Repository navigation
fix(agent): require lineage_seq on doc_reset frames - #18277
Conversation
|
Important Review skippedBot user detected. To trigger a single review, invoke the ⚙️ Run configurationConfiguration used: Repository: Comfy-Org/ComfyUI_frontend/.coderabbit.yaml Review profile: ASSERTIVE Plan: Advanced Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Comment |
🎨 Storybook: ✅ Built — View Storybook🎭 Playwright: ✅ 2160 passed, 0 failed📊 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. Fallow gates on new findings in changed files only, so everything below was introduced or moved by this PR (11 changed files).
Pre-existing findings are baselined in Reproduce locally# the base the gate compared against
npx fallow audit --root . --quiet --format json \
--changed-since 3c72065cdfefabfa70df102a8e4d43ab1fa190c4
Full docs: ADR-DEVEX-LINT-0015 🧩 Custom Nodes Ecosystem Matrix: ✅ PassedExecutes the frontend JS of every registry pack that ships any, against this change, in both renderers. The eight
On a pin-bump PR, a red gate means the ecosystem moved — not that your diff broke something. That is what the bump PR is for. Outside a pin bump or a cold rebuild of the unpinned tail, a red means the diff. Note the corpus size and the executed count are different numbers, and the verdict is against the executed count: roughly 3,100 of ~5,100 packs ship no extension-shaped JS and are skipped. Reproduce locally# build one shard's fixture, then run it
python3 scripts/registry-census/build_matrix.py --shard 1/4
MATRIX_RENDERER=vue MATRIX_VUE=1 \
npx vitest run --config vitest.matrix.config.mts
Full docs: scripts/registry-census/README.md 📦 Bundle: 10 MB gzip 🔴 +26.3 kBDetailsSummary
Category Glance App Entry Points — 3.77 kB (baseline 3.73 kB) • 🔴 +38 BMain entry bundles and manifests
Status: 1 added / 1 removed Graph Workspace — 1.45 MB (baseline 1.44 MB) • 🔴 +12.4 kBGraph editor runtime, canvas, workflow orchestration
Status: 2 added / 2 removed / 1 unchanged Views & Navigation — 133 kB (baseline 132 kB) • 🔴 +1.65 kBTop-level views, pages, and routed surfaces
Status: 16 added / 16 removed / 2 unchanged Panels & Settings — 643 kB (baseline 622 kB) • 🔴 +20.4 kBConfiguration panels, inspectors, and settings screens
Status: 28 added / 27 removed User & Accounts — 39.4 kB (baseline 26.2 kB) • 🔴 +13.2 kBAuthentication, profile, and account management bundles
Status: 8 added / 8 removed / 3 unchanged Editors & Dialogs — 126 kB (baseline 126 kB) • 🟢 -49 BModals, dialogs, drawers, and in-app editors
Status: 8 added / 8 removed UI Components — 125 kB (baseline 123 kB) • 🔴 +1.8 kBReusable component library chunks
Status: 16 added / 16 removed Data & Services — 4.05 MB (baseline 3.72 MB) • 🔴 +333 kBStores, services, APIs, and repositories
Status: 13 added / 14 removed / 2 unchanged Utilities & Hooks — 582 kB (baseline 560 kB) • 🔴 +21.7 kBHelpers, composables, and utility bundles
Status: 30 added / 30 removed / 7 unchanged Vendor & Third-Party — 17.4 MB (baseline 18.2 MB) • 🟢 -823 kBExternal libraries and shared vendor chunks
Status: 9 added / 9 removed / 9 unchanged Other — 17.9 MB (baseline 17.5 MB) • 🔴 +424 kBBundles that do not match a named category
Status: 160 added / 156 removed / 140 unchanged ⚡ Performance Report
Show regressions
All metrics
Historical variance (last 15 runs)
Trend (last 15 commits on main)
Raw data{
"timestamp": "2026-09-29T17:14:33.817Z",
"gitSha": "c327bab57aaf443f625b47dabca6d27461535dfd",
"branch": "fix/agent-crdt-first-bind-node-flicker",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 2040.8279999999763,
"styleRecalcs": 11,
"styleRecalcDurationMs": 4.718,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 450.21700000000004,
"heapDeltaBytes": -35366032,
"heapUsedBytes": 63332632,
"domNodes": -228,
"jsHeapTotalBytes": 3407872,
"scriptDurationMs": 9.288,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66639344262295,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-idle",
"durationMs": 2042.3449999999548,
"styleRecalcs": 10,
"styleRecalcDurationMs": 4.062999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 487.52099999999996,
"heapDeltaBytes": -36819828,
"heapUsedBytes": 61807824,
"domNodes": -228,
"jsHeapTotalBytes": 3932160,
"scriptDurationMs": 11.251000000000001,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.6655737704918,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1909.6440000000143,
"styleRecalcs": 74,
"styleRecalcDurationMs": 18.493000000000002,
"layouts": 12,
"layoutDurationMs": 4.969,
"taskDurationMs": 858.2070000000001,
"heapDeltaBytes": -20788276,
"heapUsedBytes": 77844436,
"domNodes": -228,
"jsHeapTotalBytes": 3932160,
"scriptDurationMs": 173.61100000000002,
"eventListeners": -164,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665789473684207,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1892.4950000000536,
"styleRecalcs": 76,
"styleRecalcDurationMs": 20.121,
"layouts": 12,
"layoutDurationMs": 5.997,
"taskDurationMs": 892.501,
"heapDeltaBytes": -40891752,
"heapUsedBytes": 57764720,
"domNodes": -228,
"jsHeapTotalBytes": 4456448,
"scriptDurationMs": 177.03300000000002,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665486725663722,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1724.7830000000022,
"styleRecalcs": 31,
"styleRecalcDurationMs": 10.740000000000002,
"layouts": 6,
"layoutDurationMs": 1.2409999999999999,
"taskDurationMs": 348.4919999999999,
"heapDeltaBytes": -18826712,
"heapUsedBytes": 79377400,
"domNodes": 53,
"jsHeapTotalBytes": 7077888,
"scriptDurationMs": 12.483999999999998,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666346153846156,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1740.0569999999789,
"styleRecalcs": 33,
"styleRecalcDurationMs": 10.068000000000001,
"layouts": 6,
"layoutDurationMs": 1.0899999999999999,
"taskDurationMs": 318.538,
"heapDeltaBytes": 1494472,
"heapUsedBytes": 80493736,
"domNodes": 53,
"jsHeapTotalBytes": 3670016,
"scriptDurationMs": 9.96,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665384615384614,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-clipping",
"durationMs": 747.7260000000001,
"styleRecalcs": 13,
"styleRecalcDurationMs": 4.475999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 464.64700000000005,
"heapDeltaBytes": -2425248,
"heapUsedBytes": 95216448,
"domNodes": 0,
"jsHeapTotalBytes": 7340032,
"scriptDurationMs": 169.094,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665909090909096,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 710.4559999999083,
"styleRecalcs": 12,
"styleRecalcDurationMs": 4.308999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 457.688,
"heapDeltaBytes": 18573600,
"heapUsedBytes": 97136196,
"domNodes": 0,
"jsHeapTotalBytes": 4194304,
"scriptDurationMs": 170.35500000000002,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1192.8209999999808,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 837.4160000000002,
"heapDeltaBytes": 9599920,
"heapUsedBytes": 85696232,
"domNodes": 0,
"jsHeapTotalBytes": 4096000,
"scriptDurationMs": 70.996,
"eventListeners": 0,
"totalBlockingTimeMs": 561,
"frameDurationMs": 35.291176470588226,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1178.4250000000611,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 833.63,
"heapDeltaBytes": 9533968,
"heapUsedBytes": 85814016,
"domNodes": 0,
"jsHeapTotalBytes": 3047424,
"scriptDurationMs": 71.99399999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 560,
"frameDurationMs": 35.857575757575766,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 878.1360000000404,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 501.8989999999999,
"heapDeltaBytes": 13617728,
"heapUsedBytes": 93221100,
"domNodes": 0,
"jsHeapTotalBytes": 266240,
"scriptDurationMs": 183.92700000000005,
"eventListeners": 0,
"totalBlockingTimeMs": 231,
"frameDurationMs": 25.23714285714289,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 896.3589999999613,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 496.9930000000002,
"heapDeltaBytes": -12194904,
"heapUsedBytes": 66560224,
"domNodes": 0,
"jsHeapTotalBytes": 266240,
"scriptDurationMs": 170.86200000000002,
"eventListeners": 0,
"totalBlockingTimeMs": 243,
"frameDurationMs": 24.997222222222263,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "extension-clean-combined-draw",
"durationMs": 10638.119999999986,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 9710.459,
"heapDeltaBytes": -11886652,
"heapUsedBytes": 69181388,
"domNodes": 0,
"jsHeapTotalBytes": -12845056,
"scriptDurationMs": 7978.192000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 1259,
"frameDurationMs": 33.02142857142857,
