@tanstack/table-core 9.0.0-beta.50 → 9.0.0-beta.52

package/dist/worker/initTableWorker.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"initTableWorker.cjs","names":["makeObjectMap","constructTable","storeReactivityBindings","serializeRowModel"],"sources":["../../src/worker/initTableWorker.ts"],"sourcesContent":["import { constructTable } from '../core/table/constructTable'\nimport { storeReactivityBindings } from '../store-reactivity-bindings'\nimport { makeObjectMap } from '../utils'\nimport { serializeRowModel } from './serializeRowModel'\nimport type { RowData } from '../types/type-utils'\nimport type { TableFeatures } from '../types/TableFeatures'\nimport type { TableOptions } from '../types/TableOptions'\nimport type { Table_Internal } from '../types/Table'\nimport type {\n  TableWorkerRequest,\n  TableWorkerResult,\n  TableWorkerStage,\n  TableWorkerStagePayload,\n} from './tableWorkerProtocol'\n\nexport type TableWorkerConfig<\n  TFeatures extends TableFeatures,\n  TData extends RowData,\n> = Omit<TableOptions<TFeatures, TData>, 'data'>\n\nfunction capitalize(stage: string) {\n  return stage.charAt(0).toUpperCase() + stage.slice(1)\n}\n\n/** Flatten column-group defs to leaf defs (mirrors core's id resolution). */\nfunction flattenColumnDefs(defs: Array<any>): Array<any> {\n  return defs.flatMap((def) =>\n    def.columns ? flattenColumnDefs(def.columns) : [def],\n  )\n}\n\n/**\n * Runs a headless \"shadow table\" inside a dedicated Web Worker.\n *\n * Call this from a user-authored worker entry file, passing the same columns\n * and processing features used on the main thread. The shadow table runs the\n * real table-core row model pipeline (real fns, real Row objects) off the\n * main thread and posts back one payload per stage the main thread requested:\n * a transferable index permutation for flat results, a serialized row tree\n * (with eagerly computed aggregates) when grouping produces synthetic rows.\n *\n * Everything passed here must be thread-portable: `accessorKey` columns or\n * accessors defined in a shared module, and fns from registries or shared\n * modules (no closures over app state).\n *\n * @example\n * ```ts\n * // table.worker.ts\n * import { initTableWorker } from '@tanstack/table-core/experimental-worker-plugin'\n * import { columns, sharedFeatures } from './tableConfig'\n *\n * initTableWorker({ features: sharedFeatures, columns })\n * ```\n */\nexport function initTableWorker<\n  TFeatures extends TableFeatures,\n  TData extends RowData,\n>(config: TableWorkerConfig<TFeatures, TData>): void {\n  let table: Table_Internal<TFeatures, TData> | undefined\n  let dataVersion = 0\n  let coreIndexById: Record<string, number> = makeObjectMap()\n  let aggregateColumnIds: Array<string> = []\n  // Last-sent model identity per stage: the memoized getters return stable\n  // objects when their inputs did not change, so identity equality is exactly\n  // \"this stage's result is unchanged\".\n  let lastSentModels: { [K in TableWorkerStage]?: unknown } = {}\n\n  self.onmessage = (event: MessageEvent<TableWorkerRequest<TData>>) => {\n    const message = event.data\n\n    if (message.type === 'data') {\n      dataVersion = message.dataVersion\n      lastSentModels = {}\n      if (!table) {\n        table = constructTable<TFeatures, TData>({\n          ...(config as TableOptions<TFeatures, TData>),\n          features: {\n            coreReactivityFeature: storeReactivityBindings(),\n            ...config.features,\n          },\n          data: message.data,\n        })\n        // Only columns with an explicit aggregation get eagerly aggregated\n        // per group. Sync tables aggregate lazily (visible cells only), so\n        // auto-aggregating every column would explode on high-cardinality\n        // grouping for values nothing renders. Read the RAW column defs:\n        // aggregationFeature injects default aggregatedCell/aggregationFn\n        // into every resolved columnDef, so the resolved defs can't tell\n        // explicit from default.\n        aggregateColumnIds = flattenColumnDefs(config.columns as Array<any>)\n          .filter(\n            (def) => def.aggregationFn != null || def.aggregatedCell != null,\n          )\n          .map(\n            (def) =>\n              def.id ??\n              (typeof def.accessorKey === 'string'\n                ? def.accessorKey.replaceAll('.', '_')\n                : undefined),\n          )\n          .filter((id): id is string => id != null)\n      } else {\n        table.setOptions((prev) => ({ ...prev, data: message.data }))\n      }\n      // Map row ids to data positions once per dataset; serialization uses it\n      // to express every stage result in terms of core row positions.\n      const coreFlatRows = table.getCoreRowModel().flatRows\n      coreIndexById = makeObjectMap()\n      for (let i = 0; i < coreFlatRows.length; i++) {\n        coreIndexById[coreFlatRows[i]!.id] = i\n      }\n      return\n    }\n\n    if (!table) return\n\n    const start = performance.now()\n\n    // Apply the serializable state slices to the shadow table's base atoms.\n    table._reactivity.batch(() => {\n      for (const [key, value] of Object.entries(message.state)) {\n        const baseAtom = (table!.baseAtoms as Record<string, any>)[key]\n        if (baseAtom && value !== undefined) {\n          baseAtom.set(value)\n        }\n      }\n    })\n\n    // Compute exactly the stages the main thread requested, skipping any this\n    // shadow table has no row model factory for (the main thread warns).