped past the single-frame sample; a removal was invisible to the one-way diff; and a leaked node plus a cleared clipboard satisfied both of the original canvas-paste assertions. The proof spec was a throwaway and is not committed. The change stays inside `browser_tests/tests/agent/` deliberately: the two helpers are local to the spec rather than added to `AgentCopyPasteDriver`, because that file sits outside the Agent path globs and moving the PR outside them would change which team's approval the PR requires. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Requested by Christian Byrne · Slack thread
Before: With the agent panel open, copying a node and pasting it sometimes pasted a different node (the one copied earlier), sometimes nothing, and sometimes the right node. The canvas cancels the default pointerdown, which is what normally collapses a text selection, so a selection left in the agent transcript survived a click on a node and
shouldIgnoreCopyPaste()then swallowed both the Ctrl+C and the Ctrl+V; the next paste fell back to the stalelocalStoragenode clipboard. The composer (a ProseMirror contenteditable) was not treated as a text input either, so pasting text into it also dropped the last copied node onto the canvas, and Ctrl+C in it overwrote the node clipboard.After: Clicking into the graph collapses a text selection made outside the clicked element (the browser default the canvas was suppressing), so the copy and paste that follow reach the graph. Contenteditable editors count as text inputs like textareas, so the composer's copy and paste are no longer treated as canvas copy/paste. With non-editable text selected, the text wins: Ctrl+C copies the text and canvas copy/paste stays out of the way. A text copy also clears the canvas clipboard slot (
localStorage['litegrapheditor_clipboard']) when the slot still holds the Ctrl+C node copy that the text copy superseded, so a canvas Ctrl+V after copying text pastes nothing instead of replaying that node. Ownership is tracked per copy act, not by payload: everyLGraphCanvas.copyToClipboard()stamps a fresh id intolocalStorage['litegrapheditor_clipboard_id'],useCopyremembers the id of its own Ctrl+C copy, and it clears the slot only while that id is still current. A copy made afterwards through any other path (context menu,Comfy.Canvas.CopySelected, duplicate, another tab), including a copy of the same unmoved selection with byte-identical output, replaces the id and survives a later text copy. Text selected inside a textarea or text/search input counts as a text copy through the control'sselectionStart/selectionEnd(the documentSelectionnever reports it), and whitespace-only document selections count too.Summary
Fix copy/paste beside the agent panel by collapsing an outside text selection on click into the graph and treating contenteditable as a text input, with the repros promoted to regression tests.
Changes
src/workbench/eventHelpers.ts:collapseTextSelectionOutside(container)collapses unless every range lies wholly inside the container;collapseOutsideSelectionOnPrimaryPointerDown(event)applies it to the clicked element on a primary-button pointerdown;shouldIgnoreCopyPaste()countsHTMLElement.isContentEditableas a text input (linearModeunchanged) and any non-collapsed document selection, whitespace included;hasTextSelection(target)reports the text a native copy puts on the clipboard, fromselectionStart/selectionEndon a textarea or input and from the documentSelectionotherwise.input[type=search]now counts as a text input for the copy/paste gate (it was wrongly listed with the non-text input types; flagged in the Cursor review round, no in-tree surface renders one today, pinned by thea search inputunit case).src/components/graph/GraphCanvas.vueregisters the pointerdown helper on the canvas element (the classic renderer has noTransformPane) and on the Vue node pane. Table-driven unit tests insrc/workbench/eventHelpers.test.ts(real happy-domSelectionandPointerEvent, no spies).src/lib/litegraph/src/LGraphCanvas.ts:copyToClipboard()writeslocalStorage['litegrapheditor_clipboard_id'](acreateUuidv4(), which also works on plain-HTTP origins wherecrypto.randomUUIDis undefined) next to the payload on every call; pinned inLGraphCanvas.clipboard.test.ts.src/composables/useCopy.tskeeps the id of its last Ctrl+C copy and, on a copy the canvas ignores that carries text (hasTextSelection(e.target)), removes the slot and the id only while that id is still current.useCopy.test.tscovers the keyboard, command, keyboard-then-command (different and equal payload), collapsed-caret, textarea and search-input interleavings.browser_tests/tests/agent/agentCopyPaste.spec.ts(cloud project,agentCopyPasteTest, keystrokes throughClipboardHelper): pastes a node clicked after transcript text was selected; Ctrl+C in the empty composer preserves the node clipboard; copying transcript text clears the keyboard node clipboard (the text is on the clipboard, a canvas paste adds nothing); a text copy preserves a newer equal command copy (Ctrl+C, thenComfy.Canvas.CopySelectedon the same unmoved node, then a text copy, then a canvas Ctrl+V pastes that node). Deferred, not in this PR: an agentadd_nodereusing a just-pasted id replaces the pasted node's type (CRDT id collision); it returns once the recorder can produce those operations, sincegraph_opsare never hand-authored.composer/transcriptlocators,graphNodes(),nodeOfType(),nodesAddedSince(),revealAndSelectNode(),replayTurn()) lives inbrowser_tests/fixtures/AgentCopyPasteDriver.ts, composed aroundagentConversationTestas theagentCopyPastefixture, together with the scenario values (conversation id, stand-in node types).agentConversationFixture.tskeeps only what every replay needs:boot()fails with a named error when the served/api/object_infolacks a type the recording uses, and the@vue-nodestag owns the replay renderer (the fixture reads it the waycomfyPageFixturedoes and fails by name when a spec omits it).browser_tests/tests/canvas/copyPasteAfterTextSelection.spec.ts(chromium project, classic renderer, standardcomfyPagefixture): selects the topbar workflow tab label, clicks a node title on the classic canvas, copies and pastes throughClipboardHelper, and asserts one more node on the graph.Topbar.getWorkflowTabLabel()exposes the label.EmptyLatentImage/SaveImagestand in.Review Focus
localStorage['litegrapheditor_clipboard']fallback inusePastestay as they are (text wins, decided in the Slack thread); context-menu, selection-toolbox and command-palette copies write only that slot and rely on the fallback for Ctrl+V. The stale replay after a text copy is fixed on the copy side instead: the text copy clears the slot only whilelitegrapheditor_clipboard_idstill names the Ctrl+C copy it supersedes, so every later slot-only copy keeps its fallback. The CRDT id-collision replace remains deferred.Screenshots (if applicable)
Videos of each case are in the Slack thread.
🤖 Generated with Claude Code
https://claude.ai/code/session_01M9ZKkgcvdGt7LHjkvJUq63