"p95FrameDurationMs": 66.69999999999982
},
{
"name": "extension-clean-combined-draw",
"durationMs": 10406.525999999985,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 9467.499,
"heapDeltaBytes": -8551000,
"heapUsedBytes": 71613272,
"domNodes": 0,
"jsHeapTotalBytes": -12582912,
"scriptDurationMs": 7859.420999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 1095,
"frameDurationMs": 31.99876923076923,
"p95FrameDurationMs": 66.59999999999854
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 10790.311999999973,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 9885.177,
"heapDeltaBytes": -9753532,
"heapUsedBytes": 71531016,
"domNodes": 0,
"jsHeapTotalBytes": -13107200,
"scriptDurationMs": 8187.549000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 1148,
"frameDurationMs": 32.478313253012054,
"p95FrameDurationMs": 66.60000000000036
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 10104.264999999941,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 9156.784,
"heapDeltaBytes": 11305288,
"heapUsedBytes": 92156692,
"domNodes": 0,
"jsHeapTotalBytes": -13893632,
"scriptDurationMs": 7562.392,
"eventListeners": 0,
"totalBlockingTimeMs": 1098,
"frameDurationMs": 31.46292834890966,
"p95FrameDurationMs": 66.59999999999945
},
{
"name": "large-graph-idle",
"durationMs": 2049.0669999999795,
"styleRecalcs": 9,
"styleRecalcDurationMs": 3.1329999999999987,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 573.1809999999999,
"heapDeltaBytes": -35631060,
"heapUsedBytes": 83055396,
"domNodes": -228,
"jsHeapTotalBytes": -360448,
"scriptDurationMs": 34.922,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66639344262295,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-idle",
"durationMs": 2044.8709999999437,
"styleRecalcs": 8,
"styleRecalcDurationMs": 3.4029999999999996,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 446.55600000000004,
"heapDeltaBytes": -36848288,
"heapUsedBytes": 82524640,
"domNodes": -228,
"jsHeapTotalBytes": -622592,
"scriptDurationMs": 39.239999999999995,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666393442622944,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-pan",
"durationMs": 2457.855000000052,
"styleRecalcs": 69,
"styleRecalcDurationMs": 11.683,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1632.307,
"heapDeltaBytes": 24728068,
"heapUsedBytes": 87021084,
"domNodes": -228,
"jsHeapTotalBytes": 225280,
"scriptDurationMs": 852.459,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665753424657538,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-pan",
"durationMs": 2436.533000000054,
"styleRecalcs": 69,
"styleRecalcDurationMs": 10.407,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1643.3010000000002,
"heapDeltaBytes": 20624944,
"heapUsedBytes": 83366576,
"domNodes": -228,
"jsHeapTotalBytes": 225280,
"scriptDurationMs": 855.995,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.78137931034482,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-zoom",
"durationMs": 3601.257999999973,
"styleRecalcs": 63,
"styleRecalcDurationMs": 10.091,
"layouts": 60,
"layoutDurationMs": 13.405,
"taskDurationMs": 1653.362,
"heapDeltaBytes": 12510960,
"heapUsedBytes": 76897168,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 781.6959999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.6662037037037,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-zoom",
"durationMs": 3549.7910000000275,
"styleRecalcs": 63,
"styleRecalcDurationMs": 10.324,
"layouts": 60,
"layoutDurationMs": 13.493,
"taskDurationMs": 1672.974,
"heapDeltaBytes": 12209284,
"heapUsedBytes": 76601476,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 783.158,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66603773584906,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "legacy-node-drag",
"durationMs": 3146.5110000000323,
"styleRecalcs": 42,
"styleRecalcDurationMs": 5.526,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1910.2069999999999,
"heapDeltaBytes": 24394604,
"heapUsedBytes": 88265336,
"domNodes": -228,
"jsHeapTotalBytes": 417792,
"scriptDurationMs": 982.897,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.755614973262034,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 3147.5370000000567,
"styleRecalcs": 45,
"styleRecalcDurationMs": 6.748999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1803.7309999999998,
"heapDeltaBytes": 29951120,
"heapUsedBytes": 94705900,
"domNodes": 0,
"jsHeapTotalBytes": 679936,
"scriptDurationMs": 1016.0659999999999,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.75531914893617,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2022.4469999999997,
"styleRecalcs": 8,
"styleRecalcDurationMs": 2.7469999999999994,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 559.1059999999999,
"heapDeltaBytes": -36774076,
"heapUsedBytes": 83000340,
"domNodes": -228,
"jsHeapTotalBytes": 950272,
"scriptDurationMs": 33.065999999999995,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66639344262295,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2040.7749999999396,
"styleRecalcs": 7,
"styleRecalcDurationMs": 3.042999999999997,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 538.196,
"heapDeltaBytes": -37412416,
"heapUsedBytes": 83765260,
"domNodes": -228,
"jsHeapTotalBytes": -360448,
"scriptDurationMs": 27.621000000000006,
"eventListeners": -164,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666393442622944,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 612.9329999999982,
"styleRecalcs": 48,
"styleRecalcDurationMs": 6.416000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 391.34000000000003,
"heapDeltaBytes": 13080892,
"heapUsedBytes": 93966748,
"domNodes": 0,
"jsHeapTotalBytes": 4194304,
"scriptDurationMs": 144.178,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 582.6439999999593,
"styleRecalcs": 46,
"styleRecalcDurationMs": 5.8320000000000025,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 392.886,
"heapDeltaBytes": 13148664,
"heapUsedBytes": 93957492,
"domNodes": 0,
"jsHeapTotalBytes": 3670016,
"scriptDurationMs": 146.696,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66571428571429,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-idle",
"durationMs": 2068.614000000025,
"styleRecalcs": 11,
"styleRecalcDurationMs": 3.9930000000000017,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 453.066,
"heapDeltaBytes": -12443220,
"heapUsedBytes": 68847068,
"domNodes": -244,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 10.717,
"eventListeners": -188,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66612903225806,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2050.942999999961,
"styleRecalcs": 10,
"styleRecalcDurationMs": 4.198,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 343.65799999999996,
"heapDeltaBytes": -13045464,
"heapUsedBytes": 66897348,
"domNodes": -244,
"jsHeapTotalBytes": -262144,
"scriptDurationMs": 7.824999999999998,
"eventListeners": -188,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66585365853659,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1773.65199999997,
"styleRecalcs": 76,
"styleRecalcDurationMs": 19.750999999999998,
"layouts": 16,
"layoutDurationMs": 6.292000000000001,
"taskDurationMs": 827.666,
"heapDeltaBytes": -22539960,
"heapUsedBytes": 58050220,
"domNodes": -244,
"jsHeapTotalBytes": 524288,
"scriptDurationMs": 117.30200000000002,
"eventListeners": -188,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666037735849052,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1728.4760000000006,
"styleRecalcs": 72,
"styleRecalcDurationMs": 18.471,
"layouts": 16,
"layoutDurationMs": 9.736000000000002,
"taskDurationMs": 683.7990000000001,
"heapDeltaBytes": -25521124,
"heapUsedBytes": 55359324,
"domNodes": -244,
"jsHeapTotalBytes": 1310720,
"scriptDurationMs": 116.496,
"eventListeners": -188,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.827184466019414,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-transition-enter",
"durationMs": 1418.7759999999798,
"styleRecalcs": 17,
"styleRecalcDurationMs": 29.561999999999998,
"layouts": 5,
"layoutDurationMs": 13.221,
"taskDurationMs": 877.5560000000003,
"heapDeltaBytes": 21416888,
"heapUsedBytes": 94297816,
"domNodes": 13667,
"jsHeapTotalBytes": 2359296,
"scriptDurationMs": 28.218,
"eventListeners": 2383,
"totalBlockingTimeMs": 129,
"frameDurationMs": 19.109333333333332,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "viewport-pan-sweep",
"durationMs": 8468.030999999995,
"styleRecalcs": 248,
"styleRecalcDurationMs": 31.857999999999997,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 5673.42,
"heapDeltaBytes": -39954828,
"heapUsedBytes": 78656016,
"domNodes": -228,
