Measure first: a big meeting, a perf spec and a desktop memory run #252

Merged

littlefox merged 2 commits from issue-225-measure into main 2026-10-21 23:40:03 +00:00

Conversation 0 Commits 2 Files changed 18 +849 -10

littlefox commented 2026-10-21 22:29:51 +00:00

Owner

Copy link

Closes #225

Measuring for 0.6 (D-057), not part of the normal e2e run:

• tools/perf.sh LABEL: the perf seed (NEOKOM_SEED_PERF=1) on a lyskomd kept in memory (LYSKOM_TMPFS=1; lyskomd fsyncs each new text), 30 ms each way between relay and server (forward.py --delay), then e2e/perf/perf.spec.ts in Chromium. Numbers in e2e/perf-results/LABEL.json; tools/perf-table.py A B compares.
• tools/desktop-perf.sh LABEL: the release desktop app under Xvfb, RSS of the app and its WebKit processes over 300 posts, idle CPU.
• The app exposes window.__neokomMemory (wasm memory) and window.__neokomCalls() (calls sent by the session).

Baseline (web, desktop Chromium, main at 2026-10-21):

	baseline
login: logged in (ms)	823
login: all 301 channels listed (ms)	41983
login: server calls	1657
read 100 one-offs: step p50 (ms)	229
read 100 one-offs: step p95 (ms)	300
read 100 one-offs: server calls	459
read 300 in a chain: step p50 (ms)	116
read 300 in a chain: step p95 (ms)	125
read 300 in a chain: slowest step (ms)	12457
read 300 in a chain: server calls	3448
read 300 in a chain: style recalcs	2304
read 300 in a chain: long tasks	0
read 300 in a chain: long-task ms	0
read 300 in a chain: JS heap / wasm MB / DOM nodes at 50…300	3.3/6.1/6334 → 3.3/7.4/6575 → 3.3/7.4/6575 → 3.4/7.4/8656 → 3.4/7.4/8656 → 3.3/7.4/6577
scroll the thread: frame p95 (ms)	17
scroll the thread: worst frame (ms)	17
scroll the thread: frames over 50 ms	0
scroll the thread: long tasks	0
idle 20 s: main-thread busy (s)	0.03
idle 20 s: server calls	0
idle 20 s: layouts	1
channel, 25× Visa äldre: posts on the page	520
channel, 25× Visa äldre: DOM nodes	14569
channel, 25× Visa äldre: JS heap (MB)	3.6
channel, 25× Visa äldre: wasm memory (MB)	89.3
channel, 25× Visa äldre: server calls	564
inbox, 2 500 letters: open (ms)	2469
inbox, 2 500 letters: server calls	209
search: hits shown (ms)	23
search: long-task ms	0
at the end: JS heap (MB)	3.8
at the end: wasm memory (MB)	89.3
at the end: DOM nodes	17143
at the end: server calls in all	6472

Desktop baseline: RSS 391 MB at start (WebKitGTK, software rendering), 401 MB after login, 406 MB after 300 posts (flat); 0.04 s CPU in 20 s idle.

Found while building this, and filed: #250 (Inkorg/Alla kanaler/Nytt inlägg covered by a channel's posts or a profile), #251 (reading a thread walks it all from the root: 12.5 s steps), #249 (two flaky specs), and a correction on #237 (the inbox reads only the newest 200 texts: 2.5 s, 213 calls).

Verified: lint four ways (incl. kom-seed); the perf spec passes end to end; normal e2e changes are opt-in (latency, PERF_SPEC/PERF_LABEL), so CI's run covers the rest.

