feat(wails): move the e2e harness and headless launch onto v3

make e2e is green on chromium: 92 passed. The harness is rebuilt on
what v3 actually offers, and three of the four things it replaced turn
out to be better than what they replaced.

The headless launch is v3's own server mode. scripts/dev-headless.sh
ran a `-tags dev` binary whose app_dev.go parsed -devserver/-assetdir
out of os.Args; that file went with v2, so the harness had no server at
all. `-tags dev,server` is a first-class mode and needs no display, so
Xvfb is gone from the script and from CI.

The bridge hooks two places, neither of them EventsOn. Inbound is
window._wails.dispatchWailsEvent, wrapped by pre-creating the object
the runtime keeps and putting an accessor on the one property.
Outbound is fetch: v3 routes every runtime call through one POST, so
the bridge sees binding calls and event emits from any module, needs no
walk of an object graph, and cannot miss a call made before it looked.

__yjEvents.call posts to that endpoint by method name, so it depends on
nothing in the app's bundle and works on a page with no init script.
That is what lets seed-sandbox.sh drop playwright-cli entirely — it
drove AddLibrary through a browser only because window.go was v2's one
way in — and with it a global npm install and a second Chromium in CI.

measure.mjs and one spec lose their window.go walks and read the
bridge's log instead; e2e/support/method-ids.mjs derives id -> name
from frontend/bindings/ (phase 6b option 1, so it cannot go stale
silently). Plain .mjs because measure.mjs runs under bare node and one
derivation beats two that can disagree.

Four bugs surfaced, and the migration is how.

The cross-service wiring never ran headless. It hung off
Common.ApplicationStarted, which server mode never emits —
setupCommonEvents is an explicit no-op there — so the queue had no
TrackLoader and playing a track changed the queue and then silently did
nothing. It is a service registered last now (backend/startup.go):
services start in registration order, which is the ordering the wiring
needs, in every mode.

Six specs called SetQueue with 3 of its 4 arguments. v2 accepted that
and filled the gap; v3 answers "expects 4 arguments, got 3".

requested-badge's cleanup read window.go and returned early on
`if (!svc)` — the silent cleanup its own comment was written to
prevent, one migration later. It posts to the runtime endpoint now,
which any page can do.

SearchIndex.Search trusted a startup latch, so rows a spec staged
afterwards were unsearchable and three specs passed only when an
earlier one happened to flip it. shelves.go fixed exactly this and left
hasCatalogRows behind; the search path now uses it as the fallback,
with the latch still the fast path.

Two spec edits are deletions of assertions about v2. harness.spec
checked Object.keys(window.go) and that a bad call *hung*; it now
checks the real runtime is loaded and that the backend rejects with a
TypeError naming the argument. album-actions asserted a tracklist
legend that dcc40b1 deleted on main — that spec has been failing since,
and what replaced it is covered in frontend/test/components.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UDCbcCZQepnpSQYJ6SxxZm
This commit is contained in:
2026-08-14 20:58:20 -04:00
co-authored by Claude Opus 5
parent 60779c41c3
commit deb3f3da7e
22 changed files with 728 additions and 328 deletions
+10 -17
View File
@@ -220,7 +220,7 @@ jobs:
apt-get install -y -qq --no-install-recommends \
ca-certificates curl git jq build-essential pkg-config \
libwebkitgtk-6.0-dev libgtk-4-dev libasound2-dev \
xvfb dbus dbus-x11 ffmpeg libasound2t64 \
dbus dbus-x11 ffmpeg libasound2t64 \
alsa-utils libasound2-plugins pulseaudio pulseaudio-utils
- name: Clone repo at this commit
@@ -247,24 +247,17 @@ jobs:
apt-get install -y -qq --no-install-recommends nodejs
corepack enable
# scripts/seed-sandbox.sh drives the real AddLibrary binding
# through playwright-cli, so the CLI has to be on PATH.
- name: Playwright CLI
run: npm install -g @playwright/cli
# @playwright/cli is gone with v2. seed-sandbox.sh drove the real
# AddLibrary binding through a browser because `window.go` was the
# only way in; v3 answers the same call over HTTP, so the seed is
# curl now and needs no CLI, no second Chromium and no shared
# PLAYWRIGHT_BROWSERS_PATH revision dance.
- name: Browsers
working-directory: /src/e2e
run: |
set -eu
# PLAYWRIGHT_BROWSERS_PATH unifies the *location*, not the
# *revisions*: @playwright/cli bundles its own playwright-core
# pinned to a different Chromium build than @playwright/test,
# so each installs its own into the shared directory. Drop
# either line and the other fails with "Browser chromium is
# not installed; expected executable at ...".
pnpm install --frozen-lockfile
npx playwright install --with-deps chromium webkit
playwright-cli install-browser chromium
# oto/v3 talks to libasound directly, and a container has no
# PulseAudio socket to fall back on — so it needs a default device
@@ -287,10 +280,10 @@ jobs:
#
# PulseAudio's null sink is timer-scheduled and does pace — the
# 0.76 s over is the buffer draining, not a rate error; 12 s of
# audio takes 13.5 s. Verified under the private session bus and
# Xvfb that dev-headless.sh runs the app in. It needs no system
# D-Bus and no kernel module, which is why it is reachable from a
# container at all.
# audio takes 13.5 s. Verified under the private session bus
# dev-headless.sh runs the app in. It needs no system D-Bus and
# no kernel module, which is why it is reachable from a container
# at all.
- name: Real-time audio sink
run: |
set -eu
+15 -6
View File
@@ -21,12 +21,21 @@ here has disappeared.
Fifteen things cost a cycle each the first time. They are here, not in a
reference, because you need them *before* the failure, not after.
- **Time out every binding call.** A bound Go method called with wrong
argument types makes the backend log `error parsing arguments` and
**never fire the callback**, so the promise hangs forever. Use
`window.__yjEvents.call(path, args, ms)` (browser) or `callBinding`
(specs), never a bare `window.go.…`. When one hangs anyway,
`make dev-logs``.dev/app.log` is the only place the reason appears.
- **Call a binding through the bridge.** `window.go` does not exist
under Wails v3 — the bindings are bundled modules, not a global — so
use `window.__yjEvents.call(path, args, ms)` (browser) or
`callBinding` (specs). Both post to the runtime's own endpoint by
method name, so they work on any page, including one with no init
script.
A bad call now *rejects*, and says why: a wrong type comes back as a
TypeError naming the argument, a wrong count as
`expects 4 arguments, got 3`, an unknown method as a ReferenceError.
Under v2 the backend logged `error parsing arguments` and never fired
the callback, so `.dev/app.log` was the only place the reason
appeared and the timeout was the only thing that made the mistake
visible. The timeout is still there, but now it means a genuinely
hung request.
- **Nothing is clickable on a fresh `YJ_HOME`.** `<first-run-wizard>`
intercepts all pointer events until a library exists, and the click
fails with a Playwright interception error that reads like a selector
@@ -74,9 +74,14 @@ behind `YJ_TESTCTL=1`, which `scripts/dev-headless.sh` sets and
- **`snapshot` writes a file, it does not print the tree.** The
command prints a path under `outputDir`; read that. Only the tail
is echoed.
- **Three separate browser caches.** `playwright-cli`, `@playwright/test`
- **Two separate browser caches.** `@playwright/test`
(`make e2e-setup`) and the Vitest provider (`make ui-setup`) each
download their own Chromium. One working is no guarantee for the next.
download their own Chromium. One working is no guarantee for the
other. There used to be a third: `playwright-cli` was a *required*
dependency because `scripts/seed-sandbox.sh` drove `AddLibrary`
through a real page, `window.go` being v2's only way in. v3 answers
the same call over HTTP, so the seed is `curl` now and the CLI is
only an exploratory convenience.
