Repository navigation
fix: render widgets on insert_workflow-materialized nodes - #18598
Conversation
Fixes two compounding bugs in the agent's insert_workflow materialization path (loading a template such as Z Image Turbo) that left every inserted node's widgets blank, despite the values being present in the CRDT document and the widget store, while positions/types/links (PM-1562) materialized correctly. 1. comfy-multi-player's insert_workflow applier remaps every inserted node's id to a derived string with colons unrelated to subgraph scoping (remap.ts's derivedId, e.g. insert:<opId>:root:node:<originalId>). Widget registration always keys on that full id, but stripGraphPrefix (src/stores/widgetValueStore.ts) unconditionally stripped everything up to the LAST colon, collapsing the id down to its trailing segment before every widget-id lookup in the node render path (LGraphNode.vue, processedWidgetRenderModel.ts, WidgetItem.vue, WidgetTextPreview.vue). Nothing was ever registered under that collapsed id, so the lookup came back empty and the node rendered with zero widget rows. Fixed by only stripping a prefix that is actually a subgraph UUID (repeated for nested subgraphs), leaving any other colon-bearing id untouched. 2. Even once the widget is discoverable, agentNodeMaterializer.ts's materialize() called node.configure() without enabling LiteGraph.namedValuesRestore. The op layer stores an ordinary node's widgets_values by name (readSemanticNode), but named restore is gated behind that flag (off by default) or a fallbackNames list that is never populated for real catalog nodes; without it, configure() falls back to positional restore against widgets_values, which for a named payload is a plain object, not an array, so nothing restores and the widget keeps its constructor default. This file already had a withNamedValuesRestore() helper for subgraph definition configuration; materialize() just never used it for the node's own configure() call. Added a regression test that drives the real pipeline (comfy-multi-player's insert_workflow applier -> EcsFollowerAdapter / AgentCrdtProjection -> graphMutations -> agentNodeMaterializer) and reads widget ids back the same way LGraphNode.vue does, plus focused unit tests for stripGraphPrefix covering the derived-id and same-trailing-segment-collision cases.
|
Important Review skippedBot user detected. To trigger a single review, invoke the ⚙️ Run configurationConfiguration used: Repository: Comfy-Org/ComfyUI_frontend/.coderabbit.yaml Review profile: ASSERTIVE Plan: Advanced Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Comment |
🎨 Storybook: ✅ Built — View Storybook🎭 Playwright: ✅ 2121 passed, 0 failed · 1 flaky📊 Browser Reports
🌾 Fallow: ✅ PassedFallow audits the diff for duplication, complexity and dead code. It gates on new findings only, so when the audit completes, a red is something this PR introduced or moved — never inherited backlog. A red can also mean the audit itself did not finish; the details below say which happened. No new findings in the changed files. Reproduce locally# the base the gate compared against
npx fallow audit --root . --quiet --format json \
--changed-since 4cf6e2d9a47f0e74527ad10b36b7827b24843c98
Full docs: ADR-DEVEX-LINT-0015 📦 Bundle: 9.98 MB gzip 🔴 +285 kBDetailsSummary
Category Glance App Entry Points — 3.87 kB (baseline 3.73 kB) • 🔴 +138 BMain entry bundles and manifests
Status: 1 added / 1 removed Graph Workspace — 1.44 MB (baseline 1.43 MB) • 🔴 +17.5 kBGraph editor runtime, canvas, workflow orchestration
Status: 2 added / 2 removed / 1 unchanged Views & Navigation — 132 kB (baseline 124 kB) • 🔴 +8.02 kBTop-level views, pages, and routed surfaces
Status: 17 added / 16 removed / 1 unchanged Panels & Settings — 639 kB (baseline 621 kB) • 🔴 +17.6 kBConfiguration panels, inspectors, and settings screens
Status: 28 added / 27 removed User & Accounts — 37.9 kB (baseline 27.8 kB) • 🔴 +10.1 kBAuthentication, profile, and account management bundles
Status: 9 added / 9 removed / 2 unchanged Editors & Dialogs — 126 kB (baseline 126 kB) • 🔴 +740 BModals, dialogs, drawers, and in-app editors
Status: 8 added / 8 removed UI Components — 124 kB (baseline 117 kB) • 🔴 +6.54 kBReusable component library chunks
Status: 16 added / 15 removed Data & Services — 3.95 MB (baseline 3.6 MB) • 🔴 +345 kBStores, services, APIs, and repositories
Status: 16 added / 16 removed Utilities & Hooks — 577 kB (baseline 559 kB) • 🔴 +18.8 kBHelpers, composables, and utility bundles
Status: 31 added / 29 removed / 7 unchanged Vendor & Third-Party — 17.4 MB (baseline 18.1 MB) • 🟢 -692 kBExternal libraries and shared vendor chunks
Status: 12 added / 12 removed / 6 unchanged Other — 17.8 MB (baseline 16.4 MB) • 🔴 +1.43 MBBundles that do not match a named category
Status: 179 added / 173 removed / 123 unchanged ⚡ Performance Report
Show regressions
All metrics
Historical variance (last 15 runs)
Trend (last 15 commits on main)
Raw data{
"timestamp": "2026-09-23T11:21:57.149Z",
"gitSha": "3dcadf989b52cf4f5a844f9e5dd791e26515fa64",
"branch": "claude/fix-insert-workflow-empty-widgets-4cf6e2d9a4",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 2008.0059999999662,
"styleRecalcs": 10,
"styleRecalcDurationMs": 7.0329999999999995,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 380.209,
"heapDeltaBytes": -7477444,
"heapUsedBytes": 60253684,
"domNodes": -260,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 7.388999999999999,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-idle",
"durationMs": 2009.085999999911,
"styleRecalcs": 11,
"styleRecalcDurationMs": 8.049000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 373.01000000000005,
"heapDeltaBytes": -8341620,
"heapUsedBytes": 59355136,
"domNodes": -259,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 8.052,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1694.2469999999616,
"styleRecalcs": 47,
"styleRecalcDurationMs": 13.811,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 659.708,
"heapDeltaBytes": -4887420,
"heapUsedBytes": 62766212,
"domNodes": -258,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 119.46,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1694.5289999999886,
"styleRecalcs": 46,
"styleRecalcDurationMs": 13.249,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 582.413,
"heapDeltaBytes": -1073304,
"heapUsedBytes": 61282372,
"domNodes": -258,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 119.37500000000001,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1702.9449999999997,
"styleRecalcs": 31,
"styleRecalcDurationMs": 13.630999999999998,
"layouts": 6,
"layoutDurationMs": 0.9940000000000001,
"taskDurationMs": 274.785,
"heapDeltaBytes": 3089520,
"heapUsedBytes": 70716216,
"domNodes": 79,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 10.134999999999998,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1692.837000000054,
"styleRecalcs": 30,
"styleRecalcDurationMs": 13.838999999999999,
"layouts": 6,
"layoutDurationMs": 1.0600000000000003,
"taskDurationMs": 275.72200000000004,
"heapDeltaBytes": 3109360,
"heapUsedBytes": 70965916,
"domNodes": 78,
"jsHeapTotalBytes": 4718592,
"scriptDurationMs": 10.513000000000002,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "dom-widget-clipping",
"durationMs": 554.5249999999555,
"styleRecalcs": 14,
"styleRecalcDurationMs": 8.41,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 377.31100000000004,
"heapDeltaBytes": 18641628,
"heapUsedBytes": 86031208,
"domNodes": 23,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 134.955,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 560.6500000000096,
"styleRecalcs": 15,
