Testing the Walkontable Rendering Engine

Walkontable follows the same freeze as the main suite. Its Playwright home is tests/e2e/walkontable/ (page objects in tests/fixtures/pages/walkontable/). Maintenance edits to existing *.spec.js are allowed; new or flaky walkontable tests go to Playwright (the presence gate blocks a new walkontable *.spec.js). In CI the legacy Jasmine/Puppeteer walkontable job and the Playwright e2e job run in parallel. This guide is for maintaining the frozen Jasmine specs.

Separate test pipeline

• Run: npm run test:walkontable --prefix handsontable (the test/spec/ specs). test:e2e does not pick them up.
• Location: src/3rdparty/walkontable/test/
  • test/spec/: Jasmine + Puppeteer with a separate Rspack config and bootstrap. Only test:walkontable runs them.
  • test/unit/: Jest tests for calculators, filters, renderers, utilities (48 *.unit.ts/*.unit.js files). The core jest.config.js picks them up, so npm run test:unit and the core Unit / test CI job run them, and test:walkontable does not.

Writing tests

• it() callbacks that call rendering APIs are async and await those calls.
• Tests are organized by subsystem (overlay/, scroll/, selection/, renderer/, table/, viewport.spec.js); place new ones in the matching directory.

What to cover

• Frozen rows and columns: the 6 overlay types are the most fragile part; include frozen scenarios for overlay positioning and sync.
• Viewport calculations: small containers that clip, containers larger than the data, dynamic resize.
• Scroll synchronization: horizontal and vertical scroll keep frozen overlays aligned.
• Large datasets: 10k+ rows. Populate arrays with forEach; arr.push(...largeArray) overflows the stack.

Common mistakes

• Running Walkontable tests via test:e2e.
• Changing the test bootstrap or Rspack config without confirming test/spec/ passes under test:walkontable and test/unit/ under the core test:unit.
• Clearing an inline overflow longhand with jQuery: $el.css('overflow-x', '') leaves an inline overflow-x in this harness, so the "clip removed" branch asserts against the wrong layout. Write el.style.overflowX = '' and assert the intermediate fact (getComputedStyle(el).overflowX === 'visible') before the behavior.
• Trusting the spec count. A failing spec whose expected/actual was a cyclic object (toBe(window), toEqual([overlay, ...])) used to be dropped at the Puppeteer bridge (Running 16 specs., then 15 specs, 0 failures, exit 0; the getOverlays spec sat like that from #12951). test/helpers/jasmine-bridge-reporter.js now reports it as a failure ([unserializable Window]). If a count comes up short, compare the Running N specs. line with the summary via npm run test:walkontable -- --testPathPattern=<file> --verbose.
• Turning on rowHeightsUniform/columnWidthsUniform in a bare Walkontable spec and asserting scrollbar-dependent row/column counts: the bare harness overflows content without a real scrollbar while the single-pass layout snapshot predicts one, so they diverge. Assert the snapshot booleans directly, or use a fixture that renders a real scrollbar.