Repository navigation
test(subgraph): pin FE-2497 nested definition loss end to end - #18159
Conversation
The existing regression test only catches the convert-to-subgraph definition teardown through its telemetry side effect. Removing its reportError assertion leaves it passing on pre-fix code, so it does not pin the reported symptom: the wrapped subgraph loses every internal link, its interior nodes lose their layout records, and the subgraph renders as an empty canvas. Assert that symptom directly instead, at both levels: - Vitest: capture interior nodes, links, layout records, the serialized round trip and reported errors, then hold them across four selection shapes and a three-deep nest. - Playwright: wrap the host in assets/subgraphs/nested-subgraph, then check the definition inventory live, after a save/reload, and by navigating into the deepest interior and asserting it still renders. All six Vitest cases and all three browser tests go red on pre-fix code (links 7 -> 0 and 6 -> 0, zero nodes rendered) and green with the fix.
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Repository: Comfy-Org/ComfyUI_frontend/.coderabbit.yaml Review profile: ASSERTIVE Plan: Team Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Comment |
🎭 Playwright: ✅ 2040 passed, 0 failed · 2 flaky📊 Browser Reports
🎨 Storybook: ✅ Built — View Storybook📦 Bundle Size
⚡ Performance Report
Absolute values
Raw data{
"timestamp": "2026-09-20T10:02:01.189Z",
"gitSha": "c0bf319fd3f17e3623d10d921b3a045e1b08df57",
"branch": "glary/fe-2497-wrap-nested-subgraph-tests",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 2018.1249999999977,
"styleRecalcs": 6,
"styleRecalcDurationMs": 6.419999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 546.025,
"heapDeltaBytes": 3166728,
"heapUsedBytes": 64281008,
"domNodes": -266,
"jsHeapTotalBytes": 4448256,
"scriptDurationMs": 12.788999999999998,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-idle",
"durationMs": 2016.86200000006,
"styleRecalcs": 8,
"styleRecalcDurationMs": 8.725,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 576.1729999999999,
"heapDeltaBytes": -2919424,
"heapUsedBytes": 59121896,
"domNodes": -265,
"jsHeapTotalBytes": 4448256,
"scriptDurationMs": 14.4,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-mouse-sweep",
"durationMs": 2162.575000000004,
"styleRecalcs": 81,
"styleRecalcDurationMs": 45.513,
"layouts": 14,
"layoutDurationMs": 5.452999999999999,
"taskDurationMs": 1184.194,
"heapDeltaBytes": 15986924,
"heapUsedBytes": 76790672,
"domNodes": -265,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 216.21099999999998,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1783.9870000000246,
"styleRecalcs": 75,
"styleRecalcDurationMs": 26.919000000000004,
"layouts": 14,
"layoutDurationMs": 4.2989999999999995,
"taskDurationMs": 831.3799999999999,
"heapDeltaBytes": 14616256,
"heapUsedBytes": 76623500,
"domNodes": -261,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 163.47699999999998,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1737.4740000000202,
"styleRecalcs": 32,
"styleRecalcDurationMs": 17.688000000000002,
"layouts": 6,
"layoutDurationMs": 1.4169999999999998,
"taskDurationMs": 334.107,
"heapDeltaBytes": 2132712,
"heapUsedBytes": 64093096,
"domNodes": 76,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 11.816,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1739.479000000074,
"styleRecalcs": 32,
"styleRecalcDurationMs": 18.39,
"layouts": 6,
"layoutDurationMs": 1.333,
"taskDurationMs": 387.35300000000007,
"heapDeltaBytes": 2094544,
"heapUsedBytes": 64116316,
"domNodes": 75,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 14.427999999999999,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-clipping",
"durationMs": 843.2899999999677,
"styleRecalcs": 11,
"styleRecalcDurationMs": 8.639,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 559.2250000000001,
"heapDeltaBytes": 17860044,
"heapUsedBytes": 79841256,
"domNodes": 18,
"jsHeapTotalBytes": 5767168,
"scriptDurationMs": 197.94500000000002,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 642.3970000000736,
"styleRecalcs": 10,
"styleRecalcDurationMs": 7.0360000000000005,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 438.046,
"heapDeltaBytes": 17202792,
"heapUsedBytes": 77590012,
"domNodes": 16,
"jsHeapTotalBytes": 4718592,
"scriptDurationMs": 166.796,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-idle",
"durationMs": 2055.1379999999995,
"styleRecalcs": 3,
"styleRecalcDurationMs": 4.151999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 771.068,
"heapDeltaBytes": 2792432,
"heapUsedBytes": 82308592,
"domNodes": -254,
"jsHeapTotalBytes": -1318912,
