Repository navigation
fix: harden workflow load against null initial state and report occurrences - #21326
Conversation
Restore null tolerance where activating a workflow reads state that is missing when the change tracker never loaded: beforeLoadNewGraph, activateLoadedWorkflow (initial state and tracker reset), and ChangeTracker.updateModified/graphEqual. Refuse to load stored content that is not a JSON object and to save an unloaded workflow, so a null state is never written over the stored file. Load an active workflow that has no tracker instead of returning it as loaded. Each guard reports through assert() with a context of booleans and enum strings (no names, paths or content) so the next occurrence shows which path produced it.
|
Important Review skippedBot user detected. To trigger a single review, invoke the
|
📦 Bundle: 11.9 MB gzip 🔴 +1.54 kBDetailsSummary
Category Glance App Entry Points — 6.61 kB (baseline 6.61 kB) • ⚪ 0 BMain entry bundles and manifests
Status: 1 added / 1 removed Graph Workspace — 3.63 MB (baseline 3.63 MB) • ⚪ 0 BGraph editor runtime, canvas, workflow orchestration
Status: 2 added / 2 removed / 2 unchanged Views & Navigation — 148 kB (baseline 148 kB) • ⚪ 0 BTop-level views, pages, and routed surfaces
Status: 8 added / 8 removed / 10 unchanged Panels & Settings — 871 kB (baseline 871 kB) • ⚪ 0 BConfiguration panels, inspectors, and settings screens
Status: 6 added / 6 removed / 29 unchanged User & Accounts — 155 kB (baseline 153 kB) • 🔴 +1.15 kBAuthentication, profile, and account management bundles
Status: 5 added / 5 removed / 11 unchanged Editors & Dialogs — 513 kB (baseline 513 kB) • ⚪ 0 BModals, dialogs, drawers, and in-app editors
Status: 10 added / 10 removed / 7 unchanged UI Components — 337 kB (baseline 337 kB) • ⚪ 0 BReusable component library chunks
Status: 6 added / 6 removed / 25 unchanged Data & Services — 1.85 MB (baseline 1.85 MB) • 🔴 +3.69 kBStores, services, APIs, and repositories
Status: 18 added / 18 removed / 23 unchanged Utilities & Hooks — 1.34 MB (baseline 1.34 MB) • ⚪ 0 BHelpers, composables, and utility bundles
Status: 24 added / 24 removed / 55 unchanged Vendor & Third-Party — 9.7 MB (baseline 9.7 MB) • ⚪ 0 BExternal libraries and shared vendor chunks Status: 15 unchanged Other — 34.6 MB (baseline 34.6 MB) • ⚪ 0 BBundles that do not match a named category
Status: 75 added / 75 removed / 673 unchanged ⚡ Performance Report
Scenario timings
Show regressions
All metrics
Historical variance (last 15 runs)
Trend (last 15 commits on main)
Raw data{
"timestamp": "2026-10-26T15:20:00.605Z",
"gitSha": "3c11b183ae066a481be6cddd72b85f3d8e59688f",
"branch": "fix/workflow-null-initial-state-hardening",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 2039.4159999999886,
"styleRecalcs": 10,
"styleRecalcDurationMs": 3.6240000000000006,
"layouts": 10,
"layoutDurationMs": 2.7709999999999995,
"taskDurationMs": 586.56,
"heapDeltaBytes": 6107668,
"heapUsedBytes": 77583072,
"domNodes": 0,
"jsHeapTotalBytes": 10747904,
"scriptDurationMs": 11.445,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665573770491804,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-idle",
"durationMs": 2021.5000000000032,
"styleRecalcs": 9,
"styleRecalcDurationMs": 2.588,
"layouts": 9,
"layoutDurationMs": 2.546,
"taskDurationMs": 539.2520000000001,
"heapDeltaBytes": 1231392,
"heapUsedBytes": 63529616,
"domNodes": 0,
"jsHeapTotalBytes": 6553600,
"scriptDurationMs": 10.142999999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665289256198346,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1827.0579999999939,
"styleRecalcs": 75,
"styleRecalcDurationMs": 12.485000000000001,
"layouts": 9,
"layoutDurationMs": 2.427,
"taskDurationMs": 951.558,
"heapDeltaBytes": 1900468,
"heapUsedBytes": 63765308,
"domNodes": -161,
"jsHeapTotalBytes": 9961472,
"scriptDurationMs": 154.98700000000002,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665137614678898,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1844.3040000000224,
"styleRecalcs": 79,
"styleRecalcDurationMs": 12.649000000000001,
"layouts": 8,
"layoutDurationMs": 2.225,
"taskDurationMs": 961.2879999999999,
"heapDeltaBytes": -17825580,
"heapUsedBytes": 51514316,
"domNodes": -161,
"jsHeapTotalBytes": 11272192,
"scriptDurationMs": 154.67999999999998,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66576576576577,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1704.4690000000173,
"styleRecalcs": 19,
"styleRecalcDurationMs": 5.002000000000001,
"layouts": 9,
"layoutDurationMs": 2.666,
"taskDurationMs": 455.306,
"heapDeltaBytes": -116488,
"heapUsedBytes": 60510636,
"domNodes": 56,
"jsHeapTotalBytes": 10223616,
"scriptDurationMs": 13.484,
"eventListeners": 43,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1721.5820000000122,
"styleRecalcs": 20,
"styleRecalcDurationMs": 5.131000000000001,
"layouts": 10,
"layoutDurationMs": 2.861,
"taskDurationMs": 449.29600000000005,
"heapDeltaBytes": 34616,
"heapUsedBytes": 61214892,
"domNodes": 56,
"jsHeapTotalBytes": 10223616,
"scriptDurationMs": 13.463999999999997,
"eventListeners": 43,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66504854368932,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 726.2770000000103,
"styleRecalcs": 12,
"styleRecalcDurationMs": 2.934,
"layouts": 8,
"layoutDurationMs": 2.504,
"taskDurationMs": 519.4309999999999,
"heapDeltaBytes": 5980368,
"heapUsedBytes": 67776184,
"domNodes": 0,
"jsHeapTotalBytes": 9961472,
"scriptDurationMs": 168.39900000000003,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665116279069764,
"p95FrameDurationMs": 16.700000000000273
},
{
"name": "dom-widget-clipping",
"durationMs": 745.1449999999795,
"styleRecalcs": 13,
"styleRecalcDurationMs": 2.708000000000001,
"layouts": 8,
"layoutDurationMs": 2.318,
"taskDurationMs": 523.069,
"heapDeltaBytes": 1556704,
"heapUsedBytes": 63095196,
"domNodes": 0,
"jsHeapTotalBytes": 5767168,
"scriptDurationMs": 170.61399999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66363636363636,
"p95FrameDurationMs": 16.799999999999727
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1324.3340000000217,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1053.4219999999998,
"heapDeltaBytes": -13636456,
"heapUsedBytes": 75379016,
"domNodes": -191,
"jsHeapTotalBytes": 24387584,
"scriptDurationMs": 74.44,
"eventListeners": -106,
"totalBlockingTimeMs": 629,
"frameDurationMs": 35.58648648648648,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1245.6020000000194,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 964.032,
"heapDeltaBytes": -14164908,
"heapUsedBytes": 74894348,
"domNodes": -191,
"jsHeapTotalBytes": 21241856,
"scriptDurationMs": 76.58200000000001,
"eventListeners": -106,
"totalBlockingTimeMs": 573,
"frameDurationMs": 33.7837837837838,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 811.2699999999791,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 470.09999999999997,
"heapDeltaBytes": -5196228,
"heapUsedBytes": 68000348,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 138.546,
"eventListeners": 0,
"totalBlockingTimeMs": 190,
"frameDurationMs": 24.017647058823542,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 823.7679999999727,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 468.45600000000064,
"heapDeltaBytes": -5207184,
"heapUsedBytes": 69108992,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 140.43499999999997,
"eventListeners": 0,
"totalBlockingTimeMs": 182,
"frameDurationMs": 24.020588235294113,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "extension-clean-combined-draw",
