Many distinct reactions used to wrap into a tall stack (16 reactions =
4 rows at phone width). Now only the first row is shown, ending in a
"+N" chip; tapping it expands inline and a "less" chip collapses again.
Expanded state is remembered per message for the session. Nothing changes
when the reactions already fit on one row.
Overflowing chips stay in the DOM (invisible, aria-hidden, untabbable,
clipped by max-height) so the container keeps its natural width — which
keeps shrink-to-fit bubble layout stable — and each chip stays measurable.
utils/reactionOverflow.ts holds the unit-tested fit calculation; a
ResizeObserver re-fits on width changes. "+N" is forced LTR for RTL UIs.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Hovering or focusing "(edited)" shows a tooltip with only the most recent
edit as a word diff — removed words struck, added words bold — plus a
+N/−N summary. Clicking still opens the full history viewer. On touch, a
long-press on the label shows the same diff as a popout (a plain tap opens
the viewer; the message's own long-press action sheet is not triggered).
utils/wordDiff.ts is a unit-tested LCS over words that ignores whitespace-
only changes and gives up past 400 words. Only plain-text bodies are
diffed; formatted edits fall back to the viewer.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
One line under the handle in the user profile popover, from the presence the
SDK already receives (currently_active + last_active_ago): 'Active now',
'Active just now', 'Last active 12 min ago / 3 hours ago / yesterday / 5 days
ago / over a month ago'. Nothing is shown without presence data or for an
offline user with no timestamp (Hide Online Status users), so nothing new is
exposed; re-renders on the existing presence events, no polling. Wording
unit-tested; verified headless ('Active now' for an online member).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
utils/linkSafety.ts compares the registrable domain the visible text claims
(when it looks like a URL/host) with the href's; a mismatch, or a punycode
(IDN) destination, renders the anchor as SuspiciousLink, whose click opens a
confirm — "It shows matrix.lotusguild.org but goes to evil.example." with the
full URL, Cancel / Open anyway (opens in a new tab with noopener). Honest links
are untouched: same registrable domain (youtube.com text over www.youtube.com,
bbc.co.uk over news.bbc.co.uk), plain-word text, mailto:, matrix.to and Lotus
permalinks, anchors with non-text children. Comparator unit-tested (incl. a
Cyrillic-а paypal homograph); verified headless that the phish and IDN
messages are flagged, the honest ones are not, the click shows the confirm and
does not navigate, Cancel keeps you put, Open anyway opens the real target.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Reported: 'Failed to save status — server may be rate limiting. Try again.'
on the first status change in a month. Root cause: Synapse rate-limits
PUT /presence/{user}/status to ONE request per 10 s per user by default
(rc_presence.per_user: per_second 0.1, burst_count 1), shared across all of
the user's devices, and our presence heartbeat (online/away on visibility and
activity changes, from every open tab/device) spends that budget — so a manual
save that lands within 10 s of a heartbeat gets a 429, which the form showed
as a dead end.
Two fixes: (1) the status save waits out Retry-After (bounded to ~25 s) via
setPresenceWithRetry instead of failing, and the error text now says what
actually happened (rate limit / server text / offline); (2) the heartbeat
dedupes — it only sends when presence or status actually changes (/sync
already keeps us online), so it stops burning the budget in the first place.
Unit-tested; reproduced headless with a routed 10 s limiter: heartbeat ok →
save 429 → retried 8 s later → saved.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
On deafen the participant set is snapshotted; on undeafen it is diffed and,
only if it changed and you were deafened for at least 10 s, one auto-dismissing
toast says 'While you were deafened: Alice, Bob joined · Cole left' (names
capped at 3 + N more). Rides the membership stream that already drives the
join/leave sounds — no new subscriptions; PTT holds don't touch deafen so they
can't trigger it. Verified headless: short deafen → nothing; bob leaves during
an 11 s deafen → 'bob left'; long deafen with no change → nothing.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Elbow-on-the-keyboard guard: a PTT hold longer than 5 min (fixed, not a
setting) is released exactly like a keyup — pttActive off, mic muted — with a
toast 'Push to talk released after 5 minutes — press the key again to keep
talking.' A fresh press re-engages normally. One watchdog shared by the DOM
path and the desktop-global hotkey path; keydown auto-repeat no longer
restarts anything; non-PTT mode, deafen and the mic button are untouched.
Helper unit-tested with fake timers; verified headless with Playwright's clock:
held → still open at +4 min → released + toast at +5 min → re-press works.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Room Settings → General → Privacy: 'Your messages in this room' with a
Delete all… flow. The confirm dialog first counts your events with a
server-side sender-filtered /messages walk (live count), then asks to confirm
with the number — typing the room name above 50 — and offers 'Leave the room
afterwards'. Files are called out as not purged by a redaction.
The job runs outside React (closing settings is fine): sequential redactEvent
with 429 back-off, 404/already-redacted skipped, progress on the tile with
Cancel, pending ids persisted per room so a reload shows Resume/Discard, a
toast when done. State events are never touched; reactions, edits and thread
replies you sent are included; encrypted rooms work the same (nothing is
decrypted). Own events need no power level, so it is purely self-service.
Unit tests cover candidate filtering, the server filter, pagination, 429/404
handling and cancel. Verified headless: 62 of bob's events (60 messages, a
reaction, a thread reply) redacted in ~34 s while alice's 10 stayed; cancel at
17/40 → reload → Resume → 'Deleted 40 messages.'