\n    const stages: { [K in TableWorkerStage]?: TableWorkerStagePayload } = {}\n    const transfer: Array<Transferable> = []\n\n    for (const stage of message.stages) {\n      if (!(config.features as Record<string, unknown>)[`${stage}RowModel`]) {\n        continue\n      }\n      const model = (table as any)[`get${capitalize(stage)}RowModel`]()\n      // Memoized getters return the same object when inputs are unchanged;\n      // skip re-serializing (and the main thread skips rebuilding). Safe under\n      // single-flight: results are never dropped within a dataVersion.\n      if (lastSentModels[stage] === model) {\n        stages[stage] = { kind: 'unchanged' }\n        continue\n      }\n      lastSentModels[stage] = model\n      stages[stage] = serializeRowModel(\n        model,\n        coreIndexById,\n        aggregateColumnIds,\n        transfer,\n      )\n    }\n\n    const response: TableWorkerResult = {\n      type: 'result',\n      requestId: message.requestId,\n      dataVersion,\n      stages,\n      computeMs: performance.now() - start,\n    }\n    postMessage(response, { transfer })\n  }\n}\n"],"mappings":";;;;;;AAoBA,SAAS,WAAW,OAAe;CACjC,OAAO,MAAM,OAAO,CAAC,CAAC,CAAC,YAAY,IAAI,MAAM,MAAM,CAAC;AACtD;;AAGA,SAAS,kBAAkB,MAA8B;CACvD,OAAO,KAAK,SAAS,QACnB,IAAI,UAAU,kBAAkB,IAAI,OAAO,IAAI,CAAC,GAAG,CACrD;AACF;;;;;;;;;;;;;;;;;;;;;;;;AAyBA,SAAgB,gBAGd,QAAmD;CACnD,IAAI;CACJ,IAAI,cAAc;CAClB,IAAI,gBAAwCA,4BAAc;CAC1D,IAAI,qBAAoC,CAAC;CAIzC,IAAI,iBAAwD,CAAC;CAE7D,KAAK,aAAa,UAAmD;EACnE,MAAM,UAAU,MAAM;EAEtB,IAAI,QAAQ,SAAS,QAAQ;GAC3B,cAAc,QAAQ;GACtB,iBAAiB,CAAC;GAClB,IAAI,CAAC,OAAO;IACV,QAAQC,sCAAiC;KACvC,GAAI;KACJ,UAAU;MACR,uBAAuBC,0DAAwB;MAC/C,GAAG,OAAO;KACZ;KACA,MAAM,QAAQ;IAChB,CAAC;IAQD,qBAAqB,kBAAkB,OAAO,OAAqB,CAAC,CACjE,QACE,QAAQ,IAAI,iBAAiB,QAAQ,IAAI,kBAAkB,IAC9D,CAAC,CACA,KACE,QACC,IAAI,OACH,OAAO,IAAI,gBAAgB,WACxB,IAAI,YAAY,WAAW,KAAK,GAAG,IACnC,OACR,CAAC,CACA,QAAQ,OAAqB,MAAM,IAAI;GAC5C,OACE,MAAM,YAAY,UAAU;IAAE,GAAG;IAAM,MAAM,QAAQ;GAAK,EAAE;GAI9D,MAAM,eAAe,MAAM,gBAAgB,CAAC,CAAC;GAC7C,gBAAgBF,4BAAc;GAC9B,KAAK,IAAI,IAAI,GAAG,IAAI,aAAa,QAAQ,KACvC,cAAc,aAAa,EAAE,CAAE,MAAM;GAEvC;EACF;EAEA,IAAI,CAAC,OAAO;EAEZ,MAAM,QAAQ,YAAY,IAAI;EAG9B,MAAM,YAAY,YAAY;GAC5B,KAAK,MAAM,CAAC,KAAK,UAAU,OAAO,QAAQ,QAAQ,KAAK,GAAG;IACxD,MAAM,WAAY,MAAO,UAAkC;IAC3D,IAAI,YAAY,UAAU,QACxB,SAAS,IAAI,KAAK;GAEtB;EACF,CAAC;EAID,MAAM,SAAgE,CAAC;EACvE,MAAM,WAAgC,CAAC;EAEvC,KAAK,MAAM,SAAS,QAAQ,QAAQ;GAClC,IAAI,CAAE,OAAO,SAAqC,GAAG,MAAM,YACzD;GAEF,MAAM,QAAS,MAAc,MAAM,WAAW,KAAK,EAAE,UAAU,CAAC;GAIhE,IAAI,eAAe,WAAW,OAAO;IACnC,OAAO,SAAS,EAAE,MAAM,YAAY;IACpC;GACF;GACA,eAAe,SAAS;GACxB,OAAO,SAASG,4CACd,OACA,eACA,oBACA,QACF;EACF;EAEA,MAAM,WAA8B;GAClC,MAAM;GACN,WAAW,QAAQ;GACnB;GACA;GACA,WAAW,YAAY,IAAI,IAAI;EACjC;EACA,YAAY,UAAU,EAAE,SAAS,CAAC;CACpC;AACF"}
+{"version":3,"file":"initTableWorker.cjs","names":["makeObjectMap","constructTable","storeReactivityBindings","serializeRowModel"],"sources":["../../src/worker/initTableWorker.ts"],"sourcesContent":["import { constructTable } from '../core/table/constructTable'\nimport { storeReactivityBindings } from '../store-reactivity-bindings'\nimport { makeObjectMap } from '../utils'\nimport { serializeRowModel } from './serializeRowModel'\nimport type { RowData } from '../types/type-utils'\nimport type { TableFeatures } from '../types/TableFeatures'\nimport type { TableOptions } from '../types/TableOptions'\nimport type { Table_Internal } from '../types/Table'\nimport type {\n  TableWorkerRequest,\n  TableWorkerResult,\n  TableWorkerStage,\n  TableWorkerStagePayload,\n} from './tableWorkerProtocol'\n\nexport type TableWorkerConfig<\n  TFeatures extends TableFeatures,\n  TData extends RowData,\n> = Omit<TableOptions<TFeatures, TData>, 'data'>\n\nfunction capitalize(stage: string) {\n  return stage.charAt(0).toUpperCase() + stage.slice(1)\n}\n\n/** Flatten column-group defs to leaf defs (mirrors core's id resolution). */\nfunction flattenColumnDefs(defs: Array<any>): Array<any> {\n  return defs.flatMap((def) =>\n    def.columns ? flattenColumnDefs(def.columns) : [def],\n  )\n}\n\n/**\n * Runs a headless \"shadow table\" inside a dedicated Web Worker.\n *\n * Call this from a user-authored worker entry file, passing the same columns\n * and processing features used on the main thread. The shadow table runs the\n * real table-core row model pipeline (real fns, real Row objects) off the\n * main thread and posts back one payload per stage the main thread requested:\n * a transferable index permutation for flat results, a serialized row tree\n * (with eagerly computed aggregates) when grouping produces synthetic rows.\n *\n * Everything passed here must be thread-portable: `accessorKey` columns or\n * accessors defined in a shared module, and fns from registries or shared\n * modules (no closures over app state).\n *\n * @example\n * ```ts\n * // table.worker.ts\n * import { initTableWorker } from '@tanstack/table-core/experimental-worker-plugin'\n * import { columns, sharedFeatures } from './tableConfig'\n *\n * initTableWorker({ features: sharedFeatures, columns })\n * ```\n */\nexport function initTableWorker<\n  TFeatures extends TableFeatures,\n  TData extends RowData,\n>(config: TableWorkerConfig<TFeatures, TData>): void {\n  let table: Table_Internal<TFeatures, TData> | undefined\n  let dataVersion = 0\n  let coreIndexById: Record<string, number> = makeObjectMap()\n  let aggregateColumnIds: Array<string> = []\n  // Last-sent model identity per stage: the memoized getters return stable\n  // objects when their inputs did not change, so identity equality is exactly\n  // \"this stage's result is unchanged\".