"durationMs": 9661.158999999998,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8784.665,
"heapDeltaBytes": 56276020,
"heapUsedBytes": 139071132,
"domNodes": -257,
"jsHeapTotalBytes": 61992960,
"scriptDurationMs": 6781.092000000001,
"eventListeners": -68,
"totalBlockingTimeMs": 439,
"frameDurationMs": 30.39716981132075,
"p95FrameDurationMs": 50.100000000000364
},
{
"name": "extension-clean-combined-draw",
"durationMs": 9741.532000000007,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8910.329000000002,
"heapDeltaBytes": -25112576,
"heapUsedBytes": 64902300,
"domNodes": -257,
"jsHeapTotalBytes": 4845568,
"scriptDurationMs": 6942.258,
"eventListeners": -68,
"totalBlockingTimeMs": 438,
"frameDurationMs": 31.600324675324675,
"p95FrameDurationMs": 50.100000000000364
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 9777.29099999999,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8930.966,
"heapDeltaBytes": 82053488,
"heapUsedBytes": 143050012,
"domNodes": 0,
"jsHeapTotalBytes": 77856768,
"scriptDurationMs": 6960.268999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 392,
"frameDurationMs": 30.19413580246914,
"p95FrameDurationMs": 50.100000000000364
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 9752.346999999987,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8987.587000000001,
"heapDeltaBytes": 38382464,
"heapUsedBytes": 132907368,
"domNodes": -257,
"jsHeapTotalBytes": 69070848,
"scriptDurationMs": 6991.857000000001,
"eventListeners": -68,
"totalBlockingTimeMs": 357,
"frameDurationMs": 30.372585669781934,
"p95FrameDurationMs": 50.09999999999991
},
{
"name": "large-combo-open",
"durationMs": 304.13899999999217,
"styleRecalcs": 25,
"styleRecalcDurationMs": 15.617999999999999,
"layouts": 11,
"layoutDurationMs": 4.013000000000001,
"taskDurationMs": 263.966,
"heapDeltaBytes": -17856624,
"heapUsedBytes": 58552808,
"domNodes": 936,
"jsHeapTotalBytes": 1835008,
"scriptDurationMs": 13.14,
"eventListeners": 205,
"totalBlockingTimeMs": 1,
"frameDurationMs": 25,
"p95FrameDurationMs": 99.90000000000009
},
{
"name": "large-combo-open",
"durationMs": 279.4410000000198,
"styleRecalcs": 24,
"styleRecalcDurationMs": 17.932000000000002,
"layouts": 12,
"layoutDurationMs": 3.9109999999999996,
"taskDurationMs": 233.118,
"heapDeltaBytes": -20407276,
"heapUsedBytes": 68029676,
"domNodes": 750,
"jsHeapTotalBytes": 3670016,
"scriptDurationMs": 13.401,
"eventListeners": 64,
"totalBlockingTimeMs": 38,
"frameDurationMs": 28.34000000000001,
"p95FrameDurationMs": 133.30000000000018
},
{
"name": "large-graph-idle",
"durationMs": 2088.8270000000375,
"styleRecalcs": 8,
"styleRecalcDurationMs": 2.583,
"layouts": 8,
"layoutDurationMs": 2.299,
"taskDurationMs": 726.6729999999999,
"heapDeltaBytes": -16744988,
"heapUsedBytes": 74197432,
"domNodes": -257,
"jsHeapTotalBytes": 2523136,
"scriptDurationMs": 42.502,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666400000000003,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-idle",
"durationMs": 2055.802999999969,
"styleRecalcs": 7,
"styleRecalcDurationMs": 2.1140000000000003,
"layouts": 7,
"layoutDurationMs": 2.101,
"taskDurationMs": 667.404,
"heapDeltaBytes": -5185104,
"heapUsedBytes": 76614964,
"domNodes": -257,
"jsHeapTotalBytes": 3796992,
"scriptDurationMs": 35.30199999999999,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665853658536584,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-pan",
"durationMs": 2614.625999999987,
"styleRecalcs": 67,
"styleRecalcDurationMs": 9.152000000000001,
"layouts": 6,
"layoutDurationMs": 1.7429999999999997,
"taskDurationMs": 1812.6779999999999,
"heapDeltaBytes": 3219204,
"heapUsedBytes": 80151132,
"domNodes": -257,
"jsHeapTotalBytes": 4808704,
"scriptDurationMs": 823.519,
"eventListeners": -66,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666025641025637,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-pan",
"durationMs": 2558.24199999995,
"styleRecalcs": 68,
"styleRecalcDurationMs": 9.119,
"layouts": 7,
"layoutDurationMs": 1.886,
"taskDurationMs": 1820.3340000000003,
"heapDeltaBytes": -14357156,
"heapUsedBytes": 61404276,
"domNodes": -257,
"jsHeapTotalBytes": 5107712,
"scriptDurationMs": 866.1,
"eventListeners": -66,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665584415584416,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-zoom",
"durationMs": 3850.5629999999655,
"styleRecalcs": 71,
"styleRecalcDurationMs": 10.82,
"layouts": 64,
"layoutDurationMs": 11.37,
"taskDurationMs": 1924.1999999999998,
"heapDeltaBytes": 8347984,
"heapUsedBytes": 79268328,
"domNodes": -233,
"jsHeapTotalBytes": 1699840,
"scriptDurationMs": 807.9340000000001,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665800865800865,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-zoom",
"durationMs": 3823.311999999987,
"styleRecalcs": 70,
"styleRecalcDurationMs": 10.38,
"layouts": 64,
"layoutDurationMs": 11.111,
"taskDurationMs": 1886.456,
"heapDeltaBytes": 2307656,
"heapUsedBytes": 79330076,
"domNodes": -233,
"jsHeapTotalBytes": 4321280,
"scriptDurationMs": 798.308,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66593886462882,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3246.5349999999944,
"styleRecalcs": 44,
"styleRecalcDurationMs": 5.829,
"layouts": 5,
"layoutDurationMs": 1.4110000000000003,
"taskDurationMs": 2155.234,
"heapDeltaBytes": -9360736,
"heapUsedBytes": 77288968,
"domNodes": -257,
"jsHeapTotalBytes": 6049792,
"scriptDurationMs": 1009.9070000000002,
"eventListeners": -65,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.839583333333334,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3274.2580000000316,
"styleRecalcs": 43,
"styleRecalcDurationMs": 5.709,
"layouts": 4,
"layoutDurationMs": 1.186,
"taskDurationMs": 2157.6279999999997,
"heapDeltaBytes": -6215688,
"heapUsedBytes": 81107232,
"domNodes": -257,
"jsHeapTotalBytes": 6311936,
"scriptDurationMs": 1007.903,
"eventListeners": -63,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.838144329896906,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "minimap-geometry-cadence",
"durationMs": 3126.0090000000105,
"styleRecalcs": 8,
"styleRecalcDurationMs": 2.013,
"layouts": 7,
"layoutDurationMs": 1.9580000000000006,
"taskDurationMs": 851.5169999999998,
"heapDeltaBytes": 4031200,
"heapUsedBytes": 74760192,
"domNodes": -257,
"jsHeapTotalBytes": 3796992,
"scriptDurationMs": 376.132,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66577540106952,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-geometry-cadence",
"durationMs": 3184.7270000000094,
"styleRecalcs": 7,
"styleRecalcDurationMs": 1.8570000000000002,
"layouts": 7,
"layoutDurationMs": 2.088,
"taskDurationMs": 874.8589999999999,
"heapDeltaBytes": -12850996,
"heapUsedBytes": 59446400,
"domNodes": -257,
"jsHeapTotalBytes": 2224128,
"scriptDurationMs": 386.405,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665445026178013,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2076.5590000000316,
"styleRecalcs": 8,
"styleRecalcDurationMs": 2.7310000000000008,
"layouts": 8,
"layoutDurationMs": 2.8060000000000005,
"taskDurationMs": 770.8190000000001,
"heapDeltaBytes": -31157124,
"heapUsedBytes": 59612608,
"domNodes": -257,
"jsHeapTotalBytes": 4620288,
"scriptDurationMs": 38.469,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66532258064516,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2059.2289999999593,
"styleRecalcs": 7,
"styleRecalcDurationMs": 2.443999999999999,
"layouts": 7,
"layoutDurationMs": 2.263,
"taskDurationMs": 713.502,