"jsHeapTotalBytes": -167936,
"scriptDurationMs": 2758.866,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.731818181818184,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "viewport-pan-sweep",
"durationMs": 8681.648999999994,
"styleRecalcs": 250,
"styleRecalcDurationMs": 32.813,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 5848.28,
"heapDeltaBytes": -34785696,
"heapUsedBytes": 84055668,
"domNodes": -228,
"jsHeapTotalBytes": 94208,
"scriptDurationMs": 2823.675,
"eventListeners": -142,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.698073217726396,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-idle",
"durationMs": 16771.301999999992,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 16040.376999999999,
"heapDeltaBytes": -30421332,
"heapUsedBytes": 167622620,
"domNodes": -3045,
"jsHeapTotalBytes": 7806976,
"scriptDurationMs": 253.913,
"eventListeners": -14915,
"totalBlockingTimeMs": 0,
"frameDurationMs": 28.8993091537133,
"p95FrameDurationMs": 50.100000000000364
},
{
"name": "vue-large-graph-idle",
"durationMs": 16751.666,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 16252.257999999998,
"heapDeltaBytes": -793600,
"heapUsedBytes": 167656156,
"domNodes": -8049,
"jsHeapTotalBytes": -20447232,
"scriptDurationMs": 262.91700000000003,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 28.602905982905984,
"p95FrameDurationMs": 50.10000000000218
},
{
"name": "vue-large-graph-pan",
"durationMs": 20708.016000000043,
"styleRecalcs": 170,
"styleRecalcDurationMs": 17.05300000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 20432.757999999998,
"heapDeltaBytes": -20443272,
"heapUsedBytes": 190812712,
"domNodes": -3045,
"jsHeapTotalBytes": 7544832,
"scriptDurationMs": 938.084,
"eventListeners": -14915,
"totalBlockingTimeMs": 33,
"frameDurationMs": 39.02358490566038,
"p95FrameDurationMs": 83.30000000000291
},
{
"name": "vue-large-graph-pan",
"durationMs": 20860.39299999993,
"styleRecalcs": 169,
"styleRecalcDurationMs": 17.543000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 20475.482,
"heapDeltaBytes": -37730132,
"heapUsedBytes": 191129912,
"domNodes": -3046,
"jsHeapTotalBytes": 2801664,
"scriptDurationMs": 889.604,
"eventListeners": -14915,
"totalBlockingTimeMs": 47,
"frameDurationMs": 38.35606617647059,
"p95FrameDurationMs": 83.29999999999927
},
{
"name": "vue-node-resize-workload",
"durationMs": 6020.976999999959,
"styleRecalcs": 20,
"styleRecalcDurationMs": 77.38200000000003,
"layouts": 20,
"layoutDurationMs": 22.420999999999996,
"taskDurationMs": 5922.293,
"heapDeltaBytes": 8650520,
"heapUsedBytes": 195876268,
"domNodes": -8049,
"jsHeapTotalBytes": 18612224,
"scriptDurationMs": 105.33200000000001,
"eventListeners": -3,
"totalBlockingTimeMs": 399,
"frameDurationMs": 34.873410404624266,
"p95FrameDurationMs": 66.70000000000073
},
{
"name": "vue-node-resize-workload",
"durationMs": 5958.718999999974,
"styleRecalcs": 20,
"styleRecalcDurationMs": 77.85300000000001,
"layouts": 20,
"layoutDurationMs": 21.983000000000004,
"taskDurationMs": 5839.362,
"heapDeltaBytes": -5465668,
"heapUsedBytes": 194096908,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 110.10199999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 404,
"frameDurationMs": 34.48843930635838,
"p95FrameDurationMs": 66.70000000000073
},
{
"name": "vue-zoom-out-idle",
"durationMs": 8577.233000000035,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8328.915,
"heapDeltaBytes": 6825524,
"heapUsedBytes": 190488968,
"domNodes": 0,
"jsHeapTotalBytes": -29097984,
"scriptDurationMs": 136.05499999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 29.905923344947734,
"p95FrameDurationMs": 50.10000000000218
},
{
"name": "vue-zoom-out-idle",
"durationMs": 8499.889000000054,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 8198.874,
"heapDeltaBytes": -22772160,
"heapUsedBytes": 164687028,
"domNodes": 0,
"jsHeapTotalBytes": -18612224,
"scriptDurationMs": 125.65599999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 30.796014492753624,
"p95FrameDurationMs": 66.59999999999854
},
{
"name": "workflow-execution",
"durationMs": 477.57400000000416,
"styleRecalcs": 13,
"styleRecalcDurationMs": 14.467,
"layouts": 3,
"layoutDurationMs": 0.9209999999999999,
"taskDurationMs": 122.63699999999997,
"heapDeltaBytes": -14497956,
"heapUsedBytes": 81349332,
"domNodes": 104,
"jsHeapTotalBytes": 4194304,
"scriptDurationMs": 11.216999999999999,
"eventListeners": 81,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.667857142857137,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "workflow-execution",
"durationMs": 476.1040000000776,
"styleRecalcs": 14,
"styleRecalcDurationMs": 15.623999999999999,
"layouts": 3,
"layoutDurationMs": 0.9279999999999999,
"taskDurationMs": 122.272,
"heapDeltaBytes": -14456140,
"heapUsedBytes": 82090856,
"domNodes": 101,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 11.274999999999999,
"eventListeners": 81,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.668965517241396,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "agent-composer-empty-conversation-typing",
"durationMs": 845.3940000000557,
"styleRecalcs": 78,
"styleRecalcDurationMs": 103.547,
"layouts": 76,
"layoutDurationMs": 20.836000000000002,
"taskDurationMs": 814.2349999999999,
"heapDeltaBytes": 17534096,
"heapUsedBytes": 88654648,
"domNodes": 76,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 331.444,
"eventListeners": 876,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "agent-composer-empty-conversation-typing",
"durationMs": 834.29000000001,
"styleRecalcs": 78,
"styleRecalcDurationMs": 102.85,
"layouts": 76,
"layoutDurationMs": 20.689,
"taskDurationMs": 798.5289999999999,
"heapDeltaBytes": 17165892,
"heapUsedBytes": 86595484,
"domNodes": 76,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 326.403,
"eventListeners": 876,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.664705882352933,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "agent-composer-long-conversation-typing",
"durationMs": 1342.5600000000486,
"styleRecalcs": 78,
"styleRecalcDurationMs": 243.9,
"layouts": 76,
"layoutDurationMs": 24.925,
"taskDurationMs": 1307.148,
"heapDeltaBytes": 13999032,
"heapUsedBytes": 132155704,
"domNodes": 76,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 448.009,
"eventListeners": 876,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.300000000000008,
"p95FrameDurationMs": 16.80000000000109
},
{
"name": "agent-composer-long-conversation-typing",
"durationMs": 1428.6070000000564,
"styleRecalcs": 78,
"styleRecalcDurationMs": 246.161,
"layouts": 76,
"layoutDurationMs": 25.006,
"taskDurationMs": 1411.2969999999998,
"heapDeltaBytes": -4890468,
"heapUsedBytes": 127528800,
"domNodes": -7943,
"jsHeapTotalBytes": -3186688,
"scriptDurationMs": 475.19100000000003,
"eventListeners": -1249,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.480487804878067,
"p95FrameDurationMs": 16.80000000000109
}
]
}🔬 E2E Coverage
Top 10 files by uncovered lines
|
f1ca0fd to
12a851e
Compare
There was a problem hiding this comment.
🔍 Cursor Review — Consolidated panel
Triggered by @christian-byrne.
Found 7 finding(s).
| Severity | Count |
|---|---|
| 🟠 High | 4 |
| 🟡 Medium | 1 |
| 🟢 Low | 2 |
Panel: 8/8 reviewers contributed findings.
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## main #18277 +/- ##
==========================================
+ Coverage 84.14% 85.00% +0.85%
==========================================
Files 3226 2332 -894
Lines 155836 133195 -22641
Branches 40278 35869 -4409
==========================================
- Hits 131131 113220 -17911
+ Misses 22977 19621 -3356
+ Partials 1728 354 -1374
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 909 files with indirect coverage changes 🚀 New features to boost your workflow:
|
d76598f to
15fdfcb
Compare
This comment has been minimized.
This comment has been minimized.
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: blocking
Two first-reconcile orderings still violate the intended ownership rule:
- P1: A frame can arrive after workflow scope exists but before the live graph is ready.
localOnlyNodeIdsreturnsnull, yet the adapter commitsremoveMissingwith empty protection and clearsreconcileNextFrame. The later graph-ready reconciliation sees the already-pruned stores and removes the hand-added node. - P1: On a cold remount,
everSeenDocNodeIdsstarts empty. A local node that the remote document held and deleted before this client observed it is therefore classified exactly like a never-synced hand-added node and retained. The existing E4 reply acknowledges that the client cannot reconstruct this history, so the absolute contract and cold-remount behavior remain unresolved.
