mirror of
https://github.com/4gray/iptvnator.git
synced 2026-10-08 09:01:03 -08:00
6e229f85b699c5223ff77c3cdebb6369274c2166
16
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
|
6e229f85b6 |
test(perf): add the J3 playback journey (#1774)
* test(perf): add the J3 playback journey J3 clicks a live channel of an Xtream portal and ends at the built-in HTML5 player's first `playing` event, with `loadedmetadata` as a secondary phase. It follows J2: every iteration is a fresh J1 launch on a copy of a profile seeded through the app's dialogs, and the click happens after the app has settled in the portal's first live category. The portal is the mock's `live-fallback` account, whose `.ts` live URLs serve the local H.264/AAC MPEG-TS fixture that mpegts.js plays through MSE on every platform. The marketing accounts' local live bytes are zero-filled and never reach `playing`. Seeding selects the HTML5 player and the `ts` stream format; the catalog's picsum.photos logos are cancelled from the test side so no request leaves the machine. Counters: renderer.ipcCallsToPlaying, renderer.httpRequestsToPlaying, renderer.domMutationsToPlaying, renderer.layoutShiftScore and renderer.longTasks; wall-clock click->loadedmetadata and click->playing. renderer.ipcSerialDepthToPlaying is listed as unavailable until the serial-depth helper lands. No baseline yet. The probe gains a media-event terminal; J2's pre-click settle moves to journey-click-settle.ts so both journeys share it unchanged, and the probe spec's jsdom fixtures move to a shared test helper. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): show J3's HTTP boundary margins and watch 1 s after playing Review follow-up. renderer.httpRequestsToPlaying compares the ledger's arrival stamps with the renderer's click and playing stamps, which come from different processes on the same host clock. Each iteration now records the distance of the nearest request on either side of both boundaries, so a count a clock difference could flip is visible. Requests after playing were a single snapshot taken right after the probe; the test now watches the ledger for a fixed 1 s after playing. A live stream never leaves the mock quiet, so J2's quiet wait does not apply. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: 4gray <fourgray@proton.me> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
4gray
|
ec8b931dbf | ci(perf): add the weekly baseline tightening workflow (#1760) | ||
|
|
58c54a3286 |
test(perf): add a settled layout-shift counter to J1 (#1756)
* test(perf): add a settled layout-shift counter to the J1 launch journey renderer.layoutShiftScore stops at the first-card cutoff, so the dashboard shift fixed in #1738 (about 0.23, roughly 15 ms after the first card) read as 0. renderer.layoutShiftScoreSettled sums the same non-input layout-shift entries until the workspace content has been quiet for 500 ms, capped at 3 s after the cutoff. The first-card counter is unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): attribute late layout shifts and record the first measurement evidence.settle.lateShifts lists the counted shifts after the first-card cutoff with the nodes the browser attributes them to. On master the settled score is 0.236: the recent-sources rail moves up 316 px about 12 ms after the first card and back down shortly after, a flicker #1738 did not cover. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): end the settle window at its scheduled deadline A cap or quiet timer delayed by a busy main thread used the moment it ran as the settle point, so shifts after the 3 s cap could enter the settled counter. The settle point is now the deadline the timer was scheduled for. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(perf): record the runner's settled layout-shift measurement Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: 4gray <fourgray@proton.me> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
|
28b022ff48 |
test(perf): add the J2 open-source journey (#1730)
* test(perf): add the J2 open-source journey Measure the click on the Xtream portal card until the category list and the first page of the opened section are painted, in the same fresh process as J1 after its counters are final and the app has settled. - journey-renderer-probe: optional click start (capture-phase listener on window, start sentinel before the app sees the click, entries before the click dropped), companion selectors, recent-input layout shifts tallied - journey-main-ipc-capture: optional start sentinel; counts calls between the two sentinels - journey-mock-request-ledger: loopback proxy that counts every request the app sends to the mock without storing credentials - open-source-journey-record: J2 counters and evidence - journey-run / journey-summary: every journey spec of one perf:journeys run adds its entry to the same summary.json - docs: J2 contract in performance-journeys.md Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): stop echoing the request URL from the ledger spec's upstream CodeQL flagged the fake upstream as reflected XSS. It now records what it received server-side and answers with a fixed text/plain body. