kayg/calternal-v0
Archived
Watch
1
0
Fork
You've already forked calternal-v0
 0
Code Issues 1.1k Pull requests Projects Releases Packages Wiki Activity Actions

PERF: Money import preview of a 15 MB / 25k-transaction Actual export takes 30–45 s #1124

Open
opened 2026-10-05 08:22:29 +00:00 by kayg · 44 comments
kayg commented 2026-10-05 08:22:29 +00:00
Owner
Copy link

From #1121: a realistic synthetic Actual export (15.3 MB ZIP, 25,000 transactions) took 33–45 s end to end through the real UI on a local server with no network limit (59–69 s at 1 MiB/s). Upload is a small part of that; preview parsing and validation dominate. Profile the preview path (unzip, SQLite read of the Actual DB, mapping, identity checks) and make a 25k-transaction import preview finish in a few seconds on this hardware, with streaming progress per stage. Add the case to bench/.

From #1121: a realistic synthetic Actual export (15.3 MB ZIP, 25,000 transactions) took 33–45 s end to end through the real UI on a local server with no network limit (59–69 s at 1 MiB/s). Upload is a small part of that; preview parsing and validation dominate. Profile the preview path (unzip, SQLite read of the Actual DB, mapping, identity checks) and make a 25k-transaction import preview finish in a few seconds on this hardware, with streaming progress per stage. Add the case to bench/.
files-390-tokyo-night.png
328 KiB
files-820-paper.png
483 KiB
files-820-tokyo-night.png
477 KiB
month-1440-paper.png
659 KiB
month-390-paper.png
286 KiB
month-820-paper.png
495 KiB
month-820-tokyo-night.png
414 KiB
week-1440-paper.png
620 KiB
week-1440-tokyo-night.png
561 KiB
month-1440-tokyo-night.png
607 KiB
month-390-tokyo-night.png
305 KiB
search-1440-paper.png
387 KiB
search-1440-tokyo-night.png
375 KiB
search-390-paper.png
137 KiB
search-390-tokyo-night.png
145 KiB
search-820-paper.png
273 KiB
search-820-tokyo-night.png
261 KiB
week-820-tokyo-night.png
380 KiB
week-390-paper.png
278 KiB
week-390-tokyo-night.png
283 KiB
week-820-paper.png
391 KiB
files-1440-paper.png
793 KiB
files-1440-tokyo-night.png
714 KiB
files-390-paper.png
328 KiB
files-390-tokyo-night.png files-820-paper.png files-820-tokyo-night.png month-1440-paper.png month-390-paper.png month-820-paper.png month-820-tokyo-night.png week-1440-paper.png week-1440-tokyo-night.png month-1440-tokyo-night.png month-390-tokyo-night.png search-1440-paper.png search-1440-tokyo-night.png search-390-paper.png search-390-tokyo-night.png search-820-paper.png search-820-tokyo-night.png week-820-tokyo-night.png week-390-paper.png week-390-tokyo-night.png week-820-paper.png files-1440-paper.png files-1440-tokyo-night.png files-390-paper.png
kayg commented 2026-10-06 05:14:59 +00:00
Author
Owner
Copy link

Starting #1124 on branch job/perf-1124 from base d0061ec3df. I will profile the Actual preview path on the locked perf VM, run the requested Home and route workloads where the existing tooling supports them, and keep measured results and code changes on this branch.

Starting #1124 on branch job/perf-1124 from base d0061ec3df127d81c86729d07d899b0bf2b6de91. I will profile the Actual preview path on the locked perf VM, run the requested Home and route workloads where the existing tooling supports them, and keep measured results and code changes on this branch.
kayg commented 2026-10-06 05:35:01 +00:00
Author
Owner
Copy link

Finding: the existing 25,000-row Actual fixture generated by bench/money-import-actual-fixture.py is 676,152 bytes, while #1124 reports a 15.3 MB ZIP. The current fixture is too small to reproduce the reported preview latency. I am extending that generator with deterministic synthetic transaction memos and adding observed stage spans to the existing benchmark. This is a benchmark-only decision; it preserves existing fixture defaults and uses no real export data.

Finding: the existing 25,000-row Actual fixture generated by bench/money-import-actual-fixture.py is 676,152 bytes, while #1124 reports a 15.3 MB ZIP. The current fixture is too small to reproduce the reported preview latency. I am extending that generator with deterministic synthetic transaction memos and adding observed stage spans to the existing benchmark. This is a benchmark-only decision; it preserves existing fixture defaults and uses no real export data.
kayg commented 2026-10-06 05:41:56 +00:00
Author
Owner
Copy link

Fixture/profile update: bench/money-import-actual-462.mjs --transactions=25000 --transaction-note-bytes=748 now creates a deterministic 15,373,905-byte Actual ZIP and records each preview sample plus observed progress-stage spans. Two generated archives had the same size and SHA-256. I will keep an importer change only if interleaved before/after runs improve p50 by at least 10% with no p95 regression, as requested; the first preview after server start is reported separately from same-process repeats.

Fixture/profile update: `bench/money-import-actual-462.mjs --transactions=25000 --transaction-note-bytes=748` now creates a deterministic 15,373,905-byte Actual ZIP and records each preview sample plus observed progress-stage spans. Two generated archives had the same size and SHA-256. I will keep an importer change only if interleaved before/after runs improve p50 by at least 10% with no p95 regression, as requested; the first preview after server start is reported separately from same-process repeats.
kayg commented 2026-10-06 05:57:23 +00:00
Author
Owner
Copy link

Before measurement on the locked perf VM (native HDD-like root disk): the deterministic 15,373,903-byte Actual ZIP with 25,000 transactions completed with p50 3,915.65 ms and p95 4,188.14 ms over five samples. Average server CPU was 2.724 s; peak server RSS was 295,079,936 bytes. The binary was built from origin/dev d0061ec3. This current synthetic run is much faster than the issue-reported 30–45 s, so I am profiling the remaining CPU path before changing importer behavior; the generated source has one Account and one Category and may not cover every production export shape.

Before measurement on the locked perf VM (native HDD-like root disk): the deterministic 15,373,903-byte Actual ZIP with 25,000 transactions completed with p50 3,915.65 ms and p95 4,188.14 ms over five samples. Average server CPU was 2.724 s; peak server RSS was 295,079,936 bytes. The binary was built from origin/dev `d0061ec3`. This current synthetic run is much faster than the issue-reported 30–45 s, so I am profiling the remaining CPU path before changing importer behavior; the generated source has one Account and one Category and may not cover every production export shape.
kayg commented 2026-10-06 06:02:27 +00:00
Author
Owner
Copy link

Finding: a locked perf record -F 99 --call-graph dwarf of the first 25k-row preview sampled calternal_plugin_money::import::ActualStreamProjection::finish_month / advance_to, calternal_money::codec::Document::parse_with_check, and ImportStreamRenderer::validate_generated_file. The streamed month currently parses its Markdown for Ledger projection and parses it again for byte-stable validation. The profile is available at /root/perf-1124/target/tmp/money-before.perf.data on the perf VM; this suggests one candidate, not yet a measured win. Before changing it, I will keep the candidate only if interleaved p50 improves at least 10% with no p95 regression.

Finding: a locked `perf record -F 99 --call-graph dwarf` of the first 25k-row preview sampled `calternal_plugin_money::import::ActualStreamProjection::finish_month` / `advance_to`, `calternal_money::codec::Document::parse_with_check`, and `ImportStreamRenderer::validate_generated_file`. The streamed month currently parses its Markdown for Ledger projection and parses it again for byte-stable validation. The profile is available at `/root/perf-1124/target/tmp/money-before.perf.data` on the perf VM; this suggests one candidate, not yet a measured win. Before changing it, I will keep the candidate only if interleaved p50 improves at least 10% with no p95 regression.
kayg commented 2026-10-06 06:28:50 +00:00
Author
Owner
Copy link

Finding: five interleaved before/after pairs on the locked perf VM used identical 15,373,903-byte ZIPs (25,000 transactions, fixture SHA256 16b9ad6292c852968a453d9796f59cea6c2fb87145cee44ac709e9e6b22167c8), with one first-preview and one warm repeat per server. The baseline p50/p95 were 3,711.33/4,898.15 ms; the parse-reuse candidate was 3,427.41/3,705.91 ms. Aggregate p50 improved 7.65%; p95 did not regress. Cold p50 improved 8.27%, warm p50 7.65%. Server CPU p50 moved 2.56 to 2.31 seconds. Since p50 did not meet the pre-stated 10% threshold, commit 943e97ca5c163bb79623f6386766024dfcc0e23e was reverted in b8b91527c; the optimization is not retained. The VM 1-minute load values ranged from 0.14 to 1.53 during samples, recorded in each raw profile.

Finding: five interleaved before/after pairs on the locked perf VM used identical 15,373,903-byte ZIPs (25,000 transactions, fixture SHA256 `16b9ad6292c852968a453d9796f59cea6c2fb87145cee44ac709e9e6b22167c8`), with one first-preview and one warm repeat per server. The baseline p50/p95 were 3,711.33/4,898.15 ms; the parse-reuse candidate was 3,427.41/3,705.91 ms. Aggregate p50 improved 7.65%; p95 did not regress. Cold p50 improved 8.27%, warm p50 7.65%. Server CPU p50 moved 2.56 to 2.31 seconds. Since p50 did not meet the pre-stated 10% threshold, commit `943e97ca5c163bb79623f6386766024dfcc0e23e` was reverted in `b8b91527c`; the optimization is not retained. The VM 1-minute load values ranged from 0.14 to 1.53 during samples, recorded in each raw profile.
kayg commented 2026-10-06 07:01:20 +00:00
Author
Owner
Copy link

The first route profile stopped before navigation timing, during Photos fixture setup. Playwright reported that the page/context closed at the upload page.evaluate call. The fixture is 194,576 bytes, and the harness had copied its Base64 body into all 1,000 entries in one Playwright argument (about 260 MB). I changed the seed to send the shared bytes once and the per-photo metadata list once, in commit a7476187c. The full route profile is retrying under the perf VM lock. This is a benchmark-fixture failure; it produced no user-facing timing results.

The first route profile stopped before navigation timing, during Photos fixture setup. Playwright reported that the page/context closed at the upload `page.evaluate` call. The fixture is 194,576 bytes, and the harness had copied its Base64 body into all 1,000 entries in one Playwright argument (about 260 MB). I changed the seed to send the shared bytes once and the per-photo metadata list once, in commit `a7476187c`. The full route profile is retrying under the perf VM lock. This is a benchmark-fixture failure; it produced no user-facing timing results.
kayg commented 2026-10-06 07:18:10 +00:00
Author
Owner
Copy link

The route profile retry passed the Photos seed but stopped before measurements while creating its Money fixture: POST /api/v1/money/budgets returned HTTP 404 because the per-User Money plug-in was not enabled. I added the real Money enable call before creating the Budget in 7efb93459. The first retry took 15m 28s to reach that assertion and produced no route timings. The next run keeps the 5k disk-backed Files folder and year of Journal entries, with smaller supplementary API Files/Notes/Photos counts; the separate Photos/Home profiles cover the large corpus.

The route profile retry passed the Photos seed but stopped before measurements while creating its Money fixture: `POST /api/v1/money/budgets` returned HTTP 404 because the per-User Money plug-in was not enabled. I added the real Money enable call before creating the Budget in `7efb93459`. The first retry took 15m 28s to reach that assertion and produced no route timings. The next run keeps the 5k disk-backed Files folder and year of Journal entries, with smaller supplementary API Files/Notes/Photos counts; the separate Photos/Home profiles cover the large corpus.
kayg commented 2026-10-06 07:23:39 +00:00
Author
Owner
Copy link

The route profile source also read apiRouteRepetitions before its const initialization and called the API sample function twice (the second call dropped the Note-link route). That would fail or overwrite the intended table after the navigation phase. Commit f6a9014fb moves the declaration before one API pass and adds the seeded Money month endpoint. I stopped the active run at the second navigation pass because its loaded module still had the defect; no timing JSON was written, and the VM lock is released.