"styleRecalcDurationMs": 9.394000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 373.86600000000004,
"heapDeltaBytes": 18921824,
"heapUsedBytes": 86364988,
"domNodes": 25,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 134.716,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1190.8200000000306,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 899.375,
"heapDeltaBytes": -17104124,
"heapUsedBytes": 73117948,
"domNodes": -295,
"jsHeapTotalBytes": 20840448,
"scriptDurationMs": 56.09199999999999,
"eventListeners": -217,
"totalBlockingTimeMs": 525,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1048.7229999999954,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 690.0699999999998,
"heapDeltaBytes": 35025548,
"heapUsedBytes": 107089224,
"domNodes": 0,
"jsHeapTotalBytes": 18771968,
"scriptDurationMs": 53.056000000000004,
"eventListeners": 0,
"totalBlockingTimeMs": 441,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 767.3169999999914,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 369.07500000000005,
"heapDeltaBytes": 14763208,
"heapUsedBytes": 90204516,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 113.32800000000002,
"eventListeners": 0,
"totalBlockingTimeMs": 154,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 765.6580000000304,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 367.4309999999998,
"heapDeltaBytes": 14633124,
"heapUsedBytes": 91591908,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 115.629,
"eventListeners": 0,
"totalBlockingTimeMs": 157,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "extension-clean-combined-draw",
"durationMs": 7010.230999999976,
"styleRecalcs": 2,
"styleRecalcDurationMs": 0.8609999999999972,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 6197.267,
"heapDeltaBytes": 11456952,
"heapUsedBytes": 93547932,
"domNodes": -14,
"jsHeapTotalBytes": 5767168,
"scriptDurationMs": 4920.525,
"eventListeners": 0,
"totalBlockingTimeMs": 71,
"frameDurationMs": 18.333333333333332,
"p95FrameDurationMs": 33.30000000000109
},
{
"name": "extension-clean-combined-draw",
"durationMs": 7550.264000000084,
"styleRecalcs": 2,
"styleRecalcDurationMs": 1.684999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 6686.7119999999995,
"heapDeltaBytes": -2721832,
"heapUsedBytes": 74778448,
"domNodes": -271,
"jsHeapTotalBytes": 6807552,
"scriptDurationMs": 5515.8240000000005,
"eventListeners": -174,
"totalBlockingTimeMs": 55,
"frameDurationMs": 17.779999999999987,
"p95FrameDurationMs": 33.29999999999927
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 8407.695999999987,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7553.1759999999995,
"heapDeltaBytes": -8030348,
"heapUsedBytes": 73880680,
"domNodes": -18,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 5965.9839999999995,
"eventListeners": 0,
"totalBlockingTimeMs": 98,
"frameDurationMs": 17.776666666666703,
"p95FrameDurationMs": 33.19999999999891
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 8481.220000000008,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7562.633999999999,
"heapDeltaBytes": -11917144,
"heapUsedBytes": 70175040,
"domNodes": -14,
"jsHeapTotalBytes": 7864320,
"scriptDurationMs": 5966.4310000000005,
"eventListeners": 0,
"totalBlockingTimeMs": 56,
"frameDurationMs": 18.333333333333332,
"p95FrameDurationMs": 33.29999999999927
},
{
"name": "large-graph-idle",
"durationMs": 2013.4279999999762,
"styleRecalcs": 10,
"styleRecalcDurationMs": 6.831,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 620.5569999999999,
"heapDeltaBytes": -3450196,
"heapUsedBytes": 83416160,
"domNodes": -234,
"jsHeapTotalBytes": -495616,
"scriptDurationMs": 34.804,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-idle",
"durationMs": 2013.7280000000146,
"styleRecalcs": 7,
"styleRecalcDurationMs": 6.721000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 666.7919999999999,
"heapDeltaBytes": -3595368,
"heapUsedBytes": 83028256,
"domNodes": -244,
"jsHeapTotalBytes": -1019904,
"scriptDurationMs": 37.993,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-pan",
"durationMs": 2494.934999999998,
"styleRecalcs": 68,
"styleRecalcDurationMs": 13.239999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1537.4280000000003,
"heapDeltaBytes": -7639732,
"heapUsedBytes": 80604584,
"domNodes": -245,
"jsHeapTotalBytes": 778240,
"scriptDurationMs": 728.915,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-pan",
"durationMs": 2352.2389999999405,
"styleRecalcs": 67,
"styleRecalcDurationMs": 13.003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1541.079,
"heapDeltaBytes": 19493684,
"heapUsedBytes": 84032140,
"domNodes": -248,
"jsHeapTotalBytes": 225280,
"scriptDurationMs": 748.5600000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-zoom",
"durationMs": 3186.882999999966,
"styleRecalcs": 64,
"styleRecalcDurationMs": 12.303999999999998,
"layouts": 60,
"layoutDurationMs": 12.773,
"taskDurationMs": 1378.742,
"heapDeltaBytes": 8442292,
"heapUsedBytes": 73992688,
"domNodes": 10,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 630.9659999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-zoom",
"durationMs": 3245.6959999999526,
"styleRecalcs": 66,
"styleRecalcDurationMs": 13.915000000000001,
"layouts": 60,
"layoutDurationMs": 12.413,
"taskDurationMs": 1486.722,
"heapDeltaBytes": -14832356,
"heapUsedBytes": 75044376,
"domNodes": -249,
"jsHeapTotalBytes": -233472,
"scriptDurationMs": 672.452,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66999999999998,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "legacy-node-drag",
"durationMs": 2796.5350000000058,
"styleRecalcs": 44,
"styleRecalcDurationMs": 8.651,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1844.8600000000001,
"heapDeltaBytes": 17239728,
"heapUsedBytes": 83303436,
"domNodes": 8,
"jsHeapTotalBytes": 417792,
"scriptDurationMs": 940.9879999999999,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "legacy-node-drag",
"durationMs": 2359.831999999983,
"styleRecalcs": 46,
"styleRecalcDurationMs": 9.238999999999997,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1655.952,
"heapDeltaBytes": -7705204,
"heapUsedBytes": 83073612,
"domNodes": -252,
"jsHeapTotalBytes": 970752,
"scriptDurationMs": 811.615,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "minimap-idle",
"durationMs": 2016.6489999999726,
"styleRecalcs": 9,
"styleRecalcDurationMs": 7.8790000000000004,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 480.775,
"heapDeltaBytes": -5709224,
"heapUsedBytes": 83478976,
"domNodes": -240,
"jsHeapTotalBytes": -495616,
"scriptDurationMs": 32.833,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2016.6719999999714,
"styleRecalcs": 7,
"styleRecalcDurationMs": 6.272000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 613.699,
"heapDeltaBytes": 20605292,
"heapUsedBytes": 84564728,
"domNodes": -244,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 35.732,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 460.7949999999619,
"styleRecalcs": 49,
"styleRecalcDurationMs": 11.134,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 301.148,
"heapDeltaBytes": 13685716,
"heapUsedBytes": 84047432,
"domNodes": 24,
"jsHeapTotalBytes": 4718592,
"scriptDurationMs": 120.223,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.799999999999727
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 509.6680000000333,
"styleRecalcs": 49,
"styleRecalcDurationMs": 11.216,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 327.86499999999995,
"heapDeltaBytes": 14073968,
"heapUsedBytes": 83912752,
"domNodes": 24,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 120.329,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.663333333333338,