Findings without an inline anchor
- P2: The PR summary still describes the pre-
6b203ebdesign as tracking workflows it has "ever bound" and comparing against the document's current node set. The final code marks a workflow only after a committed protected reconcile and compares against history observed by this client. Please update the summary to match the code and state the cold-start limit.
Existing findings, current disposition
- Incident links were omitted: confirmed, addressed. The current hook returns incident link IDs and
removeMissingretains them. The integration test keeps link 99 while sweeping unrelated link 98. - Rejected batches spent the one-shot: confirmed, addressed. Spending now requires
committed && localOnly !== null, and the rejected-batch test retries successfully. - Bind-time marking burned protection: confirmed, addressed. The adapter adds the workflow to its set only after a protected commit; the pre-frame rebind test covers this ordering.
- Current-snapshot classification can resurrect a remote deletion: confirmed, still remaining on cold remount. Accumulating IDs fixes deletions this instance observed, but a new instance has no earlier document history. The author reply explicitly notes this limit.
- No live graph can consume protection: confirmed, still remaining in a different form. The boolean is no longer cleared, but the full reconcile can commit without protection and clear
reconcileNextFrame, leaving that boolean unreachable. See the inline P1 finding. - The hook could answer for another workflow: confirmed, addressed. A mismatched workflow now returns
null. destroy()retained adapter history: confirmed, addressed.destroy()clears the set, and the reuse test confirms protection is eligible again.
Unverified candidate
- P3: The PR description gives precise claims about the half-second duration, the exact human-op round trip, a red run on unmodified
main, and earlier full-suite totals, but the supplied logs do not include those runs. Please attach or link the outputs, or qualify those statements as reported/inferred. The focused checks I ran do not settle them.
Checked and cleared
- I checked whether the conversation harness is the wrong owner for blank startup. Its consumers all need the same blank replay baseline, while
bootAgentAppalso serves tests that require normal startup state. Keeping this setup in the harness is the narrowest current owner. - The PR's "clean" provenance text lists specific local commands. I do not read it as a claim that aggregate CI is green. CI is currently 83 successful, 15 skipped, and 2 failed checks, with
playwright-tests (cloud)ande2e-statusfailing.
Coverage
Standard depth. Correctness, maintainability, idiomaticity, tests, and documentation lanes reviewed all seven changed files. The tests lane applied the unit-test, mock-quality, setup-consolidation, proof, Playwright, and flake lenses. Fallow reviewed all changed files; its public-contract prompts were traced to their consumers and its baseline complexity findings were not introduced by this PR.
Local checks on a7f2c3c84ac3a9eee4e9ac6bd52a3075021f9699: targeted Vitest 150/150 passed; pnpm typecheck passed; pnpm typecheck:browser passed; touched-file ESLint and oxfmt passed; Fallow completed. Current GitHub checks have no pending jobs, 83 successes, 15 skipped jobs, and the 2 cloud E2E failures named above.
|
Replying to the changes-requested review as a whole, on top of the inline replies (545ee8a): Both P1 blockers:
P2s: PR description rewritten to match the current code ( P3s: Full verification: Per this repo's Created by Claude Tag |
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: changes requested.
I confirmed that the null-graph deferral, explicit materializer fixture, shared blank workflow value, state rename, and non-empty authoritative rebind test address six prior findings. The cold-remount provenance defect and forbidden test double assertions remain. This head also leaves the affected cloud Playwright job red with five agent failures after retries.
The focused CRDT tests pass (150 tests), as do app/browser typechecks, touched-file ESLint and oxfmt, and knip. playwright-tests (cloud) and aggregate e2e-status fail.
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: changes requested.
I confirmed six prior findings are addressed: the unsafe graph casts, repeated setup, duplicated protection state, hook naming, retry comment, and blank-workflow field count. Three prior findings remain: fresh-page provenance loss, the red cloud agent suite, and the missing production classifier-to-adapter composition test.
P2 documentation: The known-limitation paragraph still says “cold reload/remount” and that every brand-new client instance starts empty. This revision specifically preserves history across ordinary composable remounts; the remaining boundary is a new page/Pinia lifetime that never observed the deleted node. Please describe that boundary precisely.
A new blocking reset issue is also present: doc_reset clears graph/history but does not re-arm adapter protection for the new lineage. Focused Vitest passes 156 tests, and typecheck, browser typecheck, ESLint, oxlint, formatting, knip, and Playwright test discovery complete successfully. Current CI still fails playwright-tests (cloud) with five agent failures after retries, plus aggregate e2e-status.
| * wrapping `loadGraphData` the moment `window.app` is assigned intercepts | ||
| * that startup call before it ever reaches `defaultGraph`. | ||
| */ | ||
| private async installBlankStartupGraph(): Promise<void> { |
There was a problem hiding this comment.
P1: The affected cloud lane is still red at this head: five agent tests fail after all retries, and the author independently reproduced four as regressions relative to base. The fixture change has not restored replay, delete, paste, and tab-switch behavior. Please fix the underlying regression and restore the cloud suite before this can merge.
There was a problem hiding this comment.
Following up on the verification note below — the cloud suite is now green on four of the five specs you named, with one exception I'm flagging rather than claiming fixed.
Root-caused all four via real local PLAYWRIGHT_LOCAL=1 --project=cloud runs (build:cloud, served, executed against the actual failing specs, repeated for determinism), bisecting against the PR's base commit in an isolated worktree to separate "caused by this PR's production code" from "caused by/masked by the fixture's boot sequence":
- LoadImage missing-node placeholder — not a production-code regression.
agentReplayNodeDefs.ts'sLoadImagefixture def has always lackedimage_upload: trueon itsimagecombo, soComfy.UploadImage'sbeforeRegisterNodeDefhook never wiresimageInputNameinto theuploadinput it added by hand;useImageUploadWidget's constructor then throws ('Image upload widget requires imageInputName augmentation'),LiteGraph.createNodecatches it and returnsnull, andagentNodeMaterializer.ts's?? missingNode(state)fallback degrades it. Confirmed this throws identically against the PR's base commit's own app build with the old fixture too — the old boot sequence just happened not to exercise the turn that materializes it, so the bug was already live onmain, silently. Fixed the fixture data (image: [[], { image_upload: true }]). - Both off-viewport click timeouts — genuinely caused by
installBlankStartupGraph: the old boot's stock-template load incidentally triggeredapp.ts'sloadGraphData→fitView()call, leaving the camera at a padded fit; a graph that never has content at boot never callsfitView()at all, so the camera stays at its construction-time default (scale 1, offset [0,0]), and these recordings' nodes render with headers above the viewport's top edge. AddedfitViewToContent(), called once real content exists after the first turn, computing bounds from the live nodes directly (canvas.ds.fitToBounds) rather than relying on that incidental side effect. - Paste-position case — a float-precision assertion (
toEqualon graph coordinates), exposed byfitViewToContent()choosing a different, non-"clean" scale than the old flow's incidental one. Switched totoBeCloseTo.
Still red: agentTabSwitchCatchUp.spec.ts's first case — a ckpt_name widget's invalid flag (backed by missingModelStore.isWidgetMissingModel) is false after "open a new blank workflow, then return to this tab", where it should stay true. Confirmed this is also fixture-caused, not a production-code regression (fails identically against the base-commit app build with this same fixture). Traced it to missingModelStore's scan-and-replace being global rather than per-workflow-scoped (runMissingModelPipeline's setMissingModels replaces the whole list based on whatever's in rootGraph at scan time, so opening the blank new-workflow tab wipes it), but haven't pinned down why the tab-switch-back doesn't re-populate it under this boot sequence when it does under the old one. This feels closer to "does missingModelStore need per-workflow scoping" than a clear-cut bug in this PR's own code, so I'm flagging it rather than digging further blind — happy to keep going if you'd rather I did, or to have it filed separately.
The remaining @screenshot case (agentConversationReplay.spec.ts's wire-evidence test, ~0.01 pixel-ratio diff) looks like the usual font-rendering/environment variance these carry, not a logic regression.
Created by Claude Tag
There was a problem hiding this comment.
Correction to my comment above: I said the LoadImage/image_upload: true fixture bug "was already live on main, silently" — that's wrong, and the framing in the PR description matched it. It was never silent.
