kayg/calternal
Watch
1
0
Fork
You've already forked calternal
 0
Code Issues 803 Pull requests Projects Releases Packages Wiki Activity Actions 430

Opening Settings takes about 1 s (from Notes); make it near-instant #642

New issue
Open
opened 2026-10-01 16:22:33 +00:00 by kayg · 25 comments
kayg commented 2026-10-01 16:22:33 +00:00
Owner
Copy link

Owner report (2026-10-01): opening Settings from Notes took about one full second. It was not an animation: nothing happened for one second, then Settings appeared at once. Opening Settings must be near-instant.

Expected

  • Settings opens (⌘, / Ctrl+, / G then S / the menu) from any mode, warm, in ≤ 100 ms to fully painted (p95), cold ≤ 400 ms. Same budget for every Settings section deep link.
  • Loading never waits for motion and never waits for data that a cache can serve.

Probable causes to check (measure first, with a trace)

  • The Settings route chunk (and each section chunk) loads on demand. Preload them on idle after the first paint of the app, and on hover/focus of the Settings entry. Note #542 found the Settings chunk grew +8.7%: split heavy parts (Appearance background editor, theme previews) out of the first section.
  • Section data waits on several API calls in series. Fetch in parallel or one batched call, render from the per-User cache (userStorage) first, then revalidate.
  • Notes leaving the route does synchronous teardown (editor, Yjs, observers) before the navigation commits.
  • A blocking await before the first paint (fonts, theme, preferences).

Measure

Use the #549 harness on the perf VM with HDD emulation (bench/hdd-emu.sh, flock /root/perf.lock): Notes → Settings, Calendar → Settings, Mail → Settings, Files → Settings; cold and warm; Chromium and WebKit; 1440 and 390. Post a table before and after.

**Owner report (2026-10-01):** opening Settings from Notes took about **one full second**. It was not an animation: nothing happened for one second, then Settings appeared at once. Opening Settings must be near-instant. ### Expected - Settings opens (⌘, / Ctrl+, / G then S / the menu) from any mode, warm, in **≤ 100 ms to fully painted** (p95), cold **≤ 400 ms**. Same budget for every Settings section deep link. - Loading never waits for motion and never waits for data that a cache can serve. ### Probable causes to check (measure first, with a trace) - The Settings route chunk (and each section chunk) loads on demand. Preload them on idle after the first paint of the app, and on hover/focus of the Settings entry. Note #542 found the Settings chunk grew +8.7%: split heavy parts (Appearance background editor, theme previews) out of the first section. - Section data waits on several API calls in series. Fetch in parallel or one batched call, render from the per-User cache (userStorage) first, then revalidate. - Notes leaving the route does synchronous teardown (editor, Yjs, observers) before the navigation commits. - A blocking `await` before the first paint (fonts, theme, preferences). ### Measure Use the #549 harness on the perf VM with HDD emulation (`bench/hdd-emu.sh`, `flock /root/perf.lock`): Notes → Settings, Calendar → Settings, Mail → Settings, Files → Settings; cold and warm; Chromium and WebKit; 1440 and 390. Post a table before and after.
phone-light.png
92 KiB
phone-dark.png
106 KiB
tablet-dark.png
174 KiB
desktop-light.png
395 KiB
desktop-dark.png
354 KiB
tablet-light.png
114 KiB
desktop-paper.png
286 KiB
desktop-tokyo-night.png
228 KiB
phone-paper.png
82 KiB
phone-tokyo-night.png
82 KiB
tablet-paper.png
184 KiB
tablet-tokyo-night.png
174 KiB
desktop-paper.png
270 KiB
desktop-tokyo-night.png
222 KiB
phone-paper.png
83 KiB
phone-tokyo-night.png
82 KiB
tablet-paper.png
170 KiB
tablet-tokyo-night.png
166 KiB
settings-notes-warm-trace-642.json
1 MiB
settings-open-642-scoped-filter.json
4.9 KiB
phone-light.png
100 KiB
phone-dark.png
106 KiB
tablet-light.png
143 KiB
tablet-dark.png
174 KiB
desktop-light.png
402 KiB
desktop-dark.png
354 KiB
phone-light.png phone-dark.png tablet-dark.png desktop-light.png desktop-dark.png tablet-light.png desktop-paper.png desktop-tokyo-night.png phone-paper.png phone-tokyo-night.png tablet-paper.png tablet-tokyo-night.png desktop-paper.png desktop-tokyo-night.png phone-paper.png phone-tokyo-night.png tablet-paper.png tablet-tokyo-night.png phone-light.png phone-dark.png tablet-light.png tablet-dark.png desktop-light.png desktop-dark.png
kayg commented 2026-10-01 16:50:18 +00:00
Author
Owner
Copy link

Starting #642 on job/blaze-settings at base 687ff70313. I will trace Settings navigation from Notes, Calendar, Mail, and Files, then implement and measure the warm and cold path.

Starting #642 on job/blaze-settings at base 687ff703136e71e89f8dfba139e93cd0788b25c1. I will trace Settings navigation from Notes, Calendar, Mail, and Files, then implement and measure the warm and cold path.
kayg commented 2026-10-01 17:51:06 +00:00
Author
Owner
Copy link

Baseline finding for #642. The production build ran locally because the perf VM is unreachable (ssh root@10.69.69.63 returns No route to host) and bench/hdd-emu.sh is not present. Host 1-minute load average was 18.98. Do not treat these as quiet-host perf-VM numbers.

Exact command:

PERF_LOCATION=local CALTERNAL_E2E_ASSET_OVERRIDE=1 CALTERNAL_SERVER_BIN="$CARGO_TARGET_DIR/debug/calternal-server" TMPDIR="$PWD/target/tmp" SETTINGS_OPEN_RUNS=5 node bench/settings-open-642.mjs

First-open latency is measured until the selected Account section has painted complete content. Warm p50/p95 uses five opens per source Tab after the first open in a fresh context. Results exceed both requested targets (warm ≤100 ms; cold ≤400 ms).

Source Cold open ms Warm p50/p95 ms Warm CPU p50/p95 ms Server RSS before/after KiB
Notes 1909.30 779.30 / 894.64 670.000 / 740.000 153880 / 224825.6
Calendar 1088.00 767.70 / 809.30 920.000 / 982.000 226032 / 226500
Mail 1097.50 541.80 / 651.62 690.000 / 838.000 226712 / 227216
Files 642.90 587.80 / 797.04 600.000 / 754.000 227716 / 228035.2

Full samples and heap/browser RSS data: artifacts/blaze/settings-open-642.json (local, not committed).

Baseline finding for #642. The production build ran locally because the perf VM is unreachable (`ssh root@10.69.69.63` returns `No route to host`) and `bench/hdd-emu.sh` is not present. Host 1-minute load average was 18.98. Do not treat these as quiet-host perf-VM numbers. Exact command: ```sh PERF_LOCATION=local CALTERNAL_E2E_ASSET_OVERRIDE=1 CALTERNAL_SERVER_BIN="$CARGO_TARGET_DIR/debug/calternal-server" TMPDIR="$PWD/target/tmp" SETTINGS_OPEN_RUNS=5 node bench/settings-open-642.mjs ``` First-open latency is measured until the selected Account section has painted complete content. Warm p50/p95 uses five opens per source Tab after the first open in a fresh context. Results exceed both requested targets (warm ≤100 ms; cold ≤400 ms). | Source | Cold open ms | Warm p50/p95 ms | Warm CPU p50/p95 ms | Server RSS before/after KiB | | --- | ---: | ---: | ---: | ---: | | Notes | 1909.30 | 779.30 / 894.64 | 670.000 / 740.000 | 153880 / 224825.6 | | Calendar | 1088.00 | 767.70 / 809.30 | 920.000 / 982.000 | 226032 / 226500 | | Mail | 1097.50 | 541.80 / 651.62 | 690.000 / 838.000 | 226712 / 227216 | | Files | 642.90 | 587.80 / 797.04 | 600.000 / 754.000 | 227716 / 228035.2 | Full samples and heap/browser RSS data: `artifacts/blaze/settings-open-642.json` (local, not committed).
kayg commented 2026-10-01 19:51:01 +00:00
Author
Owner
Copy link

Finished #642 on branch job/blaze-settings. Head: 94a9433216.

Added idle and pointer/focus-intent preload for the remembered Settings route, a verified User display cache, and per-User Settings resource snapshots. Settings navigation does not await the preload. The route uses cached User data while it refreshes the server value. The benchmark opens Settings from real Notes, Calendar, Mail, and Files routes and waits for complete content.

Exact production-build measurement commands:

# Before
PERF_LOCATION=local CALTERNAL_E2E_ASSET_OVERRIDE=1 CALTERNAL_SERVER_BIN="$CARGO_TARGET_DIR/debug/calternal-server" TMPDIR="$PWD/target/tmp" SETTINGS_OPEN_RUNS=5 node bench/settings-open-642.mjs

# After
PERF_LOCATION=local CALTERNAL_E2E_ASSET_OVERRIDE=1 CALTERNAL_SERVER_BIN="$CARGO_TARGET_DIR/debug/calternal-server" TMPDIR="$PWD/target/tmp" SETTINGS_OPEN_RUNS=5 SETTINGS_OPEN_OUTPUT=artifacts/blaze/settings-open-after-642.json node bench/settings-open-642.mjs
Source Cold open ms, before → after Warm latency p50/p95 ms, before → after Warm browser CPU p50/p95 ms, before → after Server RSS before/after KiB, before → after
Notes 1909.30 → 1155.40 779.30/894.64 → 689.70/757.02 670/740 → 530/560 153880/224825.6 → 148564/226535.2
Calendar 1088.00 → 1223.30 767.70/809.30 → 636.40/1277.90 920/982 → 600/670 226032/226500 → 227932/228143.2
Mail 1097.50 → 918.80 541.80/651.62 → 376.00/613.72 690/838 → 470/498 226712/227216 → 228372/228916
Files 642.90 → 711.80 587.80/797.04 → 347.00/528.10 600/754 → 460/674 227716/228035.2 → 229576/229708

The after run used a local production build at 1440×900, five warm opens per source, with load averages 24.99/26.11/25.24. The baseline load average was 18.98. SSH to root@10.69.69.63 returned “No route to host”, and bench/hdd-emu.sh is absent, so no perf-VM or HDD-emulated result is claimed. The ≤100 ms warm p95 and ≤400 ms cold goals are not met by these noisy local results. The Notes cold result improved; Calendar and Files cold results regressed in this run, and Calendar had a wide warm tail. These results need perf-VM confirmation.

The real Notes → Settings E2E passed its generous 2500 ms CI budget for both first and warm opens. The measurement profiles and JSON are local and not committed.

Final web gates, verbatim output excerpts:

$ node scripts/check-user-storage.mjs && node scripts/check-type-tokens.mjs && node scripts/check-motion-tokens.mjs && svelte-kit sync && svelte-check --tsconfig ./tsconfig.json
User browser caches use userStorage; only documented device/public-link exceptions remain.
Text sizes and UI shape values use shared role tokens.
UI transitions and animation options use shared motion tokens or documented exceptions.
Loading svelte-check in workspace: /home/kayg/Developer/calternal-wt/blaze-settings/apps/web
Getting Svelte diagnostics...

svelte-check found 0 errors and 0 warnings

$ vitest run

 Test Files  148 passed (148)
      Tests  1011 passed (1011)
   Start at  21:39:17
   Duration  258.67s (transform 52%, environment 20%, import 14%, tests 11%, setup 4%)

Environment  |component| jsdom was created 46 times · 269.31s total, 26% of tracked time

Decisions not specified in DESIGN.md: preload the remembered Settings route after the active Tab's first paint and on link intent; display a validated User snapshot and safe Settings resource caches while refreshing them from the server; keep raw Admin TOML out of persistent browser storage; and retain visited Settings panes only until the overlay closes. Loading remains independent from the entrance animation.

No Rust crate or server API changed, so no Rust gates or API adversarial probe were applicable. cargo clean removed 7237 files (4.6GiB), and apps/web/build was deleted.

Finished #642 on branch job/blaze-settings. Head: 94a9433216da2de7b0c57d1a03cfab6f1500df00. Added idle and pointer/focus-intent preload for the remembered Settings route, a verified User display cache, and per-User Settings resource snapshots. Settings navigation does not await the preload. The route uses cached User data while it refreshes the server value. The benchmark opens Settings from real Notes, Calendar, Mail, and Files routes and waits for complete content. Exact production-build measurement commands: ~~~sh # Before PERF_LOCATION=local CALTERNAL_E2E_ASSET_OVERRIDE=1 CALTERNAL_SERVER_BIN="$CARGO_TARGET_DIR/debug/calternal-server" TMPDIR="$PWD/target/tmp" SETTINGS_OPEN_RUNS=5 node bench/settings-open-642.mjs # After PERF_LOCATION=local CALTERNAL_E2E_ASSET_OVERRIDE=1 CALTERNAL_SERVER_BIN="$CARGO_TARGET_DIR/debug/calternal-server" TMPDIR="$PWD/target/tmp" SETTINGS_OPEN_RUNS=5 SETTINGS_OPEN_OUTPUT=artifacts/blaze/settings-open-after-642.json node bench/settings-open-642.mjs ~~~ | Source | Cold open ms, before → after | Warm latency p50/p95 ms, before → after | Warm browser CPU p50/p95 ms, before → after | Server RSS before/after KiB, before → after | | --- | --- | --- | --- | --- | | Notes | 1909.30 → 1155.40 | 779.30/894.64 → 689.70/757.02 | 670/740 → 530/560 | 153880/224825.6 → 148564/226535.2 | | Calendar | 1088.00 → 1223.30 | 767.70/809.30 → 636.40/1277.90 | 920/982 → 600/670 | 226032/226500 → 227932/228143.2 | | Mail | 1097.50 → 918.80 | 541.80/651.62 → 376.00/613.72 | 690/838 → 470/498 | 226712/227216 → 228372/228916 | | Files | 642.90 → 711.80 | 587.80/797.04 → 347.00/528.10 | 600/754 → 460/674 | 227716/228035.2 → 229576/229708 | The after run used a local production build at 1440×900, five warm opens per source, with load averages 24.99/26.11/25.24. The baseline load average was 18.98. SSH to root@10.69.69.63 returned “No route to host”, and bench/hdd-emu.sh is absent, so no perf-VM or HDD-emulated result is claimed. The ≤100 ms warm p95 and ≤400 ms cold goals are not met by these noisy local results. The Notes cold result improved; Calendar and Files cold results regressed in this run, and Calendar had a wide warm tail. These results need perf-VM confirmation. The real Notes → Settings E2E passed its generous 2500 ms CI budget for both first and warm opens. The measurement profiles and JSON are local and not committed. Final web gates, verbatim output excerpts: ~~~text $ node scripts/check-user-storage.mjs && node scripts/check-type-tokens.mjs && node scripts/check-motion-tokens.mjs && svelte-kit sync && svelte-check --tsconfig ./tsconfig.json User browser caches use userStorage; only documented device/public-link exceptions remain. Text sizes and UI shape values use shared role tokens. UI transitions and animation options use shared motion tokens or documented exceptions. Loading svelte-check in workspace: /home/kayg/Developer/calternal-wt/blaze-settings/apps/web Getting Svelte diagnostics... svelte-check found 0 errors and 0 warnings ~~~ ~~~text $ vitest run Test Files 148 passed (148) Tests 1011 passed (1011) Start at 21:39:17 Duration 258.67s (transform 52%, environment 20%, import 14%, tests 11%, setup 4%) Environment |component| jsdom was created 46 times · 269.31s total, 26% of tracked time ~~~ Decisions not specified in DESIGN.md: preload the remembered Settings route after the active Tab's first paint and on link intent; display a validated User snapshot and safe Settings resource caches while refreshing them from the server; keep raw Admin TOML out of persistent browser storage; and retain visited Settings panes only until the overlay closes. Loading remains independent from the entrance animation. No Rust crate or server API changed, so no Rust gates or API adversarial probe were applicable. cargo clean removed 7237 files (4.6GiB), and apps/web/build was deleted.
kayg commented 2026-10-01 20:01:27 +00:00
Author
Owner
Copy link

Round 2 starting on job/blaze-settings at 94a9433216da2de7b0c57d1a03cfab6f1500df00, based on origin/dev 687ff703136e71e89f8dfba139e93cd0788b25c1. The previous warm-open table still shows roughly 530 ms of browser CPU from Notes, so I will record a production-build CDP trace and JS sampling profile, plus distinct-paint data for Settings traversal, before changing code. I will post the attribution tables here first.

Round 2 starting on `job/blaze-settings` at `94a9433216da2de7b0c57d1a03cfab6f1500df00`, based on `origin/dev` `687ff703136e71e89f8dfba139e93cd0788b25c1`. The previous warm-open table still shows roughly 530 ms of browser CPU from Notes, so I will record a production-build CDP trace and JS sampling profile, plus distinct-paint data for Settings traversal, before changing code. I will post the attribution tables here first.
kayg commented 2026-10-01 20:30:07 +00:00
Author
Owner
Copy link

Production-build attribution before application changes. Method: one warmed Notes → Settings open on the real Notes route and local server; Chrome CDP Tracing captured devtools.timeline, v8.execute, blink.user_timing, detailed timeline and V8 CPU profiler events. A 1 ms V8 sampling profile ran with the trace. The window starts immediately before Control+, and ends after the selected Account section is complete and two frames have passed. Chromium browser, renderer, GPU and network processes were pinned to CPUs 0–1. uptime load was 8.22 before and 7.81 after (1/5/15-minute averages 8.22/20.04/27.44 before). This is one trace, not a p95 run.

