When a relay-created dm_message_created arrives for an unknown channel,
check if an existing DM has the same member (by homeUserId). If so,
route the message to the existing channel instead of adding a duplicate
sidebar entry. Prevents the transient duplicate that required page
refresh to dedup.
Refactor useGridLayout from RefObject to callback-ref API so the
ResizeObserver reattaches when the grid container remounts after
focus-mode toggle. Fixes tiles rendering at stale fullscreen dimensions
on 4:3 monitors.
Also clamp focus-mode participant strip height (max-h-[20vh] min-h-[80px])
to prevent it from crowding the focused stream on short viewports.
When a WS connection receives its ready event, push cached myActivities
to that instance so federated users see activities immediately without
waiting for the next debounce cycle.
Inserts a new tier 3 between DM member lookup and identity-only fallback.
When the current space's member list doesn't contain the voice participant
(user navigated to a different space/instance), the hook now falls back
to the cached User object from ParticipantInfo. This preserves avatar,
avatarColor, displayName, and homeUserId for federated users in PiP.
The isLocal branch was falling through to cachedUser.id when
homeUserId was null, diverging from the null-for-home-instance
contract used by the memberMatch and carry-forward branches.
Avatar.tsx already falls back to user.id internally.
Extends ParticipantInfo with cachedUser: User | null. In updateParticipants,
snapshots the previous participant list and carries forward the cached User
object when spaceStore member lookup fails (user navigated to a different
space). Local user falls back to authStore. homeUserId is now derived from
cachedUser when the direct member lookup misses. Updates EMPTY_PARTICIPANT
in PictureInPicture.tsx.
stopScreenShare() resets hwOverdrive: false, which clobbers the
user's intent when switching VP9 → H.264 HW mid-stream. Save
and restore the flag around the restart.
Add hwOverdrive: false to resetSession(), leaveVoice(), and
handleForceDisconnect() store resets. Add deactivateHwOverdrive()
to disconnect callback and cleanup effect in useLiveKit.
Ensures "resets when stream ends" contract holds for edge cases
like server shutdown, identity collision, and logout.
- Connection Info now shows encoder impl (e.g. libvpx, OpenH264,
ExternalEncoder) on send-side video tracks alongside quality
limitation reason
- Amber warning "VP9 is recommended" when H.264 is selected
- Gaming mode defaults to H.264 (NVENC hardware encoding, zero CPU)
- Text mode defaults to VP9 (better compression, CPU is free)
- Changing mode auto-updates codec to smart default
- User can manually override codec via UI pills
- Changing codec mid-stream auto-restarts the stream (~1s interruption)
- Gaming mode uses H.264 primary (hardware NVENC encoding, zero CPU
impact on games). Text mode uses VP9 primary with H.264 backup
and SIMULCAST policy for Safari compatibility.
- Fix native mode starting at 30fps by decoupling frameRate constraint
from resolution constraint in both screenShare.ts overdrive and
useLiveKit.ts updateActiveTracks.
- Filter paused backup codec tracks in Connection Info stats so dead
0kbps entries don't show alongside active codec tracks.
Replace the useEffect-based auto-connect pattern in AppLayout with
direct connect/disconnect calls from user gesture contexts. This is
required for iOS (AudioContext.resume + getUserMedia must happen in
a gesture handler) and aligns with tightening autoplay policies on
desktop browsers.
Architecture:
- voiceStore gains connectFn/disconnectFn refs, registered by AppLayout
from the single useLiveKit() instance.
- All voice join paths (ChannelSidebar, MobileSpacesScreen, MainContent,
voice_moved WS handler) pass connectFn to joinVoiceChannel().
- All disconnect paths (VoiceControls, voiceActions, MobileVoiceFullScreen,
MobileVoiceMiniBar, dm_call_ended/rejected WS handlers, ready handler)
call disconnectFn() directly.
- dm_call_accepted WS handler calls connectFn() to initiate the DM call
LiveKit connection.
- The 55-line auto-connect useEffect and lastAttemptedRef are removed.
Create useLongPress hook with 500ms delay, 10px movement cancellation
(Euclidean distance), and ghost click suppression. Integrate into
Message component to trigger the existing context menu bottom sheet
on touch devices.
Add installers/ to .gitignore and remove from tracking — binaries
should not be in git.
Also includes:
- Re-push Electron activities after WebSocket reconnect (sleep/wake)
- Merge activity map instead of replacing to preserve local state
- Nori mascot design plan
When the app is focused and the native uiohook hook is active, both the
IPC bridge and the web fallback capture the same keypress, dispatching
the toggle action twice (~0-20ms apart). This double-toggles mute/deafen
back to the original state.
Add a 100ms dedup window in dispatchKeybindAction to suppress the second
dispatch. Safe for human input (key repeat starts at ~500ms).
- Add .catch handler to scheduleNext to prevent unhandled promise
rejections from killing the scheduling loop
- Replace `as unknown as Animation` cast with Cancellable interface
for type-safe cleanup handles
Implement the <Mascot> component that renders Nori in four emotional
states (idle, sleeping, excited, lonely) with distinct SVG geometry,
color palettes, and facial expressions. Includes useMascotAnimation
hook stub for Task 2 and full test coverage (15 tests).
When the Electron main process can't start uiohook (macOS Accessibility
denied, Linux missing input group), the IPC path silently produces no
events. Previously this left zero keybind listeners active. Now the web
capture-phase fallback is always set up as a safety net, ensuring
keybinds work in-app even when the OS-level hook fails.
When running in Electron but the desktop app hasn't been rebuilt with
keybind support (no syncKeybinds/onKeybindAction in preload), the hook
now falls through to the web capture-phase listener instead of silently
doing nothing. This ensures keybinds work immediately on the web-served
UI regardless of the desktop app version.