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Message notifications shown through the service worker now carry a text-input
'Reply' action (Chrome desktop/Android). On notificationclick with
action==='reply' the SW sends the typed text itself — it already holds the
newest session's access token for authenticated media — as m.room.message
(threaded when the notification was for a thread), so it works with the tab in
the background or closed; a failed send shows a 'Reply not sent' notification
that opens the room. Not offered for encrypted rooms (the SW cannot encrypt).
The sender lives in swReply.ts so it is unit-tested; verified headless that
the SW notification carries actions + {roomId, threadId}.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
'Unable to decrypt message' now carries one sentence per matrix-js-sdk
DecryptionFailureCode (describeDecryptionFailure, unit-tested against every
code so no raw code can leak into the copy) and, where something fixes it,
one button: no key backup → 'Set up key backup'; backup exists but this
session can't open it / key withheld for an unverified session → 'Unlock key
backup' / 'Verify this session' (both open Settings → Devices via a new
settingsRequestAtom that SettingsTab consumes); backup working or unknown
session (rust-crypto re-requests keys itself) → 'Retry', which re-runs
decryptEventIfNeeded. Sender-side problems are plain text. The raw code sits
in the placeholder's tooltip for support.
Verified headless on a fresh session in the encrypted seed room: each event
shows 'Sent before you signed in here, and no key backup exists…' with
tooltip HISTORICAL_MESSAGE_NO_KEY_BACKUP; the button opens Settings → Devices.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Incident 2026-09-17: a wrong Windows clock broke calls and media keys while
the server answered 200 to everything, with no hint in the UI.
Measurement needs no extra requests and no CORS-exposed headers: every live
event carries origin_server_ts and unsigned.age (our server's now − ts at
response time), so localTimestamp − origin_server_ts is the skew. Only
RoomEvent.Timeline live events count (cache replays have stale age and are
already flagged liveEvent=false by the SDK); the initial network sync
qualifies, so a wrong clock is flagged within seconds of startup. Median of
the last 5 samples, ≥3 needed; warn at |skew| > 30 s, clear below 15 s.
UI: a banner in the sync-status slot — "Your computer's clock is 14 minutes
ahead of the server. Encrypted messages and voice calls will fail until it is
fixed." with a per-OS How-to-fix hint and Dismiss for 24 h — plus the same
line in the call status bar while in a call. Never auto-corrects anything.
Unit-tested (median, hysteresis, stale-age rejection, wording); verified
headless with Playwright's clock skewed +14 min and −3 h (banner, hint,
in-call line, dismiss) and in sync (nothing shown).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
The upload card printed the SDK's toString — 'MatrixError: [413] nope
(http://<hs>/_matrix/media/v3/upload?filename=…)'. describeUploadError() maps
the common cases to one sentence: 413/M_TOO_LARGE → 'This file is larger than
the server allows (limit N)' using m.upload.size when known, 429 → 'Slow down —
try again in a moment.', 401/403 → 'The server refused this upload: <server
text>', 5xx/transport after the retry loop → 'Couldn't reach the server. Check
your connection and retry.', other 4xx → the server's own sentence, URL
stripped. Both card renderers use it; the raw error is still console.warn-ed
by uploadContent for debugging. Unit-tested; verified headless with routed
413/403/503 responses.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
rateLimitedActions now collects non-429 failures (and a 429 that exhausted its
retries) and returns them instead of swallowing them; existing callers ignore
the return. /kick and /ban turn the list into a CommandError whose message
names who and why, using the server's own sentence (MatrixError.data.error),
never the URL-bearing toString(); RoomInput's toast shows it verbatim.
Verified headless as a non-moderator: '/kick @alice' → "Could not kick
@alice:localhost: You cannot kick user @alice:localhost."; '/ban @nobody
@alice' → "Could not ban @nobody:localhost, @alice:localhost: You don't have
permission to ban".
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Verified first on the local calls stack: a participant who left and rejoined
~3 s later played "leave" then "join" every time. Cues now go through
createCallSoundDebouncer (per USER, not per device):
- a leave cue waits 5 s; if the same user is back before it fires, the leave is
cancelled and no join cue is played either;
- a join cue is suppressed for a user who left < 60 s ago;
- same-kind cues within 250 ms collapse, so a batch of leaves still sounds once;
- a second device of someone already present is not a new arrival, and a
device switch is quiet.
Only the sound is debounced; membership UI is unchanged. Sound style, volume
and PTT interplay untouched (the style is read at play time). Timers are
injected — unit-tested with a manual clock. Re-ran the flap scenario headless:
join → flap → (silence) → real leave → one "leave" 5 s later → rejoin within
60 s → silence.
Also enables msc4133 (custom profile fields → in-call avatar decorations) on
the dev Synapse.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
matrix-js-sdk rejects an upload whose XHR ends with status 0 (offline,
connection reset, DNS) with DOMException('AbortError') to mimic fetch,
the same name mx.cancelUpload() produces. isRetryableUploadError bailed
on any AbortError, so the one failure class the retry loop was built
for was never retried. Decide by our own cancel AbortSignal instead.
Verified with Playwright routing the upload endpoint: 502 → network drop
→ ok now completes in 3 attempts (1 s, 2 s back-off) and the image sends;
413 still fails fast after 1 attempt; persistent 503 gives up after 4.
