Repository navigation
FE-2469 feat(billing): project operation state the host reads off the SDK store - #18036
Conversation
The surfaces that drive a checkout read four things off billingOperationStore that billingSdkStore does not expose: whether a subscribe is setting up, which subscription is waiting on the customer, an operation looked up by id, and whether anything is pending. With the subscription rail on, the operation lives in the SDK lifecycle, so those reads come from a store nothing writes. Project them from the lifecycle snapshot in the shapes the current consumers read, mirroring billingOperationStore's own predicates — including where it scopes to the active workspace and where it does not.
🎨 Storybook: ✅ Built — View Storybook🎭 Playwright: ✅ 2039 passed, 0 failed · 1 flaky📊 Browser Reports
📦 Bundle: 9.92 MB gzip 🔴 +396 BDetailsSummary
Category Glance App Entry Points — 3.77 kB (baseline 3.77 kB) • ⚪ 0 BMain entry bundles and manifests
Status: 1 added / 1 removed Graph Workspace — 1.45 MB (baseline 1.45 MB) • ⚪ 0 BGraph editor runtime, canvas, workflow orchestration
Status: 2 added / 2 removed / 1 unchanged Views & Navigation — 132 kB (baseline 132 kB) • ⚪ 0 BTop-level views, pages, and routed surfaces
Status: 14 added / 14 removed / 4 unchanged Panels & Settings — 637 kB (baseline 637 kB) • ⚪ 0 BConfiguration panels, inspectors, and settings screens
Status: 10 added / 10 removed / 17 unchanged User & Accounts — 37.9 kB (baseline 37.9 kB) • ⚪ 0 BAuthentication, profile, and account management bundles
Status: 6 added / 6 removed / 5 unchanged Editors & Dialogs — 126 kB (baseline 126 kB) • ⚪ 0 BModals, dialogs, drawers, and in-app editors
Status: 7 added / 7 removed / 1 unchanged UI Components — 124 kB (baseline 124 kB) • ⚪ 0 BReusable component library chunks
Status: 7 added / 7 removed / 9 unchanged Data & Services — 3.86 MB (baseline 3.85 MB) • 🔴 +2.12 kBStores, services, APIs, and repositories
Status: 13 added / 13 removed / 3 unchanged Utilities & Hooks — 565 kB (baseline 565 kB) • ⚪ 0 BHelpers, composables, and utility bundles
Status: 19 added / 19 removed / 19 unchanged Vendor & Third-Party — 17.4 MB (baseline 17.4 MB) • ⚪ 0 BExternal libraries and shared vendor chunks Status: 18 unchanged Other — 17.7 MB (baseline 17.7 MB) • ⚪ 0 BBundles that do not match a named category
Status: 73 added / 73 removed / 224 unchanged ⚡ Performance Report
Show regressions
All metrics
Historical variance (last 15 runs)
Trend (last 15 commits on main)
Raw data{
"timestamp": "2026-09-18T19:32:05.189Z",
"gitSha": "4c4c1d301cbcd44e077031e427d197c1d7cdbbc7",
"branch": "jaewon/fe-2469-project-legacy-operation-store-state",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 2021.4839999999867,
"styleRecalcs": 8,
"styleRecalcDurationMs": 5.989,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 397.021,
"heapDeltaBytes": 648856,
"heapUsedBytes": 62751748,
"domNodes": -264,
"jsHeapTotalBytes": 4710400,
"scriptDurationMs": 9.500000000000002,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-idle",
"durationMs": 1990.9609999999702,
"styleRecalcs": 8,
"styleRecalcDurationMs": 6.104,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 428.54499999999996,
"heapDeltaBytes": 2724012,
"heapUsedBytes": 64756516,
"domNodes": -263,
"jsHeapTotalBytes": 4710400,
"scriptDurationMs": 8.400999999999998,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1726.5500000000316,
"styleRecalcs": 72,
"styleRecalcDurationMs": 24.047999999999995,
"layouts": 14,
"layoutDurationMs": 4.289,
"taskDurationMs": 767.75,
"heapDeltaBytes": -6496456,
"heapUsedBytes": 55376256,
"domNodes": -263,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 137.089,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1767.6139999999805,
"styleRecalcs": 76,
"styleRecalcDurationMs": 23.817999999999998,
"layouts": 14,
"layoutDurationMs": 4.6160000000000005,
"taskDurationMs": 774.3610000000001,
"heapDeltaBytes": 12437952,
"heapUsedBytes": 74404536,
"domNodes": -261,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 133.69400000000002,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1723.8790000000108,
"styleRecalcs": 30,
"styleRecalcDurationMs": 13.334999999999999,
"layouts": 6,
"layoutDurationMs": 1.116,
"taskDurationMs": 313.821,
"heapDeltaBytes": 2155988,
"heapUsedBytes": 64243424,
"domNodes": 75,
"jsHeapTotalBytes": 4718592,
"scriptDurationMs": 11.244,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1662.7260000000206,
"styleRecalcs": 31,
"styleRecalcDurationMs": 12.579,
"layouts": 6,
"layoutDurationMs": 1.2190000000000003,
"taskDurationMs": 297.543,
"heapDeltaBytes": 2130160,
"heapUsedBytes": 64279396,
"domNodes": 75,
"jsHeapTotalBytes": 4718592,
"scriptDurationMs": 10.082999999999998,
"eventListeners": 15,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 598.906999999997,
"styleRecalcs": 11,
"styleRecalcDurationMs": 6.9750000000000005,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 398.04699999999997,
"heapDeltaBytes": 17619908,
"heapUsedBytes": 79623820,
"domNodes": 18,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 141.317,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 589.0580000000227,
"styleRecalcs": 11,
"styleRecalcDurationMs": 6.229999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 393.0400000000001,
"heapDeltaBytes": 17908688,
"heapUsedBytes": 79867824,
"domNodes": 18,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 141.485,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-idle",
"durationMs": 2031.7489999999907,
"styleRecalcs": 6,
"styleRecalcDurationMs": 5.306000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 556.3340000000001,
"heapDeltaBytes": 2169420,
"heapUsedBytes": 82048260,
"domNodes": -249,
"jsHeapTotalBytes": -8192,
"scriptDurationMs": 37.241,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-idle",
"durationMs": 2022.0679999999902,
"styleRecalcs": 5,
"styleRecalcDurationMs": 3.748000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 519.1080000000001,
"heapDeltaBytes": 2951216,
"heapUsedBytes": 82575308,
"domNodes": -249,
"jsHeapTotalBytes": -794624,
"scriptDurationMs": 32.61499999999999,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-pan",
"durationMs": 2312.234999999987,
"styleRecalcs": 67,
"styleRecalcDurationMs": 13.898999999999997,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1614.1940000000002,
"heapDeltaBytes": -1955304,
"heapUsedBytes": 79004724,
"domNodes": -253,
"jsHeapTotalBytes": 1003520,
"scriptDurationMs": 779.363,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66999999999998,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "large-graph-pan",
"durationMs": 2146.2189999999737,
"styleRecalcs": 65,
"styleRecalcDurationMs": 9.291999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1385.109,
"heapDeltaBytes": 18102648,
"heapUsedBytes": 82542968,
"domNodes": -253,
"jsHeapTotalBytes": 749568,
"scriptDurationMs": 693.1089999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-zoom",
"durationMs": 3538.4950000000117,
"styleRecalcs": 60,
"styleRecalcDurationMs": 11.945999999999998,
"layouts": 60,
"layoutDurationMs": 15.41,
"taskDurationMs": 1622.361,
"heapDeltaBytes": 10013956,
"heapUsedBytes": 73865424,
"domNodes": -263,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 721.133,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-zoom",
"durationMs": 3442.794000000049,
"styleRecalcs": 60,
"styleRecalcDurationMs": 10.686000000000002,
"layouts": 60,
"layoutDurationMs": 13.829999999999998,
"taskDurationMs": 1505.8280000000002,
"heapDeltaBytes": 8006224,
"heapUsedBytes": 73655736,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 677.347,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 2968.873999999971,
"styleRecalcs": 41,
"styleRecalcDurationMs": 6.033,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1793.6109999999999,
"heapDeltaBytes": 18199916,
"heapUsedBytes": 84044064,
"domNodes": 2,
"jsHeapTotalBytes": 417792,
"scriptDurationMs": 883.0899999999999,
"eventListeners": 180,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "legacy-node-drag",
"durationMs": 3022.5060000000212,
"styleRecalcs": 43,
"styleRecalcDurationMs": 7.752000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1821.18,
"heapDeltaBytes": -14198836,
"heapUsedBytes": 69812612,
"domNodes": -257,
"jsHeapTotalBytes": 1040384,
"scriptDurationMs": 886.024,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2032.3130000000447,
"styleRecalcs": 6,
"styleRecalcDurationMs": 4.3709999999999996,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 530.945,
"heapDeltaBytes": 1937244,
"heapUsedBytes": 82072428,
"domNodes": -251,
"jsHeapTotalBytes": -532480,
"scriptDurationMs": 31.080000000000002,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "minimap-idle",
"durationMs": 2009.1709999999239,
"styleRecalcs": 4,
"styleRecalcDurationMs": 2.5969999999999986,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 482.07599999999996,
"heapDeltaBytes": 19712620,
"heapUsedBytes": 84341396,
"domNodes": -249,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 33.44899999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 554.1419999999562,
"styleRecalcs": 47,
"styleRecalcDurationMs": 8.999000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 369.121,
"heapDeltaBytes": 13476532,
"heapUsedBytes": 76761924,
"domNodes": 20,
"jsHeapTotalBytes": 5242880,
"scriptDurationMs": 125.79599999999999,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 542.2730000000229,
"styleRecalcs": 48,
"styleRecalcDurationMs": 9.36,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 351.84400000000005,
"heapDeltaBytes": 13342036,
"heapUsedBytes": 76696060,
"domNodes": 22,
"jsHeapTotalBytes": 6029312,
"scriptDurationMs": 117.85,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-idle",
"durationMs": 1992.0930000000112,
"styleRecalcs": 10,
"styleRecalcDurationMs": 8.017,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 440.85800000000006,
"heapDeltaBytes": 16201036,
"heapUsedBytes": 79011436,
"domNodes": -276,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 9.255999999999998,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-idle",
"durationMs": 2020.4089999999724,
"styleRecalcs": 10,
"styleRecalcDurationMs": 6.973999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 426.325,
"heapDeltaBytes": 16267680,
"heapUsedBytes": 79425904,
"domNodes": -274,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 8.176999999999998,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1706.9470000000138,
"styleRecalcs": 77,
"styleRecalcDurationMs": 24.778,
"layouts": 18,
"layoutDurationMs": 5.244999999999999,
"taskDurationMs": 723.247,
"heapDeltaBytes": 16798968,
"heapUsedBytes": 79871212,
"domNodes": -274,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 91.86,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1692.075999999929,
"styleRecalcs": 78,
"styleRecalcDurationMs": 23.458000000000002,
"layouts": 18,
"layoutDurationMs": 4.949999999999999,
"taskDurationMs": 689.8420000000001,
"heapDeltaBytes": -5035948,
"heapUsedBytes": 58240700,
"domNodes": -273,
"jsHeapTotalBytes": 6283264,
"scriptDurationMs": 87.20800000000001,
"eventListeners": -198,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-transition-enter",
"durationMs": 1016.9860000000881,
"styleRecalcs": 17,
"styleRecalcDurationMs": 24.663999999999998,
"layouts": 5,
"layoutDurationMs": 9.706000000000001,
"taskDurationMs": 793.4060000000003,
"heapDeltaBytes": 19973804,
"heapUsedBytes": 102820548,
"domNodes": 13587,
"jsHeapTotalBytes": 11010048,
"scriptDurationMs": 24.298000000000002,
"eventListeners": 2383,
"totalBlockingTimeMs": 134,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "viewport-pan-sweep",
"durationMs": 8203.319000000021,
"styleRecalcs": 247,
"styleRecalcDurationMs": 38.547,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 5225.571,
"heapDeltaBytes": -13791004,
"heapUsedBytes": 66201584,
"domNodes": -246,
"jsHeapTotalBytes": 3399680,
"scriptDurationMs": 2425.5209999999997,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "viewport-pan-sweep",
"durationMs": 8153.753999999935,
"styleRecalcs": 248,
"styleRecalcDurationMs": 35.088,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 4994.01,
"heapDeltaBytes": -598960,
"heapUsedBytes": 79288672,
"domNodes": -246,
"jsHeapTotalBytes": -307200,
"scriptDurationMs": 2304.201,
"eventListeners": -152,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.80000000000109
},
{
"name": "vue-large-graph-idle",
"durationMs": 14585.800000000007,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 13930.349,
"heapDeltaBytes": -53285172,
"heapUsedBytes": 182728616,
"domNodes": -8061,
"jsHeapTotalBytes": -17047552,
"scriptDurationMs": 224.366,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-idle",
"durationMs": 14392.77000000004,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 13714.533999999998,
"heapDeltaBytes": -67789368,
"heapUsedBytes": 168819360,
"domNodes": -8061,
"jsHeapTotalBytes": -12128256,
"scriptDurationMs": 220.69,
"eventListeners": -14928,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.223333333333358,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-pan",
"durationMs": 13533.359000000019,
"styleRecalcs": 67,
"styleRecalcDurationMs": 13.428999999999997,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 13062.198,
"heapDeltaBytes": -63575272,
"heapUsedBytes": 166005980,
"domNodes": -8061,
"jsHeapTotalBytes": -10031104,
"scriptDurationMs": 752.928,
"eventListeners": -14928,
"totalBlockingTimeMs": 19,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-pan",
"durationMs": 17240.06699999995,
"styleRecalcs": 138,
"styleRecalcDurationMs": 17.350000000000005,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 16895.711,
"heapDeltaBytes": -26360128,
"heapUsedBytes": 188604720,
"domNodes": -8061,
"jsHeapTotalBytes": -17309696,
"scriptDurationMs": 793.419,
"eventListeners": -14928,
"totalBlockingTimeMs": 15,
"frameDurationMs": 17.220000000000073,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-node-resize-workload",
"durationMs": 5110.094000000004,
"styleRecalcs": 20,
"styleRecalcDurationMs": 74.96399999999997,
"layouts": 20,
"layoutDurationMs": 25.354000000000003,
"taskDurationMs": 5029.182,
"heapDeltaBytes": -50744464,
"heapUsedBytes": 179557296,
"domNodes": -8061,
"jsHeapTotalBytes": 15659008,
"scriptDurationMs": 96.565,
"eventListeners": -14928,
"totalBlockingTimeMs": 277,
"frameDurationMs": 17.779999999999987,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-node-resize-workload",
