@object-ui/plugin-report 17.5.0 → 17.6.0

package/README.md

@@ -1,16 +1,16 @@
 # @object-ui/plugin-report
 
-Report components for Object UI — build, view, render, and export reports with scheduling support.
+Report components for Object UI — render, view and export reports, with scheduling support. Report *authoring* is not part of this package.
 
 ## Features
 
 - 🧩 **Four spec report variants** — `tabular` / `summary` / `matrix` / `joined`, dispatched by a single `<ReportRenderer schema={...}>`
-- 🧮 **Server-side aggregation** — `useReportData()` posts spec `QueryAST` to `POST /api/v1/data/:object/query`; transparent in-memory fallback
-- 📅 **Date bucketing** — `dateGranularity: day|week|month|quarter|year` on `groupingsAcross` / `groupingsDown`
-- 🪜 **Multi-level grouping + totals** — row totals, column totals, grand totals for matrix; tabular/summary delegate to `ObjectGrid`
-- 🎯 **Cell drill-down** — every aggregated cell dispatches a `drill` action via `ActionRunner`; targets List view or a nested Report (M3)
-- 🧱 **Joined reports** — vertically stacked sub-reports; each block owns its own `objectName`, filter and data fetch
-- 🎨 **Type-aware cells** — `select` → Badge, `lookup` → link, `boolean` → ✓/✗, `email`/`url`/`phone` → links, `image` → thumbnail (auto-hydrated from object metadata)
+- 🧮 **Server-side aggregation** — a dataset-bound report selects measures by name through `dataSource.queryDataset`; totals come back pre-aggregated (ADR-0021)
+- 📅 **Period pivots** — the across axis is a dimension the dataset declares at the grain it needs; a report buckets nothing itself
+- 🪜 **Multi-level grouping + totals** — row subtotals, column subtotals and grand total for `matrix`, a grand-total footer for `summary`, all server-computed
+- 🎯 **Cell drill-down** — aggregated rows and matrix cells emit `DatasetDrillArgs` to the host's `onDrill` callback; the host owns navigation (ADR-0021 D2)
+- 🧱 **Joined reports** — vertically stacked sub-reports; each block names its own `dataset`, filter and ordering
+- 🎨 **Type-aware cells** — the presentation-layer `ReportViewer` renders `select` → Badge, `lookup` → link, `boolean` → ✓/✗, `email`/`url`/`phone` → links, `image` → thumbnail
 - 🖨️ **Multi-format export** — CSV, JSON, HTML, PDF, Excel; live-data and Excel-formula variants
 - 📦 **Auto-registered** — components register with `ComponentRegistry` on import; embed via `{ "type": "spec-report", "report": {...} }`
 