Unit tests in utils/uploadRetry.test.ts.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
RoomTimeline renders a numeric index window into the live timeline's
event arrays; SDK back-pagination prepends, so any side panel calling
paginateEventTimeline(room.getLiveTimeline()) shifted the visible
messages into the past on the next render and broke at-bottom tracking.
New utils/detachedTimeline.ts builds a timeline set that mirrors the
already-loaded history and paginates independently: a room-registered
filtered set (server-side contains_url / types filter) when the filter
is usable, else a private EventTimelineSet seeded from the live timeline.
useRoomMediaTimeline wraps it for the gallery (live events + redactions
handled); RoomActivityLog uses a type filter (safe in encrypted rooms);
ExportRoomHistory pages a private set so a full export no longer parks
thousands of events in the live timeline.
Verified with Playwright against a local Synapse in a 400-message plain
room and a 200-message encrypted room: timeline stays at the bottom
through gallery pages, activity load-more and a full export; live
messages keep auto-scrolling; all media found in both rooms.
Also adds scripts/dev-homeserver.sh + scripts/dev-seed.py (local
throwaway Synapse for driving the real UI) and documents them.
Closes#163
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
LOTUS_TODO.md → LOTUS_REFERENCE.md (design laws, decided deferrals,
server-blocked features, operational reference only). Every open task was
filed: cinny #195–#210, cinny-desktop #15–#18, matrix #8–#10.
LOTUS_TESTING.md keeps the automated-coverage map, the Playwright notes and
the deploy tip; every manual checklist is now a `qa` issue under the
'Manual QA backlog' (cinny #170–#194, #198) and 'Desktop QA backlog'
(cinny-desktop #11–#14, #18) milestones.
Repointed the README, LOTUS_FEATURES, CI and source comments that
referenced LOTUS_TODO.md.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
PTT and deafen were DOM key handlers and only fired while Lotus (or the EC
iframe) had focus — alt-tab into a fullscreen game and the voice controls
stopped working.
- useCallHotkeys: while a call is joined and the new device-local
`globalCallHotkeys` setting is on, register {ptt, deafen} bindings with
the desktop (`set_global_hotkeys`, cleared on leave) and act on its
`lotus-global-hotkey` press/release events. Events are ignored while
`document.hasFocus()` so the DOM handlers keep owning the in-focus case
(editable-field and interactive-element checks, no double toggles). PTT
press engages the mic exactly like the DOM path (pttActive set before
unmute), release restores; the existing blur/focus release covers a hold
that spans a focus change. Same modifier rules as the DOM path
(`shouldActOnGlobalHotkey`, tested).
- Settings → Calls: "Hotkeys Work Outside the Window" toggle, Tauri only.
- settingsSync: `globalCallHotkeys` is device-local (never synced).
- LOTUS_FEATURES: desktop section entry.
Native side lands in cinny-desktop (src-tauri/src/native/hotkeys.rs).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
The client never handled beforeinstallprompt and showed no install hint, so
phone users only got the PWA if they knew to dig through Share → Add to
Home Screen (iOS never prompts; Chromium's mini-infobar is easy to miss).
- utils/pwaInstall.ts (pure, 4 tests): show from the second visit, never in
Tauri or an installed PWA (display-mode standalone / navigator.standalone),
30-day snooze after a dismissal; kind = real prompt when the browser
handed us a deferred beforeinstallprompt, Share-sheet instructions on iOS
Safari, nothing elsewhere (Firefox desktop has no install path).
- hooks/usePwaInstallPrompt.ts: captures beforeinstallprompt/appinstalled,
counts one visit per browser session, waits 6s for the prompt event before
deciding, then enqueues a sticky toast; tap → prompt(), X → snooze.
- ToastNotif gains onDismiss (fired by the X button only) so the snooze is
recorded however the toast is closed. Mounted from ClientNonUIFeatures for
signed-in users only.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Every Lotus setting was localStorage-only, so a user on web + desktop + phone
configured theme, composer toolbar, quiet hours, call keys… three times.
- utils/settingsSync.ts (pure, 7 tests): DEVICE_LOCAL_KEYS denylist (zoom,
media auto-load, animation pause, glassmorphism, denoise tier/model,
bitrates, volumes, notification permission, developer tools, PTT mode,
camera-on-join, drawer state, and the sync toggle itself), pickSyncable,
mergeRemoteSettings (unknown keys, device-local keys and wrong-shaped
values are skipped), buildSyncedContent, shouldApplyRemote (LWW on
updatedAt; equal stamp = our own echo).
- hooks/useSettingsSync.ts: on start applies a newer remote snapshot or
pushes local if it differs; debounced push on any settingsAtom write,
skipped when the syncable subset equals the last pushed/applied snapshot
so a remote apply never echoes back; AccountData listener for live
updates; stamps forced monotonic per device; per-account lastSyncedAt
marker so another user on the same device can't inherit it; failed pushes
roll the marker back so the next change retries. Remote values are re-read
through getSettings() so enum coercion applies.
- Settings → General → Sync: toggle (device-local), "Push now", "Clear
synced copy". AccountDataEvent.LotusSettings registered.
- ClientNonUIFeatures: the #103 tracking-param subscriber moves out of
PageZoomFeature into its own TrackingParamsFeature next to
SettingsSyncFeature.