The previous hash (charCodeAt(0)*256 + charCodeAt(1)) only used the
first 2 characters of KeyboardEvent.code strings, causing all Key*
codes (KeyA, KeyB, ..., KeyZ) to collide to the same numeric value.
This made Ctrl+D and Ctrl+M appear as identical bindings (false
conflict) and broke keybind matching in the web fallback.
djb2 hashes the full string, producing unique values for all codes.
The continuous scroll was wrong for Instance — its sub-panels are
fundamentally different page types (user table, storage tool, bitrate
matrix, toggles). Revert to conditional rendering.
The hook now supports two modes:
- Scroll mode (default): smooth scroll + IntersectionObserver spy
- Tab mode (onNavigate callback): sidebar sub-links switch tabs
Instance uses tab mode. The scroll-spy infrastructure remains for
future panels that benefit from it (e.g. Account sections).
Also reverts GeneralPanel to sticky save bar since it's on its own
page again with no overlap risk.
- Memoize context value in SettingsSectionsProvider
- Use ctxRef pattern in hook to avoid depending on ctx object
in useLayoutEffect deps (ctx object changes on every render,
causing effect → setState → re-render → effect loop)
Replace the stale-flag-based voice re-registration in the WS ready
handler with a server-authoritative check. If the server's voiceStates
doesn't include us (server restarted, lost in-memory voiceRooms),
cleanly tear down the LiveKit session via leaveVoice() + disconnect().
If the server still knows about us (WS blip), do nothing.
- Task 7: Create activityStore (Zustand) with Map<userId, Activity[]>, showActivity toggle,
debounced pushActivities, and reset on logout
- Task 8: Wire WS integration — presence_update propagates activities to store,
ready payload initializes userActivities and showActivity (via setState to avoid side effects)
- Task 9: Create ActivityCard component with compact/full modes, type-colored labels,
elapsed time display, asset images, and fallback to customStatus
- Task 10: Upgrade ActivityPanel to three groups — active friends (full cards, no header),
online friends (compact), offline friends — using getPrimaryActivity for grouping
- Task 11: Upgrade MemberSidebar to show compact ActivityCards instead of raw customStatus
- Task 12: Add showActivity toggle in AccountPanel settings (Activity Status section)
with API persistence and store sync
- chatStore: normalize embeds to [] in addMessage, addRealtimeMessage, and updateMessage to guard against missing field from older servers/federation
- useWebSocket: add embeds_resolved and dm_embeds_resolved cases that patch the in-memory message cache when the server broadcasts resolved embed data
- useWebSocket: normalize embed image URLs for remote origins in message_created, message_updated, dm_message_created, dm_message_updated, and the new embeds_resolved handlers
handleVoiceDropZoneDrop called stopPropagation and clearState
unconditionally, preventing container's handleContainerDrop from
firing when dropping a channel onto a voice channel. Add early
return for non-voiceUser drags so channel/category drops bubble
through to the container handler correctly.
Port the SpaceSidebar normalization pattern: convert 'before B'
to 'after A' so only one drop indicator renders at any gap.
Fixes double-line visual glitch and unreliable drops when cursor
is between two channels. Also adjust indicator offsets from
top-0/bottom-0 to -top-[1px]/-bottom-[1px] to center in the gap.
Remove activeDrag from drag-start dependency arrays to prevent
unnecessary re-renders during drag. Wrap containerHandlers in
useMemo for consistent reference stability.
When a user joins a voice channel while already connected from another
tab, the server now sends a 'displaced' reason on voice_disconnected.
The client tears down LiveKit and shows an informational toast.
Voice channels rendered with VoiceChannel component have no text reading/acking
UI, so messages in them created phantom unread indicators on the space sidebar
that users could never clear. Root cause was a message in the counter-strike
voice channel with no read state.
Three-layer fix:
- spaceStore: track voiceChannelIds set, exclude voice channels from
channelLastMessageIds so setReadStates never marks them unread
- useWebSocket: skip markChannelUnread for voice channels on message_created,
prune orphaned unreads on every ready event
- chatStore: validate preserved unreads against channelToSpaceMap to drop
orphans that don't map to any known channel
After a server restart, navigating to a previously-visited channel showed
messages at a stale middle position instead of scrolling to the bottom.
The in-memory message cache survived the reconnect, so loadMessages()
bailed (cache hit) and the scroll-to-bottom logic never fired.
Now the ready handler clears the messages and hasMore maps for all
channels belonging to the reconnecting origin (including DMs for home).
The currently open channel is force-reloaded immediately; other channels
get fresh-fetched on next visit, triggering proper scroll-to-bottom.
Also fixes voice channel settings gear icon placement to match text
channels (flex-1 pushes icon to right edge).
The channel sidebar voice user list was maintained by a separate
voiceUsers Map (fed by WS events + fragile hydration code) that diverged
from reality after server restarts — users shown in wrong channels,
duplicated across channels. The VoiceGrid was always correct because it
reads LiveKit participants directly.
Now VoiceChannel.tsx derives its user list from LiveKit participants for
the connected channel (single source of truth) and only falls back to
server-provided voiceUsers for channels the user is not connected to.
Removed all hydration band-aids that tried to sync the two systems:
- useLiveKit ParticipantDisconnected → removeVoiceUser
- useLiveKit ConnectionStateChanged → addVoiceUser hydration loop
- useWebSocket ready handler → dynamic import LiveKit hydration
Also includes: voice channel settings gear icon on hover, persist
per-user volume/mute prefs across sessions, default screen share
audio off on Electron (no system audio capture support).
Jannis Braun