\n  let lastSentModels: { [K in TableWorkerStage]?: unknown } = {}\n\n  self.onmessage = (event: MessageEvent<TableWorkerRequest<TData>>) => {\n    const message = event.data\n\n    if (message.type === 'data') {\n      dataVersion = message.dataVersion\n      lastSentModels = {}\n      if (!table) {\n        table = constructTable<TFeatures, TData>({\n          ...(config as TableOptions<TFeatures, TData>),\n          features: {\n            coreReactivityFeature: storeReactivityBindings(),\n            ...config.features,\n          },\n          data: message.data,\n        })\n        // Only columns with an explicit aggregation get eagerly aggregated\n        // per group. Sync tables aggregate lazily (visible cells only), so\n        // auto-aggregating every column would explode on high-cardinality\n        // grouping for values nothing renders. Read the RAW column defs:\n        // rowAggregationFeature injects default aggregatedCell/aggregationFn\n        // into every resolved columnDef, so the resolved defs can't tell\n        // explicit from default.\n        aggregateColumnIds = flattenColumnDefs(config.columns as Array<any>)\n          .filter(\n            (def) => def.aggregationFn != null || def.aggregatedCell != null,\n          )\n          .map(\n            (def) =>\n              def.id ??\n              (typeof def.accessorKey === 'string'\n                ? def.accessorKey.replaceAll('.', '_')\n                : undefined),\n          )\n          .filter((id): id is string => id != null)\n      } else {\n        table.setOptions((prev) => ({ ...prev, data: message.data }))\n      }\n      // Map row ids to data positions once per dataset; serialization uses it\n      // to express every stage result in terms of core row positions.\n      const coreFlatRows = table.getCoreRowModel().flatRows\n      coreIndexById = makeObjectMap()\n      for (let i = 0; i < coreFlatRows.length; i++) {\n        coreIndexById[coreFlatRows[i]!.id] = i\n      }\n      return\n    }\n\n    if (!table) return\n\n    const start = performance.now()\n\n    // Apply the serializable state slices to the shadow table's base atoms.\n    table._reactivity.batch(() => {\n      for (const [key, value] of Object.entries(message.state)) {\n        const baseAtom = (table!.baseAtoms as Record<string, any>)[key]\n        if (baseAtom && value !== undefined) {\n          baseAtom.set(value)\n        }\n      }\n    })\n\n    // Compute exactly the stages the main thread requested, skipping any this\n    // shadow table has no row model factory for (the main thread warns).\n    const stages: { [K in TableWorkerStage]?: TableWorkerStagePayload } = {}\n    const transfer: Array<Transferable> = []\n\n    for (const stage of message.stages) {\n      if (!(config.features as Record<string, unknown>)[`${stage}RowModel`]) {\n        continue\n      }\n      const model = (table as any)[`get${capitalize(stage)}RowModel`]()\n      // Memoized getters return the same object when inputs are unchanged;\n      // skip re-serializing (and the main thread skips rebuilding). Safe under\n      // single-flight: results are never dropped within a dataVersion.\n      if (lastSentModels[stage] === model) {\n        stages[stage] = { kind: 'unchanged' }\n        continue\n      }\n      lastSentModels[stage] = model\n      stages[stage] = serializeRowModel(\n        model,\n        coreIndexById,\n        aggregateColumnIds,\n        transfer,\n      )\n    }\n\n    const response: TableWorkerResult = {\n      type: 'result',\n      requestId: message.requestId,\n      dataVersion,\n      stages,\n      computeMs: performance.now() - start,\n    }\n    postMessage(response, { transfer })\n  }\n}\n"],"mappings":";;;;;;AAoBA,SAAS,WAAW,OAAe;CACjC,OAAO,MAAM,OAAO,CAAC,CAAC,CAAC,YAAY,IAAI,MAAM,MAAM,CAAC;AACtD;;AAGA,SAAS,kBAAkB,MAA8B;CACvD,OAAO,KAAK,SAAS,QACnB,IAAI,UAAU,kBAAkB,IAAI,OAAO,IAAI,CAAC,GAAG,CACrD;AACF;;;;;;;;;;;;;;;;;;;;;;;;AAyBA,SAAgB,gBAGd,QAAmD;CACnD,IAAI;CACJ,IAAI,cAAc;CAClB,IAAI,gBAAwCA,4BAAc;CAC1D,IAAI,qBAAoC,CAAC;CAIzC,IAAI,iBAAwD,CAAC;CAE7D,KAAK,aAAa,UAAmD;EACnE,MAAM,UAAU,MAAM;EAEtB,IAAI,QAAQ,SAAS,QAAQ;GAC3B,cAAc,QAAQ;GACtB,iBAAiB,CAAC;GAClB,IAAI,CAAC,OAAO;IACV,QAAQC,sCAAiC;KACvC,GAAI;KACJ,UAAU;MACR,uBAAuBC,0DAAwB;MAC/C,GAAG,OAAO;KACZ;KACA,MAAM,QAAQ;IAChB,CAAC;IAQD,qBAAqB,kBAAkB,OAAO,OAAqB,CAAC,CACjE,QACE,QAAQ,IAAI,iBAAiB,QAAQ,IAAI,kBAAkB,IAC9D,CAAC,CACA,KACE,QACC,IAAI,OACH,OAAO,IAAI,gBAAgB,WACxB,IAAI,YAAY,WAAW,KAAK,GAAG,IACnC,OACR,CAAC,CACA,QAAQ,OAAqB,MAAM,IAAI;GAC5C,OACE,MAAM,YAAY,UAAU;IAAE,GAAG;IAAM,MAAM,QAAQ;GAAK,EAAE;GAI9D,MAAM,eAAe,MAAM,gBAAgB,CAAC,CAAC;GAC7C,gBAAgBF,4BAAc;GAC9B,KAAK,IAAI,IAAI,GAAG,IAAI,aAAa,QAAQ,KACvC,cAAc,aAAa,EAAE,CAAE,MAAM;GAEvC;EACF;EAEA,IAAI,CAAC,OAAO;EAEZ,MAAM,QAAQ,YAAY,IAAI;EAG9B,MAAM,YAAY,YAAY;GAC5B,KAAK,MAAM,CAAC,KAAK,UAAU,OAAO,QAAQ,QAAQ,KAAK,GAAG;IACxD,MAAM,WAAY,MAAO,UAAkC;IAC3D,IAAI,YAAY,UAAU,QACxB,SAAS,IAAI,KAAK;GAEtB;EACF,CAAC;EAID,MAAM,SAAgE,CAAC;EACvE,MAAM,WAAgC,CAAC;EAEvC,KAAK,MAAM,SAAS,QAAQ,QAAQ;GAClC,IAAI,CAAE,OAAO,SAAqC,GAAG,MAAM,YACzD;GAEF,MAAM,QAAS,MAAc,MAAM,WAAW,KAAK,EAAE,UAAU,CAAC;GAIhE,IAAI,eAAe,WAAW,OAAO;IACnC,OAAO,SAAS,EAAE,MAAM,YAAY;IACpC;GACF;GACA,eAAe,SAAS;GACxB,OAAO,SAASG,4CACd,OACA,eACA,oBACA,QACF;EACF;EAEA,MAAM,WAA8B;GAClC,MAAM;GACN,WAAW,QAAQ;GACnB;GACA;GACA,WAAW,YAAY,IAAI,IAAI;EACjC;EACA,YAAY,UAAU,EAAE,SAAS,CAAC;CACpC;AACF"}