- Docs: LOTUS_FEATURES entries for #103/#104; LOTUS_TODO links the new
Features 2026-Q4 milestone and #108.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Shared links routinely carry ad/analytics identifiers (utm_*, fbclid, gclid,
YouTube si=, Amazon ref=/tag=, X s=/t=, TikTok _r/_t, …) that tie every
recipient's click back to the person who shared the link. New
src/app/utils/urlTracking.ts is a pure, local stripper: a global list +
utm_/pk_/matomo_ prefixes, plus host-scoped rules so e.g. `si` is only
removed on youtube/spotify. matrix.to and non-http(s) schemes are never
rewritten; unparseable input is returned unchanged; Amazon's `th`/`psc`
variant selectors are deliberately kept. 13 unit tests.
Wired at three points, all behind a new Settings → Privacy toggle
(`stripTrackingParams`, default on):
- paste: plain-text pastes are cleaned and re-inserted through Slate's own
insertData so multi-line pastes still split into paragraphs;
- send: RoomInput submit + schedule paths and MessageEditor saves clean both
`body` and `formatted_body` (the HTML variant unescapes `&` around each
URL and re-escapes it so the markup is untouched);
- render: linkify `formatHref`/`format` and explicit `<a href>` in
formatted_body are cleaned, so links sent from other clients are safe to
click too. LINKIFY_OPTS is spread into memoised per-timeline objects, so
the toggle is a module flag kept current by ClientNonUIFeatures.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Every io.lotus.call_state push allocated a new Set, re-rendering the
app-wide call bar for the whole call. nextSpeakerSet() returns the
previous reference when membership is unchanged (pure helpers in
utils/speakerSet.ts, unit-tested), and the DOM MutationObserver fallback
in useCallSpeakers/useRemoteAllMuted is attached only while the fork's
participant list is unavailable.
Fixes#32
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
deleteSearchCacheDatabase() resolved after a 3 s "blocked" timeout while
another tab still held the DB, so decrypted rows could survive logout.
It now broadcasts lotus-logout first; every tab closes its handle and
refuses to reopen, then the delete proceeds. A boot with no session
re-runs the wipe once in case a race was still lost. Unit-tested.
Fixes#45
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Decrypted clip blob: URLs lived in an unbounded module Map for the page
lifetime and survived logout. Add clearSoundboardClipCache() (called from
both logout paths next to clearPlaintextCaches) and a 64-entry LRU that
revokes on evict. Unit-tested.
Fixes#57
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
Nothing ever removed an indexed row: redacted messages stayed searchable
with full plaintext and rendered as normal results. Now: a client-level
RoomEvent.Redaction listener deletes the row, leave/ban clears the room
(clearRoom finally has a caller), m.replace edits upsert the original
row instead of indexing the "* fallback" separately, and cached rows
whose local event is redacted render through the existing
redacted_because placeholder. Unit-tested.
Fixes#14
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
- useKeyBind refuses Tab/Enter/arrows/Home/End/Page*/Escape and bare
modifier codes, and refuses a code equal to the other call key, with an
inline message (isBindableCallKey, unit-tested).
- PTT and deafen handlers ignore events with Ctrl/Alt/Meta held (deafen
also Shift), so Cmd+M / Ctrl+M no longer toggle deafen.
- PTT only preventDefault()s when the target is not an interactive
control, so Space still activates focused buttons during a call.
Fixes#23
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
- Image compression in an encrypted room re-encoded the *plaintext* original,
uploaded it unencrypted, and reused the original's encInfo, so the media sat
on the server in the clear AND the attachment was undecryptable. The
compressed bytes are now run through encryptFile and the synthetic upload
item carries the new encInfo (buildCompressedUploadItem, unit-tested; it can
never inherit the stale encInfo).
- Scheduled messages (MSC4140) are PUT as raw m.room.message, bypassing the
SDK encryption pipeline. The Schedule button is now hidden in encrypted
rooms, handleScheduleClick no-ops there, and scheduleMessage() itself
refuses with a clear error so no caller can regress this. README notes the
limitation.
- The GIF picker uploaded the Giphy blob unencrypted into E2EE rooms; it now
mirrors the voice/attachment path (encryptFile -> upload ciphertext ->
content.file).
Fixes#6Fixes#7Fixes#11
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPmy3tPq869XDW4njjVaKA
A room could show a permanent unread that survives every cold start even
though the server considers it fully read (notification_count 0, unthreaded
read receipt at the tail). matrix-js-sdk's fixNotificationCountOnDecryption
only ever INCREMENTS an encrypted room's Total, and addReceipt's auto-clear
fires only when the tail event is the user's own — so a count inflated in an
earlier state (before a receipt covered the tail, e.g. by a since-corrupted
undecryptable event) is never decremented and keeps a genuinely-read room lit.
This is aggravated by mixing threaded-receipt clients (Element X) with
unthreaded ones (Lotus/Cinny), which split the read marker.
Add readReceiptCoversTail(room, userId): walking the live timeline newest→
oldest, if we reach the user's read-receipt event without crossing any
notification-worthy event, the room is genuinely read and a lingering Total is
suppressed to {0,0} in getUnreadInfo / getUnreadInfos. Safe by construction —
a real unread sits AFTER the receipt and stops the walk at isNotificationEvent
— and guarded against unread threads (markAsRead clears threads unconditionally)
and off-window receipts (can't confirm → don't suppress). Self-correcting: a
new message becomes the tail and the walk stops suppressing.
