Remove disabled={isUploading} from the textarea so focus() isn't silently
ignored on a disabled element. Fix ResizeObserver effect deps so the
observer is actually created after the loading spinner transitions to
message content.
Use origin-aware user ID resolution for optimistic voice add/remove
instead of always using the home instance ID. Adds getMyUserIdForOrigin
resolver (same pattern as getApiForOrigin) so joinVoiceChannel and
leaveVoice use the correct federated user ID. Also fixes WS reconnect
voice re-registration, invite URL for remote servers, and chatStore
early-load guard for unmapped channels.
The redirect effect used the shared `channels` array which could contain
stale data from a previously viewed server during async loadServerDetail.
Added channelToServerMap ownership check to ensure channels belong to
the target server before redirecting. Also persists last-visited channel
per server in localStorage for better return navigation.
The Explore page shared /channels/@me with Friends, differentiated by a
showExplore UI flag. When PiP navigated to a voice channel, the stale
flag caused MainContent to render text chat instead of the voice grid.
Replace the flag with a dedicated /explore route so the URL is the
single source of truth.
Joined discoverable servers now appear in a collapsible "Joined" section
at the bottom of the explore page, so the page never feels dead when all
servers have been joined. A mint success banner replaces the old empty
state text.
$client doesn't exist in drizzle-orm 0.33, causing a TypeError and
500 on every /api/servers/explore request. Export the raw better-sqlite3
instance directly from db/index.ts instead.
The echo gate architecture bypassed the serialized mic management chain
and introduced race conditions. Chrome's AEC handles echo properly for
both headphone and speaker users without custom intervention.
Send voice_join to WS server only after LiveKit successfully connects,
not immediately on joinVoiceChannel() or WS reconnect. Background tabs
can reconnect WebSocket but not WebRTC, causing phantom voice users.
Previously, the voice channel sidebar only showed users after a server
round-trip (voice_state_update broadcast). After a deploy/reconnect,
this left the user invisible in the sidebar despite being connected.
Now joinVoiceChannel, leaveVoice, and the WS ready handler all
optimistically update voiceUsers for the local user immediately.
Clear stale currentChannelId when clicking Explore/DMs in ServerSidebar,
and guard AppLayout route effect from clobbering showExplore. Move member
exclusion into SQL for correct pagination/totals, surface allSettled errors,
and show context-aware empty state messages.
Send explicit voice_leave to the old instance when joining voice on a
different origin, preventing stale voice state from showing the user in
two channels. Also make WS reconnect voice re-registration origin-aware
so remote reconnects properly restore voice state.
Replace the unreliable client-side DM dedup loop in populateFromReady with
a creation-time guard (findExistingDmForUser) that checks all instances
before opening a new DM. Guards added to FriendsPage, NewDmModal, and
UserProfilePopout.
Also fixes: corrupted read_states from temp_ optimistic message IDs (server
migration + client-side validation), federation-aware closeDm/addDmMember
API routing, isSelf-based DM member filtering in sidebar/header, and WS
event error isolation.
- Fix DM welcome header avatar using home identity for correct gradient color
- Register DM channel IDs in channelOriginMap so federated DM operations
(reactions, messages, typing) route to the correct instance
- Pass origin when creating DM channels from friends list and WS events
- Add server discovery/explore page with public server listings
- Add server visibility and description fields
Rewrites socialStore to aggregate friends and requests from all connected
instances using Promise.allSettled. Parses user@domain in friend requests
to route to the correct instance. Removes !isHome guards on social WS
events so remote friend requests arrive in real-time. Shows "via hostname"
labels on remote friends/requests in the UI.
Store the original home snowflake ID (homeUserId) during federation replication
so that avatar gradient colors resolve identically across instances. Previously,
replicated users got new snowflake IDs on each instance, causing different
gradient colors. Now Avatar, UserProfilePopout, VoiceUser, StreamTile, and
VoiceChannel all resolve through homeUserId when available. Includes backfill
logic for existing federated users missing the field.
Federated users get a different snowflake ID on remote instances,
causing avatar gradients to mismatch their home identity. Apply
resolveDisplayIdentity/isSelf resolution in VoiceUser, StreamTile,
VoiceChannel sidebar, and MemberSidebar so the current user's avatar
color is consistent across all views.
Move TypingIndicator from document flow (between MessageList and
MessageInput) into MessageInput's outer wrapper using absolute
bottom-full positioning. Fixes indicator being hidden behind the
floating glass input bubble on desktop.
- Include user object in reaction_added WS broadcasts for isSelf() resolution
- Use isSelf() instead of userId comparison for reaction ownership checks
- Load remote server detail after ready event to prevent empty channel list
Add identity.ts with isSelf() and resolveDisplayIdentity() — pure
stateless functions that detect replicated-self using the immutable
(username, homeInstance) composite key. No store lookups, no data
mutation. Fixes wrong avatar gradient and missing edit/delete on
own messages in remote channels.
Also includes: optimistic message dedup fix for cross-instance
messages (content-only matching), federation toast notifications,
Username component with @domain display, invite parser, and
deploy script simplification.
