useUserPresence registered 3 client listeners (Presence / CurrentlyActive /
LastPresenceTs) PER hook instance. On a large room that meant 100-250 global
listeners, every presence event fanning out across all of them, with add/remove
churn on every fast scroll.
Replace with a module-level PresenceStore singleton that registers exactly 3
listeners total (lazily, on first subscriber) and fans out to per-user
subscribers itself. The hook keeps the same public API (useState + a subscribe
effect); consumers are unchanged. Cache + subscriber sets stay bounded to
currently-mounted users; the mx-swap branch re-homes listeners on re-login.
Reviewed by two passes (SDK mutate-before-emit ordering and handler signatures
independently verified). Includes their recommended hardening: the unsubscribe
is made idempotent via a set-identity check so a double-invoke / re-subscribe
can't evict a newer subscriber.
Note: a User object that appears silently with no presence event no longer
re-seeds (deps are [mx, userId] not [mx, user]); the common presence-EDU case
is handled (and better than before). Reviewers rated this narrow case Low.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
When a single m.space.child was removed (unlinking child C from space P), the
roomToParents reducer fired the whole-room DELETE action, which wiped C's
entire parent set, stripped C as a parent from every other room, and orphaned
C's own descendants until a full resync. So removing C from space A also
dropped C's other parent B, and C's children lost C.
Add a targeted UNLINK {parent, child} action that removes only that one
parent->child edge and prunes the child entry only when its parent set
empties (matching the map's build-time invariant that zero-parent rooms have
no entry). Point the invalid-child branch of handleStateChange at it; DELETE
is unchanged for genuine room leave/delete. Unit-tested (keeps other parents,
prunes on last parent, does NOT orphan descendants, unknown pair no-op).
Verified correct + consumer-safe by two review passes.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
COR-2 (useCallEmbed): useCallJoined only reset `joined` when the embed became
undefined. Answering a 2nd call swaps the embed A->B directly (embed stays
truthy), so `joined` stayed true and call B rendered as already-joined,
skipping the loading/watchdog UI. Re-seed from `embed?.joined ?? false` on
every embed identity change.
COR-4 (ClientNonUIFeatures): the notify-dedupe used one Map<roomId,eventId>
slot shared by the main-timeline and per-thread paths, so a thread reply
overwrote the room's slot and a re-fired main message (decrypt/edit re-emit,
common in E2EE) then mismatched and double-notified. Key the slot by
`${roomId}|${threadId ?? 'main'}` so each path dedupes independently.
Both verified correct by two review passes (no missed-notification or
missed-join regressions).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Several localStorage caches held decrypted message content or user PII and
survived a normal logout, leaving residue on a shared device (the search
index was already wiped; these were not):
- cinny_scheduled_messages_v1 - decrypted IContent.body of pending sends
- cinny_recent_searches_v1 - search query text
- cinny_recent_forward_targets_v1 - recent forward contact/room graph
- cinny_recent_gifs_v1 / cinny_recent_stickers_v1 - media the user sent
- navToActivePath<userId> - per-space last-visited room paths
- (plus the translation cache added earlier)
Add a clear function per module and a single auditable clearPlaintextCaches()
aggregator, called from both logout paths (logoutClient + the server-forced
SessionLoggedOut handler) alongside the existing session/search-index wipes.
Unit-tested.
Deliberately NOT cleared (documented in the aggregator): unsent composer
drafts and the presence status message (preserved by product decision N98);
SDK sync/crypto store + io.lotus.* account data (reminders/bookmarks/notes),
already wiped by mx.clearStores(); low-sensitivity UI/metadata residue.
The forward-targets/gifs/stickers/nav-path additions and the accurate
"not covered" documentation address findings from two review passes.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add the feature to README (Messaging) and LOTUS_FEATURES (new
On-Device Message Translation section under Messaging Enhancements),
matching the existing style.
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>
The status-message emoji picker listed the guild's custom/image-pack emojis, but
clicking one did nothing — the field only wires onEmojiSelect (unicode), not
onCustomEmojiSelect, so custom picks were silently dropped (the room composer
works because it wires both). A custom emoji is an mxc image and a status is
plain-text presence status_msg, so it can't render there anyway.