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): address J2 review findings - Sentinels use cancelSourceProbe: the preload traces the call before forwarding it and SOURCE_HEALTH_CANCEL is an in-memory map lookup, so a marker no longer runs a SQLite query on the worker ahead of the measured work (Codex P1). A spec pins that handler contract. - A run is started only in the Playwright runner, replacing inherited values, and carries a random harness.runId; summaries from another invocation are never merged (Greptile P1, Codex P2). - The mock ledger tracks in-flight requests; settling and the HTTP window require none in flight (Codex P2). - clickToFirstPagePaintMs reports click to the committed paint next to the terminal-batch clickToFirstPageMs (Codex P1). - The Playwright attachment carries the whole summary (Greptile P2). - jsdom probe specs wait for the post-paint cutoff instead of a fixed 40 ms, which flaked when the harness runs all files in parallel. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(validation): describe perf:journeys as running J1 and J2 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): settle J2 on pending bridge calls and start HTTP at the click - The journey IPC capture pairs every traced start with its success or error and exposes the calls still in flight. J2 settles only when J1's capture, installed before the document loaded, has none pending, so a slow startup call cannot resolve after the click and count as J2. - The mock HTTP window starts at the renderer's click stamp instead of the test-side mark taken before Playwright's actionability checks. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): restart the J2 quiet period when pending work completes Both waits in the open-source journey (settling before the click, closing the mock window after the terminal) now use one waitForJourneyQuiet helper that compares whole samples, in-flight counts included. The poll that first sees a request or bridge call complete restarts the quiet period, so the window is never measured from a poll at which work was still pending. A fake-clock spec covers the in-flight to zero case. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): align J2 with the main-process counters from #1715 After rebasing on #1715, J1 measures main.sqlStatementsBeforeReadyToShow, so J2's reason for listing main.sqlStatementsToFirstPage as unavailable (no countable channel) was stale. State the actual limit: the running total is read from the test process and cannot be bounded at the click or the first-page batch. The performance-journeys CI job comment now names both journeys. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): launch J2 without SQL counting and stamp mock requests in sub-ms - runLaunchJourney takes the launch instrumentation; only J1 turns on the main-process counters and IPTVNATOR_PERF_COUNT_SQL, so J2's click is not measured under the hook that wraps every SQLite statement. The flags are built in journey-launch-environment.ts, which the SQL opt-in guard now expects, and a launch record without main counters is rejected. - The mock ledger stamps arrivals with performance.timeOrigin + performance.now(), the same sub-millisecond epoch as the renderer's click, so a request later in the click's millisecond is not counted before it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): reject J2 iterations with activity after settling - The open-source record compares the settle snapshot with what the probe and the IPC capture counted up to the click event, and with the mock requests between the snapshot and the click stamp. Any change means background work began during Playwright's actionability checks and could land in J2, so the iteration is rejected. - The SQL opt-in guard also checks who passes mainCounters: true: only measureLaunchJourney may, and J2 must pass false. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): count the long task that dispatches the J2 click A long task's startTime precedes the click event's timestamp when the listener runs inside it, so the start-time filter dropped the task that performs the interaction. Long tasks now count when their range overlaps the window: on one main thread only the dispatching task can overlap the click. Layout shifts keep the start-time filter. J1 is unchanged (its window starts at -Infinity). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): bound J2's late-request check by the quiet sample's mark The late-activity check compared requests against a fresh ledger mark taken after waitForQuiet returned. A request that arrived while the final quiet sample was still reading the IPC capture advanced that mark and escaped the check. The boundary is now the ledger position read by the accepted sample itself, like its DOM and IPC counts. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): end J2's HTTP window at the accepted quiet sample The post-terminal window read the ledger after waitForMockQuiet returned, so a request arriving in between was counted although its completion was never waited for. waitForMockQuiet now returns the ledger position its accepted sample read; later requests are kept as evidence (httpRequestsAfterSettledByRoute) instead of the counter. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): observe late mock requests before reading J2's ledger httpRequestsAfterSettledByRoute read the ledger right after the accepted quiet sample, so late requests had no chance to appear in it. The ledger is now read after another quiet interval; the counter stays bounded by the quiet sample's mark and late traffic shows up in the evidence. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): fail the J2 quiet wait when a sample stalls past its deadline waitForJourneyQuiet accepted a sample that returned unchanged after a stall longer than the timeout as the end of a quiet period, before the deadline check ran. The deadline is now checked first, so a stalled sample fails the wait instead of letting the click go ahead unobserved. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): detach J1's IPC capture before the J2 click J2 used J1's capture to see pending launch bridge calls while settling, but its ipcMain listener stayed attached and ran for every bridge call of the measured click. The capture can now be detached; J2 detaches J1's right after settling, before the click. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(perf): sample both J2 settle captures in one main-process snapshot The settle sample read J1's capture (pending calls) and J2's capture (call count) in two evaluate calls, so a call starting in between was counted with a stale zero in flight and its completion went unseen. Both states are now read in one synchronous pass, where no ipcMain event can be handled in between. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: 4gray <fourgray@proton.me> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
|
d60c80746b |
perf(services): share the startup inventory read (J1) (#1716)
* perf(workspace): share the startup inventory read and defer non-first-card IPC (J1) Journey J1 / renderer.ipcCallsToFirstCard: 12 -> 7. - PlaylistsService.getAllPlaylists() shares one in-flight SQLite read between concurrent callers (the playlist effect and the XMLTV source reconciliation at startup). Settled reads are never reused and every SQLite write detaches the pending read. - getM3uFavoriteChannels() stops re-reading the write-once IndexedDB -> SQLite migration receipt once it has been seen. - StartupDeferralService holds the download list, app update status and the dashboard's recent items and favorites until one task after the render that reveals the routed content (5 s safety timeout). IPTVNATOR_DISABLE_STARTUP_DEFERRAL=1 is the kill switch. - reconcileEpgSources and the two distinct migration-flag reads stay on the critical path: the first is the #1548 revision fence, the second are different keys, not duplicates. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(services): trust a migration receipt written in this session; correct J1 note - Mark the IndexedDB -> SQLite receipt as confirmed after an empty-store receipt write or a committed dbMigrateAppPlaylists, not only when it was already present, so M3U favorites skip the per-playlist re-read on the first launch after an upgrade too. - The release note no longer claims a wall-clock speedup the J1 benchmark did not show. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * perf(services): keep only the shared startup inventory read (J1) Drop the startup deferral gate, its kill switch and the deferred loads: they lowered renderer.ipcCallsToFirstCard but moved neither spawnToFirstCardMs nor load->card beyond drift, and they grew renderer.initialBytes. Keep the shared in-flight inventory read, inline it in PlaylistsService with short property names, and copy the joiner's result with structuredClone (the Electron renderer has it; the services test setup polyfills it for jsdom). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(performance): describe the shared inventory read and the J1 validation result Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(services): drop the migration receipt memo for M3U favorites Skipping the receipt read let the dashboard ask the database for M3U favorite channels before a just-toggled favorite was written: the write waits in the per-playlist queue and the cross-context lock, and the dashboard does not reload when the store's favorites are unchanged. The extra round trip had been masking that race, and the Windows E2E run hit it (epg.e2e.ts "dashboard live rails find a programme that only another playlist's XMLTV carries"). The memo was off the first card's path, so it bought nothing measurable for J1; restore the per-call read. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(services): share only the startup inventory read A pending read was shared with any concurrent caller, but not every playlist write goes through PlaylistsService: the settings reset deletes all playlists through DatabaseService, so a caller could join a read taken before that deletion (Codex review). Share only the first read, which the startup pair (playlist effect and XMLTV reconciliation) needs while the startup screen still hides every writing action; sharing ends when it settles or a PlaylistsService write starts. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: 4gray <fourgray@proton.me> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
4gray
|
de7b19aee2 | docs(performance): record the idle work audit (#1719) | ||