The route profile source also read `apiRouteRepetitions` before its `const` initialization and called the API sample function twice (the second call dropped the Note-link route). That would fail or overwrite the intended table after the navigation phase. Commit `f6a9014fb` moves the declaration before one API pass and adds the seeded Money month endpoint. I stopped the active run at the second navigation pass because its loaded module still had the defect; no timing JSON was written, and the VM lock is released.
kayg commented 2026-10-06 07:27:33 +00:00
Author
Owner
Copy link

The route inventory review showed that route-perf.mjs omitted the Notes primary Tab. Commit 89685c8b9 adds Notes and a PERF_ROUTE_IDS selector so the first-load path can run independently. The active full-route run was started before this change and will not include Notes; I will record a focused Notes route profile separately.

The route inventory review showed that `route-perf.mjs` omitted the Notes primary Tab. Commit `89685c8b9` adds Notes and a `PERF_ROUTE_IDS` selector so the first-load path can run independently. The active full-route run was started before this change and will not include Notes; I will record a focused Notes route profile separately.
kayg commented 2026-10-06 07:31:40 +00:00
Author
Owner
Copy link

The route profile's third attempt completed cold and hard-refresh samples at 390 px/light, then failed while priming the warm Files source: its Photos ready selector timed out while the URL was still the 5k Files deep link. The run did not save timing JSON. I changed warm-source setup to open the stable Photos root directly; the measured destination switch still uses the real Tab. Fix: e14551692. I will run a focused Files route profile before the full retry.

The route profile's third attempt completed cold and hard-refresh samples at 390 px/light, then failed while priming the warm Files source: its Photos ready selector timed out while the URL was still the 5k Files deep link. The run did not save timing JSON. I changed warm-source setup to open the stable Photos root directly; the measured destination switch still uses the real Tab. Fix: `e14551692`. I will run a focused Files route profile before the full retry.
kayg commented 2026-10-06 07:36:52 +00:00
Author
Owner
Copy link

The focused Files profile confirmed that a synthetic Photos root link also failed during warm-source setup: the browser remained at the Files deep link and the Photos readiness marker timed out. This is setup, before the measured Files switch. I changed the setup to use a committed full navigation to /photos, while the measured destination switch still clicks its Tab. Fix: 36f2f9b3e; retrying the focused route now.

The focused Files profile confirmed that a synthetic Photos root link also failed during warm-source setup: the browser remained at the Files deep link and the Photos readiness marker timed out. This is setup, before the measured Files switch. I changed the setup to use a committed full navigation to `/photos`, while the measured destination switch still clicks its Tab. Fix: `36f2f9b3e`; retrying the focused route now.
kayg commented 2026-10-06 07:42:05 +00:00
Author
Owner
Copy link

The focused Files profile now reaches the warm navigation, but the sample had no start time. The timer waited for a hard-coded #tab-files selector; the route became ready but the selector did not fire the timer. Commit 7c02eaec1 arms the capture listener for the next click (or popstate) immediately before the profile triggers that one navigation. This fixes the harness timing trigger, not application navigation.

The focused Files profile now reaches the warm navigation, but the sample had no start time. The timer waited for a hard-coded `#tab-files` selector; the route became ready but the selector did not fire the timer. Commit `7c02eaec1` arms the capture listener for the next click (or `popstate`) immediately before the profile triggers that one navigation. This fixes the harness timing trigger, not application navigation.
kayg commented 2026-10-06 07:46:13 +00:00
Author
Owner
Copy link

After the click listener change, the warm Files path reached its ready marker but still reported a null timing. The observer had a fixed 121-frame limit (about two seconds), shorter than the configured 60-second route-ready timeout for the 5k Files view. Commit f0881eeca keeps sampling until the same route-ready deadline. This is benchmark instrumentation; no UI code changed.

After the click listener change, the warm Files path reached its ready marker but still reported a null timing. The observer had a fixed 121-frame limit (about two seconds), shorter than the configured 60-second route-ready timeout for the 5k Files view. Commit `f0881eeca` keeps sampling until the same route-ready deadline. This is benchmark instrumentation; no UI code changed.
kayg commented 2026-10-06 07:52:04 +00:00
Author
Owner
Copy link

The warm sample still failed after the Files ready marker appeared. The browser can replace the document during the Tab navigation, which clears the in-page timing object. Commit 2fdc6f71a stores the action start in same-tab session storage and uses the real ready-and-painted time as a fallback after a document load. The report labels this timing source and leaves frame counts null when the in-page frame observer is unavailable. I will verify the focused Files run before the full rerun.

The warm sample still failed after the Files ready marker appeared. The browser can replace the document during the Tab navigation, which clears the in-page timing object. Commit `2fdc6f71a` stores the action start in same-tab session storage and uses the real ready-and-painted time as a fallback after a document load. The report labels this timing source and leaves frame counts null when the in-page frame observer is unavailable. I will verify the focused Files run before the full rerun.
kayg commented 2026-10-06 07:57:06 +00:00
Author
Owner
Copy link

The session-storage timing fallback was not the remaining failure: the next Files profile still timed out waiting for the Photos source while the page URL remained on the Files deep link. Commit 8b703f1a0 makes both Files and Photos source setup use a full URL navigation and explicitly waits for the expected route and its real ready marker before timing the destination Tab. The next focused run verifies this path.

The session-storage timing fallback was not the remaining failure: the next Files profile still timed out waiting for the Photos source while the page URL remained on the Files deep link. Commit `8b703f1a0` makes both Files and Photos source setup use a full URL navigation and explicitly waits for the expected route and its real ready marker before timing the destination Tab. The next focused run verifies this path.
kayg commented 2026-10-06 08:13:52 +00:00
Author
Owner
Copy link

Finding: the 5k Files route profile did not complete its warm-cache setup. On the perf VM at 390 px/light, cold and hard-refresh setup messages appeared, then the profile remained before its first warm sample for over 15 minutes. During the final check, the Chromium renderer had accumulated 13:06 elapsed time at 83.5% CPU; the profile still held the VM lock. I stopped the run and it exited 1 without a JSON result. This does not establish a first-paint latency, but it is evidence that this profile cannot reach the measured tab switch for the 5k folder within a practical window. I will isolate the 5k cold paint from warm navigation and continue with other workloads.

Finding: the 5k Files route profile did not complete its warm-cache setup. On the perf VM at 390 px/light, cold and hard-refresh setup messages appeared, then the profile remained before its first warm sample for over 15 minutes. During the final check, the Chromium renderer had accumulated 13:06 elapsed time at 83.5% CPU; the profile still held the VM lock. I stopped the run and it exited 1 without a JSON result. This does not establish a first-paint latency, but it is evidence that this profile cannot reach the measured tab switch for the 5k folder within a practical window. I will isolate the 5k cold paint from warm navigation and continue with other workloads.
kayg commented 2026-10-06 08:23:15 +00:00
Author
Owner
Copy link

The isolated 5k Files cold profile completed on the perf VM (load at lock entry: 0.01, 0.58, 1.36). On a 4× CPU throttle, first meaningful paint ranged from 2,282 ms (820 px/light) to 3,044 ms (1440 px/light); all six width/theme samples were under the 5,000 ms budget. Each cell has one sample. The matching warm-navigation run still did not reach its first warm sample after 15 minutes, so it has no valid warm-tab timing.

The same Home indexed 5,000 Files entries. Ten-request API p50/p95 were: calendar.range 42.0/74.2 ms, Analytics 40.9/70.3 ms, Files.recent 30.1/53.7 ms, and Files.quota 19.4/24.6 ms. Server idle RSS was 1,080,198,758 B and idle CPU 0.2%; the 24-client/12-second burst returned 84,077/84,077 HTTP 200 responses with p50/p95 3.1/6.1 ms, peak RSS 1,080,651,776 B, and mean CPU 326.99%. These route results are in docs/perf/runs/2026-10-06-1124-files-route-cold.json. They are initial baselines, not A/B changes.

The isolated 5k Files cold profile completed on the perf VM (load at lock entry: 0.01, 0.58, 1.36). On a 4× CPU throttle, first meaningful paint ranged from 2,282 ms (820 px/light) to 3,044 ms (1440 px/light); all six width/theme samples were under the 5,000 ms budget. Each cell has one sample. The matching warm-navigation run still did not reach its first warm sample after 15 minutes, so it has no valid warm-tab timing. The same Home indexed 5,000 Files entries. Ten-request API p50/p95 were: calendar.range 42.0/74.2 ms, Analytics 40.9/70.3 ms, Files.recent 30.1/53.7 ms, and Files.quota 19.4/24.6 ms. Server idle RSS was 1,080,198,758 B and idle CPU 0.2%; the 24-client/12-second burst returned 84,077/84,077 HTTP 200 responses with p50/p95 3.1/6.1 ms, peak RSS 1,080,651,776 B, and mean CPU 326.99%. These route results are in docs/perf/runs/2026-10-06-1124-files-route-cold.json. They are initial baselines, not A/B changes.
kayg commented 2026-10-06 08:34:49 +00:00
Author
Owner
Copy link

The full Home profile indexed 5,001 Photos rows, 20,000 Files rows, 1,001 Notes, and 365 Journal entries in 141,614 ms. It found 17 visible Photos tiles, but neither the first nor all visible decoded thumbnails appeared within either 120-second sample (cold tiles: 1,088 ms; warm tiles: 483 ms). The run then failed before writing JSON because its Analytics readiness selector matched both the visible dialog and its background copy. Its three scroll probes also reported zero movement: the profile still scrolled Window, while Photos now scrolls #route-content. I’m fixing the benchmark selectors and will rerun before treating scroll or Analytics results as valid. The thumbnail delay remains a product-path finding to profile.

The full Home profile indexed 5,001 Photos rows, 20,000 Files rows, 1,001 Notes, and 365 Journal entries in 141,614 ms. It found 17 visible Photos tiles, but neither the first nor all visible decoded thumbnails appeared within either 120-second sample (cold tiles: 1,088 ms; warm tiles: 483 ms). The run then failed before writing JSON because its Analytics readiness selector matched both the visible dialog and its background copy. Its three scroll probes also reported zero movement: the profile still scrolled Window, while Photos now scrolls #route-content. I’m fixing the benchmark selectors and will rerun before treating scroll or Analytics results as valid. The thumbnail delay remains a product-path finding to profile.
kayg commented 2026-10-06 08:43:58 +00:00
Author
Owner
Copy link

A repeat large-Home run indexed the full fixture, then the existing Photos key-photo profile saw a valid set request return 200 and its following Undo return 404 (expected 204 at apps/web/e2e/photos-perf.mjs:535). The first profile attempt completed the key-photo subprofile, so this may depend on background indexing/timing. I kept the expectation unchanged and filed the focused follow-up as #1193. I will skip only this separate mutation subprofile for the Photos grid rerun; its result remains an unresolved correctness finding.

A repeat large-Home run indexed the full fixture, then the existing Photos key-photo profile saw a valid set request return 200 and its following Undo return 404 (expected 204 at `apps/web/e2e/photos-perf.mjs:535`). The first profile attempt completed the key-photo subprofile, so this may depend on background indexing/timing. I kept the expectation unchanged and filed the focused follow-up as #1193. I will skip only this separate mutation subprofile for the Photos grid rerun; its result remains an unresolved correctness finding.
kayg commented 2026-10-06 08:52:49 +00:00
Author
Owner
Copy link

Photos large-Home findings on the perf VM (release server, lock held; 5,000 photos, 20,000 Files, 1,000 Notes + 1 MiB Note, 365 Journal Log entries): startup indexed 5,001 Photos in 131.0 s. The Photos API was responsive: bucket p50/p95 5.2/8.7 ms; 60-day page 20.2/48.5 ms; middle page 14.9/34.7 ms. The first 17 visible tiles appeared in 694 ms, while no visible thumbnail decoded during either 120 s cold or warm window. At 4× CPU throttle, the scroll profile reports 72 DOM tiles at most, 14.7 fps for the 400 px/frame fling, and 293.7 ms/s total main-thread task time; removing backdrop blur returned the steady scroll to 59.8 fps. I am collecting the remaining Home-view samples before ranking this path and will profile the trace before deciding whether there is a qualifying fix.