"durationMs": 5144.981999999914,
"styleRecalcs": 20,
"styleRecalcDurationMs": 68.35099999999997,
"layouts": 20,
"layoutDurationMs": 24.314000000000004,
"taskDurationMs": 5076.768000000001,
"heapDeltaBytes": -48459552,
"heapUsedBytes": 173938208,
"domNodes": -8062,
"jsHeapTotalBytes": 11640832,
"scriptDurationMs": 92.95900000000002,
"eventListeners": -14928,
"totalBlockingTimeMs": 198,
"frameDurationMs": 17.779999999999987,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-zoom-out-idle",
"durationMs": 7337.833999999986,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7000.078,
"heapDeltaBytes": -18127660,
"heapUsedBytes": 169690268,
"domNodes": 0,
"jsHeapTotalBytes": -28049408,
"scriptDurationMs": 112.86699999999993,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.223333333333358,
"p95FrameDurationMs": 16.80000000000291
},
{
"name": "vue-zoom-out-idle",
"durationMs": 7220.180000000028,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 6825.818000000002,
"heapDeltaBytes": -9972068,
"heapUsedBytes": 187530308,
"domNodes": -1,
"jsHeapTotalBytes": -24903680,
"scriptDurationMs": 161.52,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.223333333333237,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "workflow-execution",
"durationMs": 458.6069999999154,
"styleRecalcs": 16,
"styleRecalcDurationMs": 21.543,
"layouts": 4,
"layoutDurationMs": 2.803,
"taskDurationMs": 123.52100000000002,
"heapDeltaBytes": 4913092,
"heapUsedBytes": 66316940,
"domNodes": 126,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 9.939,
"eventListeners": 81,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "workflow-execution",
"durationMs": 450.04000000005817,
"styleRecalcs": 12,
"styleRecalcDurationMs": 21.084,
"layouts": 3,
"layoutDurationMs": 1.1460000000000001,
"taskDurationMs": 106.13300000000001,
"heapDeltaBytes": 4746036,
"heapUsedBytes": 66265128,
"domNodes": 117,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 10.223,
"eventListeners": 65,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
}
]
}🔬 E2E Coverage
Top 10 files by uncovered lines
|
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: ASSERTIVE Plan: Advanced Run ID: 📒 Files selected for processing (4)
Included review availability: Your plan provides up to 8 included reviews per hour; 7 remain after this review. 📝 WalkthroughWalkthroughThe billing SDK now projects lifecycle operation states into billing records. It exposes pending, setup, customer-action, and lookup state. Tests cover record projection and workspace-scoped store behavior. ChangesBilling operation state
Priority: ⬇️ Low Estimated code review effort: 3 (Moderate) | ~25 minutes Change: Refactor Sequence Diagram(s)sequenceDiagram
participant BillingOperationStore
participant projectOperationRecord
participant TeamWorkspaceStore
participant BillingSdkStore
BillingOperationStore->>projectOperationRecord: Project lifecycle operation states
projectOperationRecord-->>BillingSdkStore: Return operation records
TeamWorkspaceStore-->>BillingSdkStore: Provide activeWorkspaceId
BillingSdkStore-->>BillingSdkStore: Derive pending and customer-action state
Suggested reviewers: Merge Risk: ⚪ Minimal · up to The new billing operation projections preserve the intended lifecycle and workspace-scoped behavior, with no actionable merge-blocking risk identified. 🚥 Pre-merge checks | ✅ 7✅ Passed checks (7 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## main #18036 +/- ##
==========================================
+ Coverage 82.76% 83.79% +1.02%
==========================================
Files 2935 2242 -693
Lines 144407 127591 -16816
Branches 37118 34321 -2797
==========================================
- Hits 119519 106912 -12607
+ Misses 23631 20389 -3242
+ Partials 1257 290 -967
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 704 files with indirect coverage changes 🚀 New features to boost your workflow:
|
There was a problem hiding this comment.
Approving. The load-bearing claim here is the comment above the projections —
The four below mirror
billingOperationStore's own predicates, including where it does and does not scope to the active workspace
— so we checked all four against main rather than reading it, because "mirrors the old thing" is the kind of claim that is either exactly true or quietly not.
It is exactly true.
| predicate | main (billingOperationStore) |
this PR (billingSdkStore) |
|---|---|---|
hasPendingOperations |
status === 'pending', unscoped |
same, unscoped |
isSettingUp |
pending + not requires_action + not failed_retryable + type === 'subscription' + workspaceId === activeWorkspaceId |
same five conjuncts |
subscriptionActionOperation |
type === 'subscription' + workspaceId === activeWorkspaceId + needsCustomerAttention |
same three |
getOperation(opId) |
operations.get(opId), unscoped |
find(r => r.opId === opId), unscoped |
Including the asymmetry — two scoped, two not. Naming that in a test (finds an operation in another workspace, leaving the scope check to the caller) is better than leaving it to be rediscovered, and the reasoning is right: a consumer switching rails must not get a different answer on the other side.
Dropping superseded states in projectOperationRecord is consistent with readOnRail treating a SUPERSEDED read as nothing-to-publish, and the comment says so. Same rule in both places is worth more than the rule itself.
One finding, and it is about a coupling this PR creates rather than a defect in it.
Both scoped predicates compare workspaceId alone. That matches main today — which is exactly the problem, because main today is what #15431 exists to fix. That PR replaces the bare comparison with
operation.workspaceId === workspaceStore.activeWorkspaceId &&
operation.transitionGeneration === workspaceStore.workspaceTransitionGenerationon the grounds that an A→B→A round trip returns to the same workspaceId and a stale operation is wrongly re-adopted. #15431 is still open and CONFLICTING, so main has not had that correction yet.
So after this lands there are two implementations of the same predicate that have to change together, and nothing says so. If #15431 is revived and adds the generation check to billingOperationStore, the SDK projections keep the weaker comparison and the rails diverge precisely at a workspace switch — the case the generation check was added for. If #15431 is closed instead, they stay correctly in sync.
Not asking you to add the generation check here: mirroring main is the right call for an additive PR, and pre-empting an unmerged fix would be worse. What would close it is one line in the comment naming #15431 as the thing that must move both sides, so the second implementation is not invisible to whoever does that work.
On the shape of the PR — additive, +329/−0, consumers arriving later. We raised that shape as a tracking gap at tick 108 across #13762, #14091 and #14811, where the wiring follow-up had no recorded existence. This one names FE-2480, FE-2482, FE-2483/FE-2484 and FE-2474 in the description, which is the remedy. Worth saying we noticed, and that we will treat the projections as unfinished rather than dead if those tickets stall.
CI clean. Approved.
## Why The usage log is the last billing read still bound to `workspaceApi`. PR 2 put the reader in `account-core` and exposed `readEvents` on `billingSdkStore`; nothing calls it. This PR is that call site. With `billing_sdk_topup_enabled` or `billing_sdk_subscription_enabled` on, every other billing surface reads through the SDK and this one table still reaches for the workspace client directly — the one read left off the rail the other five already take. ## Root cause `UsageLogsTable.vue` branches on `shouldUseWorkspaceBilling` alone: the workspace client, or `customerEventService.getMyEvents` for the legacy `/customers` feed. There is no third arm, because when the readers landed (#17940 / #17942 / #17946) there was no core reader for `/billing/events` to point one at. That absence outlived the reader. `BillingReadRail` — the contract those five reads resolve through — still declares five members, and FE-2467 deliberately stopped at `billingSdkStore`, leaving the interface widening to whoever brought the caller. So even with the store method in place, the rail type does not expose it: the switch could not be written without this change. ## How **`readEvents` joins `BillingReadRail`** (`useBillingReadRail.ts:28`). It is the only member taking options, because it is the only read that asks for one page rather than one route per scope. **The workspace arm resolves through the rail** (`UsageLogsTable.vue:150`–`:155`, called at `:168`): ```ts const rail = useBillingReadRail() return rail === null ? workspaceApi.getBillingEvents(params) : readOnRail(() => rail.readEvents(params)) ``` Two choices inside that: - **The read is closed over, not passed with its arguments.** FE-2479 sketches `readOnRail(rail.readEvents, params)`, but `readOnRail` takes a zero-arg read and `useBillingCapabilities.ts:74` already closes over its options rather than widening the signature. Following the existing pattern keeps `readOnRail` — used by all five other reads — untouched. - **The rail is read once per load and held for it**, inside `readWorkspaceEvents`, per the rule on `useBillingReadRail`: a flag flip mid-read must not start on one client and publish through the other. **The `customerEventService` branch is untouched.** That is the legacy `/customers` endpoint, out of scope per M1-13, and it keeps its own error surface at `:204`. **A superseded rail read gets its own early return** (`:181`). `readOnRail` resolves `undefined` on `SUPERSEDED` rather than throwing, and the existing falsy-response branch renders `credits.loadEventsError`. Without the guard a scope change would reach the customer as "Failed to load activity". `undefined` identifies exactly one thing here: `getMyEvents` returns `CustomerEventsResponse | null` and `getBillingEvents` throws instead of returning falsy, so only a superseded rail read produces it. Nothing is published and nothing is reported — the same stance `useBillingPlans.ts:44` takes on the same result. The `latestLoadToken` guard and the `isPendingTopupCompleted` call keep their order, so a mid-checkout route flip still reports a top-up that landed on the losing fetch. ## AS IS / TO BE **Flag off, nothing changes.** `rail === null` resolves to the same `workspaceApi.getBillingEvents(params)` call the file made before, on the same branch, with the same error handling. **Flag on, the same rows arrive over the SDK transport.** The store's `readEvents` returns `BillingEventsResponse`, the type this table already consumed, so the rendered page is identical — the rail changes who fetches, not what is shown. A before/after capture would be two identical tables, so none is attached. The one new behaviour is the superseded case, and it is reachable only on the rail. AS IS, a load that resolves falsy shows "Failed to load activity". TO BE, a rail read whose scope moved on leaves the table as it was and says nothing; a genuine failure still throws and still shows the localized fallback. ## Testing `UsageLogsTable.test.ts` grows from 22 to 31 tests: - A four-row table over `shouldUseWorkspaceBilling × rail`, asserting which of the three readers serves the page and that the other two are not called. The `false × on` row is what holds the `/customers` feed off the rail. - The reader's page rendering the way the legacy rail renders its own. - The paginator moving to page 2 issuing its own read for page 2 — the reader keys requests by page, so this is the host-side half of what `events.test.ts:92` proves in the core. - `SUPERSEDED` publishing nothing and reporting nothing. - A failed read surfacing the localized fallback. - Top-up completion telemetry running off the reader page. Each new test was checked against the unpatched component: reverting the production diff fails five of them, and removing only the `undefined` guard fails the superseded row. `pnpm test:unit src/platform/workspace src/composables/billing src/platform/cloud/subscription` (118 files, 2287 tests, 1 expected-fail), `pnpm typecheck`, `pnpm lint` and `pnpm knip` all pass. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 3**, the tail of the two-PR usage-log stack. Its base is PR 2's branch, so review that one first. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | open — #18036 | | 2 | FE-2467 | billing events reader in `account-core` | open — #18042 | | **3** | **FE-2479** | **usage log reads on the SDK rail** | **this PR** | | 4 | FE-2468 | payment recovery reads on the rail | independent | | 5 | FE-2470 | team checkout on the SDK subscribe command | independent | | 6 | FE-2471 | one status reader for the cloud subscription store | independent | | 7 | FE-2480 + FE-2481 | stop registering a second poller | needs 1 | | 8 | FE-2482 | subscription panel reads the SDK store | needs 1 | | 9 | FE-2484 + FE-2483 + FE-2485 | recovery through the SDK lifecycle | needs 1 | | 10 | FE-2474 | top-up dialog reads through its adapter | independent | | 11 | FE-2481 | one `proration_at` rule | backend answer | | 12 | FE-2472 | table-driven safety net | independent | **The plan's open question for this PR is now closed.** It flagged that FE-2479 adds a sixth read with no E2E ticket, and offered two ways out: widen FE-2476 to six reads and stack it behind PR 3, or carry the events row here. Nothing under `browser_tests/` references `billingSdkTopupRailEnabled` or `billingSdkSubscriptionRailEnabled` today, so there is no reader-rail spec to add a row to — carrying it here would mean building the first rail-flag E2E harness inside an adoption PR, duplicating the scaffolding FE-2476 exists to build. FE-2476 was unstarted, so widening cost nothing. **Picked: widen FE-2476 to six reads**, now retitled and blocked on FE-2479. The cost is that PR 14 loses its independence from Step 1. ```mermaid flowchart LR PR3["PR 3 · FE-2479<br/>usage log on the rail"]:::here PR2["PR 2 · FE-2467"] --> PR3 PR1["PR 1 · FE-2469"] --> PR789["PR 7, 8, 9<br/>needs 1"] IND["PR 4, 5, 6, 10, 11, 12<br/>independent"] PR3 --> PR14["PR 14 · FE-2476<br/>six-read E2E"] PR3 --> BAKE["Step 3 · staging bake"] PR789 --> BAKE IND --> BAKE PR14 --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2479
…mfy-Org#18136) ## Why `teamSubscriptionCheckoutUtil.ts:66` calls `workspaceApi.subscribe(...)