package/dist/worker/initTableWorker.js.map

@@ -1 +1 @@
-{"version":3,"file":"initTableWorker.js","names":[],"sources":["../../src/worker/initTableWorker.ts"],"sourcesContent":["import { constructTable } from '../core/table/constructTable'\nimport { storeReactivityBindings } from '../store-reactivity-bindings'\nimport { makeObjectMap } from '../utils'\nimport { serializeRowModel } from './serializeRowModel'\nimport type { RowData } from '../types/type-utils'\nimport type { TableFeatures } from '../types/TableFeatures'\nimport type { TableOptions } from '../types/TableOptions'\nimport type { Table_Internal } from '../types/Table'\nimport type {\n  TableWorkerRequest,\n  TableWorkerResult,\n  TableWorkerStage,\n  TableWorkerStagePayload,\n} from './tableWorkerProtocol'\n\nexport type TableWorkerConfig<\n  TFeatures extends TableFeatures,\n  TData extends RowData,\n> = Omit<TableOptions<TFeatures, TData>, 'data'>\n\nfunction capitalize(stage: string) {\n  return stage.charAt(0).toUpperCase() + stage.slice(1)\n}\n\n/** Flatten column-group defs to leaf defs (mirrors core's id resolution). */\nfunction flattenColumnDefs(defs: Array<any>): Array<any> {\n  return defs.flatMap((def) =>\n    def.columns ? flattenColumnDefs(def.columns) : [def],\n  )\n}\n\n/**\n * Runs a headless \"shadow table\" inside a dedicated Web Worker.\n *\n * Call this from a user-authored worker entry file, passing the same columns\n * and processing features used on the main thread. The shadow table runs the\n * real table-core row model pipeline (real fns, real Row objects) off the\n * main thread and posts back one payload per stage the main thread requested:\n * a transferable index permutation for flat results, a serialized row tree\n * (with eagerly computed aggregates) when grouping produces synthetic rows.\n *\n * Everything passed here must be thread-portable: `accessorKey` columns or\n * accessors defined in a shared module, and fns from registries or shared\n * modules (no closures over app state).\n *\n * @example\n * ```ts\n * // table.worker.ts\n * import { initTableWorker } from '@tanstack/table-core/experimental-worker-plugin'\n * import { columns, sharedFeatures } from './tableConfig'\n *\n * initTableWorker({ features: sharedFeatures, columns })\n * ```\n */\nexport function initTableWorker<\n  TFeatures extends TableFeatures,\n  TData extends RowData,\n>(config: TableWorkerConfig<TFeatures, TData>): void {\n  let table: Table_Internal<TFeatures, TData> | undefined\n  let dataVersion = 0\n  let coreIndexById: Record<string, number> = makeObjectMap()\n  let aggregateColumnIds: Array<string> = []\n  // Last-sent model identity per stage: the memoized getters return stable\n  // objects when their inputs did not change, so identity equality is exactly\n  // \"this stage's result is unchanged\".\n  let lastSentModels: { [K in TableWorkerStage]?: unknown } = {}\n\n  self.onmessage = (event: MessageEvent<TableWorkerRequest<TData>>) => {\n    const message = event.data\n\n    if (message.type === 'data') {\n      dataVersion = message.dataVersion\n      lastSentModels = {}\n      if (!table) {\n        table = constructTable<TFeatures, TData>({\n          ...(config as TableOptions<TFeatures, TData>),\n          features: {\n            coreReactivityFeature: storeReactivityBindings(),\n            ...config.features,\n          },\n          data: message.data,\n        })\n        // Only columns with an explicit aggregation get eagerly aggregated\n        // per group. Sync tables aggregate lazily (visible cells only), so\n        // auto-aggregating every column would explode on high-cardinality\n        // grouping for values nothing renders. Read the RAW column defs:\n        // aggregationFeature injects default aggregatedCell/aggregationFn\n        // into every resolved columnDef, so the resolved defs can't tell\n        // explicit from default.\n        aggregateColumnIds = flattenColumnDefs(config.columns as Array<any>)\n          .filter(\n            (def) => def.aggregationFn != null || def.aggregatedCell != null,\n          )\n          .map(\n            (def) =>\n              def.id ??\n              (typeof def.accessorKey === 'string'\n                ? def.accessorKey.replaceAll('.', '_')\n                : undefined),\n          )\n          .filter((id): id is string => id != null)\n      } else {\n        table.setOptions((prev) => ({ ...prev, data: message.data }))\n      }\n      // Map row ids to data positions once per dataset; serialization uses it\n      // to express every stage result in terms of core row positions.\n      const coreFlatRows = table.getCoreRowModel().flatRows\n      coreIndexById = makeObjectMap()\n      for (let i = 0; i < coreFlatRows.length; i++) {\n        coreIndexById[coreFlatRows[i]!.id] = i\n      }\n      return\n    }\n\n    if (!table) return\n\n    const start = performance.now()\n\n    // Apply the serializable state slices to the shadow table's base atoms.\n    table._reactivity.batch(() => {\n      for (const [key, value] of Object.entries(message.state)) {\n        const baseAtom = (table!.baseAtoms as Record<string, any>)[key]\n        if (baseAtom && value !== undefined) {\n          baseAtom.set(value)\n        }\n      }\n    })\n\n    // Compute exactly the stages the main thread requested, skipping any this\n    // shadow table has no row model factory for (the main thread warns).\n    const stages: { [K in TableWorkerStage]?: TableWorkerStagePayload } = {}\n    const transfer: Array<Transferable> = []\n\n    for (const stage of message.stages) {\n      if (!