|
4grayandClaude Opus 5.5
|
af44e2368b |
ci(performance): give the initial-bytes ratchet slack and a labelled override (#1744)
* ci(performance): give the initial-bytes ratchet slack and a labelled override The exact renderer.initialBytes counter failed PRs for reasons outside their diff: two concurrent merges left master 108 bytes over the baseline for hours, and bundler identifier renaming moves the counter by hundreds of bytes. PRs growing it by 243 and 302 bytes had no way to pass at all, because the direction check refuses any raised baseline. - Counter entries accept `slack` (integer, entry unit): the ratchet enforces `value + slack`, reports how much slack a measurement uses, and still prints the tighten hint below `value`. renderer.initialBytes gets 4096 bytes, so growth can accumulate at most 4 KiB past the last lowered baseline while regressions such as +35 KB still fail. - check-baseline-direction.mjs compares `value + slack`, treats widened slack like a widened tolerance, and takes `--allow-increase`, which reports weakened entries as ALLOWED instead of failing. - CI passes `--allow-increase` only when the pull request (or, for a master push, the pull request merged as the pushed commit) carries the perf-baseline-increase label, read from the API so a job re-run picks up a label added later. This change widens the slack itself, so its own PR needs the label. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0166PWobUjpoiWt8E9bBdsCs * fix(performance): report slack usage only for entries that have slack A wall-clock measurement above `value` but within `value × toleranceRatio` fell into the slack branch and was reported as using "slack", conflating timing tolerance with counter slack. Only entries with `slack` report it now. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0166PWobUjpoiWt8E9bBdsCs * fix(performance): scope the push-time label and refuse raised counter values - A master push compares the whole push, but the label was read from the head commit's PR only, so one labelled PR could cover another commit's increase in the same push. The label now counts only when the push added exactly one first-parent commit (a squash or merge of one PR); any other push that weakens a baseline fails. - Raising a counter's `value` while narrowing its `slack` lowered the enforced limit and was reported as "lowered". A counter's value is the measured evidence and only moves down, so that raise is now a weakening that needs the label. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0166PWobUjpoiWt8E9bBdsCs * fix(performance): treat a counter/wall-clock type switch as a weakening Turning `{ value: 100, slack: 10 }` into `{ value: 105, toleranceRatio: 1 }` skipped the raised-counter-value rule and was reported as a lowered limit. Switching an entry between counter and wall-clock now needs the perf-baseline-increase label. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0166PWobUjpoiWt8E9bBdsCs * ci(performance): paginate the label lookups of the direction check The labels endpoint returns 30 entries per page by default, so a PR with more labels could miss perf-baseline-increase. Both lookups now request 100 per page and paginate, like the release-note gate. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0166PWobUjpoiWt8E9bBdsCs --------- Co-authored-by: Claude <noreply@anthropic.com> |
||
|
4gray
|
8b570e3390 | test(performance): measure two-byte string cost in M3U and XMLTV parsing (D1) (#1713) | ||
|
|
1d9a563d1a |
test(performance): count startup phases and SQL statements for the J1 launch journey (#1715)
* test(performance): count startup phases and SQL statements for the J1 launch journey
Implements plan item A2. With IPTVNATOR_PERF_CAPTURE=1 the main process
keeps named counters and registers a main-only performance:read-counters
IPC handler; without the flag nothing is counted and the handler does not
exist.
- debug-trace.ts owns the registry; traceStartupPhase replaces the
trace('startup', ...) sites and counts main.startupPhases.
- The database worker counts executed statements through better-sqlite3's
Statement prototype (the verbose callback expands every statement and
made bulk inserts 2-4x slower) and posts the count over its message
port, flushed before every other worker message. The main-thread shared
connection is counted through a new connection observer in the shared
database library.
- The first main window freezes main.modulesRegisteredBeforeWindow at
creation and main.sqlStatementsBeforeReadyToShow at ready-to-show.
- The journey gate drops the ready-to-show that Electron emits for the
about:blank detour, so the app sees the real document's first paint,
and taps the counters handler; the J1 record reads both counters after
the renderer probe completes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(database): require one SQL statement per exec during initialization
The performance capture counts one exec call as one statement, because
SQL cannot be split reliably in the counter (trigger bodies contain
semicolons). The historical-upgrade driver now wraps exec on every
connection initDatabase opens and fails on a batch, so that counting
assumption holds for the fresh profile and all historical schemas.