- **`getByRole('button', { name })` matches substrings.** "Play" also
matches "Add queue to playlist"; transport controls need
`exact: true`.
+214 -97
View File
@@ -7,31 +7,34 @@
* other events arrive from Go whenever they arrive. An assertion that
* sleeps and hopes is flaky; an assertion that awaits the event is not.
*
* Three things it provides on `window.__yjEvents`:
* Four things it provides on `window.__yjEvents`:
*
* record every backend -> frontend event, in order, with payloads
* wait a promise that settles on a matching event (or rejects
* with the list of events that *did* arrive, which is the
* single most useful failure message this harness can give)
* call a bound Go method that is guaranteed to settle: a binding
* invoked with wrong argument types makes the backend log
* "error parsing arguments" and never fire the callback, so
* the in-page promise hangs forever. Timing out here fixes
* that once instead of in every eval.
* call a bound Go method, by name, over the runtime's own HTTP
* endpoint — no dependence on the app's bundle
* bindings every binding call the *app* made, which is what turns
* "did that refetch the library" from an inference into a
* fact (e2e/perf/measure.mjs labels and reads these)
*
* WHERE IT HOOKS. Not EventsOn. Every backend event enters the page
* at exactly one place — wails' ipc_websocket.js does
* WHERE IT HOOKS. Two places, and neither is `EventsOn`.
*
* case "n": window.wails.EventsNotify(message)
* Inbound, `window._wails.dispatchWailsEvent`: v3's runtime assigns it
* at module scope and it is the single point every backend event enters
* the page through, so wrapping it captures all 46 whether or not the
* app subscribes to them. The runtime does
* `window._wails = window._wails || {}`, so this script creates that
* object first and puts an accessor on the *property*, wrapping at
* assignment time — v2 needed the accessor on `window` itself, because
* there the whole object was replaced.
*
* and EventsNotify fans out to listeners from there. Wrapping that
* single choke point captures all 46 events whether or not the app
* subscribes to them, and needs one wrap rather than 46.
*
* `window.wails` does not exist yet when this script runs, so we install
* an accessor on `window` and wrap at assignment time (wails' main.js
* does a plain `window.wails = {...}`), then collapse the accessor back
* to a data property so nothing downstream can tell.
* Outbound, `fetch`: v3 routes every runtime call — binding calls, event
* emits, window and dialog calls — through one POST to /wails/runtime.
* There is no global to wrap the way v2's `window.runtime` could be, and
* this is better anyway: it sees calls from any module, needs no walk of
* an object graph, and cannot miss one made before the harness looked.
*
* INSTALL EXACTLY ONCE. Listeners registered by one `eval` survive into
* the next, so a recorder that re-registers double-counts. Tests call
@@ -44,8 +47,27 @@
const LIMIT = 2000;
// Every bound service in this app lives under this Go module path,
// so specs name a binding the short way — 'queue.Queue.GetState' —
// and this is what makes that the same thing the backend calls
// 'yellowjacket/backend/queue.Queue.GetState'.
const FQN_PREFIX = "yellowjacket/backend/";
// The runtime's own object and method ids (objectNames in
// @wailsio/runtime): 0 is Call, 3 is Events, and method 0 on each is
// CallBinding and Emit respectively.
const OBJECT_CALL = 0;
const OBJECT_EVENTS = 3;
// Captured before the wrap below, and used for the harness's own
// calls: `__yjEvents.call` is this file talking to the backend, not
// the app, and counting it would make "did that action refetch the
// library" answer for the question as well as the app.
const nativeFetch = window.fetch.bind(window);
let seq = 0;
const log = [];
const bindings = [];
const waiters = new Set();
const summarize = () => {
@@ -56,6 +78,25 @@
return counts;
};
/*
* `data` is recorded as the argument list Go emitted, which is the
* shape every spec reads (`ev.data[0]`).
*
* v3's EventManager.Emit packs a variadic call into one field: no
* arguments is null, one is the value itself, more than one is the
* slice. Unpacking that back into a list is exact except for a
* single argument that is itself an array, which is indistinguishable
* from several arguments — an ambiguity v3 introduced and no
* assertion here depends on, since nothing in backend/events emits
* more than one value.
*/
const argsOf = (data) => {
if (data === null || data === undefined) {
return [];
}
return Array.isArray(data) ? data : [data];
};
const record = (name, data, dir) => {
const entry = { seq: ++seq, name, data, dir, t: Date.now() };
log.push(entry);
@@ -89,7 +130,7 @@
};
const api = {
version: 1,
version: 2,
/** Every recorded event, oldest first. */
get log() {
@@ -101,10 +142,30 @@
return seq;
},
/** Drop the buffer. Does NOT touch the recorder or waiters. */
/**
* Every binding call the app made, oldest first. Each is
* { methodID, methodName, start, ms, bytes } — the id is what the
* generated bindings send, and turning it back into a name is
* e2e/perf/measure.mjs's job, which derives the map from
* frontend/bindings/.
*/
get bindings() {
return bindings.slice();
},
/**
* Read the size of every binding response. Off by default: it
* costs a clone-and-read of each body, which only a measurement
* wants to pay. With it off, `bytes` is the Content-Length when
* the server sent one and -1 otherwise.
*/
measureBytes: false,
/** Drop the buffers. Does NOT touch the recorder or waiters. */
reset() {
const n = log.length;
log.length = 0;
bindings.length = 0;
return n;
},
@@ -177,13 +238,11 @@
async ready(timeoutMs) {
const deadline = Date.now() + (timeoutMs || 15000);
for (;;) {
if (window.go?.queue?.Queue?.GetState) {
try {
await api.call("queue.Queue.GetState", [], 2000);
return true;
} catch {
/* backend not up yet */
}
try {
await api.call("queue.Queue.GetState", [], 2000);
return true;
} catch {
/* backend not up yet */
}
if (Date.now() > deadline) {
throw new Error("__yjEvents.ready timed out");
@@ -193,38 +252,66 @@
},
/**
* Call a bound Go method by dotted path, with a timeout.
* Call a bound Go method by dotted path.
*
* await __yjEvents.call('player.Player.SetVolume', [42])
*
* A binding called with the wrong argument types never fires its
* callback — the reason appears only in .dev/app.log. Without a
* timeout the caller waits forever; with one it gets told where
* to look.
* This posts to the runtime's own endpoint rather than reaching
* into the page for a binding function, because v3 has no
* `window.go` and the generated bindings are ordinary bundled
* modules an initScript cannot import. It calls *by name*, which
* the backend resolves the same way it resolves the id the
* bundle sends.
*
* v3 rejects a bad call rather than silently never firing its
* callback the way v2 did — wrong argument types come back as a
* TypeError naming the argument, an unknown method as a
* ReferenceError. The timeout below is therefore a backstop for
* a genuinely hung request, not the mechanism that makes a
* mistake visible.