Add an EmojiBoard hideCustomEmojis (unicode-only) mode that zeroes the image
packs (removing pack groups, sidebar icons, and search results) and filters
custom entries out of Recent, and enable it on the status field. Now every emoji
shown actually inserts. Additive prop, default off — no change to other pickers.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
On cinny-desktop (Tauri/WebView2) a clicked notification navigated the web
content but never raised the OS window (a service-worker/WebView2 client.focus()
only focuses the document). The service-worker notificationClick path now calls
a new focus_main_window Tauri command via invokeTauri (no-op outside Tauri) in
addition to navigating; the rich-toast path is focused natively on the Rust side.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Review noted the number input had both an htmlFor-associated visible label
("Voters can pick up to") and an aria-label, so the aria-label won and the
visible label was not announced. Remove the aria-label so the accessible name
matches the visible label.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The poll creator only offered single (max_selections 1) or multiple = pick ALL
options — no way to run a "pick your top 2" poll, even though the display side
already enforces an arbitrary max_selections ("Select up to N"). Add a "Voters
can pick up to N of M options" control shown for multiple-choice polls. Defaults
to the option count (preserving the old select-all behavior) until lowered;
clamped to [2, filled option count] on submit.
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>
ScheduleMessageModal had a local formatSendAt(Date) byte-equivalent to the
tested formatFriendlyDateTime (utils/datetimeInput). Reuse the shared, unit-
tested helper instead of a second copy — identical output. (Also prettier-clean.)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The emoji and GIF pickers both have a "Recent" row, but the sticker tab of the
shared EmojiBoard did not — you had to hunt through packs to re-send a sticker.
Add recent stickers, mirroring recentGifs:
- New state/recentStickers.ts (localStorage cinny_recent_stickers_v1, deduped by
url, capped 16) + pure addRecentSticker with 4 unit tests.
- EmojiBoard: a "Recent" group in stickerGroupItems and a RecentClock sidebar
icon in StickerSidebar, shown only when recents exist. Entries are rebuilt into
minimal PackImageReaders (StickerItem needs only url/shortcode/body) so they
render + re-send like pack stickers.
- Recorded on select in the shared delegated click handler, covering both the
grouped and search paths.
Blast radius is the sticker tab only (reactions/status use the emoji tab).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The thread composer already showed the /command autocomplete (RoomInput.tsx:954
was never gated), but the interpreter was disabled (:523), so /me, /shrug,
/invite, etc. sent literally in threads - a confusing inconsistency and the
other half of the threads "v1" limitation.
Remove the thread gate: content-transform commands (/me, /notice, /shrug,
/tableflip, /unflip) flow into the normal send path, which already routes to the
thread via threadRootId; the rest are room-level actions. No command sends a
mis-routed timeline message (verified against useCommands). Scheduling stays
disabled in threads for now.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
From review: the per-answer voter line sat inside the radiogroup with no
association, so a screen-reader user on the radio didn't hear who voted. Add
aria-describedby from each answer to its voter line (id poll-voters-<eventId>-
<answerId>) and prefix the line with "Voted by" for a clearer announcement.
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>
isInQuietHours was a hand-rolled, untested duplicate of the overnight-window
logic. Replace it with the shared, unit-tested isWithinTimeWindow (utils/
timeWindow.ts) - identical behavior for valid HH:MM inputs, more robust on
malformed ones (returns false rather than doing NaN math), and now covered by
timeWindow.test.ts. One implementation instead of two.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The thread composer reused RoomInput's hardcoded editableName="RoomInput", so
the main timeline's global up-arrow "edit last message" handler fired while
focused in a thread composer and targeted the MAIN timeline's last message
(wrong), and there was no up-arrow edit for the thread itself.
- Make editableName a RoomInput prop (default "RoomInput"); the thread composer
passes "ThreadInput", so the two up-arrow handlers never cross-fire.
- Add an up-arrow-edit handler to ThreadTimeline (parity with RoomTimeline):
empty thread composer + Up -> edit the latest editable reply in that thread,
using thread.liveTimeline + canEditEvent + setEditId.
