peter
0d5c6273f1
- A URL mirror names the view again on every entry Back or Forward lands on, so a change made under a dialog, a drawer or a sub-view is not undone by the next reload. - A view pushed from the entry a reload under a dialog started from no longer inherits its overlay mark; a reload under two overlays closes onto the page in one step. - The linked player reopens over a live search, gives itself up when its list moves first, and plays alone over an emptied playlist; a capped Files list keeps the link it cannot resolve. - The Feed press over a channel ends the search; leaving a search ends its asked mark; Files and Playlists searches end with their page; the feed always names its view and debounces its search. - Search-box keys are declared per module and kept out of the client log below TRACE. - Writers outside lib/ go through codecs; the URL contract scans computed keys, overlay writers and two-line reads, and holds every page to a bookmark round trip (Config, Activity, Playlists and the Files sort and viewer added).
E2E characterization (golden-master) net
Playwright tests that lock the app's five core flows — feed (scroll + filter), channel (open → back), player (open/close), settings (dirty-save + guard), and page switching — so the Platform refactor can prove it changed nothing. They assert current, correct behavior; they deliberately do not encode the known Platform bugs (those get their own specs when the sprints that fix them land).
Run
siftlode e2e # against the baked localdev :8080 stack (authoritative)
siftlode e2e-ui # same, Playwright UI mode
siftlode e2e-seed # (re)seed the test account only, no tests
The localdev stack must be up (siftlode dev). global.setup.ts reseeds the account and logs in
once, sharing the session via e2e/.auth/state.json (git-ignored).
A dropped localhost connect — on a Windows host, a burst of test traffic can drop a loopback
connect now and then (every port, any program). The browser resends it and only waits (a loader
outlives an expect timeout); Node does not, so a request from the test side fails with
connect ETIMEDOUT. Those requests go through retryConnect (e2e/retryConnect.ts), which tries
that one error again and passes every other failure straight on. Don't run two e2e suites on the
same machine at once.
Browsers — every flow runs in three projects: chromium (Desktop Chrome), webkit (Desktop
Safari) and ipad (iPad gen 7 landscape: 1080×810, touch, mobile) — except a11y, member and
devices, which run on chromium only (CHROMIUM_ONLY; why is in the docs/gates.md row). Install the engines once per
machine with npx playwright install chromium webkit; run one project with
npx playwright test --project=ipad. WebKit catches what every iOS browser inherits from the engine
(JS, CSS/layout, event order). It does not reproduce iPad media refusals: Playwright's WebKit
decodes through the host OS (Media Foundation on Windows, GStreamer on Linux), and measured on
2026-09-14 it played an mp3-in-mp4 and an AV1 mp4 that a real iPad refuses. Check a media format
on a real iPad against the dev stack over the LAN (http://<desktop LAN IP>:8080, e.g. a
login-free /watch/<token> share link).
Target — PW_BASE_URL (default http://127.0.0.1:8080):
-
:8080(baked SPA) is authoritative and same-origin. After a frontend change you mustsiftlode dev(rebuild) for its baked SPA to carry the latest test hooks. -
:5173(Vite dev) serves source live — the fast refactor loop; it proxies/apito:8080, so the stack must still be up. All five flows are proxy-safe:PW_BASE_URL=http://localhost:5173 siftlode e2e
Accessibility — a11y.spec.ts runs axe-core's WCAG 2.x A + AA rules (@axe-core/playwright)
over every rail page, every admin page and the player, in chromium only (the rules read the DOM,
which the engines build alike). It does not judge colour contrast: axe files text on a gradient —
every glass surface — as "incomplete", not as a pass or a hit.
Account & fixture
Tests run as a dedicated admin account e2e@siftlode.local (never a real account;
admin so the suite can reach the admin-only pages / render branches, matching the real primary
user), reset to a deterministic baseline every run by
scripts/e2e_seed.py (run
in-container so it reuses the app's models + argon2 hasher): the top-3 organic channels subscribed,
one in-progress + one watched + one saved video, prefs pinned to language: en. A second account,
the plain member e2e-member@siftlode.local, gets the same baseline and its own session
(e2e/.auth/member.json). Two specs run as it: member.spec.ts, holding what a non-admin sees (no
admin entry, no admin call, no failed request, an admin link lands on the feed), and devices.spec.ts,
which signs it in on a desktop and an iPad at once and changes its channel layout on the server (put
back in a finally) — a spec that reads the member's server prefs must not run beside it. Credentials come from E2E_EMAIL (default
e2e@siftlode.local), E2E_MEMBER_EMAIL (default e2e-member@siftlode.local) and E2E_PASSWORD
(both accounts) — required, no default: run the suite through the ops wrapper, which exports it
from the secret store. Override the container name with
E2E_API_CONTAINER; skip the in-container seed with E2E_SKIP_SEED=1.
Selectors
The app has no data-testids of its own and all copy is i18n (en/hu), so the specs key on a thin
data-testid layer (added with this sprint) plus the locale pinned to en. Stable ARIA hooks that
already existed are used directly (role="dialog" on the player, aria-current="page" on the
active nav row, aria-pressed on the feed toggles).
Typecheck
npx tsc -p e2e/tsconfig.json --noEmit (the specs are outside the app's src tsconfig).