(config.features as Record<string, unknown>)[`${stage}RowModel`]) {\n        continue\n      }\n      const model = (table as any)[`get${capitalize(stage)}RowModel`]()\n      // Memoized getters return the same object when inputs are unchanged;\n      // skip re-serializing (and the main thread skips rebuilding). Safe under\n      // single-flight: results are never dropped within a dataVersion.\n      if (lastSentModels[stage] === model) {\n        stages[stage] = { kind: 'unchanged' }\n        continue\n      }\n      lastSentModels[stage] = model\n      stages[stage] = serializeRowModel(\n        model,\n        coreIndexById,\n        aggregateColumnIds,\n        transfer,\n      )\n    }\n\n    const response: TableWorkerResult = {\n      type: 'result',\n      requestId: message.requestId,\n      dataVersion,\n      stages,\n      computeMs: performance.now() - start,\n    }\n    postMessage(response, { transfer })\n  }\n}\n"],"mappings":";;;;;;AAoBA,SAAS,WAAW,OAAe;CACjC,OAAO,MAAM,OAAO,CAAC,CAAC,CAAC,YAAY,IAAI,MAAM,MAAM,CAAC;AACtD;;AAGA,SAAS,kBAAkB,MAA8B;CACvD,OAAO,KAAK,SAAS,QACnB,IAAI,UAAU,kBAAkB,IAAI,OAAO,IAAI,CAAC,GAAG,CACrD;AACF;;;;;;;;;;;;;;;;;;;;;;;;AAyBA,SAAgB,gBAGd,QAAmD;CACnD,IAAI;CACJ,IAAI,cAAc;CAClB,IAAI,gBAAwC,cAAc;CAC1D,IAAI,qBAAoC,CAAC;CAIzC,IAAI,iBAAwD,CAAC;CAE7D,KAAK,aAAa,UAAmD;EACnE,MAAM,UAAU,MAAM;EAEtB,IAAI,QAAQ,SAAS,QAAQ;GAC3B,cAAc,QAAQ;GACtB,iBAAiB,CAAC;GAClB,IAAI,CAAC,OAAO;IACV,QAAQ,eAAiC;KACvC,GAAI;KACJ,UAAU;MACR,uBAAuB,wBAAwB;MAC/C,GAAG,OAAO;KACZ;KACA,MAAM,QAAQ;IAChB,CAAC;IAQD,qBAAqB,kBAAkB,OAAO,OAAqB,CAAC,CACjE,QACE,QAAQ,IAAI,iBAAiB,QAAQ,IAAI,kBAAkB,IAC9D,CAAC,CACA,KACE,QACC,IAAI,OACH,OAAO,IAAI,gBAAgB,WACxB,IAAI,YAAY,WAAW,KAAK,GAAG,IACnC,OACR,CAAC,CACA,QAAQ,OAAqB,MAAM,IAAI;GAC5C,OACE,MAAM,YAAY,UAAU;IAAE,GAAG;IAAM,MAAM,QAAQ;GAAK,EAAE;GAI9D,MAAM,eAAe,MAAM,gBAAgB,CAAC,CAAC;GAC7C,gBAAgB,cAAc;GAC9B,KAAK,IAAI,IAAI,GAAG,IAAI,aAAa,QAAQ,KACvC,cAAc,aAAa,EAAE,CAAE,MAAM;GAEvC;EACF;EAEA,IAAI,CAAC,OAAO;EAEZ,MAAM,QAAQ,YAAY,IAAI;EAG9B,MAAM,YAAY,YAAY;GAC5B,KAAK,MAAM,CAAC,KAAK,UAAU,OAAO,QAAQ,QAAQ,KAAK,GAAG;IACxD,MAAM,WAAY,MAAO,UAAkC;IAC3D,IAAI,YAAY,UAAU,QACxB,SAAS,IAAI,KAAK;GAEtB;EACF,CAAC;EAID,MAAM,SAAgE,CAAC;EACvE,MAAM,WAAgC,CAAC;EAEvC,KAAK,MAAM,SAAS,QAAQ,QAAQ;GAClC,IAAI,CAAE,OAAO,SAAqC,GAAG,MAAM,YACzD;GAEF,MAAM,QAAS,MAAc,MAAM,WAAW,KAAK,EAAE,UAAU,CAAC;GAIhE,IAAI,eAAe,WAAW,OAAO;IACnC,OAAO,SAAS,EAAE,MAAM,YAAY;IACpC;GACF;GACA,eAAe,SAAS;GACxB,OAAO,SAAS,kBACd,OACA,eACA,oBACA,QACF;EACF;EAEA,MAAM,WAA8B;GAClC,MAAM;GACN,WAAW,QAAQ;GACnB;GACA;GACA,WAAW,YAAY,IAAI,IAAI;EACjC;EACA,YAAY,UAAU,EAAE,SAAS,CAAC;CACpC;AACF"}
+{"version":3,"file":"initTableWorker.js","names":[],"sources":["../../src/worker/initTableWorker.ts"],"sourcesContent":["import { constructTable } from '../core/table/constructTable'\nimport { storeReactivityBindings } from '../store-reactivity-bindings'\nimport { makeObjectMap } from '../utils'\nimport { serializeRowModel } from './serializeRowModel'\nimport type { RowData } from '../types/type-utils'\nimport type { TableFeatures } from '../types/TableFeatures'\nimport type { TableOptions } from '../types/TableOptions'\nimport type { Table_Internal } from '../types/Table'\nimport type {\n  TableWorkerRequest,\n  TableWorkerResult,\n  TableWorkerStage,\n  TableWorkerStagePayload,\n} from './tableWorkerProtocol'\n\nexport type TableWorkerConfig<\n  TFeatures extends TableFeatures,\n  TData extends RowData,\n> = Omit<TableOptions<TFeatures, TData>, 'data'>\n\nfunction capitalize(stage: string) {\n  return stage.charAt(0).toUpperCase() + stage.slice(1)\n}\n\n/** Flatten column-group defs to leaf defs (mirrors core's id resolution). */\nfunction flattenColumnDefs(defs: Array<any>): Array<any> {\n  return defs.flatMap((def) =>\n    def.columns ? flattenColumnDefs(def.columns) : [def],\n  )\n}\n\n/**\n * Runs a headless \"shadow table\" inside a dedicated Web Worker.\n *\n * Call this from a user-authored worker entry file, passing the same columns\n * and processing features used on the main thread. The shadow table runs the\n * real table-core row model pipeline (real fns, real Row objects) off the\n * main thread and posts back one payload per stage the main thread requested:\n * a transferable index permutation for flat results, a serialized row tree\n * (with eagerly computed aggregates) when grouping produces synthetic rows.\n *\n * Everything passed here must be thread-portable: `accessorKey` columns or\n * accessors defined in a shared module, and fns from registries or shared\n * modules (no closures over app state).\n *\n * @example\n * ```ts\n * // table.worker.ts\n * import { initTableWorker } from '@tanstack/table-core/experimental-worker-plugin'\n * import { columns, sharedFeatures } from './tableConfig'\n *\n * initTableWorker({ features: sharedFeatures, columns })\n * ```\n */\nexport function initTableWorker<\n  TFeatures extends TableFeatures,\n  TData extends RowData,\n>(config: TableWorkerConfig<TFeatures, TData>): void {\n  let table: Table_Internal<TFeatures, TData> | undefined\n  let dataVersion = 0\n  let coreIndexById: Record<string, number> = makeObjectMap()\n  let aggregateColumnIds: Array<string> = []\n  // Last-sent model identity per stage: the memoized getters return stable\n  // objects when their inputs did not change, so identity equality is exactly\n  // \"this stage's result is unchanged\".