Editing table data
Use the table preview grid to add, edit, and delete rows directly — with live pagination on large tables and automatic protection for masked and governed columns.
The table preview page (Analytics → Tables → [table] → Preview) shows a live data grid for any active managed table. Preview is the tab a table opens on by default, so the data is the first thing you see. Beyond read-only browsing, it lets you make row-level changes directly from the UI — no SQL required.
What you can do
Editing capability depends on whether the table has a primary key.
| Table type | Cell edit | Delete row | Add record |
|---|---|---|---|
| Table with primary key | Yes | Yes | Yes |
| Table without primary key | No | No | Yes (append) |
StarRocks — the underlying OLAP engine — requires a primary key to target a specific row for an update or delete. Tables without a primary key accept new rows (append mode) but cannot update or remove existing ones.
Editing a cell
Click any cell in an editable column to open it for editing. Type your value and press Enter (or click away) to save. The platform sends a partial upsert — only the edited cell changes; the rest of the row is untouched. Press Escape to cancel.
Adding a record
Click + Add record in the toolbar. A blank row appears at the top of the grid for you to fill in. You can tab through the cells. Click Save (or press Enter on the last cell) to commit the row. The row is appended when the table has no primary key; it is upserted on the primary key when one is defined.
Deleting rows
On a table with a primary key, hover a row to reveal the delete icon in the row gutter. Confirm to remove the row permanently. Deletes are not batched — each row is removed individually.
Viewing row details
Click any row in the grid to open the row detail panel — a side sheet that shows all column values for that row, including masked columns in their styled form. You can also open it from the row's ⋯ menu → View details.
The detail panel is read-only. Use it to inspect a row without navigating away, copy individual field values, or review governance metadata (mask, type) per column.
Per-row actions
Each row has a ⋯ (actions) menu in the trailing column. The default actions for managed tables are:
- View details — opens the row detail panel.
- Copy row — copies the row's values to the clipboard as JSON.
- Delete row — permanently removes the row (primary-key tables only).
The ⋯ menu, checkbox, and editable cells all stop propagation, so clicking them never triggers the row-click detail panel.
Masked columns
Columns with a governance mask applied show their server-supplied value (for example, a partial reveal such as the last four digits of a card number) with a muted text style and a shield glyph. The value is not withheld from the UI — the server has already decided what to reveal. The visual treatment signals that the column is governed; the cell is not editable regardless of the table's primary-key status.
If you believe a column should show a fuller value but doesn't, check your workspace role and the column's governance settings under Classification & masking.
- Hidden columns — columns your access policy excludes entirely. These do not appear in the grid at all.
Authorization is enforced server-side. The UI reflects your permissions, but the write route verifies them independently — you cannot bypass masking or column restrictions through direct API calls.
Loading state
While the first page of data is loading, the grid renders skeleton rows — animated placeholder cells in the shape of the column set. This gives the page visible structure immediately, even before any rows have arrived from the warehouse.
Pagination on large tables
The grid uses infinite scroll for browsing large tables in the preview. Rows are loaded in batches of 100. As you scroll near the bottom of the grid, the next batch is fetched automatically and appended to the grid without a page break. A loading indicator appears at the bottom while the next batch is in flight.
The toolbar shows the current loaded count (e.g. 100 of 24 830 rows). The total count is a COUNT(*) computed on the server and may take a moment to appear on very large tables.
The header row is sticky — it remains pinned to the top of the scroll region as you scroll through the rows, so column labels are always visible.
- Refresh — re-fetches from the first page. Use this after writing to see your change reflected.
Infinite scroll is used for browsing. The underlying keyset (cursor) pagination still applies — the grid never ships the full row set to the client. Each batch is a new server fetch from a stable cursor.
Query time indicator
The toolbar shows a query time chip (e.g. 42 ms) reporting how long the server-side page fetch took. This is the full round-trip inside the warehouse: the number reflects StarRocks execution time and does not include network latency between your browser and the API server.
Limitations
- Iceberg-format tables: the preview grid is read-only for tables migrated to the Iceberg lakehouse. Editing Iceberg tables is not supported in this release.
- Backward pagination: the cursor is forward-only. Clicking Previous steps back through a client-side cursor stack — navigating far backward and then forward again re-fetches intermediate pages.
- Ad-hoc / complex queries: filtering, sorting, and column show/hide are available in the preview toolbar. For multi-table joins, aggregations, or queries beyond the rule builder, use the SQL surface (
Analytics → SQL).
Related
- Managed tables — lifecycle, writes, and write modes.
- Classification & masking — how masks and access policies work.
- Querying — reading from workflows, notebooks, and SQL.