Source-row editing

Grid cells are aggregates, so editing one is ambiguous — instead you edit the source rows. Click a leaf row to open its master-detail table, click a source cell to edit it, and the engine recomputes every affected aggregate. Edits are addressed by group-key path + row index and pass through server-side validation before the frame mutates (try a revenue over 500 below — it’s rejected).

Dash (Python)

# The component emits a cell-edit event; a callback applies it via the engine and returns
# refreshed row_data. Validation rules live on the column def.
result = engine.apply_edit(
    ["region", "country"],
    {"path": ["EMEA", "Germany"], "row_index": 0,
     "column": "revenue", "old_value": 100, "new_value": 120},
    validation_rules={"revenue": [{"type": "range", "min": 0, "max": 500}]},
)

React

<TensorGrid
  data={data}
  rows={['region', 'country']}
  measures={{ revenue: 'sum', qty: 'sum' }}
  masterDetail
  editable
  validationRules={{ revenue: [{ type: 'range', min: 0, max: 500 }] }}
  onCellEdit={(e) => console.log('edited', e)}
  onEditRejected={(d) => console.warn('rejected', d.message)}
/>;

Vue

<TensorGrid
  :data="data"
  :rows="['region', 'country']"
  :measures="{ revenue: 'sum', qty: 'sum' }"
  master-detail
  editable
  :validation-rules="{ revenue: [{ type: 'range', min: 0, max: 500 }] }"
  @cell-edit="onEdit"
  @edit-rejected="onReject"
/>

Vanilla JS

mountTensorGrid(el, {
  data,
  rows: ['region', 'country'],
  measures: { revenue: 'sum', qty: 'sum' },
  masterDetail: true,
  editable: true,
  validationRules: { revenue: [{ type: 'range', min: 0, max: 500 }] },
  onCellEdit: (e) => console.log('edited', e),
  onEditRejected: (d) => console.warn('rejected', d.message),
});

Validation rules

Each column can carry validationRules — range (min/max), required, regex, or a safe-DSL predicate ({ type: 'predicate', if: 'value > 0' }) tested against the candidate row. A rejected edit leaves the frame untouched, restores the cell, and reports { code, column, rule, message }. Editing a typed column coerces the new value to the column dtype (numeric widening, temporal parsing) — never silently corrupting the column.