\n  let lastSentModels: { [K in TableWorkerStage]?: unknown } = {}\n\n  self.onmessage = (event: MessageEvent<TableWorkerRequest<TData>>) => {\n    const message = event.data\n\n    if (message.type === 'data') {\n      dataVersion = message.dataVersion\n      lastSentModels = {}\n      if (!table) {\n        table = constructTable<TFeatures, TData>({\n          ...(config as TableOptions<TFeatures, TData>),\n          features: {\n            coreReactivityFeature: storeReactivityBindings(),\n            ...config.features,\n          },\n          data: message.data,\n        })\n        // Only columns with an explicit aggregation get eagerly aggregated\n        // per group. Sync tables aggregate lazily (visible cells only), so\n        // auto-aggregating every column would explode on high-cardinality\n        // grouping for values nothing renders. Read the RAW column defs:\n        // rowAggregationFeature injects default aggregatedCell/aggregationFn\n        // into every resolved columnDef, so the resolved defs can't tell\n        // explicit from default.\n        aggregateColumnIds = flattenColumnDefs(config.columns as Array<any>)\n          .filter(\n            (def) => def.aggregationFn != null || def.aggregatedCell != null,\n          )\n          .map(\n            (def) =>\n              def.id ??\n              (typeof def.accessorKey === 'string'\n                ? def.accessorKey.replaceAll('.', '_')\n                : undefined),\n          )\n          .filter((id): id is string => id != null)\n      } else {\n        table.setOptions((prev) => ({ ...prev, data: message.data }))\n      }\n      // Map row ids to data positions once per dataset; serialization uses it\n      // to express every stage result in terms of core row positions.\n      const coreFlatRows = table.getCoreRowModel().flatRows\n      coreIndexById = makeObjectMap()\n      for (let i = 0; i < coreFlatRows.length; i++) {\n        coreIndexById[coreFlatRows[i]!.id] = i\n      }\n      return\n    }\n\n    if (!table) return\n\n    const start = performance.now()\n\n    // Apply the serializable state slices to the shadow table's base atoms.\n    table._reactivity.batch(() => {\n      for (const [key, value] of Object.entries(message.state)) {\n        const baseAtom = (table!.baseAtoms as Record<string, any>)[key]\n        if (baseAtom && value !== undefined) {\n          baseAtom.set(value)\n        }\n      }\n    })\n\n    // Compute exactly the stages the main thread requested, skipping any this\n    // shadow table has no row model factory for (the main thread warns).\n    const stages: { [K in TableWorkerStage]?: TableWorkerStagePayload } = {}\n    const transfer: Array<Transferable> = []\n\n    for (const stage of message.stages) {\n      if (!(config.features as Record<string, unknown>)[`${stage}RowModel`]) {\n        continue\n      }\n      const model = (table as any)[`get${capitalize(stage)}RowModel`]()\n      // Memoized getters return the same object when inputs are unchanged;\n      // skip re-serializing (and the main thread skips rebuilding). Safe under\n      // single-flight: results are never dropped within a dataVersion.\n      if (lastSentModels[stage] === model) {\n        stages[stage] = { kind: 'unchanged' }\n        continue\n      }\n      lastSentModels[stage] = model\n      stages[stage] = serializeRowModel(\n        model,\n        coreIndexById,\n        aggregateColumnIds,\n        transfer,\n      )\n    }\n\n    const response: TableWorkerResult = {\n      type: 'result',\n      requestId: message.requestId,\n      dataVersion,\n      stages,\n      computeMs: performance.now() - start,\n    }\n    postMessage(response, { transfer })\n  }\n}\n"],"mappings":";;;;;;AAoBA,SAAS,WAAW,OAAe;CACjC,OAAO,MAAM,OAAO,CAAC,CAAC,CAAC,YAAY,IAAI,MAAM,MAAM,CAAC;AACtD;;AAGA,SAAS,kBAAkB,MAA8B;CACvD,OAAO,KAAK,SAAS,QACnB,IAAI,UAAU,kBAAkB,IAAI,OAAO,IAAI,CAAC,GAAG,CACrD;AACF;;;;;;;;;;;;;;;;;;;;;;;;AAyBA,SAAgB,gBAGd,QAAmD;CACnD,IAAI;CACJ,IAAI,cAAc;CAClB,IAAI,gBAAwC,cAAc;CAC1D,IAAI,qBAAoC,CAAC;CAIzC,IAAI,iBAAwD,CAAC;CAE7D,KAAK,aAAa,UAAmD;EACnE,MAAM,UAAU,MAAM;EAEtB,IAAI,QAAQ,SAAS,QAAQ;GAC3B,cAAc,QAAQ;GACtB,iBAAiB,CAAC;GAClB,IAAI,CAAC,OAAO;IACV,QAAQ,eAAiC;KACvC,GAAI;KACJ,UAAU;MACR,uBAAuB,wBAAwB;MAC/C,GAAG,OAAO;KACZ;KACA,MAAM,QAAQ;IAChB,CAAC;IAQD,qBAAqB,kBAAkB,OAAO,OAAqB,CAAC,CACjE,QACE,QAAQ,IAAI,iBAAiB,QAAQ,IAAI,kBAAkB,IAC9D,CAAC,CACA,KACE,QACC,IAAI,OACH,OAAO,IAAI,gBAAgB,WACxB,IAAI,YAAY,WAAW,KAAK,GAAG,IACnC,OACR,CAAC,CACA,QAAQ,OAAqB,MAAM,IAAI;GAC5C,OACE,MAAM,YAAY,UAAU;IAAE,GAAG;IAAM,MAAM,QAAQ;GAAK,EAAE;GAI9D,MAAM,eAAe,MAAM,gBAAgB,CAAC,CAAC;GAC7C,gBAAgB,cAAc;GAC9B,KAAK,IAAI,IAAI,GAAG,IAAI,aAAa,QAAQ,KACvC,cAAc,aAAa,EAAE,CAAE,MAAM;GAEvC;EACF;EAEA,IAAI,CAAC,OAAO;EAEZ,MAAM,QAAQ,YAAY,IAAI;EAG9B,MAAM,YAAY,YAAY;GAC5B,KAAK,MAAM,CAAC,KAAK,UAAU,OAAO,QAAQ,QAAQ,KAAK,GAAG;IACxD,MAAM,WAAY,MAAO,UAAkC;IAC3D,IAAI,YAAY,UAAU,QACxB,SAAS,IAAI,KAAK;GAEtB;EACF,CAAC;EAID,MAAM,SAAgE,CAAC;EACvE,MAAM,WAAgC,CAAC;EAEvC,KAAK,MAAM,SAAS,QAAQ,QAAQ;GAClC,IAAI,CAAE,OAAO,SAAqC,GAAG,MAAM,YACzD;GAEF,MAAM,QAAS,MAAc,MAAM,WAAW,KAAK,EAAE,UAAU,CAAC;GAIhE,IAAI,eAAe,WAAW,OAAO;IACnC,OAAO,SAAS,EAAE,MAAM,YAAY;IACpC;GACF;GACA,eAAe,SAAS;GACxB,OAAO,SAAS,kBACd,OACA,eACA,oBACA,QACF;EACF;EAEA,MAAM,WAA8B;GAClC,MAAM;GACN,WAAW,QAAQ;GACnB;GACA;GACA,WAAW,YAAY,IAAI,IAAI;EACjC;EACA,YAAY,UAAU,EAAE,SAAS,CAAC;CACpC;AACF"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/table-core",
-  "version": "9.0.0-beta.50",
+  "version": "9.0.0-beta.52",
   "description": "Headless UI for building powerful tables & datagrids for TS/JS.",
   "author": "Tanner Linsley",
   "license": "MIT",