"scriptDurationMs": 45.21499999999999,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-idle",
"durationMs": 2022.4830000000793,
"styleRecalcs": 6,
"styleRecalcDurationMs": 5.293000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 559.8629999999999,
"heapDeltaBytes": 3911544,
"heapUsedBytes": 82539176,
"domNodes": -246,
"jsHeapTotalBytes": -794624,
"scriptDurationMs": 34.972,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-pan",
"durationMs": 3064.134000000024,
"styleRecalcs": 66,
"styleRecalcDurationMs": 13.879000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1963.0990000000002,
"heapDeltaBytes": -19612980,
"heapUsedBytes": 66901684,
"domNodes": -228,
"jsHeapTotalBytes": 1601536,
"scriptDurationMs": 1000.283,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.80000000000109
},
{
"name": "large-graph-pan",
"durationMs": 2261.203000000023,
"styleRecalcs": 67,
"styleRecalcDurationMs": 13.361,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1635.116,
"heapDeltaBytes": 15180468,
"heapUsedBytes": 79089396,
"domNodes": -247,
"jsHeapTotalBytes": 1273856,
"scriptDurationMs": 847.0849999999999,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-zoom",
"durationMs": 3669.027999999969,
"styleRecalcs": 60,
"styleRecalcDurationMs": 10.578000000000001,
"layouts": 60,
"layoutDurationMs": 14.054,
"taskDurationMs": 1762.554,
"heapDeltaBytes": 8117116,
"heapUsedBytes": 73971480,
"domNodes": 0,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 816.8330000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-zoom",
"durationMs": 3547.415000000001,
"styleRecalcs": 61,
"styleRecalcDurationMs": 11.105999999999998,
"layouts": 60,
"layoutDurationMs": 13.069,
"taskDurationMs": 1719.518,
"heapDeltaBytes": 8500848,
"heapUsedBytes": 74345044,
"domNodes": 2,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 813.7260000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3220.166000000006,
"styleRecalcs": 41,
"styleRecalcDurationMs": 6.236999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 2122.8610000000003,
"heapDeltaBytes": 25933376,
"heapUsedBytes": 91440276,
"domNodes": 2,
"jsHeapTotalBytes": 679936,
"scriptDurationMs": 1094.274,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3030.283000000054,
"styleRecalcs": 41,
"styleRecalcDurationMs": 5.573999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1886.8609999999999,
"heapDeltaBytes": 15236292,
"heapUsedBytes": 80635776,
"domNodes": 2,
"jsHeapTotalBytes": 1466368,
"scriptDurationMs": 982.12,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2043.6980000000062,
"styleRecalcs": 6,
"styleRecalcDurationMs": 6.116,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 614.154,
"heapDeltaBytes": -4943292,
"heapUsedBytes": 82094988,
"domNodes": -252,
"jsHeapTotalBytes": -495616,
"scriptDurationMs": 36.709,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2047.7619999999206,
"styleRecalcs": 6,
"styleRecalcDurationMs": 5.582,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 642.9549999999999,
"heapDeltaBytes": -5374092,
"heapUsedBytes": 82029440,
"domNodes": -252,
"jsHeapTotalBytes": -757760,
"scriptDurationMs": 37.784,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 595.815000000016,
"styleRecalcs": 47,
"styleRecalcDurationMs": 11.596,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 425.70399999999995,
"heapDeltaBytes": 13876952,
"heapUsedBytes": 76827456,
"domNodes": 20,
"jsHeapTotalBytes": 5767168,
"scriptDurationMs": 151.976,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 595.2619999999342,
"styleRecalcs": 47,
"styleRecalcDurationMs": 12.471000000000004,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 431.44399999999996,
"heapDeltaBytes": 13827408,
"heapUsedBytes": 77070352,
"domNodes": 20,
"jsHeapTotalBytes": 5505024,
"scriptDurationMs": 151.83200000000002,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2013.7720000000172,
"styleRecalcs": 9,
"styleRecalcDurationMs": 8.289000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 579.666,
"heapDeltaBytes": 11888540,
"heapUsedBytes": 74975884,
"domNodes": -278,
"jsHeapTotalBytes": 4448256,
"scriptDurationMs": 12.855999999999996,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2005.2759999999807,
"styleRecalcs": 10,
"styleRecalcDurationMs": 8.473,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 443.2269999999999,
"heapDeltaBytes": -7863112,
"heapUsedBytes": 55388832,
"domNodes": -274,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 7.352000000000001,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1724.536999999998,
"styleRecalcs": 77,
"styleRecalcDurationMs": 30.883,
"layouts": 18,
"layoutDurationMs": 6.300000000000001,