Photos large-Home findings on the perf VM (release server, lock held; 5,000 photos, 20,000 Files, 1,000 Notes + 1 MiB Note, 365 Journal Log entries): startup indexed 5,001 Photos in 131.0 s. The Photos API was responsive: bucket p50/p95 5.2/8.7 ms; 60-day page 20.2/48.5 ms; middle page 14.9/34.7 ms. The first 17 visible tiles appeared in 694 ms, while no visible thumbnail decoded during either 120 s cold or warm window. At 4× CPU throttle, the scroll profile reports 72 DOM tiles at most, 14.7 fps for the 400 px/frame fling, and 293.7 ms/s total main-thread task time; removing backdrop blur returned the steady scroll to 59.8 fps. I am collecting the remaining Home-view samples before ranking this path and will profile the trace before deciding whether there is a qualifying fix.
kayg commented 2026-10-06 08:55:16 +00:00
Author
Owner
Copy link

The first Photos rerun exposed a probe interaction: its temporary no-backdrop-blur CSS remained active for the later Analytics material assertion, so the profile stopped before writing JSON. I kept the material assertion unchanged, remove the style immediately after its scroll sample, and committed the probe fix as 9be024d8a. The corrected full-Home profile is running again.

The first Photos rerun exposed a probe interaction: its temporary no-backdrop-blur CSS remained active for the later Analytics material assertion, so the profile stopped before writing JSON. I kept the material assertion unchanged, remove the style immediately after its scroll sample, and committed the probe fix as 9be024d8a. The corrected full-Home profile is running again.
kayg commented 2026-10-06 09:15:39 +00:00
Author
Owner
Copy link

A locked diagnostic probe found why the Photos profile stopped in its extra Analytics overlay scenario: the shared Analytics cards compute blur(16px) saturate(1.8) brightness(1.05), while the existing #504 profile assertion requires the older saturate(1.5) value. The probe also confirmed the temporary no-blur stylesheet was absent and reduced transparency was not active. I left the existing expectation unchanged per the test rule. The Photos profile now has an explicit opt-out for this separate overlay subprofile; route-perf still measures the Analytics Tab, and the Calendar profile measures Week/Month with Events.

A locked diagnostic probe found why the Photos profile stopped in its extra Analytics overlay scenario: the shared Analytics cards compute `blur(16px) saturate(1.8) brightness(1.05)`, while the existing #504 profile assertion requires the older `saturate(1.5)` value. The probe also confirmed the temporary no-blur stylesheet was absent and reduced transparency was not active. I left the existing expectation unchanged per the test rule. The Photos profile now has an explicit opt-out for this separate overlay subprofile; route-perf still measures the Analytics Tab, and the Calendar profile measures Week/Month with Events.
kayg commented 2026-10-06 09:25:32 +00:00
Author
Owner
Copy link

Photos full Home baseline is recorded in docs/perf/2026-10-06-1124.md and docs/perf/runs/2026-10-06-1124-photos-home.json at commit 741f2f145. The perf VM indexed 5,000 Photos, 20,000 Files, 1,000 Notes and 365 Log entries in 131.4 s. Photos first tiles painted in 575 ms cold / 556 ms warm, but no visible thumbnail decoded in either bounded 120 s wait. Steady scroll was 15.4 fps (frame p50/p95 66.7/83.3 ms); fling was 14.8 fps (66.7/83.4 ms). The 1 MiB Note painted in 2.427/3.007 s p50/p95 and the 5k Files Home render in 14.912/15.358 s under 4x CPU throttle. Fixture and output are committed; Photos scroll material A/B and remaining requested profiles are still running. The VM load at lock entry was 0.92/2.21/2.67.

Photos full Home baseline is recorded in `docs/perf/2026-10-06-1124.md` and `docs/perf/runs/2026-10-06-1124-photos-home.json` at commit `741f2f145`. The perf VM indexed 5,000 Photos, 20,000 Files, 1,000 Notes and 365 Log entries in 131.4 s. Photos first tiles painted in 575 ms cold / 556 ms warm, but no visible thumbnail decoded in either bounded 120 s wait. Steady scroll was 15.4 fps (frame p50/p95 66.7/83.3 ms); fling was 14.8 fps (66.7/83.4 ms). The 1 MiB Note painted in 2.427/3.007 s p50/p95 and the 5k Files Home render in 14.912/15.358 s under 4x CPU throttle. Fixture and output are committed; Photos scroll material A/B and remaining requested profiles are still running. The VM load at lock entry was 0.92/2.21/2.67.
kayg commented 2026-10-06 09:36:49 +00:00
Author
Owner
Copy link

The Photos scroll A/B is complete and recorded at 0c3f468ed / 07f6009c9. Three baseline and three candidate 8-second scrolls alternated on the same 5k-photo Home. With the predeclared keep rule (frame p50 improves at least 10%, p95 does not regress), baseline was 66.7/100.1 ms p50/p95 and 13.7 fps; static shared-filter paint during scroll was 33.3/50.1 ms and 27.1 fps. That is 50.1% p50 improvement with p95 improvement. Separate Chrome traces are committed. They show no long tasks and similar layout object counts; the controlled filter change points to shared backdrop filters as the frame-time cause. I am keeping the candidate and applying it through the existing Photos native-scroll lifecycle, with overlays excluded.

The Photos scroll A/B is complete and recorded at `0c3f468ed` / `07f6009c9`. Three baseline and three candidate 8-second scrolls alternated on the same 5k-photo Home. With the predeclared keep rule (frame p50 improves at least 10%, p95 does not regress), baseline was 66.7/100.1 ms p50/p95 and 13.7 fps; static shared-filter paint during scroll was 33.3/50.1 ms and 27.1 fps. That is 50.1% p50 improvement with p95 improvement. Separate Chrome traces are committed. They show no long tasks and similar layout object counts; the controlled filter change points to shared backdrop filters as the frame-time cause. I am keeping the candidate and applying it through the existing Photos native-scroll lifecycle, with overlays excluded.
kayg commented 2026-10-06 10:05:27 +00:00
Author
Owner
Copy link

The Calendar measurement harness did not reach its Month timing because its readiness probe filtered /api/v1/calendar/events rows by a kind field that the provider-only CalendarEventsResponse does not define. The probe counted 0/121 rows after the forced refresh. I updated it to count that response's Event views directly, checked the script with node --check, and restarted the run under /root/perf.lock at e4544c534bfd.

The Calendar measurement harness did not reach its Month timing because its readiness probe filtered `/api/v1/calendar/events` rows by a `kind` field that the provider-only `CalendarEventsResponse` does not define. The probe counted 0/121 rows after the forced refresh. I updated it to count that response's Event views directly, checked the script with `node --check`, and restarted the run under `/root/perf.lock` at `e4544c534bfd`.
kayg commented 2026-10-06 10:12:21 +00:00
Author
Owner
Copy link

The second Calendar profile reached Week rendering with the provider fixture loaded, but gridReady then timed out waiting for offscreen .activity img elements to finish. Its failure state showed .tg.ready, 11 populated columns and 48 Calendar blocks; Activity thumbnails load lazily and are not required for the Event grid to be ready. I removed that offscreen-image gate from the performance harness and retained the populated-column check. The harness now also reports Activity image counts if grid readiness itself fails. Script syntax and whitespace checks pass; retrying on the perf VM.

The second Calendar profile reached Week rendering with the provider fixture loaded, but `gridReady` then timed out waiting for offscreen `.activity img` elements to finish. Its failure state showed `.tg.ready`, 11 populated columns and 48 Calendar blocks; Activity thumbnails load lazily and are not required for the Event grid to be ready. I removed that offscreen-image gate from the performance harness and retained the populated-column check. The harness now also reports Activity image counts if grid readiness itself fails. Script syntax and whitespace checks pass; retrying on the perf VM.
kayg commented 2026-10-06 10:20:36 +00:00
Author
Owner
Copy link

The Calendar profile completed with 121 provider Events (120 timed fixtures plus the adjacent-day Event) on October 2026's 42-cell Month grid. Month route-ready p50/p95 was 2,483/2,510 ms at 1440 px, 1,639/1,984 ms at 820 px, and 2,451/2,834 ms at 390 px; each cell has three samples on a 4× CPU throttle (6× at 390 px). On the 1440 px Week, trackpad zoom measured 100.1/216.7 ms frame p50/p95 with 42 long tasks (236 ms longest); horizontal fling measured 83.3/166.6 ms with 11 long tasks (167 ms longest). The captured Chrome trace reports 2,594.5 ms in FunctionCall slices and 1,060.5 ms in UpdateLayoutTree over the interaction trace. I have saved the full metrics and desktop Week/Month traces for root-cause review.

The Calendar profile completed with 121 provider Events (120 timed fixtures plus the adjacent-day Event) on October 2026's 42-cell Month grid. Month route-ready p50/p95 was 2,483/2,510 ms at 1440 px, 1,639/1,984 ms at 820 px, and 2,451/2,834 ms at 390 px; each cell has three samples on a 4× CPU throttle (6× at 390 px). On the 1440 px Week, trackpad zoom measured 100.1/216.7 ms frame p50/p95 with 42 long tasks (236 ms longest); horizontal fling measured 83.3/166.6 ms with 11 long tasks (167 ms longest). The captured Chrome trace reports 2,594.5 ms in FunctionCall slices and 1,060.5 ms in UpdateLayoutTree over the interaction trace. I have saved the full metrics and desktop Week/Month traces for root-cause review.
kayg commented 2026-10-06 10:57:49 +00:00
Author
Owner
Copy link

The reduced-seed warm-route rerun reproduced the Photos readiness failure. It reached the warm Files pass, then route-perf.mjs waited for Photos content while the browser URL still pointed at /files?path=Files%2FBench%2F5k-folder; the marker timed out after 60 seconds. The runner returned no JSON, so it provides no valid cold/warm timings. I am recording warm route switching as a measurement gap and continuing with focused profiles.

The reduced-seed warm-route rerun reproduced the Photos readiness failure. It reached the warm Files pass, then `route-perf.mjs` waited for Photos content while the browser URL still pointed at `/files?path=Files%2FBench%2F5k-folder`; the marker timed out after 60 seconds. The runner returned no JSON, so it provides no valid cold/warm timings. I am recording warm route switching as a measurement gap and continuing with focused profiles.
kayg commented 2026-10-06 11:15:15 +00:00
Author
Owner
Copy link

The perf VM is single-tenant and its lock is still held by another isolated run, so the queued Money profile has not started. I have prepared the remaining route and startup profiles locally. I will resume the VM measurements when the lock is released; no overlapping measurements are running.

The perf VM is single-tenant and its lock is still held by another isolated run, so the queued Money profile has not started. I have prepared the remaining route and startup profiles locally. I will resume the VM measurements when the lock is released; no overlapping measurements are running.
kayg commented 2026-10-06 11:17:04 +00:00
Author
Owner
Copy link

Correction to the reduced-seed warm-profile note: the runner logged measuring warm Files, then failed while trying to prepare Photos as the source for that target. It did not complete a warm Files sample. Only the Calendar and Notes warm samples completed before the failure, and no JSON was written.