package/skills/aggregation/SKILL.md

@@ -6,13 +6,13 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features']
 sources:
   - 'TanStack/table:docs/guide/aggregation.md'
   - 'TanStack/table:docs/framework/react/guide/aggregation.md'
-  - 'TanStack/table:packages/table-core/src/features/aggregation'
+  - 'TanStack/table:packages/table-core/src/features/row-aggregation'
   - 'TanStack/table:examples/react/aggregation'
   - 'TanStack/table:examples/react/grouped-aggregation'
 ---
@@ -27,14 +27,14 @@ for synthetic grouped rows.
 
 ```ts
 import {
-  aggregationFeature,
+  rowAggregationFeature,
   aggregationFn_mean,
   aggregationFn_sum,
   tableFeatures,
 } from '@tanstack/table-core'
 
 export const features = tableFeatures({
-  aggregationFeature,
+  rowAggregationFeature,
   aggregationFns: {
     mean: aggregationFn_mean,
     sum: aggregationFn_sum,
@@ -113,7 +113,7 @@ disables that local fallback.
 
 Wrong: registering `columnGroupingFeature` solely to total a column.
 
-Correct: register `aggregationFeature` and call
+Correct: register `rowAggregationFeature` and call
 `column.getAggregationValue()`. Grouping is only required for grouped rows.
 
 ### [HIGH] Passing a scope label and rows
@@ -136,6 +136,6 @@ grouped results sent by the worker must be structured-cloneable.
 
 ## API Discovery
 
-Inspect `node_modules/@tanstack/table-core/src/features/aggregation/` and the
+Inspect `node_modules/@tanstack/table-core/src/features/row-aggregation/` and the
 Aggregation Guide. Use `Column_Aggregation`, `AggregationFnDef`,
 `AggregationContext`, and `AggregationResult` for the typed public surface.

package/skills/api-not-found/SKILL.md

@@ -5,7 +5,7 @@ description: >
 metadata:
   type: sub-skill
   library: '@tanstack/table-core'
-  library_version: '9.0.0-beta.50'
+  library_version: '9.0.0-beta.52'
 requires: ['core', 'table-features']
 sources:
   - 'TanStack/table:packages/table-core/src/index.ts'

package/skills/client-vs-server/SKILL.md

@@ -5,7 +5,7 @@ description: >
 metadata:
   type: sub-skill
   library: '@tanstack/table-core'
-  library_version: '9.0.0-beta.50'
+  library_version: '9.0.0-beta.52'
 requires: ['core', 'table-features']
 sources:
   - 'TanStack/table:docs/guide/row-models.md'

package/skills/column-faceting/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features', 'column-filtering']
 sources:

package/skills/column-filtering/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features', 'client-vs-server']
 sources:

package/skills/column-ordering/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features']
 sources:

package/skills/column-pinning/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features', 'column-sizing']
 sources:

package/skills/column-resizing/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features', 'column-sizing']
 sources:

package/skills/column-sizing/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features']
 sources:

package/skills/column-visibility/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features']
 sources:

package/skills/core/SKILL.md

@@ -5,7 +5,7 @@ description: >
 metadata:
   type: core
   library: '@tanstack/table-core'
-  library_version: '9.0.0-beta.50'
+  library_version: '9.0.0-beta.52'
 sources:
   - 'TanStack/table:docs/overview.md'
   - 'TanStack/table:docs/guide/tables.md'

package/skills/custom-features/SKILL.md

@@ -5,7 +5,7 @@ description: >
 metadata:
   type: sub-skill
   library: '@tanstack/table-core'
-  library_version: '9.0.0-beta.50'
+  library_version: '9.0.0-beta.52'
 requires: ['core', 'table-features', 'typescript']
 sources:
   - 'TanStack/table:docs/framework/react/guide/custom-features.md'

package/skills/expanding/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features', 'client-vs-server']
 sources:

package/skills/global-filtering/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features', 'client-vs-server', 'column-filtering']
 sources:

package/skills/grouping/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features', 'client-vs-server']
 sources:

package/skills/migrate-v8-to-v9/SKILL.md

@@ -5,7 +5,7 @@ description: >
 metadata:
   type: lifecycle
   library: '@tanstack/table-core'
-  library_version: '9.0.0-beta.50'
+  library_version: '9.0.0-beta.52'
 requires: ['core', 'table-features', 'typescript']
 sources:
   - 'TanStack/table:docs/framework/react/guide/migrating.md'
@@ -68,18 +68,18 @@ V8 bundled all stock features. V9 exposes an API only when its feature is presen
 
 | Capability                  | V9 feature                |
 | --------------------------- | ------------------------- |
+| Aggregation                 | `rowAggregationFeature`   |
 | Column faceting             | `columnFacetingFeature`   |
 | Column filtering            | `columnFilteringFeature`  |
-| Aggregation                 | `aggregationFeature`      |
-| Grouping                    | `columnGroupingFeature`   |
 | Column ordering             | `columnOrderingFeature`   |
 | Column pinning              | `columnPinningFeature`    |
-| Interactive column resizing | `columnResizingFeature`   |
 | Column sizes and offsets    | `columnSizingFeature`     |
 | Column visibility           | `columnVisibilityFeature` |
 | Global filtering            | `globalFilteringFeature`  |
-| Row expansion               | `rowExpandingFeature`     |
+| Grouping                    | `columnGroupingFeature`   |
+| Interactive column resizing | `columnResizingFeature`   |
 | Pagination                  | `rowPaginationFeature`    |
+| Row expansion               | `rowExpandingFeature`     |
 | Row pinning                 | `rowPinningFeature`       |
 | Row selection               | `rowSelectionFeature`     |
 | Sorting                     | `rowSortingFeature`       |
@@ -91,7 +91,7 @@ Honor feature prerequisites in the same `tableFeatures` call:
 - `columnResizingFeature` requires `columnSizingFeature`.
 - `globalFilteringFeature` requires `columnFilteringFeature`.
 - Every row-model or function-registry slot requires its associated feature.
-- `aggregationFns` requires `aggregationFeature`; grouped aggregation uses both `aggregationFeature` and `columnGroupingFeature`.
+- `aggregationFns` requires `rowAggregationFeature`; grouped aggregation uses both `rowAggregationFeature` and `columnGroupingFeature`.
 - Put prerequisite feature properties before dependent slots so inference and diagnostics remain clear.
 
 ### 2. Move row models into feature slots and rename factories
@@ -122,7 +122,7 @@ Move registries from table options or factory arguments into these feature slots
 
 Register only the built-ins the table references by string name, importing each individually (`filterFn_includesString`, `sortFn_alphanumeric`, `aggregationFn_sum`, and so on) alongside any custom functions. The full registry objects (`filterFns`, `sortFns`, `aggregationFns` exports) still work but bundle every built-in. A slot's keys become the valid string names in column definitions, and `'auto'` resolves only registered functions.
 
-Aggregation is independent from grouping. Add `aggregationFeature` for
+Aggregation is independent from grouping. Add `rowAggregationFeature` for
 `aggregationFn`, `aggregatedCell`, `column.getAggregationValue(options?)`, and
 `cell.getIsAggregated`. A root total does not require grouping. Convert legacy
 custom callables `(columnId, leafRows, childRows) => result` to
@@ -300,7 +300,7 @@ Do not confuse new capabilities with required breakages. After the table works,
 - [ ] Move all remaining `get*RowModel()` options or earlier-beta `rowModels` entries to `create*RowModel()` feature slots.
 - [ ] Register each dependent feature before its row-model slot.
 - [ ] Move `filterFns`, `sortingFns`/`sortFns`, and `aggregationFns` into feature slots, registering individually imported built-ins; pass no registries to factories.
-- [ ] Register `aggregationFeature` independently and migrate custom aggregation callables to context-based `AggregationFnDef` definitions.
+- [ ] Register `rowAggregationFeature` independently and migrate custom aggregation callables to context-based `AggregationFnDef` definitions.
 - [ ] Register `columnFilteringFeature` before global filtering and filter/facet dependencies.
 - [ ] Register `columnSizingFeature` before `columnResizingFeature`.
 - [ ] Replace `table.getState()` and top-level `onStateChange` according to the adapter state guide.

package/skills/pagination/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features', 'client-vs-server']
 sources:

package/skills/row-pinning/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features']
 sources:

package/skills/row-selection/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features']
 sources:

package/skills/sorting/SKILL.md

@@ -6,7 +6,7 @@ metadata:
   {
     type: sub-skill,
     library: '@tanstack/table-core',
-    library_version: '9.0.0-beta.50',
+    library_version: '9.0.0-beta.52',
   }
 requires: ['core', 'table-features', 'client-vs-server']
 sources:

package/skills/table-features/SKILL.md

@@ -5,7 +5,7 @@ description: >
 metadata:
   type: sub-skill
   library: '@tanstack/table-core'
-  library_version: '9.0.0-beta.50'
+  library_version: '9.0.0-beta.52'
 requires: ['core']
 sources:
   - 'TanStack/table:docs/guide/row-models.md'
@@ -22,7 +22,7 @@ This skill builds on `core`. Read it first for the headless model and stable inp
 