Rank Self cost Function / event File / thread
1 69.8 ms RasterTask (sum across raster workers) Chrome renderer raster workers
2 49.6 ms V8 (program) samples without a named call frame renderer JS profile
3 31.0 ms UpdateLayoutTree (style recalc) renderer main thread
4 29.2 ms getBoundingClientRect (nearest JS caller p) /_app/immutable/chunks/DE3A531x.js
5 19.3 ms LocalFrameView::performLayout renderer main thread
6 11.8 ms ao /_app/immutable/chunks/DE3A531x.js
7 5.8 ms anonymous function /_app/immutable/nodes/46.DZAzbUuc.js
8 5.3 ms anonymous function /_app/immutable/chunks/DE3A531x.js
9 3.2 ms fetch /_app/immutable/chunks/DK5Urzxf.js
10 3.0 ms PrePaint renderer main thread
11 2.7 ms before /_app/immutable/chunks/Df0kZKHp.js
12 2.6 ms Paint renderer main thread
13 2.4 ms xo /_app/immutable/chunks/DE3A531x.js
14 2.3 ms URL /_app/immutable/chunks/DlmC6AZe.js
15 2.2 ms destroy /_app/immutable/chunks/B2bumzYI.js
16 2.2 ms cloneNode /_app/immutable/chunks/Df0kZKHp.js
17 2.1 ms window.fetch /_app/immutable/chunks/DK5Urzxf.js
18 1.7 ms blur /_app/immutable/chunks/DK5Urzxf.js
19 1.6 ms ii /_app/immutable/chunks/Df0kZKHp.js
20 1.5 ms Ka /_app/immutable/chunks/Df0kZKHp.js

Trace self time totals for the measured window: style recalc 31.0 ms; layout 19.6 ms; main-thread PrePaint/Paint 5.6 ms; raster-worker paint 69.8 ms; compositing (Layerize/Commit/UpdateLayer) 1.9 ms; GC 0 ms. The JS profiler's named-function samples plus the 49.6 ms V8 (program) samples account for about 157 ms of JS self time. Raster worker time is CPU across threads and can overlap wall time; do not add it to the main-thread total.

The renderer PerformanceObserver and trace both saw one main-thread long task: 97.6 ms, starting 106.3 ms after the open mark (observer reported 97 ms). Chrome also traced GPU-process RunTask spans of 61.4 ms at +44.5 ms, 66.0 ms at +116.9 ms and 53.3 ms at +183.3 ms; these are separate from renderer main-thread Long Tasks. Complete Account content was ready in 294.3 ms. The raw trace is in the local ignored artifact artifacts/blaze/settings-notes-warm-trace-642.json.

Production-build attribution before application changes. Method: one warmed Notes → Settings open on the real Notes route and local server; Chrome CDP Tracing captured `devtools.timeline`, `v8.execute`, `blink.user_timing`, detailed timeline and V8 CPU profiler events. A 1 ms V8 sampling profile ran with the trace. The window starts immediately before `Control+,` and ends after the selected Account section is complete and two frames have passed. Chromium browser, renderer, GPU and network processes were pinned to CPUs 0–1. `uptime` load was 8.22 before and 7.81 after (1/5/15-minute averages 8.22/20.04/27.44 before). This is one trace, not a p95 run. | Rank | Self cost | Function / event | File / thread | | ---: | ---: | --- | --- | | 1 | 69.8 ms | RasterTask (sum across raster workers) | Chrome renderer raster workers | | 2 | 49.6 ms | V8 `(program)` samples without a named call frame | renderer JS profile | | 3 | 31.0 ms | UpdateLayoutTree (style recalc) | renderer main thread | | 4 | 29.2 ms | getBoundingClientRect (nearest JS caller `p`) | `/_app/immutable/chunks/DE3A531x.js` | | 5 | 19.3 ms | LocalFrameView::performLayout | renderer main thread | | 6 | 11.8 ms | `ao` | `/_app/immutable/chunks/DE3A531x.js` | | 7 | 5.8 ms | anonymous function | `/_app/immutable/nodes/46.DZAzbUuc.js` | | 8 | 5.3 ms | anonymous function | `/_app/immutable/chunks/DE3A531x.js` | | 9 | 3.2 ms | `fetch` | `/_app/immutable/chunks/DK5Urzxf.js` | | 10 | 3.0 ms | PrePaint | renderer main thread | | 11 | 2.7 ms | `before` | `/_app/immutable/chunks/Df0kZKHp.js` | | 12 | 2.6 ms | Paint | renderer main thread | | 13 | 2.4 ms | `xo` | `/_app/immutable/chunks/DE3A531x.js` | | 14 | 2.3 ms | `URL` | `/_app/immutable/chunks/DlmC6AZe.js` | | 15 | 2.2 ms | `destroy` | `/_app/immutable/chunks/B2bumzYI.js` | | 16 | 2.2 ms | `cloneNode` | `/_app/immutable/chunks/Df0kZKHp.js` | | 17 | 2.1 ms | `window.fetch` | `/_app/immutable/chunks/DK5Urzxf.js` | | 18 | 1.7 ms | `blur` | `/_app/immutable/chunks/DK5Urzxf.js` | | 19 | 1.6 ms | `ii` | `/_app/immutable/chunks/Df0kZKHp.js` | | 20 | 1.5 ms | `Ka` | `/_app/immutable/chunks/Df0kZKHp.js` | Trace self time totals for the measured window: style recalc 31.0 ms; layout 19.6 ms; main-thread PrePaint/Paint 5.6 ms; raster-worker paint 69.8 ms; compositing (Layerize/Commit/UpdateLayer) 1.9 ms; GC 0 ms. The JS profiler's named-function samples plus the 49.6 ms V8 `(program)` samples account for about 157 ms of JS self time. Raster worker time is CPU across threads and can overlap wall time; do not add it to the main-thread total. The renderer PerformanceObserver and trace both saw one main-thread long task: 97.6 ms, starting 106.3 ms after the open mark (observer reported 97 ms). Chrome also traced GPU-process RunTask spans of 61.4 ms at +44.5 ms, 66.0 ms at +116.9 ms and 53.3 ms at +183.3 ms; these are separate from renderer main-thread Long Tasks. Complete Account content was ready in 294.3 ms. The raw trace is in the local ignored artifact `artifacts/blaze/settings-notes-warm-trace-642.json`.
kayg commented 2026-10-01 21:07:26 +00:00
Author
Owner
Copy link

Round 2 follow-up trace (after rail geometry cache)

The Settings rail's own getBoundingClientRect() call no longer appears in the profile. The production warm Notes → Settings trace still took 708.3 ms to complete, so the 100 ms target is not met. Chromium processes were pinned to CPUs 0,1. Host load average was 32.79 / 31.62 / 29.09, above the requested quiet-host preference; treat latency as contended local data.

Cost Self / summed duration Attribution
(program) 189 ms sampled Main renderer JavaScript without a named frame
Svelte ao 37 ms sampled fade transition helper in BIcm8obb.js; reads computed opacity
getBoundingClientRect 15 ms sampled Remaining route/overlay geometry reads; no FloatingSidebar caller in this trace
xo 12 ms sampled Theme canvas color resolution in themeColor.ts; appends a probe and reads computed style
focus 11 ms sampled Native focus handling
fetch 7 ms sampled Native fetch dispatch
cloneNode 5 ms sampled DOM clone
before 4 ms sampled Minified callback in Df0kZKHp.js
mark 3 ms sampled User timing marker
blur 3 ms sampled Native focus handling
removeAttribute 3 ms sampled DOM update
get cookie 3 ms sampled Cookie read
window.fetch 3 ms sampled D5C0KbR5.js
Ka 3 ms sampled Df0kZKHp.js
(anonymous) 2 ms sampled Df0kZKHp.js
Oe 2 ms sampled D5C0KbR5.js
ownKeys 2 ms sampled Df0kZKHp.js
u 2 ms sampled nodes/46.BSKJW9FZ.js
n 2 ms sampled BIcm8obb.js
m 2 ms sampled DlmC6AZe.js

Main-thread trace totals: style recalc 72.5 ms, layout 32.1 ms, pre-paint 6.4 ms, paint 11.1 ms, compositing commit 20.1 ms, layerize 2.4 ms, GC safepoint 0.4 ms. Raster work summed to 86.3 ms across worker threads. The open contained 74 ms and 229 ms long tasks. The previous trace table remains the pre-change attribution table requested before code changes; the raw follow-up trace is artifacts/blaze/settings-notes-warm-trace-642.json in the worktree.

### Round 2 follow-up trace (after rail geometry cache) The Settings rail's own `getBoundingClientRect()` call no longer appears in the profile. The production warm Notes → Settings trace still took **708.3 ms** to complete, so the 100 ms target is not met. Chromium processes were pinned to CPUs 0,1. Host load average was **32.79 / 31.62 / 29.09**, above the requested quiet-host preference; treat latency as contended local data. | Cost | Self / summed duration | Attribution | | --- | ---: | --- | | `(program)` | 189 ms sampled | Main renderer JavaScript without a named frame | | Svelte `ao` | 37 ms sampled | `fade` transition helper in `BIcm8obb.js`; reads computed opacity | | `getBoundingClientRect` | 15 ms sampled | Remaining route/overlay geometry reads; no FloatingSidebar caller in this trace | | `xo` | 12 ms sampled | Theme canvas color resolution in `themeColor.ts`; appends a probe and reads computed style | | `focus` | 11 ms sampled | Native focus handling | | `fetch` | 7 ms sampled | Native fetch dispatch | | `cloneNode` | 5 ms sampled | DOM clone | | `before` | 4 ms sampled | Minified callback in `Df0kZKHp.js` | | `mark` | 3 ms sampled | User timing marker | | `blur` | 3 ms sampled | Native focus handling | | `removeAttribute` | 3 ms sampled | DOM update | | `get cookie` | 3 ms sampled | Cookie read | | `window.fetch` | 3 ms sampled | `D5C0KbR5.js` | | `Ka` | 3 ms sampled | `Df0kZKHp.js` | | `(anonymous)` | 2 ms sampled | `Df0kZKHp.js` | | `Oe` | 2 ms sampled | `D5C0KbR5.js` | | `ownKeys` | 2 ms sampled | `Df0kZKHp.js` | | `u` | 2 ms sampled | `nodes/46.BSKJW9FZ.js` | | `n` | 2 ms sampled | `BIcm8obb.js` | | `m` | 2 ms sampled | `DlmC6AZe.js` | Main-thread trace totals: style recalc **72.5 ms**, layout **32.1 ms**, pre-paint **6.4 ms**, paint **11.1 ms**, compositing commit **20.1 ms**, layerize **2.4 ms**, GC safepoint **0.4 ms**. Raster work summed to **86.3 ms** across worker threads. The open contained **74 ms** and **229 ms** long tasks. The previous trace table remains the pre-change attribution table requested before code changes; the raw follow-up trace is `artifacts/blaze/settings-notes-warm-trace-642.json` in the worktree.
kayg commented 2026-10-01 21:36:18 +00:00
Author
Owner
Copy link

Production-build CDP follow-up after the cached FloatingSidebar geometry and transition/style-probe fixes. I opened Notes → Settings from a warmed Notes route, with Tracing categories devtools.timeline, v8.execute, blink.user_timing, detailed timeline/layout, and the 1 ms V8 sampling profiler. Chromium processes were pinned to CPUs 0–1. Local uptime before the run was 14.34, 17.28, 20.82 (1/5/15-minute averages), so this was not a quiet-host run. The complete Account pane was ready in 417 ms.

Top 20 V8 sample buckets by self time in the measured open window. Generated bundle paths are shown because this production build has no source maps; (program) has no script URL in V8's profile.

Rank Self time Function / sample File
1 117.5 ms (program) V8 profile, no script URL
2 36.0 ms focus browser DOM API
3 16.9 ms fetch browser API
4 11.1 ms garbage collector V8
5 9.1 ms blur browser DOM API
6 8.3 ms get cookie browser DOM API
7 4.5 ms setAttribute browser DOM API
8 4.2 ms before /_app/immutable/chunks/Df0kZKHp.js:1
9 4.0 ms getBoundingClientRect browser DOM API
10 3.9 ms get aria-busy /_app/immutable/nodes/46.CgLMEUpY.js:1
11 3.9 ms children /_app/immutable/nodes/46.CgLMEUpY.js:1
12 3.4 ms p /_app/immutable/chunks/DlmC6AZe.js:1
13 2.9 ms anonymous /_app/immutable/nodes/0.hOni4qPx.js:6
14 2.7 ms anonymous /_app/immutable/nodes/46.CgLMEUpY.js:6
15 2.7 ms matchMedia browser DOM API
16 2.7 ms remove browser DOM API
17 2.3 ms databases browser IndexedDB API
18 2.3 ms Ar /_app/immutable/chunks/Df0kZKHp.js:1
19 2.1 ms send browser API
20 2.1 ms append browser DOM API

Trace event totals (separate from and overlapping V8 samples): style recalc 29.8 ms; layout 20.3 ms; PrePaint 10.0 ms; Paint 17.4 ms; Layerize + Commit + UpdateLayer 2.7 ms; raster workers 47.0 ms; minor GC 10.4 ms. The renderer had one 147.2 ms RunTask beginning 129.8 ms into the open. It contains a 103.3 ms animation-frame task / 102.6 ms Svelte microtask flush; the largest nested FunctionCall is an anonymous function in Df0kZKHp.js:1:4903.

The earlier named costs removed from the first trace are no longer prominent: getComputedStyle-based fade/theme-color work and the dashboard fly function do not appear in the top samples. Warm open remains above the 100 ms target; this run attributes the largest remaining slice to an anonymous Svelte/runtime task and the renderer still had a long task. The local Settings profile after these changes is being measured separately across Notes, Calendar, Mail and Files.

Production-build CDP follow-up after the cached FloatingSidebar geometry and transition/style-probe fixes. I opened Notes → Settings from a warmed Notes route, with Tracing categories `devtools.timeline`, `v8.execute`, `blink.user_timing`, detailed timeline/layout, and the 1 ms V8 sampling profiler. Chromium processes were pinned to CPUs 0–1. Local `uptime` before the run was `14.34, 17.28, 20.82` (1/5/15-minute averages), so this was not a quiet-host run. The complete Account pane was ready in 417 ms. Top 20 V8 sample buckets by self time in the measured open window. Generated bundle paths are shown because this production build has no source maps; `(program)` has no script URL in V8's profile. | Rank | Self time | Function / sample | File | | ---: | ---: | --- | --- | | 1 | 117.5 ms | `(program)` | V8 profile, no script URL | | 2 | 36.0 ms | `focus` | browser DOM API | | 3 | 16.9 ms | `fetch` | browser API | | 4 | 11.1 ms | garbage collector | V8 | | 5 | 9.1 ms | `blur` | browser DOM API | | 6 | 8.3 ms | `get cookie` | browser DOM API | | 7 | 4.5 ms | `setAttribute` | browser DOM API | | 8 | 4.2 ms | `before` | `/_app/immutable/chunks/Df0kZKHp.js:1` | | 9 | 4.0 ms | `getBoundingClientRect` | browser DOM API | | 10 | 3.9 ms | `get aria-busy` | `/_app/immutable/nodes/46.CgLMEUpY.js:1` | | 11 | 3.9 ms | `children` | `/_app/immutable/nodes/46.CgLMEUpY.js:1` | | 12 | 3.4 ms | `p` | `/_app/immutable/chunks/DlmC6AZe.js:1` | | 13 | 2.9 ms | anonymous | `/_app/immutable/nodes/0.hOni4qPx.js:6` | | 14 | 2.7 ms | anonymous | `/_app/immutable/nodes/46.CgLMEUpY.js:6` | | 15 | 2.7 ms | `matchMedia` | browser DOM API | | 16 | 2.7 ms | `remove` | browser DOM API | | 17 | 2.3 ms | `databases` | browser IndexedDB API | | 18 | 2.3 ms | `Ar` | `/_app/immutable/chunks/Df0kZKHp.js:1` | | 19 | 2.1 ms | `send` | browser API | | 20 | 2.1 ms | `append` | browser DOM API | Trace event totals (separate from and overlapping V8 samples): style recalc 29.8 ms; layout 20.3 ms; PrePaint 10.0 ms; Paint 17.4 ms; Layerize + Commit + UpdateLayer 2.7 ms; raster workers 47.0 ms; minor GC 10.4 ms. The renderer had one 147.2 ms `RunTask` beginning 129.8 ms into the open. It contains a 103.3 ms animation-frame task / 102.6 ms Svelte microtask flush; the largest nested `FunctionCall` is an anonymous function in `Df0kZKHp.js:1:4903`. The earlier named costs removed from the first trace are no longer prominent: `getComputedStyle`-based `fade`/theme-color work and the dashboard `fly` function do not appear in the top samples. Warm open remains above the 100 ms target; this run attributes the largest remaining slice to an anonymous Svelte/runtime task and the renderer still had a long task. The local Settings profile after these changes is being measured separately across Notes, Calendar, Mail and Files.
kayg commented 2026-10-01 22:20:47 +00:00
Author
Owner
Copy link

Round 2 production trace after bounded Settings blur

I captured a second source-mapped Chrome CDP trace after the cached sidebar geometry, computed-style-free transitions, deferred theme-color probe, and bounded Settings blur changes. This is the real production build and a warmed Notes → Settings open. Chromium processes were pinned to CPUs 0–1. Local uptime was 15.29, 13.80, 14.84 (1/5/15-minute averages), so this was a contended host run. The Account pane was complete in 601.7 ms. The renderer recorded long tasks of 65 ms at +27.7 ms and 154 ms at +201.4 ms.