` with no rail branch, so a team checkout from the marketing deep link `/cloud/subscribe?tier=team` always rides the legacy transport — even with `billing_sdk_subscription_enabled` on and every other subscribe going through the SDK. That leaves one entry point issuing a charge on a transport the milestone is retiring, and it is the entry point a customer arrives on from outside the app, where nothing else about the session is warm. ## Root cause This util predates the rail and was never a call site the rail work looked at: it lives under `platform/cloud/subscription/utils`, not under `platform/workspace`, and it is a plain function rather than a composable — so it was not reached by either the `useWorkspaceBilling` adapter change or the checkout one. The SDK command has accepted `teamCreditStopId` since `subscriptionCommands.ts:156`; nothing was missing but the branch. ## How **`issueTeamSubscribe(planSlug, options)`** takes `useSubscriptionRail()` and falls through to `workspaceApi.subscribe` exactly as `useWorkspaceBilling.subscribe` does — `ok` returns the settled response, `error` throws, and `unavailable` runs the legacy call, which is the 404 latch every other rail call site uses for a backend gate that is still closed. It is a local function rather than a call into `useWorkspaceBilling`, because that composable is marked `@internal` and builds fresh reactive state per call: invoking it from a plain util would create an orphaned `isLoading`/`error` pair that nothing renders. **`subscribeInputFrom` moves to `platform/workspace/billing/subscribeInput.ts`** and is imported by both call sites. It was private to `useWorkspaceBilling`; duplicating a request-body mapping into a second issuer is how the two drift, and this mapping carries the two fields that must be dropped rather than sent empty. **Everything the util owns is preserved.** A `needs_payment_method` response without `payment_method_url` still throws rather than stranding the customer mid-payment — reachable only on the legacy path, since the rail settles before returning. `trackWorkspaceCheckoutStarted` still receives the `billingOpId` from whichever rail answered. Telemetry names and shapes are unchanged, including the `subscription_checkout` failure event the outer wrapper emits. ## AS IS / TO BE | | AS IS | TO BE | | --- | --- | --- | | Flag off | legacy `POST /api/billing/subscribe` | unchanged | | Flag on | legacy `POST /api/billing/subscribe` | SDK transport, with an idempotency key | | Flag on, route 404 | n/a | SDK attempt, then the legacy call | No capture is attached: there is no visual difference between the two rails here — the same Stripe page opens or the same landing happens, and what changed is which transport issued the charge. The assertions below are on the request, which is what separates them. ## Testing - `teamSubscriptionCheckoutUtil.test.ts` — three rail rows beside the existing legacy ones: the generated body issued through the SDK with the plan slug, stop id and both URLs, landing back in the app and reporting the settled operation id to `trackBeginCheckout`; the 404 case running the legacy call with the options it always sent; and a rail refusal surfacing as the checkout failure without a legacy retry, since a declined card is an answer rather than a closed gate. `pnpm test:unit src/platform/cloud src/platform/workspace src/composables/billing` (145 files, 2599 tests), `pnpm typecheck`, `pnpm lint` and `pnpm knip` pass. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 5**, independent of every other PR in the step. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | merged — Comfy-Org#18036 | | 2 | FE-2467 | billing events reader in `account-core` | open — Comfy-Org#18042 | | 3 | FE-2479 | usage log reads on the SDK rail | open — Comfy-Org#18047, needs 2 | | 4 | FE-2468 | payment recovery reads on the rail | open — Comfy-Org#18060 | | **5** | **FE-2470** | **team checkout on the SDK subscribe command** | **this PR** | | 6 | FE-2471 | one status reader for the cloud subscription store | open — Comfy-Org#18092 | | 7 | FE-2480 + FE-2481 | checkout reads its operation off the rail | open — Comfy-Org#18134 | | 8 | FE-2482 | subscription panel reads the SDK store | open — Comfy-Org#18133 | | 9 | FE-2484 + FE-2483 + FE-2485 | recovery through the SDK lifecycle | open — Comfy-Org#18135 | | 10 | FE-2474 | top-up dialog reads through its adapter | open — Comfy-Org#18132 | | 11 | FE-2481 | one `proration_at` rule | backend answer | | 12 | FE-2472 | table-driven safety net | independent | ```mermaid flowchart LR PR5["PR 5 · FE-2470<br/>team checkout on the rail"]:::here PR1["PR 1 · FE-2469<br/>merged"] --> PR789["PR 7, 8, 9"] PR2["PR 2 · FE-2467"] --> PR3["PR 3 · FE-2479"] IND["PR 4, 6, 10, 11, 12<br/>independent"] PR5 --> BAKE["Step 3 · staging bake"] PR789 --> BAKE PR3 --> BAKE IND --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2470 --------- Co-authored-by: GitHub Action <action@github.com>
## Why The usage log is the last billing read still bound to `workspaceApi`. PR 2 put the reader in `account-core` and exposed `readEvents` on `billingSdkStore`; nothing calls it. This PR is that call site. With `billing_sdk_topup_enabled` or `billing_sdk_subscription_enabled` on, every other billing surface reads through the SDK and this one table still reaches for the workspace client directly — the one read left off the rail the other five already take. ## Root cause `UsageLogsTable.vue` branches on `shouldUseWorkspaceBilling` alone: the workspace client, or `customerEventService.getMyEvents` for the legacy `/customers` feed. There is no third arm, because when the readers landed (#17940 / #17942 / #17946) there was no core reader for `/billing/events` to point one at. That absence outlived the reader. `BillingReadRail` — the contract those five reads resolve through — still declares five members, and FE-2467 deliberately stopped at `billingSdkStore`, leaving the interface widening to whoever brought the caller. So even with the store method in place, the rail type does not expose it: the switch could not be written without this change. ## How **`readEvents` joins `BillingReadRail`** (`useBillingReadRail.ts:28`). It is the only member taking options, because it is the only read that asks for one page rather than one route per scope. **The workspace arm resolves through the rail** (`UsageLogsTable.vue:150`–`:155`, called at `:168`): ```ts const rail = useBillingReadRail() return rail === null ? workspaceApi.getBillingEvents(params) : readOnRail(() => rail.readEvents(params)) ``` Two choices inside that: - **The read is closed over, not passed with its arguments.** FE-2479 sketches `readOnRail(rail.readEvents, params)`, but `readOnRail` takes a zero-arg read and `useBillingCapabilities.ts:74` already closes over its options rather than widening the signature. Following the existing pattern keeps `readOnRail` — used by all five other reads — untouched. - **The rail is read once per load and held for it**, inside `readWorkspaceEvents`, per the rule on `useBillingReadRail`: a flag flip mid-read must not start on one client and publish through the other. **The `customerEventService` branch is untouched.** That is the legacy `/customers` endpoint, out of scope per M1-13, and it keeps its own error surface at `:204`. **A superseded rail read gets its own early return** (`:181`). `readOnRail` resolves `undefined` on `SUPERSEDED` rather than throwing, and the existing falsy-response branch renders `credits.loadEventsError`. Without the guard a scope change would reach the customer as "Failed to load activity". `undefined` identifies exactly one thing here: `getMyEvents` returns `CustomerEventsResponse | null` and `getBillingEvents` throws instead of returning falsy, so only a superseded rail read produces it. Nothing is published and nothing is reported — the same stance `useBillingPlans.ts:44` takes on the same result. The `latestLoadToken` guard and the `isPendingTopupCompleted` call keep their order, so a mid-checkout route flip still reports a top-up that landed on the losing fetch. ## AS IS / TO BE **Flag off, nothing changes.** `rail === null` resolves to the same `workspaceApi.getBillingEvents(params)` call the file made before, on the same branch, with the same error handling. **Flag on, the same rows arrive over the SDK transport.** The store's `readEvents` returns `BillingEventsResponse`, the type this table already consumed, so the rendered page is identical — the rail changes who fetches, not what is shown. A before/after capture would be two identical tables, so none is attached. The one new behaviour is the superseded case, and it is reachable only on the rail. AS IS, a load that resolves falsy shows "Failed to load activity". TO BE, a rail read whose scope moved on leaves the table as it was and says nothing; a genuine failure still throws and still shows the localized fallback. ## Testing `UsageLogsTable.test.ts` grows from 22 to 31 tests: - A four-row table over `shouldUseWorkspaceBilling × rail`, asserting which of the three readers serves the page and that the other two are not called. The `false × on` row is what holds the `/customers` feed off the rail. - The reader's page rendering the way the legacy rail renders its own. - The paginator moving to page 2 issuing its own read for page 2 — the reader keys requests by page, so this is the host-side half of what `events.test.ts:92` proves in the core. - `SUPERSEDED` publishing nothing and reporting nothing. - A failed read surfacing the localized fallback. - Top-up completion telemetry running off the reader page. Each new test was checked against the unpatched component: reverting the production diff fails five of them, and removing only the `undefined` guard fails the superseded row. `pnpm test:unit src/platform/workspace src/composables/billing src/platform/cloud/subscription` (118 files, 2287 tests, 1 expected-fail), `pnpm typecheck`, `pnpm lint` and `pnpm knip` all pass. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 3**, the tail of the two-PR usage-log stack. Its base is PR 2's branch, so review that one first. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | open — #18036 | | 2 | FE-2467 | billing events reader in `account-core` | open — #18042 | | **3** | **FE-2479** | **usage log reads on the SDK rail** | **this PR** | | 4 | FE-2468 | payment recovery reads on the rail | independent | | 5 | FE-2470 | team checkout on the SDK subscribe command | independent | | 6 | FE-2471 | one status reader for the cloud subscription store | independent | | 7 | FE-2480 + FE-2481 | stop registering a second poller | needs 1 | | 8 | FE-2482 | subscription panel reads the SDK store | needs 1 | | 9 | FE-2484 + FE-2483 + FE-2485 | recovery through the SDK lifecycle | needs 1 | | 10 | FE-2474 | top-up dialog reads through its adapter | independent | | 11 | FE-2481 | one `proration_at` rule | backend answer | | 12 | FE-2472 | table-driven safety net | independent | **The plan's open question for this PR is now closed.** It flagged that FE-2479 adds a sixth read with no E2E ticket, and offered two ways out: widen FE-2476 to six reads and stack it behind PR 3, or carry the events row here. Nothing under `browser_tests/` references `billingSdkTopupRailEnabled` or `billingSdkSubscriptionRailEnabled` today, so there is no reader-rail spec to add a row to — carrying it here would mean building the first rail-flag E2E harness inside an adoption PR, duplicating the scaffolding FE-2476 exists to build. FE-2476 was unstarted, so widening cost nothing. **Picked: widen FE-2476 to six reads**, now retitled and blocked on FE-2479. The cost is that PR 14 loses its independence from Step 1. ```mermaid flowchart LR PR3["PR 3 · FE-2479<br/>usage log on the rail"]:::here PR2["PR 2 · FE-2467"] --> PR3 PR1["PR 1 · FE-2469"] --> PR789["PR 7, 8, 9<br/>needs 1"] IND["PR 4, 5, 6, 10, 11, 12<br/>independent"] PR3 --> PR14["PR 14 · FE-2476<br/>six-read E2E"] PR3 --> BAKE["Step 3 · staging bake"] PR789 --> BAKE IND --> BAKE PR14 --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2479
…owns it (Comfy-Org#18134) ## Why `useSubscriptionCheckout` decides what the checkout dialog shows about the operation in flight by reading `billingOperationStore` at `:197`–`:199`, with no rail branch. Six things hang off that one read: the parked-checkout recovery prompt, the authentication state, the declined-attempt message, the reconciliation prompt, the busy state that keeps the pay button locked, and the hosted-step URL. With `billing_sdk_subscription_enabled` on, the operation is owned by the SDK lifecycle and the legacy store has no record of it. So the dialog's answer to all six is "there is no operation" — the pay button unlocks mid-checkout, a parked checkout offers no way to supply a card, and a declined attempt shows no reason. `activeCheckoutActionUrl` is the one that was already patched: `:209` falls back to `subscriptionRail?.subscriptionActionUrl`. That fallback is the shape of the fix; this PR gives the other five the same thing properly. ## Root cause `billingOperationStore` was the poller *and* the view-model. Comfy-Org#17853 and Comfy-Org#17665 railed the commands, so a subscribe already issues through the SDK — but the projections the checkout reads had no counterpart on the SDK store, so the reads could not follow the writes. FE-2469 (Comfy-Org#18036) added them; this is the switch. ## How **`SubscriptionRail` gains the two operation reads** (`subscriptionOperationView.ts`) — `subscriptionActionOperation` and `getOperation(opId)` — next to `subscriptionActionUrl`, which was already a read on that interface. `useSubscriptionRail()` then answers everything the checkout needs from one place, rather than the checkout holding a command rail and a second reader. **`useSubscriptionCheckout` reads `subscriptionRail ?? billingOperationStore`,** captured once at setup for the same reason the rail itself is: a flip mid-checkout must not have the dialog observe on a rail the operation was not issued on. The workspace comparison at `:203` stays where it is — both sides hand back an unscoped record on purpose. **`projectOperationRecord` gains `errorMessage`** (`operationRecordView.ts`), projected from a `failed_retryable` attempt's decline reason through the same `declineDetail` the top-up projection uses. Without it `authenticationError` reads `undefined` on the rail, which is the declined-card message going missing. ### The write half is deliberately unchanged, and that is a correction to both tickets FE-2480 asks to "skip `startOperation` when the rail is on" at `:1542` / `:1548`, and FE-2481 asks the same of `useDowngradeToPersonal.ts:335` / `:355`. **Implementing that would strand a customer.