"taskDurationMs": 884.4699999999999,
"heapDeltaBytes": 1175536,
"heapUsedBytes": 64318124,
"domNodes": -277,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 125.35,
"eventListeners": -220,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1709.38799999999,
"styleRecalcs": 77,
"styleRecalcDurationMs": 28.373999999999995,
"layouts": 18,
"layoutDurationMs": 5.53,
"taskDurationMs": 800.821,
"heapDeltaBytes": 1025588,
"heapUsedBytes": 64391336,
"domNodes": -274,
"jsHeapTotalBytes": 6283264,
"scriptDurationMs": 112.553,
"eventListeners": -220,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-transition-enter",
"durationMs": 1387.008999999921,
"styleRecalcs": 17,
"styleRecalcDurationMs": 29.07,
"layouts": 5,
"layoutDurationMs": 12.848000000000003,
"taskDurationMs": 932.5220000000003,
"heapDeltaBytes": -6421372,
"heapUsedBytes": 80139172,
"domNodes": 13587,
"jsHeapTotalBytes": 10223616,
"scriptDurationMs": 31.216,
"eventListeners": 2383,
"totalBlockingTimeMs": 139,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "viewport-pan-sweep",
"durationMs": 8309.091000000024,
"styleRecalcs": 247,
"styleRecalcDurationMs": 35.178000000000004,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 5804.104,
"heapDeltaBytes": -12506944,
"heapUsedBytes": 66845688,
"domNodes": -248,
"jsHeapTotalBytes": 4186112,
"scriptDurationMs": 2798.7329999999997,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "viewport-pan-sweep",
"durationMs": 8539.839999999913,
"styleRecalcs": 248,
"styleRecalcDurationMs": 36.388000000000005,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 5987.159000000001,
"heapDeltaBytes": -4040584,
"heapUsedBytes": 75347828,
"domNodes": -247,
"jsHeapTotalBytes": -339968,
"scriptDurationMs": 2874.8010000000004,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-idle",
"durationMs": 17464.743,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 17033.768000000004,
"heapDeltaBytes": -54098180,
"heapUsedBytes": 167900140,
"domNodes": -8061,
"jsHeapTotalBytes": -14356480,
"scriptDurationMs": 277.097,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-idle",
"durationMs": 17005.414000000088,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 16501.32,
"heapDeltaBytes": -31123596,
"heapUsedBytes": 190380620,
"domNodes": -8061,
"jsHeapTotalBytes": -13000704,
"scriptDurationMs": 267.177,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.219999999999953,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-pan",
"durationMs": 22870.81599999999,
"styleRecalcs": 174,
"styleRecalcDurationMs": 24.71399999999996,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 22727.03,
"heapDeltaBytes": -66800932,
"heapUsedBytes": 166890048,
"domNodes": -8061,
"jsHeapTotalBytes": -17047552,
"scriptDurationMs": 948.7829999999999,
"eventListeners": -14928,
"totalBlockingTimeMs": 99,
"frameDurationMs": 17.779999999999806,
"p95FrameDurationMs": 16.799999999995634
},
{
"name": "vue-large-graph-pan",
"durationMs": 20892.100000000028,
"styleRecalcs": 170,
"styleRecalcDurationMs": 17.769000000000005,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 20508.682,
"heapDeltaBytes": -35845520,
"heapUsedBytes": 179671884,
"domNodes": -8061,
"jsHeapTotalBytes": -15847424,
"scriptDurationMs": 918.4929999999999,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-node-resize-workload",
"durationMs": 6695.886000000087,
"styleRecalcs": 20,
"styleRecalcDurationMs": 91.63700000000003,
"layouts": 20,
"layoutDurationMs": 31.29799999999999,
"taskDurationMs": 6675.445999999999,
"heapDeltaBytes": -58948072,
"heapUsedBytes": 171154064,
"domNodes": -8061,
"jsHeapTotalBytes": 14086144,
"scriptDurationMs": 130.179,
"eventListeners": -14928,
"totalBlockingTimeMs": 588,
"frameDurationMs": 17.776666666666763,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-node-resize-workload",
"durationMs": 4730.498000000011,
"styleRecalcs": 20,
"styleRecalcDurationMs": 80.17099999999996,
"layouts": 20,
"layoutDurationMs": 22.864999999999995,
"taskDurationMs": 4615.459000000001,
"heapDeltaBytes": -47282896,
"heapUsedBytes": 168115284,
"domNodes": -8061,
"jsHeapTotalBytes": 6049792,
"scriptDurationMs": 120.80300000000001,
"eventListeners": -14928,
"totalBlockingTimeMs": 84,
"frameDurationMs": 17.77333333333336,
"p95FrameDurationMs": 33.29999999999927
},
{
"name": "vue-zoom-out-idle",
"durationMs": 9063.351000000011,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8904.633,
"heapDeltaBytes": 15365104,
"heapUsedBytes": 197228176,
"domNodes": 0,
"jsHeapTotalBytes": -10223616,
"scriptDurationMs": 149.32299999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666763,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-zoom-out-idle",