🤖 Generated with Claude Code

https://claude.ai/code/session_01XJW2ocwLunyaLiUwCBHEP8

Closes #225 Measuring for 0.6 (D-057), not part of the normal e2e run: - `tools/perf.sh LABEL`: the perf seed (`NEOKOM_SEED_PERF=1`) on a lyskomd kept in memory (`LYSKOM_TMPFS=1`; lyskomd fsyncs each new text), 30 ms each way between relay and server (`forward.py --delay`), then `e2e/perf/perf.spec.ts` in Chromium. Numbers in `e2e/perf-results/LABEL.json`; `tools/perf-table.py A B` compares. - `tools/desktop-perf.sh LABEL`: the release desktop app under Xvfb, RSS of the app and its WebKit processes over 300 posts, idle CPU. - The app exposes `window.__neokomMemory` (wasm memory) and `window.__neokomCalls()` (calls sent by the session). Baseline (web, desktop Chromium, main at 2026-10-21): | | baseline | |---|---| | login: logged in (ms) | 823 | | login: all 301 channels listed (ms) | 41983 | | login: server calls | 1657 | | read 100 one-offs: step p50 (ms) | 229 | | read 100 one-offs: step p95 (ms) | 300 | | read 100 one-offs: server calls | 459 | | read 300 in a chain: step p50 (ms) | 116 | | read 300 in a chain: step p95 (ms) | 125 | | read 300 in a chain: slowest step (ms) | 12457 | | read 300 in a chain: server calls | 3448 | | read 300 in a chain: style recalcs | 2304 | | read 300 in a chain: long tasks | 0 | | read 300 in a chain: long-task ms | 0 | | read 300 in a chain: JS heap / wasm MB / DOM nodes at 50…300 | 3.3/6.1/6334 → 3.3/7.4/6575 → 3.3/7.4/6575 → 3.4/7.4/8656 → 3.4/7.4/8656 → 3.3/7.4/6577 | | scroll the thread: frame p95 (ms) | 17 | | scroll the thread: worst frame (ms) | 17 | | scroll the thread: frames over 50 ms | 0 | | scroll the thread: long tasks | 0 | | idle 20 s: main-thread busy (s) | 0.03 | | idle 20 s: server calls | 0 | | idle 20 s: layouts | 1 | | channel, 25× Visa äldre: posts on the page | 520 | | channel, 25× Visa äldre: DOM nodes | 14569 | | channel, 25× Visa äldre: JS heap (MB) | 3.6 | | channel, 25× Visa äldre: wasm memory (MB) | 89.3 | | channel, 25× Visa äldre: server calls | 564 | | inbox, 2 500 letters: open (ms) | 2469 | | inbox, 2 500 letters: server calls | 209 | | search: hits shown (ms) | 23 | | search: long-task ms | 0 | | at the end: JS heap (MB) | 3.8 | | at the end: wasm memory (MB) | 89.3 | | at the end: DOM nodes | 17143 | | at the end: server calls in all | 6472 | Desktop baseline: RSS 391 MB at start (WebKitGTK, software rendering), 401 MB after login, 406 MB after 300 posts (flat); 0.04 s CPU in 20 s idle. Found while building this, and filed: #250 (Inkorg/Alla kanaler/Nytt inlägg covered by a channel's posts or a profile), #251 (reading a thread walks it all from the root: 12.5 s steps), #249 (two flaky specs), and a correction on #237 (the inbox reads only the newest 200 texts: 2.5 s, 213 calls). Verified: lint four ways (incl. kom-seed); the perf spec passes end to end; normal e2e changes are opt-in (latency, PERF_SPEC/PERF_LABEL), so CI's run covers the rest. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01XJW2ocwLunyaLiUwCBHEP8

littlefox added 1 commit 2026-10-21 22:29:51 +00:00

Measure first: a big meeting, a perf spec and a desktop memory run ... c5abe1fe63

0.6 is about performance (D-057), and every fix should come with a
before/after number. tools/perf.sh seeds a big meeting (NEOKOM_SEED_PERF:
~2 800 texts, a 1 500-long chain, 300 more channels, 2 500 letters) on a
lyskomd kept in memory, puts 30 ms each way between relay and server,
and runs e2e/perf/perf.spec.ts in Chromium: memory (JS heap and the
app's wasm), DOM nodes, main-thread time, long tasks, frames while
scrolling, step times and server calls. tools/desktop-perf.sh measures
the desktop app's RSS and idle CPU. tools/perf-table.py compares runs.

The app exposes its wasm memory and the session's call count to the page
for this; Chromium shows no WebSocket frames sent from wasm.

Refs #225

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XJW2ocwLunyaLiUwCBHEP8

littlefox added 1 commit 2026-10-21 22:59:05 +00:00

Perf: a scenario for reading cards by scrolling past them ... 91bc19544c

Reading with space marks few cards by being seen, the case #228 is
about. This scrolls slowly through a post's 120 unread replies.

Refs #225

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XJW2ocwLunyaLiUwCBHEP8

littlefox referenced this pull request 2026-10-21 23:11:07 +00:00

Read by being seen without reloading the thread; redraw only what changed #254

littlefox merged commit f6fda0f337 into main 2026-10-21 23:40:03 +00:00

littlefox referenced this pull request from a commit 2026-10-21 23:40:04 +00:00

Merge pull request 'Measure first: a big meeting, a perf spec and a desktop memory run' (#252) from issue-225-measure into main

littlefox referenced this pull request 2026-10-22 00:14:53 +00:00

Inkorg, Alla kanaler and Nytt inlägg do nothing while a channel's posts or a profile is open #250

littlefox referenced this pull request 2026-10-22 00:14:54 +00:00

Measure first: a meeting with thousands of posts, and numbers to compare #225

littlefox referenced this pull request 2026-10-22 14:11:05 +00:00

Release 0.6.0 #281

littlefox referenced this pull request from a commit 2026-10-22 14:11:05 +00:00

Release 0.6.0

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

area/ci

Forgejo Actions, runner, tests

  

area/desktop

Desktop build

  

area/packaging

Arch, Debian, Fedora, NixOS packages

  

area/protocol

kom-proto, wire format, charset

  

area/reading

Unread tracking, reading order, review

  

area/ui

Dioxus app, layout, interaction

  

area/web

Web build and relay

  

area/writing

Composing and sending texts

  

kind/bug

Something is broken

  

kind/feature

New functionality

  

kind/infra

Tooling, tests, CI, dev environment

  

kind/request

requests for features or changes from users

  

needs/user

Waiting for a decision or action from littlefox

  

parked

Kept for reference; no work planned until the user says so

  

perf

Run the perf workflow and its budgets on this pull request (#287)

  

security

Security-relevant

  

upstream

A proposal for Protocol A or lyskomd, for the LysKOM development group

No labels

area/ci

area/desktop

area/packaging

area/protocol

area/reading

area/ui

area/web

area/writing

kind/bug

kind/feature

kind/infra

kind/request

needs/user

parked

perf

security

upstream

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

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

littlefox/neokom!252

No description provided.