Top measured costs, combining 1 ms V8 self-samples with non-overlapping renderer trace spans. The V8 (program) bucket has no script URL. Raster worker time is summed across workers and can overlap main-thread time.

Rank Self time Cost Attribution
1 159.2 ms JavaScript samples V8 (program), no script URL
2 67.5 ms RasterTask raster workers, summed CPU time
3 62.8 ms Commit renderer trace span self time
4 57.2 ms focus browser DOM API
5 45.9 ms UpdateLayoutTree renderer style recalculation
6 17.6 ms fetch browser API
7 13.7 ms get cookie browser DOM/API work
8 12.2 ms Layout renderer layout
9 11.6 ms send browser API
10 9.3 ms garbage collector V8 sample
11 9.0 ms anonymous src/components/IconLabel.svelte:34
12 8.1 ms cloneNode browser DOM API
13 7.0 ms Paint renderer paint
14 6.0 ms anonymous @tiptap/core/dist/index.js:4786
15 4.4 ms setItem browser storage API
16 4.3 ms uo src/lib/overlay/scrollLock.ts:23
17 3.9 ms createElement browser DOM API
18 3.9 ms before browser DOM API
19 3.3 ms anonymous no source URL
20 2.4 ms PrePaint renderer pre-paint

The same trace's largest attributed FunctionCall spans were 66.1 ms in Svelte's src/internal/client/dom/task.js:19 microtask queue and 57.4 ms at src/lib/a11y/focusTrap.ts:231. These spans overlap the V8 samples above, so they are attribution details and are not added again. Other trace self totals were Layerize 1.1 ms and parallel scavenger GC 7.7 ms. The full raw trace remains in the worktree at artifacts/blaze/settings-notes-warm-trace-642.json.

Latest five-run-per-source warm profile, production build, 1440×900 Chromium. Load before the run was 17.65, 14.49, 15.04 (1/5/15-minute averages). The local perf VM was offline.

Source Cold open ms Warm p50/p95 ms Warm browser CPU p50/p95 ms
Notes 1817.90 463.50 / 640.26 530 / 776
Calendar 1182.00 404.80 / 569.92 630 / 762
Mail 852.30 297.00 / 319.82 440 / 488
Files 937.40 364.20 / 404.86 500 / 582

The ≤100 ms p95 target is not met. The bounded blur reduced measured raster worker time from 79.8 ms to 67.5 ms in these individual traces, but the warm timing profile remains far over target under the recorded local load.

Production Settings screenshots from a real temporary server and real Account/Appearance APIs (each is attached to this issue):

Viewport Light Dark
Phone 390×844 phone-paper.png phone-tokyo-night.png
Tablet 820×1000 tablet-paper.png tablet-tokyo-night.png
Desktop 1440×900 desktop-paper.png desktop-tokyo-night.png

Decision where DESIGN is silent: Settings keeps its full scrim tint, but only the glass surface blurs the content beneath it. The surface keeps the same glass treatment and the full page still dims. The goal is to bound repeated raster work without changing the visible overlay material.

### Round 2 production trace after bounded Settings blur I captured a second source-mapped Chrome CDP trace after the cached sidebar geometry, computed-style-free transitions, deferred theme-color probe, and bounded Settings blur changes. This is the real production build and a warmed Notes → Settings open. Chromium processes were pinned to CPUs 0–1. Local `uptime` was `15.29, 13.80, 14.84` (1/5/15-minute averages), so this was a contended host run. The Account pane was complete in **601.7 ms**. The renderer recorded long tasks of **65 ms at +27.7 ms** and **154 ms at +201.4 ms**. Top measured costs, combining 1 ms V8 self-samples with non-overlapping renderer trace spans. The V8 `(program)` bucket has no script URL. Raster worker time is summed across workers and can overlap main-thread time. | Rank | Self time | Cost | Attribution | | ---: | ---: | --- | --- | | 1 | 159.2 ms | JavaScript samples | V8 `(program)`, no script URL | | 2 | 67.5 ms | RasterTask | raster workers, summed CPU time | | 3 | 62.8 ms | Commit | renderer trace span self time | | 4 | 57.2 ms | `focus` | browser DOM API | | 5 | 45.9 ms | UpdateLayoutTree | renderer style recalculation | | 6 | 17.6 ms | `fetch` | browser API | | 7 | 13.7 ms | `get cookie` | browser DOM/API work | | 8 | 12.2 ms | Layout | renderer layout | | 9 | 11.6 ms | `send` | browser API | | 10 | 9.3 ms | garbage collector | V8 sample | | 11 | 9.0 ms | anonymous | `src/components/IconLabel.svelte:34` | | 12 | 8.1 ms | `cloneNode` | browser DOM API | | 13 | 7.0 ms | Paint | renderer paint | | 14 | 6.0 ms | anonymous | `@tiptap/core/dist/index.js:4786` | | 15 | 4.4 ms | `setItem` | browser storage API | | 16 | 4.3 ms | `uo` | `src/lib/overlay/scrollLock.ts:23` | | 17 | 3.9 ms | `createElement` | browser DOM API | | 18 | 3.9 ms | `before` | browser DOM API | | 19 | 3.3 ms | anonymous | no source URL | | 20 | 2.4 ms | PrePaint | renderer pre-paint | The same trace's largest attributed `FunctionCall` spans were **66.1 ms** in Svelte's `src/internal/client/dom/task.js:19` microtask queue and **57.4 ms** at `src/lib/a11y/focusTrap.ts:231`. These spans overlap the V8 samples above, so they are attribution details and are not added again. Other trace self totals were Layerize 1.1 ms and parallel scavenger GC 7.7 ms. The full raw trace remains in the worktree at `artifacts/blaze/settings-notes-warm-trace-642.json`. Latest five-run-per-source warm profile, production build, 1440×900 Chromium. Load before the run was `17.65, 14.49, 15.04` (1/5/15-minute averages). The local perf VM was offline. | Source | Cold open ms | Warm p50/p95 ms | Warm browser CPU p50/p95 ms | | --- | ---: | ---: | ---: | | Notes | 1817.90 | 463.50 / 640.26 | 530 / 776 | | Calendar | 1182.00 | 404.80 / 569.92 | 630 / 762 | | Mail | 852.30 | 297.00 / 319.82 | 440 / 488 | | Files | 937.40 | 364.20 / 404.86 | 500 / 582 | The ≤100 ms p95 target is not met. The bounded blur reduced measured raster worker time from 79.8 ms to 67.5 ms in these individual traces, but the warm timing profile remains far over target under the recorded local load. Production Settings screenshots from a real temporary server and real Account/Appearance APIs (each is attached to this issue): | Viewport | Light | Dark | | --- | --- | --- | | Phone 390×844 | [phone-paper.png](https://git.kayg.org/attachments/46023d6e-4b3f-44e6-a2c2-2d9c6edb7fd1) | [phone-tokyo-night.png](https://git.kayg.org/attachments/d942f868-70c9-4e03-a550-945ce72f299c) | | Tablet 820×1000 | [tablet-paper.png](https://git.kayg.org/attachments/8aee2461-ceac-4fd5-9c28-3a2f7534ebc8) | [tablet-tokyo-night.png](https://git.kayg.org/attachments/1268fcf1-05cc-4f1f-ac5d-7b9770df75a1) | | Desktop 1440×900 | [desktop-paper.png](https://git.kayg.org/attachments/356e099f-9388-4b73-8add-b27814518166) | [desktop-tokyo-night.png](https://git.kayg.org/attachments/61eb76b8-9d8a-4196-97c8-075bf6512ea5) | Decision where DESIGN is silent: Settings keeps its full scrim tint, but only the glass surface blurs the content beneath it. The surface keeps the same glass treatment and the full page still dims. The goal is to bound repeated raster work without changing the visible overlay material.
kayg commented 2026-10-02 00:06:41 +00:00
Author
Owner
Copy link

#642 final profile and report

Head: 98bcdcdc2ffd8a25179e36eec8668ad27804e81c (job/blaze-settings; no push or merge).

Work

  • Recorded a production Chrome tracing profile and JS sampling profile for warm Notes → Settings. The raw trace and five-run profile are attached below.
  • Bounded Settings backdrop blur to the Settings glass plane while keeping the viewport tint. The scrim now disables filters only on shared glass owners, instead of invalidating every Notes descendant.
  • Kept desktop dashboard scroll locking out of body layout. Desktop dashboards use root overflow clipping with a stable scrollbar gutter; touch sheets use the fixed-body lock. Stacked locks keep the stronger strategy and restore the saved styles.
  • Fixed the #641 arrow queue race. One queue consumer now waits for the selected route callback to resolve before another step begins.
  • Captured six production screenshots with real Settings and Note data at phone, tablet and desktop widths in both themes.

Latest Chrome trace

One warm Notes → Settings open took 308.2 ms at load 17.03, 19.55, 17.45, with Chrome pinned to CPUs 0–1. The trace recorded one 75 ms long task, starting 122.7 ms after the open mark. JS sampling covered 327.59 ms; 152.29 ms was idle, leaving about 175.30 ms of non-idle sampled main-thread stacks. The target of about 50 ms main-thread work was not met.

Rank Self time Cost
1 77.41 ms V8 (program); no source URL in the profile
2 37.68 ms compositor RasterTask total
3 20.12 ms UpdateLayoutTree style recalculation
4 15.55 ms native focus; called by focusQuietly in apps/web/src/lib/a11y/inputModality.ts:103
5 9.42 ms native fetch
6 8.31 ms Paint
7 8.21 ms Svelte render.js:258
8 7.17 ms anonymous script frame; no mapped source
9 7.16 ms p in src/index.ts:357
10 6.95 ms IconLabel.svelte:34
11 4.29 ms scrollLock.ts:82
12 3.08 ms native pushState
13 2.99 ms Layerize
14 2.96 ms native getBoundingClientRect
15 2.02 ms window.fetch in src/runtime/client/fetcher.js:70
16 2.00 ms PrePaint
17 1.80 ms Svelte attributes.js:512
18 1.55 ms native before; no mapped source
19 1.50 ms Svelte blocks/if.js:21
20 1.44 ms native removeAttribute

Other trace self times: Layout 0 ms, Commit 0.71 ms, Layerize 2.99 ms, PrePaint 2.00 ms, GC 0 ms. Raster time is on the compositor workers. The prior trace after the scroll-lock change had 46.64 ms of style recalculation; the latest trace has 20.12 ms. These runs used different host loads, so treat that comparison as directional.

Five-run Settings-open profile

Local production build, Chromium 1440×900, pinned to CPUs 0–1. Load average recorded by the profile: 18.19, 19.33, 17.49.

Source Cold open Warm p50 / p95 Browser CPU p50 / p95
Notes 1723.70 ms 588.80 / 646.82 ms 540 / 678 ms
Calendar 1339.50 ms 652.30 / 674.26 ms 470 / 750 ms
Mail 1477.40 ms 424.30 / 514.16 ms 400 / 470 ms
Files 1579.70 ms 468.20 / 633.00 ms 450 / 536 ms

The ≤100 ms warm p95 target is still missed. The perf VM was offline; these are local results under a loaded host, not a quiet-host result.

Production screenshots

Trace: settings-notes-warm-trace-642.json. Five-run profile: settings-open-642-scoped-filter.json.

Decisions not stated in DESIGN

  • Keep the full viewport tint, and apply live blur only to the Settings glass plane.
  • Use root overflow clipping for the desktop dashboard and fixed-body locking for touch sheets; use a stable scrollbar gutter to avoid shifting the desktop page.
  • Limit scrim filter suspension to the shared filter-owner classes listed in the style comment. A future shared blur owner must join that list.
  • Serialize repeated Settings arrows until each route has a complete painted frame. Boundary repeats keep the selected section in place.

Gates

bun run check:

User browser caches use userStorage; only documented device/public-link exceptions remain.
Text sizes and UI shape values use shared role tokens.
UI transitions and animation options use shared motion tokens or documented exceptions.
Loading svelte-check in workspace: /home/kayg/Developer/calternal-wt/blaze-settings/apps/web
Getting Svelte diagnostics...

svelte-check found 0 errors and 0 warnings

bun run test:

Test Files  149 passed (149)
      Tests  1014 passed (1014)
   Start at  01:49:18
   Duration  122.55s (transform 49%, environment 19%, import 16%, tests 11%, setup 4%)

Environment  |component| jsdom was created 46 times · 124.49s total, 28% of tracked time

cargo build -p calternal-server (to embed the production assets): Finished dev profile [unoptimized + debuginfo] target(s) in 2m 53s. No Rust source or API contract changed, so Rust clippy/test gates were not run.

Known gap: the requested warm p95 ≤100 ms and ≤50 ms main-thread work are not achieved. The latest trace and profile remain attached for follow-up.

## #642 final profile and report Head: `98bcdcdc2ffd8a25179e36eec8668ad27804e81c` (`job/blaze-settings`; no push or merge). ### Work - Recorded a production Chrome tracing profile and JS sampling profile for warm Notes → Settings. The raw trace and five-run profile are attached below. - Bounded Settings backdrop blur to the Settings glass plane while keeping the viewport tint. The scrim now disables filters only on shared glass owners, instead of invalidating every Notes descendant. - Kept desktop dashboard scroll locking out of body layout. Desktop dashboards use root overflow clipping with a stable scrollbar gutter; touch sheets use the fixed-body lock. Stacked locks keep the stronger strategy and restore the saved styles. - Fixed the #641 arrow queue race. One queue consumer now waits for the selected route callback to resolve before another step begins. - Captured six production screenshots with real Settings and Note data at phone, tablet and desktop widths in both themes. ### Latest Chrome trace One warm Notes → Settings open took **308.2 ms** at load `17.03, 19.55, 17.45`, with Chrome pinned to CPUs 0–1. The trace recorded one 75 ms long task, starting 122.7 ms after the open mark. JS sampling covered 327.59 ms; 152.29 ms was idle, leaving about 175.30 ms of non-idle sampled main-thread stacks. The target of about 50 ms main-thread work was not met. | Rank | Self time | Cost | | ---: | ---: | --- | | 1 | 77.41 ms | V8 `(program)`; no source URL in the profile | | 2 | 37.68 ms | compositor `RasterTask` total | | 3 | 20.12 ms | `UpdateLayoutTree` style recalculation | | 4 | 15.55 ms | native `focus`; called by `focusQuietly` in `apps/web/src/lib/a11y/inputModality.ts:103` | | 5 | 9.42 ms | native `fetch` | | 6 | 8.31 ms | `Paint` | | 7 | 8.21 ms | Svelte `render.js:258` | | 8 | 7.17 ms | anonymous script frame; no mapped source | | 9 | 7.16 ms | `p` in `src/index.ts:357` | | 10 | 6.95 ms | `IconLabel.svelte:34` | | 11 | 4.29 ms | `scrollLock.ts:82` | | 12 | 3.08 ms | native `pushState` | | 13 | 2.99 ms | `Layerize` | | 14 | 2.96 ms | native `getBoundingClientRect` | | 15 | 2.02 ms | `window.fetch` in `src/runtime/client/fetcher.js:70` | | 16 | 2.00 ms | `PrePaint` | | 17 | 1.80 ms | Svelte `attributes.js:512` | | 18 | 1.55 ms | native `before`; no mapped source | | 19 | 1.50 ms | Svelte `blocks/if.js:21` | | 20 | 1.44 ms | native `removeAttribute` | Other trace self times: `Layout` 0 ms, `Commit` 0.71 ms, `Layerize` 2.99 ms, `PrePaint` 2.00 ms, GC 0 ms. Raster time is on the compositor workers. The prior trace after the scroll-lock change had 46.64 ms of style recalculation; the latest trace has 20.12 ms. These runs used different host loads, so treat that comparison as directional. ### Five-run Settings-open profile Local production build, Chromium 1440×900, pinned to CPUs 0–1. Load average recorded by the profile: `18.19, 19.33, 17.49`. | Source | Cold open | Warm p50 / p95 | Browser CPU p50 / p95 | | --- | ---: | ---: | ---: | | Notes | 1723.70 ms | 588.80 / 646.82 ms | 540 / 678 ms | | Calendar | 1339.50 ms | 652.30 / 674.26 ms | 470 / 750 ms | | Mail | 1477.40 ms | 424.30 / 514.16 ms | 400 / 470 ms | | Files | 1579.70 ms | 468.20 / 633.00 ms | 450 / 536 ms | The ≤100 ms warm p95 target is still missed. The perf VM was offline; these are local results under a loaded host, not a quiet-host result. ### Production screenshots - Phone, light: [phone-paper.png](https://git.kayg.org/attachments/66d5779c-eb31-489c-87ed-e02b89eda4fa) - Phone, dark: [phone-tokyo-night.png](https://git.kayg.org/attachments/e4c0cc53-f171-4fcc-af6e-4189d24b6eb0) - Tablet, light: [tablet-paper.png](https://git.kayg.org/attachments/d5b888ea-ebc3-4000-948f-94ed525408eb) - Tablet, dark: [tablet-tokyo-night.png](https://git.kayg.org/attachments/02b0e92f-23f0-4edf-990c-1480d444f1e3) - Desktop, light: [desktop-paper.png](https://git.kayg.org/attachments/8b7b0e74-cc6b-4524-af9c-0a1a28d533f6) - Desktop, dark: [desktop-tokyo-night.png](https://git.kayg.org/attachments/f851c01d-2ede-4d7d-abd1-2e394be307f4) Trace: [settings-notes-warm-trace-642.json](https://git.kayg.org/attachments/01d670b7-c6e1-49bc-9a2e-cf495052e9b7). Five-run profile: [settings-open-642-scoped-filter.json](https://git.kayg.org/attachments/6302f347-5256-4aad-a183-3a38f37aa8b7). ### Decisions not stated in DESIGN - Keep the full viewport tint, and apply live blur only to the Settings glass plane. - Use root overflow clipping for the desktop dashboard and fixed-body locking for touch sheets; use a stable scrollbar gutter to avoid shifting the desktop page. - Limit scrim filter suspension to the shared filter-owner classes listed in the style comment. A future shared blur owner must join that list. - Serialize repeated Settings arrows until each route has a complete painted frame. Boundary repeats keep the selected section in place. ### Gates `bun run check`: ```text User browser caches use userStorage; only documented device/public-link exceptions remain. Text sizes and UI shape values use shared role tokens. UI transitions and animation options use shared motion tokens or documented exceptions. Loading svelte-check in workspace: /home/kayg/Developer/calternal-wt/blaze-settings/apps/web Getting Svelte diagnostics... svelte-check found 0 errors and 0 warnings ``` `bun run test`: ```text Test Files 149 passed (149) Tests 1014 passed (1014) Start at 01:49:18 Duration 122.55s (transform 49%, environment 19%, import 16%, tests 11%, setup 4%) Environment |component| jsdom was created 46 times · 124.49s total, 28% of tracked time ``` `cargo build -p calternal-server` (to embed the production assets): `Finished dev profile [unoptimized + debuginfo] target(s) in 2m 53s`. No Rust source or API contract changed, so Rust clippy/test gates were not run. Known gap: the requested warm p95 ≤100 ms and ≤50 ms main-thread work are not achieved. The latest trace and profile remain attached for follow-up.
kayg commented 2026-10-02 00:14:53 +00:00
Author
Owner
Copy link

Final head after the documentation audit: 357e2a12c422d8fb948d5c890b338611951856fa.