"p95FrameDurationMs": 16.799999999999727
},
{
"name": "subgraph-idle",
"durationMs": 1998.4989999999812,
"styleRecalcs": 12,
"styleRecalcDurationMs": 8.319999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 376.40399999999994,
"heapDeltaBytes": -2398072,
"heapUsedBytes": 66863484,
"domNodes": -273,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 6.979000000000003,
"eventListeners": -220,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-idle",
"durationMs": 2006.4780000000155,
"styleRecalcs": 11,
"styleRecalcDurationMs": 7.998999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 388.868,
"heapDeltaBytes": -3787664,
"heapUsedBytes": 59079672,
"domNodes": -275,
"jsHeapTotalBytes": 4710400,
"scriptDurationMs": 8.227,
"eventListeners": -220,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1693.0479999999761,
"styleRecalcs": 46,
"styleRecalcDurationMs": 14.533,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 615.733,
"heapDeltaBytes": -7913040,
"heapUsedBytes": 61880820,
"domNodes": -274,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 85.92899999999999,
"eventListeners": -220,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1739.375999999993,
"styleRecalcs": 46,
"styleRecalcDurationMs": 15.373,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 693.37,
"heapDeltaBytes": -6965040,
"heapUsedBytes": 62971528,
"domNodes": -275,
"jsHeapTotalBytes": 6545408,
"scriptDurationMs": 98.813,
"eventListeners": -220,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-transition-enter",
"durationMs": 867.4839999999904,
"styleRecalcs": 18,
"styleRecalcDurationMs": 22.029,
"layouts": 5,
"layoutDurationMs": 9.389,
"taskDurationMs": 695.1719999999999,
"heapDeltaBytes": -6271888,
"heapUsedBytes": 81437272,
"domNodes": 13667,
"jsHeapTotalBytes": 11010048,
"scriptDurationMs": 22.219999999999995,
"eventListeners": 2383,
"totalBlockingTimeMs": 84,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "viewport-pan-sweep",
"durationMs": 8226.386999999988,
"styleRecalcs": 250,
"styleRecalcDurationMs": 37.684000000000005,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 4614.643,
"heapDeltaBytes": -21383356,
"heapUsedBytes": 66059888,
"domNodes": -214,
"jsHeapTotalBytes": 3960832,
"scriptDurationMs": 2199.226,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "viewport-pan-sweep",
"durationMs": 8230.564999999955,
"styleRecalcs": 250,
"styleRecalcDurationMs": 37.319,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 4689.635,
"heapDeltaBytes": -6164448,
"heapUsedBytes": 80195052,
"domNodes": -236,
"jsHeapTotalBytes": 1826816,
"scriptDurationMs": 2208.191,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-idle",
"durationMs": 15280.324000000008,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 15165.552000000001,
"heapDeltaBytes": -21667820,
"heapUsedBytes": 191531184,
"domNodes": -8062,
"jsHeapTotalBytes": -11780096,
"scriptDurationMs": 238.419,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-idle",
"durationMs": 14785.28300000005,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 14594.497000000001,
"heapDeltaBytes": -58969388,
"heapUsedBytes": 175854876,
"domNodes": -8061,
"jsHeapTotalBytes": -21630976,
"scriptDurationMs": 233.02500000000003,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-pan",
"durationMs": 17335.27700000002,
"styleRecalcs": 125,
"styleRecalcDurationMs": 14.603000000000005,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 17107.968,
"heapDeltaBytes": -66207012,
"heapUsedBytes": 180382124,
"domNodes": -8062,
"jsHeapTotalBytes": -21446656,
"scriptDurationMs": 697.736,
"eventListeners": -14928,
"totalBlockingTimeMs": 16,
"frameDurationMs": 17.220000000000073,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-pan",
"durationMs": 17131.354999999985,
"styleRecalcs": 125,
"styleRecalcDurationMs": 14.592999999999995,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 16998.852,
"heapDeltaBytes": -34296108,
"heapUsedBytes": 192902492,
"domNodes": -8062,
"jsHeapTotalBytes": -20422656,
"scriptDurationMs": 716.892,
"eventListeners": -14928,
"totalBlockingTimeMs": 7,
"frameDurationMs": 17.223333333333237,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-node-resize-workload",
"durationMs": 5479.497999999978,
"styleRecalcs": 20,
"styleRecalcDurationMs": 62.597999999999985,
"layouts": 20,
"layoutDurationMs": 26.469000000000005,
"taskDurationMs": 5464.518999999999,
"heapDeltaBytes": -10529340,
"heapUsedBytes": 177922252,
"domNodes": -8062,
"jsHeapTotalBytes": 17563648,
"scriptDurationMs": 101.758,
"eventListeners": -14928,
"totalBlockingTimeMs": 278,
"frameDurationMs": 17.223333333333297,
"p95FrameDurationMs": 16.80000000000109
},
{
"name": "vue-node-resize-workload",
"durationMs": 5648.680000000013,
"styleRecalcs": 20,
"styleRecalcDurationMs": 65.08100000000003,
"layouts": 20,
"layoutDurationMs": 27.467999999999993,
"taskDurationMs": 5618.339000000001,
"heapDeltaBytes": -40146932,
"heapUsedBytes": 196979556,
"domNodes": -8061,
"jsHeapTotalBytes": 19025920,
"scriptDurationMs": 99.62,
"eventListeners": -14928,
"totalBlockingTimeMs": 295,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-zoom-out-idle",
"durationMs": 7119.160999999963,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 6995.16,
"heapDeltaBytes": -12590764,
"heapUsedBytes": 178149500,
"domNodes": 0,
"jsHeapTotalBytes": -24117248,
"scriptDurationMs": 101.95899999999996,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.223333333333358,
"p95FrameDurationMs": 16.80000000000291
},
{
"name": "vue-zoom-out-idle",
"durationMs": 7150.26499999999,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7015.713000000002,
"heapDeltaBytes": 132000,
"heapUsedBytes": 170753712,
"domNodes": 0,
"jsHeapTotalBytes": -17563648,
"scriptDurationMs": 104.10599999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.219999999999953,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "workflow-execution",
"durationMs": 91.63100000000668,
"styleRecalcs": 6,
"styleRecalcDurationMs": 12.337,
"layouts": 3,
"layoutDurationMs": 1.398,
"taskDurationMs": 62.160999999999994,
"heapDeltaBytes": 3123788,
"heapUsedBytes": 69075872,
"domNodes": 123,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 6.486000000000002,
"eventListeners": 17,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999727
},
{
"name": "workflow-execution",
"durationMs": 86.22300000001815,
"styleRecalcs": 6,
"styleRecalcDurationMs": 12.915000000000001,
"layouts": 3,
"layoutDurationMs": 1.4460000000000002,
"taskDurationMs": 60.09700000000001,
"heapDeltaBytes": 3045688,
"heapUsedBytes": 67815560,
"domNodes": 104,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 5.511,
"eventListeners": 33,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000273
}
]
}🔬 E2E Coverage
Top 10 files by uncovered lines
🧩 Custom Nodes Ecosystem Matrix: ✅ PassedExecutes the frontend JS of every registry pack that ships any, against this change, in both renderers. The eight
On a pin-bump PR, a red gate means the ecosystem moved — not that your diff broke something. That is what the bump PR is for. Outside a pin bump or a cold rebuild of the unpinned tail, a red means the diff. Note the corpus size and the executed count are different numbers, and the verdict is against the executed count: roughly 3,100 of ~5,100 packs ship no extension-shaped JS and are skipped. Reproduce locally# build one shard's fixture, then run it
python3 scripts/registry-census/build_matrix.py --shard 1/4
MATRIX_RENDERER=vue MATRIX_VUE=1 \
npx vitest run --config vitest.matrix.config.mts
Full docs: scripts/registry-census/README.md |
christian-byrne
left a comment
There was a problem hiding this comment.