"heapDeltaBytes": -24404280,
"heapUsedBytes": 68581436,
"domNodes": -257,
"jsHeapTotalBytes": 5144576,
"scriptDurationMs": 34.69,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-progress-execution-cadence",
"durationMs": 2852.9230000000043,
"styleRecalcs": 24,
"styleRecalcDurationMs": 10.318999999999999,
"layouts": 5,
"layoutDurationMs": 1.2950000000000002,
"taskDurationMs": 668.76,
"heapDeltaBytes": -5387284,
"heapUsedBytes": 77925804,
"domNodes": -239,
"jsHeapTotalBytes": 913408,
"scriptDurationMs": 203.38899999999998,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66549707602339,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-progress-execution-cadence",
"durationMs": 2753.1920000000127,
"styleRecalcs": 24,
"styleRecalcDurationMs": 11.393,
"layouts": 5,
"layoutDurationMs": 1.608,
"taskDurationMs": 676.4789999999999,
"heapDeltaBytes": -10252072,
"heapUsedBytes": 73359888,
"domNodes": -240,
"jsHeapTotalBytes": 5107712,
"scriptDurationMs": 215.36799999999997,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665454545454548,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-topology-cadence",
"durationMs": 3389.7650000000112,
"styleRecalcs": 7,
"styleRecalcDurationMs": 2.1519999999999992,
"layouts": 6,
"layoutDurationMs": 1.8210000000000002,
"taskDurationMs": 1073.283,
"heapDeltaBytes": 3117588,
"heapUsedBytes": 87269620,
"domNodes": -257,
"jsHeapTotalBytes": 4583424,
"scriptDurationMs": 411.41900000000004,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66600985221675,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "minimap-topology-cadence",
"durationMs": 3368.804999999952,
"styleRecalcs": 5,
"styleRecalcDurationMs": 1.3479999999999994,
"layouts": 5,
"layoutDurationMs": 1.287,
"taskDurationMs": 1037.3429999999998,
"heapDeltaBytes": -6485296,
"heapUsedBytes": 69775444,
"domNodes": -257,
"jsHeapTotalBytes": 4321280,
"scriptDurationMs": 403.598,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.748756218905474,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 663.5559999999714,
"styleRecalcs": 47,
"styleRecalcDurationMs": 5.852,
"layouts": 10,
"layoutDurationMs": 2.856,
"taskDurationMs": 463.07699999999994,
"heapDeltaBytes": 1792096,
"heapUsedBytes": 65233684,
"domNodes": 0,
"jsHeapTotalBytes": 9961472,
"scriptDurationMs": 149.93599999999998,
"eventListeners": 46,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.664102564102567,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 648.0320000000575,
"styleRecalcs": 46,
"styleRecalcDurationMs": 5.202000000000001,
"layouts": 8,
"layoutDurationMs": 2.343,
"taskDurationMs": 454.476,
"heapDeltaBytes": 1739516,
"heapUsedBytes": 64645020,
"domNodes": 0,
"jsHeapTotalBytes": 11010048,
"scriptDurationMs": 149.093,
"eventListeners": 46,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.664102564102567,
"p95FrameDurationMs": 16.700000000000273
},
{
"name": "subgraph-idle",
"durationMs": 2017.0300000000054,
"styleRecalcs": 9,
"styleRecalcDurationMs": 2.486,
"layouts": 8,
"layoutDurationMs": 2.361,
"taskDurationMs": 596.445,
"heapDeltaBytes": 4424028,
"heapUsedBytes": 68126528,
"domNodes": 0,
"jsHeapTotalBytes": 9437184,
"scriptDurationMs": 11.168,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666942148760334,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2039.0929999999798,
"styleRecalcs": 9,
"styleRecalcDurationMs": 2.606,
"layouts": 9,
"layoutDurationMs": 2.7030000000000007,
"taskDurationMs": 576.272,
"heapDeltaBytes": 534092,
"heapUsedBytes": 64149116,
"domNodes": -183,
"jsHeapTotalBytes": 7077888,
"scriptDurationMs": 9.383000000000001,
"eventListeners": -115,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.6655737704918,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1740.998999999988,
"styleRecalcs": 77,
"styleRecalcDurationMs": 15.184,
"layouts": 9,
"layoutDurationMs": 2.5260000000000002,
"taskDurationMs": 908.344,
"heapDeltaBytes": 16448380,
"heapUsedBytes": 79690308,
"domNodes": 0,
"jsHeapTotalBytes": 6815744,
"scriptDurationMs": 110.44699999999999,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665384615384614,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1758.5149999999885,
"styleRecalcs": 81,
"styleRecalcDurationMs": 15.438,
"layouts": 11,
"layoutDurationMs": 3.03,
"taskDurationMs": 913.177,
"heapDeltaBytes": -4471564,
"heapUsedBytes": 67750908,
"domNodes": 0,
"jsHeapTotalBytes": 11534336,
"scriptDurationMs": 107.798,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665714285714284,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-transition-enter",
"durationMs": 1434.5199999999636,
"styleRecalcs": 19,
"styleRecalcDurationMs": 22.788000000000004,
"layouts": 5,
"layoutDurationMs": 14.078999999999997,
"taskDurationMs": 1073.766,
"heapDeltaBytes": 7417152,
"heapUsedBytes": 91731260,
"domNodes": 16030,
"jsHeapTotalBytes": 16392192,
"scriptDurationMs": 32.07299999999999,
"eventListeners": 3246,
"totalBlockingTimeMs": 161,
"frameDurationMs": 19.594594594594593,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "viewport-pan-sweep",
"durationMs": 9192.196000000024,
"styleRecalcs": 251,
"styleRecalcDurationMs": 31.124000000000002,
"layouts": 8,
"layoutDurationMs": 2.27,
"taskDurationMs": 6453.086,
"heapDeltaBytes": 2695644,
"heapUsedBytes": 95056772,
"domNodes": -257,
"jsHeapTotalBytes": 6680576,
"scriptDurationMs": 2888.862,
"eventListeners": -66,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.696363636363635,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "viewport-pan-sweep",
"durationMs": 9177.973000000065,
"styleRecalcs": 251,
"styleRecalcDurationMs": 30.149999999999995,
"layouts": 8,
"layoutDurationMs": 2.34,
"taskDurationMs": 6401.941,
"heapDeltaBytes": 1093604,
"heapUsedBytes": 82267020,
"domNodes": -257,
"jsHeapTotalBytes": 5038080,
"scriptDurationMs": 2890.9069999999997,
"eventListeners": -66,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.696357012750457,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-idle",
"durationMs": 20978.215000000033,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 20505.032,
"heapDeltaBytes": -23398340,
"heapUsedBytes": 258990076,
"domNodes": -3343,
"jsHeapTotalBytes": 11264000,
"scriptDurationMs": 323.61600000000004,
"eventListeners": -653,
"totalBlockingTimeMs": 0,
"frameDurationMs": 37.33523131672598,
"p95FrameDurationMs": 66.79999999999927
},
{
"name": "vue-large-graph-idle",
"durationMs": 20722.034000000007,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 20244.684000000005,
"heapDeltaBytes": -12416412,
"heapUsedBytes": 269583024,
"domNodes": -3343,
"jsHeapTotalBytes": 7069696,
"scriptDurationMs": 290.14300000000003,
"eventListeners": -653,
"totalBlockingTimeMs": 0,
"frameDurationMs": 38.433766233766235,
"p95FrameDurationMs": 83.29999999999927
},
{
"name": "vue-large-graph-pan",
"durationMs": 25726.511000000017,
"styleRecalcs": 181,
"styleRecalcDurationMs": 22.865999999999996,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 25412.357,
"heapDeltaBytes": -14749316,
"heapUsedBytes": 261918952,
"domNodes": -3343,
"jsHeapTotalBytes": 14934016,
"scriptDurationMs": 908.339,
"eventListeners": -651,
"totalBlockingTimeMs": 28,
"frameDurationMs": 48.125280898876404,
"p95FrameDurationMs": 100
},
{
"name": "vue-large-graph-pan",
"durationMs": 25459.824000000026,
"styleRecalcs": 179,
"styleRecalcDurationMs": 21.921999999999997,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 25189.512,
"heapDeltaBytes": 5913784,
"heapUsedBytes": 282533464,
"domNodes": -3343,
"jsHeapTotalBytes": 11456512,
"scriptDurationMs": 930.388,
