From 1ddca424732e58da2f7fbed0ac543c7c6731e057 Mon Sep 17 00:00:00 2001 From: litruv Date: Tue, 3 Nov 2026 13:22:39 +1100 Subject: [PATCH] Apply the #109 re-review: crash assertion, artifact hook on launch failure, dropped redundant lines. (#109) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit tests/e2e/login.spec.mjs: assert the render process is alive with the `crash` event instead of counting pageerrors; the failed-login test now accepts the rate-limited message as well as the wrong-password one; the failure hook attaches the console before the app guard so a launch failure still produces an artifact, wraps `screenshot()` because it throws on a crashed page, and reaches `fs.rmSync` on launch failure; dropped the duplicate login-form assertion. playwright.config.mjs: pin `trace: 'off'` explicitly rather than relying on the version default, since a trace captures the login POST body. .gitea/workflows/tests.yml: dropped `--prefer-offline`; corrected the xvfb comment (the guard is live for the host runner, which already has xvfb); dropped an unresolvable CI run-number citation. Checks: `node --check` on both JS files, `npm run test:unit` — 2 passed. E2E cannot be run locally (no `E2E_PASSWORD` in this session); CI is the check. --- .gitea/workflows/tests.yml | 10 +++++--- playwright.config.mjs | 4 +++ tests/e2e/login.spec.mjs | 52 +++++++++++++++++++++++--------------- 3 files changed, 41 insertions(+), 25 deletions(-) diff --git a/.gitea/workflows/tests.yml b/.gitea/workflows/tests.yml index 12f7f07..2fc634c 100644 --- a/.gitea/workflows/tests.yml +++ b/.gitea/workflows/tests.yml @@ -58,17 +58,19 @@ jobs: # The E2E run drives the vite dev server, so cinny's dependencies are needed too. - name: Install dependencies (cinny) working-directory: ./cinny - run: npm ci --prefer-offline + run: npm ci # Electron needs system libraries the runner image does not ship. Do not narrow this to # chromium: Playwright's chromium dep set covers its own bundled browser, not Electron, # and drops libgtk-3-0t64, libegl1, libegl-mesa0, libgles2, libsecret-1-0, libxtst6 and - # libxss1, which Electron needs to launch. Run 930 failed with "Process failed to launch!" - # on the narrowed set. + # libxss1, which Electron needs to launch. Narrowing it fails with "Process failed to + # launch!". - name: Install system libraries for Electron run: npx playwright install-deps - # The runner image has no xvfb and no fresh apt lists. + # The container runner image has no xvfb and no fresh apt lists. The host runner already + # has xvfb, so the guard skips the install there; both share the ubuntu-latest label, so + # which one runs is not fixed. - name: Ensure a display for Electron run: command -v xvfb-run >/dev/null || (sudo apt-get update && sudo apt-get install -y xvfb) diff --git a/playwright.config.mjs b/playwright.config.mjs index 2300866..fa2bb37 100644 --- a/playwright.config.mjs +++ b/playwright.config.mjs @@ -10,7 +10,11 @@ export default defineConfig({ workers: 1, // The suite logs in to a live homeserver, where Synapse login rate limiting is a real flake. retries: process.env.CI ? 1 : 0, + // An explicit reporter replaces the default, so list stays for the CI log. reporter: [['list'], ['html']], + // A trace captures the login POST body, which contains the password. Pin it off rather than + // relying on the version default. + trace: 'off', // cinny/vite.config.js binds 0.0.0.0 with fs.allow: ['..'], which is exposed on a shared // CI runner, so pin the dev server to loopback. The flag has to reach vite itself: // `npm run dev:vite -- --host` lands on sh's positional params, not on vite. diff --git a/tests/e2e/login.spec.mjs b/tests/e2e/login.spec.mjs index a1d4b4b..9e1141a 100644 --- a/tests/e2e/login.spec.mjs +++ b/tests/e2e/login.spec.mjs @@ -11,6 +11,7 @@ let app; let page; let logs = []; let profile; +let crashed = false; // ServerPicker gives its input no name or test id, so position is the only handle on it; // this breaks if the form ever gains a field