** Those four registrations sit behind `response.status === 'needs_payment_method'` or `'pending_payment'`. The rail's own subscribe settles before it returns — `projectSubscribeResult` maps a settled command to `status: 'subscribed'` and everything else to an error — so those two statuses are shapes only `workspaceApi.subscribe` produces. And `workspaceApi.subscribe` still runs with the flag on: `useWorkspaceBilling.ts:445` falls through to it when the SDK route answers 404, which is the whole point of the `NOT_AVAILABLE` latch. So the flag is the wrong predicate. On the 404 fallback it is on while the response came off the legacy transport, and skipping the registration would leave that operation with no poller at all — the customer opens the hosted page and nothing ever settles it. There is no double-poll to remove here: the adapter shape Comfy-Org#17665 chose already prevents it, which FE-2480's own description says. The invariant is now pinned by test at both call sites instead, so a later change cannot reintroduce the second registration the tickets were guarding against. ## AS IS / TO BE With `billing_sdk_subscription_enabled` on and a subscribe parked on the customer: | | AS IS | TO BE | | --- | --- | --- | | Pay button during the operation | unlocks — `isPolling` reads no operation | stays busy until the operation settles | | Checkout parked awaiting a card | no recovery prompt | recovery prompt, as on the legacy rail | | Declined attempt | no reason shown | the decline reason, as on the legacy rail | | Reconciliation needed | no prompt | the reconciliation prompt | No capture is attached: reaching this state needs an authenticated unified-cloud session with the flag on and a live subscribe parked on a hosted step, which is not reproducible from a local checkout — `operations` is a private ref inside `billingSdkStore`, written only by the lifecycle, and there is no dev harness for it. Reviewers should exercise it in a preview environment with the flag on by starting a subscribe that lands on a verification step and watching the dialog before completing it. Flag off, every one of these reads is the legacy store exactly as today. ## Testing - `useSubscriptionCheckout.test.ts` — a parked operation read off the lifecycle on the rail and off the poller when the rail is null, asserting `parkedCheckoutRecovery`, `authenticationState` and `authenticationError` on each; plus a two-row table over the flag proving a legacy-transport `pending_payment` registers exactly one poller, with the second row being the 404-fallback case the section above turns on. - `useDowngradeToPersonal.test.ts` — the pending-payment registration tightened from "called with" to "called once", so a rail branch that added a second registration at that site fails. - `operationRecordView.test.ts` and the store suite pass unchanged with `errorMessage` added. `pnpm test:unit src/platform/workspace src/composables/billing` (91 files, 1956 tests), `pnpm typecheck` and `pnpm lint` pass. Note for whoever merges second: this touches `operationRecordView.ts` alongside PR 8 (Comfy-Org#18133), which adds `actionUrl` validation to the same projection. Different lines, trivial rebase. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 7**, one of the three switches PR 1 unblocked. Parallel to PRs 8 and 9. FE-2481's other half — the single `proration_at` rule — stays in PR 11, which waits on the backend answer. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | merged — Comfy-Org#18036 | | 2 | FE-2467 | billing events reader in `account-core` | open — Comfy-Org#18042 | | 3 | FE-2479 | usage log reads on the SDK rail | open — Comfy-Org#18047, needs 2 | | 4 | FE-2468 | payment recovery reads on the rail | open — Comfy-Org#18060 | | 5 | FE-2470 | team checkout on the SDK subscribe command | independent | | 6 | FE-2471 | one status reader for the cloud subscription store | open — Comfy-Org#18092 | | **7** | **FE-2480 + FE-2481 (poller half)** | **checkout reads its operation off the rail** | **this PR** | | 8 | FE-2482 | subscription panel reads the SDK store | open — Comfy-Org#18133 | | 9 | FE-2484 + FE-2483 + FE-2485 | recovery through the SDK lifecycle | needs 1 | | 10 | FE-2474 | top-up dialog reads through its adapter | open — Comfy-Org#18132 | | 11 | FE-2481 | one `proration_at` rule | backend answer | | 12 | FE-2472 | table-driven safety net | independent | ```mermaid flowchart LR PR1["PR 1 · FE-2469<br/>merged"] --> PR7["PR 7 · FE-2480 + FE-2481<br/>checkout reads on the rail"]:::here PR1 --> PR8["PR 8 · FE-2482"] PR1 --> PR9["PR 9 · FE-2484 + FE-2483 + FE-2485"] PR2["PR 2 · FE-2467"] --> PR3["PR 3 · FE-2479"] IND["PR 4, 5, 6, 10, 11, 12<br/>independent"] PR7 --> BAKE["Step 3 · staging bake"] PR8 --> BAKE PR9 --> BAKE PR3 --> BAKE IND --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2480 - Part of FE-2481
## Why No spec under `browser_tests` sets `billing_sdk_topup_enabled`. Comfy-Org#17665 shipped the top-up rail with 48 dual-path unit tests and no browser coverage, so the first flow due to be flagged on in staging (FE-2486 ramps top-up first) is the one flow with no end-to-end proof. Plan change and cancel both have off / on / 404 specs; top-up had none. ## How `browser_tests/tests/dialogs/topUpSdkRail.spec.ts`, modelled on `planChangeSdkRail.spec.ts`. Both rails POST `/api/billing/topup`, so the request body cannot separate them. Two things can, and the spec asserts both rather than only the outcome: - **transport** — the SDK runs on `fetch`, the legacy workspace client on axios' XHR adapter - **`Idempotency-Key`** — only the SDK sends one, and the body's `idempotency_key` must match the header The flag goes through `FeatureFlagHelper.seedServerFlags`, never by stubbing `GET /api/features` — that response carries `firebase_config`, and a stub signs the session out silently. Three rows: the legacy call with the rail off, the SDK call with it on, and a closed backend gate. Each also asserts the balance is re-read after the purchase settles, which is the customer-visible end of the flow on both rails. ## What this found: the top-up rail has no 404 fallback FE-2475's third scope bullet asked for "the `NOT_AVAILABLE` latch falls back to the legacy call once, and the purchase still completes". **That latch does not exist.** `account-core` does mint the code — `topup.ts:226` maps a 404 to `NOT_AVAILABLE` — but nothing in `src/` reads it. There is no occurrence of `NOT_AVAILABLE` anywhere in the app. Compare the subscription rail, which has `subscriptionRouteAvailable` in `billingSdkStore.ts:334` and `onSubscriptionRail` → `DECLINED` → `workspaceApi` in `useWorkspaceBilling.ts:445`. Top-up has neither the latch nor the retry: `useTopupOperation.topup` returns `sdkStore.createTopup(...)` and a 404 comes back out as a `WorkspaceApiError`. So with the flag on before the backend route is deployed, **every top-up fails** and the customer sees an error toast. That is a ramp-ordering risk on the exact flag FE-2486 turns on first. The third row pins that behaviour as it actually is rather than as the ticket assumed, because a spec that asserted the intended fallback would simply be red and tell nobody anything. Closing the gap is a source change and needs its own ticket; when it lands, this row is the one that gets rewritten, and it is written to make that obvious. ## AS IS / TO BE **No product change** — this PR adds a spec file and touches no source. There is nothing to capture; what changes is what CI proves about code already on `main`. ## Testing - `topUpSdkRail.spec.ts` (new, `@cloud`) — three rows as above. - `pnpm typecheck`, `pnpm typecheck:browser` and `pnpm lint` pass. **This spec has not been executed locally.** It needs a cloud backend on `:8188`, which this checkout has no access to, so CI is its first run. Reviewers should treat a red `@cloud` shard here as the spec's own problem rather than a regression, and I will chase it. ## Plan position Milestone 1, Step 1b — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 13**, backfill for code already on `main` (Comfy-Org#17665), independent of every Step 1 PR. | PR | Step | Tickets | What | Status | | --- | --- | --- | --- | --- | | 1 | 1 | FE-2469 | operation-state projections | merged — Comfy-Org#18036 | | 2 | 1 | FE-2467 | billing events reader | open — Comfy-Org#18042 | | 3 | 1 | FE-2479 | usage log on the rail | open — Comfy-Org#18047 | | 4 | 1 | FE-2468 | payment recovery on the rail | open — Comfy-Org#18060 | | 5 | 1 | FE-2470 | team checkout on the rail | open — Comfy-Org#18136 | | 6 | 1 | FE-2471 | one status reader | open — Comfy-Org#18092 | | 7 | 1 | FE-2480 + FE-2481 | checkout reads on the rail | open — Comfy-Org#18134 | | 8 | 1 | FE-2482 | subscription panel on the rail | open — Comfy-Org#18133 | | 9 | 1 | FE-2484 + FE-2483 + FE-2485 | recovery on the rail | open — Comfy-Org#18135 | | 10 | 1 | FE-2474 | top-up dialog adapter | open — Comfy-Org#18132 | | 11 | 1 | FE-2481 | one `proration_at` rule | backend answer | | 12 | 1 | FE-2472 | table-driven safety net | open — Comfy-Org#18137 | | **13** | **1b** | **FE-2475** | **top-up rail E2E** | **this PR** | | 14 | 1b | FE-2476 | six reads E2E | needs 3 | | 15 | 1b | FE-2477 | resubscribe + portal E2E | independent | ```mermaid flowchart LR PR13["PR 13 · FE-2475<br/>top-up rail E2E"]:::here STEP1["Step 1 · PRs 1-12"] PR3["PR 3 · FE-2479"] --> PR14["PR 14 · FE-2476"] PR15["PR 15 · FE-2477"] STEP1 --> BAKE["Step 3 · staging bake"] PR13 --> BAKE PR14 --> BAKE PR15 --> BAKE BAKE --> RAMP["Step 4 · production ramp<br/>top-up first"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2475
…SDK store (Comfy-Org#18042) ## Why The usage log is the last billing surface with no rail to move to. `UsageLogsTable.vue:158` picks its backend per load — `workspaceApi.getBillingEvents` on the workspace rail, `customerEventService.getMyEvents` otherwise — and there is no third branch to add, because `account-core` has no reader for `/billing/events`. Every other billing read has one. FE-2479 moves that table onto the SDK rail. It cannot until the core can answer the read. This PR makes it answerable; the call site moves in FE-2479. ## Root cause The events read is unlike the five readers already in the core — credits, status, plans, payment methods, capabilities — in one way that matters: its request varies per call. The others ask for one route per scope, so `scopedReader` keyed an in-flight read on scope alone and a caller arriving mid-flight joined it. That key is wrong for a paginated feed. A caller wanting page 1 shares a scope with a read of page 2 already in flight, so it would have joined that read and been handed page 2's events as the answer to its own request. The skeleton had no way to say *same scope, different question*. The feed is also the one endpoint that can answer a question other than the one asked: it clamps a page past the end and reports the page it actually served. `UsageLogsTable.vue:176`–`:182` already relies on this, writing `response.page` and `response.limit` back over what it requested. ## How **`route` widens to `string | ((options) => string)`.** A reader with a fixed route passes the string it always passed, unchanged. `events.ts` passes a builder for `/billing/events?page=&limit=` with a fixed key order, so the same page asked for twice is the same string — and omitting either option omits it from the query, leaving the endpoint's own default in force rather than guessing at it. **The resolved route joins the scope in identifying an in-flight read.** `InFlightRead` carries its route and a caller joins only on a match of both (`scopedReader.ts:223`). A caller for a different route on the same scope starts its own read and takes the slot; the displaced read still settles and publishes, so its own caller is answered — the snapshot carries which page it is — but nobody can join it any more. One slot is enough for a feed read one page at a time. **`createBillingEventsReader` adds `scope` and `readAt` to the decoded page, and nothing else.** There is no int64 in this response to widen, so the projection is the identity plus the two fields every snapshot carries. **`billingSdkStore.readEvents` drops exactly those two** (`billingSdkStore.ts:437`) and returns `BillingEventsResponse` — the type `UsageLogsTable` already consumes — so FE-2479 is a call swap rather than a shape change. A failed read passes through untouched. **Every `BillingClient` composition wires the reader**, since the interface now requires it: `createBillingSdk` for the workspace host, the `account-ui` harness, and billing-web. billing-web's hosted surfaces do not read the feed, so its test fake declares `events.read` unused there, next to `credits` and `status`. ## AS IS / TO BE **No user-visible change, on either rail.** Nothing calls `readEvents` yet — `UsageLogsTable` still branches between `workspaceApi` and `customerEventService` exactly as before, and renders the same rows from the same backend. A before/after capture would show two identical tables, so none is attached. The one behavioural change is internal, and reviewers should look at it: two concurrent reads on one scope. AS IS, a read of page 1 issued while page 2 is in flight is served page 2's body. TO BE, it issues its own request and is served page 1. No shipped caller does this yet; `events.test.ts:92` is what holds it. ## Testing - `events.test.ts` — the route built for each combination of `page` and `limit`, the scope the snapshot reports, an event the decoder rejects surfacing as `MALFORMED_RESPONSE`, two pages in flight not serving one from the other, and two callers on one page sharing a single request. - `scopedReader.test.ts` — the events reader joins the shared table of reader cases, so the scope fence, the publish fence and the denial rules are proved for it rather than restated. - `billingSdkStore.test.ts` — `readEvents` forwarding its options and returning the page without `scope`/`readAt`, and a failure passing through untouched. `pnpm test:unit src/platform/workspace` (82 files, 1831 tests), `@comfyorg/account-core` (56 files, 925 tests), `@comfyorg/account-ui` (18 files, 120 tests) and `@comfyorg/billing-web` (13 files, 125 tests) pass, as do `pnpm typecheck`, `pnpm lint` and `pnpm knip`. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 2**, the head of the two-PR usage-log stack: PR 3 collapses to a call swap once it lands. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | merged — Comfy-Org#18036 | | **2** | **FE-2467** | **billing events reader in `account-core`** | **this PR** | | 3 | FE-2479 | usage log reads on the SDK rail | needs 2 | | 4 | FE-2468 | payment recovery reads on the rail | independent | | 5 | FE-2470 | team checkout on the SDK subscribe command | independent | | 6 | FE-2471 | one status reader for the cloud subscription store | independent | | 7 | FE-2480 + FE-2481 | stop registering a second poller | needs 1 | | 8 | FE-2482 | subscription panel reads the SDK store | needs 1 | | 9 | FE-2484 + FE-2483 + FE-2485 | recovery through the SDK lifecycle | needs 1 | | 10 | FE-2474 | top-up dialog reads through its adapter | independent | | 11 | FE-2481 | one `proration_at` rule | backend answer | | 12 | FE-2472 | table-driven safety net | independent | ```mermaid flowchart LR PR2["PR 2 · FE-2467<br/>events reader"]:::here PR1["PR 1 · FE-2469"] --> PR7["PR 7, 8, 9<br/>needs 1"] PR2 --> PR3["PR 3 · FE-2479<br/>usage log on the rail"] IND["PR 4, 5, 6, 10, 11, 12<br/>independent"] PR3 --> BAKE["Step 3 · staging bake"] PR7 --> BAKE IND --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2467