@@ -26,23 +26,55 @@ npm install @object-ui/plugin-report
 
 ## Quick Start
 
+`ReportRenderer` is the one entry point — a dispatcher that routes any report
+shape to the right renderer. It takes the report as `schema`; there is no
+report-authoring component in this package (see
+[Legacy presentation layer](#legacy-presentation-layer)).
+
 ```tsx
-import { ReportBuilder, ReportViewer, ReportRenderer } from '@object-ui/plugin-report';
+import { ReportRenderer } from '@object-ui/plugin-report';
 
-function ReportEditorPage() {
-  return <ReportBuilder report={initialReport} />;
+// `dataSource` and `router` are host-provided.
+function ReportPage() {
+  return (
+    <ReportRenderer
+      schema={{
+        name: 'opp_by_stage',
+        type: 'summary',
+        dataset: 'opportunity_pipeline',
+        rows: ['stage'],
+        values: ['amount_sum', 'deal_count'],
+      }}
+      dataSource={dataSource}
+      onDrill={({ object }) => router.push(`/records/${object}`)}
+    />
+  );
 }
+```
+
+The dataset-bound renderer can also be addressed directly when the host has
+already decided the shape — it takes the report as `report`, not `schema`:
+
+```tsx
+import { DatasetReportRenderer, isDatasetReport } from '@object-ui/plugin-report';
 
-function ReportViewPage() {
-  return <ReportViewer report={reportDefinition} showToolbar />;
+if (isDatasetReport(stored)) {
+  return <DatasetReportRenderer report={stored} dataSource={dataSource} />;
 }
+```
+
+`ReportViewer` renders a presentation-layer report, and takes a whole
+`ReportViewerSchema` as `schema` — `report` / `showToolbar` are keys *inside*
+that schema, not props of the component:
 
-function EmbeddedReport() {
+```tsx
+import { ReportViewer } from '@object-ui/plugin-report';
+
+function LegacyReportPage() {
   return (
-    <ReportRenderer
-      title="Monthly Sales"
-      description="Sales performance overview"
-      chart={chartConfig}
+    <ReportViewer
+      schema={{ type: 'report-viewer', report: reportDefinition, data: rows, showToolbar: true }}
+      onRefresh={refetch}
     />
   );
 }
@@ -50,87 +82,201 @@ function EmbeddedReport() {
 
 ## Spec Reports — the four variants
 
-The plugin renders any `Report` defined by `@objectstack/spec`:
+A report is **dataset-bound**: it names a semantic-layer `dataset` and selects
+that dataset's measures (`values`) grouped by its dimensions (`rows`, and for a
+matrix also `columns`). Every name in a report is a name the dataset defines —
+the report declares no object, no field, no aggregate and no date bucket of its
+own. `defineReport` validates at authoring time and returns the parsed shape
+`ReportRenderer` takes as `schema`:
 
 ```ts
-import type { ReportInput } from '@objectstack/spec/ui';
+import { defineReport } from '@objectstack/spec/ui';
 import { ReportRenderer } from '@object-ui/plugin-report';
 
-const report: ReportInput = {
+const report = defineReport({
   name: 'opp_by_stage',
-  objectName: 'opportunity',
+  label: 'Opportunities by Stage',
   type: 'summary',
-  columns: [
-    { field: 'stage' },
-    { field: 'amount', aggregate: 'sum' },
-    { field: 'id', label: 'Deals', aggregate: 'count' },
-  ],
-  groupingsDown: [{ field: 'stage', sortOrder: 'asc' }],
-};
+  dataset: 'opportunity_pipeline',
+  rows: ['stage'],
+  values: ['amount_sum', 'deal_count'],
+  order: [{ by: 'amount_sum', direction: 'desc' }],
+});
 
 <ReportRenderer schema={report} dataSource={ds} />
 ```
 
-| `type`    | Description                                          |
-| --------- | ---------------------------------------------------- |
-| `tabular` | Flat record list                                     |
-| `summary` | Single-axis grouped + aggregated                     |
-| `matrix`  | Row × column pivot with cell aggregates and totals   |
-| `joined`  | Vertically stacked sub-reports, each with own data   |
+| Key             | Type                                           | Meaning                                                          |
+| --------------- | ---------------------------------------------- | ---------------------------------------------------------------- |
+| `name`          | `string` (required)                            | Identifier, at least 2 characters.                                |
+| `label`         | `string` \| `Record< string, string >` (required) | Display title; the record form is the i18n shape.              |
+| `type`          | `tabular` \| `summary` \| `matrix` \| `joined` | Defaults to `tabular`.                                          |
+| `dataset`       | `string`                                       | The dataset this report reads.                                    |
+| `rows`          | `string[]`                                     | Dimension **names** to group down.                                |
+| `columns`       | `string[]`                                     | Dimension **names** across — the matrix pivot axis.               |
+| `values`        | `string[]`                                     | Measure **names** to display.                                     |
+| `runtimeFilter` | filter condition                               | Scope filter, merged with the filter the host passes at render.    |
+| `order`         | `{ by, direction }[]`                          | Result ordering, most significant key first.                      |
+| `drilldown`     | `boolean`                                      | Defaults to `true`.                                               |
+| `chart`         | object                                         | Embedded visualization over the same selection.                   |
+| `blocks`        | report[]                                       | `joined` only — the stacked sub-reports.                          |
+
+The schema is **strict**: a key it does not declare is rejected, not ignored.
+
+| `type`    | Description                                                                    |
+| --------- | ------------------------------------------------------------------------------ |
+| `tabular` | The selection as a flat list — `rows` + `values`, no totals                     |
+| `summary` | The same table plus a server-computed grand-total footer                        |
+| `matrix`  | Cross-tab: `rows` down × `columns` across, measures in the cells, plus subtotals |
+| `joined`  | A vertical stack of blocks, each its own dataset-bound table                    |
+
+The **declared type** picks the presentation — never the shape of the data that
+comes back.
 
 ### Matrix (row × column pivot)
 
+`columns` is a second list of dimension **names**, used as the across axis:
+
 ```ts
-{
+import { defineReport } from '@objectstack/spec/ui';
+
+defineReport({
   name: 'pipeline_by_quarter',
-  objectName: 'opportunity',
+  label: 'Pipeline Coverage by Quarter',
   type: 'matrix',
-  columns: [{ field: 'amount', label: 'Pipeline', aggregate: 'sum' }],
-  groupingsDown:   [{ field: 'forecast_category' }],
-  groupingsAcross: [{ field: 'close_date', dateGranularity: 'quarter' }],
-}
+  dataset: 'opportunity_pipeline',
+  rows: ['forecast_category'],
+  columns: ['close_quarter'],
+  values: ['amount_sum'],
+});
 ```
 
-`dateGranularity` accepts `day | week | month | quarter | year` and is
-pushed down to the server-side aggregator.
+`close_quarter` is a time dimension the **dataset** declares at that grain. A
+report cannot bucket a date itself: the grain a measure is valid at belongs to
+the semantic layer, not to one report. A `matrix` with no `columns` degrades to
+the flat grouped table.
 
-### Joined (M3)
+### Joined
 