A careful check just now (isolated worktree, main completely unmodified, running agentConversationReplay.spec.ts's agent-workflow-editing-05 case exactly as it exists on main, twice for consistency) shows that test already fails on main today, reproducibly, with the identical LoadImage-materializes-as-missing-node-placeholder error.
So the accurate framing is: the image_upload: true omission is pre-existing on main and unrelated to this PR's production code, and it already causes that spec to fail on main right now — this PR's fixture change just happens to also fix it as a side effect of fixing the image_upload: true data gap, not because it "exposed" a previously-silent bug. I've updated the PR description's Cloud E2E fixture bullet to say this plainly.
Generated by Claude Code
|
Replying to this round's summary (review #5271649262) — individual replies are posted on each of the eight inline threads; a rundown here:
Verification: New head: 16e00c2. Created by Claude Tag |
|
Follow-up: CI's Fallow check went from pass-with-a-pre-existing-finding to a hard fail after 16e00c2, flagging Extracted the Created by Claude Tag |
Playwright videos for new testsDownload the recorded videos and HTML report Recorded projects: cloud Recorded test titles:
|
|
Investigated the three red Root cause: CI infra flake, not a code regression. Each of the three failed at the same step, before any Playwright test ran: This PR's diff doesn't touch any Verification: re-ran just the failed jobs on run 35658342012 with zero code changes. All three (shard 6, 7, 12) are now green. That confirms it was transient and unrelated to this branch. No commit pushed, since there was nothing in the code to fix. Separately, Generated by Claude Code |
|
Updating Playwright Expectations |
|
Ready for re-review. CI is green across the board except the one already-documented Generated by Claude Code |
|
The Generated by Claude Code |
christian-byrne
left a comment
There was a problem hiding this comment.
Verdict: changes requested. The two redundant setup comments are gone, but the production and coverage blockers from the previous review remain on 355cbd45f5.
- Keep
lineage_seqrequired through the event boundary.DocResetDetail.lineageSeqis still optional, anddetail.lineageSeq ?? detail.seq ?? 0can merge distinct document lineages when an older or malformed event omits the authoritative lineage sequence. Update the generated contract and reject incomplete reset events instead of manufacturing lineage identity. - Prove follower replacement clears non-empty semantic state. The current test only verifies calls and an empty replacement document, so it can pass while stale nodes or links survive the replacement sweep. Seed stores and the live graph, replace the follower, and assert both are empty before the replacement state materializes.
- Wait for CRDT catch-up before asserting the reload result.
agentHumanAddReload.spec.tschecks node visibility beforesendPrompt()and waits only for a new subscribe afterward, so the final assertions can run before the first reconcile that the test claims to cover. Add a deterministic signal for the catch-up frame or reconcile completion, then assert the node survives. - Bound and complete provenance tracking.
agentCrdtDocHistoryStore.tsstill creates append-only reset and per-node keys and scans all localStorage entries on reads, with no cleanup policy. It also records node provenance only, so a locally added link between previously observed nodes is not protected bycomputeLocalOnlyGraphIds()and can be swept on first reconcile. Add safe lineage cleanup and independent link provenance with regression coverage.
Verification: all required checks currently pass, including lint/format, unit tests, website E2E, and CLA. The broader browser/ecosystem matrix is still running, and the AI co-author trailer check is failing. Source verification on the pinned head confirms the latest commit changes only the two review-requested comments.
|
Playwright Cause: 7851a4c made Fix is one line: add Generated by Claude Code |
|
Merged |
christian-byrne
left a comment
There was a problem hiding this comment.
Verdict: changes requested.
The lineage_seq parser and follower behavior look correct, and the relevant unit, typecheck/lint, and browser E2E checks pass. However, commit bd058252fbf8c36bf54b9469fce9c7f3eb03b1fa includes a Claude-Session trailer, so Check for AI agent co-author trailers fails.
Please rewrite that commit without the attribution trailer and let the check rerun. No code changes are otherwise requested from this review.
|
@christian-byrne thanks. The Rewriting the message alone won't clear it, and any commit I make is authored as Claude, so this needs a human identity on the commits. The smallest fix: I squash the branch into one commit on top of current Generated by Claude Code |
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: changes requested. The protocol fix and its regression coverage are correct, but the current head still fails the attribution check. One contract-ownership issue and three PR-description corrections also remain.
- P1: I confirm the existing attribution finding.
bd058252fbcontains aClaude-Sessiontrailer, contrary to the repository's commit policy, and the exact-headCheck for AI agent co-author trailersjob fails. Rewrite that commit without the trailer. - P2: The inline finding asks the parser and valid test frames to derive their field names from generated
DocResetDatarather than maintaining another copy of the server contract.
PR-description findings
- P2: The summary says the follower or its consumers manufactured lineage identity from
seq. At the base revision,LayoutFollowerBridgereplaced the document on a valid reset but did not interpretseqas lineage identity, and no consumer derived one from it. State the actual compatibility fix: the local parser accepted resets that violated the generated contract by omittinglineage_seq. - P3: The description says a rejected frame "gets reported," but
DocFrameClientreports only the first malformed frame of each type per client. Later malformed resets are dropped without another report. Name that rate limit. - P3 Nit: Remove the decorative robot footer and bare session URL. The repository asks PR descriptions to be concise and forbids emojis. I am not treating attribution in the PR body itself as prohibited policy.
Prior findings closed on this head
- Duplicate
resyncWidget: addressed by6c6dca8630. - Required
lineage_seq: addressed by7851a4c76c. - Non-empty follower replacement coverage: superseded by the graph-API projection merged at
de662f65e1. - Reload catch-up readiness: addressed by
22395334c9. - Duplicate provenance callback: superseded when the old adapter was removed.
- Unbounded history storage: superseded when the history store was removed.
- Missing link provenance: superseded when the local-only sweep was removed.
- Repeated document snapshots: superseded by frame-delta projection.
- Startup-helper claim: superseded when the helper was removed.
- Adapter test naming: superseded by the replacement test suites.
Verification and coverage
Reviewed at bd058252fb, depth standard, across correctness, maintainability, idiomaticity, tests, and documentation.
- Focused Vitest: 3 files, 89 tests passed.
- App and browser typechecks passed.
- Changed-file ESLint and formatting passed.
- Playwright discovery found all 4 affected tests in the 2 changed specs.
- Focused malformed-reset composition test passed.
- Fallow audited all 7 changed files and found no new gated issue.
- Exact-head CI has 103 successful, 28 skipped, and 1 failed check. The only failure is the attribution check above.
Correctness and test-design review found no behavioral defect. A malformed reset is rejected before dispatch, zero and a lineage distinct from seq are covered, follower state remains unchanged, and both browser reset payloads now pass through the production parser.
| v?: unknown | ||
| workflow_id?: unknown | ||
| seq?: unknown | ||
| lineage_seq?: unknown |
There was a problem hiding this comment.
P2: @comfyorg/ingest-types already owns this required field in generated DocResetData, but this parser, the shared test fixture, and the two valid E2E reset payloads maintain separate copies. This PR shows the drift cost: the generated contract required lineage_seq while every local copy still compiled until runtime validation exposed the omission. Keep malformed boundary values as unknown, but derive the accepted field names and valid fixture payloads from generated DocResetData / DocResetFrame so the next contract change fails at compile time.
There was a problem hiding this comment.
Addressed in 224387688a: the parser and shared fixture now derive reset wire keys from generated DocResetData, and both browser reset payloads are checked with DocResetFrame.
bd05825 to
2243876
Compare
christian-byrne
left a comment
There was a problem hiding this comment.
Verdict: approved.
The current head keeps lineage_seq required at the wire boundary, rejects missing or invalid values before dispatch, and types the shared and browser-test reset payloads against the generated ingest contract. The attribution check is also green.
Verified on 224387688a9a4523c02bb17280607e81f2148cab:
- Parser, numeric-domain, and follower-subscription suites: 89 tests passed.
- Frontend and browser-test typechecks passed.
- Targeted lint passed for all changed files.
- All required GitHub checks passed, including unit tests, browser E2E, lint/format, website E2E, and CLA.
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: changes requested.
The reset contract and tests are now correct, and the generated-type ownership finding is fixed. The exact head still fails a required repository check, despite the intervening approval's statement that all checks pass.
- P1:
Check for AI agent co-author trailersfails on224387688a; GitHub annotates the run with "AI agent commit authorship detected in PR commits." The PR range still contains commits authored or committed by Claude/Amp and fourAmp-Thread-IDtrailers. Rewrite every flagged commit's author/committer identity and remove prohibited trailers, then verify.github/scripts/check-ai-co-authors.shexits 0 against the exact base and head.