Also recognize polls (m.poll.start / msc3381) as notification events so a
poll-only unread is never walked past (closes a pre-existing gap in the
tail scans), and factor the unread-thread guard into roomHasUnreadThread.
Reviewed by 3 agents (false-suppression safety, unread-system regression,
SDK behavior): no real unread is hidden for any standard content, no
regression to the atom/PUT-DELETE paths, and the fix produces {0,0} for the
target scenario and stays resolved.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A completed in-room device-verification request is a plain m.room.message
(msgtype m.key.verification.request) that matches the default DM push rule
with no recency gate, so the server/SDK notification count stays > 0 and the
DM re-lights as unread on every fresh sync until the room is opened twice.
Two-part fix:
- Display suppression: getUnreadInfo/getUnreadInfos return {0,0} for a room
whose ENTIRE unread span (tail -> read receipt) is verification-flow events,
via new pure helpers isVerificationFlowEvent + unreadIsOnlyVerification.
Conservative: never suppresses when the read marker is off-window, the tail
is still encrypted, or a highlight is present.
- Durable auto-read: useAutoMarkVerificationRead sends a read receipt covering
the request (the only SDK-durable lever), once per room per session, gated on
the same verification-only predicate so it can never ack a real message.
unreadIsOnlyVerification also rejects any room with an unread thread, because
markAsRead clears every thread unconditionally — otherwise a verification-only
main timeline with a genuine unread thread reply would be hidden/auto-acked.
Reviewed by 5 agents; the thread-scope guard closes the one bug they found.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
On the Windows/Tauri build, clicking a message notification opened the app but
didn't navigate to the message. showOsNotification preferred the service worker
(registration.showNotification) and returned early; WebView2 has a service
worker, so the SW-owned toast always won and its click (SW notificationclick →
client.focus + postMessage → navigate) focused the app but the navigate didn't
complete in WebView2.
The desktop build injects a window.Notification shim that routes tagged message
toasts to the native rich WinRT toast, whose click emits lotus-notification-
activate with the path → useTauriToastActions → navigate. But the SW path
shadowed `new Notification()`, so that shim (and show_rich_toast) never ran on
desktop. Skipping the SW path under Tauri lets the shim take over and navigate.
Web browsers are unchanged (isDesktopApp() is false → SW path as before). Two
review agents verified the diagnosis + no web regression across both repos.
DESKTOP-QA REQUIRED — this activates a previously-dead code path. Known desktop
follow-ups it exposes (documented in LOTUS_TODO, both in cinny-desktop Rust):
- tag-coalescing is lost (rapid same-room messages stack toasts instead of
collapsing) — show_rich_toast doesn't dedupe by room.
- thread/invite quick-reply misroutes: the reply target is the coalescing tag
(roomId:threadId / 'lotus-invites'), not a real room id → sendMessage fails.
Navigation itself (body click) is correct for all cases.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The on-device search index grew unbounded over a long session. putRows now
prunes each touched room to MAX_ROWS_PER_ROOM (5000) — deleting the oldest rows
by [roomId, ts] via a self-chaining IDB cursor within the same write tx (never
awaits a non-IDB promise mid-tx, so the transaction can't auto-commit and
truncate the prune). Exposed a pure, unit-tested evictCount() for the decision;
the cursor path itself is browser-only (node --test has no IndexedDB).
Deliberate tradeoff (documented in code): the coverage window keeps claiming the
evicted tail so the search doesn't re-fetch → re-evict it forever. Net effect —
in a room past 5000 cached rows, an evicted old message is silently unsearchable
rather than churning. Clear cached index / logout still wipe everything.
Two review agents verified the IndexedDB-spec correctness (cursor delete+continue
semantics, put-then-count ordering, roomRange bracketing with no prefix bleed,
tx liveness, abort→cache-miss) since CI can't. Gate-green (tsc, eslint, prettier,
925 tests, build).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Toast queue: a burst of notifications appended unboundedly and could cover the
viewport. Cap at 5 in the atom writer, dropping the OLDEST non-sticky toast
(sticky = action toasts requiring a click, never dropped). The drop scan
excludes the just-appended newest (`length - 1` bound) so a fresh toast is
never the one eaten when the cap is full of stickies — it stretches instead.
Container gains a maxHeight + overflowY safety net and scrolls the newest
(bottom) toast into view if the stack ever overflows. +4 unit tests incl. the
cap-full-of-stickies boundary.
- "Unread First" room sort left the entire read tail (all counts tie at 0) in
arbitrary Map order. factoryRoomIdByUnread now breaks ties by recent activity.
Relocated from Home.tsx (module-private) to utils/sort.ts (exported, pure) and
unit-tested (equal-count and read-tail cases fall back to activity).
Bug-hunt findings from LOTUS_TODO. Three review passes: the second caught that
the cap could silently drop the newest notification when full of stickies (real
bug, untested boundary) — fixed and covered; a third traced the corrected loop.
