test(dashboard): inspect compact widget controls in accessibility triage
This commit is contained in:
@@ -0,0 +1,16 @@
|
||||
# Dashboard diagnostics compatibility recovery
|
||||
|
||||
29 October 2026. Test-only correction to an inherited diagnostic fixture; application source/runtime remain unchanged.
|
||||
|
||||
The source-bound `integration-05/` continuation passed the nine-case smoke recovery, remaining 95 ordinary cases and seven specialized configurations through RD21. The diagnostics configuration then recorded 14 passes/four failures. Each failed EN/Bokmål light/dark accessibility case expected nine permanently visible size selectors in the former large dashboard editing form. Workstream C intentionally replaced that form with direct widgets and compact native disclosures. The assertion stopped those four cases before their final editor axe scan; no accessibility pass is claimed for the failed attempt.
|
||||
|
||||
The corrected fixture verifies the editor and all nine actual widget disclosures, checks the compact default, scans it with axe, opens each disclosure using Enter, and retains the exact nine-size-control requirement. It additionally checks all nine height controls. The existing badge, drawer visibility, keyboard focus, settled-state and axe assertions remain unchanged. No application code, test count or accessibility threshold was removed or relaxed.
|
||||
|
||||
Evidence under `.artifacts/pre-phase10-ux/`:
|
||||
|
||||
- `integration-05/{process.json,diagnostics.json,diagnostics-results/}`: original four failures and 14 passes, with unchanged captured application/ordinary-test source and runtime.
|
||||
- `diagnostic-compatibility-01/{original-accessibility.spec.ts,original-source.json}`: original diagnostic source archived and matched to Git before correction. Diagnostic sources lie outside the ordinary runner's source-prefix inventory, so this additional binding is explicit.
|
||||
- `diagnostic-compatibility-01/green.json`: all four updated cases passed with retries disabled, including compact and expanded editor axe scans.
|
||||
- Follow-up review: no assertion weakening or application change found; the new keyboard and height checks add coverage.
|
||||
|
||||
Final qualification still requires the complete 18-case diagnostic configuration and the remaining volume lanes. Their results must be reconciled separately from the ordinary 380-case census and the focused four-case recovery. Historical failures remain evidence. This checkpoint does not promote human/device accessibility or production acceptance.
|
||||
@@ -65,7 +65,18 @@ for (const language of ['en', 'nb'] as const) for (const theme of ['light', 'dar
|
||||
await scan('drawer-closed-settled');
|
||||
await page.getByRole('button', { name: t.dashboardCustomize, exact: true }).click();
|
||||
const editor = page.getByRole('region', { name: t.dashboardEditing, exact: true });
|
||||
await expect(editor.getByRole('combobox')).toHaveCount(9);
|
||||
await expect(editor).toBeVisible();
|
||||
const preview = page.getByRole('region', { name: t.dashboardPreview, exact: true });
|
||||
const disclosures = preview.locator('[data-dashboard-widget] summary');
|
||||
await expect(disclosures).toHaveCount(9);
|
||||
await expect(preview.getByRole('combobox')).toHaveCount(0);
|
||||
await scan('editor-compact');
|
||||
for (const disclosure of await disclosures.all()) {
|
||||
await disclosure.focus();
|
||||
await disclosure.press('Enter');
|
||||
}
|
||||
await expect(preview.getByRole('combobox')).toHaveCount(9);
|
||||
await expect(preview.getByRole('spinbutton')).toHaveCount(9);
|
||||
await scan('editor-settled');
|
||||
await info.attach('phase9-settled-axe', { body: JSON.stringify({ language, theme, badge, drawerAfterEscape, hiddenDrawer, scans,
|
||||
limits: 'Finite native DOM/keyboard evidence; remaining overlapping background contrast and actual screen-reader perception still require human assessment.' }, null, 2), contentType: 'application/json' });
|
||||
|
||||
Reference in New Issue
Block a user