Prior findings
- P2 generated contract ownership: addressed by
224387688a; I closed the loop in the original thread. - P2/P3 PR description: addressed. The current body states the actual parser compatibility fix, names the once-per-frame-type reporting limit, and removes the decorative footer and session URL.
- P1 attribution: the earlier
Claude-Sessiontrailer is gone, but the required attribution check remains red for the commit identities/trailers above. - I contest the approval's claim that the attribution check is green: REST check run
109347398501is completed with conclusionfailureat this exact SHA.
Checked and cleared
- The actual merge-base diff contains the seven reset-contract files described by the PR, not the unrelated files visible in a two-dot comparison against the newer base.
Partial<Record<keyof DocResetData, unknown>>intentionally derives referenced wire keys without rejecting unknown extra fields at an external boundary; that is compatible with the PR's compile-time coupling claim.- No correctness, type-safety, or test-design defect remains in the parser, fixture, or Playwright payload changes.
Verification
Reviewed 224387688a at standard depth across correctness, maintainability, idiomaticity, tests, and documentation.
- Focused Vitest: 3 files, 89 tests passed.
- App and browser typechecks passed.
- Changed-file ESLint and formatting passed.
- Playwright discovery found all 4 affected tests in the 2 changed specs.
- Exact-head checks: 103 successful, 18 skipped, and 1 failed across the paginated REST response; the failure is the attribution check above.
| // No `lineage_seq` on the wire — an older or malformed host frame must not | ||
| // have its lineage identity manufactured from `seq`. |
There was a problem hiding this comment.
P3 Nit: Delete this comment. The test title already says the reset is missing lineage_seq and must not fall back to seq; the input visibly omits the field, and the assertions prove rejection. Keeping a third prose copy adds only a stale-text risk.
There was a problem hiding this comment.
2243876 to
b0ff2ef
Compare
|
@DrJKL squashed the branch into a single trailer-free commit, b0ff2ef, on top of the merge base (tree identical to 2243876 apart from the test-comment nit). The Amp commits and Generated by Claude Code |
parseServerDocFrame now rejects doc_reset frames missing lineage_seq, matching the generated DocResetData contract. The shared fixture and the E2E reset payloads are typed against DocResetData/DocResetFrame so a contract change fails typecheck at every site.
b0ff2ef to
e42f886
Compare
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: ready to approve. e42f8863ec clears the last two findings from my previous review. No new findings.
The branch is now one commit authored and committed by Christian Byrne, and Check for AI agent co-author trailers passes at this SHA. The seven PR files match the tree I reviewed at 224387688a, except that the redundant two-line comment in followerSubscription.test.ts is gone. The PR now requires lineage_seq on doc_reset, rejects missing or invalid values before dispatch, and checks the parser keys, shared fixture, and both browser reset payloads against generated DocResetData / DocResetFrame.
One item remains outside this code: the PR body asks whether every host path that emits doc_reset sends lineage_seq. This frontend diff cannot settle that. Before merge, the backend owner should confirm it, because a host that omits the field will have its resets dropped.
Review record
Every finding I raised on this PR, grouped by review. "Superseded by #18700" means main's CRDT projection rework deleted or replaced the code in question, and the squashed branch no longer changes it. For those threads, I linked this round's reply.
Body-level findings
- 2026-09-21,
a7f2c3c84a. P1, a frame before live-graph readiness spent protection. P1, cold remount lost deletion history. P2, stale PR summary. P3, unverifiable timing and red-on-main claims. All four were superseded by #18700, and the current body no longer makes those claims. - 2026-09-21,
545ee8a83d. Cold-remount provenance, forbidden test-double assertions, and red cloud agent E2E. The first two were superseded by #18700. The cloud agent suites passed on later heads, and no check has failed at this SHA. - 2026-09-21,
72bd0363f1. P2, inaccurate description of the limitation boundary; the review also saiddoc_resetdid not re-arm protection. Both were superseded by #18700. - 2026-09-22,
1d2f5ea1cf. P3 Nit, overstated reload verification in the body; tab close deleted history. The body was rewritten, and the history store was removed with #18700. - 2026-09-22,
328c1ce93cand supplement. A TTL reintroduced resurrection; the lifecycle regression stopped short; P2, mixed scope; P3 Nit, transcript-style body. The TTL and lifecycle code were superseded by #18700. The scope is now 7 reset-contract files, and the body is concise. - 2026-09-23,
e0ad99aa73, 2026-09-24,bc0df27762, and 2026-09-24,bb2648bda6. Provenance compaction, then cross-tab loss; exact-head cloud failure from seven startup nodes; P2, 29-file mixed scope, raised three times; P2, stale body, raised three times; P3 Nit, body length and attribution residue, raised three times. The provenance code was superseded by #18700. Blank-startup fixes cleared the cloud failure. Scope and body were resolved ate42f8863ec. - 2026-09-24,
99977c702dand 2026-09-24,58eb368286. These approvals were later dismissed. They included fixes I pushed and structural follow-ups: oversizedapplyQueuedFrameandstartAgentCrdtFollower, append-only provenance, inferred link provenance, a 2,000+ line adapter suite, and reload-test staging. All of those were superseded by #18700, which removed the adapter, history store, and classifier. - 2026-09-29,
4e31c7623c. The merge broke browser typecheck with a duplicateresyncWidget; P2, scattered first-frame lifecycle ownership; P2, append-only storage; P2, 2,939-line integration suite; P3, private-seam test names; unverified P3, an unpinned green-check claim.6c6dca8630fixed the typecheck break; #18700 superseded the rest. - 2026-09-29,
6c6dca8630. This review carried nine findings; its inline record below shows how each ended. - 2026-09-29,
bd058252fb. P1,Claude-Sessiontrailer. P2, duplicated generated reset contract. P2, false claim that consumers derived lineage fromseq. P3, missing report rate limit. P3 Nit, robot footer and session URL.224387688afixed the contract and PR-body items (reply);e42f8863ecfixed attribution. - 2026-09-29,
224387688a. P1, the attribution check still failed on Claude/Amp commit identities andAmp-Thread-IDtrailers; I also contested the approval's green-check claim. The single human-authored commit fixed this ate42f8863ec, and the check passes.
Other reviewers' findings
- christian-byrne raised findings in
e53608ea03,355cbd45f5, andbd058252fb. These covered requiredlineage_seq, which is fixed and retained. They also covered replacement coverage, reload catch-up, and provenance cleanup/link provenance, all superseded by #18700. Two redundant setup comments were removed, and the attribution trailer is fixed ate42f8863ec.
Inline findings
2026-09-21 review at a7f2c3c84a
- P1,
ecsFollowerAdapter.ts:nullhere does not defer the destructive reconcile.. Ended: author answered (reply); code later superseded by #18700. - P3,
agentConversationFixture.ts: This creates a third hand-maintained copy of the nine-field blank workflow, alongsidedefaultGraph.tsand `custom…. Ended: author answered (reply); code later superseded by #18700. - P2,
agentNodeMaterializer.test.ts: This test copies a narrower version of the production classifier.. Ended: author answered (reply); code later superseded by #18700. - P3,
agentNodeMaterializer.test.ts: This preamble restates the test title and the implementation path exercised immediately below.. Ended: author answered (reply); code later superseded by #18700. - P3,
ecsFollowerAdapter.ts:everBoundno longer records whether a workflow was bound.. Ended: author answered (reply); code later superseded by #18700. - P3,
useAgentCrdtFollower.test.ts: This adds the forbiddenas unknown aspattern twice in these tests, even though this file already uses `fromParti…. Ended: author answered (reply); code later superseded by #18700. - P2,
ecsFollowerAdapter.integration.test.ts: The new cases do not falsify an implementation that protects every rebind.. Ended: author answered (reply); code later superseded by #18700.
2026-09-21 review at 545ee8a83d
- P1,
useAgentCrdtFollower.ts: This still cannot distinguish a remotely deleted node after a cold remount from a genuinely hand-added node.. Ended: author answered (reply); code later superseded by #18700. - P1,
agentConversationFixture.ts: The shared fixture path changed here, and the current cloud job now fails five agent tests after all retries:. Ended: author answered (reply); code later superseded by #18700. - P2,
useAgentCrdtFollower.test.ts: The two new fixtures still use the repository-forbiddenas unknown asescape hatch and do not satisfy `Materializ…. Ended: author answered (reply); code later superseded by #18700. - P2,
ecsFollowerAdapter.integration.test.ts: This proves the adapter half with a fixed callback, whileuseAgentCrdtFollower.test.tsproves the classifier half…. Ended: author answered (reply); code later superseded by #18700. - P2,
ecsFollowerAdapter.integration.test.ts: This describe block is now 407 lines and repeats the same mutation store, host, follower, adapter, frame, and teard…. Ended: author answered (reply); code later superseded by #18700. - P3,
ecsFollowerAdapter.ts: Protection completion is represented twice:. Ended: superseded by #18700, reply. - P3,
ecsFollowerAdapter.ts: This hook returns both node and required incident-link IDs, solocalOnlyNodeIdshides half of its contract and co…. Ended: superseded by #18700, reply. - P2,
ecsFollowerAdapter.ts: This comment says only the first full reconcile asks the hook at all, but null and rejected attempts deliberately l…. Ended: superseded by #18700, reply. - P3,
blankWorkflowGraph.ts: This literal has eight fields, not nine.. Ended: superseded by #18700, reply.