Correction to the reduced-seed warm-profile note: the runner logged `measuring warm Files`, then failed while trying to prepare Photos as the source for that target. It did not complete a warm Files sample. Only the Calendar and Notes warm samples completed before the failure, and no JSON was written.
kayg commented 2026-10-06 12:17:55 +00:00
Author
Owner
Copy link

Correction to the measurement scope: the first #1124 profiles used /root/perf-1124/target/tmp, which is on native /dev/sda1; those numbers are diagnostic and are not HDD-emulated results. I recorded this in docs/perf/2026-10-06-1124.md at head c2eafe427.

The corrected profiles place server state under /srv/hdd-emu and run under the existing 8 ms / 200 IOPS I/O limit. The Money profile also records its in-lock load and rechecks the emulator with direct-I/O fio. The perf lock is currently held by another job, so these reruns have not started.

Correction to the measurement scope: the first #1124 profiles used `/root/perf-1124/target/tmp`, which is on native `/dev/sda1`; those numbers are diagnostic and are not HDD-emulated results. I recorded this in `docs/perf/2026-10-06-1124.md` at head `c2eafe427`. The corrected profiles place server state under `/srv/hdd-emu` and run under the existing 8 ms / 200 IOPS I/O limit. The Money profile also records its in-lock load and rechecks the emulator with direct-I/O fio. The perf lock is currently held by another job, so these reruns have not started.
kayg commented 2026-10-06 12:21:43 +00:00
Author
Owner
Copy link

The corrected Money run is now inside /root/perf.lock with state under /srv/hdd-emu. Its direct-I/O QD1 qualification measured 86.1 IOPS, p50 11.21 ms and p99 31.06 ms; the lock-entry load average was 6.00 / 6.60 / 5.88. The emulator is configured for 8 ms / 200 IOPS, but this loaded result is slower than the prior qualification. I am recording the observed qualification and load with the profile; QD16 and the Money samples are still running.

The corrected Money run is now inside `/root/perf.lock` with state under `/srv/hdd-emu`. Its direct-I/O QD1 qualification measured 86.1 IOPS, p50 11.21 ms and p99 31.06 ms; the lock-entry load average was 6.00 / 6.60 / 5.88. The emulator is configured for 8 ms / 200 IOPS, but this loaded result is slower than the prior qualification. I am recording the observed qualification and load with the profile; QD16 and the Money samples are still running.
kayg commented 2026-10-06 12:36:25 +00:00
Author
Owner
Copy link

The qualified-HDD Actual and Money Budget profile completed under /root/perf.lock at head 21194cd30. The 25,000-row fidelity ZIP was 15,389,723 bytes; five preview samples measured 5,955.02 / 6,118.53 ms p50/p95, 4.268 s average server CPU and 335,618,048 bytes peak RSS. The issue-reported 30–45 s did not reproduce. The two-request burst admitted one preview (200) and returned 429 for the concurrent request. This is synthetic and is not a direct comparison with the simpler native-filesystem profile.

The imported 2025-12 Budget month, fully painted (three cold and three warm samples per cell):

Width Theme Cold p50/p95 (ms) Warm p50/p95 (ms)
390 light 1,127.8 / 1,227.8 1,399.5 / 2,047.2
390 dark 878.9 / 2,109.6 1,166.8 / 1,444.3
820 light 660.1 / 1,703 1,363.4 / 1,597.3
820 dark 1,040.1 / 2,340.9 1,466.7 / 1,638.9
1,440 light 1,020.7 / 2,901.7 1,574.4 / 1,594.3
1,440 dark 1,225.6 / 2,139.9 1,509.7 / 1,764.3

Fio qualification in the same lock measured QD1 125.0 IOPS at 7.96/8.29 ms p50/p99 and QD16 200.9 IOPS at 100.14/104.33 ms. See docs/perf/2026-10-06-1124.md and its linked raw artifacts. The earlier 400 responses were benchmark setup failures: the fidelity export needs an Account-kind selection/re-preview, and the server expects no JSON body when no opening-balance acknowledgement is needed. The benchmark now mirrors both UI steps (#462, #1130).

The qualified-HDD Actual and Money Budget profile completed under `/root/perf.lock` at head `21194cd30`. The 25,000-row fidelity ZIP was 15,389,723 bytes; five preview samples measured 5,955.02 / 6,118.53 ms p50/p95, 4.268 s average server CPU and 335,618,048 bytes peak RSS. The issue-reported 30–45 s did not reproduce. The two-request burst admitted one preview (200) and returned 429 for the concurrent request. This is synthetic and is not a direct comparison with the simpler native-filesystem profile. The imported 2025-12 Budget month, fully painted (three cold and three warm samples per cell): | Width | Theme | Cold p50/p95 (ms) | Warm p50/p95 (ms) | | ---: | --- | ---: | ---: | | 390 | light | 1,127.8 / 1,227.8 | 1,399.5 / 2,047.2 | | 390 | dark | 878.9 / 2,109.6 | 1,166.8 / 1,444.3 | | 820 | light | 660.1 / 1,703 | 1,363.4 / 1,597.3 | | 820 | dark | 1,040.1 / 2,340.9 | 1,466.7 / 1,638.9 | | 1,440 | light | 1,020.7 / 2,901.7 | 1,574.4 / 1,594.3 | | 1,440 | dark | 1,225.6 / 2,139.9 | 1,509.7 / 1,764.3 | Fio qualification in the same lock measured QD1 125.0 IOPS at 7.96/8.29 ms p50/p99 and QD16 200.9 IOPS at 100.14/104.33 ms. See `docs/perf/2026-10-06-1124.md` and its linked raw artifacts. The earlier 400 responses were benchmark setup failures: the fidelity export needs an Account-kind selection/re-preview, and the server expects no JSON body when no opening-balance acknowledgement is needed. The benchmark now mirrors both UI steps (#462, #1130).
kayg commented 2026-10-06 13:45:59 +00:00
Author
Owner
Copy link

#1124 performance review report — head c0c3bf2331d044ed7709e3c526023ddbefc77a60 (branch job/perf-1124; no push).

The measured runs, environment qualification, raw profiles, and traces are in docs/perf/2026-10-06-1124.md and docs/perf/runs/2026-10-06-1124-*.

Ranked paths and changes

Path Before After / disposition
Large Home Files render, 5,000 entries 14,912 / 15,358 ms p50/p95 at 4× CPU throttle; 27 virtual DOM rows, 13,442 ms renderer marker No code change. A 5k warm-route attempt hit its 12-minute setup limit before a sample; derived-view cost is a likely cause, but no Chrome trace isolated it.
Calendar Week zoom 8.5 fps; 100.1 / 216.7 ms frame p50/p95; 42 long tasks, 236 ms max No code change. Trace recorded 1,060.5 ms UpdateLayoutTree and 530.1 ms Layout; per-frame --hour grid-height updates are a likely layout cause.
Photos feed scroll 13.7 fps; 66.7 / 100.1 ms frame p50/p95 Kept 685b1105f: 27.1 fps; 33.3 / 50.1 ms, 50.1% p50 improvement and lower p95. The controlled A/B points to shared backdrop filters as the main cost.
Actual import, parsed-document reuse candidate Baseline aggregate p50/p95 3,711.33 / 4,898.15 ms Candidate 3,427.41 / 3,705.91 ms; p50 improved 7.65%, below the pre-set 10% keep rule, so the candidate was reverted.

Files, Calendar and Photos A/B profiles ran on the VM's native filesystem and are diagnostic, not HDD results. The synthetic HDD Actual preview used a 25,000-transaction, 15,389,723-byte archive and measured 5,955.02 / 6,118.53 ms p50/p95 over five samples. It did not reproduce the reported 30–45 seconds and is not comparable to the simpler native fixture.

The HDD emulator passed qualification (QD1 125.0 IOPS, 7.96/8.29 ms p50/p99; QD16 200.9 IOPS, 100.14/104.33 ms). The populated Budget month fully-painted ready p50/p95, cold / warm, was: 390 light 1,127.8/1,227.8 / 1,399.5/2,047.2 ms; 390 dark 878.9/2,109.6 / 1,166.8/1,444.3; 820 light 660.1/1,703 / 1,363.4/1,597.3; 820 dark 1,040.1/2,340.9 / 1,466.7/1,638.9; 1440 light 1,020.7/2,901.7 / 1,574.4/1,594.3; 1440 dark 1,225.6/2,139.9 / 1,509.7/1,764.3.

The common-route sample covered 24 routes. The slowest p50/p95 were Calendar range 42.0/74.2 ms, Analytics 40.9/70.3 ms, Files recent 30.1/53.7 ms and Files quota 19.4/24.6 ms. Idle RSS was 1,080,198,758 bytes and CPU 0.2%. A 12-second, concurrency-24 burst completed 84,077 requests with all 200 responses and 3.1/6.1 ms response p50/p95.

Built and committed

  • Scroll-scoped shared-filter suppression with one root Photos scroll marker and scrollend / 140 ms fallback: apps/web/src/lib/photos/PhotoTimeline.svelte, packages/ui/src/tokens.css.
  • Production-route Photos A/B, state verification and Mac-platform screenshot capture harness: apps/web/e2e/photos-perf.mjs.
  • Money import benchmark follows Account-kind review, discard/re-preview and inferred-opening-balance confirmation: bench/money-import-actual-462.mjs.
  • Profile report, JSON and Chrome trace artifacts under docs/perf/; exact perf-lint pins refreshed after merging origin/dev.

Commits: 685b1105f, 95b98a770, 21194cd30, merge 3c3881845, and report c0c3bf233. Decisions: keep the Photos change because it passed the 10% p50 rule without p95 regression; revert document reuse because it did not. No DESIGN decision was needed.

Gates (verbatim summaries)

cargo fmt --check: exit 0, no output.

cd apps/web && bun run check:

perf-lint: PASS; 0 violations; 22359 scoped exceptions
svelte-check found 0 errors and 2 warnings in 2 files

The two warnings are empty CSS rulesets in packages/ui/src/components/calendar/AttachmentDeck.svelte:1055 and AgendaList.svelte:1277. No Rust source changed, so per-crate clippy/test gates did not apply.

cd apps/web && bun run test:

Test Files  274 passed (274)
     Tests  1907 passed (1907)
  Duration  444.10s (transform 32%, environment 25%, import 22%, tests 17%, setup 5%)

bun run build passed after the merge. cargo clean removed 15,539 files / 6.3 GiB; apps/web/build was deleted after verification.

Known gaps and UX gaps

  • The post-change Photos A/B on HDD and six Mac-platform screenshots (390/820/1440, light/dark) did not run: the perf VM lock remained held by another job until the four-hour time limit. The checked-in harness can capture them; no screenshots are attached.
  • Warm Tab switching, Search typing, and startup time after a migration remain unmeasured. The top-20 route set was not collected; 24 routes were sampled, with the slowest four listed above.
  • Files and Calendar causes remain likely diagnoses, not fixes. Photos A/B is native-filesystem evidence only. The money fixture is synthetic and does not reproduce the reported latency.
  • Existing follow-ups: Files/Calendar #549, Actual import #462, Search #258, startup #1161. The Paths #549 and #462 have this review's measurements in their comments.
  • UX gap closed: the Photos optimization applies to scroll input through the shared feed event and clears on scrollend or a short fallback; it adds no input-specific motion branch. UX gap left: required Mac-rendered screenshot review is pending because the locked VM run did not start.

For the merge round: capture the pending Photos screenshots and HDD A/B under /root/perf.lock; measure warm Tabs, Search typing and startup after migration. No API route or Rust contract changed in this job.