inspected exact head 18d9c30097f3: all four changed files and the complete review-thread set. The UUID-gated stripGraphPrefix preserves insert_workflow derived IDs while retaining nested subgraph unscoping, and withNamedValuesRestore restores the named widget payload with its existing try/finally state reset. The regression coverage exercises both discoverability and value restoration through the real follower/materialization path, including same-original-ID isolation. I found no additional blocking issue in this pass; hosted checks are still running.
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## main #18598 +/- ##
==========================================
+ Coverage 83.58% 84.55% +0.97%
==========================================
Files 2999 2293 -706
Lines 148394 131067 -17327
Branches 38321 33466 -4855
==========================================
- Hits 124029 110821 -13208
+ Misses 23071 19945 -3126
+ Partials 1294 301 -993
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 713 files with indirect coverage changes 🚀 New features to boost your workflow:
|
There was a problem hiding this comment.
🔍 Cursor Review — Consolidated panel
Triggered by @christian-byrne.
Found 7 finding(s).
| Severity | Count |
|---|---|
| 🟠 High | 1 |
| 🟡 Medium | 3 |
| 🟢 Low | 2 |
| ⚪ Nit | 1 |
Panel: 8/8 reviewers contributed findings.
| rest = rest.slice(separatorIndex + 1) | ||
| separatorIndex = rest.indexOf(':') | ||
| } | ||
| return parseNodeId(rest) |
There was a problem hiding this comment.
🟠 High — Returning the colon-bearing id fixes the store lookup, but computeProcessedWidgets derives nodeExecId from createNodeExecutionId([state.id]) and requireNodeIdSegment rejects any id containing ':', so both the isGraphReady and fallback branches yield null and the function early-returns [] — the node still renders zero widget rows even though widgetIds is now populated. The new regression test only replays the store lookup, not useProcessedWidgets, so it goes green either way; confirm the rendered path and extend coverage through computeProcessedWidgets. Raised by 1 of 8 reviewers (claude-opus-5-thinking-max edge-case).
There was a problem hiding this comment.
Tracked as FE-2847. The PR merged with this thread unresolved, so the finding is carried in Linear rather than lost here. Filed automatically; close the ticket if the finding is no longer true at main.
| let separatorIndex = rest.indexOf(':') | ||
| while ( | ||
| separatorIndex !== -1 && | ||
| isUuidShapedSubgraphId(rest.slice(0, separatorIndex)) |
There was a problem hiding this comment.
🟡 Medium — A UUID-shaped first segment does not prove the segment is a subgraph scope: NodeId accepts any non-empty string and registration (attachNodeToStores / BaseWidget.setNodeId) always keys on the raw id, so a node whose own id begins with <uuid>: still strips to its trailing segment and reproduces the PM-1580 empty-lookup symptom. Worse, the collapse aliases distinct ids in one root graph — <uuidA>:5, <uuidB>:5 and a plain 5 all resolve to 5, letting one node render and overwrite another's widget values. All four call sites pass raw unscoped ids, so scope should come from graph context (or the strip be limited to callers known to hold a NodeLocatorId) rather than from string shape. Raised by 6 of 8 reviewers (gpt-5.6-sol-max adversarial, gpt-5.6-sol-max edge-case, claude-opus-5-thinking-max adversarial, gemini-3.1-pro adversarial, kimi-k3-max adversarial, kimi-k3-max edge-case).
| try { | ||
| const savedInputs = serialised.inputs?.map((input) => ({ ...input })) | ||
| node.configure(withNamedWidgetValues(serialised, widgets)) | ||
| withNamedValuesRestore(() => |
There was a problem hiding this comment.
🟡 Medium — Forcing LiteGraph.namedValuesRestore on around configure() also flips SubgraphNode.configure's guard if (!info.widgets_values_named || !LiteGraph.namedValuesRestore) this._applyPromotedWidgetValues(info.widgets_values); because withNamedWidgetValues always sets widgets_values_named for a name-keyed payload, _applyPromotedWidgetValues is now skipped for every agent-materialized subgraph instance. That is the only path that re-applies proxyWidgetErrorQuarantine host values and the positional restore for promoted inputs, and super.configure's named restore does not cover them, so those values are silently dropped. Raised by 3 of 8 reviewers (claude-opus-5-thinking-max adversarial, claude-opus-5-thinking-max edge-case, kimi-k3-max edge-case).
There was a problem hiding this comment.
Tracked as FE-2847. The PR merged with this thread unresolved, so the finding is carried in Linear rather than lost here. Filed automatically; close the ticket if the finding is no longer true at main.
| expect(node.widgets?.[0]?.value).toBe('my_custom_prefix') | ||
| }) | ||
|
|
||
| it("two inserted nodes sharing an original id do not steal each other's widgets", () => { |
There was a problem hiding this comment.
🟡 Medium — This test cannot observe the collision it names: graphA and graphB are separate LGraphs, so the widget store's root-graph key already partitions the two nodes and a regression reintroducing the old unconditional strip would keep it green. Insert both nodes into a single graph so two derived ids sharing the trailing :9 segment actually compete for the same store key. Raised by 2 of 8 reviewers (claude-opus-5-thinking-max adversarial, claude-opus-5-thinking-max edge-case).
There was a problem hiding this comment.
Tracked as FE-2847. The PR merged with this thread unresolved, so the finding is carried in Linear rather than lost here. Filed automatically; close the ticket if the finding is no longer true at main.
| * widgets). Stopping as soon as the next segment fails the UUID check keeps | ||
| * the rest of a non-scoped id intact. | ||
| */ | ||
| export function stripGraphPrefix(scopedId: SerializedNodeId): NodeId | null { |
There was a problem hiding this comment.