2026-09-21 review at 72bd0363f1
- P1,
agentCrdtDocHistoryStore.ts: This moves provenance across composable remounts but still loses it on a new page/Pinia lifetime, as the comment ac…. Ended: author answered (reply); code later superseded by #18700. - P1,
agentConversationFixture.ts: The affected cloud lane is still red at this head:. Ended: author answered (reply); code later superseded by #18700. - P2,
ecsFollowerAdapter.integration.test.ts: This still reimplements the production seam with localeverSeenDocNodeIds,liveNodeIds, andlocalOnlyGraphIds…. Ended: author answered (reply); code later superseded by #18700. - P1,
ecsFollowerAdapter.ts:doc_resetstarts a new document lineage, butclearForReset()never removes this workflow ID, so the new lineage…. Ended: author answered (reply); code later superseded by #18700. - P2,
agentConversationFixture.ts: This is described as a startup interception, but the accessor and wrapper remain for the page lifetime and rewrite…. Ended: author answered (reply); code later superseded by #18700. - P2,
agentCrdtDocHistoryStore.ts: This map and each per-workflow set only grow untildoc_reset;. Ended: author answered (reply); code later superseded by #18700. - P3,
ecsFollowerAdapter.integration.test.ts: These three adjacent helper JSDocs only paraphrase the function names and next few statements (mint, encode, regi…. Ended: author answered (reply); code later superseded by #18700. - P3,
agentCrdtDocHistoryStore.ts: This references the removedlocalOnlyNodeIdssymbol.. Ended: author answered (reply); code later superseded by #18700.
2026-09-22 review at 1d2f5ea1cf
- P1,
agentWorkflowTabBindingStore.ts: This reintroduces the deletion-resurrection gap when a workflow is closed and later rebound.. Ended: author answered (reply); code later superseded by #18700. - P1,
agentConversationFixture.ts: This startup interception currently leaves the required cloud suite deterministically red.. Ended: author answered (reply); code later superseded by #18700. - P2,
ecsFollowerAdapter.integration.test.ts: This still reconstructs the production classifier inside the test.. Ended: author answered (reply); code later superseded by #18700. - P2,
agentCrdtDocHistoryStore.ts: This trusts the shape of persisted browser data.. Ended: author answered (reply); code later superseded by #18700. - P2,
agentCrdtDocHistoryStore.ts: The persisted set only grows for the lifetime of a workflow lineage.. Ended: author answered (reply); code later superseded by #18700. - P3,
agentConversationFixture.ts: RestoringloadGraphDatafixes the method wrapper, butwindow.appremains this accessor for the page lifetime.. Ended: author answered (reply); code later superseded by #18700. - P3,
useAgentCrdtFollower.ts: This comment is now stale:. Ended: author answered (reply); code later superseded by #18700. - P3,
agentConversationFixture.ts: This block records the implementation's review history and repeats details already encoded by the helper and nearby…. Ended: author answered (reply); code later superseded by #18700.
2026-09-22 review at 328c1ce93c
- P1,
agentCrdtDocHistoryStore.ts: This TTL only postpones the resurrection bug.. Ended: author answered (reply); code later superseded by #18700. - P2,
agentWorkflowTabBindingStore.test.ts: Despite the name, this test closes a tab and then manually calls the history store and classifier;. Ended: author answered (reply); code later superseded by #18700. - P3,
useAgentCrdtFollower.ts: The previous stale claim remains here:. Ended: author answered (reply); code later superseded by #18700.
2026-09-22 review at 328c1ce93c
- P3,
useAgentCrdtFollower.ts: The comment explicitly says this production symbol is exported so a test can call it.. Ended: author answered (reply); code later superseded by #18700. - P3,
ecsFollowerAdapter.integration.test.ts: This test spends roughly forty lines narrating its own setup and private implementation state (`protectedWorkflowId…. Ended: author answered (reply); code later superseded by #18700. - P3,
agentCrdtDocHistoryStore.test.ts: This is not a "genuinely new page load";. Ended: author answered (reply); code later superseded by #18700. - P3,
agentCrdtDocHistoryStore.test.ts: TwouseStore()calls under the same active Pinia return the same store instance by Pinia contract, so this tests…. Ended: author answered (reply); code later superseded by #18700. - P3,
agentWorkflowTabBindingStore.ts: Nine lines explain a call that deliberately does not exist, duplicate the history-store documentation, and end with…. Ended: author answered (reply); code later superseded by #18700. - P3,
agentConversationFixture.ts: This comment paraphrases the immediately followingObject.definePropertyin six lines, after the method-level com…. Ended: author answered (reply); code later superseded by #18700.
2026-09-23 review at e0ad99aa73
- P1,
useAgentCrdtFollower.ts: This checkpoint corrupts the provenance classifier in both directions.. Ended: superseded by #18700, reply. - P1,
agentConversationFixture.ts: The exact-head cloud suite fails deterministically because this blank-startup hook is only installed by selected co…. Ended: superseded by #18700, reply. - P2,
agentCloseReopenRemoteDelete.spec.ts: This does not perform the workflow-tab close/reopen claimed by the describe text and PR body;. Ended: superseded by #18700, reply. - P2,
agentCrdtDocHistoryStore.ts:remember()rewrites the complete persisted history synchronously for every non-empty doc frame even when no ID ch…. Ended: superseded by #18700, reply. - P3,
MultiAutogrowRealignHarness.ts: This generic blank-startup utility is imported from the large conversation fixture, so an unrelated autogrow harnes…. Ended: superseded by #18700, reply. - P3,
agentCrdtDocHistoryStore.ts: This says a workflow ID "is a lineage ID," butdoc_resetstarts a new lineage under the same workflow ID elsewher…. Ended: superseded by #18700, reply. - P3,
ecsFollowerAdapter.ts: This private one-caller wrapper only hides one guard and two statements;. Ended: superseded by #18700, reply. - P3,
ecsFollowerAdapter.integration.test.ts: This test reuses an adapter afterdestroy()and asserts hook call counts, pinning privateprotectedWorkflowIds…. Ended: superseded by #18700, reply. - P3,
agentLocalOnlyGraphIds.ts: This comment explains why the function was moved for tests rather than documenting its behavioral contract.. Ended: superseded by #18700, reply.
2026-09-24 review at bc0df27762
- P1,
useAgentCrdtFollower.ts: This checkpoint still corrupts provenance in both directions:. Ended: superseded by #18700, reply. - P2,
agentCloseReopenRemoteDelete.spec.ts: This still reloads the page rather than closing and reopening a workflow tab, despite the test and PR claiming that…. Ended: superseded by #18700, reply. - P2,
agentCrdtDocHistoryStore.ts:remember()still changes the unusedupdatedAton every non-empty frame, forcing a synchronous rewrite and seria…. Ended: superseded by #18700, reply. - P3,
agentCrdtDocHistoryStore.ts: A workflow ID is not a lineage ID:. Ended: superseded by #18700, reply. - P3,
ecsFollowerAdapter.ts: This one-caller wrapper hides only one guard and two statements without creating a reusable policy or coherent resp…. Ended: superseded by #18700, reply. - P3,
ecsFollowerAdapter.integration.test.ts: This test reuses an adapter afterdestroy()and asserts hook call counts, pinning private bookkeeping rather than…. Ended: superseded by #18700, reply. - P3,
agentLocalOnlyGraphIds.ts: This still narrates the review-driven extraction rather than the function's behavioral contract.. Ended: superseded by #18700, reply. - P2,
agentInsertWorkflowCanvasResult.spec.ts: Installing the init script before navigation fixes the seven-node failure, butbootAgentApp()returns after `exte…. Ended: superseded by #18700, reply. - P3,
blankStartupGraph.ts: This six-line type explanation is unnecessary and partly inaccurate:. Ended: superseded by #18700, reply. - P3,
agentInsertWorkflowCanvasResult.spec.ts: This eight-line block duplicates the shared helper's contract immediately before a self-describing call.. Ended: superseded by #18700, reply.