#1124 performance review report — head `c0c3bf2331d044ed7709e3c526023ddbefc77a60` (branch `job/perf-1124`; no push). The measured runs, environment qualification, raw profiles, and traces are in `docs/perf/2026-10-06-1124.md` and `docs/perf/runs/2026-10-06-1124-*`. ## Ranked paths and changes | Path | Before | After / disposition | | --- | --- | --- | | Large Home Files render, 5,000 entries | 14,912 / 15,358 ms p50/p95 at 4× CPU throttle; 27 virtual DOM rows, 13,442 ms renderer marker | No code change. A 5k warm-route attempt hit its 12-minute setup limit before a sample; derived-view cost is a likely cause, but no Chrome trace isolated it. | | Calendar Week zoom | 8.5 fps; 100.1 / 216.7 ms frame p50/p95; 42 long tasks, 236 ms max | No code change. Trace recorded 1,060.5 ms `UpdateLayoutTree` and 530.1 ms `Layout`; per-frame `--hour` grid-height updates are a likely layout cause. | | Photos feed scroll | 13.7 fps; 66.7 / 100.1 ms frame p50/p95 | Kept `685b1105f`: 27.1 fps; 33.3 / 50.1 ms, 50.1% p50 improvement and lower p95. The controlled A/B points to shared backdrop filters as the main cost. | | Actual import, parsed-document reuse candidate | Baseline aggregate p50/p95 3,711.33 / 4,898.15 ms | Candidate 3,427.41 / 3,705.91 ms; p50 improved 7.65%, below the pre-set 10% keep rule, so the candidate was reverted. | Files, Calendar and Photos A/B profiles ran on the VM's native filesystem and are diagnostic, not HDD results. The synthetic HDD Actual preview used a 25,000-transaction, 15,389,723-byte archive and measured 5,955.02 / 6,118.53 ms p50/p95 over five samples. It did not reproduce the reported 30–45 seconds and is not comparable to the simpler native fixture. The HDD emulator passed qualification (QD1 125.0 IOPS, 7.96/8.29 ms p50/p99; QD16 200.9 IOPS, 100.14/104.33 ms). The populated Budget month fully-painted ready p50/p95, cold / warm, was: 390 light 1,127.8/1,227.8 / 1,399.5/2,047.2 ms; 390 dark 878.9/2,109.6 / 1,166.8/1,444.3; 820 light 660.1/1,703 / 1,363.4/1,597.3; 820 dark 1,040.1/2,340.9 / 1,466.7/1,638.9; 1440 light 1,020.7/2,901.7 / 1,574.4/1,594.3; 1440 dark 1,225.6/2,139.9 / 1,509.7/1,764.3. The common-route sample covered 24 routes. The slowest p50/p95 were Calendar range 42.0/74.2 ms, Analytics 40.9/70.3 ms, Files recent 30.1/53.7 ms and Files quota 19.4/24.6 ms. Idle RSS was 1,080,198,758 bytes and CPU 0.2%. A 12-second, concurrency-24 burst completed 84,077 requests with all 200 responses and 3.1/6.1 ms response p50/p95. ## Built and committed - Scroll-scoped shared-filter suppression with one root Photos scroll marker and `scrollend` / 140 ms fallback: `apps/web/src/lib/photos/PhotoTimeline.svelte`, `packages/ui/src/tokens.css`. - Production-route Photos A/B, state verification and Mac-platform screenshot capture harness: `apps/web/e2e/photos-perf.mjs`. - Money import benchmark follows Account-kind review, discard/re-preview and inferred-opening-balance confirmation: `bench/money-import-actual-462.mjs`. - Profile report, JSON and Chrome trace artifacts under `docs/perf/`; exact perf-lint pins refreshed after merging `origin/dev`. Commits: `685b1105f`, `95b98a770`, `21194cd30`, merge `3c3881845`, and report `c0c3bf233`. Decisions: keep the Photos change because it passed the 10% p50 rule without p95 regression; revert document reuse because it did not. No DESIGN decision was needed. ## Gates (verbatim summaries) `cargo fmt --check`: exit 0, no output. `cd apps/web && bun run check`: ``` perf-lint: PASS; 0 violations; 22359 scoped exceptions svelte-check found 0 errors and 2 warnings in 2 files ``` The two warnings are empty CSS rulesets in `packages/ui/src/components/calendar/AttachmentDeck.svelte:1055` and `AgendaList.svelte:1277`. No Rust source changed, so per-crate clippy/test gates did not apply. `cd apps/web && bun run test`: ``` Test Files 274 passed (274) Tests 1907 passed (1907) Duration 444.10s (transform 32%, environment 25%, import 22%, tests 17%, setup 5%) ``` `bun run build` passed after the merge. `cargo clean` removed 15,539 files / 6.3 GiB; `apps/web/build` was deleted after verification. ## Known gaps and UX gaps - The post-change Photos A/B on HDD and six Mac-platform screenshots (390/820/1440, light/dark) did not run: the perf VM lock remained held by another job until the four-hour time limit. The checked-in harness can capture them; no screenshots are attached. - Warm Tab switching, Search typing, and startup time after a migration remain unmeasured. The top-20 route set was not collected; 24 routes were sampled, with the slowest four listed above. - Files and Calendar causes remain likely diagnoses, not fixes. Photos A/B is native-filesystem evidence only. The money fixture is synthetic and does not reproduce the reported latency. - Existing follow-ups: Files/Calendar #549, Actual import #462, Search #258, startup #1161. The Paths #549 and #462 have this review's measurements in their comments. - UX gap closed: the Photos optimization applies to scroll input through the shared feed event and clears on scrollend or a short fallback; it adds no input-specific motion branch. UX gap left: required Mac-rendered screenshot review is pending because the locked VM run did not start. For the merge round: capture the pending Photos screenshots and HDD A/B under `/root/perf.lock`; measure warm Tabs, Search typing and startup after migration. No API route or Rust contract changed in this job.
kayg commented 2026-10-06 18:14:11 +00:00
Author
Owner
Copy link

Starting round 2 on job/perf2-1124, base c0c3bf2331d044ed7709e3c526023ddbefc77a60 (job/perf-1124). Read the first-round report and DESIGN §§28–32, 39, 59. Priorities: 5k Files, busy Calendar, search keystrokes; then warm Tabs and migration startup. Keep only interleaved perf-VM A/B changes with ≥10% p50 gain and no p95 regression. New measurements use perf-run calternal perf2-1124 (current shared-host lock rule), with in-lock load and production builds. No push or deploy.

Starting round 2 on `job/perf2-1124`, base `c0c3bf2331d044ed7709e3c526023ddbefc77a60` (job/perf-1124). Read the first-round report and DESIGN §§28–32, 39, 59. Priorities: 5k Files, busy Calendar, search keystrokes; then warm Tabs and migration startup. Keep only interleaved perf-VM A/B changes with ≥10% p50 gain and no p95 regression. New measurements use `perf-run calternal perf2-1124` (current shared-host lock rule), with in-lock load and production builds. No push or deploy.
kayg commented 2026-10-06 18:27:45 +00:00
Author
Owner
Copy link

Finding: Files builds CollectionItem metadata for every loaded entry on each page and sort. Current-day labels call shared formatTime, which calls calendarTimeZone → systemTimeZone → a new Intl.DateTimeFormat plus resolvedOptions for each label, despite the shared formatter cache. Candidate leaves an omitted zone implicit in Intl; confirmed User zones and explicit caller zones still take precedence.

Focused regression: 5,000 timestamp labels make no formatter construction after priming; explicit/confirmed zone precedence is covered. First provisional perf-VM pair (same production builds, 4× CPU throttle, 5k real files, /srv/hdd-emu/tmp, shared perf-run lock): Files open 63,395.94 → 6,727.09 ms; sort 2,682.60 → 466.85 ms. Later baseline opens are about 15–17 s, so the first open includes startup activity. Five interleaved pairs are still running; no slow sample is removed. Trace captures run separately after timing.

Synced once with origin/dev at merge 8a4507c55; resolved the Actual fixture conflict by preserving round-1 transaction memos and dev multi-Account/padding options. A combined fixture generated successfully. No added dependencies.

Finding: Files builds CollectionItem metadata for every loaded entry on each page and sort. Current-day labels call shared `formatTime`, which calls `calendarTimeZone` → `systemTimeZone` → a new Intl.DateTimeFormat plus resolvedOptions for each label, despite the shared formatter cache. Candidate leaves an omitted zone implicit in Intl; confirmed User zones and explicit caller zones still take precedence. Focused regression: 5,000 timestamp labels make no formatter construction after priming; explicit/confirmed zone precedence is covered. First provisional perf-VM pair (same production builds, 4× CPU throttle, 5k real files, `/srv/hdd-emu/tmp`, shared perf-run lock): Files open 63,395.94 → 6,727.09 ms; sort 2,682.60 → 466.85 ms. Later baseline opens are about 15–17 s, so the first open includes startup activity. Five interleaved pairs are still running; no slow sample is removed. Trace captures run separately after timing. Synced once with origin/dev at merge `8a4507c55`; resolved the Actual fixture conflict by preserving round-1 transaction memos and dev multi-Account/padding options. A combined fixture generated successfully. No added dependencies.
kayg commented 2026-10-06 18:48:59 +00:00
Author
Owner
Copy link

Round 2 finding: production Files timestamps construct an Intl formatter only to read the device time zone on every current-day label. Ten 500-entry pages format 27,500 labels although the list has 27 DOM rows. A separate CPU trace attributes 20,929.2 ms of baseline self time to that device-zone lookup.

Five interleaved A/B pairs on the perf VM, Chromium 4× CPU throttle, same real 5k folder: all entries painted p50/p95 15,524.23/63,395.94 ms → 6,203.18/8,168.22 ms (60.04% p50 gain); sort 2,089.35/2,682.60 → 514.98/690.22 ms (75.35% gain). Keep commit a296c8689. Omitted device zone now stays implicit in the existing cached formatter; confirmed User zones and explicit caller zones retain precedence. The focused tests pass (37 tests).

Evidence and qualification limits: docs/perf/2026-10-06-1124-round2.md and its committed raw timing/CPU-stack JSON. This phase used the same 8 ms HDD-delay mount but lacked the 200 IOPS scope, so it is labelled an HDD-delay diagnostic. The next Calendar/Search comparison adds that scope and uses matched builds after the one origin/dev merge, 8a4507c55.

Current head: 0d00bece6. The full web suite produced 1,979 passing tests and two unrelated 5-second timeouts; rerun uses a 30-second timeout and the same assertions. No API changes or Rust changes.

Round 2 finding: production Files timestamps construct an Intl formatter only to read the device time zone on every current-day label. Ten 500-entry pages format 27,500 labels although the list has 27 DOM rows. A separate CPU trace attributes 20,929.2 ms of baseline self time to that device-zone lookup. Five interleaved A/B pairs on the perf VM, Chromium 4× CPU throttle, same real 5k folder: all entries painted p50/p95 15,524.23/63,395.94 ms → 6,203.18/8,168.22 ms (60.04% p50 gain); sort 2,089.35/2,682.60 → 514.98/690.22 ms (75.35% gain). Keep commit a296c8689. Omitted device zone now stays implicit in the existing cached formatter; confirmed User zones and explicit caller zones retain precedence. The focused tests pass (37 tests). Evidence and qualification limits: docs/perf/2026-10-06-1124-round2.md and its committed raw timing/CPU-stack JSON. This phase used the same 8 ms HDD-delay mount but lacked the 200 IOPS scope, so it is labelled an HDD-delay diagnostic. The next Calendar/Search comparison adds that scope and uses matched builds after the one origin/dev merge, 8a4507c55. Current head: 0d00bece6. The full web suite produced 1,979 passing tests and two unrelated 5-second timeouts; rerun uses a 30-second timeout and the same assertions. No API changes or Rust changes.
kayg commented 2026-10-06 19:22:43 +00:00
Author
Owner
Copy link

Round 2 build audit: the transform-only override did not produce the intended Calendar/Search baseline. Compiled Calendar code in both builds uses the vertical/zoom root predicate, and both compiled Search observers have the four-element intrinsic set. Treat calendar-search-ab.json as A/A diagnostic variance, not candidate retention evidence. No Calendar or Search change is retained on that evidence. The earlier Files timestamp A/B used normal builds; compiled code confirms its intended difference.