"eventListeners": -651,
"totalBlockingTimeMs": 13,
"frameDurationMs": 48.01679245283018,
"p95FrameDurationMs": 100
},
{
"name": "vue-node-resize-workload",
"durationMs": 7736.477000000036,
"styleRecalcs": 20,
"styleRecalcDurationMs": 57.43500000000002,
"layouts": 20,
"layoutDurationMs": 23.129000000000012,
"taskDurationMs": 7637.817,
"heapDeltaBytes": -35208272,
"heapUsedBytes": 257214136,
"domNodes": -3343,
"jsHeapTotalBytes": 17813504,
"scriptDurationMs": 138.336,
"eventListeners": -653,
"totalBlockingTimeMs": 0,
"frameDurationMs": 42.612500000000004,
"p95FrameDurationMs": 83.39999999999964
},
{
"name": "vue-node-resize-workload",
"durationMs": 7416.174999999953,
"styleRecalcs": 20,
"styleRecalcDurationMs": 79.042,
"layouts": 20,
"layoutDurationMs": 24.061,
"taskDurationMs": 7330.397000000001,
"heapDeltaBytes": -16388668,
"heapUsedBytes": 265654672,
"domNodes": -3343,
"jsHeapTotalBytes": 16506880,
"scriptDurationMs": 110.98300000000002,
"eventListeners": -653,
"totalBlockingTimeMs": 0,
"frameDurationMs": 42.86878612716764,
"p95FrameDurationMs": 83.39999999999964
},
{
"name": "vue-zoom-out-idle",
"durationMs": 10509.338999999954,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10221.955999999998,
"heapDeltaBytes": 2849112,
"heapUsedBytes": 261637252,
"domNodes": 0,
"jsHeapTotalBytes": 12845056,
"scriptDurationMs": 147.436,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 37.904693140794215,
"p95FrameDurationMs": 83.29999999999927
},
{
"name": "vue-zoom-out-idle",
"durationMs": 10183.641999999963,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 9923.991999999998,
"heapDeltaBytes": 14120816,
"heapUsedBytes": 277179588,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 138.299,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 37.16423357664234,
"p95FrameDurationMs": 83.20000000000073
},
{
"name": "workflow-execution",
"durationMs": 159.0929999999844,
"styleRecalcs": 12,
"styleRecalcDurationMs": 11.289,
"layouts": 10,
"layoutDurationMs": 4.635,
"taskDurationMs": 110.80799999999999,
"heapDeltaBytes": 4795764,
"heapUsedBytes": 65921792,
"domNodes": 126,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 11.552000000000001,
"eventListeners": 5,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000273
},
{
"name": "workflow-execution",
"durationMs": 147.49600000004648,
"styleRecalcs": 11,
"styleRecalcDurationMs": 9.597,
"layouts": 10,
"layoutDurationMs": 4.205,
"taskDurationMs": 109.204,
"heapDeltaBytes": -8268692,
"heapUsedBytes": 64112076,
"domNodes": 124,
"jsHeapTotalBytes": 10223616,
"scriptDurationMs": 6.021,
"eventListeners": 5,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999727
},
{
"name": "agent-composer-empty-conversation-typing",
"durationMs": 482.00099999996837,
"styleRecalcs": 5,
"styleRecalcDurationMs": 0.63,
"layouts": 76,
"layoutDurationMs": 15.703000000000001,
"taskDurationMs": 442.815,
"heapDeltaBytes": -4245552,
"heapUsedBytes": 38648068,
"domNodes": 4,
"jsHeapTotalBytes": 2359296,
"scriptDurationMs": 203.99,
"eventListeners": 292,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665517241379316,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "agent-composer-empty-conversation-typing",
"durationMs": 467.0370000000048,
"styleRecalcs": 5,
"styleRecalcDurationMs": 0.6130000000000001,
"layouts": 76,
"layoutDurationMs": 15.744000000000002,
"taskDurationMs": 427.256,
"heapDeltaBytes": -4346768,
"heapUsedBytes": 39637152,
"domNodes": 4,
"jsHeapTotalBytes": 3670016,
"scriptDurationMs": 200.123,
"eventListeners": 292,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.668965517241396,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "agent-composer-long-conversation-typing",
"durationMs": 612.1749999999793,
"styleRecalcs": 5,
"styleRecalcDurationMs": 1.9680000000000002,
"layouts": 76,
"layoutDurationMs": 16.580000000000002,
"taskDurationMs": 565.2120000000001,
"heapDeltaBytes": 971192,
"heapUsedBytes": 98966364,
"domNodes": 4,
"jsHeapTotalBytes": 790528,
"scriptDurationMs": 213.28900000000002,
"eventListeners": 292,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "agent-composer-long-conversation-typing",
"durationMs": 576.7379999999775,
"styleRecalcs": 5,
"styleRecalcDurationMs": 1.936,
"layouts": 76,
"layoutDurationMs": 16.642,
"taskDurationMs": 535.335,
"heapDeltaBytes": 924244,
"heapUsedBytes": 90236232,
"domNodes": 4,
"jsHeapTotalBytes": 266240,
"scriptDurationMs": 208.40599999999998,
"eventListeners": 296,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66571428571429,
"p95FrameDurationMs": 16.799999999999272
}
]
}🔬 E2E CoverageLine coverage: 60.4% Coverage totals
Top 10 files by uncovered lines
🎭 Playwright: ✅ 3112 passed, 0 failed · 4 flaky📊 Browser Reports
🎨 Storybook: ✅ Built — View Storybook🧩 Custom Nodes Ecosystem Matrix: 🟡 Withheld — could not measureExecutes the frontend JS of every registry pack that ships any, against this change, in both renderers. The eight
Findings and check detailsThe matrix runs on same-repository version-bump PRs with relevant changes, main-branch pushes, and manual dispatches. This candidate does not qualify, so every expensive stage was 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 🌾 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. Findings and check detailsNo new findings in the changed files. Reproduce locally# the base the gate compared against
npx fallow audit --root . --quiet --format json \
--changed-since 6d00a3cd274dc6322481ab54e976696ab9036aa0
Full docs: ADR-DEVEX-LINT-0015 🏛️ Architecture: ✅ PassedFindings and check detailsWarnings (204)
Reproduce locallypnpm architecture:checkFull docs: Frontend domain catalog |
Codecov Report❌ Patch coverage is
@@ Coverage Diff @@
## main #21326 +/- ##
==========================================
+ Coverage 86.63% 87.57% +0.93%
==========================================
Files 3942 2782 -1160
Lines 185024 154557 -30467
Branches 48201 39526 -8675
==========================================
- Hits 160304 135356 -24948
+ Misses 22257 18526 -3731
+ Partials 2463 675 -1788
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 1170 files with indirect coverage changes 🚀 New features to boost your workflow:
|
Seeds a saved workflow whose stored content is JSON null with a newer local draft, reloads, and reuses it by name by dropping default.json. On the parent this throws the production signatures from CLOUD-FRONTEND-PROD-2Z2/-2Z3/-28Y/-233 (reading 'nodes' from graphEqual, reading 'extra' on the next load); with the hardening it loads without page errors, leaves the stored file untouched and keeps the draft. Adds a unit case for the same state: stored null overlaid by a newer draft is rejected and the draft is kept.
Closing a persisted tab unloads it and drops its draft but left isModified set, so the workflow stayed in modifiedWorkflows. Cloud sign-out's "Save" iterates modifiedWorkflows and saved it, writing JSON.stringify(null) over the stored file. That stored null is the state behind CLOUD-FRONTEND-PROD-2Z2/-2Z3/-28Y/-233.
|
Root cause now confirmed; e2e red on parent, green here; one gap left for #16960.
Full context for agent readersProduction evidence (1.57.0, 2026-10-21T18:32:33Z, one iPhone session, trace
Why a stored Writer (verified at store level on the parent Added to this PR
Residual for #16960 (Matt Miller): not fixed here. Run on this branch, the dropped graph stays on the canvas but the previously selected "Unsaved Workflow" tab stays active and captures it (7 nodes, modified). A |
dante01yoon
left a comment
There was a problem hiding this comment.