 ```ts
-{
+import { defineReport } from '@objectstack/spec/ui';
+
+defineReport({
   name: 'churn_signals',
-  objectName: 'account',          // container default
+  label: 'Customer Churn Signals',
   type: 'joined',
-  columns: [],
   blocks: [
-    { name: 'at_risk', type: 'summary', columns: [...], filter: {...} },
-    { name: 'lost',    type: 'summary', objectName: 'opportunity', columns: [...], filter: {...} },
+    {
+      name: 'at_risk',
+      label: 'At-Risk Accounts',
+      type: 'summary',
+      dataset: 'account_health',
+      rows: ['industry'],
+      values: ['account_count'],
+      runtimeFilter: { is_active: true },
+    },
+    {
+      name: 'lost',
+      label: 'Recently Lost',
+      type: 'summary',
+      dataset: 'opportunity_pipeline',
+      rows: ['owner'],
+      values: ['amount_sum'],
+      runtimeFilter: { stage: 'closed_lost' },
+    },
   ],
-}
+});
 ```
 
-Block rules: `objectName` falls back to the container; `filter` is ANDed
-with the container's; each block runs an isolated `useReportData()` call;
-`block.type` must not be `joined` (no recursion).
+Block rules: every block names its **own** `dataset` (there is no container
+dataset to inherit); the container's `runtimeFilter` is merged into each block
+and the block's own keys win; each block orders itself through its own `order`;
+a block's declared `type` picks its presentation; `block.type` must not be
+`joined` (no recursion).
 
-## Server-side aggregation + drill-down
+### Ordering
+
+`order` is a list, most significant key first, lowered onto the dataset
+selection so the **server** orders the query — which is why ordering by a
+derived measure works and the sort applies to the whole result rather than to
+one fetched page. `by` names a dimension the report groups by or a measure it
+displays. For a matrix the across axis reads left-to-right in row-arrival
+order, so ordering by the across dimension is what fixes the column order;
+declaring nothing still reads correctly, because a selected time dimension
+defaults to ascending.
 
-`useReportData()` translates a `Report` into spec `QueryAST` and posts it
-to `POST /api/v1/data/:object/query`. If the endpoint is unavailable it
-falls back transparently to `dataSource.find()` + client-side aggregation.
+### Embedded chart
+
+`chart.type` is one of the spec's chart names; `xAxis` / `yAxis` are **bare
+dimension and measure names** from the report's own selection, not field paths
+and not expressions:
+
+```ts
+import { defineReport } from '@objectstack/spec/ui';
+
+defineReport({
+  name: 'pipeline_by_stage',
+  label: 'Pipeline by Stage',
+  type: 'summary',
+  dataset: 'opportunity_pipeline',
+  rows: ['stage'],
+  values: ['amount_sum'],
+  chart: {
+    type: 'bar',
+    title: 'Pipeline by Stage',
+    xAxis: 'stage',
+    yAxis: 'amount_sum',
+    showLegend: false,
+  },
+});
+```
+
+`series`, `colors`, `height`, `showDataLabels`, `annotations` and `interaction`
+are passed through to the chart as authored.
+
+## Server-side aggregation + drill-down
 
-Every aggregated cell dispatches a `drill` action through `ActionRunner`:
+A dataset-bound report selects its dimensions (`rows`, and for `matrix` also
+`columns`) and measures (`values`) **by name** and runs them through
+`dataSource.queryDataset` — the same governed semantic-layer path that
+dataset-bound dashboard widgets and the dataset preview use, so the numbers and
+the server-resolved dimension labels match everywhere. Totals (row subtotals,
+column subtotals, grand total) are **server-computed**: the renderer places the
+pre-aggregated rows it is handed and never re-aggregates bucketed values
+client-side, since measures like `avg` cannot be recombined without drifting
+from the semantic layer.
+
+Drill-down is a **host callback, not a registered handler**: pass `onDrill` and
+every aggregated row / matrix cell becomes clickable. The report emits *what was
+clicked*; the host decides where that goes, because the renderer only knows
+dimension names (ADR-0021 D2).
 