I rebuilt the normal production baseline from restored source files. Its compiled Calendar predicate is horizontal-only and its Search observer includes the surface and results viewport. Future comparisons use this audited build.

The new benchmark creates 240 Logs through one real batch API call and asserts 240 rows in Calendar/range. It also asserts 121 provider Events. Earlier raw fixture metadata described written Log lines without verifying their projection; do not treat those rows as a verified 240-Log Calendar fixture.

Separate trace on the verified fixture: Calendar's date-settling function has 2,677.68 ms of self time across the recorded interactions. Vertical scrollend runs the horizontal date callback and repeated offset reads even when the date did not move. Candidate: skip this path when the horizontal offset and glide target did not change. Regression checks 20 vertical completions, retained column identity, and an early horizontal completion; 17 TimeGrid tests pass. The shared day-surface proposal is not retained.

Search candidate now batches state and observer height invalidations at one frame boundary and observes intrinsic content rather than its own animated surface. Content/input still render without waiting for that read; CSS reduced-motion and shared motion remain.

Head 7dcc23561. Full web test rerun passed with unchanged assertions and a 30-second timeout: Test Files 282 passed (282); Tests 1981 passed (1981). Production rebuild and measured candidate decision follow.

Round 2 build audit: the transform-only override did not produce the intended Calendar/Search baseline. Compiled Calendar code in both builds uses the vertical/zoom root predicate, and both compiled Search observers have the four-element intrinsic set. Treat calendar-search-ab.json as A/A diagnostic variance, not candidate retention evidence. No Calendar or Search change is retained on that evidence. The earlier Files timestamp A/B used normal builds; compiled code confirms its intended difference. I rebuilt the normal production baseline from restored source files. Its compiled Calendar predicate is horizontal-only and its Search observer includes the surface and results viewport. Future comparisons use this audited build. The new benchmark creates 240 Logs through one real batch API call and asserts 240 rows in Calendar/range. It also asserts 121 provider Events. Earlier raw fixture metadata described written Log lines without verifying their projection; do not treat those rows as a verified 240-Log Calendar fixture. Separate trace on the verified fixture: Calendar's date-settling function has 2,677.68 ms of self time across the recorded interactions. Vertical scrollend runs the horizontal date callback and repeated offset reads even when the date did not move. Candidate: skip this path when the horizontal offset and glide target did not change. Regression checks 20 vertical completions, retained column identity, and an early horizontal completion; 17 TimeGrid tests pass. The shared day-surface proposal is not retained. Search candidate now batches state and observer height invalidations at one frame boundary and observes intrinsic content rather than its own animated surface. Content/input still render without waiting for that read; CSS reduced-motion and shared motion remain. Head 7dcc23561. Full web test rerun passed with unchanged assertions and a 30-second timeout: Test Files 282 passed (282); Tests 1981 passed (1981). Production rebuild and measured candidate decision follow.
kayg commented 2026-10-06 19:58:50 +00:00
Author
Owner
Copy link

Round 2 finding, HEAD 29c709c4a: audited five interleaved A/B pairs on perf-test (4× Chromium CPU, Mac platform, HDD 8 ms + 200 IOPS + 150 MB/s, asserted 5k Files / 240 API Log projections / 121 CalDAV Events).

Search input-to-two-painted-frames: p50 177.2 → 145.1 ms (18.12% gain); p95 297 → 296 ms. Maximum 314.6 → 354.8 ms: the worst tail remains a gap. Keep the shared-frame intrinsic-height batching. A separate marked CPU trace attributes typing Layout 281.88 → 241.84 ms and Paint 317.99 → 231.08 ms; these are diagnostic trace costs, not latency samples.

Reject the Calendar date-settle candidate: Week open p50 7386 → 6842 ms (7.36%, below threshold), p95 9978 → 15171 ms; vertical frame p50 75.1 → 68.7 ms (8.52%, below threshold). It is reverted in c0fe714d1, including its own regression and exact syntax pins. No existing test expectation was changed.

The next Calendar candidate preserves horizontal scrollend and sticky labels, and keeps the existing shared cheaper material through vertical steps using a 140 ms quiet interval. Two new ownership/quiet tests plus all existing TimeGrid tests pass (18 tests). It is not retained until a separate A/B meets the threshold. Files is being re-qualified under the full throttled disk profile, including first paint and genuinely warm destination Tabs.

Round 2 finding, HEAD 29c709c4a: audited five interleaved A/B pairs on perf-test (4× Chromium CPU, Mac platform, HDD 8 ms + 200 IOPS + 150 MB/s, asserted 5k Files / 240 API Log projections / 121 CalDAV Events). Search input-to-two-painted-frames: p50 177.2 → 145.1 ms (18.12% gain); p95 297 → 296 ms. Maximum 314.6 → 354.8 ms: the worst tail remains a gap. Keep the shared-frame intrinsic-height batching. A separate marked CPU trace attributes typing Layout 281.88 → 241.84 ms and Paint 317.99 → 231.08 ms; these are diagnostic trace costs, not latency samples. Reject the Calendar date-settle candidate: Week open p50 7386 → 6842 ms (7.36%, below threshold), p95 9978 → 15171 ms; vertical frame p50 75.1 → 68.7 ms (8.52%, below threshold). It is reverted in c0fe714d1, including its own regression and exact syntax pins. No existing test expectation was changed. The next Calendar candidate preserves horizontal scrollend and sticky labels, and keeps the existing shared cheaper material through vertical steps using a 140 ms quiet interval. Two new ownership/quiet tests plus all existing TimeGrid tests pass (18 tests). It is not retained until a separate A/B meets the threshold. Files is being re-qualified under the full throttled disk profile, including first paint and genuinely warm destination Tabs.
kayg commented 2026-10-06 20:53:46 +00:00
Author
Owner
Copy link

Round 2 progress, HEAD d30996533. The retained Files formatting slice qualifies for all-5k readiness (p50 14944.70 →7027.09 ms, p95 74272.21 →7978.86) and Name reverse sort (2883.07 →683.00, p95 3523.63 →823.46). First paint gains only 5.37%. Files synthetic scroll p95 rises 91.5 →97.1 ms; #1227 records the remaining paint/paging work and that adverse result. Raw data and costs are committed in docs/perf/2026-10-06-1124-round2.md.

Search remains qualified (177.2 →145.1 ms p50, 297 →296 p95), with its worse maximum disclosed in #1228. New preview-presence sizing is covered by the production review flow.

Calendar scroll-paint prototypes improve Week/Month scroll p50 and p95, but fail the opening p95 requirement. All samples are kept, including repeatable ~24 s first-after Week opens. Simpler selectors did not resolve that cost. Vertical scrolling now keeps its original clip and hour-label material; only existing shared chrome filters change. Loading-time initial positioning is also excluded from gesture paint, matching the horizontal tracker contract. Its focused 29 tests pass. This is the last qualification candidate; it will be reverted if it fails the rule.

Completed gates on the prior working slice, quoted:

perf-lint: PASS; 0 violations; 22615 scoped exceptions
svelte-check found 0 errors and 2 warnings in 2 files
 Test Files  283 passed (283)
      Tests  1982 passed (1982)

The loading guard's current check/test and final VM comparison are running. No Rust code or dependencies were changed. No push, deploy or merge beyond the one authorized origin/dev refresh was done. Startup follow-up is #1225; it explicitly reports same-release pending backfills rather than a prior-release upgrade.

Round 2 progress, HEAD d30996533. The retained Files formatting slice qualifies for all-5k readiness (p50 14944.70 →7027.09 ms, p95 74272.21 →7978.86) and Name reverse sort (2883.07 →683.00, p95 3523.63 →823.46). First paint gains only 5.37%. Files synthetic scroll p95 rises 91.5 →97.1 ms; #1227 records the remaining paint/paging work and that adverse result. Raw data and costs are committed in docs/perf/2026-10-06-1124-round2.md. Search remains qualified (177.2 →145.1 ms p50, 297 →296 p95), with its worse maximum disclosed in #1228. New preview-presence sizing is covered by the production review flow. Calendar scroll-paint prototypes improve Week/Month scroll p50 and p95, but fail the opening p95 requirement. All samples are kept, including repeatable ~24 s first-after Week opens. Simpler selectors did not resolve that cost. Vertical scrolling now keeps its original clip and hour-label material; only existing shared chrome filters change. Loading-time initial positioning is also excluded from gesture paint, matching the horizontal tracker contract. Its focused 29 tests pass. This is the last qualification candidate; it will be reverted if it fails the rule. Completed gates on the prior working slice, quoted: ``` perf-lint: PASS; 0 violations; 22615 scoped exceptions svelte-check found 0 errors and 2 warnings in 2 files Test Files 283 passed (283) Tests 1982 passed (1982) ``` The loading guard's current check/test and final VM comparison are running. No Rust code or dependencies were changed. No push, deploy or merge beyond the one authorized origin/dev refresh was done. Startup follow-up is #1225; it explicitly reports same-release pending backfills rather than a prior-release upgrade.
kayg commented 2026-10-06 21:16:34 +00:00
Author
Owner
Copy link

Calendar paint candidates were reverted: all four paired phases regressed opening p95. The last resource trace records /api/v1/calendar/items at 7897.5 ms. A new client candidate overlaps the existing Task read with Range and Items, and still publishes one complete Calendar snapshot. It adds no endpoint, request or partial rendering. A regression checks request overlap and delayed cache publication. Qualification is pending a five-pair run; rejected trials and raw samples are committed in 009046aed.

Calendar paint candidates were reverted: all four paired phases regressed opening p95. The last resource trace records `/api/v1/calendar/items` at 7897.5 ms. A new client candidate overlaps the existing Task read with Range and Items, and still publishes one complete Calendar snapshot. It adds no endpoint, request or partial rendering. A regression checks request overlap and delayed cache publication. Qualification is pending a five-pair run; rejected trials and raw samples are committed in `009046aed`.
kayg commented 2026-10-06 21:41:34 +00:00
Author
Owner
Copy link

The global overlap was reverted: Week p50/p95 4943.63/6272.14 → 6453.27/11272.06 ms; Month 2971.95/4298.34 → 2669.95/3607.41 ms (10.16% p50 gain). Follow-up #1232 covers opening and #1230 covers scroll. Source inspection distinguishes Week fan-out of seven-day chunks from Month’s one four-to-six-week window. Commit 7e2eb76b9 overlaps only the 28–42-day window and retains the shorter Grid pipeline, with regressions for both scheduling paths and complete cache publication. Web check passes (0 errors, 2 unchanged empty-rule warnings), 45 adapter tests pass, and the production build passes. The new five-pair comparison remains pending; this is not yet a retained performance gain. Warm Tab readiness was corrected after the separate API snapshot reported three entries while the loaded UI showed four; it now waits for the completed accessible Files label and still asserts matching before/after destination identity.

The global overlap was reverted: Week p50/p95 4943.63/6272.14 → 6453.27/11272.06 ms; Month 2971.95/4298.34 → 2669.95/3607.41 ms (10.16% p50 gain). Follow-up #1232 covers opening and #1230 covers scroll. Source inspection distinguishes Week fan-out of seven-day chunks from Month’s one four-to-six-week window. Commit 7e2eb76b9 overlaps only the 28–42-day window and retains the shorter Grid pipeline, with regressions for both scheduling paths and complete cache publication. Web check passes (0 errors, 2 unchanged empty-rule warnings), 45 adapter tests pass, and the production build passes. The new five-pair comparison remains pending; this is not yet a retained performance gain. Warm Tab readiness was corrected after the separate API snapshot reported three entries while the loaded UI showed four; it now waits for the completed accessible Files label and still asserts matching before/after destination identity.
kayg commented 2026-10-06 22:06:00 +00:00
Author
Owner
Copy link

Performance round 2: partial outcome at the time limit. Head a47a4c854133dc2e15608d94eb2641f60fcfd0ab on job/perf2-1124. No push, deploy or promotion. Worktree is clean. Origin/dev was merged once, before final gates.