Approving. This started as hypothesis-driven hardening and ended with the cause confirmed and a black-box test that reproduces it, which is the order I want these in — the body is explicit that the first version was "hardening plus instrumentation, not a verified fix," and the 2026-10-21 update says the stored file is literally null. That's the difference between guarding a symptom and knowing what you're guarding.
workflowNullStoredContent.spec.ts is the strongest part. It seeds the actual production state — workflows/default.json stored as null with a newer local draft — drives real gestures (reload, drag-drop, a canvas drag, then New), and asserts pageErrors is empty. Then it asserts the two things I'd most want and would not have expected a test to check:
expect(storedContent).toBe('null')
expect(await readDraft(page)).not.toBeNull()The stored file is not rewritten and the draft holding the user's last edits survives. The failure mode a fix like this invites is quietly healing the broken file by overwriting it, which would destroy whatever the user still had; this pins that it doesn't. Good instinct, and it also answered the question I'd formed while reading parseStoredWorkflow — that throwing on null content might turn a degraded open into a hard failure. For the draft case it doesn't, and load() could already throw on unparseable content before this, so the new throw joins an existing error path rather than creating one.
Telemetry hygiene checks out: reportWorkflowStateAnomaly sends method, booleans and originalContentKind only. workflow.path appears solely in the local dedup key and never reaches assert, so the "no names, paths or content" claim in the body holds as written.
I also agree with throwing from save()/saveAs() rather than returning quietly. workflowStore.saveWorkflow wraps it in try/finally with no catch, so it propagates to the command layer and surfaces instead of marking the workflow clean — which is the right trade when the alternative is JSON.stringify(null) landing on the user's file. The refusal is only reachable in the broken state, so a visible error there is strictly better than a silent one.
No blocking findings. One thing worth watching rather than changing:
Unanchored
nit-pick (non-blocking):assertthrows in DEV, so each of these guards converts a recoverable production path into an immediate dev-build failure. The body names this and it matches the existing inactive-tracker assertion, so it's consistent rather than new. The one to keep an eye on isChangeTracker.graphEqual— it sits on an ordinary canvas-capture path, so if the null state turns out to be more reachable in dev than expected, that guard is the one that will make itself known loudly. Not a reason to change it; the whole point is to find the trigger. Just the first place I'd look if dev builds start tripping after this lands.
|
|
|
|
…state and report occurrences (#21977) <!-- claude-slack-attribution --> _Requested via [Slack thread](https://comfy-organization.slack.com/archives/C09KFK2ABB8/p1792312102906099?thread_ts=1792312102.906099&cid=C09KFK2ABB8)_ **Before:** if a workflow became the active one without its change tracker ever loading, activating it threw `Cannot read properties of null (reading 'extra')` (and a sibling `reading 'nodes'` of null in `ChangeTracker.graphEqual`). A save from that state could also write `null` over the stored workflow file. **After:** the missing state is tolerated where it is read, an unloaded workflow is loaded instead of being returned as loaded, a save from an unloaded workflow is refused, and each guard hit is reported with enough structure to identify the real trigger. Manual backport of #21326 (the automatic backport hit conflicts). Conflict resolved: `src/platform/workflow/core/services/workflowService.test.ts` only, in the `vi.mock` block near the top. Kept the new `@/base/assert` mock and dropped the two `lateCustomNodes` mocks, since that module does not exist on this branch. The cherry-pick commit message was reworded; no code changes beyond the original PR. Local checks: typecheck (app and browser), oxfmt, oxlint and eslint on touched files, and vitest for `src/platform/workflow` and `src/scripts/changeTracker` (64 files, 1232 passed). The new Playwright spec `browser_tests/tests/workflowNullStoredContent.spec.ts` is included but was only typechecked locally, not run. --- _Generated by [Claude Code](https://claude.ai/code/session_01QBengj382AMebTiwLDtHtr)_ Co-authored-by: Claude <noreply@anthropic.com>
…tate and report occurrences (#21974) <!-- claude-slack-attribution --> _Requested via [Slack thread](https://comfy-organization.slack.com/archives/C09KFK2ABB8/p1792312102906099?thread_ts=1792312102.906099&cid=C09KFK2ABB8)_ **Before:** if a workflow became the active one without its change tracker ever loading, activating it threw `Cannot read properties of null (reading 'extra')` (and a sibling `reading 'nodes'` of null in `ChangeTracker.graphEqual`). A save from that state could also write `null` over the stored workflow file. **After:** the missing state is tolerated where it is read, an unloaded workflow is loaded instead of being returned as loaded, a save from an unloaded workflow is refused, and each guard hit is reported with enough structure to identify the real trigger. Manual backport of #21326 (the automatic backport hit conflicts). Conflict resolved: `src/platform/workflow/core/services/workflowService.test.ts` only, in the `vi.mock` block near the top. Kept the new `@/base/assert` mock and dropped the two `lateCustomNodes` mocks, since that module does not exist on this branch. The cherry-pick commit message was reworded; no code changes beyond the original PR. Local checks: typecheck (app and browser), oxfmt, oxlint and eslint on touched files, and vitest for `src/platform/workflow` and `src/scripts/changeTracker` (64 files, 1238 passed). The new Playwright spec `browser_tests/tests/workflowNullStoredContent.spec.ts` is included but was only typechecked locally, not run. --- _Generated by [Claude Code](https://claude.ai/code/session_01QBengj382AMebTiwLDtHtr)_ Co-authored-by: Claude <noreply@anthropic.com>
…state (#21326) (#22397) ## Summary - Backports #21326 to `cloud/1.58`, the release line currently serving the production events tracked in [PM-2409](https://linear.app/comfy/issue/PM-2409). - Prevents literal `null` workflow content from entering workflow load/save and makes ChangeTracker comparison fail safe when either graph state is null. - Adds the same black-box regression from #21326, covering close-without-save, sign-out/save, workflow-name reuse, and reopen without a crash. ## Production evidence Sentry issue `CLOUD-FRONTEND-PROD-36D` contains one production throw captured twice (global and Vue handlers): 2 raw events / 1 user on release `1.58.0`, Windows/Electron. The exact `graphEqual` failure family spans 15 Sentry groups and 614 raw events since 2026-10-05 across releases 1.55.17–1.58.0. Raw counts include handled/unhandled and dual-capture groups, so they are not unique crashes or users. The PM-2409 path is node shape customization: the node shape is assigned, then `captureCanvasState()` throws while comparing against a null stored graph. The visible mutation may occur, but the dirty/undo checkpoint and subsequent menu completion are interrupted. This is the same root cause already owned by [FRONT-3169](https://linear.app/comfy/issue/FRONT-3169) and fixed on `main` by #21326. `cloud/1.58` still had the unguarded implementation. ## Backport notes - Production changes and the black-box spec are byte-identical to #21326. - The only cherry-pick conflict was in a test mock block. Two `lateCustomNodes` mocks from `main` were omitted because those modules do not exist on `cloud/1.58`; the branch's existing `mockAssert` setup was retained. ## Verification - `NODE_OPTIONS=--no-experimental-webstorage pnpm exec vitest run src/platform/workflow/core/services/workflowService.test.ts src/platform/workflow/management/stores/comfyWorkflow.test.ts src/platform/workflow/management/stores/workflowStore.test.ts src/scripts/changeTracker.test.ts` — 350 passed - `pnpm typecheck` — passed - `pnpm typecheck:browser` — passed - changed-file `oxlint --type-aware` and `oxfmt --check` — passed - `VITE_USE_LEGACY_DEFAULT_GRAPH=true pnpm build` — passed - `scripts/test-browser-offline.sh --project=chromium workflowNullStoredContent.spec.ts` — 1 passed This PR is a carrier only; production remains affected until the release line is deployed. Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>
…he workflow picker saves it (FRONT-3774) (#22819) Backport of #22649 then #22648 to `cloud/1.58`. Fixes [FRONT-3774](https://linear.app/comfyorg/issue/FRONT-3774): picking an unopened unsaved tab in the Agent workflow picker after a reload could not save it, and the tab's draft was deleted. Found in release QA case C-163 on 1.59.3; the same code is on 1.58. - **Applies here:** `cloud/1.58` has both defects. With the two fixes reverted, all five new tests fail on this branch (2 draft-deletion, 3 picker); with them, they pass. - **Same symptom as 1.59:** `cloud/1.58` has #21326's guard (as #22397), so the pick fails with "Save failed · Cannot save workflow … before it is loaded". - **Two commits, source order.** Conflicts were imports only; behaviour hunks apply unchanged. <details><summary>Full context for agent readers</summary> Commits, in order: the two `core/1.58` carrier commits (#22818, each `cherry-pick -x` from `main`), which apply cleanly here. Their resolutions: 1. #22649 `9580dff303`, fix(workflow): stop deleting an unsaved tab's draft on a same-path save or its first load. - `useWorkflowPersistenceV2.test.ts` import conflict: kept the train's `vue` import (this change adds no `computed`/`ref` use) and took the source's `vitest` import with `onTestFinished`, which the new test uses. - `useWorkflowPersistenceV2.ts` imports `ComfyWorkflowJSON` from `@/platform/workflow/validation/schemas/workflowSchema`, because 1.58 predates the `@comfyorg/graph-model` package. Same resolution the bot used on `core/1.59` (#22667). 2. #22648 `816ea98f48`, fix(agent): load an unopened unsaved tab before the workflow picker saves it. - `workflowService.test.ts` import conflict: kept the train's imports plus `blankGraph`, which the new `loadWorkflowInBackground` test uses (`getDefaultWorkflow` from `main` is not used by this change). - `agentWorkflowSelection.spec.ts`: same `ComfyWorkflowJSON` import-path change as above. Every `browser_tests/` hunk from both sources is carried; none is dropped. Local checks on this tree (Node 24): `typecheck` (pre-commit) and `typecheck:browser` pass; pre-push depcruise and knip pass. Vitest on `useAgentWorkflowSelection.test.ts`, `workflowService.test.ts`, `AgentPanelRoot.test.ts` and `src/platform/workflow/persistence/`: 807 passed, 1 expected fail, 14 files. The new browser case `picks an unsaved tab a reload left unopened` runs in CI (`e2e-status`). Authority: the program's release-backport grant for the current trains (`cloud/1.58` is listed). </details> Justification: Release backport of FRONT-3774 fixes (#22648, plus #22649 where the train lacked it) to cloud/1.58 under the standing agent backport review-and-merge grant for the current trains. Production hunks match the source PRs line for line (only a train-local import path differs); the source unit tests fail on the train tip without the carrier and pass with it; all required checks pass at 8634d19; no unresolved threads.