"durationMs": 8447.080000000027,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8230.805999999999,
"heapDeltaBytes": 10575556,
"heapUsedBytes": 182986220,
"domNodes": 0,
"jsHeapTotalBytes": -8126464,
"scriptDurationMs": 121.946,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "workflow-execution",
"durationMs": 134.6769999998969,
"styleRecalcs": 8,
"styleRecalcDurationMs": 21.845,
"layouts": 5,
"layoutDurationMs": 3.737,
"taskDurationMs": 111.66799999999999,
"heapDeltaBytes": 3275872,
"heapUsedBytes": 64723608,
"domNodes": 131,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 14.571000000000003,
"eventListeners": 17,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "workflow-execution",
"durationMs": 451.4910000000327,
"styleRecalcs": 12,
"styleRecalcDurationMs": 19.371,
"layouts": 3,
"layoutDurationMs": 0.8039999999999998,
"taskDurationMs": 107.16399999999999,
"heapDeltaBytes": 4871420,
"heapUsedBytes": 66297848,
"domNodes": 119,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 10.227000000000002,
"eventListeners": 81,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
}
]
}🔬 E2E Coverage
Top 10 files by uncovered lines
|
|
@coderabbitai review No code change needed for this one — the skip is a base-branch configuration effect, not a finding. This PR deliberately targets Triggering a single review explicitly above instead. Once #18040 merges, this PR's base falls back to |
Playwright videos for new testsDownload the recorded videos and HTML report Recorded projects: chromium Recorded test titles:
|
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## fix/subgraph-release-serialization-mismatch #18159 +/- ##
============================================================================
Coverage 83.78% 83.79%
============================================================================
Files 2242 2242
Lines 127636 127636
Branches 32469 32469
============================================================================
+ Hits 106943 106948 +5
+ Misses 20403 20398 -5
Partials 290 290
Flags with carried forward coverage won't be shown. Click here to find out more. 🚀 New features to boost your workflow:
|
Re-verified: before/after on the current branch headRe-ran both suites against
"BEFORE" = reverting only the two product lines from BEFORE — Vitest (6 failed)BEFORE — Playwright (3 failed, reported symptom verbatim)AFTER — both greenWider regression check: Why nothing to push
A commit here would have been an empty or no-op change, so I did not fabricate one. |
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: follow-up requested
I found no product correctness blocker in this test-only increment. The focused tests do go red against the reverted base fix and green at this head. I found two material test or repository-boundary gaps and two minor Playwright structure issues inline.
Pre-merge blockers: none.
Findings without an anchorable line
- [P2] The verification section calls all three
import-x/no-restricted-pathswarnings pre-existing. The base already has the tworeportErrorwarnings, but this increment adds thelayoutStorewarning. Please distinguish the new warning from the inherited two. - [P3] The visual-proof and notes sections claim pixel-for-pixel equivalence plus delete/undo and copy/paste probe results, but the linked material does not include the reporter's source image or those probe logs. Please link the evidence or narrow those claims to what the checked-in tests establish.
Existing findings confirmed or contested
No prior review findings existed on this head.
Unverified candidate
- [P2] I could not independently reproduce the claimed three red Playwright cases with the base fix reverted, or the two full-suite failures on the effective base. The local targeted command discovered all three new tests but stopped in fixture setup because this orb has no ComfyUI backend on port 8188. Head CI is green. A backend-enabled targeted revert run plus the two named tests on
12e6a724would settle the remaining claim.
Checked and cleared
- I independently derived the wrapper's four links from the fixture and conversion behavior. Existing promotion tests own
filename_prefixsemantics, so I did not ask this nested-definition regression to duplicate them. - The new
SubgraphHelpermethods are additive. Existing construction and methods are unchanged. - Reversing only the
12e6a724product patch makes all six focused Vitest cases fail. At this exact head, the focused file passes 6/6 and the full subgraph unit directory passes 361/361. - The live and serialized inventories retain duplicate definition names as separate rows and derive expectations from literal fixture topology rather than production logic.