 ```ts
 import {
-  aggregationFeature,
+  rowAggregationFeature,
   aggregationFn_sum,
   columnGroupingFeature,
   createFilteredRowModel,
@@ -38,7 +38,7 @@ export const features = tableFeatures({
   columnFilteringFeature,
   filteredRowModel: createFilteredRowModel(),
   filterFns: { includesString: filterFn_includesString },
-  aggregationFeature,
+  rowAggregationFeature,
   columnGroupingFeature,
   aggregationFns: { sum: aggregationFn_sum },
   rowSortingFeature,
@@ -68,7 +68,7 @@ const features = tableFeatures({
   filterFns: { includesString: filterFn_includesString },
   rowSortingFeature,
   sortFns: { alphanumeric: sortFn_alphanumeric },
-  aggregationFeature,
+  rowAggregationFeature,
   columnGroupingFeature,
   aggregationFns: { sum: aggregationFn_sum },
 })
@@ -76,7 +76,7 @@ const features = tableFeatures({
 
 `filterFns`, `sortFns`, and `aggregationFns` are feature slots, not table
 options. They respectively require `columnFilteringFeature`,
-`rowSortingFeature`, and `aggregationFeature`. Import individual built-ins
+`rowSortingFeature`, and `rowAggregationFeature`. Import individual built-ins
 (`filterFn_*`, `sortFn_*`, `aggregationFn_*`) and register them under their
 conventional keys; the full registry objects (`filterFns`, `sortFns`,
 `aggregationFns` exports) still work but bundle every built-in. A registered

package/skills/typescript/SKILL.md

@@ -5,7 +5,7 @@ description: >
 metadata:
   type: sub-skill
   library: '@tanstack/table-core'
-  library_version: '9.0.0-beta.50'
+  library_version: '9.0.0-beta.52'
 requires: ['core', 'table-features']
 sources:
   - 'TanStack/table:docs/guide/helpers.md'

package/src/features/column-grouping/createGroupedRowModel.ts

@@ -5,10 +5,10 @@ import { table_autoResetExpanded } from '../row-expanding/rowExpandingFeature.ut
 import { table_autoResetPageIndex } from '../row-pagination/rowPaginationFeature.utils'
 import {
   aggregateColumnValue,
-  normalizeAggregationRows,
-} from '../aggregation/aggregationFeature.utils'
-import { row_getGroupingValue } from './columnGroupingFeature.utils'
+  normalizeUniqueAggregationRows,
+} from '../row-aggregation/rowAggregationFeature.utils'
 import type { Row_ColumnGrouping } from './columnGroupingFeature.types'
+import type { Column_Internal } from '../../types/Column'
 import type { TableFeatures } from '../../types/TableFeatures'
 import type { RowModel } from '../../core/row-models/coreRowModelsFeature.types'
 import type { Table, Table_Internal } from '../../types/Table'
@@ -20,7 +20,7 @@ import type { RowData } from '../../types/type-utils'
  *
  * The factory reads the relevant table state atoms and options, then returns a row model function used by the table row-model pipeline.
  *
- * When aggregationFeature is also registered, grouped rows use its shared
+ * When rowAggregationFeature is also registered, grouped rows use its shared
  * executor for non-group values. Grouping remains useful without aggregation.
  */
 export function createGroupedRowModel<
@@ -103,7 +103,7 @@ function _createGroupedRowModel<
     const columnId = existingGrouping[depth] as string
 
     // Group the rows together for this level
-    const rowGroupsMap = groupBy(rows, columnId)
+    const rowGroupsMap = groupBy(table, rows, columnId)
 
     // Perform aggregations for each group
     const aggregatedGroupedRows = Array.from(rowGroupsMap.entries()).map(
@@ -118,7 +118,13 @@ function _createGroupedRowModel<
           subRow.parentId = id
         })
 
-        const leafRows = normalizeAggregationRows(groupedRows, Infinity)
+        // Rows produced by groupBy are disjoint members of the pre-grouped
+        // row tree, so the duplicate-id guard is unnecessary; flat groups
+        // reuse the partition array as-is.
+        const leafRows = normalizeUniqueAggregationRows(
+          groupedRows,
+          Infinity,
+        ) as Array<Row<TFeatures, TData>>
 
         const row = constructRow(
           table,
@@ -171,6 +177,7 @@ function _createGroupedRowModel<
               column,
               groupingRow: row,
               rows: groupedRows,
+              uniqueRows: true,
             })
             return cache[colId]
           },
@@ -203,19 +210,49 @@ function _createGroupedRowModel<
 }
 
 function groupBy<TFeatures extends TableFeatures, TData extends RowData = any>(
+  table: Table_Internal<TFeatures, TData>,
   rows: Array<Row<TFeatures, TData>>,
   columnId: string,
 ) {
   const groupMap = new Map<any, Array<Row<TFeatures, TData>>>()
 
-  return rows.reduce((map, row) => {
-    const resKey = `${row_getGroupingValue(row, columnId)}`
-    const previous = map.get(resKey)
+  // Resolve the column once instead of per row: `table.getColumn` goes
+  // through the memoized-API dispatcher, which is far too expensive to sit
+  // inside this per-row loop. The branches below mirror
+  // `row_getGroupingValue`'s caching contract exactly.
+  const column = table_getColumn(table, columnId) as
+    | Column_Internal<TFeatures, TData, unknown>
+    | undefined
+  const getGroupingValue = column?.columnDef.getGroupingValue
+
+  for (let i = 0; i < rows.length; i++) {
+    const row = rows[i]!
+    let groupingValue
+    if (getGroupingValue) {
+      const cache = (row as any)._groupingValuesCache as
+        | Record<string, unknown>
+        | undefined
+      if (cache && hasOwn(cache, columnId)) {
+        groupingValue = cache[columnId]
+      } else if (cache) {
+        groupingValue = cache[columnId] = getGroupingValue(
+          row.original,
+          row.index,
+          row,
+        )
+      }
+    } else {
+      groupingValue = row.getValue(columnId)
+    }
+
+    const resKey = `${groupingValue}`
+    const previous = groupMap.get(resKey)
     if (!previous) {
-      map.set(resKey, [row])
+      groupMap.set(resKey, [row])
     } else {
       previous.push(row)
     }
-    return map
-  }, groupMap)
+  }
+
+  return groupMap
 }