Documents the definition in the counter and the architecture docs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(performance): count SQL statements only for the launch journey
Codex review: the M3U import, refresh-cancellation and Xtream benchmarks
also run with IPTVNATOR_PERF_CAPTURE=1, so the statement hook wrapped
every row of their bulk inserts and changed what they measure.
SQL counting now also needs IPTVNATOR_PERF_COUNT_SQL=1, which only the
launch journey sets; a harness test fails if another source sets it.
Startup phases, the window snapshot and the read handler stay on the
capture flag. Without SQL counting no ready-to-show listener is attached,
so a zero is never reported for statements nobody counted.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: 4gray <fourgray@proton.me>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
|
69056487bc |
ci(performance): run the performance journeys on the Linux runner (#1717)
* ci(performance): run the performance journeys on the Linux runner Adds a warn-only performance-journeys job to ci.yml that builds the electron-performance configuration, runs the journey benchmarks under xvfb and uploads dist/performance/journeys/ as evidence. Pull requests run it only when they touch journey-relevant paths. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(performance): document the journeys CI job and runner evidence Describes the performance-journeys job and its path gate, and records why the J1 runtime counters are not baselined yet: on the Linux runner the launch journey is bimodal (13/576 vs 16/939 bridge calls/DOM mutations), so the counters are not deterministic. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * ci(performance): run the journeys unless a PR changes only safe paths The scope filter listed the paths that can move a journey, so a PR that changed only a root build input (.nvmrc, nx.json, tsconfig.base.json) skipped the measurement. List the paths that cannot instead: the E2E workflow's ignore list plus release notes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: 4gray <fourgray@proton.me> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
|
f0e51d2806 |
perf(web): keep lazy-only services and SafePipe out of main.js (#1729)
* perf(web): keep lazy-only services and SafePipe out of main.js The eager shell imported barrels that re-export Angular injectables and a pipe it never uses, and their static definitions keep those modules in main.js: PlaylistFileImportService came with PlaylistContextFacade, normalizeDateLocale with SafePipe, and the workspace-shell-util barrel with SettingsContextService, which #1714 grew with match counts. That growth put master 108 bytes over the renderer.initialBytes baseline #1712 had measured on a branch without #1714. Add file-level entries for the three modules and use them from the eager and settings code: renderer.initialBytes 1,626,127 -> 1,619,993 bytes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * chore(performance): lower the initial-bytes baseline to 1,619,993 bytes Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: 4gray <fourgray@proton.me> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
|
0f1cad4b16 |
test(performance): add the J1 launch-to-usable journey benchmark (#1698)
* test(performance): add the J1 launch-to-usable journey benchmark Implements plan items A1, A3 (J1 only) and the minimal A4 from .plans/2026-09-25-performance-journeys-ratchet.md. - journey-renderer-probe.ts: init-script probe counting DOM mutations, layout shifts and long tasks until the first source card is visible on /workspace with the splash removed; unit-tested with jsdom fixtures. - journey-main-ipc-capture.ts: counts bridge invocations from the preload's renderer-API trace channel up to a sentinel call the probe fires, so the IPC counter is exact without touching production code. - launch.journey.ts + playwright.journeys.config.ts: seeded profile (one M3U source, one Xtream portal on the loopback mock), one warm-up and five measured iterations, fresh process and data directory each, writing dist/performance/journeys/<timestamp>/summary.json with exact counters and P50/P90 wall-clock. - Nx target electron-backend-e2e:journeys and root script perf:journeys. - docs/architecture/performance-journeys.md, README, context and validation map entries. renderer.cdTicksToFirstCard and main.sqlStatementsBeforeReadyToShow are reported as unavailable with the reason instead of being faked. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * test(performance): gate the renderer load so the probes never race startup Review follow-up for #1698. - journey-renderer-gate.cjs: a main-process hook loaded with `-r` (the mechanism Playwright uses for its own loader) makes the first loadFile/loadURL navigate to about:blank and holds the real load until the test releases it. Playwright reports no page before a navigation commits, so this is what lets the renderer probe be registered on the page before the real document exists; the IPC capture is installed before the release too. A safety timeout releases the gate on its own and marks the iteration invalid. Unit-tested with a fake BrowserWindow. - launch-journey-app.ts: registers the probe on the parked page, releases the gate, waits for the real document to commit, fails fast when the probe is missing, and refuses an iteration whose gate timed out, saw a second load, or released before the probe was in place. - journey-renderer-probe.ts: entries delivered live after the terminal batch are buffered and filtered by the same cutoff as queued ones, and the cutoff is sampled in a timer queued from the first rAF, i.e. after the card's frame is painted, so the render task's long task and layout shift are consistently included. - launch-journey-record.ts: layoutShiftScore rounded to three decimals; a 0.0001 shift flipped in and out of the cutoff between iterations. - playwright.journeys.config.ts: reuse a mock server left on the journeys port locally (its fixtures are deterministic); CI still starts its own. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * test(performance): validate the journey gate after the probe completes Codex follow-up on #1698: the gate state returned by release() cannot see a reload or recovery navigation that happens before the first card. Re-read the live state once the renderer probe has finished and validate that instead, so an iteration spanning an extra navigation is rejected. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * test(performance): fail an iteration whose performance observers were unavailable Codex follow-up on #1698: a renderer that cannot observe layout-shift or longtask entries used to pass the probe with zero counters, which a ratchet could not tell apart from a genuine zero. The probe assertion now rejects such an iteration and names the missing observer. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> --------- Co-authored-by: 4gray <fourgray@proton.me> Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
4grayandClaude Fable 5.1
|
512e9787a8 |
perf(web): load Angular date locales lazily per language (#1695)
Plan thread C1, journey **J1 `launch`**, counter **`renderer.initialBytes`**. Stacked on #1694 → #1693 → #1692; merge in order. `apps/web/src/app/app-date-locales.ts` imported the locale data of all 18 supported languages eagerly, so every user shipped and parsed all of it at startup. Each locale is now a dynamic import keyed by the Angular locale id that `normalizeDateLocale()` derives from the app language (`by` → `be`, `ary` → `ar-MA`, `zhtw` → `zh-Hant`); English needs no data. Ordering is preserved so no template renders a locale whose data has not arrived (Angular throws in that case): - `main.ts` awaits the initial language's data (from `getInitialLanguage()`) before `bootstrapApplication`. - Both `TranslateService.use()` call sites, `AppComponent.initSettings()` and `SettingsFormFacade.applySavedSettings()`, register the data first through the new `AppDateLocaleService`. - A failed load never leaves the locale without data: English formatting is registered under the requested id (eager 1.1 KB `@angular/common/locales/en`), so `DatePipe` renders instead of throwing; the locale is not marked registered, so the next call retries and a success replaces the fallback (review follow-up). - `AppDateLocaleService.use()` orders switches by request, not by completion: a switch whose data arrives after a newer request is dropped, so the language chosen last wins (review follow-up). No kill switch: behavior is identical once the locale resolves, and the only new failure mode (a same-origin chunk failing to load) is shared with every lazy route. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
4grayandClaude Fable 5.1
|
d3b6e548cc |
ci(performance): fail when the web app's initial bytes grow (#1694)
Third step of the performance-journeys ratchet, stacked on #1693 (which is stacked on #1692; merge in order, GitHub retargets each to `master`). - New `Initial bytes ratchet` job in `.github/workflows/ci.yml` (ubuntu-latest): install, `pnpm nx build web --skip-nx-cache` (production configuration, the one users download), then `pnpm run perf:initial-bytes:check`. The job fails when `renderer.initialBytes` exceeds `tools/performance/journey-baselines.json`. - `dist/performance/` is uploaded as the `performance-journey-summary` artifact on every run, so a failing or tightenable run carries its evidence. - After review: the job first runs the new `tools/performance/check-baseline-direction.mjs`, which compares `journey-baselines.json` with the revision the change is measured against (the target branch of a pull request, `github.event.before` for a `master` push, `master` for a manual dispatch) and fails on any raised enforced limit (`value × toleranceRatio`), any widened or newly added tolerance, or any removed entry, so a PR cannot grow the payload and raise the baseline to match (lowered limits and new entries pass; a target branch without the file has nothing to weaken). Node tests cover it. - Docs: the performance-journeys contract and the validation map name the job, and the contract now states that this runner is the canonical measurer (take baseline values from its output, not from a local build). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