Built and retained: shared timestamp formatter reuse and batched intrinsic Search height measurement, including preview-presence invalidation. Qualified full-HDD five-pair results:

Operation Before p50 / p95 ms After p50 / p95 ms p50 gain
Files: all 5,000 entries 14944.70 / 74272.21 7027.09 / 7978.86 52.98%
Files: reverse Name sort 2883.07 / 3523.63 683.00 / 823.46 76.31%
Search: input to painted frame 177.20 / 297.00 145.10 / 296.00 18.12%

This does not claim a Files scroll gain: frame p95 rose 91.5 → 97.1 ms (6.12%), and whole-condition browser CPU p50 rose 3.44 → 3.88 s. #1227 contains the adverse figures. Search maximum rose 314.6 → 354.8 ms; #1228 tracks the tail. CPU/RSS clocks and limits are in the report, including whole-condition counters that cannot be attributed to one operation.

Calendar's requested third fix is unfinished. All paint, settle and request-overlap candidates were reverted. Global overlap improved Month by 10.16% in one phase but regressed Week. The final scoped Month phase did not reproduce that gain: Month p50/p95 3978.13/4831.62 → 4353.00/5031.05 ms; Week 7312.59/7602.58 → 6766.77/13383.60 ms. Its functional gates passed, but that did not qualify it. No request-window or paint prototype remains.

Warm Tab browser input p50/p95/max is 747.5/995.5/995.5 ms; the automation clock is separate. Prepared and restored Files identities match (root, four rows). Pending-backfill startup ready p50/p95 is 1906.53/2920.83 ms; first usable Files is 3568.52/4396.51 ms. This is not a previous-release SQL upgrade result.

Files: packages/ui/src/time.ts, apps/web/src/lib/time.test.ts, apps/web/src/lib/components/search-dialog.svelte; exact live pins in contracts/perf/{exceptions,adoption-1058,registry}.json; bench/interaction-1124.mjs, bench/chrome-phase-1124.py and its regression; docs/perf/2026-10-06-1124-round2.md and raw docs/perf/runs/ results. Merge conflict resolution preserved both Money fixture changes. No Rust product source or dependency changed in this round.

Final gates, verbatim:

perf-lint: PASS; 0 violations; 22615 scoped exceptions
svelte-check found 0 errors and 2 warnings in 2 files

 Test Files  282 passed (282)
      Tests  1980 passed (1980)
   Start at  23:54:45
   Duration  315.71s (environment 30%, transform 24%, import 23%, tests 16%, setup 6%)

Commands: bun run check; bun run test -- --maxWorkers=2 --testTimeout=30000. Two unchanged empty focus-rule warnings remain in AttachmentDeck and AgendaList. The first final test process received SIGTERM (143), without an assertion failure; one retry on the retained source passed. cargo fmt --check passed with no output; no per-crate clippy/test was needed. cargo clean ran (one file, 356 bytes). Web build output and saved local builds were removed. No existing product-test expectation was changed.

Chrome profile join regression (python bench/chrome-phase-1124.test.py):

Ran 1 test in 0.469s

OK

UX gaps closed: remove Search's animated-surface size feedback; batch intrinsic reads; remeasure preview presence. All six real API contexts passed keyboard selection, Escape, real empty state and offline Retry; tablet/desktop also passed preview hide/re-show. All compact bounds fit the viewport. Twenty-four macOS-platform screenshots (390/820/1440 × light/dark × Files/Week/Month/Search) are attached in the preceding comment. Claude performs visual review. Search reuses SearchField/SearchResultRow plus shared OverlaySurface, Pill/PillGroup, Checkbox, CopyLink, Kbd, Menu, ProgressiveBlur and QuickLook; Files uses FileCollection, Calendar its TimeGrid/MonthGrid. No local styling or new visual primitive was added.

Known/UX gaps left: third Calendar fix; full standard-action/touch matrices and visual review; actual previous-release upgrade; warm Tab source flamegraph; isolated per-operation CPU/RSS. The full raw traces of one rejected ready-only prototype were accidentally overwritten while collecting the latest traces. Its committed phase/latency summaries and the accepted Files/Search traces remain.

Decisions outside DESIGN: use the shared formatter's confirmed User-zone precedence with Intl's implicit device zone; coalesce intrinsic Search reads per frame; distinguish first paint from complete metadata and browser input clocks from automation readiness. Test Month separately from Grid fan-out, then remove it when it fails. Keep every slow sample, synthetic-scroll labels and maxima. Baseline.json has no exact matching profile, so use matching interleaved A/B baselines. No claim that all three requested paths were fixed.

Remaining slow-path issues with numbers: #1227 Files; #1228 Search tail; #1230 Calendar scroll; #1232 Calendar opening; #1233 warm Tabs; #1225 pending-backfill startup. For the merge round, run one browser at a time: cd apps/web && bun run test:e2e:files, bun run test:e2e:search, bun run test:e2e:calendar; prove primary actions, Copy link, keyboard/touch and shared edits. Use startup-1011 with prebuilt previous/merged binaries and the shared perf-run lock for a real SQL upgrade. Details and exact environment options are in the report. Never build on the perf VM.