Gate-green (tsc, eslint, prettier, 920 tests, build).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
GIF previews rendered but never played: Synapse's /thumbnail endpoint
flattens animated GIFs to a still first frame. GifCard and the generic OG
card now request the original via /download (no width/height) for GIFs, so
they animate. Guarded with shouldServeGifOriginal(): a matrix:image:size cap
(10 MB) keeps a huge self-hosted GIF on the frozen thumbnail, and the generic
card's eager <img> gains loading="lazy" (it was the one preview image missing
it) so originals stay off the wire until near the viewport.
Also adds Mixcloud + Deezer inline media embeds (iframe widgets via
parseMediaEmbed/MediaEmbedCard, matching the existing click-to-play pattern),
and fixes Deezer podcast links: they live at /show/<id>, not /podcast/<id>
(the latter 404s on Deezer's own oEmbed) — verified against the live API.
Reviewed by two agents; both findings (Deezer /show, GIF eager-load) fixed
and covered by tests. Desktop Tauri frame-src CSP updated separately.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Recognize store.steampowered.com content URLs and render each richly, within
the existing privacy-first facade. 2-agent reviewed (both SHIP).
- getSteamTarget / steamWidgetEmbedUrl (videoEmbed.ts, +tests): classify
/app/{id}, /news/app/{id}/view/{gid}, /(bundle|sub|dlc)/{id}; non-content
pages (home/search/wishlist) and other hosts fall through to the generic card.
- SteamCard now dispatches:
- app → OG capsule header + click-to-play facade → Steam's OFFICIAL store
widget iframe (store.steampowered.com/widget/{id}): live region-aware price,
discount %, Buy on Steam. Nothing loads from Steam until "Show price &
store" is pressed; gated by the inlineMediaEmbeds setting. App pages use the
wide card so the ~646px widget has room.
- news → rich announcement card (banner + headline + body preview + link) —
your example URL previously fell through to the plain generic card.
- bundle/sub/dlc → the OG store card.
Grounded in our CSP: the widget works via frame-src https: (no infra change),
images route through the homeserver (img-src excludes Steam), and there is NO
client-side Steam API call (connect-src + Steam CORS both block it) — which is
also the honest ceiling: no review scores/genres client-side, price/buy come
from the official widget.
Runtime QA still needed: the live widget iframe rendering (height/fit) can't be
verified headlessly.
Gates: tsc 0, eslint 0, prettier clean, 912 tests, build ok.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Bugs found by a 3-agent audit of the inline-embed system (core posture —
sandbox, postMessage origin+source, XSS, noreferrer, oEmbed — verified sound);
fixes reviewed by 2 agents on the staged diff (both SHIP).
Parsing (videoEmbed.ts, + tests):
- Twitch/Kick/SoundCloud/Streamable reserved-path exclusion — their own utility
pages (twitch.tv/directory, kick.com/browse, soundcloud.com/discover/…,
streamable.com/login, bare /videos) no longer render as broken player embeds.
- SoundCloud: `/<artist>/<tab>` profile-tab listings excluded; `/<artist>/sets/<slug>`
real sets still detected.
- Vimeo: unlisted-hash capture constrained to lowercase-hex, so a normal video's
trailing segment (/likes, /settings, a slug) isn't captured as a bogus `h=`
param that Vimeo then rejects.
Rendering (UrlPreviewCard.tsx, RenderMessageContent.tsx):
- Spotify/Steam/Discord/IMDb route og:image through mxcUrlToHttp like every other
card — a raw og:image is an mxc:// URI (broken <img> on standard Synapse) or an
off-homeserver request that defeats the click-to-play facade.
- `wide` card class now follows the RESOLVED embed (incl. the og:url short-link
fallback), so an og:url-resolved player gets the wide layout, not a cramped one.
- Twitter host detection (isTwitter/isTwitterTweet) aligned with getTweetId —
mobile.twitter.com and legacy /statuses/ now route to the Twitter card/embed.
- De-dupe preview URLs so a message repeating a link doesn't render sibling
cards with identical React keys.
Gates: tsc 0, eslint 0, prettier clean, 910 tests, build ok.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Test-coverage batch 2 (2-agent reviewed, both SHIP; isolation verified — Node
runs each test file in its own process, so the console patch can't leak).
- cryptoDiagLog.test.ts: the E2EE KE-cluster diagnostics tool — KE-signature
capture vs ignore, most-specific-first match order, KE-3/KE-4 rows, Error /
object / circular-arg serialization (String() fallback never throws), the
200-entry ring-buffer eviction, getCryptoDiagEntries copy semantics,
install idempotency, and buildCryptoDiagReport's client metadata + LOCKED
PII-safe key set (no field can silently leak) + no-client/throwing-getter
fallbacks. Silences console pass-through so the ring-buffer test stays quiet.
- closedLobbyCategories.test.ts: mirrors closedNavCategories — id join,
hydrate, PUT/DELETE, idempotent PUT, no-op DELETE, array persistence,
per-user key namespacing.
Also: mark the EC in-call mobile UI audit done in LOTUS_TODO (stale entry;
shipped as element-call:lotus e36aef8a).
Gates: tsc 0, eslint 0, prettier clean, 911 tests, build ok.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Test-coverage expansion (2-agent reviewed, both SHIP). The named candidates
(roomToUnread, markedUnread, serverAcl, plaintextCaches, recent*) were already
tested, so this targets genuinely-untested pure logic.