Coverage
Standard-depth correctness, maintainability, idiomaticity, tests, and documentation lanes reviewed all three changed files against the effective stacked base. The tests review inventoried 535 fixture methods, traced 24 interactions, applied the Playwright rulebook, and reviewed the Vitest mock and fixture boundaries. Typecheck and browser typecheck passed. Changed-file ESLint had zero errors and three warnings. Formatting passed. Diff-scoped Fallow found no new complexity or duplication. The only local execution limitation was the absent ComfyUI backend for Playwright.
|
|
||
| test.use({ initialSettings: { 'Comfy.UseNewMenu': 'Disabled' } }) | ||
|
|
||
| const WRAPPED_SELECTION = ['subgraph 2', 'VAE Decode', 'Save Image'] |
There was a problem hiding this comment.
[P3] The browser-test guide says specs should keep free-standing constants out of module scope. These values are local to this scenario, so please move them into the describe or the tests that use them. That keeps the expected data beside its assertion instead of adding another spec-level API.
There was a problem hiding this comment.
Done in 991c418 — WRAPPED_SELECTION, NESTED_DEFINITIONS_INTACT and NESTED_DEFINITIONS_AFTER_WRAP now live inside the test.describe callback, next to the assertions that consume them. Module scope holds only imports and test.use.
Worth noting this lines up with a separate point from the earlier review pass: hoisting WRAPPED_SELECTION away from the assertions is arguably what let an incorrect expected link count survive the first draft, since the selection is the input that determines the expected counts. Keeping them together makes that dependency visible.
| 'still renders the nested interior after wrapping', | ||
| { tag: ['@vue-nodes'] }, | ||
| async ({ comfyPage }) => { | ||
| await comfyPage.nodeOps.selectNodes(WRAPPED_SELECTION) |
There was a problem hiding this comment.
[P3] This case has three distinct phases: wrap, descend, and assert the deepest render. Only descent is named in the trace. Please put the wrapping and final rendered outcome in semantic test.step blocks too, so a failure identifies which phase broke without reconstructing it from line numbers.
There was a problem hiding this comment.
Done in 991c418. All three phases of that case are named steps now:
Wrap the subgraph host together with its siblings
Descend through the wrapper into the deepest nested subgraph
Assert the deepest interior still renders its three nodes
I applied the same "Wrap the subgraph host together with its siblings" step to the other two cases as well, so the wrap phase is labelled consistently across the file rather than only where it happened to be split out.
It already pays off — in the pre-fix run the trace attributes the failure to the descend phase rather than to a bare line number:
3) ... still renders the nested interior after wrapping
› Descend through the wrapper into the deepest nested subgraph
| } from '@/lib/litegraph/src/litegraph' | ||
| import { createTestNode } from '@/lib/litegraph/src/__fixtures__/nodeHelpers' | ||
| import { reportError } from '@/platform/telemetry/reportError' | ||
| import { layoutStore } from '@/renderer/core/layout/store/layoutStore' |
There was a problem hiding this comment.
[P2] This adds a new src/lib to src/renderer dependency, and the repository's import-x/no-restricted-paths rule reports it. Please keep the Litegraph test inside the leaf boundary, or move the layout-integrated assertion to the renderer/service-layer test that owns layoutStore. The two existing reportError warnings are inherited; this renderer warning is incremental.
There was a problem hiding this comment.
Fixed in 991c418 — the layoutStore import is gone and the litegraph test is back inside the leaf boundary.
I removed the interiorNodesWithLayout field from DefinitionHealth and the layoutStore.resetForTests() call that field required, rather than relocating the assertion. The remaining fields still take every case red on pre-fix code, so nothing was lost:
BEFORE (fix reverted): Tests 6 failed (6) <- via links, serialisedLinks, reportedErrors
AFTER (fix present): Tests 6 passed (6)
eslint on the file now reports only the two inherited reportError warnings; the incremental renderer one is gone:
✖ 2 problems (0 errors, 2 warnings) # both @/platform/telemetry/reportError, pre-existing
The rendering claim that field was standing in for is now owned entirely by the Playwright case, which is the layer that actually renders.
There was a problem hiding this comment.
Follow-up: I have now taken the second option you offered here as well, in 90f5b011.
My earlier reply only did the first half — removing the layoutStore import so the litegraph test returns to its leaf boundary — and dropped the assertion rather than relocating it. Your comment explicitly allowed "move the layout-integrated assertion to the renderer/service-layer test that owns layoutStore", and that is the better answer, because it also satisfies 4056573223's geometry request at the same time.