Performance round 2: partial outcome at the time limit. Head `a47a4c854133dc2e15608d94eb2641f60fcfd0ab` on `job/perf2-1124`. No push, deploy or promotion. Worktree is clean. Origin/dev was merged once, before final gates. Built and retained: shared timestamp formatter reuse and batched intrinsic Search height measurement, including preview-presence invalidation. Qualified full-HDD five-pair results: | Operation | Before p50 / p95 ms | After p50 / p95 ms | p50 gain | | --- | --- | --- | --- | | Files: all 5,000 entries | 14944.70 / 74272.21 | 7027.09 / 7978.86 | 52.98% | | Files: reverse Name sort | 2883.07 / 3523.63 | 683.00 / 823.46 | 76.31% | | Search: input to painted frame | 177.20 / 297.00 | 145.10 / 296.00 | 18.12% | This does not claim a Files scroll gain: frame p95 rose 91.5 → 97.1 ms (6.12%), and whole-condition browser CPU p50 rose 3.44 → 3.88 s. #1227 contains the adverse figures. Search maximum rose 314.6 → 354.8 ms; #1228 tracks the tail. CPU/RSS clocks and limits are in the report, including whole-condition counters that cannot be attributed to one operation. Calendar's requested third fix is unfinished. All paint, settle and request-overlap candidates were reverted. Global overlap improved Month by 10.16% in one phase but regressed Week. The final scoped Month phase did not reproduce that gain: Month p50/p95 3978.13/4831.62 → 4353.00/5031.05 ms; Week 7312.59/7602.58 → 6766.77/13383.60 ms. Its functional gates passed, but that did not qualify it. No request-window or paint prototype remains. Warm Tab browser input p50/p95/max is 747.5/995.5/995.5 ms; the automation clock is separate. Prepared and restored Files identities match (root, four rows). Pending-backfill startup ready p50/p95 is 1906.53/2920.83 ms; first usable Files is 3568.52/4396.51 ms. This is not a previous-release SQL upgrade result. Files: `packages/ui/src/time.ts`, `apps/web/src/lib/time.test.ts`, `apps/web/src/lib/components/search-dialog.svelte`; exact live pins in `contracts/perf/{exceptions,adoption-1058,registry}.json`; `bench/interaction-1124.mjs`, `bench/chrome-phase-1124.py` and its regression; `docs/perf/2026-10-06-1124-round2.md` and raw `docs/perf/runs/` results. Merge conflict resolution preserved both Money fixture changes. No Rust product source or dependency changed in this round. Final gates, verbatim: ```text perf-lint: PASS; 0 violations; 22615 scoped exceptions svelte-check found 0 errors and 2 warnings in 2 files Test Files 282 passed (282) Tests 1980 passed (1980) Start at 23:54:45 Duration 315.71s (environment 30%, transform 24%, import 23%, tests 16%, setup 6%) ``` Commands: `bun run check`; `bun run test -- --maxWorkers=2 --testTimeout=30000`. Two unchanged empty focus-rule warnings remain in AttachmentDeck and AgendaList. The first final test process received SIGTERM (143), without an assertion failure; one retry on the retained source passed. `cargo fmt --check` passed with no output; no per-crate clippy/test was needed. `cargo clean` ran (one file, 356 bytes). Web build output and saved local builds were removed. No existing product-test expectation was changed. Chrome profile join regression (`python bench/chrome-phase-1124.test.py`): ```text Ran 1 test in 0.469s OK ``` UX gaps closed: remove Search's animated-surface size feedback; batch intrinsic reads; remeasure preview presence. All six real API contexts passed keyboard selection, Escape, real empty state and offline Retry; tablet/desktop also passed preview hide/re-show. All compact bounds fit the viewport. Twenty-four macOS-platform screenshots (390/820/1440 × light/dark × Files/Week/Month/Search) are attached in the preceding comment. Claude performs visual review. Search reuses SearchField/SearchResultRow plus shared OverlaySurface, Pill/PillGroup, Checkbox, CopyLink, Kbd, Menu, ProgressiveBlur and QuickLook; Files uses FileCollection, Calendar its TimeGrid/MonthGrid. No local styling or new visual primitive was added. Known/UX gaps left: third Calendar fix; full standard-action/touch matrices and visual review; actual previous-release upgrade; warm Tab source flamegraph; isolated per-operation CPU/RSS. The full raw traces of one rejected ready-only prototype were accidentally overwritten while collecting the latest traces. Its committed phase/latency summaries and the accepted Files/Search traces remain. Decisions outside DESIGN: use the shared formatter's confirmed User-zone precedence with Intl's implicit device zone; coalesce intrinsic Search reads per frame; distinguish first paint from complete metadata and browser input clocks from automation readiness. Test Month separately from Grid fan-out, then remove it when it fails. Keep every slow sample, synthetic-scroll labels and maxima. Baseline.json has no exact matching profile, so use matching interleaved A/B baselines. No claim that all three requested paths were fixed. Remaining slow-path issues with numbers: #1227 Files; #1228 Search tail; #1230 Calendar scroll; #1232 Calendar opening; #1233 warm Tabs; #1225 pending-backfill startup. For the merge round, run one browser at a time: `cd apps/web && bun run test:e2e:files`, `bun run test:e2e:search`, `bun run test:e2e:calendar`; prove primary actions, Copy link, keyboard/touch and shared edits. Use startup-1011 with prebuilt previous/merged binaries and the shared perf-run lock for a real SQL upgrade. Details and exact environment options are in the report. Never build on the perf VM.
Commenting is not possible because the repository is archived.
No Branch/Tag specified
Branches Tags
dev
wip/cards2-1083
wip/cards-1083
wip/canvas-visual
wip/canvasvis2-976
wip/calhdr-1112
wip/blocks-1125
wip/allday-1107
wip/agenda-1086
wip/trayicons2-1095
wip/trayicons-1095
wip/sidebar3-1094
wip/previewcard-1098
wip/palette2-1123
wip/palette-1093
wip/nlpchip-1127
wip/mchrome-1084
wip/mailghost2-1094
wip/mailghost-1094
wip/kbpreview2-1118
wip/kbpreview-1118
wip/kanban-1092
wip/editreg-1132
wip/editorrail3-1113
wip/editorrail2-1113
wip/editorrail-1113
wip/draw4-1101
wip/draw3-1101
wip/draw2-1101
wip/draw-1101
wip/delete-1119
job/txentry-1198
job/rev1586
job/rev2-money-ident
job/rev2-webperf
job/morph-1104
job/previewcard-1098
job/importhang-1121
job/kanban-1092
job/merge-round-7b
job/merge-round-7c
job/calcards-1115
job/collabrev3-1197
job/e2e-b-1071
job/agenda-1086
job/agenda-decks
job/allday-1107
job/browserfix
job/calact9
job/adv7c-1105
wip/collabrev3-1197
wip/calcards-1115
wip/calact9
wip/calact10
wip/browserfix
wip/agenda-decks
wip/adv7c-1105
wip/txentry-1198
wip/rev2-webperf
wip/rev2-money-ident
wip/rev1586
wip/morph-1104
wip/merge-round-7c5
wip/merge-round-7c4
wip/merge-round-7c3
wip/merge-round-7c2
wip/merge-round-7c
wip/importhang-1121
wip/fix1586
wip/e2e-b2-1071
wip/e2e-b-1071
job/noncanon-1618
wip/noncanon-1618
job/edrevert-1616
job/partial-1605
wip/partial-1605
job/calact6
job/cold-1586
job/perfvm-1587
wip/cold-1586
job/cal404-1581
wip/calact6
wip/cal404-1581
job/dname-1584
wip/dname-1584
job/merge41
wip/merge41b
wip/merge41
job/rev1573c
job/hscroll-1579
wip/rev1573c
job/rev1573b
job/rev1573
wip/fix1573
wip/hscroll2
wip/hscroll-1579
job/rev1574
wip/fix1574
wip/rev1574
job/startup-1573c
job/merge40
wip/startup-1573d
wip/startup-1573c
job/calact5
wip/remtext2
wip/merge40c
wip/merge40b
wip/merge40
job/calact4
job/startup-1573
job/calprof
wip/calprof
wip/startup-1573
job/fs1575
job/fix1570
job/calact3
job/rec1540c
wip/fix1570
job/rev1570-review
wip/calact3
job/calact2
wip/calact2
job/calcold-1560
wip/calcold-1560
wip/rec1540c
job/fix1569
wip/fix1569
job/calact-1557
wip/calact-1557
job/remtext-1568
wip/remtext-1568
job/fix1566
wip/fix1566
job/rec1540b
job/reconcile-1540
job/money2rev
wip/money2rev
job/merge39
wip/merge39
wip/reconcile-1540
job/tabswitch2
job/merge38
wip/merge38
job/autherr-1553
wip/autherr-1553
job/money-688
wip/money688rev3
wip/money688rev2
wip/money688rev
wip/money688l
wip/money688k
wip/money688j
wip/money688i
wip/money688h
wip/money688g
wip/money688f
wip/money688e
wip/money688d
wip/money688c
wip/money688b
wip/money-688
wip/tabswitch2
wip/merge37
job/merge37
job/jobnames-1538
wip/jobnames-1538
job/tabcomp-perf
job/calscript-1492
wip/tabcomp-perf
wip/calscript-1492
job/mailbody-1533
wip/mailbody-1533
job/intlfmt-1461
job/merge36
wip/intlfmt2
wip/intlfmt-1461
wip/merge36
job/merge35c
job/tagperf-1186
wip/tagrev2
wip/tagrev
wip/tagperf4
wip/tagperf2-1186
wip/tagperf-1186
wip/tagfix5
wip/tagfix3
wip/tagfast2
wip/tagfast
wip/merge35c
job/notesel-1493
wip/notesel2
wip/notesel-1493
job/merge35b
wip/merge35b
wip/m35btriage
wip/m35bratchet
wip/calsel-1491
job/calsel-1491
job/notifread-1387
job/bulkwrite-1393
wip/notifread2
wip/notifread-1387
job/slowsel-1392
wip/bulkwrite-1393
wip/bulkrev4
wip/bulkrev3
wip/bulkrev2
wip/bulkrev
wip/bulkfix4
wip/bulkfix3
wip/bulkfix2
wip/bulkfix
job/mail7-1340
job/bigscroll-1481
job/searchperf-1425
wip/mailrev4
wip/mailrev3
wip/mailrev2
wip/mailrev
wip/mail7f
wip/mail7e
wip/mail7d
wip/searchperf2
wip/searchperf-1425
wip/mail7c
wip/mail7b
wip/mail7-1340
job/photoopen-1367
job/mailsel-1449
wip/bigscroll-1481
wip/slowsel3
wip/slowsel2
wip/slowsel-1392
wip/selroot
job/journaldup
wip/journaldup8
wip/journaldup7
wip/journaldup6
wip/journaldup5
wip/journaldup4
wip/journaldup3
wip/journaldup2
wip/journaldup
wip/jdrev5
wip/jdrev4
wip/jdrev3
wip/jdrev2
wip/jdrev
wip/jdfix9
job/bignote-1349
wip/mailsel-1449
wip/bignote3
wip/bignote2
wip/bignote-1349
job/statusmark-1457
wip/statusmark-1457
wip/photoopen4
wip/photoopen3
wip/photoopen2
wip/photoopen-1367
job/authflash-1473
job/merge35
wip/authflash-1473
job/phototags-1370
wip/phototags-1370
wip/merge35
wip/m35triage
job/tick-1446
wip/tick-1446
job/restartidx-1450
job/exmove-1333
job/coldnote-1451
wip/restartidx-1450
wip/exmove-1333
wip/coldnote-1451
job/undo412-1430
wip/undo412-1430
job/calwrite-1441
job/xcancel-1443
job/calscroll-1362
wip/xcancel-1443
wip/calwrite-1441
job/xsub-1445
wip/calscroll3
wip/calscroll2
wip/calscroll-1362
wip/xsub-1445
job/linkdup-1440
wip/linkdup2
wip/linkdup-1440
wip/tabswitch4
wip/tabswitch3
wip/tabswitch-1233
wip/navrev
job/tabswitch-1233
job/mdpaste-1431
job/taskundo-1433
job/directory-1199
wip/taskundo-1433
job/remind-1432
job/nfd-1411
wip/remind-1432
wip/dirfix4
wip/dirfix3
wip/dirfix2
wip/dirfix-1199
wip/directory-1199-r
wip/directory-1199
job/readerx-1427
wip/readerx-1427
job/scrollfade-1416
job/tipphone-1439
wip/nfd-1411
job/purge-1420
job/notifrow-1414
wip/purge-1420
wip/notifrow-1414
job/slop-1385
wip/slop2
wip/slop-1385
job/tbover-1409
job/qltitle-1408
job/grantlive-1405
job/taskclick-1434
wip/tbover-1409
wip/taskclick-1434
wip/qltitle-1408
job/scrub-1422
job/hotfix35
job/ticks-1361
wip/scrub-1422
job/taskprev-1401
wip/grantlive-1405
job/ghost-1323
wip/ticks3
wip/ticks2
wip/ticks-1361
wip/taskprev-1401
job/bash3-settings
wip/ghost-1323
job/mailmorph-1368
wip/bash3-settings
job/apiaudit-1374
wip/bash3-filesphotos
job/bash3-filesphotos
job/xuser39-1402
job/bash3-money
job/mailhtml-1372
wip/mailhtml2
wip/mailhtml-1372
job/bash3-interop
job/bash3-calendar
wip/bash3-money
wip/bash3-interop
wip/bash3-calendar
wip/mailmorph2
wip/mailmorph-1368
wip/bash3-tasks
job/outro-1391
job/linkpick-1379
job/linkdel-1384
wip/linkpick2
wip/linkpick-1379
wip/linkdel3
wip/linkdel2
wip/linkdel-1384
job/mailrows-1390
wip/mailrows-1390
wip/apiaudit2
wip/apiaudit-1374
job/bash-core
wip/outro-1391
wip/bash-core2
wip/bash-core
job/bash3-notes
job/bash3-tasks
job/merge34
job/notifdel-1363
wip/merge34
wip/notifdel2
wip/notifdel-1363
job/wklink-1371
wip/wklink-1371
job/searchbar-1360
wip/searchbar2
wip/searchbar-1360
job/merge33-1122
wip/merge33c
wip/merge33b
wip/merge33-1122
job/bash-security
job/mailglass-1373
wip/bash-security2
wip/bash-security
wip/mailglass-1373
job/instaudit-1369
wip/instaudit-1369
job/tabstorm-1338
wip/tabstorm2
wip/tabstorm-1338
wip/bash-files2
wip/bash-files
job/bash-files
job/bash-money
job/editfuzz-1335
wip/bash-money
wip/moneyrev
job/tasksperf-1286
wip/tasksperf3
wip/tasksperf2
wip/tasksperf-1286
wip/editfuzz2
wip/editfuzz-1335
review/dirrev-1199
job/merge32
wip/merge32b
wip/merge32
job/bash-chrome
job/tagdnd-1187
job/bash-notes
wip/bash-notes2
wip/bash-notes
wip/bash-chrome
wip/reconcile-c1
wip/adv-c1
wip/tagdnd2-1187
wip/tagdnd-1187
job/bash-calendar
job/deploy-c1
job/editbreak-1334
job/bash-perf
wip/bash-calendar2
wip/bash-calendar
wip/bash-perf
job/hunt-20261006
job/bash-interop
job/collabloss-1197
wip/collabrev-1197
wip/collabloss5-1197
wip/collabloss4-1197
wip/collabloss3-1197
wip/collabloss2-1197
wip/collabloss-1197
wip/bash-interop
job/bash-mail
wip/bash-mail
job/vis-1293
wip/vis-1293
wip/restyle-search-r
wip/restyle-search-fix2
wip/restyle-search-fix
wip/restyle-search
job/restyle-settings
job/restyle-search
wip/restyle-settings-fix
wip/restyle-settings
job/adv31
wip/advfix-a
wip/adv31
job/bash-tasks
wip/bash-tasks
job/restyle-files
wip/restyle-files-fix
wip/restyle-files
job/fmtbar-1210
wip/fmtbar2-1210
wip/fmtbar-1210
job/perf2-1124
wip/perf2-1124
job/proflog-1204
job/merge31
wip/restyle-notes
wip/merge31
wip/proflog-1204
job/restyle-notes
job/fix-1213
wip/fix-1213
job/hide-1153
job/restyle-mailmoney
job/notestatus-1209
wip/hide-1153
wip/hiderev2-1153
wip/hiderev-1153
wip/hide5-1153
wip/hide4-1153
wip/hide3-1153
wip/hide2-1153
job/sidebarblur-1208
wip/notestatus-1209
wip/sidebarblur-1208
wip/restyle-mailmoney
job/onboard-1141
wip/onboard2-1141
wip/onboard-1141.aborted-early
wip/onboard-1141
job/adv-1202
job/segmented-1200
wip/segmented-1200
job/merge30
job/notifloop-1194
job/toastmorph-1201
wip/notifloop-1194
job/perf-1124
job/tocrail-1191
job/cards-1179
wip/cards2-1179
wip/cards-1179
wip/tocrail-1191
wip/perf-1124
wip/merge30j
job/wizchoices-1140
wip/wizchoices-1140
wip/restyle-1190
job/moneyfmt-1180
wip/moneyfmt2-1180
wip/moneyfmt-1180-r
wip/moneyfmt-1180
job/pillglass-1189
job/flags-1181
wip/flags-1181
job/restyle-1190
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/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
fixbatch-25
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
fixbatch-23
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/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
hotfix-sidebar2
job/cards-1083
job/palette-1093
job/mchrome-1084
job/e2e-a-1071
job/canvas-visual
wip/e2e-a2-1071
wip/e2e-a-1071
backup/canvas-visual-before-reconcile
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
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/staging-imap
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-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/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/theme-detail
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
job/merge-round-3-pre-rebuild
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
backup/doc-stack-r3-reset-source
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
job/dedup-375-clean
job/dedup-375-before-rebase
wip/modes-424
job/recog-417
wip/recog-417
wip/bounce-425
wip/ab-384-luna
backup/doc-stack-round3-before-rebuild
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
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-v0#1124
No description provided.