4grayandClaude Fable 5.1
|
7abaad29e7 |
chore(performance): commit the initial-bytes baseline and ratchet checker (#1693)
Second step of the performance-journeys ratchet, stacked on #1692 (merge that first; this PR retargets to `master` automatically). - `tools/performance/journey-baselines.json`: J1 `launch` / `renderer.initialBytes` = **2,739,510 bytes**, the ubuntu runner's production build of `apps/web` at this content (after #1692 stopped bundling `package.json` into `main.js`). A local macOS build of this pre-#1695 code is 2 bytes smaller in `main.js` (the eager locale imports); once #1695 removes them the two are byte-identical. Correction to an earlier version of this description: the "556-byte macOS vs Linux difference" was almost entirely `package.json` text embedded in `main.js`, which moved with every script edit in this stack, plus this 2-byte residue. The runner is the canonical measurer; the CI run on the stacked #1694 branch (this content plus the job) is where the number is confirmed. - `tools/performance/check-journey-ratchet.mjs` compares a journey summary with the baselines: a counter above its value fails (exact, no slack), wall-clock entries fail above `value × toleranceRatio`, a baseline without a measurement fails so dropping a measurement cannot disable the ratchet, values below baseline print a "tighten" hint, and measured counters without a baseline are noted only. After review: checking nothing (empty file, or `--only` naming a missing entry) fails; a counter is read only from `counters` and a wall-clock entry only from `wallClock`; the repeatable `--only <journey>/<counter>` flag scopes a check. - Root scripts: `perf:initial-bytes:check` (measures into its own `dist/performance/initial-bytes.summary.json`, then checks `--only launch/renderer.initialBytes`) and `perf:ratchet:check` (full check); `perf:tools:test` runs both test files, as does `pnpm nx test performance-tools`. - `docs/architecture/performance-journeys.md` gains the Ratchet section (file format, rules, "baselines only move down"); the validation map lists the check. The CI job that runs the check on every PR is #1694; C1 (lazy Angular date locales, #1695) then lowers the baseline with the measured output as evidence. Note: `ci.yml` only triggers on pull requests targeting `master`, so this stacked PR shows no Actions runs until #1692 merges. The evidence runs above were dispatched with `gh workflow run ci.yml --ref <branch>`. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
4grayandClaude Fable 5.1
|
8bc877b625 |
chore(performance): measure initial bytes of the built web app (#1692)
First step of the performance-journeys ratchet (plan thread: J1 `launch`, counter `renderer.initialBytes`).
- `tools/performance/measure-initial-bytes.mjs` reads the built `dist/apps/web/index.html` and sums `index.html` plus every same-origin `<script src>`, `<link rel="stylesheet">` and `<link rel="modulepreload">` it references. Manifest, icons, external URLs and lazy chunks are not counted. A referenced file missing from the build fails the measurement instead of counting as zero bytes.
- `--json` prints the breakdown; `--summary <file>` writes the `journeys.<journey>.counters` shape a ratchet checker will consume (next PR).
- New Nx project `performance-tools` (test + lint targets), Tier B in `tools/coverage/coverage-policy.json`, root scripts `perf:initial-bytes` and `perf:tools:test`.
- New contract `docs/architecture/performance-journeys.md`, linked from the validation map, the agent context map and the README.
- **Review follow-ups:** resources are deduplicated by request URL (query kept, fragment dropped); `index.html` is parsed with parse5 (already a repository dependency, scripting enabled), so comments, bogus comments, raw-text bodies (script/style/noscript/title/textarea), inert `<template>` contents and character references in attributes all follow the HTML5 algorithm instead of a hand-written scanner; the review's edge cases stay as regression tests; docs show the `pnpm --silent` form for JSON output and explain how the counter relates to Angular's rounded "Initial total".
- **Found while measuring:** the environment files and the playback diagnostic panel imported the whole `package.json` (`import packageJson from '@package'`), which esbuild cannot tree-shake, so `main.js` carried the complete file and the counter moved with every script or dependency edit. They now import `{ version }` only (
|