…e workflow picker saves it (FRONT-3774) (#22818) Backport of #22649 then #22648 to `core/1.58`. Fixes [FRONT-3774](https://linear.app/comfyorg/issue/FRONT-3774): picking an unopened unsaved tab in the Agent workflow picker after a reload could not save it, and the tab's draft was deleted. Found in release QA case C-163 on 1.59.3; the same code is on 1.58. - **Applies here:** `core/1.58` has both defects. With the two fixes reverted, all five new tests fail on this branch (2 draft-deletion, 3 picker); with them, they pass. - **Worse symptom on this train:** `core/1.58` lacks #21326's "refuse to save an unloaded workflow" guard, so the picker's save of an unloaded tab writes an empty (`null`) workflow instead of failing. Loading the tab first fixes both shapes. #21326 itself is not carried. - **Two commits, source order.** Conflicts were imports only; behaviour hunks apply unchanged. <details><summary>Full context for agent readers</summary> Commits, in order (both `cherry-pick -x` from `main`): 1. #22649 `9580dff303`, fix(workflow): stop deleting an unsaved tab's draft on a same-path save or its first load. - `useWorkflowPersistenceV2.test.ts` import conflict: kept the train's `vue` import (this change adds no `computed`/`ref` use) and took the source's `vitest` import with `onTestFinished`, which the new test uses. - `useWorkflowPersistenceV2.ts` imports `ComfyWorkflowJSON` from `@/platform/workflow/validation/schemas/workflowSchema`, because 1.58 predates the `@comfyorg/graph-model` package. Same resolution the bot used on `core/1.59` (#22667). 2. #22648 `816ea98f48`, fix(agent): load an unopened unsaved tab before the workflow picker saves it. - `workflowService.test.ts` import conflict: kept the train's imports plus `blankGraph`, which the new `loadWorkflowInBackground` test uses (`getDefaultWorkflow` from `main` is not used by this change). - `agentWorkflowSelection.spec.ts`: same `ComfyWorkflowJSON` import-path change as above. Every `browser_tests/` hunk from both sources is carried; none is dropped. Local checks on this tree (Node 24): `typecheck` (pre-commit) and `typecheck:browser` pass; pre-push depcruise and knip pass. Vitest on `useAgentWorkflowSelection.test.ts`, `workflowService.test.ts`, `AgentPanelRoot.test.ts` and `src/platform/workflow/persistence/`: 808 passed, 1 expected fail, 14 files. The new browser case `picks an unsaved tab a reload left unopened` runs in CI (`e2e-status`). Authority: the program's release-backport grant for the current trains (`core/1.58` is listed). </details> Justification: FRONT-3774 backport to core/1.58 under the standing agent backport grant for the current trains. Production hunks match main #22648 (and #22649 persistence) line for line; all 132 exact-head checks at d0d9c5c are green after rerunning one flaky Playwright shard; no unresolved threads. Sibling carriers #22816, #22817 and #22819 merged the same way.
…-3C3 (Comfy-Org#22877) ## Summary The `openWorkflow` state-anomaly report added in Comfy-Org#21326 is covered only for a remote-synced **persisted** workflow. The first production occurrence was a different shape — a **temporary, never-loaded** instance aliasing the active path — and it reports a different payload. This pins that payload so the next occurrence stays decidable. This is test-only. No production code changes. Comfy-Org#21326 was explicitly "hardening plus instrumentation, not a verified fix", with the stated purpose that "every guard hit is reported with enough structure to identify the real trigger on the next occurrence". That occurrence has now happened (`CLOUD-FRONTEND-PROD-3C3`, release `1.59.5`, 1 event / 1 user, handled and recovered), and its payload rules out two of the three candidate explanations in Comfy-Org#21326's hypothesis: | Field reported | What it rules out | | --- | --- | | `originalContentKind: 'object'` | the "stored content parses to JSON `null`" branch | | `originalContentKind: 'object'` | the "has since unloaded its change tracker" branch — `unload()` sets `originalContent = null`, which classifies as `'unavailable'`, so an unloaded instance **cannot** report `'object'` | | `isTemporary: true`, `isPersisted: false`, `isLoaded: false` | a persisted workflow, and any instance that ever completed a load | What survives is the same-path-different-instance branch, narrowed to a freshly minted temporary: `createNewWorkflow` produces exactly this state (`size: -1`, `originalContent` set from the default graph, `changeTracker` still `null`), and `isActive` matches by path rather than identity. That discrimination depends entirely on the `originalContentKind` contract holding. If `unload()` ever stops clearing `originalContent`, or `createNewWorkflow` stops seeding it, the three cases silently collapse back into one undifferentiated hypothesis and the next occurrence tells us nothing. Nothing currently pins that, which is what this test adds. **What this test does not claim.** The alias is constructed, not reproduced. What gets two instances onto one path in production is still unknown, and this is not that trigger — it pins the telemetry contract, not the cause. That caveat lives here rather than in test comments, which the repository forbids. ## Verification - `src/platform/workflow/management/stores/workflowStore.test.ts` — **95/95 pass** (`vitest run`, Node 26.10.0, `NODE_OPTIONS=--no-experimental-webstorage`). - **Discrimination proven, not assumed.** With `originalContentKind` flipped to `'unavailable'` the new test fails with `expected "unavailable" / received "object"`, so the field is load-bearing rather than incidentally satisfied by `objectContaining`. - `oxfmt --check` clean; `oxlint --type-aware` clean; `typecheck:app` clean (all three also enforced by the pre-commit hook on this commit). - No production file is touched, so there is no user-visible behaviour to exercise through the black-box harness. ## After release `CLOUD-FRONTEND-PROD-3C3` is 1 event / 1 user and self-recovered, so this PR deliberately does not attempt a fix. Two sibling groups from the same Comfy-Org#21326 instrumentation are larger and user-visible, and are tracked on FRONT-3169 rather than here: `CLOUD-FRONTEND-PROD-397` ("Workflow content is not a JSON object", 13 events / 8 users) and `CLOUD-FRONTEND-PROD-393` ("Workflow saved without an active state", 7 events / 7 users), both still firing on `1.59.5`.