- dom.test.ts: getThumbnailDimensions (scaling math incl. just-over-cap
boundaries), tryDecodeURIComponent, syntaxErrorPosition, and the three
scroll-view geometry helpers (via duck-typed element mocks — no jsdom).
- emoji.test.ts: getHexcodeForEmoji (astral codepoints, 4-digit zero-pad,
FE0F/FE0E/200D stripping on and off, keycap sequences, degenerate inputs)
and the pre-load `undefined` contract for getShortcode(s)For.
Fix (found while writing the tests): syntaxErrorPosition required whitespace
AFTER the digits (`/position\s(\d+)\s/`), but real V8/Node JSON.parse errors
put the number at end-of-string ("... at position N"), so it returned
undefined for every real error and the three dev-tools JSON editors silently
pointed their cursor at position 0. Dropped the trailing `\s`; tests now assert
extraction at end-of-string.
Gates: tsc 0, eslint 0, prettier clean, 891 tests, build ok.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Verify-then-fix batch of minor bugs; each staged diff reviewed by 2 agents
(both SHIP). Two listed items (N6 receipt-avatar refresh, H10 room-name
length reject) were already handled and left unchanged.
Threads:
- T5: a just-sent reply no longer under-notifies — `participated` also checks
the local thread timeline for our own events, since the server-bundle
`hasCurrentUserParticipated` lags.
- T6: a room set to "Mentions & Keywords only" no longer over-notifies Default
thread replies — new `roomMentionsOnly` gate (behavior-identical when false;
+4 unit tests).
- T7: thread-mode account-data writes are serialized with content carried
forward (setAccountData is a bare PUT whose result lags the /sync echo, so
plain serialization wouldn't stop the lost update); carry only on success.
Calls / audio:
- C-L2: a real incoming ring cancels a lingering Settings ringtone preview.
- C-L3: the ringtone AudioContext is primed on the first page gesture (via the
always-mounted CallEmbedProvider) so the first ring after a cold load isn't
silent.
- C-L5: useCallSpeakers depends on a stable boolean, so the tile MutationObserver
+ io.lotus.call_state subscription aren't rebuilt on every membership change.
Crypto:
- F5: the OIDC refresher forwards the freshly-refreshed token expiry
(passed on the tokens object at runtime) as expiresInMs, so the persisted
expiresAt no longer goes stale across reloads.
Gates: tsc 0, eslint 0, prettier clean, 860/860 tests, build ok.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
COR-3 (CallEmbedProvider): the incoming-call lifetime guard distrusted a
caller's sender_ts only when it was >20s AHEAD of the server ts. A caller clock
that ran SLOW left sender_ts in the past, so the ring auto-dismissed/never
showed for a fresh invite. Trust sender_ts only within ±20s of the server ts,
else fall back to it (also fixes a NaN path when sender_ts is missing).
COR-6 (CallControl): forceState rebuilt CallControlState with 5 args, silently
defaulting screenshareAudioMuted to false; pass this.screenshareAudioMuted.
COR-5 (uploadContent + useBindUploadAtom): cancelling during the retry back-off
was a no-op (mx.cancelUpload only aborts an in-flight request), so the upload
resurrected on the next attempt. Thread an AbortSignal: the back-off sleep
resolves early on abort and the loop stops with an abort error; the hook aborts
a per-upload AbortController on cancel (alongside mx.cancelUpload for the
in-flight case).
All verified by two review passes (no double-settle / no resurrection); includes
their suggested abort-listener cleanup on normal sleep resolution.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
SEC-3: add `noopener,noreferrer` to the 5 `window.open(_blank)` sites that
don't use the returned handle (UserChips, OidcManageAccount, OtherDevices x2,
Verification), closing reverse tab-nabbing. SSOStage is intentionally excluded —
it needs the window handle + intact opener for its origin-checked SSO
postMessage handshake.
SEC-4: guard the `/acl` slash command against bricking the room.
- Extract the ACL glob helpers (isValidServerPattern/globToRegExp/matchesAnyGlob)
from RoomServerACL into a shared utils/serverAcl.ts (+ unit test) so the
command and the settings editor validate identically.
- Default a MISSING allow list to `*` only when the room has NO existing ACL
(a first `/acl -d x` otherwise sent `allow: []`, which bricks the room); an
existing ACL's absent/empty allow is preserved, not silently widened.
- Reject invalid globs; fail CLOSED on the universally-catastrophic cases
(empty allow, or a `*` deny) even when the local domain is unknown; and reject
any change that would ban this homeserver (self-lockout).
Guard hardened per two review passes (fail-closed on unknown domain; no silent
federation widening).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Follow-up hardening from two review passes on the on-device translation
feature:
- Privacy (HIGH): the translation cache is decrypted message plaintext,
but logout did not clear it (unlike the search index), leaving up to
300 cleartext bodies in localStorage on shared devices. Add
clearTranslationCache() and call it from both logout paths
(logoutClient and the server-forced SessionLoggedOut handler).
- Edited messages (MEDIUM): the cache key was eventId:target with no
content dependence, so an edit reused the pre-edit translation. Fold a
content fingerprint into the key, and re-arm the auto-translate
one-shot when the text changes.
- Settings (LOW): coerce a persisted translateTargetLang to a supported
curated code so the hook never targets a language the engine can't
produce (previously only the UI clamped it).