…cycle (Comfy-Org#18135) ## Why Recovery is the path a customer takes after a redirect checkout, a reload mid-payment, or a flag flip while an operation is pending. It has two entry points, and with `billing_sdk_subscription_enabled` on both of them hand the operation to the wrong driver. - `useWorkspaceBilling.resumePendingOperation` (`:304`) gives a pending operation to the SDK **only** when the server calls it a top-up. A pending `subscription` at load starts the legacy poller, for an operation whose writes went through the SDK. - `useSubscriptionDialog.resumePendingCheckout` (`:293`) has no rail branch at all. That is the path a customer takes when they return to a parked checkout, so on the rail it starts a legacy poller for an operation the lifecycle already owns. Shipping either alone leaves recovery half-railed, which is the exact ambiguity FE-2484 exists to remove — so the two ship together, with the browser proof FE-2485 asks for. ## Root cause Comfy-Org#17665 railed top-up and railed its recovery with it, because top-up was the flow being moved. The subscription commands were railed later (Comfy-Org#17853, and the checkout in Comfy-Org#18036's follow-ups) and nothing went back to the resume paths. `pending_billing_op_type === 'topup'` was never a rule — it was the shape of the one flow that had a rail at the time, left in place as the condition. ## How **`railOwnsResume(type)`** replaces that condition in `useWorkspaceBilling`. Each kind of pending operation follows the rail that issues it: `topup` on `billingSdkTopupRailEnabled`, everything else on `billingSdkSubscriptionRailEnabled`. A server predating `pending_billing_op_type` only ever had subscriptions to hand back, so an absent field follows the subscription rail — the same reading `resumeModeFor` already takes of it. `'cancel'` is named in the ticket but is not a value this field can hold: the response type is `'topup' | 'subscription' | undefined`, and `resumeModeFor`'s `never` branch is what holds that. A cancel pending at load is reported as a subscription operation and is covered by the same branch. **`billingSdkStore.recoverPendingOperation(opId)`** is new, for the caller that has something to decide on the outcome. `lifecycle.recover()` resolves at *adoption*, not at settlement, and the dialog needs the terminal status to know whether to reopen the pricing table. So this awaits the projection settling, and returns `undefined` when there is nothing to adopt — either the read failed, or the server named a different operation, which means the pointer this caller held is stale. **`useSubscriptionDialog.resumePendingCheckout` branches on the rail** and gets one new behaviour out of it: an `undefined` result clears the stale pointer instead of leaving it parked. The owner/workspace guard above it is untouched, and so is the host pointer `comfy:pending-subscription-checkout` — it carries the tier/cycle selection the pricing dialog restores, which the SDK's scope-keyed pointer deliberately does not. Only who drives the operation moves. ## AS IS / TO BE With `billing_sdk_subscription_enabled` on and a subscription operation pending at load: | | AS IS | TO BE | | --- | --- | --- | | Server reports a pending subscription | legacy poller starts on an SDK-owned operation | `lifecycle.recover()` adopts it; no legacy poller | | Customer returns to a parked checkout | legacy poller starts | `recoverPendingOperation` adopts it | | Pointer left by a pre-flip session | polled on the rail it was not issued on | adopted by the lifecycle | | Recovery finds nothing to adopt | pointer stays parked indefinitely | pointer cleared | No still capture is attached: there is no visual difference to photograph — both rails render the same checkout dialog, and what changed is which transport polls the operation behind it. `operationRecoverySdkRail.spec.ts` is the evidence, and it asserts exactly that discriminator. Flag off, both entry points behave as they do today, which the spec's first row holds. ## Testing - `operationRecoverySdkRail.spec.ts` (new, `@cloud`) — three browser rows against the same mocked pending operation. Both rails poll `GET /api/billing/ops/{id}`, so the request body cannot separate them; the transport can, as in `planChangeSdkRail`: the SDK runs on `fetch`, the legacy workspace client on axios' XHR. Rail off, no poll is a `fetch`. Rail on, the operation is adopted **exactly once, on one rail** — see the race note below for why this row cannot assert *which* rail without flaking. And a failed recovery still reopens the checkout on the Creator annual plan the host pointer recorded, then clears it. - `useWorkspaceBilling.sdk.test.ts` — a five-row table over (op type × topup rail × subscription rail) pinning which side adopts, including the two rows that prove the rails do not cross: a pending top-up with only the subscription rail on stays on the poller, and a pending subscription with only the top-up rail on stays on the poller. - `useSubscriptionDialog.test.ts` — a two-row table over the rail: each adopts through its own driver and neither through the other's, and the pricing dialog reopens on the recorded tier/cycle either way. Plus the stale-pointer row, which is the new behaviour. - `billingSdkStore.test.ts` — `recoverPendingOperation` resolving with the settled record, and returning `undefined` both when the server names another operation and when the recovery read fails. `pnpm test:unit src/platform/workspace src/composables/billing src/platform/cloud` (145 files, 2607 tests), `pnpm typecheck`, `pnpm typecheck:browser` and `pnpm lint` pass. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 9**, the last of the three switches PR 1 unblocked. It carries FE-2485 rather than leaving it for Step 1b, because it proves new code: E1–E3 stay separate PRs because they backfill code already on `main`. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | merged — Comfy-Org#18036 | | 2 | FE-2467 | billing events reader in `account-core` | open — Comfy-Org#18042 | | 3 | FE-2479 | usage log reads on the SDK rail | open — Comfy-Org#18047, needs 2 | | 4 | FE-2468 | payment recovery reads on the rail | open — Comfy-Org#18060 | | 5 | FE-2470 | team checkout on the SDK subscribe command | independent | | 6 | FE-2471 | one status reader for the cloud subscription store | open — Comfy-Org#18092 | | 7 | FE-2480 + FE-2481 | checkout reads its operation off the rail | open — Comfy-Org#18134 | | 8 | FE-2482 | subscription panel reads the SDK store | open — Comfy-Org#18133 | | **9** | **FE-2484 + FE-2483 + FE-2485** | **recovery through the SDK lifecycle** | **this PR** | | 10 | FE-2474 | top-up dialog reads through its adapter | open — Comfy-Org#18132 | | 11 | FE-2481 | one `proration_at` rule | backend answer | | 12 | FE-2472 | table-driven safety net | independent | ```mermaid flowchart LR PR1["PR 1 · FE-2469<br/>merged"] --> PR9["PR 9 · FE-2484 + 2483 + 2485<br/>recovery on the rail"]:::here PR1 --> PR7["PR 7 · FE-2480 + FE-2481"] PR1 --> PR8["PR 8 · FE-2482"] PR2["PR 2 · FE-2467"] --> PR3["PR 3 · FE-2479"] IND["PR 4, 5, 6, 10, 11, 12<br/>independent"] PR9 --> BAKE["Step 3 · staging bake"] PR7 --> BAKE PR8 --> BAKE PR3 --> BAKE IND --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2484 - Part of FE-2483 - Part of FE-2485
…Comfy-Org#18060) ## Why `recoverOutstandingPayment` in `useSubscriptionCheckout.ts` is the path a customer lands on when a plan change is rejected because their card failed. It read billing status and the portal URL straight off `workspaceApi` with no rail branch: - `:555` — `(await workspaceApi.getBillingStatus()).billing_status === 'payment_failed'`, the second check behind `TRANSITION_NOT_ALLOWED` - `:566` — `workspaceApi.getPaymentPortalUrl(returnUrl)` With `billing_sdk_subscription_enabled` on, the subscribe that failed ran on the SDK, and the recovery from that failure ran on the legacy transport. The customer is sent to a portal the rail that rejected them never issued. ## Root cause Recovery was written as an error branch, not as a flow, so it was not on the list when the surrounding call sites were railed. Everything around it moved: `subscribe` and `previewSubscribe` take `useSubscriptionRail()`, the reads take `useBillingReadRail()`, and `manageSubscription` already opens the portal through `rail.openPaymentPortal`. `recoverOutstandingPayment` sat between them, reached only on a rejected preview, and kept both of its own calls on `workspaceApi`. ## How Two reads, two rails, because the two answers come from two different places. **The status read takes `useBillingReadRail()`, not `useSubscriptionRail()`.** This is a deliberate departure from FE-2468's scope line, which says to gate both on `useSubscriptionRail()`. `readStatus` is not on `SubscriptionRail` — it is on `BillingReadRail`, and gating it on the subscription rail does not typecheck without reaching past the interface into the store. It is also the wrong rail: `useBillingReadRail()` is on when *either* flag is on, and `fetchStatus` in `useWorkspaceBilling` already reads status through it. Gating this one read on the subscription flag would make it the only status read in the app that disagrees with `fetchStatus` while the top-up flag is on alone. The read follows the pattern this same file already uses for saved payment methods at `:364`. `readOnRail` returns `undefined` on `SUPERSEDED`, so the status is now optional and `status?.billing_status` decides. A superseded read means the scope moved on underneath it — there is no subscription left to recover, which is the same nothing the legacy path produces from a stale read. **The portal takes `useSubscriptionRail()`,** through a small `readPaymentPortalUrl` helper so the call site keeps reading as one expression. It mirrors `manageSubscription`: `ok` hands back the URL, `unavailable` means the route is not deployed here and the legacy client answers, and `error` throws. Throwing is what keeps the legacy branch byte-identical — the existing `catch (portalError)` already reports `billing_portal_open_failure` and surfaces the toast, and a rail failure now lands exactly where a legacy throw did. Going through the rail also picks up the `paymentMethods.invalidate()` the SDK command does on success, which the legacy call has no equivalent for. The rail is the one held at `:162` rather than a fresh `useSubscriptionRail()`, both because the portal open happens after an `await` and because holding it is what `useSubscriptionRail`'s own contract asks for — a flag flip mid-recovery must not read status on one rail and open the portal on the other. ## AS IS / TO BE **No user-visible change with the flags off**, and none in the shape of the flow with them on — the customer sees the same Stripe portal in the same tab, and the same toast when it cannot open. What changes is who issues the portal session. | | AS IS (flag on) | TO BE (flag on) | | --- | --- | --- | | second `TRANSITION_NOT_ALLOWED` check | `workspaceApi.getBillingStatus()` | `readStatus` on the read rail | | portal session | `workspaceApi.getPaymentPortalUrl()` | `sdk.commands.openPaymentPortal` | | superseded scope | stale read, recovery may proceed | no recovery | | saved-card list after the portal | unchanged, may be stale | invalidated by the rail | A before/after capture would show the same two screens, so none is attached. The difference is in which transport issued the session, which is what the tests below pin. ## Testing `useSubscriptionCheckout.test.ts` — a new `payment recovery on the SDK rail` suite, seven cases: - the second `TRANSITION_NOT_ALLOWED` check reads `readStatus` and does not touch `getBillingStatus` - a table over the three answers that recover nothing — healthy status, `SUPERSEDED`, and a hard read failure — each with the **legacy client answering `payment_failed`**, so recovering nothing is only possible if the rail's answer is the one being read - the portal URL the rail hands back is the one opened, with no legacy call - `unavailable` falls through to the legacy client - a rail portal failure reports `billing_portal_open_failure` and shows the server guidance, where a legacy throw already landed The existing legacy-path suite is unchanged and still passes, which is the flag-off half of the dual path. Reverting the source change and re-running fails all 7. It used to fail only 6: the `unavailable` fall-through case discarded the rail's return value, so it passed on legacy-only code too. `b7318d7` asserts that the rail was consulted before the fallback, which is what makes the seventh case discriminate. **No Playwright regression test was added; the milestone plan routes this rail's browser coverage elsewhere.