Final verification after that documentation-only commit:

  • bun run check: svelte-check found 0 errors and 0 warnings
  • git diff --check: passed
  • cargo clean: Removed 7238 files, 4.6GiB total
  • Removed apps/web/build, apps/web/.svelte-kit, and target/tmp after verification.

The production measurements and attached screenshots in the report above remain unchanged. Known gap: the Settings warm-open target (≤100 ms p95 / about 50 ms main-thread work) is still not met in the local pinned run; the perf VM was offline and the local host load was high. No Rust source changed.

Final head after the documentation audit: `357e2a12c422d8fb948d5c890b338611951856fa`. Final verification after that documentation-only commit: - `bun run check`: `svelte-check found 0 errors and 0 warnings` - `git diff --check`: passed - `cargo clean`: `Removed 7238 files, 4.6GiB total` - Removed `apps/web/build`, `apps/web/.svelte-kit`, and `target/tmp` after verification. The production measurements and attached screenshots in the report above remain unchanged. Known gap: the Settings warm-open target (≤100 ms p95 / about 50 ms main-thread work) is still not met in the local pinned run; the perf VM was offline and the local host load was high. No Rust source changed.
kayg commented 2026-10-02 00:44:06 +00:00
Author
Owner
Copy link

Round 3 started on job/blaze-settings, based on origin/dev at 687ff703136e71e89f8dfba139e93cd0788b25c1 (current head 357e2a12c422d8fb948d5c890b338611951856fa). I am tracing a source-mapped production build, then splitting Settings first paint from deferred content and checking the #641 repeat-key contract. I will record CPU pinning, uptime, frame results, and gates here when complete.

Round 3 started on `job/blaze-settings`, based on `origin/dev` at `687ff703136e71e89f8dfba139e93cd0788b25c1` (current head `357e2a12c422d8fb948d5c890b338611951856fa`). I am tracing a source-mapped production build, then splitting Settings first paint from deferred content and checking the #641 repeat-key contract. I will record CPU pinning, uptime, frame results, and gates here when complete.
kayg commented 2026-10-02 01:24:57 +00:00
Author
Owner
Copy link

Temporary source-map production profile (cd apps/web && bun run vite build --sourcemap; maps remain in ignored build output) ran on Chromium 153 pinned to CPUs 0–1 at 1440×900. Host uptime: 03:15:23 up 4 days 20:27, 0 users, load average: 22.62, 24.25, 24.12. This is a local, noisy baseline. Warm Notes → Settings readiness was 608.5 ms in this run; the registered first shell/content frame was 608.2 ms.

The resolved top 20 CPU sample locations (ms) were: packages/api-client/src/index.ts:357 p 11.24; Svelte dom/elements/class.js:14 Sa 7.54; Svelte runtime.js:548 get 6.73; src/lib/overlay/scrollLock.ts:82 ho 5.12; SvelteKit runtime/client/fetcher.js:70 window.fetch 4.90; src/routes/settings/api.svelte.ts:83 E 3.43; Svelte DOM events 181 li 1.94; Svelte bindings/this.js:42 anonymous 1.83; a remaining minified generated chunk uBIymjUX.js:1:489 anonymous 1.79; packages/ui/src/components/FloatingSidebar.svelte:186 D 1.76; src/routes/settings/account/StorageGroup.svelte:66 anonymous 1.75; packages/api-client/src/index.ts:362 m 1.51; Svelte proxy 70 f 1.30; SvelteKit client 1647 It 1.29; packages/ui/src/components/Pill.svelte:132 anonymous 1.13; Svelte runtime 225 Ar 1.11; packages/ui/src/components/LinkedHeading.svelte:134 get label 0.92; src/lib/navigation.ts:82 Fl 0.89; Svelte batch 329 #_ 0.81; src/routes/settings/parts/SettingsCard.svelte:90 children 0.75.

The Long Animation Frame record is more actionable: 230 ms total; the import continuation maps to generated/client-optimized/app.js:58 for 36.1 ms, then the reveal FrameRequestCallback maps to packages/ui/src/components/OverlaySurface.svelte:382 for 113.8 ms, including 45.5 ms forced style/layout. At that line present = true mounts all seven Account groups in one Svelte flush. CPU Profile's synthetic (program) frame has scriptId=0, no URL and no source map; its 612.65 ms cannot be mapped directly. Stack-associated frames were mapped above and the attributed focus path resolves to src/lib/a11y/inputModality.ts:103 (55.97 ms). This is retained as profiler limitation rather than misattributed to a component.

Temporary source-map production profile (`cd apps/web && bun run vite build --sourcemap`; maps remain in ignored build output) ran on Chromium 153 pinned to CPUs 0–1 at 1440×900. Host `uptime`: `03:15:23 up 4 days 20:27, 0 users, load average: 22.62, 24.25, 24.12`. This is a local, noisy baseline. Warm Notes → Settings readiness was 608.5 ms in this run; the registered first shell/content frame was 608.2 ms. The resolved top 20 CPU sample locations (ms) were: `packages/api-client/src/index.ts:357 p 11.24`; Svelte `dom/elements/class.js:14 Sa 7.54`; Svelte `runtime.js:548 get 6.73`; `src/lib/overlay/scrollLock.ts:82 ho 5.12`; SvelteKit `runtime/client/fetcher.js:70 window.fetch 4.90`; `src/routes/settings/api.svelte.ts:83 E 3.43`; Svelte DOM events `181 li 1.94`; Svelte `bindings/this.js:42 anonymous 1.83`; a remaining minified generated chunk `uBIymjUX.js:1:489 anonymous 1.79`; `packages/ui/src/components/FloatingSidebar.svelte:186 D 1.76`; `src/routes/settings/account/StorageGroup.svelte:66 anonymous 1.75`; `packages/api-client/src/index.ts:362 m 1.51`; Svelte proxy `70 f 1.30`; SvelteKit client `1647 It 1.29`; `packages/ui/src/components/Pill.svelte:132 anonymous 1.13`; Svelte runtime `225 Ar 1.11`; `packages/ui/src/components/LinkedHeading.svelte:134 get label 0.92`; `src/lib/navigation.ts:82 Fl 0.89`; Svelte batch `329 #_ 0.81`; `src/routes/settings/parts/SettingsCard.svelte:90 children 0.75`. The Long Animation Frame record is more actionable: 230 ms total; the import continuation maps to `generated/client-optimized/app.js:58` for 36.1 ms, then the reveal `FrameRequestCallback` maps to `packages/ui/src/components/OverlaySurface.svelte:382` for 113.8 ms, including 45.5 ms forced style/layout. At that line `present = true` mounts all seven Account groups in one Svelte flush. CPU Profile's synthetic `(program)` frame has `scriptId=0`, no URL and no source map; its 612.65 ms cannot be mapped directly. Stack-associated frames were mapped above and the attributed `focus` path resolves to `src/lib/a11y/inputModality.ts:103` (55.97 ms). This is retained as profiler limitation rather than misattributed to a component.
kayg commented 2026-10-02 01:58:39 +00:00
Author
Owner
Copy link

Local Chromium (CPU affinity 0–1; host uptime at probe start: 03:56:37 up 4 days 21:09, 0 users, load average: 18.75, 19.12, 18.28) has the signed-in User cache before opening Settings. The real Notes route probe reports cold first Settings open: shell 352 ms, heading/first two interactive cards 396 ms; warm repeat: shell 231 ms, heading/first two cards 245 ms. Both first frames had two Account groups and an enabled control. Mounting and loading the remaining five groups took 2,628 ms in the cold run. This is a local, noisy host measurement. The idle split reduced mounted Account groups, but the warm opening still exceeds 100 ms; I am attributing the route/reveal task with the source-map build before deciding the next code change.

Local Chromium (CPU affinity 0–1; host uptime at probe start: `03:56:37 up 4 days 21:09, 0 users, load average: 18.75, 19.12, 18.28`) has the signed-in User cache before opening Settings. The real Notes route probe reports cold first Settings open: shell 352 ms, heading/first two interactive cards 396 ms; warm repeat: shell 231 ms, heading/first two cards 245 ms. Both first frames had two Account groups and an enabled control. Mounting and loading the remaining five groups took 2,628 ms in the cold run. This is a local, noisy host measurement. The idle split reduced mounted Account groups, but the warm opening still exceeds 100 ms; I am attributing the route/reveal task with the source-map build before deciding the next code change.
kayg commented 2026-10-02 02:03:17 +00:00
Author
Owner
Copy link

Source-map profiling result for #642. This was a temporary production build from vite build --sourcemap; maps are for profiling only and will not ship. Chromium 153 was pinned to CPUs 0–1. At the profile run, uptime was 03:59:54 up 4 days 21:12, 0 users, load average: 24.10, 20.46, 18.87.

Warm Notes → Settings: shell frame 414.8 ms, Account content frame 414.9 ms, ready frame 420.2 ms, 2 Account groups mounted. The opening task observer saw 56 ms, 50 ms and 102 ms tasks before that frame. Entrance-end was not captured by this trace. A later focused open test is needed for animation-tail separation.

Top CPU frames (ms, source-map original positions where available):

  1. (program) — 529.36; profiler scriptId 0 has no URL or source-map location.
  2. svelte/internal/client/dom/blocks/branches.js:76 #a — 10.86
  3. (program):175 serialize — 8.77; minified pseudo-frame has no map entry.
  4. svelte/internal/client/runtime.js:225 Ar — 8.56
  5. apps/web/src/lib/overlay/scrollLock.ts:82 ho — 8.25
  6. svelte/internal/client/reactivity/props.js:262 xo — 4.39
  7. packages/ui/src/components/FloatingSidebar.svelte:285 get title — 3.30
  8. svelte/internal/client/reactivity/props.js:237 ownKeys — 3.02
  9. apps/web/src/routes/settings/[...path]/+page.svelte:261 (anonymous) — 2.95
  10. svelte/internal/client/dom/operations.js:81 L — 2.85
  11. svelte/internal/client/dom/blocks/if.js:36 o — 2.53
  12. svelte/internal/client/dom/elements/transitions.js:338 la — 2.43
  13. apps/web/src/lib/userStorage.ts:15 n — 2.37
  14. svelte/internal/client/reactivity/props.js:81 ownKeys — 2.35
  15. svelte/internal/client/dom/blocks/each.js:266 (anonymous) — 1.95
  16. lib0/encoding.js:130 HT — 1.87
  17. packages/ui/src/components/OverlaySurface.svelte:10 rs — 1.81
  18. svelte/internal/client/reactivity/sources.js:354 Cn — 1.64
  19. packages/ui/src/components/CopyLink.svelte:84 S — 1.53
  20. svelte/internal/client/dom/elements/attributes.js:512 (anonymous) — 1.47

(program) is a V8 pseudo-frame (scriptId 0), so source maps cannot assign it to a Svelte function. Its nearest mapped stack-associated cost was AppToaster.svelte:74 getBoundingClientRect at 32.79 ms; other mapped associations included Svelte attribute updates (9.46 ms), Svelte transition geometry (9.20 ms), and SvelteKit pushState (8.06 ms). The profile maps the 36 ms pre-paint focus cost from the previous trace to 1.87 ms in SvelteKit focus here after the focus deferral change.

Source-map profiling result for #642. This was a temporary production build from `vite build --sourcemap`; maps are for profiling only and will not ship. Chromium 153 was pinned to CPUs 0–1. At the profile run, `uptime` was `03:59:54 up 4 days 21:12, 0 users, load average: 24.10, 20.46, 18.87`. Warm Notes → Settings: shell frame 414.8 ms, Account content frame 414.9 ms, ready frame 420.2 ms, 2 Account groups mounted. The opening task observer saw 56 ms, 50 ms and 102 ms tasks before that frame. Entrance-end was not captured by this trace. A later focused open test is needed for animation-tail separation. Top CPU frames (ms, source-map original positions where available): 1. `(program)` — 529.36; profiler scriptId 0 has no URL or source-map location. 2. `svelte/internal/client/dom/blocks/branches.js:76 #a` — 10.86 3. `(program):175 serialize` — 8.77; minified pseudo-frame has no map entry. 4. `svelte/internal/client/runtime.js:225 Ar` — 8.56 5. `apps/web/src/lib/overlay/scrollLock.ts:82 ho` — 8.25 6. `svelte/internal/client/reactivity/props.js:262 xo` — 4.39 7. `packages/ui/src/components/FloatingSidebar.svelte:285 get title` — 3.30 8. `svelte/internal/client/reactivity/props.js:237 ownKeys` — 3.02 9. `apps/web/src/routes/settings/[...path]/+page.svelte:261 (anonymous)` — 2.95 10. `svelte/internal/client/dom/operations.js:81 L` — 2.85 11. `svelte/internal/client/dom/blocks/if.js:36 o` — 2.53 12. `svelte/internal/client/dom/elements/transitions.js:338 la` — 2.43 13. `apps/web/src/lib/userStorage.ts:15 n` — 2.37 14. `svelte/internal/client/reactivity/props.js:81 ownKeys` — 2.35 15. `svelte/internal/client/dom/blocks/each.js:266 (anonymous)` — 1.95 16. `lib0/encoding.js:130 HT` — 1.87 17. `packages/ui/src/components/OverlaySurface.svelte:10 rs` — 1.81 18. `svelte/internal/client/reactivity/sources.js:354 Cn` — 1.64 19. `packages/ui/src/components/CopyLink.svelte:84 S` — 1.53 20. `svelte/internal/client/dom/elements/attributes.js:512 (anonymous)` — 1.47 `(program)` is a V8 pseudo-frame (scriptId 0), so source maps cannot assign it to a Svelte function. Its nearest mapped stack-associated cost was `AppToaster.svelte:74 getBoundingClientRect` at 32.79 ms; other mapped associations included Svelte attribute updates (9.46 ms), Svelte transition geometry (9.20 ms), and SvelteKit `pushState` (8.06 ms). The profile maps the 36 ms pre-paint focus cost from the previous trace to 1.87 ms in SvelteKit focus here after the focus deferral change.
kayg commented 2026-10-02 02:30:26 +00:00
Author
Owner
Copy link

Updated #642 source-map profile from the current production source. This used a temporary vite build --sourcemap; source maps are profiling-only and will not ship. Chromium 153 was pinned to CPUs 0–1. Start uptime: 04:27:51 up 4 days 21:40, 0 users, load average: 21.58, 22.93, 20.07.

Warm Notes → Settings: shell frame 294.6 ms, Profile content frame 294.7 ms, ready 296.9 ms; the first frame contains one Account group. One 62 ms long task ran before that frame. The entrance end is still not captured by the profiler; I am correcting the E2E animation sampler separately.