*/
call(path, args, timeoutMs) {
const parts = String(path).split(".");
let fn = window.go;
for (const p of parts) {
fn = fn?.[p];
}
if (typeof fn !== "function") {
return Promise.reject(
new Error(`__yjEvents.call: no such binding: ${path}`),
);
}
const request = nativeFetch("/wails/runtime", {
method: "POST",
headers: {
"Content-Type": "application/json",
"x-wails-client-id": window._wails?.clientId ?? "",
},
body: JSON.stringify({
object: OBJECT_CALL,
method: 0,
args: {
"call-id": `yj-${Math.random().toString(36).slice(2)}`,
methodName: FQN_PREFIX + String(path),
args: args || [],
},
}),
}).then(async (res) => {
const type = res.headers.get("Content-Type") || "";
const json = type.includes("application/json");
if (!res.ok) {
const body = json ? await res.json() : { message: await res.text() };
throw new Error(
`__yjEvents.call(${path}) failed: ` +
`${body.kind || "Error"}: ${body.message}`,
);
}
return json ? res.json() : res.text();
});
return Promise.race([
Promise.resolve(fn(...(args || []))),
request,
new Promise((_, reject) =>
setTimeout(
() =>
reject(
new Error(
`__yjEvents.call(${path}) did not settle in ` +
`${timeoutMs || 10000}ms — almost always wrong ` +
`argument types; check .dev/app.log for ` +
`"error parsing arguments"`,
`${timeoutMs || 10000}ms — the runtime endpoint ` +
`hung, which is not how a bad argument fails; ` +
`check .dev/app.log`,
),
),
timeoutMs || 10000,
@@ -241,62 +328,92 @@
writable: false,
});
// Wrap `obj[method]` once, routing every invocation through `tap`.
const wrap = (obj, method, tap) => {
const original = obj[method];
if (typeof original !== "function" || original.__yjWrapped) {
return;
}
const wrapped = function (...args) {
try {
tap(args);
} catch {
/* a broken recorder must never break the app */
}
return original.apply(this, args);
};
wrapped.__yjWrapped = true;
obj[method] = wrapped;
};
// ── Inbound ──────────────────────────────────────────────────────
//
// The runtime keeps whatever `window._wails` already is, so creating
// it here and defining an accessor on the one property we care about
// means the wrap happens the moment the runtime module is evaluated.
window._wails = window._wails || {};
// Install an accessor that wraps on first assignment, then collapses
// back into an ordinary property.
const hookOnAssign = (name, onAssign) => {
let value;
Object.defineProperty(window, name, {
configurable: true,
enumerable: true,
get: () => value,
set: (v) => {
value = v;
let dispatch;
Object.defineProperty(window._wails, "dispatchWailsEvent", {
configurable: true,
enumerable: true,
get: () => dispatch,
set: (fn) => {
dispatch = function (event) {
try {
onAssign(v);
record(event?.name, argsOf(event?.data), "in");
} catch {
/* ditto */
/* a broken recorder must never break the app */
}
Object.defineProperty(window, name, {
value: v,
configurable: true,
enumerable: true,
writable: true,
});
},
return fn.apply(this, arguments);
};
},
});
// ── Outbound ─────────────────────────────────────────────────────
//
// One POST per runtime call. Only two of the thirteen object ids
// are interesting here; the rest (window, dialogs, clipboard) pass
// through untouched and unrecorded.
window.fetch = function (input, init) {
let call = null;
try {
// The runtime passes a **URL object**, not a string — it
// builds `new URL(runtimeURL())` — and a URL has no `.url`,
// only a Request does. Reading the wrong one matched
// nothing and recorded no calls at all, which looks
// identical to an app that made none.
const url =
input && typeof input === "object" && "url" in input
? input.url
: String(input ?? "");
if (
url.includes("/wails/runtime") &&
init?.method === "POST" &&
typeof init.body === "string"
) {
const body = JSON.parse(init.body);
if (body.object === OBJECT_EVENTS && body.method === 0) {
record(body.args?.name, argsOf(body.args?.data), "out");
} else if (body.object === OBJECT_CALL && body.method === 0) {
call = {
methodID: body.args?.methodID ?? null,
methodName: body.args?.methodName ?? null,
start: performance.now(),
};
}
}
} catch {
/* ditto */
}
const response = nativeFetch(input, init);
if (!call) {
return response;
}
return response.then(async (res) => {
try {
call.ms = performance.now() - call.start;
call.bytes = api.measureBytes
? (await res.clone().text()).length
: Number(res.headers.get("Content-Length") ?? -1);
bindings.push(call);
if (bindings.length > LIMIT) {
bindings.splice(0, bindings.length - LIMIT);
}
} catch {
/* ditto */
}
return res;
});
};
// Inbound: every backend -> frontend event.
hookOnAssign("wails", (w) => {
wrap(w, "EventsNotify", ([message]) => {
const parsed = JSON.parse(message);
record(parsed.name, parsed.data, "in");
});
});
// Outbound: events the frontend emits, so a flow that round-trips
// through Go is legible from one buffer.
hookOnAssign("runtime", (r) => {
wrap(r, "EventsEmit", (args) => {
record(args[0], args.slice(1), "out");
});
});
})();
+9
View File
@@ -247,6 +247,15 @@ func NewYellowJacketApp(
)
}
// Last, deliberately: services start in registration order, so this
// runs once every service above has taken its context. See
// startup.go for why the wiring is a service rather than an
// application-event hook.
yjApp.Services = append(
yjApp.Services,
application.NewService(&startupService{app: yjApp}),
)
return yjApp, nil
}
+12 -1
View File
@@ -1414,7 +1414,18 @@ func (si *SearchIndex) ExactMatches(query string, perCategory int) []SearchIndex
// by relevance (popularity-blended). Returns nil when the index
// hasn't finished its initial build.
func (si *SearchIndex) Search(ctx context.Context, query string, limit int) []SearchIndexResult {
if !si.IsReady() {
// IsReady is latched once at startup, so it is right in the app and
// wrong for anything that fills the table afterwards — including the
// e2e suite staging a catalog, which is how search gets tested in
// CI, where the artifact URL points at a dead address on purpose.
// shelves.go learned this already; the search path did not, and the
// symptom was three specs that passed only when an earlier one
// happened to flip the flag first.
//
// The latch stays as the fast path — once it is true nothing can
// make it false — and the probe is one indexed `SELECT 1 … LIMIT 1`
// on the only path that could otherwise answer "no catalog" wrongly.
if !si.IsReady() && !si.hasCatalogRows(ctx) {
return nil
}
+53
View File
@@ -0,0 +1,53 @@
package backend
import (
"context"
"github.com/wailsapp/wails/v3/pkg/application"
)
// startupService is the cross-service wiring, wearing a service's
// clothes so the runtime starts it like everything else.
//
// The wiring belongs to no single service — it is the hooks, adapters
// and callbacks that make one package drive another — so v2 put it in
// OnStartup and the first v3 port hung it off
// events.Common.ApplicationStarted, which fires after every service's
// own ServiceStartup and is therefore the right *moment*.
//
// It is the wrong *mechanism*, because server mode emits no
// application events at all: v3's setupCommonEvents is an explicit
// no-op under `-tags server` ("server mode has no platform-specific
// events to map"). So the desktop build wired itself and the headless
// build did not, which showed up as "No player set, cannot load track"
// — the queue had no TrackLoader, so a track played from the UI
// changed the queue and then silently did nothing.
//
// Registering last is what preserves the ordering the wiring depends
// on: services start in registration order, on the main goroutine,
// before the platform run loop (application.Run's startup closure), so
// every service this touches has taken its context by the time this
// runs. A service is also the honest description of what this is —
// something with a lifecycle the app owns — and it costs no bindings,
// since ServiceStartup and ServiceShutdown are excluded from them.
type startupService struct {
app *YellowJacketApp
}
// ServiceStartup runs the app-level wiring.