 ```tsx
-import { registerDrillHandler } from '@object-ui/plugin-report';
-registerDrillHandler(actionRunner, { navigate: router.push });
+import { ReportRenderer, type DatasetDrillArgs } from '@object-ui/plugin-report';
+
+<ReportRenderer
+  schema={report}
+  dataSource={dataSource}
+  onDrill={(args: DatasetDrillArgs) => {
+    // The host resolves dataset → object and dimension → field, then navigates.
+    const filter = { ...args.objectFilter, ...args.runtimeFilter };
+    router.push(`/records/${args.object}?filter=${encodeURIComponent(JSON.stringify(filter))}`);
+  }}
+/>
 ```
 
-Drill targets:
-1. **List view** — default; navigates to the filtered records.
-2. **Report drawer** — if the host widget declares `drillDown.report`,
-   the click opens a side drawer that renders that report scoped to the
-   cell's group key (composes dashboard → report → record).
+What a drill click carries:
+
+| `DatasetDrillArgs` | Meaning |
+| ------------------ | ------- |
+| `dataset`      | Dataset the clicked aggregate was computed over. |
+| `groupKey`     | Dimension **name** → clicked bucket value (row dims, plus across dims for a matrix cell). |
+| `runtimeFilter`| The effective render-time scope filter, if any. |
+| `object`       | The dataset's base object, when the server supplied it. |
+| `objectFilter` | Exact record-list filter (object **field** name → raw stored value) for the clicked bucket. Present only when the server returned the dimension→field mapping plus raw grouped values — that is what lets select/lookup dimensions filter precisely instead of by display label. |
+
+Supply no `onDrill` and nothing is clickable; a report can also opt out with
+`drilldown: false`.
 
 ## Filter-time date helpers — current limitation
 
@@ -144,21 +290,76 @@ const daysAgo = (n: number): string => {
   return d.toISOString().slice(0, 10);
 };
 
-filter: { close_date: { $gte: daysAgo(30) } }
+runtimeFilter: { close_date: { $gte: daysAgo(30) } }
 ```
 
 See the bundled CRM `customer_churn_signals` demo for the full pattern.
 Native filter-time CEL evaluation is tracked for a future major version.
 