Top 20 CPU frames (ms; original source positions when maps resolve them):

  1. (program) — 309.68; V8 pseudo-frame, scriptId 0, no source location.
  2. (program):186 innerSerialize — 10.92; generated pseudo-frame, no map entry.
  3. apps/web/src/lib/overlay/scrollLock.ts:82 — 5.41
  4. svelte/internal/client/reactivity/sources.js:189 — 3.85
  5. packages/ui/src/components/Pill.svelte:2 — 3.44
  6. (program):29 sample — 2.90; profiler sampler frame, no map entry.
  7. svelte/internal/client/dom/blocks/branches.js:186 ensure — 2.76
  8. svelte/internal/client/dom/blocks/svelte-element.js:31 — 2.69
  9. svelte/internal/client/reactivity/sources.js:354 — 2.36
  10. svelte/internal/client/runtime.js:548 — 1.85
  11. svelte/internal/client/dom/template.js:335 — 1.82
  12. svelte/internal/client/dom/elements/style.js:31 — 1.79
  13. svelte/internal/client/reactivity/props.js:82 — 1.76
  14. packages/api-client/src/index.ts:357 — 1.71
  15. svelte/internal/client/reactivity/props.js:226 — 1.55
  16. svelte/internal/client/dom/elements/attributes.js:158 — 1.38
  17. @sveltejs/kit/runtime/client/client.js:1712 — 1.36
  18. packages/ui/src/components/FloatingSidebar.svelte:169 — 1.22
  19. @sveltejs/kit/runtime/client/utils.js:26 — 1.21
  20. apps/web/src/lib/appearance/background.svelte.ts:472 — 1.15

Stack-associated (program) samples remain unlocatable as functions. The largest mapped associations were SvelteKit fetch 14.19 ms, AppToaster.svelte:74 getBoundingClientRect 13.17 ms, and userStorage.ts:62 setItem 10.18 ms. The deferred focus work sampled 1.89 ms in inputModality.ts:103 and 0.85 ms in SvelteKit focus.

Updated #642 source-map profile from the current production source. This used a temporary `vite build --sourcemap`; source maps are profiling-only and will not ship. Chromium 153 was pinned to CPUs 0–1. Start `uptime`: `04:27:51 up 4 days 21:40, 0 users, load average: 21.58, 22.93, 20.07`. Warm Notes → Settings: shell frame 294.6 ms, Profile content frame 294.7 ms, ready 296.9 ms; the first frame contains one Account group. One 62 ms long task ran before that frame. The entrance end is still not captured by the profiler; I am correcting the E2E animation sampler separately. Top 20 CPU frames (ms; original source positions when maps resolve them): 1. `(program)` — 309.68; V8 pseudo-frame, scriptId 0, no source location. 2. `(program):186 innerSerialize` — 10.92; generated pseudo-frame, no map entry. 3. `apps/web/src/lib/overlay/scrollLock.ts:82` — 5.41 4. `svelte/internal/client/reactivity/sources.js:189` — 3.85 5. `packages/ui/src/components/Pill.svelte:2` — 3.44 6. `(program):29 sample` — 2.90; profiler sampler frame, no map entry. 7. `svelte/internal/client/dom/blocks/branches.js:186 ensure` — 2.76 8. `svelte/internal/client/dom/blocks/svelte-element.js:31` — 2.69 9. `svelte/internal/client/reactivity/sources.js:354` — 2.36 10. `svelte/internal/client/runtime.js:548` — 1.85 11. `svelte/internal/client/dom/template.js:335` — 1.82 12. `svelte/internal/client/dom/elements/style.js:31` — 1.79 13. `svelte/internal/client/reactivity/props.js:82` — 1.76 14. `packages/api-client/src/index.ts:357` — 1.71 15. `svelte/internal/client/reactivity/props.js:226` — 1.55 16. `svelte/internal/client/dom/elements/attributes.js:158` — 1.38 17. `@sveltejs/kit/runtime/client/client.js:1712` — 1.36 18. `packages/ui/src/components/FloatingSidebar.svelte:169` — 1.22 19. `@sveltejs/kit/runtime/client/utils.js:26` — 1.21 20. `apps/web/src/lib/appearance/background.svelte.ts:472` — 1.15 Stack-associated `(program)` samples remain unlocatable as functions. The largest mapped associations were SvelteKit `fetch` 14.19 ms, `AppToaster.svelte:74 getBoundingClientRect` 13.17 ms, and `userStorage.ts:62 setItem` 10.18 ms. The deferred focus work sampled 1.89 ms in `inputModality.ts:103` and 0.85 ms in SvelteKit focus.
kayg commented 2026-10-02 03:23:56 +00:00
Author
Owner
Copy link

Round 3 source-map attribution for #642. I rebuilt with bun run vite build --sourcemap for profiling only; maps are temporary and will not ship. Chromium 153.0.8010.12 was pinned to CPUs 0-1. Start uptime: 05:17:56 up 4 days 22:30, 0 users, load average: 10.22, 10.84, 15.85.

The instrumented ready frame was 252.5 ms (shell/content frame 250.8/250.9 ms) and included a 101 ms Long Task starting at 113.9 ms. The uninstrumented settings-open-642 run on the same build measured warm content at 187.9 ms, with a 0 ms first-frame Long Task. It still misses the 100 ms content target by 87.9 ms. Its animation capture saw prefers-reduced-motion: false, but the only surface Animation was already finished with duration 0 ms. The measured 187.9 ms therefore has 0 ms observed entrance animation and 187.9 ms of route/mount/frame scheduling work; this is a 0/100 split for animation/work in this run. The focus stacks are down to about 1.19 ms each for inputModality.ts:103 focus and SvelteKit client.js:3232 focus.

Resolved top 20 sampled CPU frames:

# Source CPU ms
1 V8 (program) pseudo-frame, scriptId 0; no source-map URL 343.77
2 V8 (program) line 47, column 48, anonymous; unmapped 10.64
3 Svelte internal/client/reactivity/effects.js:471 sr 8.20
4 Svelte internal/client/dom/operations.js:195 In 7.48
5 V8 (program) line 175 serialize; unmapped 6.97
6 apps/web/src/lib/actions/magnetic.ts:186 destroy 4.64
7 packages/ui/src/actions/portal.ts:13 3.96
8 apps/web/src/lib/overlay/scrollLock.ts:82 ho 3.51
9 SvelteKit runtime/client/client.js:1647 It 2.67
10 SvelteKit runtime/client/client.js:895 Tt 1.33
11 SvelteKit runtime/client/client.js:1712 Z 1.11
12 packages/ui/src/components/OverlaySurface.svelte:382 0.90
13 Generated chunk uBIymjUX.js:1:489; unmapped 0.79
14 apps/web/src/lib/overlay/scrollLock.ts:57 mo 0.63
15 apps/web/generated/client-optimized/app.js:11 0.61
16 SvelteKit runtime/client/fetcher.js:70 window.fetch 0.60
17 apps/web/src/lib/shortcuts/registry.ts:352 pn 0.60
18 packages/ui/src/components/UIHeading.svelte:26 0.59
19 Svelte internal/client/dom/elements/transitions.js:417 h.onfinish 0.59
20 V8 (program) line 78, column 40, anonymous; unmapped 0.59

Source maps resolve named application and Svelte frames, but V8's scriptId 0 (program) samples have no script URL or map. The profile includes 213.30 ms of those pseudo-frame samples in the separate programFrames aggregation and 318.15 ms with no mapped parent in its stack aggregation; it cannot truthfully assign that cost to a component. The first-frame tree now has one Profile group; the other six Account cards and live User verification are queued after the first paint.

Round 3 source-map attribution for #642. I rebuilt with `bun run vite build --sourcemap` for profiling only; maps are temporary and will not ship. Chromium 153.0.8010.12 was pinned to CPUs 0-1. Start `uptime`: `05:17:56 up 4 days 22:30, 0 users, load average: 10.22, 10.84, 15.85`. The instrumented ready frame was 252.5 ms (shell/content frame 250.8/250.9 ms) and included a 101 ms Long Task starting at 113.9 ms. The uninstrumented `settings-open-642` run on the same build measured warm content at 187.9 ms, with a 0 ms first-frame Long Task. It still misses the 100 ms content target by 87.9 ms. Its animation capture saw `prefers-reduced-motion: false`, but the only surface Animation was already finished with duration 0 ms. The measured 187.9 ms therefore has 0 ms observed entrance animation and 187.9 ms of route/mount/frame scheduling work; this is a 0/100 split for animation/work in this run. The focus stacks are down to about 1.19 ms each for `inputModality.ts:103 focus` and SvelteKit `client.js:3232 focus`. Resolved top 20 sampled CPU frames: | # | Source | CPU ms | | ---: | --- | ---: | | 1 | V8 `(program)` pseudo-frame, scriptId 0; no source-map URL | 343.77 | | 2 | V8 `(program)` line 47, column 48, anonymous; unmapped | 10.64 | | 3 | Svelte `internal/client/reactivity/effects.js:471 sr` | 8.20 | | 4 | Svelte `internal/client/dom/operations.js:195 In` | 7.48 | | 5 | V8 `(program)` line 175 `serialize`; unmapped | 6.97 | | 6 | `apps/web/src/lib/actions/magnetic.ts:186 destroy` | 4.64 | | 7 | `packages/ui/src/actions/portal.ts:13` | 3.96 | | 8 | `apps/web/src/lib/overlay/scrollLock.ts:82 ho` | 3.51 | | 9 | SvelteKit `runtime/client/client.js:1647 It` | 2.67 | | 10 | SvelteKit `runtime/client/client.js:895 Tt` | 1.33 | | 11 | SvelteKit `runtime/client/client.js:1712 Z` | 1.11 | | 12 | `packages/ui/src/components/OverlaySurface.svelte:382` | 0.90 | | 13 | Generated chunk `uBIymjUX.js:1:489`; unmapped | 0.79 | | 14 | `apps/web/src/lib/overlay/scrollLock.ts:57 mo` | 0.63 | | 15 | `apps/web/generated/client-optimized/app.js:11` | 0.61 | | 16 | SvelteKit `runtime/client/fetcher.js:70 window.fetch` | 0.60 | | 17 | `apps/web/src/lib/shortcuts/registry.ts:352 pn` | 0.60 | | 18 | `packages/ui/src/components/UIHeading.svelte:26` | 0.59 | | 19 | Svelte `internal/client/dom/elements/transitions.js:417 h.onfinish` | 0.59 | | 20 | V8 `(program)` line 78, column 40, anonymous; unmapped | 0.59 | Source maps resolve named application and Svelte frames, but V8's `scriptId 0` `(program)` samples have no script URL or map. The profile includes 213.30 ms of those pseudo-frame samples in the separate `programFrames` aggregation and 318.15 ms with no mapped parent in its stack aggregation; it cannot truthfully assign that cost to a component. The first-frame tree now has one Profile group; the other six Account cards and live User verification are queued after the first paint.
kayg commented 2026-10-02 03:40:15 +00:00
Author
Owner
Copy link

Updated #642 opening measurement with a requestAnimationFrame sampler. The prior MutationObserver sampler inspected the Settings tree on every DOM mutation and could distort the opening flush.

Pinned local run: Chromium and the harness on CPUs 0–1. uptime: 05:33:55 up 4 days 22:46, 0 users, load average: 10.26, 16.12, 16.46.

Warm Notes → Settings: shell 376.4 ms, heading and first content frame 377.4 ms, one of seven Account groups mounted, first frame interactive. The frame capture found 66 ms and 99 ms long tasks before first content. The ≤100 ms content and ≤50 ms first-task targets still fail. Cold content appeared at 1103.9 ms, with a 546 ms task before it.

Motion capture: prefers-reduced-motion=false; one entrance animation with 200 ms computed duration. It starts in the first content frame, so observed pre-content time is work and scheduling (0% entrance, 100% work/scheduling). The animation finished at 816.4 ms, 439 ms after content appeared on this loaded host. The content is visible and interactive while it runs.

This sampler result replaces the earlier optimistic 187.9 ms estimate. The high local load and the failed 100 ms target remain gaps; I have not treated them as passing.

Updated #642 opening measurement with a requestAnimationFrame sampler. The prior MutationObserver sampler inspected the Settings tree on every DOM mutation and could distort the opening flush. Pinned local run: Chromium and the harness on CPUs 0–1. `uptime`: `05:33:55 up 4 days 22:46, 0 users, load average: 10.26, 16.12, 16.46`. Warm Notes → Settings: shell 376.4 ms, heading and first content frame 377.4 ms, one of seven Account groups mounted, first frame interactive. The frame capture found 66 ms and 99 ms long tasks before first content. The ≤100 ms content and ≤50 ms first-task targets still fail. Cold content appeared at 1103.9 ms, with a 546 ms task before it. Motion capture: `prefers-reduced-motion=false`; one entrance animation with 200 ms computed duration. It starts in the first content frame, so observed pre-content time is work and scheduling (0% entrance, 100% work/scheduling). The animation finished at 816.4 ms, 439 ms after content appeared on this loaded host. The content is visible and interactive while it runs. This sampler result replaces the earlier optimistic 187.9 ms estimate. The high local load and the failed 100 ms target remain gaps; I have not treated them as passing.
kayg commented 2026-10-02 05:06:39 +00:00
Author
Owner
Copy link

Round 3 complete. Branch: job/blaze-settings; merged origin/dev once at c4a61e8cf; head: bf3ad5f29d1108c34e87c0e9d20b1dbe67ffc5c3.

Built

  • Settings now paints the overlay shell and Account Profile group first. Retained panes and the remaining Account groups use one shared idle-slice queue. User verification remains default-deny until its background result arrives.
  • Rail highlight and selected content follow the latest repeated key immediately; only URL writes serialize. Opening focus runs after the first painted frame with preventScroll. Toast placement geometry reads also wait until the first frame.
  • Added real-production open, frame-repeat and six-size/theme review probes under apps/web/e2e/, plus the shared bench/blaze.mjs profile.

Round 3 measurements

Chromium and its harness were pinned to CPUs 0–1. The low-overhead frame capture recorded uptime: 05:33:55 up 4 days 22:46, 0 users, load average: 10.26, 16.12, 16.46.

Open Shell frame First content frame Groups mounted then Long tasks before content
Warm Notes → Settings 376.4 ms 377.4 ms 1 / 7 66 ms, 99 ms
Cold Notes → Settings — 1103.9 ms — 546 ms

The ≤100 ms content target and ≤50 ms first-task target still fail. The 200 ms entrance animation starts in the first content frame, so it accounts for 0% of the pre-content delay; work and scheduling account for 100%. Content is visible and interactive while it runs. The animation ended at 816.4 ms, 439 ms after warm content appeared on this loaded host.

The source-map top-20 is posted earlier in this issue. It resolves Svelte, SvelteKit and app frames. V8's largest (program) frame (343.77 ms, scriptId 0) has no script URL or source-map location, so it remains unattributed. The updated profile puts each focus stack around 1 ms; focus now occurs after first paint.

The 15 ms warm arrow trace had zero highlight-lag, content-lag, mismatch or incomplete frames. The cold trace had 21 incomplete frames; see the #641 closeout. This does not make the cold path pass.

Production screenshots

The merged production build showed all seven live Account groups in all six captures. Captures are attached here:

Gates

  • cargo fmt --check: exit 0; no output.
  • cargo clippy -p calternal-plugin --all-targets -- -D warnings: passed, exit 0.
  • cargo clippy -p calternal-server --all-targets -- -D warnings: passed, exit 0.
  • cargo clippy -p calternal-plugin-calendar --all-targets -- -D warnings: passed, exit 0.
  • cargo clippy -p calternal-plugin-files --all-targets -- -D warnings: passed, exit 0.
  • cargo clippy -p calternal-plugin-mail --all-targets -- -D warnings: passed, exit 0.
  • cargo clippy -p calternal-plugin-notes --all-targets -- -D warnings: passed, exit 0.
  • Rust tests: calternal-plugin 25 passed; calternal-server 107 passed, 3 ignored; calendar 83 unit + 1 cache integration + 3 protocol tests passed; files 146 passed, 1 ignored; mail 45 passed, 2 ignored. The Notes suite reported 166 passed and 1 failed: daily_and_composer_preserve_unrelated_bytes received 404 instead of 200. The isolated test passed (1 passed, 166 filtered out). I left the existing expectation and Notes code unchanged.
  • bun run check: 
    User browser caches use userStorage; only documented device/public-link exceptions remain.
Text sizes and UI shape values use shared role tokens.
UI transitions and animation options use shared motion tokens or documented exceptions.
Loading svelte-check in workspace: /home/kayg/Developer/calternal-wt/blaze-settings/apps/web
Getting Svelte diagnostics...

svelte-check found 0 errors and 0 warnings
  • bun run test: 
     Test Files  154 passed (154)
      Tests  1056 passed (1056)
   Start at  06:33:31
   Duration  126.00s (transform 55%, environment 17%, import 15%, tests 10%, setup 3%)

  Transform  |component| transforming modules took 243.53s · 49% of tracked time, re-done on every run
             persist transforms across runs with fsModuleCache: true
             learn more: https://vitest.dev/guide/improving-performance#caching-between-reruns

One API-only adversarial round kept the server alive and reported no server crash or 5xx. The non-SLOW observations and the API-only selector's missing MCP fixture are recorded in #664. Most other findings were marked SLOW under the shared host load. I did not change any existing test expectation.

Decisions not specified in DESIGN

  • Use one shared Settings mount queue: yield after a painted frame, use scheduler.yield() / requestIdleCallback where available, then a timer fallback. Keep Account Profile synchronous and keep Admin hidden until authorization is verified.
  • Defer overlay opening focus by one painted frame and call focus({ preventScroll: true }).
  • Follow the 2026-10-01 owner override: keyboard actions animate like pointer actions; reduced-motion still applies.

Known gap: warm content is still 377.4 ms, and cold content is 1103.9 ms. The largest source-map pseudo-frame remains unresolved.