//
// OnDomReady no longer means the DOM is ready — nothing in v3 offers
// that — and it does not need to: what it does is start the soft
// rescan and report a startup failure, neither of which wants a
// frontend. The frontend drives its own state synchronisation by
// calling EmitCurrentState once its stores are listening, which is
// what makes the rename harmless.
func (s *startupService) ServiceStartup(
ctx context.Context,
_ application.ServiceOptions,
) error {
s.app.OnStartup(ctx)
s.app.OnDomReady(ctx)
return nil
}
+74 -62
View File
@@ -56,6 +56,8 @@ import { dirname, resolve } from 'node:path';
import { fileURLToPath } from 'node:url';
import { chromium } from '@playwright/test';
import { methodIDs } from '../support/method-ids.mjs';
const HERE = dirname(fileURLToPath(import.meta.url));
const REPO = resolve(HERE, '../..');
const OUT_DIR = resolve(REPO, '.dev/perf');
@@ -63,6 +65,16 @@ const BRIDGE = resolve(REPO, '.playwright/init-events.js');
const BASE_URL = process.env.YJ_URL ?? 'http://localhost:34115';
/**
* methodID -> 'pkg.Type.Method', derived from frontend/bindings/.
*
* A binding call carries only the id, so this is what turns a
* measurement's "which bindings did that provoke" back into names. It
* is derived rather than written down for the reason plan 009 phase 6b
* gives: a hand-maintained list goes stale silently.
*/
const METHOD_NAMES = Object.fromEntries(methodIDs());
// The browse script visited for the heap measurement. Deliberately the
// views the audit named as retaining: explore (two unbounded caches),
// artists and genres (per-frame work), settings (the 3 s ticker).
@@ -123,64 +135,41 @@ function parseArgs(argv) {
/* -------------------------------------------------------------------- */
/**
* Wrap every bound Go method so a measurement can say which bindings a
* user action provoked and how much they returned.
* Say which bindings a user action provoked, and how much they
* returned.
*
* Post-hoc wrapping is safe because `frontend/wailsjs` looks its target
* up at call time (`window['go']['library']['Library']['GetAllTracks']()`),
* so a store holding an imported wrapper still lands here.
* This used to walk `window.go` and wrap every bound method in place,
* which worked because v2's generated stubs looked their target up at
* call time. v3 has no such object — the bindings are bundled modules
* — so `.playwright/init-events.js` records every call off the single
* POST v3 routes them all through, and this reads that log. It is
* strictly better: it needs no walk, sees calls from any module, and
* cannot miss one made before a wrapper was installed, which is what
* the old "runs twice" dance was working around.
*
* `bytes` is the response size, so `measureBytes` is turned on here —
* only a measurement wants to pay for a clone-and-read of every body.
*/
const INSTRUMENT = `() => {
// Runs twice: once as an initScript (before window.go exists, which
// is the only moment early enough to catch the long-task observer's
// first entries) and once after the bridge reports ready. So the
// state is created at most once and the *walk* happens every time —
// getting that backwards silently measures zero binding calls.
const first = !window.__yjPerf;
const INSTRUMENT = `(names) => {
if (window.__yjPerf) return;
if (first) {
const calls = [];
window.__yjPerf = {
calls,
reset: () => { calls.length = 0; },
since: (t) => calls.filter((c) => c.start >= t),
longtasks: [],
};
}
window.__yjEvents.measureBytes = true;
const calls = window.__yjPerf.calls;
const wrap = (obj, key, path) => {
const fn = obj[key];
if (typeof fn !== 'function' || fn.__yjPerfWrapped) return;
const wrapped = function (...args) {
const start = performance.now();
let out;
try { out = fn.apply(this, args); } catch (e) { throw e; }
return Promise.resolve(out).then((v) => {
let bytes = 0;
try { bytes = JSON.stringify(v ?? null).length; } catch { bytes = -1; }
calls.push({ path, start, ms: performance.now() - start, bytes });
return v;
});
};
wrapped.__yjPerfWrapped = true;
obj[key] = wrapped;
window.__yjPerf = {
get calls() {
return window.__yjEvents.bindings.map((c) => ({
path: names[c.methodID] || ('#' + c.methodID),
methodID: c.methodID,
start: c.start,
ms: c.ms,
bytes: c.bytes,
}));
},
reset: () => { window.__yjEvents.reset(); },
since: (t) => window.__yjPerf.calls.filter((c) => c.start >= t),
longtasks: [],
};
const walk = (obj, prefix, depth) => {
if (!obj || depth > 4) return;
for (const key of Object.keys(obj)) {
const v = obj[key];
if (typeof v === 'function') wrap(obj, key, prefix + key);
else if (v && typeof v === 'object') walk(v, prefix + key + '.', depth + 1);
}
};
walk(window.go, '', 0);
if (!first) return;
// Long tasks are the honest form of "the app stalls": a 25 MB JSON
// parse on the main thread shows up here and nowhere else.
try {
@@ -352,7 +341,15 @@ async function measureTrackChange(page) {
const paths = (tracks ?? []).slice(0, 4).map((t) => t.FilePath);
if (paths.length < 2) return { error: 'library too small to measure' };
await ev.call('queue.Queue.SetQueue', [paths, 0, false], 15000);
await ev.call(
'queue.Queue.SetQueue',
// The fourth argument is the queue's source; these are
// ad-hoc tracks, so it is the empty one. v3 rejects a
// call with the wrong argument count where v2 filled the
// gap with a zero value.
[paths, 0, false, { type: '', id: 0, label: '' }],
15000,
);
await ev.call('queue.Queue.PlayIndex', [0], 15000);
// Settle mid-track before starting to record. Starting playback
@@ -1046,14 +1043,15 @@ const SCROLL_SETTLE_MS = 260;
async function measureScroll(page) {
// -- M3: the track list, with the Art column staged on. --
const priorColumns = await page.evaluate(async () => {
const prior = await window.go.config.Config.GetTrackListColumns();
const ev = window.__yjEvents;
const prior = await ev.call('config.Config.GetTrackListColumns', [], 15000);
await window.go.config.Config.SetTrackListColumns(
await ev.call('config.Config.SetTrackListColumns', [
[{ id: 'albumArt' }, { id: 'trackName' },
{ id: 'artistName' }, { id: 'trackLength' }],
);
], 15000);
return prior.map((c) => ({ id: c.id }));
return (prior ?? []).map((c) => ({ id: c.id }));
});
const scrollView = async (view, tag) => {
@@ -1141,7 +1139,9 @@ async function measureScroll(page) {
const artists = await scrollView('artists', 'artists-view');
await page.evaluate(
(cols) => window.go.config.Config.SetTrackListColumns(cols),
(cols) => window.__yjEvents.call(
'config.Config.SetTrackListColumns', [cols], 15000,
),
priorColumns,
);
@@ -1605,7 +1605,15 @@ async function measurePlayerBarPass(page) {
if (paths.length < 2) return { error: 'library too small to measure' };
await ev.call('queue.Queue.SetQueue', [paths, 0, false], 15000);
await ev.call(
'queue.Queue.SetQueue',
// The fourth argument is the queue's source; these are
// ad-hoc tracks, so it is the empty one. v3 rejects a
// call with the wrong argument count where v2 filled the
// gap with a zero value.