-## Legacy components
-
-`ReportBuilder`, `ReportViewer` and the export functions below remain
-available for legacy presentation-layer reports and are not affected by
-the spec-driven pipeline above.
-
-
-
-Export reports in multiple formats:
+## Stored pre-9.0 documents — migration only
+
+> ⚠️ **This section is about documents that already exist in storage. The shape
+> it describes is NOT available for authoring** — everything above is how a
+> report is written, nothing here is.
+
+Before the ADR-0021 cutover a report carried its own query: an `objectName`,
+`columns` as column *definition objects* with `aggregate`, and
+`groupingsDown` / `groupingsAcross` with `sortOrder` and `dateGranularity`.
+That form is **rejected by the current schema** — `objectName` and
+`groupingsDown` come back as unrecognized keys, and object-shaped `columns`
+fails as the wrong type (`columns` is now a list of dimension names).
+
+Stored documents in that shape still render, through a deliberately lossy
+bridge: `ReportRenderer` converts them to a presentation report and hands that
+to `ReportViewer`. The converter reads the *current* keys (`values`, `rows`), so
+a pre-9.0 document arrives with none of them and what comes out is the report's
+title over an **empty column set**. The document renders; its columns,
+groupings, aggregates and sort do not, and nothing reports an error. That is why
+the bridge is a waiting room rather than a supported way to write a report.
+
+Migrating one is a re-expression against the dataset:
+
+| Pre-9.0 key                        | Where it goes now                                                       |
+| ---------------------------------- | ----------------------------------------------------------------------- |
+| `objectName`                       | Gone from the report — the `dataset` owns the object.                    |
+| `columns: [{ field, aggregate }]`  | `values: string[]`; the aggregate becomes a **measure** in the dataset.  |
+| `columns: [{ field }]`             | `rows: string[]`, as a dimension.                                        |
+| `groupingsDown: [{ field }]`       | `rows: string[]`.                                                        |
+| `groupingsAcross: [{ field }]`     | `columns: string[]`.                                                     |
+| `sortOrder` on a grouping          | `order: [{ by, direction }]` on the report.                              |
+| `dateGranularity` on a grouping    | A time dimension declared at that grain **in the dataset**.              |
+| `filter`                           | `runtimeFilter`.                                                         |
+
+The part with no mechanical equivalent is the dataset itself: measures and time
+dimensions the old report declared inline must exist in the semantic layer
+before the migrated report can name them.
+
+The query-form renderers that used to draw these documents inline
+(`SpecReportGrid`, `MatrixRenderer`, `JoinedReportRenderer`) were removed at the
+cutover and have no replacement export — the bridge above is what renders a
+stored old-shape document now.
+
+## Legacy presentation layer
+
+`ReportViewer`, `LegacyReportRenderer` and the export functions below remain
+available for pre-spec presentation reports (`{ report, data, columns, chart }`)
+and are not affected by the dataset-bound pipeline above.
+
+The authoring components that used to sit alongside them — `ReportBuilder`,
+`ScheduleConfig`, `ChartConfig`, the columns/groupings editors — and the
+`registerDrillHandler`-style drill helpers were **removed** in the 9.0 cutover
+(see `CHANGELOG.md`). They have no replacement export in this package: report
+authoring lives in the console designer, and drill-down is the `onDrill`
+callback above.
+
+### Export
+
+Every format exporter takes the **report schema first and the rows second** —
+`(report: ReportComponentSchema, data: any[], config?: ReportExportConfig)`. The
+schema is where the exporter reads the columns (`report.fields`) and the download
+name (`config.filename`, else `report.title`) from; there is **no filename
+parameter**. All five return `void` synchronously — they trigger a browser
+download — so there is nothing to `await`:
 
 ```tsx
 import {
@@ -169,51 +370,151 @@ import {
   exportAsPDF,
   exportAsExcel,
 } from '@object-ui/plugin-report';
+import type { ReportComponentSchema } from '@object-ui/types';
+
+const report: ReportComponentSchema = {
+  type: 'report',
+  title: 'Sales Report',
+  fields: [
+    { name: 'account', label: 'Account' },
+    { name: 'amount', label: 'Amount', type: 'currency', aggregation: 'sum' },
+  ],
+};
+const rows = [{ account: 'Acme', amount: 12000 }];
+
+exportAsCSV(report, rows); // downloads "Sales Report.csv"
+exportAsJSON(report, rows); // "Sales Report.json"
+exportAsHTML(report, rows); // "Sales Report.html"
+exportAsPDF(report, rows); // opens a print window; falls back to .html if blocked
+exportAsExcel(report, rows, { format: 'excel', filename: 'sales-2026-q1.tsv' });
 
-await exportAsCSV(reportData, 'sales-report.csv');
-await exportAsPDF(reportData, 'sales-report.pdf');
-await exportAsExcel(reportData, 'sales-report.xlsx');
+// `exportReport` is the router — the format comes FIRST, then the same triple.
+exportReport('csv', report, rows);
 ```
 