- Chinese (LOW): restore canonical BCP-47 case (zh-Hant / zh-Hans) at
the Translator API boundary, since normalizeLang lowercases the script
subtag for internal keys.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add per-message translation that runs entirely on-device via the
Chromium built-in Translator + LanguageDetector APIs. Message text
never leaves the machine and never touches a cloud service, preserving
the E2EE guarantee. When the on-device engine is unavailable
(non-Chromium / mobile) the feature simply hides itself; there is no
network fallback.
- Engine abstraction (utils/translation): TranslationEngine interface
plus a chromeTranslationEngine implementation (feature-detected,
caches translator/detector instances, download-progress monitor).
Pure lang-code helpers (normalize/sameLanguage/curated targets) with
unit tests.
- Settings: translateTargetLang (default English) + autoTranslate
(opt-in), with a Messages settings tile — a target-language select
and an auto-translate switch, disabled with a note where unsupported.
- useMessageTranslation hook + shared per-event toggle atom-family and a
persisted LRU cache so scrollback never re-translates.
- UI: a Translate / Show Original message-menu action, an inline
"Translated from <lang> - Show original" chip, and a body swap in
m.text/m.emote/m.notice that renders the translated text through the
plain-text path (linkify + emoji) inside a dir=auto span for RTL.
- Auto-translate flips foreign messages whose model is already
downloaded; first-time downloads keep the manual chip (user gesture).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
check:prettier was not part of my gate routine, so formatting drift accumulated
across the session's touched files (and a few older ones). Run prettier --write
to bring the repo back to 'All matched files use Prettier code style!'.
Formatting only — no logic changes. tsc/tests/build all green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The poll card showed vote counts but never who voted, even though
computePollState already parses a sender for every response. Surface it:
- tallyResponses now also returns voters: Map<answerId, senderId[]>, built in the
same latest-response-per-sender loop as the counts, so voters can never disagree
with the numbers (voters.get(id).length === counts.get(id)). +5 unit tests.
- PollContent adds a "Show who voted" toggle, shown only when results are visible
(disclosed live, or undisclosed after end — so a secret ballot stays secret).
When on, each answer lists its voters' display names (getMemberName), rendered
as a sibling of the answer button so the radiogroup keyboard model is untouched.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Manual DND only existed via the desktop tray (manualDndAtom), so web/mobile
users had no way to pause notifications, and there was no snooze-for-a-duration
anywhere. Add a "Pause Notifications" control in Settings > Notifications:
- Presets: 30 min / 1 hour / 4 hours / Until 8 AM / Until I resume, plus Resume;
live "Paused until ..." status that flips back on when the snooze lapses.
- Persisted snooze instant (cinny_notification_snooze_until_v1) so it survives a
reload; 0 = off, SNOOZE_INDEFINITE = until resumed.
- Feeds the existing notification gate (ClientNonUIFeatures, both the message and
invite monitors) alongside Focus Assist / manual DND / Quiet Hours, suppressing
notify() and playSound().
- Pure helpers isSnoozeActive/nextTimeAtHour/SNOOZE_INDEFINITE in utils/snooze.ts
(+5 unit tests); persisted atom in state/notificationSnooze.ts.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Night Light was all-or-nothing. Add an optional schedule with From/To time
inputs so the warm overlay only shows during set hours and toggles itself on/off
automatically (the overlay re-checks every minute; no reload). Overnight windows
that wrap midnight (e.g. 21:00 -> 07:00) are handled.
Window logic is the pure, unit-tested isWithinTimeWindow/parseHHMM in
utils/timeWindow.ts. New settings: nightLightSchedule/Start/End (default
21:00-07:00).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
There was no way to see or cancel a reminder once set (removeReminder was only
called by the fire-and-forget monitor), and addReminder didn't dedupe, so a
message could silently accumulate duplicate reminders. The Remind Me dialog now
lists the reminders already set on that message (soonest first) each with a
cancel button.
- New shared, tested formatFriendlyDateTime(ts, now?) in utils/datetimeInput.ts
(Today/Tomorrow/date + time).
- Per-row cancel busy-guard; inline "Could not cancel" on failure.
Also applies two nits from the custom-time review: focus the date input when the
custom picker is revealed, and clear the error when editing date/time.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The Remind Me dialog only offered four fixed presets, so you couldn't set a
reminder for an arbitrary time. Add a "Custom time…" option that reveals date +
time pickers (validated >= 1 minute in the future) and sets the reminder at that
absolute timestamp.
Also extract the local date/time <input> helpers (toLocalDate, toLocalTime,
parseLocalDateTime, pickerInputStyle) into a shared, unit-tested
utils/datetimeInput.ts and reuse them in ScheduleMessageModal (deduped from an
inline copy) — identical output, now covered by tests.
Documents the previously-undocumented Message Reminders feature in
LOTUS_FEATURES.md.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Fire a pending scheduled message immediately via MSC4140 action:'send'
(the server dispatches the stored delayed event now, as a normal timeline
event) instead of having to cancel and retype.
- sendScheduledMessageNow(mx, delayId) mirrors cancel/restart with action:'send'
- handleSendNow reuses the per-row busy guard; prunes local state only once the
server confirms; a failed send shows an inline "Could not send now" error with
the message still sendable/editable/cancellable
- Send-now IconButton (Icons.Send) added before Edit/Cancel in each row
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
jaredandClaude Opus 5