[paths, 0, false, { type: '', id: 0, label: '' }],
15000,
);
await ev.call('queue.Queue.PlayIndex', [0], 15000);
await ev.call('player.Player.Pause', [], 5000).catch(() => {});
await new Promise((r) => setTimeout(r, 600));
@@ -1804,7 +1812,9 @@ async function run(label) {
const browser = await chromium.launch();
const context = await browser.newContext({ viewport: { width: 1440, height: 900 } });
await context.addInitScript({ path: BRIDGE });
await context.addInitScript(`(${INSTRUMENT})()`);
await context.addInitScript(
`(${INSTRUMENT})(${JSON.stringify(METHOD_NAMES)})`,
);
const page = await context.newPage();
const client = await context.newCDPSession(page);
@@ -1813,8 +1823,10 @@ async function run(label) {
const t0 = Date.now();
await page.goto(BASE_URL, { waitUntil: 'load' });
await page.evaluate(() => window.__yjEvents.ready(30000));
// Wrapping runs before `window.go` exists; re-run now that it does.
await page.evaluate(`(${INSTRUMENT})()`);
// No second instrumentation pass. The old one existed because
// wrapping had to happen after `window.go` appeared, yet the long
// task observer had to start before it; the bridge now records every
// binding call from the initScript onward, so one pass does both.
const report = {
label,
+2 -2
View File
@@ -1,8 +1,8 @@
import { defineConfig, devices } from '@playwright/test';
/**
* These specs drive the *real* application: the Wails dev server on
* :34115 serves the real frontend with real bindings on `window.go`,
* These specs drive the *real* application: Wails v3's server mode on
* :34115 serves the real frontend with the real generated bindings,
* bridged to the same Go backend a desktop window would use. Nothing
* here is mocked.
*
+8 -10
View File
@@ -16,6 +16,14 @@ import type { Page } from '@playwright/test';
* none, so Play was wired, labelled correctly, clicked cleanly and
* queued **nothing**. Every component test still passed.
*/
// There is no test for the tracklist legend, and there should not be:
// `dcc40b1` inverted the mark — rows *not* in the library are dimmed in
// place and nothing marks the ones that are — and deleted
// `.tracklist-legend` with it. The spec asserting it survived that
// commit and has been failing on main ever since. What replaced it
// (the dimming, and the `aria-disabled` that carries it to anyone not
// seeing the page) is covered at the component tier, in
// frontend/test/components/album-actions.test.ts.
test.describe('playing an album from its page', () => {
test.beforeEach(async ({ app }) => {
await openFirstAlbum(app);
@@ -55,16 +63,6 @@ test.describe('playing an album from its page', () => {
await expect.poll(() => queueLength(app)).toBe(before * 2);
});
test('the ticks against the tracks have a legend', async ({ app }) => {
// `H-13` calls them unexplained. They were never *unlabelled* — the
// indicator has carried a title and an aria-label reading
// "Track “X” is in your library" all along — but a sighted user
// scanning the page got a column of green circles and no key.
await expect(
app.locator('explore-album-details').locator('.tracklist-legend'),
).toContainText('in your library');
});
test('the ticks are badges, not keyboard stops', async ({ app }) => {
// Every one of them was a <button> whose click handler was a
// stopPropagation() and a comment saying to wire up the download
+25 -10
View File
@@ -15,12 +15,18 @@ import {
*/
test.describe('harness', () => {
test('the app is the real app, not a mock', async ({ app }) => {
// All 11 bound services land on window.go through the dev server.
const services = await app.evaluate(() => Object.keys(window.go));
// v2 landed every bound service on `window.go`, so "is this real"
// could be asked of an object. v3 has no such global — the
// bindings are ordinary bundled modules — so the question is asked
// of the runtime instead, which is a better question anyway: the
// real runtime is loaded, and it answers for real methods and
// refuses invented ones.
const runtime = await app.evaluate(() => ({
dispatch: typeof window._wails?.dispatchWailsEvent,
client: typeof window._wails?.clientId,
}));
expect(services).toEqual(
expect.arrayContaining(['queue', 'player', 'library', 'explore']),
);
expect(runtime).toEqual({ dispatch: 'function', client: 'string' });
const state = await callBinding<{ tracks: unknown[] }>(
app,
@@ -28,6 +34,12 @@ test.describe('harness', () => {
);
expect(state).toHaveProperty('tracks');
const unknown = await app
.evaluate(() => window.__yjEvents.call('queue.Queue.Nope', [], 2_000))
.catch((err: Error) => err.message);
expect(unknown).toContain('unknown bound method');
});
test('backend events are recorded, in order, with payloads', async ({
@@ -58,10 +70,12 @@ test.describe('harness', () => {
});
test('a binding called with wrong types fails fast', async ({ app }) => {
// player.UserVolume is an int. Passing a float makes the backend
// log "error parsing arguments" and never fire the callback; without
// a timeout the promise never settles and the spec hangs until the
// suite gives up.
// player.UserVolume is an int. Under v2 a float made the backend
// log "error parsing arguments" and never fire the callback, so
// this asserted that the harness's own timeout fired — the failure
// was visible only because the harness invented a deadline. v3
// answers 422 with a TypeError naming the argument, so the
// assertion is now on the backend's own words.
const failure = await app.evaluate(async () => {
try {
await window.__yjEvents.call(
@@ -76,7 +90,8 @@ test.describe('harness', () => {
}
});
expect(failure).toContain('did not settle');
expect(failure).toContain('TypeError');
expect(failure).toContain('player.UserVolume');
});
test('the control surface is mounted and seeded', async ({ testctl }) => {
+25 -25
View File
@@ -1,4 +1,13 @@
import { test, expect, resetEvents, callBinding } from '../support/fixtures.js';
import type { Page } from '@playwright/test';
import {
test,
expect,
resetEvents,
callBinding,
bindingCalls,
NO_QUEUE_SOURCE,
} from '../support/fixtures.js';
/**
* Finishing a track is cheap, and does not disturb the user.
@@ -19,23 +28,17 @@ import { test, expect, resetEvents, callBinding } from '../support/fixtures.js';
/** Long enough for a fixture track (26 s) to finish by itself. */
const FINISH_TIMEOUT = 60_000;
/** Instrument the library bindings so "was anything refetched" is a fact. */
const COUNT_LIBRARY_CALLS = `(() => {
const w = window;
if (w.__yjCalls) { w.__yjCalls.length = 0; return; }
w.__yjCalls = [];
const lib = w.go.library.Library;
for (const key of Object.keys(lib)) {
const fn = lib[key];
if (typeof fn !== 'function' || fn.__counted) continue;
const wrapped = function (...args) {
w.__yjCalls.push(key);
return fn.apply(this, args);
};
wrapped.__counted = true;
lib[key] = wrapped;
}
})()`;
/**
* "Was anything refetched" is a fact, not an inference.
*
* This used to wrap every method on `window.go.library.Library` in
* place. v3 has no such object, and does not need one: the harness
* bridge records every binding call off the single POST they all go
* through, so the question is answered by reading that log rather than
* by instrumenting a target first. `resetEvents` clears it.
*/
const libraryCalls = async (app: Page): Promise<string[]> =>
(await bindingCalls(app)).filter((c) => c.startsWith('library.Library.'));
/**
* Select rows by dispatching on the row rather than clicking it.