+`exportAsExcel` writes a BOM-prefixed **TSV** that Excel opens, so its default
+name is `report.title` + `.tsv`, not `.xlsx`. `config.includeHeaders: false` drops the
+header row; `orientation` / `pageSize` apply to the HTML and PDF paths only.
+
 ### Live Export
 
-Export with real-time data and Excel formulas:
+`exportWithLiveData` fetches the rows itself, so `dataSource` and `resource` are
+**required** in `LiveExportOptions` — an options object carrying only `format`
+does not compile. It is the one `async` export function, resolving to a
+`LiveExportResult`:
 
 ```tsx
-import { exportWithLiveData, exportExcelWithFormulas } from '@object-ui/plugin-report';
+import { exportWithLiveData } from '@object-ui/plugin-report';
 
-await exportWithLiveData(reportConfig, { format: 'pdf' });
-await exportExcelWithFormulas(reportConfig, {
-  columns: [{ field: 'total', formula: 'SUM(B2:B100)' }],
+// `myAdapter` is a host-provided DataSource (see @object-ui/data-*).
+const result = await exportWithLiveData(report, {
+  dataSource: myAdapter,
+  resource: 'orders',
+  format: 'pdf',
 });
+// -> { success: true, recordCount: 42, format: 'pdf' }
 ```
 
-### ScheduleConfig
+The format falls back to `report.defaultExportFormat`, then `'pdf'`. Failures are
+returned, not thrown: `{ success: false, recordCount: 0, format, error }`.
 
-Configure recurring report generation:
+`exportExcelWithFormulas` is a separate, synchronous three-parameter function —
+`(report, data, options)` — with the column list inside the third argument. A
+column is `{ name, header, width?, numberFormat?, formula? }`: `name` is the key
+read off each row (there is no `field` key), `header` is required, and a `formula`
+is a template in which the `{ROW}` placeholder is replaced by that row's
+spreadsheet line number:
 
 ```tsx
-import { ScheduleConfig, createScheduleTrigger } from '@object-ui/plugin-report';
+import { exportExcelWithFormulas } from '@object-ui/plugin-report';
 
-<ScheduleConfig
-  reportId="monthly-sales"
-  onSave={(schedule) => saveSchedule(schedule)}
-/>
+exportExcelWithFormulas(report, rows, {
+  columns: [
+    { name: 'amount', header: 'Amount', numberFormat: '#,##0.00' },
+    { name: 'total', header: 'Total', formula: '=B{ROW}*C{ROW}' },
+  ],
+  includeAggregationRow: true,
+});
+```
+
+Omitting `columns` derives them from `report.fields`. `includeAggregationRow`
+appends a totals row built from each field's `aggregation`.
+
+### Scheduled export
+
+A schedule is **data on the report schema**, not a component: it is
+`ReportComponentSchema.schedule`, typed `ReportScheduleConfig`. Both types live
+in `@object-ui/types` and are *not* re-exported here — importing them from this
+package is a TS2305.
+
+`createScheduleTrigger` turns that declared schedule into a callable a workflow
+engine can invoke. It takes the report, the data source, the resource to query
+and a completion callback:
 
-const trigger = createScheduleTrigger((reportId) => generateReport(reportId));
+```ts
+import { createScheduleTrigger } from '@object-ui/plugin-report';
+import type { ReportComponentSchema, ReportScheduleConfig } from '@object-ui/types';
+
+// `dataSource` and `notify` are host-provided.
+
+const report: ReportComponentSchema = {
+  type: 'report',
+  title: 'Monthly Sales',
+  schedule: {
+    enabled: true,
+    frequency: 'monthly',
+    dayOfMonth: 1,
+    time: '07:00',
+    formats: ['pdf', 'excel'],
+    recipients: ['sales@example.com'],
+  },
+};
+
+const trigger = createScheduleTrigger(
+  report,
+  dataSource,
+  'orders',
+  (exported: ReportComponentSchema, schedule: ReportScheduleConfig) => {
+    notify(schedule.recipients ?? [], schedule.subject ?? exported.title);
+  },
+);
+
+const results = await trigger(); // LiveExportResult[] — one entry per scheduled format
 ```
 