Round 3 complete. Branch: `job/blaze-settings`; merged `origin/dev` once at `c4a61e8cf`; head: `bf3ad5f29d1108c34e87c0e9d20b1dbe67ffc5c3`. ## Built - Settings now paints the overlay shell and Account Profile group first. Retained panes and the remaining Account groups use one shared idle-slice queue. User verification remains default-deny until its background result arrives. - Rail highlight and selected content follow the latest repeated key immediately; only URL writes serialize. Opening focus runs after the first painted frame with `preventScroll`. Toast placement geometry reads also wait until the first frame. - Added real-production open, frame-repeat and six-size/theme review probes under `apps/web/e2e/`, plus the shared `bench/blaze.mjs` profile. ## Round 3 measurements Chromium and its harness were pinned to CPUs 0–1. The low-overhead frame capture recorded `uptime`: `05:33:55 up 4 days 22:46, 0 users, load average: 10.26, 16.12, 16.46`. | Open | Shell frame | First content frame | Groups mounted then | Long tasks before content | | --- | ---: | ---: | ---: | --- | | Warm Notes → Settings | 376.4 ms | 377.4 ms | 1 / 7 | 66 ms, 99 ms | | Cold Notes → Settings | — | 1103.9 ms | — | 546 ms | The ≤100 ms content target and ≤50 ms first-task target still fail. The 200 ms entrance animation starts in the first content frame, so it accounts for 0% of the pre-content delay; work and scheduling account for 100%. Content is visible and interactive while it runs. The animation ended at 816.4 ms, 439 ms after warm content appeared on this loaded host. The source-map top-20 is posted earlier in this issue. It resolves Svelte, SvelteKit and app frames. V8's largest `(program)` frame (343.77 ms, `scriptId 0`) has no script URL or source-map location, so it remains unattributed. The updated profile puts each focus stack around 1 ms; focus now occurs after first paint. The 15 ms warm arrow trace had zero highlight-lag, content-lag, mismatch or incomplete frames. The cold trace had 21 incomplete frames; see the #641 closeout. This does not make the cold path pass. ## Production screenshots The merged production build showed all seven live Account groups in all six captures. Captures are attached here: - Phone light: https://git.kayg.org/attachments/282abd19-70c0-4190-994b-370ccc5d3118 - Phone dark: https://git.kayg.org/attachments/7d9875db-3d18-4a9c-96d7-b6b0acbf6dc2 - Tablet light: https://git.kayg.org/attachments/597ab3f2-22c0-4963-abc2-ca46b5fe1d4d - Tablet dark: https://git.kayg.org/attachments/4207908d-a956-4cf3-a593-cc38f3246b36 - Desktop light: https://git.kayg.org/attachments/d3d4f4cd-7163-48ca-9317-a12e8dddbba7 - Desktop dark: https://git.kayg.org/attachments/0909ec7e-7686-43c1-913b-634d59967a06 ## Gates - `cargo fmt --check`: exit 0; no output. - `cargo clippy -p calternal-plugin --all-targets -- -D warnings`: passed, exit 0. - `cargo clippy -p calternal-server --all-targets -- -D warnings`: passed, exit 0. - `cargo clippy -p calternal-plugin-calendar --all-targets -- -D warnings`: passed, exit 0. - `cargo clippy -p calternal-plugin-files --all-targets -- -D warnings`: passed, exit 0. - `cargo clippy -p calternal-plugin-mail --all-targets -- -D warnings`: passed, exit 0. - `cargo clippy -p calternal-plugin-notes --all-targets -- -D warnings`: passed, exit 0. - Rust tests: `calternal-plugin` 25 passed; `calternal-server` 107 passed, 3 ignored; calendar 83 unit + 1 cache integration + 3 protocol tests passed; files 146 passed, 1 ignored; mail 45 passed, 2 ignored. The Notes suite reported 166 passed and 1 failed: `daily_and_composer_preserve_unrelated_bytes` received 404 instead of 200. The isolated test passed (1 passed, 166 filtered out). I left the existing expectation and Notes code unchanged. - `bun run check`: ``` User browser caches use userStorage; only documented device/public-link exceptions remain. Text sizes and UI shape values use shared role tokens. UI transitions and animation options use shared motion tokens or documented exceptions. Loading svelte-check in workspace: /home/kayg/Developer/calternal-wt/blaze-settings/apps/web Getting Svelte diagnostics... svelte-check found 0 errors and 0 warnings ``` - `bun run test`: ``` Test Files 154 passed (154) Tests 1056 passed (1056) Start at 06:33:31 Duration 126.00s (transform 55%, environment 17%, import 15%, tests 10%, setup 3%) Transform |component| transforming modules took 243.53s · 49% of tracked time, re-done on every run persist transforms across runs with fsModuleCache: true learn more: https://vitest.dev/guide/improving-performance#caching-between-reruns ``` One API-only adversarial round kept the server alive and reported no server crash or 5xx. The non-SLOW observations and the API-only selector's missing MCP fixture are recorded in #664. Most other findings were marked SLOW under the shared host load. I did not change any existing test expectation. ## Decisions not specified in DESIGN - Use one shared Settings mount queue: yield after a painted frame, use `scheduler.yield()` / `requestIdleCallback` where available, then a timer fallback. Keep Account Profile synchronous and keep Admin hidden until authorization is verified. - Defer overlay opening focus by one painted frame and call `focus({ preventScroll: true })`. - Follow the 2026-10-01 owner override: keyboard actions animate like pointer actions; reduced-motion still applies. Known gap: warm content is still 377.4 ms, and cold content is 1103.9 ms. The largest source-map pseudo-frame remains unresolved.
kayg commented 2026-10-02 13:14:41 +00:00
Author
Owner
Copy link

#663 bundle/loading audit adds static graph evidence to the existing #642 owner. Production build at origin/dev c4a61e8cf0: Settings total 452,529 B gzip JS and 70,263 B gzip CSS; incremental versus shell is 82,857 B JS. The single Settings node contributes 76,896 B gzip JS. settings/[...path]/+page.svelte:44–56 imports every User/Admin section before it chooses one. Round 7a adds Connected Accounts to the same static list; round 7b blaze-settings retains this boundary while it improves mounted-view retention. Split sections and large pickers by the selected destination, then warm only a bounded likely destination after first usable paint. Preserve the current deep links, retained views and fail-closed Admin guard. These bytes do not establish a current HDD timing regression; reuse #642/#641 for that proof. #556 remains the shortcut data owner. No new Settings issue filed.

#663 bundle/loading audit adds static graph evidence to the existing #642 owner. Production build at origin/dev c4a61e8cf090170f35b1bed3350d9de20c83ecd5: Settings total 452,529 B gzip JS and 70,263 B gzip CSS; incremental versus shell is 82,857 B JS. The single Settings node contributes 76,896 B gzip JS. settings/[...path]/+page.svelte:44–56 imports every User/Admin section before it chooses one. Round 7a adds Connected Accounts to the same static list; round 7b blaze-settings retains this boundary while it improves mounted-view retention. Split sections and large pickers by the selected destination, then warm only a bounded likely destination after first usable paint. Preserve the current deep links, retained views and fail-closed Admin guard. These bytes do not establish a current HDD timing regression; reuse #642/#641 for that proof. #556 remains the shortcut data owner. No new Settings issue filed.
kayg commented 2026-10-02 13:14:43 +00:00
Author
Owner
Copy link

Round 4 starting on job/blaze-settings at bf3ad5f29d1108c34e87c0e9d20b1dbe67ffc5c3, based on origin/dev c4a61e8cf090170f35b1bed3350d9de20c83ecd5. I will measure the production build on the locked perf VM first. If warm Settings content still exceeds 100 ms, I will capture the requested V8 compile and Blink timeline trace, attribute the unattributed work, and make a measured fix.

Round 4 starting on `job/blaze-settings` at `bf3ad5f29d1108c34e87c0e9d20b1dbe67ffc5c3`, based on `origin/dev` `c4a61e8cf090170f35b1bed3350d9de20c83ecd5`. I will measure the production build on the locked perf VM first. If warm Settings content still exceeds 100 ms, I will capture the requested V8 compile and Blink timeline trace, attribute the unattributed work, and make a measured fix.
kayg commented 2026-10-02 13:37:31 +00:00
Author
Owner
Copy link

The current production client manifest maps Settings to nodes/46.Rflp4JPq.js. Its transitive static-import closure is 36 JavaScript files, 983,765 raw bytes / 319,785 gzip bytes; the largest shared chunk is 481,999 raw / 151,032 gzip bytes. This proves a large static dependency graph, but not that all of it parses on the open path. The opening trace will determine that.

The perf VM is reachable. Inside /root/perf.lock, uptime reported load averages 0.02 / 0.13 / 0.53. The HDD-emulation mount is active on /dev/mapper/calternal-hdd-emu with 8 ms read and write delay. I am using that mount for the real server fixture data.

The current production client manifest maps Settings to `nodes/46.Rflp4JPq.js`. Its transitive static-import closure is 36 JavaScript files, 983,765 raw bytes / 319,785 gzip bytes; the largest shared chunk is 481,999 raw / 151,032 gzip bytes. This proves a large static dependency graph, but not that all of it parses on the open path. The opening trace will determine that. The perf VM is reachable. Inside `/root/perf.lock`, `uptime` reported load averages 0.02 / 0.13 / 0.53. The HDD-emulation mount is active on `/dev/mapper/calternal-hdd-emu` with 8 ms read and write delay. I am using that mount for the real server fixture data.
kayg commented 2026-10-02 14:11:05 +00:00
Author
Owner
Copy link

Round 4 perf-VM baseline (before application changes)

The production SPA ran on the perf VM in five sessions under /root/perf.lock, with HDD emulation active. Each session recorded one cold first open and five warm opens from Notes, Calendar, Mail and Files. This gives 5 cold and 25 warm samples per source. Load at the start was 0.10, 0.31, 0.36 (1/5/15 minute averages); the five-session run ended at 3.38, 1.61, 0.84. The 1-minute load includes this measurement workload.

Source Cold latency p50 / p95 / max ms Warm latency p50 / p95 / max ms Warm browser CPU p50 / p95 / max ms
Notes 462.90 / 884.12 / 954.80 277.60 / 399.78 / 426.40 380 / 600 / 600
Calendar 394.20 / 567.70 / 597.40 268.90 / 353.10 / 482.50 390 / 488 / 660
Mail 534.10 / 596.88 / 612.30 236.30 / 371.60 / 458.40 340 / 488 / 590
Files 337.60 / 602.90 / 612.50 270.80 / 332.28 / 346.80 350 / 440 / 460

Warm p95 exceeds the 100 ms target for all four source Tabs. The first cold Notes, Calendar and Files opens requested the 269,888-byte Settings route node and four small JavaScript chunks (282,971 bytes total, uncompressed). The Mail cold route had already loaded those scripts. No warm open requested a new JavaScript resource in 100 warm samples. This narrows the warm delay to work after code is present; I am capturing the requested V8/Blink trace next.

## Round 4 perf-VM baseline (before application changes) The production SPA ran on the perf VM in five sessions under `/root/perf.lock`, with HDD emulation active. Each session recorded one cold first open and five warm opens from Notes, Calendar, Mail and Files. This gives 5 cold and 25 warm samples per source. Load at the start was `0.10, 0.31, 0.36` (1/5/15 minute averages); the five-session run ended at `3.38, 1.61, 0.84`. The 1-minute load includes this measurement workload. | Source | Cold latency p50 / p95 / max ms | Warm latency p50 / p95 / max ms | Warm browser CPU p50 / p95 / max ms | | --- | ---: | ---: | ---: | | Notes | 462.90 / 884.12 / 954.80 | 277.60 / 399.78 / 426.40 | 380 / 600 / 600 | | Calendar | 394.20 / 567.70 / 597.40 | 268.90 / 353.10 / 482.50 | 390 / 488 / 660 | | Mail | 534.10 / 596.88 / 612.30 | 236.30 / 371.60 / 458.40 | 340 / 488 / 590 | | Files | 337.60 / 602.90 / 612.50 | 270.80 / 332.28 / 346.80 | 350 / 440 / 460 | Warm p95 exceeds the 100 ms target for all four source Tabs. The first cold Notes, Calendar and Files opens requested the 269,888-byte Settings route node and four small JavaScript chunks (282,971 bytes total, uncompressed). The Mail cold route had already loaded those scripts. No warm open requested a new JavaScript resource in 100 warm samples. This narrows the warm delay to work after code is present; I am capturing the requested V8/Blink trace next.
kayg commented 2026-10-02 14:49:42 +00:00
Author
Owner
Copy link

Round 4 trace finding and attribution correction. My earlier 343 ms V8 (program) attribution was wrong: V8 Profile starts on renderer TID 87830, while ProfileChunk samples are delivered on profiler TID 87867. The first aggregation keyed the sample clock by TID and included pre-open samples. Recomputed from the profile start time by PID, the 185.53 ms marked warm Notes open contains 185.26 ms of sampled time: 124.97 ms (idle), 36.91 ms (program), 4.26 ms blur, and 1.89 ms in the named go frame from 6sHylKWU.js. The compile spans sum to 20.32 ms in the interval, mostly background compilation; 5 V8.ParseFunction spans sum to 0.06 ms. This trace does not support a large synchronous script parse/compile cost.

The open trace had 14 UpdateLayoutTree spans totaling 6.03 ms (maximum subtree 146 elements) and 8 Layout spans totaling 2.48 ms (maximum 267 layout objects). The first Settings render measured 103 descendants in the saved benchmark runs; the trace DOMStats grew from 157 to 232 document elements. Warm Settings open requested no JavaScript resources. The traced Notes view had four existing API responses complete by 58 ms; the Settings /api/v1/auth/me and /api/v1/auth/me/security requests returned by 86 ms and 122 ms. Neither is waited on by the Settings content-ready condition. blink.debug was requested but emitted no events. VizCompositorThread task spans overlap the frame pipeline, so this capture does not isolate backdrop-filter raster cost.

The quiet perf-VM baseline remains above the 100 ms warm target for every source (aggregate warm p95: Notes 399.78 ms, Calendar 353.10 ms, Mail 371.60 ms, Files 332.28 ms). No application optimization is justified by this trace alone: DOM, style and layout are small; the prior dominant-CPU claim was an analyzer error; and the remaining compositor/frame wait is not attributed well enough to risk a visual regression. I am retaining Round 3 production behavior and adding accurate script/resource and compile-trace evidence to the benchmark profile.

Round 4 trace finding and attribution correction. My earlier 343 ms V8 `(program)` attribution was wrong: V8 `Profile` starts on renderer TID 87830, while `ProfileChunk` samples are delivered on profiler TID 87867. The first aggregation keyed the sample clock by TID and included pre-open samples. Recomputed from the profile start time by PID, the 185.53 ms marked warm Notes open contains 185.26 ms of sampled time: 124.97 ms `(idle)`, 36.91 ms `(program)`, 4.26 ms `blur`, and 1.89 ms in the named `go` frame from `6sHylKWU.js`. The compile spans sum to 20.32 ms in the interval, mostly background compilation; 5 `V8.ParseFunction` spans sum to 0.06 ms. This trace does not support a large synchronous script parse/compile cost. The open trace had 14 `UpdateLayoutTree` spans totaling 6.03 ms (maximum subtree 146 elements) and 8 Layout spans totaling 2.48 ms (maximum 267 layout objects). The first Settings render measured 103 descendants in the saved benchmark runs; the trace DOMStats grew from 157 to 232 document elements. Warm Settings open requested no JavaScript resources. The traced Notes view had four existing API responses complete by 58 ms; the Settings `/api/v1/auth/me` and `/api/v1/auth/me/security` requests returned by 86 ms and 122 ms. Neither is waited on by the Settings content-ready condition. `blink.debug` was requested but emitted no events. VizCompositorThread task spans overlap the frame pipeline, so this capture does not isolate backdrop-filter raster cost. The quiet perf-VM baseline remains above the 100 ms warm target for every source (aggregate warm p95: Notes 399.78 ms, Calendar 353.10 ms, Mail 371.60 ms, Files 332.28 ms). No application optimization is justified by this trace alone: DOM, style and layout are small; the prior dominant-CPU claim was an analyzer error; and the remaining compositor/frame wait is not attributed well enough to risk a visual regression. I am retaining Round 3 production behavior and adding accurate script/resource and compile-trace evidence to the benchmark profile.
kayg commented 2026-10-02 15:32:09 +00:00
Author
Owner
Copy link

Round 4 finished at 90bec5ab15fa4d23960a52cd456468fc7dcdc343 (after the required merge from origin/dev). I did not push, deploy or merge this branch.

Quiet perf-VM result

Five locked production-build sessions ran with HDD emulation. Each session had one cold open and five warm opens for each source Tab. The aggregate has five cold and 25 warm samples per source. uptime at the first lock was 0.10 / 0.31 / 0.36 (1/5/15 minutes). The workload raised the one-minute value; the final five-session value was 3.38 / 1.61 / 0.84.

Source Cold p50 / p95 / max ms Warm p50 / p95 / max ms
Notes 462.90 / 884.12 / 954.80 277.60 / 399.78 / 426.40
Calendar 394.20 / 567.70 / 597.40 268.90 / 353.10 / 482.50
Mail 534.10 / 596.88 / 612.30 236.30 / 371.60 / 458.40
Files 337.60 / 602.90 / 612.50 270.80 / 332.28 / 346.80

All warm p95 values remain above 100 ms. The cold Settings node 46 request was 269,888 bytes; the first-open script set was 282,971 raw bytes. The full route closure was 983,765 raw / 319,785 gzip bytes across 36 files. Warm opens fetched no JavaScript.

Trace result

I corrected the earlier 343 ms V8 (program) claim. V8 profile start is on renderer TID 87830; sample chunks arrive on profiler TID 87867. The first aggregation used the sample TID for the clock and included pre-open samples. With the profile start aligned by PID, the 185.53 ms marked warm Notes open has 185.26 ms sampled time: 124.97 ms (idle), 36.91 ms (program), 4.26 ms blur, and 1.89 ms in go from 6sHylKWU.js. Five V8.ParseFunction spans total 0.06 ms. The compile spans total 20.32 ms, mostly background compilation.