** The [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709) asks a flag-on PR for its own dual-path unit suite, and for a cloud E2E in the same PR *when the code is new* — which is why PR 9 absorbs FE-2485. This PR adds a rail branch to recovery that already ships on `main`, not new behaviour, and the plan's own E2E audit flagged exactly one gap in Step 1 (FE-2479, PR 3) and did not flag this ticket. The payment portal on the subscription rail has its own scoped spec, FE-2477 (PR 15), in the Step 1b backfill. To correct an earlier version of this paragraph: it claimed a browser test could not observe which transport answered a call. That is wrong in this repo. `browser_tests/tests/dialogs/planChangeSdkRail.spec.ts`, written for this same FE-2216 rail work, separates the two rails precisely that way — *"the recorded requests are what separate them: only the SDK transport sends an `Idempotency-Key`... the SDK runs on `fetch`, the workspace client on axios' XHR"* — and `checkoutRecovery.spec.ts` already routes `**/api/billing/status` inside a recovery flow. So an E2E here is feasible, and the reason it is not in this PR is the plan's allocation above, not a technical limit. If a reviewer would rather it ship here than in FE-2477, say so and I will add it. `pnpm test:unit` (src/platform/workspace + src/composables/billing: 91 files, 1960 tests), `pnpm typecheck`, `pnpm lint` and `pnpm knip` all pass. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 4**, independent: it does not depend on PR 1 and blocks nothing in Step 1. It is a prerequisite for FE-2490, which deletes the legacy branches in `useSubscriptionCheckout` once the flag has been at 100% for a release cycle — the legacy branch this PR leaves byte-identical is the one FE-2490 removes. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | merged (Comfy-Org#18036) | | 2 | FE-2467 | billing events reader in `account-core` | independent | | 3 | FE-2479 | usage log reads on the SDK rail | stacked on 2 | | **4** | **FE-2468** | **payment recovery reads on the rail** | **this PR** | | 5 | FE-2470 | team checkout on the SDK subscribe command | independent | | 6 | FE-2471 | one status reader for the cloud subscription store | independent | | 7 | FE-2480 + FE-2481 | stop registering a second poller | needs 1 | | 8 | FE-2482 | subscription panel reads the SDK store | needs 1 | | 9 | FE-2484 + FE-2483 + FE-2485 | recovery through the SDK lifecycle | needs 1 | | 10 | FE-2474 | top-up dialog reads through its adapter | independent | | 11 | FE-2481 | one `proration_at` rule | backend answer | | 12 | FE-2472 | table-driven safety net | independent | ```mermaid flowchart LR PR4["PR 4 · FE-2468<br/>payment recovery"]:::here PR1["PR 1 · FE-2469<br/>merged"] --> PR7["PR 7 · FE-2480 + FE-2481"] PR1 --> PR8["PR 8 · FE-2482"] PR1 --> PR9["PR 9 · FE-2484 + FE-2483 + FE-2485"] PR2["PR 2 · FE-2467"] --> PR3["PR 3 · FE-2479"] IND["PR 5, 6, 10, 11, 12<br/>independent"] PR4 --> BAKE["Step 3 · staging bake"] PR7 --> BAKE PR8 --> BAKE PR9 --> BAKE PR3 --> BAKE IND --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal<br/>FE-2490 deletes this legacy branch"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` ## Review focus The `useBillingReadRail()` choice for the status read is the one call that departs from the ticket's literal scope line. If the intent really was to keep that read on the subscription flag alone, this is the line to push back on — but it would then be the only status read in the app gated differently from `fetchStatus`. - Part of FE-2468
…r rail (Comfy-Org#18140) ## Why Comfy-Org#17940 (status, balance), Comfy-Org#17942 (plans), Comfy-Org#17946 (capabilities, saved payment methods) and Comfy-Org#18047 (billing events) moved every billing read onto the SDK readers behind `useBillingReadRail`. **None of the four shipped a browser spec.** Nothing under `browser_tests/` referenced `billingSdkTopupRailEnabled` or `billingSdkSubscriptionRailEnabled` before this stack, so the reader rail — six routes feeding five rendered surfaces — has never been exercised end to end. The int64 → number projections (`safeInt64.asSafeNumber`, `projectBillingStatus`, `projectBillingPlans`) are well unit-tested but had never been run against a rendered value, which is the only place a silently rounded amount would show. ## How `browser_tests/tests/dialogs/billingReadsSdkRail.spec.ts`. Both rails GET the same six routes, so the URL cannot separate them; the transport can, as in the write-rail specs — the SDK runs on `fetch`, the legacy workspace client on axios' XHR. Settings ▸ Workspace ▸ Plan & Credits is the one surface that drives all six, so the spec opens it once per row and asserts across the whole set rather than route by route. **Rows 1 and 2 are the parity pair.** Rail off: every route this flow drives is read and *no* read went out on `fetch`. Rail on: the same routes are read and *no* read went out on `xhr` — one leaking read would mean a surface reading a backend the rail did not settle. Both rows assert the same rendered values (the credits total, the usage-log row), so the projections are pinned against a rendered surface rather than against each other. **Five of the six reads, not all six.** Saved payment methods is not read by this panel — the top-up dialog reads it for its saved-card note — so it belongs to `topUpSdkRail.spec.ts` (Comfy-Org#18138) rather than here. Asserting it from this flow would mean driving a dialog this spec has no other reason to open. The other five are all here, billing events included, which is the read this ticket was widened for. **The usage log is behind the panel's Activity tab.** The Credits tab it opens on never mounts `UsageLogsTable`, so the events read does not go out until that click — which is what the first CI runs of this spec showed. **Row 3 is the int64 case.** A balance of `9007199254740993` micros is beyond `Number.MAX_SAFE_INTEGER`; the projection must refuse the read rather than render a number off by an unknown amount. The fixture is sent as **raw JSON rather than a JS literal** — written as a literal the digits would be lost in the test file before the app ever saw them, which would make the row assert nothing. `no-loss-of-precision` flagged exactly that, and the raw body is the fix. **Row 4 is the paging case Comfy-Org#18047 introduced.** The events reader is scoped *and* paged, and `scopedReader` used to key an in-flight read on scope alone — so a caller wanting page 2 could be handed page 1's answer. The row pages forward and asserts page 2 issued its own request and rendered its own row. ## Widened to six reads FE-2476 originally covered five. The milestone plan left a choice open before PR 3 could open — widen this ticket to six and stack it behind PR 3, or carry the events row inside PR 3 — and **widening won**: there was no reader-rail spec to add a row to, so "the events row inside PR 3" would have meant building the first rail-flag E2E harness inside an adoption PR. The cost is that this PR is no longer independent of Step 1; it stacks on PR 3. The `SUPERSEDED` row the ticket also names is not here. A superseded read is a scope change under an in-flight request, which needs a workspace switch mid-load; driving that reliably in a browser needs a harness this spec does not have, and `UsageLogsTable.vue:181` already discards an undefined response under unit test. Left for a follow-up rather than written as a flaky row. ## AS IS / TO BE **No product change** — this PR adds a spec file and touches no source. There is nothing to capture; what changes is what CI proves about code already on `main` and on the PR 2/3 stack. ## Testing - `billingReadsSdkRail.spec.ts` (new, `@cloud`) — four rows as above. - `pnpm typecheck`, `pnpm typecheck:browser` and `pnpm lint` pass. **This spec has not been executed locally.** It needs a cloud backend on `:8188`, which this checkout has no access to, so CI is its first run. Reviewers should treat a red `@cloud` shard here as the spec's own problem rather than a regression, and I will chase it. ## Plan position Milestone 1, Step 1b — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 14**. It targets PR 2's branch, which now carries PR 3 (Comfy-Org#18047 merged into it), because the sixth read only exists there. | PR | Step | Tickets | What | Status | | --- | --- | --- | --- | --- | | 1 | 1 | FE-2469 | operation-state projections | merged — Comfy-Org#18036 | | 2 | 1 | FE-2467 | billing events reader | open — Comfy-Org#18042 | | 3 | 1 | FE-2479 | usage log on the rail | merged into 2 — Comfy-Org#18047 | | 4 | 1 | FE-2468 | payment recovery on the rail | open — Comfy-Org#18060 | | 5 | 1 | FE-2470 | team checkout on the rail | open — Comfy-Org#18136 | | 6 | 1 | FE-2471 | one status reader | open — Comfy-Org#18092 | | 7 | 1 | FE-2480 + FE-2481 | checkout reads on the rail | open — Comfy-Org#18134 | | 8 | 1 | FE-2482 | subscription panel on the rail | open — Comfy-Org#18133 | | 9 | 1 | FE-2484 + FE-2483 + FE-2485 | recovery on the rail | open — Comfy-Org#18135 | | 10 | 1 | FE-2474 | top-up dialog adapter | open — Comfy-Org#18132 | | 11 | 1 | FE-2481 | one `proration_at` rule | backend answer | | 12 | 1 | FE-2472 | table-driven safety net | open — Comfy-Org#18137 | | 13 | 1b | FE-2475 | top-up rail E2E | open — Comfy-Org#18138 | | **14** | **1b** | **FE-2476** | **six reads E2E** | **this PR** | | 15 | 1b | FE-2477 | resubscribe + portal E2E | open — Comfy-Org#18139 | ```mermaid flowchart LR PR2["PR 2 · FE-2467<br/>+ PR 3 · FE-2479"] --> PR14["PR 14 · FE-2476<br/>six reads E2E"]:::here STEP1["Step 1 · PRs 1, 4-12"] PR13["PR 13 · FE-2475"] PR15["PR 15 · FE-2477"] STEP1 --> BAKE["Step 3 · staging bake"] PR13 --> BAKE PR14 --> BAKE PR15 --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2476
…rg#18133) ## Why `SubscriptionPanelContentWorkspace.vue` decides whether to show the "setting up your subscription" banner, and whether to offer the Complete verification button inside it, by reading `billingOperationStore` — `:445` for `isSettingUp`, `:447` for `subscriptionActionOperation?.actionUrl`. Neither read has a rail branch. With `billing_sdk_subscription_enabled` on, the operation those two values describe is owned by the SDK lifecycle. The legacy store has no record of it, so the panel renders the plan tile as though nothing were in flight: no banner while the subscription is being set up, and no way through when the customer still owes a hosted verification step. ## Root cause FE-2469 named this the last half of `billingOperationStore` to move: the store was the poller *and* the view-model, and the SDK replaced the poller only. The commands were railed by Comfy-Org#17853 and Comfy-Org#17665, so a subscribe already issues through the SDK — but the two predicates the panel reads had no counterpart on the SDK store, so they could not follow. PR 1 (Comfy-Org#18036) added them. This PR is the switch they were added for. ## How **`useSubscriptionOperationView()`** joins `useSubscriptionRail()` in `useSubscriptionRail.ts` — the same flag, read once, for the same reason the commands read it once: a flip mid-operation must not have the panel observe on a rail the operation was not issued on. It answers the two questions the panel asks and nothing else, from `billingSdkStore` when the rail is on and from `billingOperationStore` when it is off. It is a read-side sibling rather than a widening of `SubscriptionRail`, because `SubscriptionRail` is the command interface `useSubscriptionCheckout` holds; the panel issues nothing. **The panel loses its `useBillingOperationStore` import and both computeds** for one destructure. `openSubscriptionVerification` and the template are untouched. **`projectOperationRecord` now validates `actionUrl`** (`operationRecordView.ts`). This is a parity gap PR 1 left: the legacy store validates the URL at registration (`billingOperationStore.ts:253`), so a legacy record's `actionUrl` is https-or-null by the time a surface reads it, while the projection carried the SDK's value through raw. Both this panel and `useSubscriptionCheckout` hand that value to `window.open`, so the projection has to make the same guarantee the record it mirrors makes. `account-core` exports the same `validateActionUrl` the SDK store already uses for `subscriptionActionUrl`, so this is one predicate, not a second copy. ## AS IS / TO BE No screenshot — the state this PR changes is not reachable on a local build. `billingSdkSubscriptionRailEnabled` is `billingSdkSubscriptionEnabled && unifiedCloudAuthEnabled`, so the rail is only on inside an authenticated unified-cloud session, and `isSettingUp` / `subscriptionActionOperation` are computed off `billingSdkStore`'s private `operations`, which only the SDK lifecycle's poll of a live backend operation writes. There is no story for `SubscriptionPanelContentWorkspace`, no dev harness for the lifecycle, and no writable surface on the store, so nothing short of a real subscribe parked on a hosted verification step renders the banner. Rather than stage a fake, the two frames are described: | | | | --- | --- | | **AS IS** | The plan tile as normal while an SDK-owned subscribe is in flight: no "setting up your subscription" spinner banner and no **Complete verification** button, because the legacy store holds no record of the operation. | | **TO BE** | The spinner banner (`billingOperation.subscriptionProcessing`), and inside it the **Complete verification** button for a role with `canManageSubscription`, opening the hosted step. | To exercise it in a preview environment: with `billing_sdk_subscription_enabled` on, start a subscribe that lands on a hosted verification/3DS step, then open the subscription panel before completing it. On this branch the banner and the button are there; on `main` the panel shows the plan tile as though nothing were pending. Flag off, the panel is unchanged — the legacy store answers both reads exactly as it does today. ## Testing - `useSubscriptionOperationView.test.ts` — the poller answers both reads with the rail off and the SDK is never constructed; the lifecycle answers both with the rail on while the poller holds nothing; a two-row table over the rail proving neither rail offers another workspace's hosted step (the lifecycle keeps operations from a scope it has left, so the filter is load-bearing, not incidental); and a non-https step refused on the SDK rail. - `operationRecordView.test.ts` — a row for the `javascript:` step the projection now refuses, alongside the https step it carries. `pnpm test:unit src/platform/workspace src/composables/billing` (92 files, 1958 tests), `pnpm typecheck` and `pnpm lint` pass. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 8**, one of the three switches PR 1 unblocked. Parallel to PRs 7 and 9; no ordering between them. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | merged — Comfy-Org#18036 | | 2 | FE-2467 | billing events reader in `account-core` | open — Comfy-Org#18042 | | 3 | FE-2479 | usage log reads on the SDK rail | open — Comfy-Org#18047, needs 2 | | 4 | FE-2468 | payment recovery reads on the rail | open — Comfy-Org#18060 | | 5 | FE-2470 | team checkout on the SDK subscribe command | independent | | 6 | FE-2471 | one status reader for the cloud subscription store | open — Comfy-Org#18092 | | 7 | FE-2480 + FE-2481 | stop registering a second poller | needs 1 | | **8** | **FE-2482** | **subscription panel reads the SDK store** | **this PR** | | 9 | FE-2484 + FE-2483 + FE-2485 | recovery through the SDK lifecycle | needs 1 | | 10 | FE-2474 | top-up dialog reads through its adapter | open — Comfy-Org#18132 | | 11 | FE-2481 | one `proration_at` rule | backend answer | | 12 | FE-2472 | table-driven safety net | independent | ```mermaid flowchart LR PR1["PR 1 · FE-2469<br/>merged"] --> PR8["PR 8 · FE-2482<br/>panel on the rail"]:::here PR1 --> PR7["PR 7 · FE-2480 + FE-2481"] PR1 --> PR9["PR 9 · FE-2484 + FE-2483 + FE-2485"] PR2["PR 2 · FE-2467"] --> PR3["PR 3 · FE-2479"] IND["PR 4, 5, 6, 10, 11, 12<br/>independent"] PR8 --> BAKE["Step 3 · staging bake"] PR7 --> BAKE PR9 --> BAKE PR3 --> BAKE IND --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2482