The assertion now lives in src/renderer/core/layout/operations/subgraphConversionLayout.test.ts, next to the existing graphLayoutAttachment.test.ts that owns this area. Renderer importing lib is the permitted direction, so there is no restricted-path violation — eslint reports nothing at all on the new file:
$ npx eslint src/renderer/core/layout/operations/subgraphConversionLayout.test.ts
(no output)
The litegraph test is unchanged from my previous reply and still carries only the two inherited reportError warnings. Net effect: the layer boundary holds, and the layout behaviour is covered at the layer that owns it — with a test that goes red pre-fix, which the version inside the litegraph test never would have proven on its own.
| return { | ||
| nodes: definition.nodes.length, | ||
| links: definition.links.size, | ||
| interiorNodesWithLayout: definition.nodes.filter(({ id }) => |
There was a problem hiding this comment.
[P2] Counting layout entries proves that keys survive, but not that each interior node keeps its geometry. A conversion that swaps the two records or replaces both with default geometry still passes. Please give the nodes distinct, hand-checkable positions or sizes and compare the ID-to-layout mapping before and after wrapping in one representative nested case.
There was a problem hiding this comment.
Good catch on the weakness — key-presence counting was too weak, and it is gone. But I have to push back on the geometry comparison as the replacement: I built it, and it does not bite.
I added getDefinitionNodeGeometry() returning id -> { pos, size } and a case asserting the deepest interior against hand-checkable literals from the asset, before and after wrapping:
const DEEPEST_INTERIOR_GEOMETRY = {
'4': { pos: [26, 474], size: [315, 98] },
'5': { pos: [473, 609], size: [315, 106] },
'7': { pos: [413, 389], size: [425, 181] }
}Run against pre-fix code, that case passed, while the other three failed:
1) ... keeps the wrapped and nested definitions populated FAILED
2) ... keeps every nested interior node at its own position PASSED <-- pins nothing
3) ... keeps the nested definitions populated in the saved wf FAILED
4) ... still renders the nested interior after wrapping FAILED
The reason is the shape of this bug. releaseSubgraphs wipes the link store and the layout-store records, but the Subgraph instance survives with its _nodes array and each node's own pos/size untouched. That matches the reported workflow exactly — in purz's corrupted file all 28 interior nodes kept byte-identical pos/size while links went 77 -> 0. So geometry is simply not a channel this regression travels through, and a geometry assertion here would be a test that never goes red — the exact failure mode this PR exists to correct in the base branch's test.
The geometry that does change is the layout store, which is what you were pointing at. I checked whether an E2E could read it: it is a module singleton, not exposed on window, and no browser test touches it (grep -rn "layoutStore" browser_tests/ returns one unrelated comment). Reaching it would mean adding a test-only global to production, and asserting it from the unit test is precisely what your other comment asks me to stop doing — the two requests point opposite ways, so I resolved in favour of the layer boundary.
What I did instead, to close the "swapped or default content still passes" hole at the layer that owns rendering: the render case now asserts all three interior nodes by their distinct titles, not just a count.
await expect(comfyPage.vueNodes.nodes).toHaveCount(3)
await expect(comfyPage.vueNodes.getNodeByTitle('Load Checkpoint')).toBeVisible()
await expect(comfyPage.vueNodes.getNodeByTitle('Empty Latent Image')).toBeVisible()
await expect(comfyPage.vueNodes.getNodeByTitle('CLIP Text Encode (Prompt)')).toBeVisible()Three distinct identities cannot be satisfied by placeholders, a collapsed set, or one node rendered three times, and it goes red pre-fix because nothing renders at all. The non-biting case and its now-unused helper were removed (pnpm knip clean). Happy to revisit if you would rather expose the layout store to E2E deliberately — that is a bigger call than this PR should make on its own.
There was a problem hiding this comment.
Revisiting this — you were right and my earlier pushback was wrong. Now implemented in 90f5b011.
My mistake was treating this comment and 4056573221 as irreconcilable. 4056573221 spelled out the reconciliation I skipped: "or move the layout-integrated assertion to the renderer/service-layer test that owns layoutStore." I only took the first branch (remove it from the litegraph test) and dropped the assertion instead of relocating it.
The earlier geometry attempt failed because I read the wrong source. node.pos is never perturbed by this bug, so it could not go red. The geometry that does change is the layout store, which detachGraphLayouts clears — exactly the "ID-to-layout mapping" you named. Asserting it belongs at the renderer layer, where importing both litegraph and layoutStore is the legal direction, so the litegraph test stays leaf-clean and both comments are satisfied at once.