2026-09-24 review at bb2648bda6
- P1,
agentCrdtDocHistoryStore.ts: This read-modify-write is monotonic only inside one Pinia instance.. Ended: closed, reply. - P2,
agentConversationFixture.ts: `state:. Ended: closed, reply. - P2,
agentConversationFixture.ts: This duplicatesWorkflowHelper.openPersistedWorkflow()but omits its workflow-idle and node-readiness waits.. Ended: closed, reply. - P2,
agentCloseReopenRemoteDelete.spec.ts: This does not wait for the workflow tab to finish closing.. Ended: closed, reply. - P2,
useAgentCrdtFollower.test.ts: This test claims reload-before-add-ack coverage but manually calls the commit hook, unmounts, and remounts against…. Ended: superseded by #18700, reply. - P3,
useAgentCrdtFollower.test.ts: The new cases force bothadapterState.intent!andlocalOnly!instead of proving setup succeeded.. Ended: closed, reply. - P3,
agentCloseReopenRemoteDelete.spec.ts: This multi-phase Playwright case performs save, remote deletion, close, reopen, selection, resubscription, catch-up…. Ended: closed, reply. - P3,
ecsFollowerAdapter.ts: The previous one-caller wrapper was inlined, then this four-argument one-caller wrapper moved one guard and one ass…. Ended: author answered (reply); code later superseded by #18700. - P2,
useAgentCrdtFollower.ts: This comment describesids.nodeIds, but this function no longer has anidsparameter.. Ended: closed, reply. - P2,
agentCrdtDocHistoryStore.test.ts: This title says a genuine page reload is covered byagentCloseReopenRemoteDelete.spec.ts, but that spec now close…. Ended: closed, reply. - P3,
useAgentCrdtFollower.test.ts: These new regression cases narrate nearly every setup action and assertion across roughly 25 lines.. Ended: closed, reply. - P3,
agentCloseReopenRemoteDelete.spec.ts: The test now paraphrases almost every self-describing harness call with a multi-line block.. Ended: closed, reply. - P3,
agentCrdtDocHistoryStore.test.ts: This records the field's implementation history and elapsed-time review discussion.. Ended: closed, reply. - P3,
agentCrdtDocHistoryStore.ts: This store-wide retention policy is attached to the plainHistoryEntryshape and repeats the store docblock below…. Ended: closed, reply. - P3,
useAgentCrdtFollower.test.ts: This comment paraphrases the helper body and records a removed type-workaround (no cast needed).. Ended: closed, reply.
2026-09-29 review at 4e31c7623c
- P1,
agentConversationFixture.ts: The merge left a secondresyncWidgetimplementation at line 1181.. Ended: closed, reply. - P2,
docFrameClient.ts:lineage_seqis required by generatedDocResetData, but the normalized contract makes it optional and silently f…. Ended: closed, reply. - P2,
ecsFollowerAdapter.integration.test.ts: This proves replacement does not rearm protection, but starts from an empty semantic graph, so a no-op clear also p…. Ended: closed, reply. - P2,
agentHumanAddReload.spec.ts:subscribeCount()rises when the host sends catch-up, not when the page applies and reconciles it.. Ended: closed, reply. - P3,
useAgentCrdtFollower.ts: This post-commit callback repeats the same document snapshot and provenance write already performed at the start of…. Ended: closed, reply. - P2,
agentCrdtDocHistoryStore.ts: This remains unbounded:. Ended: closed, reply. - P2,
agentLocalOnlyGraphIds.ts: Link provenance is not tracked.. Ended: closed, reply. - P3,
useAgentCrdtFollower.ts: One update snapshots all document node IDs here, again intrackNodeChanges, again for non-agent cleanup, and once…. Ended: closed, reply. - P3,
startupGraph.ts: This overstates the helper:. Ended: closed, reply. - P3,
ecsFollowerAdapter.integration.test.ts: This name catalogs private collaborators rather than behavior.. Ended: closed, reply.
2026-09-29 review at bd058252fb
- P2,
docFrameClient.ts:@comfyorg/ingest-typesalready owns this required field in generatedDocResetData, but this parser, the shared…. Ended: closed, reply.
2026-09-29 review at 224387688a
- P3,
followerSubscription.test.ts: Delete this comment.. Ended: author answered (reply), deleted ate42f8863ec.
Checks at e42f8863ec
- Focused Vitest for
docFrameClient.numeric,docFrameClient, andfollowerSubscription: 3 files, 89 tests passed. pnpm typecheckandpnpm typecheck:browserpassed.- ESLint and oxfmt passed on the 7 changed files.
playwright test --listdiscovered all 4 tests in the 2 changed specs..github/scripts/check-ai-co-authors.shexited 0 for the merge base and this head.- The branch merges cleanly into current main, and main has not changed the PR files or
packages/ingest-typessince the merge base. - The five review lanes (correctness, maintainability, idiomaticity, tests, documentation) ran against the identical tree at
224387688a. Their only finding was the deleted comment, so I reused that evidence here. - Hosted CI at review time: 73 successful, 40 skipped, none failed. 23 were still running, including the Chromium and cloud Playwright shards, ecosystem matrix, and perf tests.
## Summary Remove the AI commit attribution check that blocks PRs such as Comfy-Org#18277. ## Changes - **What**: Delete the Check AI Co-Authors workflow and its now-unused script. ## Review Focus Confirm the attribution check should no longer be required. Failed run: https://github.com/Comfy-Org/ComfyUI_frontend/actions/runs/36543901331/job/109325571458?pr=18277
christian-byrne
left a comment
There was a problem hiding this comment.
Verdict: approved.
The current head keeps lineage_seq required at the wire boundary, rejects missing or invalid values before dispatch, and leaves the existing follower and subscription unchanged for malformed resets. Shared fixtures and both browser-test producers are typed against the generated ingest contract.
Verified on e42f8863ecc4f5df7e48e7febab4a907ba798e97:
- Parser, numeric-domain, and follower-subscription suites: 89 tests passed.
- Browser-test typecheck passed.
- Targeted lint passed for all seven changed files.
- All required GitHub checks pass.
Summary
Make the local
doc_resetparser enforce the generated contract, which requireslineage_seq, and derive its accepted fields and valid test payloads from the generatedDocResetDatatype.Before:
parseServerDocFrameaccepted adoc_resetframe that omittedlineage_seq, even though the generatedDocResetDatacontract in@comfyorg/ingest-typesrequires it. The parser, the shared test fixture, and the hand-built E2E reset payloads each kept their own copy of the field list, so the omission compiled everywhere and only surfaced at runtime.After: a
doc_resetframe parses only whenlineage_seqis a valid sequence (a non-negative integer;0is allowed and can differ fromseq), and the parsed reset carrieslineageSeq. A frame without it is rejected before dispatch and leaves the followed doc and subscription untouched.DocFrameClientreports only the first malformed frame of each type per client asagent_crdt_invalid_server_frame; later malformed resets on the same client are dropped without another report.How: the
doc_resetbranch reads the wire payload throughPartial<Record<keyof DocResetData, unknown>>, so field names come from the generated type while values stayunknownuntil validated. ThedocResetFramefixture default and the two E2E reset payloads are checked withsatisfies DocResetData/satisfies DocResetFrame. Renaming or removing a requiredDocResetDatafield now failspnpm typecheckandpnpm typecheck:browserat all four sites.Changes
src/workbench/extensions/agent/crdt/docFrameClient.tsrequireslineage_seqondoc_reset, exposes it aslineageSeq, and keys the reset branch offkeyof DocResetData.__fixtures__/docFrameClient.ts(valid default typed asDocResetData),docFrameClient.test.ts,docFrameClient.numeric.test.ts,followerSubscription.test.ts.browser_tests/tests/agent/agentCanvasBlanksDuringRunTurn.spec.tsandbrowser_tests/tests/agent/agentFirstMintResetSweepsTemplate.spec.tsaddlineage_seqto their hand-built reset frames and type them asDocResetFrame.doc_resetwithoutlineage_seqnow has those frames dropped instead of treated as a lineage break.Review Focus
This PR originally protected hand-added nodes through the first CRDT reconcile and persisted doc provenance in localStorage. That work was superseded by #18700 (#18700), which removed the reconcile layer, so after the merge of main the net diff against main is the
lineage_seqrequirement above. Please confirm that every host path emittingdoc_resetsendslineage_seqbefore this lands.