+`trigger()` reads `report.schedule` itself: a schedule that is missing or
+`enabled: false` exports nothing and resolves to `[]`. Otherwise it exports once
+per entry in `schedule.formats` (falling back to `report.defaultExportFormat`,
+then `'pdf'`) and calls the completion callback with the report and the schedule
+it ran.
+
 ### Schema-Driven Usage
 
-Components auto-register with `ComponentRegistry`:
+Importing the package registers three component types with `ComponentRegistry`:
+
+| `type`          | Component        | Notes                                              |
+| --------------- | ---------------- | -------------------------------------------------- |
+| `report`        | `ReportRenderer` | The dispatcher.                                    |
+| `spec-report`   | `ReportRenderer` | Spec-native alias; the report goes under `report`.  |
+| `report-viewer` | `ReportViewer`   | Presentation-layer viewer.                         |
 
 ```json
 {
-  "type": "report-builder",
-  "report": { "sections": [] }
+  "type": "spec-report",
+  "report": {
+    "name": "opp_by_stage",
+    "type": "summary",
+    "dataset": "opportunity_pipeline",
+    "rows": ["stage"],
+    "values": ["amount_sum"]
+  }
 }
 ```
 
+There is no `report-builder` component type — the authoring component that name
+addressed was removed in the 9.0 cutover, so a node declaring it resolves to
+nothing.
+
 ### Type-aware cell rendering
 
 `ReportViewer` delegates cell rendering to the shared `getCellRenderer`
@@ -237,40 +538,11 @@ component appropriate for its type — instead of `String(value)`.
 | `richtext` / `html` / `markdown` | Sanitised inline content                   |
 | `json`                        | Collapsed code preview                        |
 
-Authors do **not** need to repeat type metadata on every report column:
-when a report binds an `objectName`, the runtime auto-hydrates each
-column's `type`, `options`, `referenceTo`, and `label` from the
-corresponding `ObjectField`. Author-provided values always win.
-
-Minimal report leveraging type-aware cells:
-
-```ts
-import type { ReportInput } from '@objectstack/spec/ui';
-
-export const ContactsReport: ReportInput = {
-  name: 'contacts_by_account',
-  label: 'Contacts by Account',
-  objectName: 'contact', // ← enables auto-hydration
-  type: 'tabular',
-  columns: [
-    { field: 'full_name', label: 'Name' },
-    { field: 'email',      label: 'Email' },     // → mailto:
-    { field: 'phone',      label: 'Phone' },     // → tel:
-    { field: 'is_primary', label: 'Primary' },   // → ✓/✗
-    { field: 'account',    label: 'Account' },   // → linked record
-    { field: 'status',     label: 'Status' },    // → Badge with option color
-  ],
-};
-```
-
-Override per column when needed:
-
-```ts
-columns: [
-  { field: 'tier', label: 'Tier', type: 'select',
-    options: [{ value: 'gold', label: 'Gold', color: 'amber' }] },
-]
-```
+These are presentation-layer columns: each entry in `report.fields` carries its
+own `name`, `label` and `type`, and that declared `type` is what selects the
+renderer. This package resolves nothing from object metadata — there is no
+`objectName` on a report to hydrate from, and a column with no `type` renders
+as plain text.
 
 Legacy `renderAs: 'badge'` + `colorMap` is still honoured for plain
 string columns.