Production Checklist

Use this list right before you ship. Each item is one line; click through to the deep guide for the how and why. If you can tick every box, you’re production-ready.

Security

• Sanitize external data before assigning to grid.rows — validate types, ensure unique row IDs, strip unexpected fields.
• No innerHTML from user input in custom renderers. Use textContent, or sanitize with DOMPurify first. See Renderer security.
• CSP allows style-src 'self' — the grid injects styles via adoptedStyleSheets. See Styles not applying (CSP).
• Target browsers meet the minimum — Chrome/Edge 123, Firefox 121, Safari 17.5. See Browser support.
• Trusted Types policy installed if you enforce require-trusted-types-for 'script'. See Trusted Types.

Performance

• Only the features you use are imported (or use /features/* side-effect imports for tree-shaking). Avoid @toolbox-web/grid/all in production.
• Fixed rowHeight set for datasets over a few thousand rows. See Performance guide.
• Column virtualization enabled if you render 50+ columns. See ColumnVirtualizationPlugin.
• Server-side loading considered for million-row datasets. See ServerSidePlugin.
• Bundle size measured — your final grid bundle is within the bundle budget.

Reliability

• Async data fetches handle failure — grid.loading toggled, errors caught, user shown a message.
• Grid configuration wrapped in try/catch as a safety net — TBW configuration errors include import hints, but should not crash your app.
• Stable getRowId if your rows can be re-fetched / re-sorted — keeps selection, focus, and edit state correct.
• Automated tests cover the critical paths (sort, filter, edit, selection) and use the stable selector contract — not visual indices or framework-generated class names.

Accessibility

• Grid has an accessible name — via <tbw-grid-header title="…">, gridAriaLabel, gridAriaLabelledBy, or aria-label.
• Tab → arrow keys → Enter → Escape all work as described in the Accessibility guide.
• Tested with at least one screen reader (NVDA, VoiceOver, or JAWS) — see Testing with screen readers.
• Forced-colors / Windows High Contrast verified visually.

Monitoring (optional but recommended)

• Track sort / filter / selection events (sort-change, filter-change, selection-change) to understand how users use the grid in production.
• Sentry / equivalent captures any uncaught grid errors with the TBW error code intact (it’s in the message).

See also

PerformanceProfiling, virtualization tuning, and scale-tier reference

AccessibilityWCAG 2.1 AA, screen reader testing, focus management

TroubleshootingCommon deployment issues and fixes

Automated testingPlaywright, Cypress, and WebdriverIO recipes

Error CodesTBW error code reference