🟢 Low — For agent-inserted nodes this now returns a colon-bearing id, which createNodeLocatorId rejects, so widgetNodeLocatorId produces null at runtime even though its public overload is declared non-nullable and SimplifiedWidget.nodeLocatorId is typed NodeLocatorId | undefined. Locator-keyed features (favorited widgets, etc.) will silently no-op for those nodes — worth handling explicitly rather than parking a null in a non-nullable slot. Raised by 1 of 8 reviewers (claude-opus-5-thinking-max edge-case).
There was a problem hiding this comment.
Tracked as FE-2847. The PR merged with this thread unresolved, so the finding is carried in Linear rather than lost here. Filed automatically; close the ticket if the finding is no longer true at main.
| @@ -367,7 +373,9 @@ function materialize( | |||
|
|
|||
| try { | |||
| const savedInputs = serialised.inputs?.map((input) => ({ ...input })) | |||
There was a problem hiding this comment.
🟢 Low — The process-wide flag window now spans a full node.configure(), which synchronously fires extension onConfigure hooks and widget callbacks, so any other node configured from inside those callbacks silently inherits named-restore semantics despite the experimental Comfy.Workflow.NamedValuesRestore setting being off — while work those callbacks defer to a microtask sees the flag already restored. Passing the intent through the ISerialisedNode payload or a configure() option would scope it to the node being materialized. Raised by 1 of 8 reviewers (claude-opus-5-thinking-max adversarial).
There was a problem hiding this comment.
Tracked as FE-2847. The PR merged with this thread unresolved, so the finding is carried in Linear rather than lost here. Filed automatically; close the ticket if the finding is no longer true at main.
| return parseNodeId(String(scopedId).replace(/^(.*:)+/, '')) | ||
| let rest = String(scopedId) | ||
| let separatorIndex = rest.indexOf(':') | ||
| while ( |
There was a problem hiding this comment.
⚪ Nit — The loop runs a Zod safeParse per colon-separated segment, and stripGraphPrefix is called from render computeds (LGraphNode.vue's widgetIds, once per widget in processedWidgetRenderModel.ts), so a many-segment id re-parses on every re-render. An anchored regex test plus a tracked start offset (rather than re-slicing each iteration) removes both the parse and the per-segment allocation. Raised by 2 of 8 reviewers (claude-opus-5-thinking-max adversarial, kimi-k3-max adversarial).
There was a problem hiding this comment.
Tracked as FE-2847. The PR merged with this thread unresolved, so the finding is carried in Linear rather than lost here. Filed automatically; close the ticket if the finding is no longer true at main.
…tion fails (PM-1580 follow-up) #18598 fixed stripGraphPrefix's widget-id lookup for comfy-multi-player's insert_workflow remapped ids (insert:<opId>:<scope>:node:<originalId>), but computeProcessedWidgets derives its own nodeExecId separately, via executionIdFromState/createNodeExecutionId. Both reject any node id containing a colon, including the derived id itself, so nodeExecId comes back null for these nodes and computeProcessedWidgets early-returns [] -- the same "zero widget rows" symptom the store-lookup fix addressed, from a different call site the existing regression test never exercised (it only replayed the raw store lookup, not computeProcessedWidgets). Adds createLeafNodeExecutionId, which keeps a root-level node's raw id whole instead of rejecting a colon inside it, and uses it at the two fallback call sites (executionIdFromState, computeProcessedWidgets) that only need a single, ancestor-free segment. Also scopes agentNodeMaterializer's withNamedValuesRestore (added in #18598 to fix ordinary nodes' named widget restore) away from SubgraphNode instances: SubgraphNode.configure() skips _applyPromotedWidgetValues -- the only place a proxyWidgetErrorQuarantine override wins over a stale value -- whenever widgets_values_named is set and namedValuesRestore is on, so forcing the flag for every configure() call would silently drop a quarantined promoted-input value on agent-materialized subgraph instances. Also fixes the insert_workflow regression test's collision case, which put the two colliding-id nodes in separate LGraphs and so never exercised a same-graph collision. Fixes PM-1580.
<!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1790058330774449?thread_ts=1790058330.774449&cid=C0BGH348Z0C)_ Fixes a bug where an agent's `insert_workflow` batch (e.g. the in-app agent loading a template like "Z Image Turbo") materialized nodes with correct positions and types, and links between them materialized correctly (#18458 / PM-1562), but every inserted node's **widgets rendered as empty/blank bodies** even though the widget values, `MarkdownNote` text, `SaveImage` `filename_prefix`, etc. were present in the CRDT document and in the widget store. Two independent, compounding bugs, both specific to how `insert_workflow` differs from `add_node`: **Bug 1 — widgets undiscoverable (blank node body, zero widget rows).** `comfy-multi-player`'s `insert_workflow` applier remaps every inserted node's id to a derived string carrying colons for reasons that have nothing to do with subgraph scoping (`remap.ts`'s `derivedId`, e.g. `insert:<opId>:root:node:<originalId>`). `add_node` never does this — it keeps the id comfy-cli minted. Widget registration (`attachNodeToStores`/`BaseWidget.setNodeId` in `nodeShellLifecycle.ts`/`BaseWidget.ts`) always keys on that full id. The node-render path reads widget ids back through `stripGraphPrefix` (`src/stores/widgetValueStore.ts`) — used by `LGraphNode.vue`, `processedWidgetRenderModel.ts`, `WidgetItem.vue`, and `WidgetTextPreview.vue`. It used to strip *everything up to the last colon* unconditionally, which is correct for a genuine subgraph-scoped id (`<uuid>:<localId>`) but wrong for `insert_workflow`'s derived id: it collapsed `insert:<opId>:root:node:9` down to bare `9`, a key nothing was ever registered under (and, if a second `insert_workflow` op happened to reuse the same original id, a key two unrelated nodes would collapse onto). The lookup came back empty, so the node rendered with **zero widget rows**. Fixed `stripGraphPrefix` to only strip a prefix that is actually a subgraph UUID (repeating for nested subgraphs), leaving any other colon-bearing id untouched. **Bug 2 — widget discoverable but stuck at its constructor default.** Once bug 1 is fixed, `agentNodeMaterializer.ts`'s `materialize()` still didn't restore the real value. The op layer stores an ordinary node's `widgets_values` **by name** (`readSemanticNode`), but `LGraphNode.configure()`'s named-value restore is gated behind the experimental `Comfy.Workflow.NamedValuesRestore` setting (off by default) or a `fallbackNames` list that is never populated for real catalog nodes. Without either, `configure()` falls back to *positional* restore against `widgets_values` — for a named payload that's a plain object, not an array, so nothing restores and the widget keeps whatever value its constructor set. This file already has a `withNamedValuesRestore()` helper that flips `LiteGraph.namedValuesRestore` on for the duration of a call — it was used only when configuring a subgraph *definition*, never for the node's own `configure()` call in `materialize()`. `add_node`-created nodes mostly dodge this because the agent's add-node tool typically follow-ups with discrete `set_widget` ops (a direct, always-correct value write), whereas `insert_workflow` bakes every widget value into the node's initial creation payload, hitting this path on every single widget of every inserted node. **Fix:** wrap `materialize()`'s `node.configure(...)