New test: src/renderer/core/layout/operations/subgraphConversionLayout.test.ts, alongside the existing graphLayoutAttachment.test.ts that owns this area. It gives the two interior nodes distinct positions, then compares the id-to-layout mapping either side of wrapping their host:
setNodePosition(origin, [110, 220])
setNodePosition(target, [330, 440])
// ... wrap innerHost into a new subgraph ...
expect({ before, after: interiorGeometry(rootGraph, inner) }).toEqual({
before: { 1: { x: 110, y: 220 }, 2: { x: 330, y: 440 } },
after: { 1: { x: 110, y: 220 }, 2: { x: 330, y: 440 } }
})And unlike the node.pos version, it bites:
BEFORE (fix reverted): AFTER (fix present):
- "1": { "x": 110, "y": 220 } Tests 1 passed (1)
- "2": { "x": 330, "y": 440 }
+ "1": null
+ "2": null
Tests 1 failed (1)
Because the positions are distinct and keyed by id, this also fails on the two cases you called out: a swap surfaces as 1 -> {330,440}, and a reset to defaults surfaces as neither matching. eslint reports nothing on the new file — renderer importing lib is the permitted direction. pnpm typecheck clean; src/renderer/core/layout/operations/ 28 passed.
- Restore the src/lib leaf boundary: drop the layoutStore import from the litegraph test, along with the interiorNodesWithLayout field and the resetForTests call it required. The remaining fields (links, serialisedLinks, reportedErrors) still take all six cases red on pre-fix code, so no signal is lost. - Scope the spec's selection and expectation constants into the describe block, beside the assertions that consume them. - Name every phase of the render case with test.step, so a failure identifies the phase. - Assert all three interior nodes by title after descending, so the case cannot pass on placeholder or default content. A geometry comparison was tried for the interior nodes and removed: this bug does not perturb node.pos or node.size. The teardown wipes links and layout-store records while the surviving Subgraph keeps its node objects untouched, which matches the reported workflow where all 28 interior nodes kept byte-identical positions. The geometry test passed on pre-fix code, so it pinned nothing.
|
@DrJKL status summary — all four review points are addressed in Each thread has a detailed reply; the short version:
On the one I pushed back on: I built the requested Current state on Two process notes: I have not resolved these threads, per One open question worth your call, from the geometry thread: the geometry that does change here is the layout store, but it is a module singleton not exposed on |
…ubgraph Reviewer asked for a geometry comparison of the interior nodes, and offered moving a layout assertion to the renderer test that owns layoutStore. My first attempt read node.pos, which this bug never perturbs, so it passed on pre-fix code and pinned nothing. The geometry that does change is the layout store, which detachGraphLayouts clears. Asserting it belongs at the renderer layer, where importing both litegraph and layoutStore is the legal direction, so the litegraph test stays inside its leaf boundary. Gives the two interior nodes distinct positions, then compares the id-to-layout mapping either side of wrapping their host. Pre-fix both records read null; with the fix both keep their own position, so a swap or a reset to defaults fails too.
|
@DrJKL thanks — the two summary findings were both correct and I've acted on them. One heads-up first, then the evidence you asked for. This review ran against the first commit, not the current headThe tell is "three warnings": All four inline findings were already fixed before this review ran, in Summary P2 — warning attribution: you're right, correctedThe description said "3 pre-existing" warnings. That was wrong: two Summary P3 — unevidenced claims: narrowed
I could not apply these to the PR body directly — Unverified candidate — settled with a backendThis sandbox has a real ComfyUI backend, so I ran what you couldn't. A. Current head B. Same head, only the C. The two named specs checked out directly on So Also worth noting for the loop we seem to be in: all four threads are |
3185562
into
fix/subgraph-release-serialization-mismatch
A `test.fixme`/`.skip` landed to unblock a red `main` is a promise to come back, and the PR body is the only place that promise survives. #16907 disabled `agentPanelLifecycle.spec.ts` with a completely empty body, so recovering the intent later meant reconstructing the failure direction from commit timestamps. This fails a pull request that newly disables a test unless its body names a tracking issue or follow-up PR, and prints the file and line so the author does not have to hunt for it. Scoped to avoid crying wolf, since a guard that fires on ordinary work gets routed around: - Only DISABLING declarations, matched as a first argument that is a string literal. Playwright's conditional `test.skip(condition, 'reason')` and bare `test.skip()` are runtime controls, not a test taken out of service. - Only lines this PR ADDS. A disable already on the base branch is not this author's debt to justify. - Only test sources, so a doc or changelog mentioning `test.fixme` is inert. Verified against the case that motivated it: replaying #16907's real base and head with its real (empty) body fails and names `agentPanelLifecycle.spec.ts: test.fixme('supports keyboard activation ...`. The same replay over the last six merged spec-touching PRs (#18213, #18201, #18197, #18183, #18159, #18174) passes, as do synthetic cases for conditional skip, bare skip, docs-only mention, and an unrelated edit to a file that already had a disabled test. shellcheck clean.