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>
The timeline image viewer only zoomed via the -/+ buttons and the % toggle. Add
scroll-to-zoom (wheel up = in, down = out) for parity with the media-gallery
lightbox.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The timeline image viewer's transform applied translate inside scale(), so
dragging a zoomed image moved it by zoom x the cursor distance (panning outran
the pointer). Divide the pan offset by zoom so it tracks 1:1 - matching the
media-gallery lightbox fix.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Jump to Time uses timestampToEvent (MSC3030). On a homeserver that doesn't
support it, the dialog showed the raw "M_UNRECOGNIZED: Unrecognized request"
error. Show a clear explanation instead when errcode is M_UNRECOGNIZED.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Review noted media without a caption has a filename body, so Copy Text showed
and copied the filename. Gate to m.text/m.emote/m.notice so it only appears for
actual text messages (matching the intended behavior).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The message menu had Copy Link (permalink) but no way to copy the message text
itself. Add a Copy Text item that copies the plain-text body with the reply
fallback stripped (trimReplyFromBody). It renders nothing when there is no
usable text body (e.g. media without a caption), so the caller can list it
unconditionally next to Copy Link. Uses Icons.Text.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Constrain the rendered location description width and break long tokens so an
oversized description from another client can't overflow the message bubble.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The m.location renderer only read the legacy top-level geo_uri, so a location
from a client that sends only the MSC3488 shape (uri under
org.matrix.msc3488.location / m.location) showed as broken. Fall back to that
uri, and display an MSC3488 description above the coordinates when present.
Closes the consume-side gap noted in review of the send-side MSC3488 change.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
From review: equal start/end silently disables the tint (zero-length window).
Show an inline hint so the user isn't left wondering why nothing happens.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Location sharing sent only a legacy geo_uri with a "Location: geo:..." body, so
other clients often rendered it as plain text. Include the MSC3488 blocks
(org.matrix.msc3488.location/asset/ts + m.ts) alongside geo_uri and a readable
body, so Element and others render a proper location pin. Local rendering is
unchanged (still reads geo_uri).
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>
From review: sanitizeFilename now denylists only path-hostile + control chars
(preserving Unicode room names instead of collapsing CJK/emoji names to "room"),
converts whitespace to dashes, trims separator runs, and caps length. Also
surface an inline error when canvas.toBlob returns null so a failed PNG encode
isn't a silent no-op.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The invite QR was display-only. Add a "Download QR" button that exports it as a
PNG via an offscreen high-resolution (1024px, spec 4-module quiet zone)
QRCodeCanvas + canvas.toBlob, saved through useSaveFile (filename from the room
name) with the standard download toast. The visible code stays an SVG so it
renders crisply at any size/theme.
Also corrects the stale LOTUS_FEATURES note (the QR is generated locally via
qrcode.react, not api.qrserver.com).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The shared account-data store removes optimistically with no rollback, so the
reminder row vanished the instant Cancel was clicked. Showing a "could not
cancel" error beside the already-gone row (and it reappearing on next sync) was
self-contradictory. Match the removeBookmark convention: fire-and-forget
optimistic removal, no inline error. Drops the now-moot cancelling busy-guard
and uses a collision-safe React key for same-minute custom reminders.
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>
Review noted the 500-entry io.lotus.bookmarks blob approaches the ~65KB event
limit. The senderId I stored was never read back (the panel re-resolves the
live sender via the event and uses the senderName snapshot otherwise), so it
was dead payload. Keep only senderName.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Saved-message cards showed the room and time but not the author, so in a busy
room you couldn't tell who said it without jumping. Now each card shows
"{sender} - {time ago}":
- Bookmark gains optional senderId/senderName (snapshotted at save time in
Message.tsx from the already-computed sender display name); optional so
existing stored bookmarks stay valid.
- The panel re-resolves the author's current display name live from the event
when the room is joined, falling back to the stored snapshot for left rooms.
- Search now also matches the author name.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- mediaFilename: only skip appending an extension when the body already ends in
a plausible short alphanumeric extension, so "Screenshot 2024.01.05" still
gets a real extension appended for the saved file.