` call in `withNamedValuesRestore()` too, so a named payload actually lands. **Tests:** - `agentInsertWorkflowWidgets.regression.test.ts` (new): drives the real pipeline (comfy-multi-player's `insert_workflow` applier → `EcsFollowerAdapter`/`AgentCrdtProjection` → `graphMutations` → `agentNodeMaterializer`), then reads widget ids back exactly the way `LGraphNode.vue` does (`stripGraphPrefix` + `getNodeWidgetIds`), asserting the widget is both discoverable and correctly-valued. Also covers two `insert_workflow` ops that reuse the same original node id, proving they no longer collide. - `widgetValueStore.test.ts`: focused unit coverage for `stripGraphPrefix` — bare id passthrough, single/nested subgraph-uuid stripping, the derived-id passthrough case, the same-trailing-segment non-collision case, and the empty-id case. **Verified:** `pnpm typecheck`, `pnpm lint` (eslint + oxlint), `pnpm format:check` (via the repo's pre-commit hook) and the full `src/workbench/extensions/agent/crdt` + `src/stores` Vitest suites (2148 passed, 16 pre-existing expected-fails unrelated to this change, 1 skipped) all pass, plus the render-adjacent `LGraphNode.widgetOrder.test.ts`, `WidgetItem.test.ts`, and `WidgetTextPreview.test.ts`. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY --- _Generated by [Claude Code](https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY)_
<!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1790058330774449?thread_ts=1790058330.774449&cid=C0BGH348Z0C)_ Fixes a bug where an agent's `insert_workflow` batch (e.g. the in-app agent loading a template like "Z Image Turbo") materialized nodes with correct positions and types, and links between them materialized correctly (#18458 / PM-1562), but every inserted node's **widgets rendered as empty/blank bodies** even though the widget values, `MarkdownNote` text, `SaveImage` `filename_prefix`, etc. were present in the CRDT document and in the widget store. Two independent, compounding bugs, both specific to how `insert_workflow` differs from `add_node`: **Bug 1 — widgets undiscoverable (blank node body, zero widget rows).** `comfy-multi-player`'s `insert_workflow` applier remaps every inserted node's id to a derived string carrying colons for reasons that have nothing to do with subgraph scoping (`remap.ts`'s `derivedId`, e.g. `insert:<opId>:root:node:<originalId>`). `add_node` never does this — it keeps the id comfy-cli minted. Widget registration (`attachNodeToStores`/`BaseWidget.setNodeId` in `nodeShellLifecycle.ts`/`BaseWidget.ts`) always keys on that full id. The node-render path reads widget ids back through `stripGraphPrefix` (`src/stores/widgetValueStore.ts`) — used by `LGraphNode.vue`, `processedWidgetRenderModel.ts`, `WidgetItem.vue`, and `WidgetTextPreview.vue`. It used to strip *everything up to the last colon* unconditionally, which is correct for a genuine subgraph-scoped id (`<uuid>:<localId>`) but wrong for `insert_workflow`'s derived id: it collapsed `insert:<opId>:root:node:9` down to bare `9`, a key nothing was ever registered under (and, if a second `insert_workflow` op happened to reuse the same original id, a key two unrelated nodes would collapse onto). The lookup came back empty, so the node rendered with **zero widget rows**. Fixed `stripGraphPrefix` to only strip a prefix that is actually a subgraph UUID (repeating for nested subgraphs), leaving any other colon-bearing id untouched. **Bug 2 — widget discoverable but stuck at its constructor default.** Once bug 1 is fixed, `agentNodeMaterializer.ts`'s `materialize()` still didn't restore the real value. The op layer stores an ordinary node's `widgets_values` **by name** (`readSemanticNode`), but `LGraphNode.configure()`'s named-value restore is gated behind the experimental `Comfy.Workflow.NamedValuesRestore` setting (off by default) or a `fallbackNames` list that is never populated for real catalog nodes. Without either, `configure()` falls back to *positional* restore against `widgets_values` — for a named payload that's a plain object, not an array, so nothing restores and the widget keeps whatever value its constructor set. This file already has a `withNamedValuesRestore()` helper that flips `LiteGraph.namedValuesRestore` on for the duration of a call — it was used only when configuring a subgraph *definition*, never for the node's own `configure()` call in `materialize()`. `add_node`-created nodes mostly dodge this because the agent's add-node tool typically follow-ups with discrete `set_widget` ops (a direct, always-correct value write), whereas `insert_workflow` bakes every widget value into the node's initial creation payload, hitting this path on every single widget of every inserted node. **Fix:** wrap `materialize()`'s `node.configure(...)` call in `withNamedValuesRestore()` too, so a named payload actually lands. **Tests:** - `agentInsertWorkflowWidgets.regression.test.ts` (new): drives the real pipeline (comfy-multi-player's `insert_workflow` applier → `EcsFollowerAdapter`/`AgentCrdtProjection` → `graphMutations` → `agentNodeMaterializer`), then reads widget ids back exactly the way `LGraphNode.vue` does (`stripGraphPrefix` + `getNodeWidgetIds`), asserting the widget is both discoverable and correctly-valued. Also covers two `insert_workflow` ops that reuse the same original node id, proving they no longer collide. - `widgetValueStore.test.ts`: focused unit coverage for `stripGraphPrefix` — bare id passthrough, single/nested subgraph-uuid stripping, the derived-id passthrough case, the same-trailing-segment non-collision case, and the empty-id case. **Verified:** `pnpm typecheck`, `pnpm lint` (eslint + oxlint), `pnpm format:check` (via the repo's pre-commit hook) and the full `src/workbench/extensions/agent/crdt` + `src/stores` Vitest suites (2148 passed, 16 pre-existing expected-fails unrelated to this change, 1 skipped) all pass, plus the render-adjacent `LGraphNode.widgetOrder.test.ts`, `WidgetItem.test.ts`, and `WidgetTextPreview.test.ts`. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY --- _Generated by [Claude Code](https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY)_
…s, subgraphs, templates) Extends the PM-1562/#18458 (link materialization) and PM-1580/#18598/#18604 (widget rendering) regression coverage with broader integration tests for insert_workflow across widget types, link topologies, subgraph promotion, and representative workflow-template shapes. - agentInsertWorkflow.widgetTypes.test.ts (13 cases): combo/number/text/ toggle/multiline widgets on inserted nodes, partial widgets_values, cross-graph and same-graph id collisions. - agentInsertWorkflow.linkTopology.test.ts (8 cases): fan-out, fan-in, parallel chains, links crossing into/out of a subgraph boundary, a fully interior subgraph link, batch id-collision, incremental inserts. - agentInsertWorkflow.subgraphs.test.ts (7 cases): a promoted widget surviving a later set_widget, multiple widget types promoted together, subgraph-instance id collisions, several instances of one definition in one op, coexistence with a pre-existing instance, multiple definitions in one op. Documents a real gap (not fixed here): a subgraph instance nested as an interior node of another subgraph does not yet materialize through this pipeline. - agentInsertWorkflow.templates.test.ts (7 cases): representative txt2img, img2img, ControlNet, LoRA, video and compound pipelines, plus the same template inserted twice without cross-wiring.