PR Created by the Glary-Bot Agent
Targets #18040 (
fix/subgraph-release-serialization-mismatch), notmain. Test-only — no product code touched.Requested in #bug-dump as follow-up to the FE-2497 RCA. #18040 already fixes the root cause; this adds the coverage that pins the user-visible symptom.
Why
#18040's regression test catches the bug only through its telemetry side effect. Deleting its one
expect(reportError).not.toHaveBeenCalled()line leaves the test passing on pre-fixmain, becauseserialiseStoredNodesfalls back to live-node serialization and its remaining assertions only check node types:So the thing FE-2497 was actually reported for — every internal link gone, interior unrenderable — is unpinned. If the mismatch detector is ever moved, throttled, or reworded, the regression walks straight back in.
What this adds
Vitest (
SubgraphConversion.nestedDefinitions.test.ts) — asserts the symptom directly via acaptureHealthsnapshot of interior nodes, links, layout records, the serialized round trip, and reported errors, held across a table of four selection shapes plus a three-deep nest:collectSubgraphsPostOrderreleases the whole subtreePlaywright (
browser_tests/tests/subgraph/subgraphWrapNestedDefinition.spec.ts) — uses the existingassets/subgraphs/nested-subgraphasset, whose shape matches the report exactly (root → subgraph 2 → subgraph 3). Three lenses on one wrap: live definition inventory, the saved workflow afterserializeAndReload(), and navigating into the deepest interior to assert it still renders.Fixture (
SubgraphHelper) —getDefinitionInventory()/getSerializedDefinitionInventory()return one row per definition sorted by name (an array, not a name-keyed object, sinceConvert to Subgraphnames every new definitionNew Subgraph), anddescendSubgraphs(levels)which requires exactly one subgraph node per level before descending.Red / green
Reverting only #18040's two-line product change, with the tests unchanged:
The browser failures are the reported symptom verbatim —
subgraph 2links7 -> 0,subgraph 3links6 -> 0, and zero Vue nodes rendered in the interior:Visual proof
Captured against a live ComfyUI backend + dev server, navigating to the deepest nested subgraph after the wrap. The broken capture reproduces the reporter's screenshot pixel for pixel — empty canvas with only the subgraph IO node stub.
Notes for review
'New Subgraph': links: 4is correct, verified by dumping the live wrapper. The 4th is a boundary link,-10:0 -> 9:1 (STRING): conversion promotesSave Image'sfilename_prefixwidget to a wrapper input. Counting only the asset's three explicit root links gives 3 and is wrong — there is a comment on the constant so it doesn't get "corrected".layoutStore.resetForTests()was added to thebeforeEach. Fixture UUIDs reset per test, so every test reused the same layout-store keys and stale entries could have masked a regression where conversion fails to recreate layout rather than deleting it.middlesubgraph is now built from two connected nodes so it carries a real link — previouslyinteriorNodesWithLayoutwas its only load-bearing field.reportedErrorsfolded intocaptureHealthrestores fix: keep released subgraph definitions consistent across convert-to-subgraph #18040's telemetry guard across all five new cases.test.failmarkers: fix: keep released subgraph definitions consistent across convert-to-subgraph #18040 fixes this class completely. I probed the adjacent paths it deferred — delete+undo and copy/paste of a subgraph host — and both are healthy (delete correctly releases, undo fully restores 7/6 links). Nothing is left failing to mark.Verification
pnpm typecheck/pnpm typecheck:browser: cleaneslint,oxlint,oxfmt --check: 0 errors (3 pre-existingimport-x/no-restricted-pathswarnings;LGraph.tsitself importslayoutStore, as do 25 files undersrc/lib/litegraph/)vitest run src/lib/litegraph/src/subgraph/: 19 files, 361 passedplaywright test browser_tests/tests/subgraph/ --project=chromium: 145 passed, 2 failed —subgraphDraftPositionsandsubgraphSeed, both of which fail identically on this branch with my changes stashed (sandbox reload timeout and a generation-dependent test)deep-reviewersub-agent; all actionable findings applied. Its one blocking finding claimedlinks: 4should be3— disproved empirically above, and it had flagged the browser suite as unrunnable in its own session.Screenshots