…ned (Comfy-Org#22946) ## Summary Opening a saved workflow whose stored file is empty or JSON that is not an object (files written as `null` by frontends before Comfy-Org#21326) now shows the user an error toast with a **Delete** action, instead of rejecting silently. Before: clicking such a workflow in the sidebar did nothing visible. `openWorkflow` rejected, nobody caught it, and Sentry got two groups for one click: the static anomaly report from Comfy-Org#21326 (`[Assertion failed]: Workflow content is not a JSON object`) plus an unhandled `Error` whose title contained the workflow path, so it split into one group per file name and carried the user's file name. ## Changes - `ComfyWorkflow.load` throws a typed `UnreadableWorkflowError` (`reason: 'empty' | 'not-object'`, `workflowPath`). The message is static; the path is a property only. - The empty-content case now goes through the same static, deduplicated anomaly report as the non-object case, so swallowing the throw does not lose that signal. - `workflowService.openWorkflow` catches only that error, ends the navigation, returns `false` (its existing "did not open" result) and shows one toast per file: "Can't open this workflow" / "The saved file for {name} is empty or damaged, so it can't be opened. You can delete it from your workflows." Delete goes through the normal `deleteWorkflow` path (confirm dialog when the setting is on) and dismisses the toast when the file is gone. Repeated clicks reuse the same toast. Every other load failure still rejects as before. - The canvas is not touched: the failure happens before `loadGraphData`. - Other paths that open the same file also recover instead of rejecting: - Loading a graph under the name of an unreadable stored file (startup draft restore, dropping a file with the same name) opens it as a new workflow and shows the toast. The store no longer leaves a tab behind for a file that failed to load. - Closing the active tab falls back to the default graph when the next tab is the unreadable file, instead of refusing to close. - On startup, when the session's workflow does not open, the app falls back to the latest draft or the default graph instead of reporting it restored with an empty canvas. - No toast for an open the caller has already moved away from. Context: Sentry `CLOUD-FRONTEND-PROD-397` ("Workflow content is not a JSON object", tracked on FRONT-3169). No new `null` files should be written since Comfy-Org#21326's save guard; this handles files corrupted before it. ## Tests - Red first: every new unit test failed before its fix and passes with it. - `workflowStore.test.ts`: typed error, message without the path, empty content reported. - `workflowService.test.ts`: returns `false` with a toast and no graph load; Delete deletes and dismisses; one toast per file; no toast for a superseded open; same-name graph load opens as a new workflow; tab close falls back to the default graph; other errors still reject. - `workflowStore.test.ts`: no tab left behind after a failed load. `useWorkflowPersistenceV2.test.ts`: startup falls back when the session workflow does not open. - e2e `workflowNullStoredContent.spec.ts`: stores `null` for a workflow, clicks it in the sidebar, expects the toast with the file name and no page errors, deletes from the toast through the confirm dialog, and checks the file is gone from the sidebar. - Local: `src/platform/workflow` unit tests (2365 pass across `src/platform/workflow` and the sidebar tests; `missingModelScan.test.ts` times out on external network loads, unrelated), `typecheck:app`, `typecheck:browser`, `oxlint --type-aware`, `oxfmt --check`. ## Not in this PR - A restored tab whose file is unreadable stays in the tab bar (clicking it shows the toast with Delete). Before this PR the same case aborted the rest of startup. - Agent panel callers show their own "workflow unavailable" message on `false`, so an agent-initiated open of a broken file shows both messages. - Server-side cleanup of existing `null` files: not done. The in-app Delete covers the affected users, and the backend owns stored user files. - "Workflow saved without an active state" (`CLOUD-FRONTEND-PROD-393`) is a separate path and gets its own PR. --------- Co-authored-by: GitHub Action <action@github.com>
…ead of failing (#22971) (#22983) Backport of #22971 to `cloud/1.59`. Fixes [FRONT-3837](https://linear.app/comfyorg/issue/FRONT-3837): cloud prod alert "Cannot save workflow … before it is loaded" (Sentry CLOUD-FRONTEND-PROD-394 / PROD-393). - Save and Save As load a tab that is not loaded first (with its draft), instead of hitting the #21326 guard. The guard stays for an active tab with no content. - Workflow sync no longer unloads or re-stamps a modified background tab, which used to lose its draft. - If the edits are already gone: nothing is written, a warning toast shows, and a static report is sent. - The guard message no longer contains the file path. - **Conflict:** only `src/locales/en/main.json`. Kept this branch's keys and added the two new `workflowService` keys; the `unreadableWorkflow*` keys from #22946 are not added here. - Auto-backport failed on merge conflicts; cherry-picked `03bbb78ba2` with `-x`. --------- Co-authored-by: GitHub Action <action@github.com>
…ead of failing (#22971) (#22981) Backport of #22971 to `cloud/1.60`. Fixes [FRONT-3837](https://linear.app/comfyorg/issue/FRONT-3837): cloud prod alert "Cannot save workflow … before it is loaded" (Sentry CLOUD-FRONTEND-PROD-394 / PROD-393). - Save and Save As load a tab that is not loaded first (with its draft), instead of hitting the #21326 guard. The guard stays for an active tab with no content. - Workflow sync no longer unloads or re-stamps a modified background tab, which used to lose its draft. - If the edits are already gone: nothing is written, a warning toast shows, and a static report is sent. - The guard message no longer contains the file path. - **Conflict:** only `src/locales/en/main.json`. Kept this branch's keys and added the two new `workflowService` keys; the `unreadableWorkflow*` keys from #22946 are not added here. - Auto-backport failed on merge conflicts; cherry-picked `03bbb78ba2` with `-x`. --------- Co-authored-by: GitHub Action <action@github.com>
…ad of failing (#22971) (#22980) Backport of #22971 to `core/1.60`. Fixes [FRONT-3837](https://linear.app/comfyorg/issue/FRONT-3837): cloud prod alert "Cannot save workflow … before it is loaded" (Sentry CLOUD-FRONTEND-PROD-394 / PROD-393). - Save and Save As load a tab that is not loaded first (with its draft), instead of hitting the #21326 guard. The guard stays for an active tab with no content. - Workflow sync no longer unloads or re-stamps a modified background tab, which used to lose its draft. - If the edits are already gone: nothing is written, a warning toast shows, and a static report is sent. - The guard message no longer contains the file path. - **Conflict:** only `src/locales/en/main.json`. Kept this branch's keys and added the two new `workflowService` keys; the `unreadableWorkflow*` keys from #22946 are not added here. - Auto-backport failed on merge conflicts; cherry-picked `03bbb78ba2` with `-x`. --------- Co-authored-by: GitHub Action <action@github.com>
…ad of failing (#22971) (#22982) Backport of #22971 to `core/1.59`. Fixes [FRONT-3837](https://linear.app/comfyorg/issue/FRONT-3837): cloud prod alert "Cannot save workflow … before it is loaded" (Sentry CLOUD-FRONTEND-PROD-394 / PROD-393). - Save and Save As load a tab that is not loaded first (with its draft), instead of hitting the #21326 guard. The guard stays for an active tab with no content. - Workflow sync no longer unloads or re-stamps a modified background tab, which used to lose its draft. - If the edits are already gone: nothing is written, a warning toast shows, and a static report is sent. - The guard message no longer contains the file path. - **Conflict:** only `src/locales/en/main.json`. Kept this branch's keys and added the two new `workflowService` keys; the `unreadableWorkflow*` keys from #22946 are not added here. - Auto-backport failed on merge conflicts; cherry-picked `03bbb78ba2` with `-x`. --------- Co-authored-by: GitHub Action <action@github.com>
…ad of failing (#22971) (#22984) Backport of #22971 to `core/1.57`. Fixes [FRONT-3837](https://linear.app/comfyorg/issue/FRONT-3837): cloud prod alert "Cannot save workflow … before it is loaded" (Sentry CLOUD-FRONTEND-PROD-394 / PROD-393). - Save and Save As load a tab that is not loaded first (with its draft), instead of hitting the #21326 guard. The guard stays for an active tab with no content. - Workflow sync no longer unloads or re-stamps a modified background tab, which used to lose its draft. - If the edits are already gone: nothing is written, a warning toast shows, and a static report is sent. - The guard message no longer contains the file path. - **Manual adaptation:** `core/1.57` does not have #22648, so this backport also adds its `workflowService.loadWorkflowInBackground` helper (same code as on `main`), which the save path uses to load a tab. The Agent picker test changes from #22971 are left out because the picker tests from #22648 are not on this branch. - Auto-backport failed on merge conflicts; cherry-picked `03bbb78ba2` with `-x`. --------- Co-authored-by: GitHub Action <action@github.com>