@@ -82,8 +85,7 @@ test.describe('a finished track', () => {
expect(paths.length).toBeGreaterThanOrEqual(2);
await callBinding(app, 'queue.Queue.SetQueue', [paths, 0, false]);
await app.evaluate(COUNT_LIBRARY_CALLS);
await callBinding(app, 'queue.Queue.SetQueue', [paths, 0, false, NO_QUEUE_SOURCE]);
await resetEvents(app);
await callBinding(app, 'queue.Queue.PlayIndex', [0]);
@@ -111,12 +113,10 @@ test.describe('a finished track', () => {
'a play emitted the retag event, which invalidates every cache',
).toBe(0);
const refetched = await app.evaluate(
() => (window as unknown as { __yjCalls: string[] }).__yjCalls,
);
const refetched = await libraryCalls(app);
expect(
refetched.filter((c) => c.startsWith('GetAll')),
refetched.filter((c) => c.startsWith('library.Library.GetAll')),
'a play refetched a collection',
).toEqual([]);
@@ -136,7 +136,7 @@ test.describe('a finished track', () => {
const paths = await app.evaluate(firstPaths, 2);
await callBinding(app, 'queue.Queue.SetQueue', [paths, 0, false]);
await callBinding(app, 'queue.Queue.SetQueue', [paths, 0, false, NO_QUEUE_SOURCE]);
await resetEvents(app);
await callBinding(app, 'queue.Queue.PlayIndex', [0]);
+3 -1
View File
@@ -7,6 +7,7 @@ import {
resetEvents,
waitForEvent,
LONG_TRACK,
NO_QUEUE_SOURCE,
} from '../support/fixtures.js';
import type { Page } from '@playwright/test';
@@ -161,6 +162,7 @@ test.describe('a finished queue keeps its context', () => {
[rows[0].file_path],
0,
false,
NO_QUEUE_SOURCE,
]);
await waitForEvent(app, 'QueueChanged');
await callBinding(app, 'queue.Queue.Play');
@@ -199,7 +201,7 @@ test.describe('a track that will not play says so', () => {
try {
await callBinding(app, 'queue.Queue.Clear');
await resetEvents(app);
await callBinding(app, 'queue.Queue.SetQueue', [paths, 0, false]);
await callBinding(app, 'queue.Queue.SetQueue', [paths, 0, false, NO_QUEUE_SOURCE]);
await waitForEvent(app, 'QueueChanged');
await callBinding(app, 'queue.Queue.Play');
+7 -2
View File
@@ -1,4 +1,9 @@
import { test, expect, callBinding } from '../support/fixtures.js';
import {
test,
expect,
callBinding,
NO_QUEUE_SOURCE,
} from '../support/fixtures.js';
import type { Page } from '@playwright/test';
/**
@@ -35,7 +40,7 @@ async function queueFourAndOpen(app: Page): Promise<string[]> {
return (tracks as { FilePath: string }[]).slice(0, 4).map((t) => t.FilePath);
});
await callBinding(app, 'queue.Queue.SetQueue', [paths, 0, false]);
await callBinding(app, 'queue.Queue.SetQueue', [paths, 0, false, NO_QUEUE_SOURCE]);
// A closed panel renders no list at all, so there is no row to focus.
await app.locator('#queue-button').click();
+8 -2
View File
@@ -1,4 +1,10 @@
import { test, expect, callBinding, waitForEvent } from '../support/fixtures.js';
import {
test,
expect,
callBinding,
waitForEvent,
NO_QUEUE_SOURCE,
} from '../support/fixtures.js';
import type { Page } from '@playwright/test';
/**
@@ -64,7 +70,7 @@ async function playTheLongOne(app: Page): Promise<void> {
expect(paths.length).toBeGreaterThan(0);
await callBinding(app, 'queue.Queue.SetQueue', [paths, 0, false]);
await callBinding(app, 'queue.Queue.SetQueue', [paths, 0, false, NO_QUEUE_SOURCE]);
await waitForEvent(app, 'TrackChanged');
// The scroll cycle is armed 1500 ms after the geometry is measured,
+33 -23
View File
@@ -199,39 +199,49 @@ function addRequest(
}
/**
* Drop any request for this album, through the raw binding.
* Drop any request for this album, over the runtime endpoint.
*
* Deliberately not `callBinding`: that goes through `window.__yjEvents`,
* which only exists on a page the `app` fixture created — a bare
* `browser.newPage()` has no init script, so the bridge is undefined and
* the cleanup throws where nobody is looking. The first version of this
* did exactly that and left the request behind, which failed the *next*
* run of this same spec.
* Deliberately not `callBinding`: that goes through
* `window.__yjEvents`, which only exists on a page the `app` fixture
* created — a bare `browser.newPage()` has no init script, so the
* bridge is undefined and the cleanup throws where nobody is looking.
* The first version of this did exactly that and left the request
* behind, which failed the *next* run of this same spec.
*
* v2's answer was `window.go`, which every page had. v3 has no such
* global, and the version of this that kept reading it did not throw —
* it returned early on `if (!svc)`, which is the same silent cleanup
* with a different cause. A POST to `/wails/runtime` needs neither:
* it is the same request the bundle makes, and any page can make it.
*/
async function clearRequest(page: import('@playwright/test').Page) {
await page.evaluate(async (mbid) => {
const go = (
window as unknown as {
go?: {
download?: {
Service?: {
ListRequests(): Promise<{ id: number; mbid: string }[]>;
RemoveRequest(id: number): Promise<void>;
};
};
};
}
).go;
const call = async (method: string, args: unknown[]) => {
const res = await fetch('/wails/runtime', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
object: 0,
method: 0,
args: {
'call-id': `clear-${method}`,
methodName: `yellowjacket/backend/${method}`,
args,
},
}),
});
const svc = go?.download?.Service;
if (!res.ok) throw new Error(`${method}: ${await res.text()}`);
if (!svc) return;
return res.json();
};
const rows = (await svc.ListRequests()) ?? [];
const rows: { id: number; mbid: string }[] =
(await call('download.Service.ListRequests', [])) ?? [];
for (const row of rows) {
if (row.mbid?.toLowerCase() === mbid.toLowerCase()) {
await svc.RemoveRequest(row.id);
await call('download.Service.RemoveRequest', [row.id]);
}
}
}, MBID);
+49 -6
View File
@@ -3,12 +3,26 @@ import { dirname, resolve } from 'node:path';
import { test as base, expect, type Page } from '@playwright/test';
import { nameOf } from './method-ids.mjs';
const here = dirname(fileURLToPath(import.meta.url));
/** The same bridge `playwright-cli` loads, so an exploratory session and
* a committed spec see an identical page. */
const INIT_SCRIPT = resolve(here, '../../.playwright/init-events.js');
/**
* The fourth argument to `queue.Queue.SetQueue`, for a queue with no
* single source — the whole library, or a handful of ad-hoc tracks,
* which is what every spec here builds.
*
* It is passed explicitly because v3 rejects a call with the wrong
* argument count (`expects 4 arguments, got 3`) where v2 accepted one
* and filled the gap with a zero value. The specs had been three-arg
* since the parameter was added; nothing said so.
*/
export const NO_QUEUE_SOURCE = { type: '', id: 0, label: '' };
/**
* The 90-second fixture track (`cmd/gentestdata`, case `edge-lengths`).
*
@@ -61,12 +75,13 @@ export async function eventNames(
}
/**
* Call a bound Go method with a timeout.
* Call a bound Go method by name, over the runtime's own endpoint.
*
* Wrong argument types make the backend log "error parsing arguments"
* and never fire the callback, so an unguarded call hangs until the
* whole spec times out with no clue why. This fails in seconds and
* says where to look.
* v3 rejects a bad call rather than never firing its callback the way
* v2 did: a wrong argument type comes back as a TypeError naming the
* argument, a wrong count as `expects 4 arguments, got 3`, an unknown
* method as a ReferenceError. The timeout is a backstop for a hung
* request, not the mechanism that makes a mistake visible.
*/
export async function callBinding<T = unknown>(
page: Page,
@@ -81,6 +96,21 @@ export async function callBinding<T = unknown>(
) as Promise<T>;
}
/**
* The binding calls the *app* made, newest last, as `pkg.Type.Method`.
*
* This is what replaces v2's trick of wrapping `window.go` in place:
* `.playwright/init-events.js` records every call off the one POST v3
* routes them all through, and `method-ids.ts` names them from the
* generated tree. It sees calls from any module and cannot miss one
* made before a wrapper was installed.