- Lightbox pan: divide the translate by zoom (it runs nested inside scale), so
dragging tracks the cursor 1:1 instead of moving `zoom`x too far.
- Wheel: ignore deltaY === 0 (pure horizontal scroll no longer zooms out).
- Zoom-out button disables at the real reachable minimum (0.2, not dead 0.1).
- Zoom-controls group gets role="group" so its aria-label is announced.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The timeline image viewer supports zoom/pan, but the gallery's own lightbox
rendered a plain object-fit:contain image. Add the same affordances for images
(videos keep their native controls):
- scroll wheel or header -/+ buttons to zoom; +/-/0 keys; double-click or the
% chip toggles 1x<->2x
- drag to pan when zoomed; zoom/pan reset when navigating to another item
Reuses the shared useZoom/usePan hooks (usePan already cleans up drag listeners
on unmount and resets pan when zoom returns to 1x).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The gallery's File and Audio tabs already had download buttons, but images and
videos could only be saved by jumping to the source message. Add:
- a Download button in the lightbox header (full-resolution source), and
- a hover/focus download button on each image/video grid tile
Both reuse the shared FileDownloadButton (decrypts E2EE media client-side, saves
via useSaveFile, spinner/check/retry states). The tile download control is a
sibling of the tile button (not nested — avoids interactive-in-interactive) and
stays visible on touch (hover:none) devices. Download always targets the
full-res file/url, not the thumbnail.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
handleEdit already clears cancelErrors for the old row; also clear sendErrors
so a prior failed "Send now" doesn't leave a stale inline error after editing.
Cosmetic hygiene, matching the existing cancelErrors handling.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The folds Input container is display:flex with no width, so inside the row's
grow="Yes" wrapper the wrapper grew but the input collapsed to content width.
Set width:100% on the Input so the name field uses the space the row gives it.
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>
Address findings from 2 review agents (pause/resume duration model +
meter lifecycle were verified correct):
- Mic-stream leak (HIGH, pre-existing but in this change's blast radius):
the mic tracks were only stopped inside mr.onstop, which cancel() nulls
and the unmount effect never triggered — so cancelling or unmounting
mid-recording/pause left the mic live (OS indicator on). Hold the stream
in a ref and release its tracks explicitly (stopStream) on cancel and on
unmount, independent of onstop. Normal stop still releases via onstop.
- Defensive: startMeters now cancels any existing rAF/interval before
starting, so it can never spawn a second loop.
- a11y/UX: the finish button (checkmark, advances to the preview step) is
relabeled "Finish recording"/"Finish" so the label matches the check
glyph (was "Stop recording" with the old pause icon). The three
recording-control buttons get flexShrink:0 so they don't squish the
waveform at narrow composer widths.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The voice recorder was a single continuous take — an interruption meant
stopping early or starting over (and the Stop button confusingly used a
Pause icon). Add real pause/resume.
- MediaRecorder.pause()/resume() with a new 'paused' state.
- Duration now accumulates only active-recording time: an accumulate-on-
pause model (accumulatedMsRef + segmentStartRef) replaces the wall-clock
startTime, so paused time is excluded from both the live timer and the
finalized preview duration.
- Extracted startMeters/stopMeters so the waveform rAF + timer interval are
reused across start/resume; stopMeters keeps the audio graph alive for
resume while stopAll tears it down.
- Recording view now also renders the 'paused' state: a Pause/Resume toggle
(Pause vs Play icon), the record dot stops pulsing (dimmed), and the
waveform/timer freeze. Stop/Cancel/unmount all handle a paused recorder.
- Fixed the mislabeled finish button: it now shows a checkmark (it advances
to the preview step) instead of a pause icon.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Address findings from 2 review agents (core edit path was verified
correct — media preservation incl. encrypted, threading, no-op guard):
- Require a filename (utils/room.ts): canEditCaption now also checks the
MSC2530 `filename` exists. Fixes media from clients that omit filename,
where the editor prefilled the filename as a caption and clearing it
wrote an empty body. Such media simply isn't caption-editable (matches
renderCaption never showing a caption for it).
- Carry m.mentions (MessageEditor): a caption edit now unions typed
@-mentions with prior mentions like the text-edit path, so mentioning
someone in a caption edit notifies them.