The trace had 14 style recalculation spans totaling 6.03 ms and 8 Layout spans totaling 2.48 ms. Its Settings view had 103 descendants in the benchmark runs. Warm opens fetched no JavaScript. The Notes page behind Settings did not show a large style or layout cost in this trace. blink.debug emitted no events, and VizCompositorThread spans overlap the frame pipeline, so this run does not prove that backdrop blur owns the remaining delay.

I made no production UI change. The trace does not identify a safe optimization that can account for the remaining time without risking the visual material. The before/after app comparison is therefore unchanged: Round 3 production code remains in place, and the warm target still misses. The branch adds raw resource and DOM measurements and opts into V8 compile/Blink trace categories for the next attribution pass.

Files and decisions

Round 4 commits:

  • cd1212c82 perf(bench): capture Settings VM evidence
  • b61a444e9 perf(bench): report Settings preload bytes

These update apps/web/e2e/animation-trace.mjs and bench/settings-open-642.mjs. The required origin/dev merge also brought changes to deploy/media-sandbox and tests/adversarial/prepare-media-runtime.sh.

Decision: no undocumented DESIGN choice was needed. I kept the existing Settings glass and route behavior because the trace did not attribute the remaining frame/compositor wait to a specific production cost. The earlier 343 ms CPU claim was an analyzer error, not a code regression.

UX gaps closed: no UI behavior changed. The benchmark now records the script resources requested during an open, decoded/encoded sizes, and Settings DOM count. UX gaps left: warm p95 remains above target; backdrop-filter cost is not isolated.

Gates

bun run check output:

$ node scripts/check-user-storage.mjs && node scripts/check-type-tokens.mjs && node scripts/check-motion-tokens.mjs && svelte-kit sync && svelte-check --tsconfig ./tsconfig.json
User browser caches use userStorage; only documented device/public-link exceptions remain.
Text sizes and UI shape values use shared role tokens.
UI transitions and animation options use shared motion tokens or documented exceptions.
Loading svelte-check in workspace: /home/kayg/Developer/calternal-wt/blaze-settings/apps/web
Getting Svelte diagnostics...
svelte-check found 0 errors and 0 warnings

bun run test output:

 Test Files  19 failed | 135 passed (154)
      Tests  34 failed | 1022 passed (1056)
   Start at  17:08:38
   Duration  1140.74s (transform 54%, environment 15%, import 15%, tests 11%, setup 4%)

  Transform  |component| transforming modules took 1759.87s · 44% of tracked time, re-done on every run
             persist transforms across runs with fsModuleCache: true
             learn more: https://vitest.dev/guide/improving-performance#caching-between-reruns

error: script "test" exited with code 1

The 34 failures are mostly configured 5-second timeouts across unrelated files; the theme contrast test reached its 30-second timeout. Later Composer timer/state assertions followed a timed-out test and cleanup. This is consistent with the severe shared-host CPU and I/O pressure during the run, but I did not run a second test pass to prove the cause. I did not change any test expectation. Rust gates were not applicable because no Rust source changed. cargo clean removed 1142 files / 555.6 MiB. I also deleted apps/web/build and apps/web/.svelte-kit.

Round 4 finished at `90bec5ab15fa4d23960a52cd456468fc7dcdc343` (after the required merge from `origin/dev`). I did not push, deploy or merge this branch. ## Quiet perf-VM result Five locked production-build sessions ran with HDD emulation. Each session had one cold open and five warm opens for each source Tab. The aggregate has five cold and 25 warm samples per source. `uptime` at the first lock was 0.10 / 0.31 / 0.36 (1/5/15 minutes). The workload raised the one-minute value; the final five-session value was 3.38 / 1.61 / 0.84. | Source | Cold p50 / p95 / max ms | Warm p50 / p95 / max ms | | --- | ---: | ---: | | Notes | 462.90 / 884.12 / 954.80 | 277.60 / 399.78 / 426.40 | | Calendar | 394.20 / 567.70 / 597.40 | 268.90 / 353.10 / 482.50 | | Mail | 534.10 / 596.88 / 612.30 | 236.30 / 371.60 / 458.40 | | Files | 337.60 / 602.90 / 612.50 | 270.80 / 332.28 / 346.80 | All warm p95 values remain above 100 ms. The cold Settings node 46 request was 269,888 bytes; the first-open script set was 282,971 raw bytes. The full route closure was 983,765 raw / 319,785 gzip bytes across 36 files. Warm opens fetched no JavaScript. ## Trace result I corrected the earlier 343 ms V8 `(program)` claim. V8 profile start is on renderer TID 87830; sample chunks arrive on profiler TID 87867. The first aggregation used the sample TID for the clock and included pre-open samples. With the profile start aligned by PID, the 185.53 ms marked warm Notes open has 185.26 ms sampled time: 124.97 ms `(idle)`, 36.91 ms `(program)`, 4.26 ms `blur`, and 1.89 ms in `go` from `6sHylKWU.js`. Five `V8.ParseFunction` spans total 0.06 ms. The compile spans total 20.32 ms, mostly background compilation. The trace had 14 style recalculation spans totaling 6.03 ms and 8 Layout spans totaling 2.48 ms. Its Settings view had 103 descendants in the benchmark runs. Warm opens fetched no JavaScript. The Notes page behind Settings did not show a large style or layout cost in this trace. `blink.debug` emitted no events, and VizCompositorThread spans overlap the frame pipeline, so this run does not prove that backdrop blur owns the remaining delay. I made no production UI change. The trace does not identify a safe optimization that can account for the remaining time without risking the visual material. The before/after app comparison is therefore unchanged: Round 3 production code remains in place, and the warm target still misses. The branch adds raw resource and DOM measurements and opts into V8 compile/Blink trace categories for the next attribution pass. ## Files and decisions Round 4 commits: - `cd1212c82` `perf(bench): capture Settings VM evidence` - `b61a444e9` `perf(bench): report Settings preload bytes` These update `apps/web/e2e/animation-trace.mjs` and `bench/settings-open-642.mjs`. The required `origin/dev` merge also brought changes to `deploy/media-sandbox` and `tests/adversarial/prepare-media-runtime.sh`. Decision: no undocumented DESIGN choice was needed. I kept the existing Settings glass and route behavior because the trace did not attribute the remaining frame/compositor wait to a specific production cost. The earlier 343 ms CPU claim was an analyzer error, not a code regression. UX gaps closed: no UI behavior changed. The benchmark now records the script resources requested during an open, decoded/encoded sizes, and Settings DOM count. UX gaps left: warm p95 remains above target; backdrop-filter cost is not isolated. ## Gates `bun run check` output: ```text $ node scripts/check-user-storage.mjs && node scripts/check-type-tokens.mjs && node scripts/check-motion-tokens.mjs && svelte-kit sync && svelte-check --tsconfig ./tsconfig.json User browser caches use userStorage; only documented device/public-link exceptions remain. Text sizes and UI shape values use shared role tokens. UI transitions and animation options use shared motion tokens or documented exceptions. Loading svelte-check in workspace: /home/kayg/Developer/calternal-wt/blaze-settings/apps/web Getting Svelte diagnostics... svelte-check found 0 errors and 0 warnings ``` `bun run test` output: ```text Test Files 19 failed | 135 passed (154) Tests 34 failed | 1022 passed (1056) Start at 17:08:38 Duration 1140.74s (transform 54%, environment 15%, import 15%, tests 11%, setup 4%) Transform |component| transforming modules took 1759.87s · 44% of tracked time, re-done on every run persist transforms across runs with fsModuleCache: true learn more: https://vitest.dev/guide/improving-performance#caching-between-reruns error: script "test" exited with code 1 ``` The 34 failures are mostly configured 5-second timeouts across unrelated files; the theme contrast test reached its 30-second timeout. Later Composer timer/state assertions followed a timed-out test and cleanup. This is consistent with the severe shared-host CPU and I/O pressure during the run, but I did not run a second test pass to prove the cause. I did not change any test expectation. Rust gates were not applicable because no Rust source changed. `cargo clean` removed 1142 files / 555.6 MiB. I also deleted `apps/web/build` and `apps/web/.svelte-kit`.
kayg commented 2026-10-02 15:33:44 +00:00
Author
Owner
Copy link

Additional trace check: the marked 185.53 ms interval contains no GC-named trace events. Along with the low V8.ParseFunction and style/layout totals, this capture does not identify GC as the warm-open delay.