…tion fails (PM-1580 follow-up) (Comfy-Org#18604) ## Summary Follow-up to Comfy-Org#18598, which fixed `stripGraphPrefix`'s widget-id lookup for comfy-multi-player's `insert_workflow` remapped ids (`insert:<opId>:<scope>:node:<originalId>`) but left two more places rejecting the same colon-bearing ids: 1. **`computeProcessedWidgets` still returns `[]` for these nodes.** It derives its own `nodeExecId` via `executionIdFromState`/`createNodeExecutionId`, both of which reject *any* colon in a node id — including the derived id itself — so `nodeExecId` comes back `null` and the function early-returns `[]` regardless of how correct the widget-id lookup is. This is the actual rendering path (`useProcessedWidgets` → `LGraphNode.vue`), so the "zero widget rows" symptom Comfy-Org#18598 aimed to fix was not actually fixed. Comfy-Org#18598's regression test only replayed the raw store lookup (`stripGraphPrefix` + `getNodeWidgetIds`), never `computeProcessedWidgets`, so it went green either way. Fix: added `createLeafNodeExecutionId`, which keeps a root-level node's raw id whole instead of rejecting a colon inside it (there's no ancestor segment to disambiguate it from), and used it at the two fallback call sites in `executionIdFromState` and `computeProcessedWidgets`. 2. **`agentNodeMaterializer`'s new `withNamedValuesRestore` wrap around every `node.configure()` call drops quarantined promoted-widget values on subgraph instances.** `SubgraphNode.configure()` skips `_applyPromotedWidgetValues` — the only place a `proxyWidgetErrorQuarantine` override wins over a stale value — whenever `widgets_values_named` is set **and** `namedValuesRestore` is on. Comfy-Org#18598 started forcing that flag for every materialized node's `configure()` call (previously only subgraph *definitions* used it), so an agent-materialized subgraph *instance* carrying a quarantined value now silently loses it to the stale value instead. Fix: scoped the forced flag away from `SubgraphNode` instances, which don't need it (their promoted values restore through `_applyPromotedWidgetValues`, not the named-values path). 3. **The new "collision" regression test never exercised a same-graph collision.** It put the two colliding-id nodes in separate `LGraph`s, so even a `stripGraphPrefix` that collapsed both to their trailing segment (the pre-Comfy-Org#18598 bug) would still pass, since the per-graph maps are keyed by graph id first. Changed it to insert both nodes into the same graph. Found while verifying an automated review panel's findings on Comfy-Org#18598 after it was merged via the merge queue; all three were confirmed against the current code with reproducing tests before this fix, and each new assertion fails without the fix. ## Test plan - [x] `createLeafNodeExecutionId`/`executionIdFromState` regression tests in `src/types/nodeIdentification.test.ts` and `src/utils/graphTraversalUtil.test.ts` - [x] Same-graph collision case in `agentInsertWorkflowWidgets.regression.test.ts` - [x] Confirmed each new assertion fails on `main` (pre-fix) and passes with the fix - [x] `pnpm vitest run` on all touched/related test files (340 passed) - [x] `vue-tsc --noEmit` clean - [x] pre-commit hooks (oxfmt/oxlint/eslint/typecheck) passed 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY --- _Generated by [Claude Code](https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY)_ Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
…ets when execId derivation fails (PM-1580 follow-up) (#18608) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1790058330774449?thread_ts=1790058330.774449&cid=C0BGH348Z0C)_ Backport of #18604 to `cloud/1.54`. Opened manually: the `PR Backport` workflow (`pr-backport.yaml`) only runs when the PR carries the `needs-backport` label — `core/1.54` + `cloud/1.54` alone (added manually before merge) don't trigger it, so it never ran at merge time. After adding `needs-backport`, the workflow did run and this target (`cloud/1.54`) cherry-picked `035d9056b0` cleanly, but the run's overall job still failed because the `core/1.54` target hit a real conflict (that target is missing #18598 entirely, since its own backport #18600 is still open/unmerged against `core/1.54`) — and the workflow's "Cleanup stranded backport branches" step then deleted the `cloud/1.54` branch it had just successfully pushed, because it runs on `failure()` for the *job*, not per-target. So this clean backport never got a PR from the bot despite succeeding. Re-cherry-picked `035d9056b0` onto `cloud/1.54` (which already has #18598 via #18601) by hand — clean apply, no conflicts. `core/1.54` is intentionally not backported here: it needs #18600 (the #18598 backport) merged first. --- 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY --- _Generated by [Claude Code](https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY)_ Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
…s, subgraphs, templates) (Comfy-Org#18607) ## Before/After **Before:** `insert_workflow` regression coverage was two narrow pins — one for the PM-1562 link-materialization bug (`agentInsertWorkflowLinks.regression.test.ts`) and one for the PM-1580 widget-rendering bug (`agentInsertWorkflowWidgets.regression.test.ts`), plus a single subgraph-blueprint-paste test. Real-world `insert_workflow` usage (agent-driven template inserts, pastes, multi-node/subgraph batches) had no broader integration coverage: no check that every widget *type* renders correctly, no coverage of link topologies beyond the exact shapes those two bugs hit, and no coverage of subgraph promotion beyond one happy path. **After:** 35 new integration test cases across 4 files, all driving the real pipeline (comfy-multi-player's `insert_workflow` applier → `AgentCrdtProjection`/`EcsFollowerAdapter` → `graphMutations` → `agentNodeMaterializer`), same technique as the existing regression files (either literal `WorkflowJSON` ops, or a real litegraph graph built with `createTestSubgraph`/`createTestSubgraphNode` and round-tripped through `serialize()`): - **`agentInsertWorkflow.widgetTypes.test.ts`** (13 cases): combo, number, text, toggle and multiline widgets on inserted nodes; a combo value outside its declared options; a multi-widget node rendering in declared order; a node whose `widgets_values` is shorter than its widget list; two same-original-id nodes inserted into **different** graphs; three same-original-id nodes inserted into the **same** graph (extends the existing 2-way collision test). - **`agentInsertWorkflow.linkTopology.test.ts`** (8 cases): one output fanning out to two inputs; three independent chains in one batch; a link crossing **into** a freshly-inserted subgraph's promoted input; a link crossing **out of** one; a fully **interior** subgraph link; two batch-inserted ops that collide on node ids without cross-wiring their links; a two-source fan-in; an incrementally-inserted link on an already-open graph. - **`agentInsertWorkflow.subgraphs.test.ts`** (7 cases): a promoted widget materialized via `insert_workflow` surviving a later `set_widget` op; two different widget types (combo + number) promoted together on one host; two subgraph instances colliding on their original host id in the same graph; several instances of the *same* definition inserted in one op; a new subgraph insert coexisting with a pre-existing, unrelated instance; two distinct subgraph definitions carried by one op. - **`agentInsertWorkflow.templates.test.ts`** (7 cases): representative checkpoint txt2img, img2img, ControlNet, LoRA, video and compound (ControlNet+LoRA) pipelines, plus the same template inserted twice with no cross-wiring. This repo vendors no local fixture set for the GCS-hosted templates catalogue, so these are literal graphs built to the same node/widget shapes as the well-known