…Org#22971) ## Summary Saving a tab whose content was not loaded failed with "Cannot save workflow '<path>' before it is loaded". That is the guard from Comfy-Org#21326, which stops the save from writing `null` over the file. On cloud production (1.59.5) it fires from the close dialog's **Save** on a saved file that was unloaded but still marked modified (Sentry `CLOUD-FRONTEND-PROD-394`, 12 events / 11 users), and from saves of unsaved tabs a reload restored without opening (`CLOUD-FRONTEND-PROD-393`, 34 / 32; the agent-picker part of that is Comfy-Org#22648). Linear: [FRONT-3837](https://linear.app/comfyorg/issue/FRONT-3837). Changes: - `saveWorkflow` / `saveWorkflowAs` load an unloaded tab in the background first (the serialized load queue, as Comfy-Org#22648 does for the picker), which applies its draft. This covers every caller: close dialog, Ctrl+S, autosave, workspace switch, picker. The guard in `ComfyWorkflow.save` stays as the backstop. - If loading finds no usable draft for a saved file marked modified, its edits were already gone: nothing is written, the modified mark is cleared, the user gets a warning toast ("Unsaved changes were not saved"), and it is reported (`Modified saved workflow lost its edits before saving`, static message, no path) so any remaining source stays visible. Before, the save failed; a plain write would have reported Save as done. - An **active** tab with no loaded content still hits the guard: a background copy is not what is on the canvas. - Root cause fixed for one path: workflow sync no longer unloads (or re-stamps) a **modified background tab** whose file changed remotely. Unloading made its draft older than the file, so the next load dropped the edits and left the modified mark, which is the state the close-dialog Save then failed on. The tab is now treated like the active one: its in-memory edits win until the user saves or discards them. - The guard's error message no longer contains the workflow path (alerts showed user file names in their titles). Design notes: - Sync keeping local edits means a remote change to a modified background tab (another device, the agent writing the file) is not shown until the user saves or discards. Saving then overwrites it, as it already does for the active tab. Before, the remote copy silently replaced the edits. - `discardSupersededLoad` is unchanged: it restores the pre-open state, and saving that state now loads first. ## Verification - Red first: on `main` the new service tests failed with the exact production message ("Cannot save workflow 'workflows/background.json' before it is loaded"), and the sync test failed with the tab unloaded. - `workflowService.test.ts` "saving a tab that is not loaded": close-dialog Save writes the draft; direct save writes the draft; no usable draft writes nothing, warns and reports; an active unloaded tab still refuses; a restored unsaved tab closes through Save As and is written. - `workflowStore.test.ts`: a modified background tab keeps its edits and metadata through a sync with a changed file; an unmodified one is still unloaded; guard message is static. - Local: all `src/platform/workflow` tests and `useAgentWorkflowSelection.test.ts` pass (2360), `typecheck:app`, `oxlint --type-aware`, `oxfmt`. - Adversarial review (fresh reviewer): its majors are fixed in the second commit (no write when edits are gone, active tab keeps the guard, kept tab's draft stays usable, Save As loads after the prompt). - After release: `CLOUD-FRONTEND-PROD-394` and `-393` should stop on the release carrying this; any `Modified saved workflow lost its edits before saving` events show what else still drops edits. --------- Co-authored-by: GitHub Action <action@github.com>
Requested via Slack thread
Before: if a workflow became the active one without its change tracker ever loading, activating it threw
TypeError: Cannot read properties of null (reading 'extra')(and a siblingreading 'nodes'of null inChangeTracker.graphEqual). Nothing recorded which path got it into that state, and a save from that state could writenullover the stored workflow file.After: the same situation is tolerated at each point that reads the missing state, an unloaded workflow is loaded instead of being handed back as loaded, a save from an unloaded workflow is refused, and every guard hit is reported with enough structure to identify the real trigger on the next occurrence.
Summary
This is hardening plus instrumentation, not a verified fix. A production alert shows
Cannot read properties of null (reading 'extra')inactivateLoadedWorkflowand a siblingreading 'nodes'of null inChangeTracker.graphEqual, but there was no usable stack, and nothing has reproduced it locally. The root cause below is a hypothesis.Hypothesis:
ComfyWorkflow.initialStateischangeTracker?.initialState ?? null, so it is null whenever the tracker is not loaded. That can happen when:workflowStore.openWorkflowreturns early becauseisActive(workflow)is true.isActivecompares by path, so a same-path instance that never loaded its tracker is returned as if it were loaded.null(load()builds a tracker around it).#16802 removed several
?.guards that used to absorb this (initialState,activeState?.id,changeTracker?.deactivate()inbeforeLoadNewGraph), turning the missing state into a hard throw.What changes:
activateLoadedWorkflow: tolerates a missinginitialState(falls back to the mode of the load being processed) and a missing tracker (skips the reset/restore instead of throwing).beforeLoadNewGraph: tolerates an active workflow with no tracker.ChangeTracker.graphEqual: returnsfalse(not equal) when either side is missing.updateModifiedleavesisModifieduntouched when the tracker has no initial state.ComfyWorkflow.load(): rejects stored content that is not a JSON object (null, number, string, array) instead of building a tracker around it.ComfyWorkflow.save()/saveAs(): refuse (throw, before touching content or storage) when there is no active state, sonullis never persisted. Throwing is deliberate here: returning quietly would let callers mark the workflow clean without saving.workflowStore.openWorkflow: an active workflow that is not loaded is now loaded instead of returned as loaded.assert(false, message, context)(the same reporting path as the inactive-tracker assertion), through a small shared helper incomfyWorkflow.ts,reportWorkflowStateAnomaly. The message is static per call site. The context carriesmethod,isLoaded,hasChangeTracker,hasInitialState,hasActiveState,isPersisted,isTemporary,originalContentKind(unavailable | empty | null | object | other | unparseable), andisActivewhere known. No names, paths or content are sent. Reports are deduplicated per call site and workflow.Known tradeoffs:
assertthrows in DEV builds, so in dev these guards surface the problem immediately instead of recovering, same as the existing inactive-tracker assertion. Production reports and recovers.activateLoadedWorkflowimmediately resets the tracker with the graph just loaded onto the canvas, so the canvas state is what is kept.Verification
workflowStore.test.ts,workflowService.test.ts,changeTracker.test.tsandcomfyWorkflow.test.ts. With the source changes stashed, 16 of the new tests fail.pnpm exec vitest run src/platform/workflow src/scripts/changeTracker.test.ts(all pass),pnpm typecheck(clean), oxlint (including type-aware), eslint and oxfmt on the changed files (clean), and the pre-commit hook.Workflow content is not a JSON object,Workflow saved without an active state,openWorkflow called on an active workflow that is not loaded,Activated workflow has no initial state,Activated workflow has no change tracker,Active workflow has no change tracker before loading a new graph,ChangeTracker has no initial state,ChangeTracker.graphEqual() called with a missing state) and theirmethodandoriginalContentKindcontext. Which one fires first, and with which flags, identifies the trigger.Generated by Claude Code
Update 2026-10-21: root cause confirmed, CI green at
a324947null. Details and production evidence: comment.aaec0a4adds a black-box spec,browser_tests/tests/workflowNullStoredContent.spec.ts, which fails on the parent and passes here.ad78d13adds the fix for the verified writer: closing a tab now clears the modified mark.a3249478867b383f3b04e5ca6d902bf868b307cb, which already contains currentmain. The new spec ran and passed in CI. The "Not run" line above is out of date.Full context for agent readers
a324947mergesmainand changes nothing else. Its tree equalsgit merge-treeofad78d13andmain@606fbc634c.a324947:test,lint-and-format,e2e-status,website-e2eandcla-assistantsucceeded. Playwright: 2996 passed, 0 failed. Unit: 8 of 8 shards passed.playwright-video-new-testsran the new spec: 1 passed in 17.4 s.agentReplyDrafts,modelImportProgressDialoganddynamicGroup. None of them touches this diff.updateModified.comfyWorkflow.tswithout coverage.codecov/patchstill passes.nullis never overwritten and never silently replaced.