Additional trace check: the marked 185.53 ms interval contains no GC-named trace events. Along with the low `V8.ParseFunction` and style/layout totals, this capture does not identify GC as the warm-open delay.
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
dev
wip/palette2-1123
wip/palette-1093
wip/onboard2-1141
wip/onboard-1141.aborted-early
wip/onboard-1141
wip/notifloop-1194
wip/nlpchip-1127
wip/morph-1104
wip/merge-round-7c5
wip/merge-round-7c4
wip/merge-round-7c3
wip/merge-round-7c2
wip/merge-round-7c
wip/mchrome-1084
wip/mailghost2-1094
wip/mailghost-1094
wip/kbpreview2-1118
wip/kbpreview-1118
wip/kanban-1092
wip/importhang-1121
wip/hiderev-1153
wip/hide4-1153
wip/hide3-1153
wip/hide2-1153
wip/hide-1153
wip/editreg-1132
wip/editorrail3-1113
wip/editorrail2-1113
wip/editorrail-1113
wip/e2e-b2-1071
wip/e2e-b-1071
wip/draw4-1101
wip/draw3-1101
wip/draw2-1101
wip/draw-1101
wip/directory-1199-r
wip/directory-1199
wip/delete-1119
wip/collabrev-1197
wip/collabloss-1197
wip/cards2-1083
wip/cards-1083
wip/canvas-visual
wip/canvasvis2-976
wip/calhdr-1112
wip/calcards-1115
wip/browserfix
wip/blocks-1125
job/onboard-1141
wip/allday-1107
wip/agenda-decks
wip/agenda-1086
wip/adv7c-1105
wip/txentry-1198
wip/trayicons2-1095
wip/trayicons-1095
wip/tagperf-1186
wip/sidebar3-1094
wip/rev2-webperf
wip/rev2-money-ident
job/adv-1202
wip/restyle-notes
wip/previewcard-1098
job/merge30
job/notifloop-1194
job/collabloss-1197
job/hide-1153
job/perf-1124
job/perf2-1124
job/tocrail-1191
job/restyle-settings
wip/restyle-settings
job/segmented-1200
job/tagperf-1186
wip/segmented-1200
job/restyle-files
job/tagdnd-1187
job/cards-1179
wip/cards2-1179
wip/cards-1179
wip/tocrail-1191
wip/tagdnd-1187
wip/restyle-files
wip/perf-1124
wip/merge30j
job/restyle-notes
job/wizchoices-1140
wip/wizchoices-1140
wip/restyle-1190
job/moneyfmt-1180
job/txentry-1198
wip/moneyfmt2-1180
wip/moneyfmt-1180-r
wip/moneyfmt-1180
job/pillglass-1189
job/flags-1181
wip/flags-1181
job/restyle-1190
job/restyle-mailmoney
job/restyle-search
job/settingsreg-1195
job/wizard-1140
site/website
wip/wizardrev2-1140
wip/wizardrev-1140
wip/wizard5-1140
wip/wizard4-1140
wip/wizard3-1140
wip/wizard2-1140
wip/wizard-1140
wip/pillglass-1189
wip/settingsreg-1195
job/merge29
job/fu-1171
wip/merge29j
wip/fu-1171
job/fu-1166
job/directory-1199
job/proflog-1204
job/txresearch-1188
wip/fu-1166
job/merge28
job/search-1066
wip/search-1066
wip/merge28j
job/gateslot-1182
job/bulkimport-1157
job/mailnet-1160
wip/mailnetrev-1160
wip/mailnet-1160
wip/bulkrev-1157
wip/bulkimport-1157
job/startup-1161
wip/startup-1161
job/merge27
job/linkcards-1151
wip/linkcards3-1151
wip/linkcards2-1151
wip/linkcards-1151
job/traydate-1144
wip/traydate3-1144
wip/traydate2-1144
wip/traydate-1144
job/draw-1101
wip/merge27j
job/blockpill-1152
wip/blockpill3-1152
wip/blockpill2-1152
wip/blockpill-1152
job/minihover-1149
wip/minihover2-1149
wip/minihover-1149
job/merge25
wip/merge25-r
wip/merge25b
wip/merge25
job/inspector-1129
job/tags-1110
wip/inspector3-1129
wip/inspector2-1129
wip/inspector-1129
wip/tagsrev-1110
wip/tags2-1110
wip/tags-1110
job/dates-1148
wip/datesrev-1148
wip/dates2-1148
wip/dates-1148
job/licence-1145
wip/licence2-1145
wip/licence-1145
job/selfhost-1156
job/merge23
wip/merge23
job/tagfilter-1109
wip/tagfilter2-1109
wip/tagfilter-1109
job/kbd-1134
wip/kbd2-1134
wip/kbd-1134
job/palfoot-1137
wip/selfhost-1156
wip/palfoot2-1137
wip/palfoot-1137
job/toggle-1158
wip/toggle-1158
job/kbpreview-1118
job/docratchet-1155
job/perflint-1133
job/devtests-1159
wip/docratchet-1155
wip/devtests-1159
job/segv-1136
wip/toast-1142
wip/segv-1136
job/toast-1142
job/blockreload-1147
wip/blockreload-1147
job/font-1150
wip/font-1150
job/importui-1120
job/minimonth-1149
wip/importui-1120
wip/minimonth-1149
job/depcheck-1146
wip/perflint-1133
wip/depcheck-1146
job/calcards-1115
job/blocks-1125
job/plus-1128
job/shift-1138
wip/plus2-1128
wip/plus-1128
wip/shift-1138
job/moneyfid-1130
job/editorrail-1113
wip/moneyrev-1130
wip/moneyfid-1130
job/noext-851
wip/noext-851
wip/noext3-851
wip/noext2-851
job/week-1135
wip/week-1135
job/editreg-1132
job/smoke-1122
wip/smoke-1122
job/docs-1143
job/palette2-1123
job/calhdr-1112
job/nlpchip-1127
job/mailghost-1094
job/reconnect-1131
wip/reconnect-1131
job/trayicons-1095
job/delete-1119
job/importhang-1121
job/cards-1083
job/palette-1093
job/mchrome-1084
job/e2e-a-1071
job/canvas-visual
job/previewcard-1098
job/allday-1107
wip/e2e-a2-1071
wip/e2e-a-1071
job/e2e-b-1071
job/adv7c-1105
job/kanban-1092
job/agenda-1086
job/merge-round-7c
job/morph-1104
wip/surfaces-p2
job/merge-round-9
wip/merge-round-9
job/7cfix-small
wip/7cfix-small
job/mailui-1078
job/merge-round-8
wip/merge-round-8
wip/mailui-1078
job/mailround-1038
job/applemail-accept
wip/settitle-1068
wip/mailround2-1038
wip/mailround-1038
wip/e2e-7b
job/crash-1069
wip/crash-1069
job/searchlost-1066
wip/searchlost-1066
job/7b-reconcile
job/flake-1065
wip/flake-1065
wip/merge-round-7b7
wip/merge-round-7b6
wip/merge-round-7b5
wip/merge-round-7b4
wip/7b-reconcile
job/appupdate-1059
job/nfd-1044
wip/appupdate-1059
job/e2e-7b
job/loop-1062
wip/loop-1062
job/pdfprev-1045
job/invtoggle-1053
wip/pdfprev-1045
wip/nfd-1044
wip/invtoggle-1053
job/7bfix-e2e
job/mailstress-b
wip/7bfix-e2e
wip/mailstress-b
job/7bfix-adv
wip/7bfix-adv
job/mailstress-a
job/stack-1054
wip/stack-1054
wip/mailstress-a
job/mailstress-1038
wip/mailstress-1038
job/upload500-1051
wip/upload500-1051
job/share-1034
wip/share-1034
job/syncerr-1037
job/7bfix-photos
wip/7bfix-photos
job/paste-1036
job/setside-1039
wip/setside-1039
wip/paste-1036
job/lease-1042
wip/syncerr-1037
wip/lease-1042
job/7bfix-data
job/passkeybind-1043
wip/apprevoke-1041
job/invite-1035
wip/invite-1035
job/merge-round-7b2
wip/merge-round-7b2
job/mailproxy-486
job/apprevoke-1041
job/rebuild-1033
job/pillborder-1029
wip/pillborder-1029
wip/mailproxy-486
wip/applemail-486
job/headless-998
wip/headless-998
job/groups-1028
wip/groups-1028
job/rebuildwarn-1016
wip/rebuildwarn-1016
job/startup-1011
wip/startup-1011
job/monthpill-1009
job/bgthumb-1025
job/sharetitle-1012
wip/monthpill-1009
wip/bgthumb-1025
wip/sharetitle-1012
job/canvas-cards-977
wip/canvas-cards-977
job/canvas-pencil-978
job/canvas-sketch-990
wip/canvas-sketch-990
wip/canvas-pencil-978
job/canvas-files-989
wip/canvas-files-989
job/canvas-collab-991
wip/canvas-collab-991
job/weekscroll-1018
wip/weekscroll-1018
wip/canvas-core-976
job/canvas-core-976
job/round-drag
wip/round-drag
job/round-settings
job/browserfix
wip/oapi-974
job/oapi-974
job/hist2-integrate
job/mailhtml-726
wip/mailhtml-726
wip/hist2-integrate
job/moneyfu-984
job/drag-1015
wip/drag-1015
job/rename-1017
wip/rename-1017
job/hist2-api
wip/hist2-api
job/oneacct-1014
wip/oneacct-1014
wip/moneyfu-984
job/hist2-bench
job/hist2-restore
wip/hist2-bench
job/hist2-write
job/hotfix-724
wip/hotfix-724
wip/hist2-write
wip/hist2-restore
job/hist2-store
job/hist2-ui
wip/hist2-ui
wip/hist2-store
job/searchstarve-965
job/shutdown-963
wip/shutdown-963
wip/pubedit-981
job/pubedit-981
job/analytics-973
wip/searchstarve-965
job/authflash-850
job/weeklane-969
job/pvtitle-1004
job/hist-975
wip/authflash-850
job/voicepill-617
wip/pvtitle-1004
job/headring-1003
wip/weeklane-969
wip/voicepill-617
wip/headring-1003
wip/analytics-973
job/agentscope-980
wip/thumbsandbox-988
job/thumbsandbox-988
wip/hist-975
job/links-856
wip/links-856
job/davetag-966
wip/davetag-966
job/filesstorm-1000
job/hoverpad-725
wip/filesstorm-1000
job/ffmpegblas-993
job/merge-round-7a
wip/hoverpad-725
wip/ffmpegblas-993
job/nowdot-1002
wip/verify-7a
job/noteid-857
wip/nowdot-1002
wip/noteid-857
wip/merge-round-7a
wip/agentscope-980
job/imapedge
job/a11yfix2
wip/imapedge-941
wip/imapedge
wip/a11yfix2
job/notetask-986
job/logheading
wip/logheading-998
job/textthumb-652
job/photolive-987
wip/photolive-987
job/davactive-983
job/savefix-985
job/tabicons-607
wip/davactive-983
wip/tabicons-607
wip/notetask-986
wip/savefix-985
job/dirid-627
job/buildspeed-1007
wip/dirid-627
job/agenda-decks
job/perfguards-impl
job/undo-a11y
wip/undo-a11y
job/mailperf
job/wal-824
wip/settings-50
job/settings-50
job/notesfilter-606
wip/notesfilter-606
job/surfaces-p2
wip/wal-824
job/maillayouts
wip/mailperf
wip/maillayouts
job/taskmeta-659
job/money-ident
wip/money-ident
wip/taskmeta-659
job/errstates
wip/perfguards-impl
job/headings-881
wip/headings-881
wip/errstates
job/voice-619
job/gaps-827
job/notesperf
wip/notesperf
wip/voice-619
job/hddsql-549
job/perf-stream-668
wip/perf-stream-668
wip/deeplinks-fix
job/deeplinks-fix
job/authfix
job/docsfix-rust
wip/docsfix-rust
job/webperf
job/docsfix-web
job/datafix2
job/webdav-lock-476
job/copyfix
wip/copyfix
wip/webperf
job/focus-658
wip/protofix
job/mediafix
job/protofix
wip/mediafix
job/agentfix
job/hhmm-724
wip/agentfix
job/undo-722
job/reuse
wip/webdav-lock-476
wip/reuse
job/scopefix
job/datafix
wip/hhmm-724
wip/undo-722
job/surfaces-p1
wip/hddsql-549
job/voicememos-618
wip/datafix2
wip/surfaces-p1
job/fix-940
wip/fix-940
job/blaze-surfaces
wip/datafix
wip/blaze-surfaces
job/taskday-655
job/linknav-639
wip/linknav-639
wip/gaps-827
job/isolation-707
job/audiophotos-720
wip/audiophotos-720
job/advfind-664
wip/voicememos-618
wip/taskday-655
wip/isolation-707
wip/advfind-664
wip/scopefix
wip/focus-658
job/testgaps
wip/testgaps
job/overscroll-718
wip/authfix
job/deps
wip/overscroll-718
job/rev2-agentfix
job/rev2-money-ident
job/rev2-mailperf
wip/deps
job/hardening-728
wip/hardening-728
job/searchgen-832
wip/searchgen-832
job/photopw-849
job/mailsql-825
wip/photopw-849
job/sharefix
wip/sharefix
job/rev2-mailhtml-726
job/rev2-perfguards
job/copyval-723
job/lightglass-r2
wip/lightglass-r2
wip/docsfix-web
job/copy-audit
job/macinterop-staging-r2
job/design-sync
job/rev2-taskmeta-659
job/rev2-webperf
job/docs-audit
job/rev2-advfind-664
job/rev2-mailproxy-486
job/states-audit
job/rev2-datafix
job/design-drift
job/test-gaps
job/rev2-voicememos-618
job/rev2-mediafix
job/rev2-deps
job/rev2-datafix2
job/licence-audit
job/issue-hygiene
job/rev2-protofix
job/rev2-voice-619
job/rev2-isolation-707
job/rev2-surfaces-p1
job/deeplink-audit2
job/rev2-audiophotos-720
wip/test-gaps
job/rev2-overscroll-718
job/rev2-undo-722
wip/states-audit
job/rev2-dropmd-719
job/rev2-linknav-639
job/merge-7b-plan
wip/merge-7b-plan
job/rev2-taskday-655
wip/mailsql-825
job/rev2-webdav-lock-476
job/rev2-browserfix
wip/design-drift
job/rev2-hddsql-549
wip/deeplink-audit2
job/rev2-scopefix
job/rev2-authfix
job/rev2-hardening-728
job/rev2-wal-824
job/rev2-sharefix
job/calsidebar-638
job/chrome-audit
job/ioperf
wip/ioperf
wip/chrome-audit
wip/calsidebar-638
job/dropmd-719
wip/dropmd-719
job/ocr-build
wip/ocr-build
job/blaze-settings
wip/copyval-723
job/toastring-721
wip/toastring-721
job/deployfix-732
wip/deployfix-732
wip/blaze-settings
job/money-import-recheck
job/rev-a11y
job/perf-arch-db
job/rev-7b-data
wip/textthumb-652
wip/perf-arch-db
job/sec-protocols
job/sidehdr-660
job/rev-7b-security
job/research-surfaces
job/rev-design-gaps
job/rev-mcp-api
wip/sidehdr-660
job/perf-arch-memory
wip/sec-protocols
job/perf-arch-bundle
job/snapedge-714
wip/rev-mcp-api
job/sec-supplychain
wip/research-surfaces
job/perf-arch-sync
job/rev-consistency
job/perf-arch-server
wip/perf-arch-server
wip/perf-arch-memory
job/perf-arch-io
job/perf-arch-client
job/sec-fs
job/sec-mcp-scopes
job/sec-sharing
job/perf-guards
job/sec-browser
job/sec-admin-deploy
job/sec-auth
wip/snapedge-714
job/bgpicker-717
wip/perf-arch-bundle
wip/money-import-recheck
job/advsetup-654
wip/bgpicker-717
wip/advsetup-654
job/burst-709
job/kbdcaps-710
job/app-pw-chooser
wip/burst-709
wip/app-pw-chooser
job/imaptest-625
wip/kbdcaps-710
job/fix-499
wip/fix-499
job/perf-mut-667
job/calimg-589
job/perf-snap-666
wip/calimg-589
wip/perf-snap-666
wip/perf-mut-667
job/perf-cache-665
wip/perf-cache-665
job/voicefiles-620
wip/voicefiles-620
job/admin-burst-705
wip/admin-burst-705
job/voicememos-review
wip/voicememos-review
wip/ryw-653
job/ryw-653
job/writeonopen-661
job/instant-663
wip/writeonopen-661
job/money-import-review
wip/money-import-review
wip/importjs-610
review/integrations-407-round6
wip/integrations-review
job/dragghost-612
wip/dragghost-612
job/integrations
wip/integrations
job/decider-656
job/merge-round-6
job/perf-rerun
wip/merge-round-6
job/integrations-review-round5
job/selalign-576
wip/selalign-576
job/mcp-events-491
job/files-631
job/cal-e2e-569
wip/cal-e2e-569
job/reload-423
wip/reload-423
wip/mcp-events-491
wip/files-631
job/notesbridge-644
wip/notesbridge-644
job/editor-series
job/calcard-series
wip/calcard-series
job/mcp-events-review-491
wip/mcp-events-review
wip/editor-series
job/quirks-546
job/integrations-recheck
job/tocrail-636
wip/tocrail-636
wip/quirks-546
wip/reminders-643
job/reminders-643
wip/davscale-573
job/davscale-573
job/integrations-review
wip/ocr-eval-584
job/ocr-eval-584
job/esc-537
wip/esc-537
job/toastname-586
wip/toastname-586
job/submenu-579
wip/submenu-579
job/tasks-mode
wip/tasks-mode
job/agentdocs-630
job/dupwrite-634
wip/agentdocs-630
wip/dupwrite-634
job/lightglass-588
wip/lightglass-588
job/tabswitch-549
job/ghosttask-623
wip/ghosttask-623
job/toaststack-616
job/weekstate-609
job/mailsync-613
wip/mailsync-613
wip/weekstate-609
job/maildup-626
wip/tabswitch-549
wip/maildup-626
wip/toaststack-616
job/motion-611
wip/motion-611
job/tlstest-601
wip/tlstest-601
job/perf-495
job/floating-sheet
wip/floating-sheet
job/remdup-585
wip/remdup-585
job/fix-502
wip/fix-502
job/attachplay-622
job/perf-batch
wip/perf-batch-563
wip/perf-495
hotfix/mail-sync-diag
job/mail-m3
wip/mail-m3
job/attach-poof-603
job/calhover-608
job/editorbar-604
job/mentions-605
job/merge-round-4
job/allday-514
wip/merge-round-4
wip/allday-514
job/merge-round-4a
wip/merge-round-4a
job/sharestack-580
job/fix-501
wip/sharestack-580
wip/fix-501
job/perf-batch-563
job/apw-cache-review
wip/apw-cache-review
job/probe-520
wip/probe-520
job/mac-393
wip/mac-393
job/header-571
job/flake-513
wip/flake-513
job/docs-thumb-547
wip/header-571
job/webcal-572
wip/webcal-572
wip/shortcuts-542
job/shortcuts-542
wip/docs-thumb-547
job/caldav-stress
wip/caldav-stress
wip/sweep-478
job/apw-cache-512
wip/apw-cache-512
job/money-empty-540
wip/restart-505
wip/money-empty-540
wip/fix-510
job/restart-505
job/fix-503
job/perf-496
wip/perf-496
job/fix-498
wip/fix-498
job/info-inspector-465
wip/info-inspector-465
job/fix-510
job/fix-507
wip/fix-507
wip/fix-503
job/fix-493
job/money-kinds
wip/money-kinds
job/hygiene-548
job/merge-round-3
wip/fix-493
job/drag-snap-536
wip/merge-round-3
wip/merge-round-0930
wip/drag-snap-536
job/align-538
wip/align-538
job/bg-flash
wip/bg-flash
job/money-import
job/search-count-544
wip/search-count-544
wip/money-import
job/settings-key-541
wip/settings-key-541
job/toast-539
job/preview-421
wip/preview-421
wip/toast-539
job/tasks-500-531
job/title-plain-526
wip/title-plain-526
wip/tasks-500-531
job/notes-bridge
wip/parity-484
job/parity-484
job/files-slow
job/crash-525
wip/notes-bridge
wip/files-slow
wip/crash-525
job/kbd-motion-527
wip/bg-422
job/analytics-504
wip/analytics-504
wip/kbd-motion-527
job/upload-pill-523
wip/upload-pill-523
wip/tray-order
job/tray-order
wip/overflow-mid
wip/merge-round-2
job/perf-494
wip/perf-494
wip/mcp-fast-492
wip/motion-477
wip/asr-ab-489
wip/theme-variants-506
wip/overflow-511
wip/week-header-508
wip/attach-427
job/dav-delete-471
job/iso-435
wip/iso-435
wip/files-sel-keys
wip/dav-delete-471
job/align-253
job/siwc-490
wip/siwc-490
job/money-kinds-review
wip/align-253
wip/money-kinds-review
job/small-bugs-3
wip/overlay-title-487
wip/multiget-500
wip/hidden-420
wip/webcal-ui
wip/webcal-431
job/perf-367
job/location
wip/small-bugs-3
wip/location
wip/perf-367
wip/admin-deny-483
job/tag-unicode-473
wip/tag-unicode-473
job/blur-436
wip/photos-470
wip/blur-436
wip/small-bugs-4
wip/hunt-20260930
wip/settings-hdr-482
wip/chips-416
job/dedup-375
wip/dedup-375
job/doc-stack
wip/doc-stack
job/tokens-literals
wip/tokens-literals
job/jobs-leftovers
wip/send-fast
wip/paste-467
wip/money-numbers
job/money-plugin
wip/money-plugin
job/break-dav
wip/merge-batch
wip/crossday-469
wip/mac-verify
wip/mail-m2
wip/break-dav
wip/money-review2
job/money-md
job/modes-424
wip/money-md
wip/jobs-leftovers
job/agenda-413
wip/agenda-413
wip/modes-424
job/recog-417
wip/recog-417
wip/bounce-425
wip/ab-384-luna
job/webdav-perf
wip/webdav-perf
job/toast-ring
wip/toast-ring
job/money-review
wip/money-review
wip/micro-motion
wip/settings-card
wip/minical
job/notes-imap-428
job/least-priv
wip/ui-small-2
wip/flaky-426
wip/drag-end-418
job/jank
wip/jank
wip/least-priv
wip/docs-site
job/agenda
job/sec-batch
wip/sec-batch
wip/per-user-index
job/area-calendars
wip/area-calendars
job/parity
wip/parity
job/documents-research
wip/documents-research
job/test-infra
job/reminders-sync
wip/small-bugs-2
wip/reminders-sync
wip/gestures
job/google-oauth
wip/tags-merge
wip/tags
job/e2e-theme
wip/e2e-theme
job/icon-align
wip/test-infra
wip/select-align
wip/editor-385
job/voice
wip/webdav
job/webdav
job/app-pw-ui
job/editor-integrity
wip/editor-integrity
wip/voice
wip/quota
wip/cal-followups
wip/icon-align
job/composer-scale
wip/composer-scale
job/jobs-page
wip/jobs-page
job/hig-type
wip/hig-type
wip/app-pw-ui
job/motion-spring
job/mcp
wip/motion-spring
wip/mcp
job/small-bugs
wip/push-hosts
job/profile-sign
wip/touch-369
wip/profile-sign
job/mobile-focus
wip/mobile-focus
wip/ui-polish-354
wip/small-bugs
wip/dup-task
job/toast-polish
job/app-pw-scopes
wip/toast-polish
wip/app-pw-scopes
wip/cli-agent
wip/selection-pills
job/preview-attach
wip/preview-attach
job/dav-proppatch
wip/dav-proppatch
wip/cal-switcher
job/atomic-race
wip/atomic-race
job/photos-shared
wip/photos-shared
wip/cal-grid
wip/note-rewrite
wip/search-rebuild
job/mail-m1
job/paperless-import
wip/paperless-import
wip/mail-m1
wip/hidden-activity
wip/search-d
wip/pricing-research
wip/cursors
wip/auto-scheme
job/single-pills
wip/single-pills
wip/xuser-matrix
wip/money-format
wip/app-pw-setup
wip/purge-dos
wip/vault-health
wip/caldav-apple
wip/xuser-audit
wip/e2e-green
wip/tabbar
wip/adv-harness
wip/maple-mono
job/search-fix
wip/search-fix
wip/search-perf-c
job/adv-harness
wip/sidebar-headers
job/glass
wip/temp-index
job/polish
wip/polish
wip/file-protocols
wip/money-research
wip/glass
wip/voice-models
wip/collab-redo
job/voice-research
wip/hunt-20260928
wip/notes-actions-research
wip/search-pad
wip/search-perf
wip/search-sticky
wip/editor-undo
wip/chrome-rules
wip/motion
wip/appearance-research
wip/appearance
wip/audit-bugs
wip/cal-glass
wip/block-actions
wip/authz-order
wip/event-stripes
wip/chrome-sidebar
wip/auth-flaky
wip/robust-2
wip/gate-fix
wip/menu-blur
wip/import-calternaljs
wip/tray-fix
job/import-calternaljs
wip/index-order
wip/audit-fixes
wip/search-chevrons
research/mail
wip/phone-chrome
wip/dedup-break
wip/csp
wip/ui-audit
wip/select-toast
wip/perf
wip/flat-layout
wip/fonts
wip/event-tint
wip/sync-converge
wip/data-split
wip/glass-audit
wip/robustness
wip/sync-chaos
wip/search-thumbs
wip/fuzz
wip/menu-icons
wip/search-pill
wip/sync-changing
wip/heading-links
wip/date-formats
wip/a11y
wip/break-editor
wip/e2e-fix
wip/settings-sections
wip/sync-root-guard
wip/search-palette
wip/share-edit
job/toasts
wip/toasts
wip/cont-analytics
wip/authz-review
wip/popovers
wip/overlay-glass
wip/change-feed
wip/editor-modes
wip/composer-align
wip/cont-agenda
wip/agenda-merge
job/agent-conventions
wip/agent-conventions
wip/backend-misc
job/route-audit
wip/route-audit
wip/ui-batch
wip/heif-hardening
wip/grid-resize
wip/ask-page
wip/webmcp
job/deeplink-audit
wip/deeplinks
wip/shortcuts
wip/cont-tz-days
main
No results found.
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
kayg/calternal#642
No description provided.