*/
export async function bindingCalls(page: Page): Promise<string[]> {
const calls = await page.evaluate(() => window.__yjEvents.bindings);
return calls.map(nameOf);
}
/** Thin client for the dev-only /__test/ surface (backend/testctl). */
export class TestCtl {
constructor(private readonly baseURL: string) {}
@@ -166,7 +196,20 @@ declare global {
): Promise<YjEvent>;
ready(timeoutMs?: number): Promise<boolean>;
call(path: string, args?: unknown[], timeoutMs?: number): Promise<any>;
/** Every binding call the app itself made; see init-events.js. */
bindings: {
methodID: number | null;
methodName: string | null;
start: number;
ms: number;
bytes: number;
}[];
measureBytes: boolean;
};
/** The v3 runtime's own namespace, installed by @wailsio/runtime. */
_wails?: {
dispatchWailsEvent?: (event: unknown) => void;
clientId?: string;
};
go: Record<string, Record<string, Record<string, (...a: any[]) => any>>>;
}
}
+110
View File
@@ -0,0 +1,110 @@
import { readFileSync, readdirSync, statSync } from 'node:fs';
import { fileURLToPath } from 'node:url';
import { dirname, join, resolve } from 'node:path';
/**
* Turns the method ids a binding call carries back into the names a
* spec (or a measurement) wants to read.
*
* Plain JavaScript, not TypeScript, because `e2e/perf/measure.mjs` runs
* under bare `node` and both it and the specs need exactly this map —
* and one derivation with a `.mjs` extension is better than two that
* can disagree.
*
* v2 put every bound method on `window.go`, so a harness could walk
* that object to wrap or enumerate them. v3 has no such surface: the
* generated bindings are ordinary bundled modules calling
* `$Call.ByID(<fnv hash of the fully-qualified name>)`, and what
* reaches the wire — and therefore what `.playwright/init-events.js`
* can record — is the number.
*
* Plan 009 phase 6b named two ways to get the list back. This is the
* preferred one: derive it from `frontend/bindings/`, which is a real
* generated tree and is already gated by `make bindings-check`. The
* alternative — a hand-maintained list — goes stale silently, which is
* the failure mode that check exists to prevent.
*
* Nothing here hashes anything. The generated source carries the id as
* a literal beside the function that sends it, so this reads the two
* together rather than recomputing one from the other and hoping the
* hash still matches.
*/
const here = dirname(fileURLToPath(import.meta.url));
const BINDINGS_ROOT = resolve(here, '../../frontend/bindings/yellowjacket');
/** `export function Name(…) { return $Call.ByID(123, …) }` */
const BOUND_METHOD = /export function (\w+)\([\s\S]*?\$Call\.ByID\((\d+)/g;
/** `import * as Library from "./library.js";` — the only place the Go
* type's casing survives; the file is `library.ts`, and
* `frontendutil.ts` cannot tell you it is `FrontendUtil`. */
const SERVICE_EXPORT = /import \* as (\w+) from "\.\/([\w.]+)\.js"/g;
function* walk(dir) {
for (const entry of readdirSync(dir)) {
const path = join(dir, entry);
if (statSync(path).isDirectory()) yield* walk(path);
else if (entry.endsWith('.ts')) yield path;
}
}
let cached;
/**
* methodIDs maps a binding's method id to `pkg.Type.Method` — the same
* short path `callBinding` takes, so a spec never has to know that the
* backend calls it `yellowjacket/backend/queue.Queue.GetState`.
*/
export function methodIDs() {
if (cached) return cached;
const byID = new Map();
for (const file of walk(BINDINGS_ROOT)) {
if (!file.endsWith('index.ts')) continue;
const dir = dirname(file);
const pkg = dir.split('/').pop() ?? '';
const index = readFileSync(file, 'utf8');
for (const [, typeName, base] of index.matchAll(SERVICE_EXPORT)) {
let source;
try {
source = readFileSync(join(dir, `${base}.ts`), 'utf8');
} catch {
continue; // a models-only re-export, which binds nothing
}
for (const [, method, id] of source.matchAll(BOUND_METHOD)) {
byID.set(Number(id), `${pkg}.${typeName}.${method}`);
}
}
}
if (byID.size === 0) {
throw new Error(
`method-ids: no bound methods under ${BINDINGS_ROOT}; ` +
`run 'make bindings'`,
);
}
cached = byID;
return byID;
}
/** Names one recorded binding call, or `#<id>` if the map has no entry
* — which fails the assertion naming it rather than passing quietly. */
export function nameOf(call) {
if (call.methodName) {
return call.methodName.replace(/^yellowjacket\/backend\//, '');
}
return call.methodID === null
? '#unknown'
: (methodIDs().get(call.methodID) ?? `#${call.methodID}`);
}
+8 -2
View File
@@ -8,7 +8,13 @@
"noEmit": true,
"skipLibCheck": true,
"types": ["node"],
"allowImportingTsExtensions": true
"allowImportingTsExtensions": true,
"allowJs": true
},
"include": ["specs/**/*.ts", "support/**/*.ts", "playwright.config.ts"]
"include": [
"specs/**/*.ts",
"support/**/*.ts",
"support/**/*.mjs",
"playwright.config.ts"
]
}
-10
View File
@@ -81,16 +81,6 @@ func main() {
OnShutdown: yjApp.OnShutdown,
})
// The services' own ServiceStartup hooks have run by the time this
// fires; what is left is the wiring between them.
app.Event.OnApplicationEvent(
events.Common.ApplicationStarted,
func(*application.ApplicationEvent) {
yjApp.OnStartup(app.Context())
yjApp.OnDomReady(app.Context())
},
)
window := app.Window.NewWithOptions(application.WebviewWindowOptions{
Title: "yellowjacket",
Width: winCfg.Width,
+30 -23
View File
@@ -2,28 +2,36 @@
#
# Start YellowJacket headless, in the background, and return.
#
# `wails dev` binds an HTTP + WebSocket dev server on :34115 that serves
# the real frontend with the real generated bindings on `window.go` and
# bridges every call and every runtime.EventsEmit to the *same* Go
# backend the desktop window uses. A browser pointed at it is not a
# mock. That is what makes this app drivable by a coding agent.
# Wails v3's server mode binds an HTTP server on :34115 that serves the
# real frontend with the real generated bindings and bridges every call
# and every event to the *same* Go backend a desktop window would use.
# A browser pointed at it is not a mock. That is what makes this app
# drivable by a coding agent.
#
# Three details are load-bearing:
#
# * We run the dev *binary*, not `wails dev`. app_dev.go parses
# -devserver / -assetdir / -loglevel straight from os.Args, so a
# `go build -tags "dev"` binary serves the identical
# devserver with no file watcher, no rebuild supervisor and no
# reload broadcast: one process, one PID, deterministic startup.
# * We build with `-tags dev,server` and run the binary. This is a
# first-class Wails mode ("a pure HTTP server without native GUI
# dependencies"), not something hand-rolled: v2 had no such thing,
# so this script used to run a dev binary whose app_dev.go parsed
# -devserver / -assetdir out of os.Args. That file is gone with v2.
# The port comes from WAILS_SERVER_PORT.
#
# * Xvfb is not optional. devserver.Run ends in Frontend.Run(ctx),
# which opens the GTK window and blocks; no flag suppresses it.
# * Xvfb is gone, and that is the point of server mode. v2's
# devserver.Run ended in Frontend.Run(ctx), which opened the GTK
# window and blocked with no flag to suppress it; server mode opens
# no window, so there is no display to fake.
#
# * dbus-run-session is not incidental. A private session bus makes
# backend/mediacontrols register MPRIS for real, so it becomes
# assertable with busctl. It replaces the bus, not /run/user, so
# PulseAudio still works and InitSpeaker succeeds.
#
# * The frontend is *embedded*, not served from disk. main.go has
# //go:embed all:frontend/dist, so a frontend change needs the
# rebuild this script does anyway; there is no -assetdir to point
# at a live directory.
#
# Usage:
# scripts/dev-headless.sh [--seed NAME|--fresh] [--port N] [--no-build]
#
@@ -47,7 +55,7 @@ LOG_LEVEL="${YJ_LOG_LEVEL:-debug}"
STARTUP_TIMEOUT=60
usage() {
sed -n '3,28p' "$0" | sed 's/^# \{0,1\}//'
sed -n '3,38p' "$0" | sed 's/^# \{0,1\}//'
exit "${1:-0}"
}
@@ -129,9 +137,9 @@ echo "$YJ_HOME" >"$HOME_FILE"
# ── Build ────────────────────────────────────────────────────────────
if [ "$BUILD" = 1 ]; then
echo "dev-headless: building frontend + dev binary..."