…ption store (Comfy-Org#18092) ## Why `performFetchSubscriptionStatus` in `useSubscription.ts` read billing status straight off `workspaceApi.getBillingStatus()` with no rail branch, in parallel with `useWorkspaceBilling.fetchStatus` (`:336`) which has taken `useBillingReadRail()` since FE-2216. Two status readers, one railed and one not. With either billing SDK flag on, that is two clients answering the same question. The unrailed one is the earlier of the two: it fires on the auth watch at bootstrap, and its response is what writes `workspaceStore.setWorkspaceBillingRail`. So the read that feeds the legacy-versus-workspace routing decision was the one read still going through the client the rail replaced. ## The decision FE-2471 is a decision ticket. It offers two options, and this PR takes **option 1 — read on the rail here as well, keeping two readers but one transport.** **Option 2, delegating to `useWorkspaceBilling.fetchStatus` and deleting this fetch, is rejected on layering.** This read's response is the input to the routing decision: `setWorkspaceBillingRail` feeds `workspaceStore.activeWorkspaceBillingRail`, which `useBillingRouting.ts:31` reads to choose between `useLegacyBilling` and `useWorkspaceBilling`. Delegating puts that input inside one of the two adapters being chosen between — and `useLegacyBilling.ts:49` reads this very composable, so the legacy adapter would have to call the workspace adapter to learn which adapter it should be. At bootstrap `activeWorkspace?.type` is undefined, `useBillingRouting.type` is therefore `'legacy'`, and the value that would settle it has not been read yet. Two further costs make it the wrong move for Step 1: - **`useWorkspaceBilling.fetchStatus` carries side effects a bootstrap read must not acquire.** It writes `seatCapacity` (`:349`) and calls `resumePendingOperation` (`:304`), which registers a legacy poller or calls `sdk.recover()`. That is precisely the code PRs 7 and 9 are rewriting; making app bootstrap enter it would widen their blast radius for no gain. - **Its staleness fence is workspace-shaped.** `isStaleStatusRead` (`:326`) drops any read whose captured `activeWorkspace?.id` no longer matches, where this composable fences on `(userId, activeWorkspaceId)` — the identity pair a user-scoped bootstrap read actually needs, and the one the "identity switch" test in this file pins. What the ticket's title complains about is better read as two *transports* than two readers, and one transport is what this PR leaves. **The consequence for the plan: FE-2495 (M1-26) closes as not needed** — its own description says to close it if option 1 is chosen. ## How The read comes out of `performFetchSubscriptionStatus` into `readSubscriptionStatus`, which takes `useBillingReadRail()` and holds it for that read — a flag flip mid-read must not start on one client and publish through the other: ```ts async function readSubscriptionStatus(): Promise<BillingStatusResponse | undefined> { const rail = useBillingReadRail() try { return rail ? await readOnRail(rail.readStatus) : await workspaceApi.getBillingStatus() } catch (error) { throw new AuthStoreError( t('toastMessages.failedToFetchSubscription', { … }) ) } } ``` Splitting it is worth a word, since the rail branch alone would have been three lines inline. Reading and publishing are two concerns, and only the first one grew: which client answers and how its failure is shaped now sit in one named function, leaving the caller with the scope guards and the publish. It also drops a `let` that was assigned in a `try`. Inline, the branch tipped `performFetchSubscriptionStatus` over fallow's CRAP threshold (37.1 against 30, `introduced: true`); extracting puts it back under without touching a test. `readOnRail` returns `undefined` on `SUPERSEDED`, so the status is optional and a superseded read returns `null` before the publish — no status, no `setWorkspaceBillingRail`, no `syncPendingSubscriptionSuccess`. That is the same nothing the identity guard directly below produces from a read whose owner or workspace has changed, which is why it sits next to it rather than in its own branch. Nothing else moved. The error path is behaviourally untouched: a failed rail read throws `WorkspaceApiError` out of `readOnRail` and hits the same `catch`, wrapped in the same `AuthStoreError` with the same `failedToFetchSubscription` message the legacy throw produced. The in-flight dedupe, the pending-checkout telemetry and the `billing_rail` write are all as they were. This is the same shape FE-2216 gave `useBillingPlans.fetchPlans` in Comfy-Org#17942 — the closest precedent, a cloud subscription composable reading through `useBillingReadRail` with a superseded read publishing nothing. ## AS IS / TO BE **No user-visible change, on either rail.** Same status, same panels, same gate. What changes is which client answers. | | AS IS (flag on) | TO BE (flag on) | | --- | --- | --- | | bootstrap status read | `workspaceApi.getBillingStatus()` | `readStatus` on the read rail | | `setWorkspaceBillingRail` source | legacy client's response | the rail's response | | superseded scope | stale read published | nothing published | | flag off | `workspaceApi.getBillingStatus()` | unchanged | Worth stating explicitly, since the read is on the bootstrap path: **the rail read has no weaker auth precondition than the legacy one.** Under `unified_cloud_auth` — the flag these billing rails ramp behind — `getWorkspaceAuthHeader` returns the unified token or null, and a null throws `userNotAuthenticated` without waiting; the SDK's scope source reads the same unified session and answers `NOT_AUTHENTICATED` when it is not established. Both paths need a minted unified session, so railing this read adds no new unauthenticated-bootstrap window. A before/after capture would show the same screen, so none is attached. ## Testing `useSubscription.test.ts` — the two behaviours that hold on both clients are now rows over both, so the rail path is pinned by the same assertions as the legacy path rather than by a copy: - **publishes a status read through `$reader` and updates the workspace billing rail** — over the workspace client and the SDK reader. Each row also asserts the *other* client was never called, which is the one-transport claim the ticket is about. - **reports a failed read through `$reader` in the same message** — a legacy rejection and a rail `ACCESS_DENIED` both surface as `Failed to fetch subscription status: …`. - **keeps the published status when a rail read is superseded** — rail-only: a good status is published first, then a `SUPERSEDED` read resolves `null` and leaves it standing. These replace the three legacy-only cases they subsume (`should fetch subscription status successfully`, `should handle fetch errors gracefully`, `updates the active workspace billing rail from status`) with no loss of legacy coverage — the legacy row asserts strictly more than the case it replaced. Reverting the source change and re-running fails exactly the three rail cases; the legacy rows pass on both sides, which is the flag-off half of the dual path. `pnpm vitest run src/platform/cloud src/platform/workspace src/composables/billing` (145 files, 2598 tests), `pnpm typecheck`, `pnpm lint` and `pnpm knip` all pass. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 6**, independent: it depends on nothing in Step 1 and blocks nothing in it. Its output is a plan change rather than a dependency — it closes FE-2495 out of Step 5. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | merged (Comfy-Org#18036) | | 2 | FE-2467 | billing events reader in `account-core` | open (Comfy-Org#18042) | | 3 | FE-2479 | usage log reads on the SDK rail | open (Comfy-Org#18047), stacked on 2 | | 4 | FE-2468 | payment recovery reads on the rail | open (Comfy-Org#18060) | | 5 | FE-2470 | team checkout on the SDK subscribe command | independent | | **6** | **FE-2471** | **one status transport for the cloud subscription store** | **this PR** | | 7 | FE-2480 + FE-2481 | stop registering a second poller | needs 1 | | 8 | FE-2482 | subscription panel reads the SDK store | needs 1 | | 9 | FE-2484 + FE-2483 + FE-2485 | recovery through the SDK lifecycle | needs 1 | | 10 | FE-2474 | top-up dialog reads through its adapter | independent | | 11 | FE-2481 | one `proration_at` rule | backend answer | | 12 | FE-2472 | table-driven safety net | independent | ```mermaid flowchart LR PR6["PR 6 · FE-2471<br/>one status transport"]:::here PR1["PR 1 · FE-2469<br/>merged"] --> PR7["PR 7 · FE-2480 + FE-2481"] PR1 --> PR8["PR 8 · FE-2482"] PR1 --> PR9["PR 9 · FE-2484 + FE-2483 + FE-2485"] PR2["PR 2 · FE-2467"] --> PR3["PR 3 · FE-2479"] IND["PR 4, 5, 10, 11, 12<br/>independent"] PR6 --> BAKE["Step 3 · staging bake"] PR7 --> BAKE PR8 --> BAKE PR9 --> BAKE PR3 --> BAKE IND --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal<br/>FE-2495 closes as not needed"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` ## Review focus **The decision is the review, not the diff.** The source change is small and follows Comfy-Org#17942; what needs a second opinion is choosing option 1 over delegation, because that choice retires FE-2495. If you read the routing cycle differently — or think the two bootstrap round trips are worth paying the layering cost to remove — this is the PR to say so on, and FE-2495 should stay open. Second, the superseded branch returning `null` rather than rescheduling: a superseded read does not call `syncPendingSubscriptionSuccess`, so it neither clears nor reschedules pending-checkout recovery. That matches what a stale legacy read does today at the identity guard, and the recovery retry is driven separately by the storage/visibility listeners — but it is the one behavioural edge the rail introduces here. - Part of FE-2471
…Comfy-Org#18137) ## Why `useSubscriptionCheckout.ts` is 1817 lines behind a 4840-line suite, and FE-2490 deletes roughly half of it — every legacy branch, once `billing_sdk_subscription_enabled` has been at 100% for a release cycle. The existing suite asserts a great deal, but not as a matrix over the two rails. So a removal PR can go green by deleting the cases that covered the branch it removed, rather than by showing the behaviour survived. That is the failure mode this PR exists to close. ## How A table over **rail × checkoutType × outcome**, added to the existing suite rather than a new file so it reuses the shared billing fixtures — the ticket's own constraint, and the reason there are no new mocks here. - **rail** — legacy and SDK, set through the same `mockSubscriptionRail` seam the rest of the suite uses. - **checkoutType** — `new` and `change`, driven by the quote's `transition_type` exactly as the composable derives it, plus reactivation as its own pair of rows since it is a flag on a change rather than a third type. - **outcome** — the four terminal shapes the projection can hand back: a subscribe the server settled on the spot, one parked on a billing operation that then succeeds, one parked on an operation that then fails, and no response at all. Each row asserts the same two user-visible things on both rails: the step the customer is left on, and that the attempt released. The second is not incidental — a checkout left busy is a dialog the customer can neither leave nor retry from, and it is the state a half-removed branch most easily produces. The rows are sparse. A row names its outcome and the step that outcome must leave behind; the rail and the checkout type are crossed over it rather than restated per case, so the table reads as four behaviours rather than eighteen bodies. **What gives it teeth against FE-2490.** The two parked rows are the ones that run `advanceToSuccessOnOperation` — the legacy registration and the terminal await that Step 5 removes. If that path goes without a replacement that settles the operation, "parked then settles" stops reaching the success step on both rails and this table fails, which is exactly the signal a removal PR needs and does not have today. ## AS IS / TO BE **No product change.** This PR adds test rows and touches no source file, so there is nothing to capture; the suite is green before and after on both rails. What changes is what a future removal has to survive. ## Testing - `useSubscriptionCheckout.test.ts` — 18 matrix rows (2 rails × 2 checkout types × 4 outcomes) plus 2 reactivation rows, asserting the step and the released attempt on each. The file goes from 162 to 180 tests, all passing. `pnpm test:unit src/platform/workspace src/composables/billing` (91 files, 1970 tests), `pnpm typecheck` and `pnpm lint` pass. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 12**, independent of every other PR in the step, and the one Step 5's first removal is checked against. Note for whoever merges second: this touches `useSubscriptionCheckout.test.ts` alongside PR 7 (Comfy-Org#18134), which widens the rail stub in the same file. Different blocks, trivial rebase. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | merged — Comfy-Org#18036 | | 2 | FE-2467 | billing events reader in `account-core` | open — Comfy-Org#18042 | | 3 | FE-2479 | usage log reads on the SDK rail | open — Comfy-Org#18047, needs 2 | | 4 | FE-2468 | payment recovery reads on the rail | open — Comfy-Org#18060 | | 5 | FE-2470 | team checkout on the SDK subscribe command | open — Comfy-Org#18136 | | 6 | FE-2471 | one status reader for the cloud subscription store | open — Comfy-Org#18092 | | 7 | FE-2480 + FE-2481 | checkout reads its operation off the rail | open — Comfy-Org#18134 | | 8 | FE-2482 | subscription panel reads the SDK store | open — Comfy-Org#18133 | | 9 | FE-2484 + FE-2483 + FE-2485 | recovery through the SDK lifecycle | open — Comfy-Org#18135 | | 10 | FE-2474 | top-up dialog reads through its adapter | open — Comfy-Org#18132 | | 11 | FE-2481 | one `proration_at` rule | backend answer | | **12** | **FE-2472** | **table-driven safety net** | **this PR** | ```mermaid flowchart LR PR12["PR 12 · FE-2472<br/>safety net"]:::here PR1["PR 1 · FE-2469<br/>merged"] --> PR789["PR 7, 8, 9"] PR2["PR 2 · FE-2467"] --> PR3["PR 3 · FE-2479"] IND["PR 4, 5, 6, 10, 11<br/>independent"] PR789 --> BAKE["Step 3 · staging bake"] PR3 --> BAKE IND --> BAKE PR12 --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal<br/>FE-2490 checked against PR 12"] PR12 -.-> REMOVE classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2472