before the ones below. @@ -43,6 +44,7 @@ async function submitLogin(username, password) { test.beforeEach(async () => { logs = []; + crashed = false; if (!PASSWORD) { throw new Error('E2E_PASSWORD is not set. Export it, or add it as the CI secret.'); @@ -63,29 +65,37 @@ test.beforeEach(async () => { } page.on('console', (msg) => logs.push(`[renderer] ${msg.type()}: ${msg.text()}`)); page.on('pageerror', (err) => logs.push(`[renderer] pageerror: ${err}`)); + // crash is render-process-gone, so it does not fire on the MatrixError or the ServiceWorker error. + page.on('crash', () => { + crashed = true; + }); await chooseServer(HOMESERVER); }); // Config-based screenshots do not fire for a page taken from app.firstWindow(), and console -// capture is not covered by config at all, so this hook stays manual. Traces stay off because a -// trace captures the login POST body. +// capture is not covered by config at all, so this hook stays manual. test.afterEach(async ({}, testInfo) => { - if (!app) return; - if (testInfo.status !== testInfo.expectedStatus) { - if (page && !page.isClosed()) { - await testInfo.attach('failure.png', { body: await page.screenshot(), contentType: 'image/png' }); - } + // Console first: a failed launch has no app and no page, so this is the only artifact that + // path can produce. await testInfo.attach('console.log', { body: logs.join('\n'), contentType: 'text/plain' }); + if (page && !page.isClosed()) { + // screenshot() throws on a crashed page, which would mask the original failure. + const shot = await page.screenshot().catch(() => null); + if (shot) await testInfo.attach('failure.png', { body: shot, contentType: 'image/png' }); + } } - await app - .close() - // main.js preventDefault()s window close to minimise to tray, so close can hang. - .catch(() => app.process().kill('SIGTERM')); + if (app) { + await app + .close() + // main.js preventDefault()s window close to minimise to tray, so close can hang. + .catch(() => app.process().kill('SIGTERM')); + } - fs.rmSync(profile, { recursive: true, force: true }); + // Reachable on launch failure, so the temp profile does not leak. + if (profile) fs.rmSync(profile, { recursive: true, force: true }); }); async function expectAuthenticated() { @@ -108,9 +118,6 @@ async function expectAuthenticated() { return res.status; }) .toBe(200); - - // The login form is gone, so the app is on the authenticated shell. - await expect(usernameInput()).toHaveCount(0); } test('test account can log in and reaches the authenticated app', async () => { @@ -118,14 +125,17 @@ test('test account can log in and reaches the authenticated app', async () => { await expectAuthenticated(); }); -test('failed login reports an error without crashing the app', async () => { +test('failed login reports an error and keeps the renderer alive', async () => { await submitLogin(USER, 'not-the-password'); - await expect(page.getByText('Invalid Username or Password.')).toBeVisible(); + // A rate-limited run renders LoginError.RateLimited instead of the wrong-password message, + // so accept both branches rather than asserting one exact string. + await expect(page.getByText(/Invalid Username or Password\.|rate-limited by server/)).toBeVisible(); - // "Without crashing" is proven by the form still rendering: a dead renderer cannot render it. - // Pageerrors go into the artifact but are not asserted — the failed login itself produces - // MatrixError, and dev mode produces a ServiceWorker registration error. No second login - // here: every run costs Synapse rc_login budget, and the re-login was the main flake source. + // A dead renderer cannot render the form. Pageerrors go into the artifact but are not + // asserted — the failed login itself produces MatrixError, and dev mode produces a + // ServiceWorker registration error. No second login here: every run costs Synapse rc_login + // budget, and the re-login was the main flake source. await expect(usernameInput()).toBeVisible(); + await expect(crashed).toBe(false); });