echo "dev-headless: building frontend + dev server binary..."
(cd frontend && pnpm install --silent && pnpm build >/dev/null)
go build -tags "dev" -o "$BIN" .
go build -tags "dev,server" -o "$BIN" .
fi
if [ ! -x "$BIN" ]; then
@@ -141,7 +149,7 @@ fi
# ── Launch ───────────────────────────────────────────────────────────
# setsid puts the app in its own process group so dev-stop can kill the
# whole tree (xvfb-run, dbus-daemon, the app) by group id. Never
# whole tree (dbus-daemon, the app) by group id. Never
# `pkill -f`: the pattern matches the invoking shell's own command line
# and silently drops the rest of the chain.
: >"$LOG_FILE"
@@ -150,11 +158,9 @@ fi
# rather than implied by the dev build so that a human's `make dev` does
# not carry an arbitrary-SQL endpoint on a listening port.
YJ_TESTCTL=1 \
YJ_LOG_LEVEL="$LOG_LEVEL" setsid dbus-run-session -- xvfb-run -a \
WAILS_SERVER_PORT="$PORT" \
YJ_LOG_LEVEL="$LOG_LEVEL" setsid dbus-run-session -- \
"$BIN" \
-devserver "localhost:$PORT" \
-assetdir "$REPO_ROOT/frontend/dist" \
-loglevel Debug \
>>"$LOG_FILE" 2>&1 &
APP_PID=$!
@@ -191,7 +197,8 @@ Drive it with:
playwright-cli -s=yj snapshot
playwright-cli -s=yj eval "() => window.__yjEvents.call('queue.Queue.GetState', [], 5000)"
A binding call that never settles means wrong argument types: the
backend logs 'error parsing arguments' and never fires the callback.
The app log is the only place that shows up, so always use a timeout.
A bad binding call now rejects rather than hanging: v3 answers wrong
argument types with a TypeError naming the argument and an unknown
method with a ReferenceError. The timeout in __yjEvents.call is a
backstop for a hung request, not how a mistake becomes visible.
EOF
+26 -27
View File
@@ -36,7 +36,6 @@ MANIFEST="$REPO_ROOT/test_data/music_library_test.manifest.json"
NAME="default"
PORT=34115
SESSION="yj-seed"
BUILD_ARGS=()
# Replaced below once the manifest says how many tracks are coming: a
# 50 000-track scan is minutes, and a fixed 180 s deadline would abort
@@ -74,7 +73,7 @@ need() {
exit 1
}
}
need playwright-cli
need curl
need jq
if [ ! -f "$MANIFEST" ]; then
@@ -89,7 +88,6 @@ SCAN_TIMEOUT=$((180 + WANT_TRACKS / 50))
FIXTURE_HASH="$(jq -r .hash "$MANIFEST")"
cleanup() {
playwright-cli -s="$SESSION" close >/dev/null 2>&1 || true
./scripts/dev-stop.sh >/dev/null 2>&1 || true
}
trap cleanup EXIT
@@ -108,24 +106,34 @@ YJ_CORE_INDEX_URL="http://127.0.0.1:1/none.tar.zst" \
YJ_HOME="$(cat "$RUN_DIR/app.home")"
playwright-cli -s="$SESSION" open "http://localhost:$PORT" >/dev/null
# Every binding call gets a timeout. A call with wrong argument types
# makes the backend log 'error parsing arguments' and never fire the
# callback, so the in-page promise never settles and a naive await
# hangs forever.
# A binding is called over the runtime's own HTTP endpoint, by method
# name — the same request the bundle makes, minus the browser.
#
# This used to drive a real page through playwright-cli, because v2's
# only way in was `window.go`. v3 answers the same call over HTTP, so a
# browser (and a global npm install of the CLI, in CI) buys nothing
# here: seeding needs the *app* to run and the *real* scanner to finish,
# which it still does. It also fails properly now — v3 answers a bad
# argument with a 422 and a TypeError naming it, where v2 logged
# "error parsing arguments" and never fired the callback.
call() {
playwright-cli -s="$SESSION" eval "async () => {
const timeout = new Promise((_, reject) =>
setTimeout(() => reject(new Error('binding timeout')), 15000));
return await Promise.race([(async () => { $1 })(), timeout]);
}"
local method="$1"
local args="${2:-[]}"
local body
body="$(jq -nc --arg m "yellowjacket/backend/$method" --argjson a "$args" \
'{object: 0, method: 0, args: {"call-id": "seed", methodName: $m, args: $a}}')"
curl -sS --fail-with-body --max-time 30 \
-X POST "http://localhost:$PORT/wails/runtime" \
-H 'Content-Type: application/json' \
-d "$body"
}
echo "seed-sandbox: registering library $LIBRARY_DIR"
if ! call "return await window.go.library.Library.AddLibrary(
${LIBRARY_DIR@Q});" >/dev/null; then
if ! call library.Library.AddLibrary \
"$(jq -nc --arg d "$LIBRARY_DIR" '[$d]')" >/dev/null; then
echo "seed-sandbox: AddLibrary failed; app log:" >&2
tail -n 40 "$LOG_FILE" >&2
exit 1
@@ -137,19 +145,12 @@ fi
# a fixture is not being ingested and the seed is wrong.
echo "seed-sandbox: waiting for the scan to reach $WANT_TRACKS tracks"
# The result is tagged rather than scraped for bare digits:
# playwright-cli echoes the evaluated source back, and that source
# contains numbers of its own (the binding timeout, for one).
deadline=$((SECONDS + SCAN_TIMEOUT))
got=0
while [ "$SECONDS" -lt "$deadline" ]; do
got="$(call "const libs =
await window.go.library.Library.GetAllLibrariesWithTrackCounts();
const total = (libs ?? []).reduce(
(n, l) => n + (l.trackCount ?? 0), 0);
return 'YJTRACKS' + '=' + total;" |
grep -oE 'YJTRACKS=[0-9]+' | head -n 1 | cut -d= -f2)"
got="$(call library.Library.GetAllLibrariesWithTrackCounts |
jq '[(. // [])[].trackCount // 0] | add // 0')"
got="${got:-0}"
[ "$got" = "$WANT_TRACKS" ] && break
@@ -170,8 +171,6 @@ if [ "$got" != "$WANT_TRACKS" ]; then
exit 1
fi
playwright-cli -s="$SESSION" close >/dev/null 2>&1 || true
# SIGTERM, so OnBeforeClose / OnShutdown persist window, player and
# queue state. A seed built from a killed app is missing exactly the
# state those hooks write.