…omfy-Org#18132) ## Why `useTopupOperation` owns the rail choice for the top-up dialog — it picks the SDK store or the legacy poller once, when the dialog opens, so a flag flip mid-purchase cannot issue on one rail and observe on the other. The dialog then reaches past it anyway: it imports `useBillingOperationStore` directly (`:314`), instantiates it (`:343`), and calls `startOperation` on it after a purchase comes back pending (`:677`). With `billing_sdk_topup_enabled` on, that is a direct reference from a dialog on the SDK rail to a store nothing is driving its operation with. FE-2474 removes the three lines. ## Root cause The adapter was extracted from the dialog in Comfy-Org#17665 for the reads and the commands — `isAddingCredits`, `topupOperation`, `topup`, `retryPaymentAuthentication`, `dismissOperation`. The pending-registration was left behind because it is not a read or a command: it is the dialog asking to be told when an operation it just issued settles, and on the legacy rail `startOperation` is both the registration and that notification. That one call kept the import alive, and with it the dialog's binding to the legacy store. ## How **`useTopupOperation.adoptPendingOperation(opId, { attemptStartedAt })`** is the sixth thing the adapter answers. On the legacy rail it is the `startOperation` call the dialog made, unchanged — same operation type, same `autoHandleRequiresAction: true`, same promise that resolves when the operation settles. On the SDK rail it resolves immediately: the lifecycle adopted the operation when `createTopup` issued it, so there is nothing to register, and registering would be a second poller on one operation. **The dialog destructures it alongside the other five and drops the three lines.** The `.then` / `.catch` around the call is unchanged — it still clears `paymentSubmitted` on settlement and reports a purchase error on rejection. One thing worth naming for review: **this call site is already unreachable on the SDK rail.** `projectTopupResult` maps a settled result to `completed` or `failed` and an unsettled one to `undefined`, so `response.status === 'pending'` is a legacy-only shape. The fix is therefore not a behaviour change on either rail — it is removing the dialog's ability to reach a store it is not using, which is what makes the legacy branches deletable in FE-2492 without a search through component files. ## AS IS / TO BE **No user-visible change, on either rail.** The legacy rail issues the same registration with the same arguments; the SDK rail takes a branch it could not reach before and cannot reach now. A before/after capture would show two identical top-up dialogs, so none is attached. ## Testing - `useTopupOperation.test.ts` — a two-row table over the rail asserting exactly one poller registration per pending top-up: the legacy rail forwards to `startOperation` with the operation id, type and metadata, and the SDK rail registers nothing. This is the dual-path assertion the milestone asks of every rail-touching change. - `TopUpCreditsDialogContentWorkspace.test.ts` (37 tests) passes unchanged, which is the point — the dialog's behaviour is identical through the adapter. `pnpm test:unit src/platform/workspace` , `pnpm typecheck` and `pnpm lint` pass. ## Plan position Milestone 1, Step 1 — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 10**, independent of every other PR in the step. | PR | Tickets | What | Status | | --- | --- | --- | --- | | 1 | FE-2469 | operation-state projections on the SDK store | merged — Comfy-Org#18036 | | 2 | FE-2467 | billing events reader in `account-core` | open — Comfy-Org#18042 | | 3 | FE-2479 | usage log reads on the SDK rail | open — Comfy-Org#18047, needs 2 | | 4 | FE-2468 | payment recovery reads on the rail | open — Comfy-Org#18060 | | 5 | FE-2470 | team checkout on the SDK subscribe command | independent | | 6 | FE-2471 | one status reader for the cloud subscription store | open — Comfy-Org#18092 | | 7 | FE-2480 + FE-2481 | stop registering a second poller | needs 1 | | 8 | FE-2482 | subscription panel reads the SDK store | needs 1 | | 9 | FE-2484 + FE-2483 + FE-2485 | recovery through the SDK lifecycle | needs 1 | | **10** | **FE-2474** | **top-up dialog reads through its adapter** | **this PR** | | 11 | FE-2481 | one `proration_at` rule | backend answer | | 12 | FE-2472 | table-driven safety net | independent | ```mermaid flowchart LR PR10["PR 10 · FE-2474<br/>top-up dialog adapter"]:::here PR1["PR 1 · FE-2469<br/>merged"] --> PR789["PR 7, 8, 9"] PR2["PR 2 · FE-2467"] --> PR3["PR 3 · FE-2479"] IND["PR 4, 5, 6, 11, 12<br/>independent"] PR10 --> BAKE["Step 3 · staging bake"] PR789 --> BAKE PR3 --> BAKE IND --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2474
…rtal (Comfy-Org#18139) ## Why Comfy-Org#17853 put cancel, resubscribe **and** the payment portal on the subscription rail. `cancelSubscriptionSdkRail.spec.ts` covered cancel only, so two of the three shipped with no browser case on either rail — and the portal is the one with the most branching, because Comfy-Org#17922 added a three-destination cascade behind it. ## How `browser_tests/tests/dialogs/subscriptionLifecycleSdkRail.spec.ts`, a sibling of the cancel spec rather than an extension of it: both actions need a cancelled subscription and a populated plan catalog, which the cancel spec's active-subscription fixture does not have. Six rows plus the cascade. **Resubscribe — off / on / 404.** Same discriminator as cancel: the SDK transport runs on `fetch` and sends an `Idempotency-Key` matching the body's `idempotency_key`; the legacy workspace client runs on axios' XHR and sends neither. The 404 row proves the fallback fires once and the second attempt is the legacy one. **The six-row `/api/billing/plans` fixture** is the gap Comfy-Org#17853's evidence run recorded. The Reactivate plan button reads the catalog to name the plan being resumed, so an empty or one-row `plans` response never renders it — which is why resubscribe had no browser case to extend in the first place. **Payment portal — off / on / 404,** with one assertion that is deliberately a non-defect: `openPaymentPortal` carries **no** idempotency key on either rail, because it is a plain POST outside the operation lifecycle. The spec asserts its absence rather than treating it as missing, so a later change that adds one is noticed rather than assumed correct. **The blocked-popup cascade.** `manageSubscription` tries three destinations in order — the hosted billing-web page, then the rail, then the legacy call — and each opens a different page, so a browser refusing one says nothing about the next. The last row refuses every `window.open` and asserts the rail is still asked and the legacy call still runs after it. A refused tab must not be read as "the portal is handled", which is the regression Comfy-Org#17922 fixed. The `window.open` stand-in is the one from `hostedBillingDestination.spec.ts` — a proxy that records the destination instead of navigating, so a disowned-tab open cannot steer the test page — extended to record more than one destination, since the cascade opens several. ## AS IS / TO BE **No product change** — this PR adds a spec file and touches no source. There is nothing to capture; what changes is what CI proves about code already on `main`. ## Testing - `subscriptionLifecycleSdkRail.spec.ts` (new, `@cloud`) — 7 rows as above. - `pnpm typecheck`, `pnpm typecheck:browser` and `pnpm lint` pass. **This spec has not been executed locally.** It needs a cloud backend on `:8188`, which this checkout has no access to, so CI is its first run. Reviewers should treat a red `@cloud` shard here as the spec's own problem rather than a regression, and I will chase it. ## Plan position Milestone 1, Step 1b — [PR plan](https://linear.app/comfyorg/document/milestone-1-pr-plan-605f42c5e709). This is **PR 15**, backfill for code already on `main` (Comfy-Org#17853, Comfy-Org#17922), independent of every Step 1 PR and of the other two Step 1b PRs. | PR | Step | Tickets | What | Status | | --- | --- | --- | --- | --- | | 1 | 1 | FE-2469 | operation-state projections | merged — Comfy-Org#18036 | | 2 | 1 | FE-2467 | billing events reader | open — Comfy-Org#18042 | | 3 | 1 | FE-2479 | usage log on the rail | open — Comfy-Org#18047 | | 4 | 1 | FE-2468 | payment recovery on the rail | open — Comfy-Org#18060 | | 5 | 1 | FE-2470 | team checkout on the rail | open — Comfy-Org#18136 | | 6 | 1 | FE-2471 | one status reader | open — Comfy-Org#18092 | | 7 | 1 | FE-2480 + FE-2481 | checkout reads on the rail | open — Comfy-Org#18134 | | 8 | 1 | FE-2482 | subscription panel on the rail | open — Comfy-Org#18133 | | 9 | 1 | FE-2484 + FE-2483 + FE-2485 | recovery on the rail | open — Comfy-Org#18135 | | 10 | 1 | FE-2474 | top-up dialog adapter | open — Comfy-Org#18132 | | 11 | 1 | FE-2481 | one `proration_at` rule | backend answer | | 12 | 1 | FE-2472 | table-driven safety net | open — Comfy-Org#18137 | | 13 | 1b | FE-2475 | top-up rail E2E | open — Comfy-Org#18138 | | 14 | 1b | FE-2476 | six reads E2E | needs 3 | | **15** | **1b** | **FE-2477** | **resubscribe + portal E2E** | **this PR** | ```mermaid flowchart LR PR15["PR 15 · FE-2477<br/>resubscribe + portal E2E"]:::here STEP1["Step 1 · PRs 1-12"] PR13["PR 13 · FE-2475"] PR3["PR 3 · FE-2479"] --> PR14["PR 14 · FE-2476"] STEP1 --> BAKE["Step 3 · staging bake"] PR13 --> BAKE PR14 --> BAKE PR15 --> BAKE BAKE --> RAMP["Step 4 · production ramp"] --> REMOVE["Step 5 · legacy removal"] classDef here fill:#f2994a,stroke:#b35c00,color:#1a1a1a,font-weight:bold ``` - Part of FE-2477
Why
Four surfaces that drive a checkout read state off
billingOperationStorewith no rail branch: the subscription panel's "setting up" banner, the checkout's active-operation lookup, the pending-operation guard inuseWorkspaceBilling, and the top-up dialog. Withbilling_sdk_subscription_enabledon, the operation those surfaces describe lives in the SDK lifecycle, so they read a store nothing is writing.They cannot move until the SDK store answers the same questions. This PR makes them answerable; the call sites move in FE-2480, FE-2482, FE-2483/FE-2484 and FE-2474.
Root cause
billingOperationStorewas two things at once — the poller and the view-model the billing UI reads. The SDK replaced the poller half only:billingSdkStoreexposes the commands and the top-up projections, but nothing equivalent toisSettingUp,subscriptionActionOperation,getOperation(id)orhasPendingOperations. That missing half is what has kept the legacy store alive after every command already had a rail.How
projectOperationRecord(new,operationRecordView.ts) maps oneBillingOperationStateonto the record shape the legacy poller handed out — the fields the consumers actually read, and no more.supersededprojects to nothing: the scope moved on under the operation, the same outcomereadOnRailgives a superseded read.billingSdkStorethen exposeshasPendingOperations,isSettingUp,subscriptionActionOperationandgetOperationover that projection.Two deliberate choices:
isSettingUpandsubscriptionActionOperationfilter on the active workspace;hasPendingOperationsdoes not. That asymmetry isbillingOperationStore's, and copying it is the point — a consumer switching rails must not find a different answer on the other side. The lifecycle keeps operations from a scope it has left, so the filter is load-bearing here rather than incidental.getOperationdoes not filter by workspace, and exposesworkspaceIdinstead.useSubscriptionCheckout.ts:203already does its own comparison against the active workspace; leaving that comparison where it is keeps FE-2480 a one-line switch instead of a semantics change.subscriptionActionOperationis beyond the three projections FE-2469 lists. It is included becauseuseSubscriptionCheckout.ts:197reads it as the fallback foractiveCheckoutOperationand then reads.status,.isAuthenticating,.phaseand.authenticationStateoff the result at:239–:251. The existingsubscriptionActionUrlcovers only the URL, so without this FE-2480 would not be a switch.AS IS / TO BE
No user-visible change, on either rail. Nothing consumes these projections yet — this PR only adds exports to
billingSdkStoreand a new pure module. The flag-off path is untouched, and the flag-on path gains no new behaviour until the call sites move in the follow-ups above. A before/after capture would show two identical screens, so none is attached.The first user-visible change in this chain is FE-2482, which will carry its own AS IS / TO BE.
Testing
operationRecordView.test.ts— 12 table-driven rows over the projection: pending detail, the hosted step, the server phase, both challenge states, each settled phase (includingtimed_out→timeoutand reconciliation's own authentication state), andsupersededprojecting to nothing.billingSdkStore.test.ts— the four store predicates, including a table overisSettingUpthat pins the customer-attention exclusions and the cross-workspace case, andgetOperationreturning another workspace's record so the caller keeps its own scope check.pnpm test:unit(src/platform/workspace + src/composables/billing: 91 files, 1952 tests),pnpm typecheck,pnpm lintandpnpm knipall pass.Plan position
Milestone 1, Step 1 — PR plan. This is PR 1, the keystone: it unblocks PRs 7, 8 and 9, each of which collapses to a short switch once it lands.
account-coreproration_atrule