- Double caption: revert to the editor replacing the content while editing
(as text edits do) instead of rendering the media + its caption above an
editor prefilled with the same caption — removes the confusing duplicate.
- Removed-caption "(edited)" marker (RenderMessageContent): when a media
message is edited but has no caption (e.g. the caption was removed),
render the standalone "(edited)" affordance so Edit History stays
reachable (previously it lived only inside the caption and vanished).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Captions could be attached at upload but never changed — canEditEvent
only allowed m.text/emote/notice, so a typo in an image caption meant
delete + re-upload. Add caption editing for image/video messages.
- utils/room.ts: canEditCaption (own image/video RoomMessage, no
non-thread relation) + canEditEventOrCaption. canEditEvent unchanged.
- Message.tsx: gate the Edit affordance (quick-actions + menu) on
canEditEventOrCaption; label it "Edit caption" for media; keep the media
rendered above the editor while editing.
- MessageEditor.tsx: for a media message, seed the editor from the caption
(not the filename), allow an empty caption (removes it), and build the
m.replace so m.new_content spreads the original media content
(url/info/encrypted file/filename/msgtype) and only sets body +
format/formatted_body. Outer content is the full media (not a "* text"
fallback) so non-edit-aware clients still render the media. No-op guard
when the caption is unchanged. Placeholder "Add a caption…".
Rendering + Edit History need no changes: getEditedEvent's m.new_content
flows to renderCaption, and the word-diff already diffs body (the
caption). Encrypted media keeps its file/key (no re-upload).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Address findings from 2 review agents on the recent-GIFs row:
- Motion/perf: the Recent row rendered up to 16 full animated GIFs at
once (autoplaying). Capture a small still image at pick time
(fixed_width_small_still / *_still) into RecentGif.previewUrl and render
that for the thumbnail, so recents no longer autoplay. Pre-existing
recents without a preview fall back to the animated url. Re-send still
uses the animated url, so the sent m.image is unchanged.
- a11y: the recent buttons all had the identical label "Send recent GIF".
Give them positional labels ("Send recent GIF N of M") and wrap the grid
in a role="group" labelled by the "Recent" section heading, so the row
is a distinguishable, announced group.
Correctness review found no bugs (write-before-unmount, term gating,
dedupe, re-send fidelity all verified).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The GIF picker was a bare Giphy search grid with no memory of what you've
sent, so re-sending a go-to reaction GIF meant re-typing the search every
time. Add a "Recent" row at the top of the picker (default view; hidden
while searching) for one-click re-sending.
- New persisted state state/recentGifs.ts: recentGifsAtom (localStorage,
cinny_recent_gifs_v1, getOnInit) + pure addRecentGif (dedupe-by-url
move-to-front, cap 16, ignore empty url), with 5 unit tests.
- GifPicker records every sent GIF (from search or the Recent row) to the
front, and renders a 3-up thumbnail grid of recents above the search
grid when there are recents and no active search term. Section label
matches the picker's existing `// GIF_SEARCH` treatment (lotusTerminal)
or a muted label otherwise.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Address findings from 2 review agents on the room-nav draft indicator:
- Icon semantics (a11y): Icons.Message read as message activity and
collided with two existing bubble uses in the same row (call-chat
toggle, mark-unread), especially next to the unread badge. Replace it
with the composer's shared DraftDot (a small color.Success.Main dot),
so both draft surfaces share one visual language. Rendered as a
role="img" span with aria-label "Unsent draft" (reliably announced,
unlike a bare aria-labelled svg).
- Precise thread-key filter: hydration skipped any draftKey containing
'::', which would also skip an IPv6-literal server name in a roomId.
Match '::$' (thread root is an event id) so only real thread drafts are
skipped.
- Defensive hasMsgDraft: guard toPlainText so a corrupted/foreign draft
value can't throw during a nav render.
- Clear the draft atom on send: the send / scheduled-send handlers reset
the editor and localStorage but left the jotai draft atom set, so the
composer DraftIndicator could show a stale dot after sending a restored
draft. Add setMsgDraft([]) to both.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
jaredandClaude Opus 4.8