public templates rather than pulled fixtures. **Real gap found, not fixed here:** while writing the subgraph tests I found that an interior node which is itself a subgraph instance (a subgraph nested inside another subgraph, as opposed to two sibling definitions in one op) does not currently materialize correctly through `insert_workflow` — its `type` still names the un-remapped blueprint definition id, which nothing registers on the live side, so it falls back to a plain, widget-less node. That's called out in a comment at the top of `agentInsertWorkflow.subgraphs.test.ts` rather than asserted around or silently skipped. Worth its own follow-up ticket. ## How - All four files reuse the same harness pattern as the existing regression files (`bindProjection`/`insertOp` helpers, `LiteGraph.registerNodeType` fixture classes, `useWidgetValueStore`/`stripGraphPrefix` for reading rendered widgets the way `LGraphNode.vue` does). - `pnpm vitest run` on the four new files: **35/35 passing**. - `pnpm vitest run src/workbench/extensions/agent/crdt/`: all 64 files in the directory pass (925 passed, 15 pre-existing expected-fails), confirming no regression. - `pnpm typecheck:app`: clean. - `pnpm exec oxlint --type-aware` / `eslint` on the new files: clean. - Branched from current `main` (`a937e31e`, includes Comfy-Org#18598's widget-rendering fix). Comfy-Org#18604 (the PM-1580 follow-up for colon-bearing root-level ids) is still open as of this PR and not included; nothing in this suite specifically re-pins that narrow fallback path, since its own regression test already lives in Comfy-Org#18604. <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · Slack thread (channel C0BGH348Z0C, thread 1790058330.774449)_ 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY --- _Generated by [Claude Code](https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY)_ --------- Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>
…ets when execId derivation fails (PM-1580 follow-up) (#18608) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1790058330774449?thread_ts=1790058330.774449&cid=C0BGH348Z0C)_ Backport of #18604 to `cloud/1.54`. Opened manually: the `PR Backport` workflow (`pr-backport.yaml`) only runs when the PR carries the `needs-backport` label — `core/1.54` + `cloud/1.54` alone (added manually before merge) don't trigger it, so it never ran at merge time. After adding `needs-backport`, the workflow did run and this target (`cloud/1.54`) cherry-picked `035d9056b0` cleanly, but the run's overall job still failed because the `core/1.54` target hit a real conflict (that target is missing #18598 entirely, since its own backport #18600 is still open/unmerged against `core/1.54`) — and the workflow's "Cleanup stranded backport branches" step then deleted the `cloud/1.54` branch it had just successfully pushed, because it runs on `failure()` for the *job*, not per-target. So this clean backport never got a PR from the bot despite succeeding. Re-cherry-picked `035d9056b0` onto `cloud/1.54` (which already has #18598 via #18601) by hand — clean apply, no conflicts. `core/1.54` is intentionally not backported here: it needs #18600 (the #18598 backport) merged first. --- 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY --- _Generated by [Claude Code](https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY)_ (cherry picked from commit a55ce3d)
<!-- ccr-slack-attribution -->
Requested by Christian Byrne · Slack thread
Fixes a bug where an agent's
insert_workflowbatch (e.g. the in-app agent loading a template like "Z Image Turbo") materialized nodes with correct positions and types, and links between them materialized correctly (#18458 / PM-1562), but every inserted node's widgets rendered as empty/blank bodies even though the widget values,MarkdownNotetext,SaveImagefilename_prefix, etc. were present in the CRDT document and in the widget store.Two independent, compounding bugs, both specific to how
insert_workflowdiffers fromadd_node:Bug 1 — widgets undiscoverable (blank node body, zero widget rows).
comfy-multi-player'sinsert_workflowapplier remaps every inserted node's id to a derived string carrying colons for reasons that have nothing to do with subgraph scoping (remap.ts'sderivedId, e.g.insert:<opId>:root:node:<originalId>).add_nodenever does this — it keeps the id comfy-cli minted. Widget registration (attachNodeToStores/BaseWidget.setNodeIdinnodeShellLifecycle.ts/BaseWidget.ts) always keys on that full id.The node-render path reads widget ids back through
stripGraphPrefix(src/stores/widgetValueStore.ts) — used byLGraphNode.vue,processedWidgetRenderModel.ts,WidgetItem.vue, andWidgetTextPreview.vue. It used to strip everything up to the last colon unconditionally, which is correct for a genuine subgraph-scoped id (<uuid>:<localId>) but wrong forinsert_workflow's derived id: it collapsedinsert:<opId>:root:node:9down to bare9, a key nothing was ever registered under (and, if a secondinsert_workflowop happened to reuse the same original id, a key two unrelated nodes would collapse onto). The lookup came back empty, so the node rendered with zero widget rows.Fixed
stripGraphPrefixto only strip a prefix that is actually a subgraph UUID (repeating for nested subgraphs), leaving any other colon-bearing id untouched.Bug 2 — widget discoverable but stuck at its constructor default.
Once bug 1 is fixed,
agentNodeMaterializer.ts'smaterialize()still didn't restore the real value. The op layer stores an ordinary node'swidgets_valuesby name (readSemanticNode), butLGraphNode.configure()'s named-value restore is gated behind the experimentalComfy.Workflow.NamedValuesRestoresetting (off by default) or afallbackNameslist that is never populated for real catalog nodes. Without either,configure()falls back to positional restore againstwidgets_values— for a named payload that's a plain object, not an array, so nothing restores and the widget keeps whatever value its constructor set.This file already has a
withNamedValuesRestore()helper that flipsLiteGraph.namedValuesRestoreon for the duration of a call — it was used only when configuring a subgraph definition, never for the node's ownconfigure()call inmaterialize().add_node-created nodes mostly dodge this because the agent's add-node tool typically follow-ups with discreteset_widgetops (a direct, always-correct value write), whereasinsert_workflowbakes every widget value into the node's initial creation payload, hitting this path on every single widget of every inserted node.Fix: wrap
materialize()'snode.configure(...)call inwithNamedValuesRestore()too, so a named payload actually lands.Tests:
agentInsertWorkflowWidgets.regression.test.ts(new): drives the real pipeline (comfy-multi-player'sinsert_workflowapplier →EcsFollowerAdapter/AgentCrdtProjection→graphMutations→agentNodeMaterializer), then reads widget ids back exactly the wayLGraphNode.vuedoes (stripGraphPrefix+getNodeWidgetIds), asserting the widget is both discoverable and correctly-valued. Also covers twoinsert_workflowops that reuse the same original node id, proving they no longer collide.widgetValueStore.test.ts: focused unit coverage forstripGraphPrefix— bare id passthrough, single/nested subgraph-uuid stripping, the derived-id passthrough case, the same-trailing-segment non-collision case, and the empty-id case.Verified:
pnpm typecheck,pnpm lint(eslint + oxlint),pnpm format:check(via the repo's pre-commit hook) and the fullsrc/workbench/extensions/agent/crdt+src/storesVitest suites (2148 passed, 16 pre-existing expected-fails unrelated to this change, 1 skipped) all pass, plus the render-adjacentLGraphNode.widgetOrder.test.ts,WidgetItem.test.ts, andWidgetTextPreview.test.ts.🤖 Generated with Claude Code
https://claude.ai/code/session_01UqyourCBfi97xzpBx2W7oY
Generated by Claude Code