Batch 1 — Bug fixes:
- Fix stale voice state on remote reconnect (clearVoiceUsersForOrigin)
- Fix logout not cleaning remote servers from store
- Fix reply-to asset normalization for remote messages
- Fix HTTPS hardcoded in autoConnectAll (store full origin, legacy fallback)
Batch 2 — Password enforcement + replication flow:
- Add connectToRemote() with home password verification before remote auth
- Auto-cascade: verify home password → register on remote → login fallback
- Replace Register/Login tabs with single password field in ConnectedInstances
- Add DifferentPasswordError for typed fallback-login UI transition
- Fallback login form shown only when remote has different password
The env_val() function uses grep which returns exit code 1 when no
match is found. Under set -euo pipefail, this kills the script.
Adding || true prevents this on upgrade installs with older .env files.
The node -e command to set instance name needs to run from
/app/packages/server where pnpm's symlinked node_modules resolve
better-sqlite3, not from /app root.
Redesign deployment as a single docker-compose with Backspace, Caddy
(auto-HTTPS), and LiveKit (voice/video) using hybrid networking:
Backspace+Caddy on isolated bridge, LiveKit on host mode for WebRTC.
- Add install.sh: interactive installer that handles Docker setup,
domain/DNS verification, secret generation, LiveKit config, and
deployment with health-check wait
- Add Caddyfile: static reverse proxy config using Caddy env vars,
routes /livekit/* to host-mode LiveKit via host.docker.internal
- Rewrite docker-compose.yml: all-in-one with profiles (voice),
no external volumes/networks, bind mount ./data for visibility
- Fix livekit.ts: use LIVEKIT_URL env var directly instead of
Host-header derivation that made the env var dead code
- Fix Dockerfile: health check reads $PORT dynamically
- Update .env.example: add DOMAIN, COMPOSE_PROFILES documentation
- Update .gitignore: add livekit.yaml (contains secrets)
Phase 5 of multi-instance federation. Adds a two-layer fix:
Layer 1 — Data ingestion normalization: Remote instance user avatars,
server icons, and attachment filenames are rewritten to absolute URLs
when entering the app (via WebSocket events or API responses), so all
downstream components render them correctly without changes.
Layer 2 — Outbound action routing: wsSend calls (voice join/leave/status,
typing) and file uploads in UI components now route through the correct
instance based on the active channel's origin.
Two fixes addressing architectural review feedback:
1. Snowflake ID collisions: Replace process.pid-based worker ID with a
cryptographically random value (0-1023) generated once at first boot
and persisted to instance_settings.worker_id. Eliminates deterministic
ID collisions between Docker instances that all run as PID 1.
2. Reaction API leak: Revert addReaction/removeReaction signatures to
(messageId, emoji) — the store now resolves the channel internally by
scanning its message cache, keeping routing logic out of the UI layer.
Refactor the WebSocket layer from a singleton connection to a connection
map supporting N concurrent instances. Each connection has its own
heartbeat worker, reconnect state, and token. Stores are now
instance-aware: serverStore merges servers by origin, chatStore routes
API calls through the correct client, and instanceStore triggers WS
connect/disconnect on add/remove. DM, social, and voice events remain
home-only.
Introduce instanceStore (Zustand) with full federation lifecycle:
probe remote instances, register with username collision fallback,
login to existing accounts, sync instance list across all connected
instances, auto-reconnect from cached tokens on login/page load, and
cleanup on logout. Add ConnectedInstances component to User Settings
with home instance card, remote instance management, and inline
add-instance flow (URL probe → register/login → connected).
Refactor singleton api object into a class with constructor-scoped
closures parameterized by baseUrl and getToken. Export backward-
compatible api singleton, createApiClient factory for remote instances,
and new instance.info() and users.verifyPassword() methods for
federation support.
Add multi-instance support foundation: shared federation types
(ReplicatedInstance, InstanceInfoResponse, VerifyPasswordRequest),
database schema changes (home_instance, replicated_instances on users,
instance_name on settings), public instance info endpoint, auth
registration with homeInstance and username@domain collision fallback,
password verification endpoint, and replicatedInstances sync on user
profile. Extract duplicated sanitizeUser into shared utility across
8 server files.
Root cause fix for all broken CSS variable opacity modifiers. Converted
every hex CSS variable to space-separated RGB channels (e.g. #fda4af →
253 164 175) and updated Tailwind config to use rgb(var(...) / <alpha-value>).
This makes bg-accent-rose/10, border-border-hard/50, etc. work natively
everywhere — no more currentColor fallback. Reverted all previous explicit
rgba() workarounds back to clean Tailwind syntax. Also fixed the voice
connection wifi icon SVG (was bottom-left aligned, now centered).
Server-side: handleReactionAdd/Remove now check dm_messages table when
message isn't found in server messages, enabling DM emoji reactions
via the same unified client event path.
Client-side: replaced all remaining broken opacity modifiers on CSS
variables (border-border-soft/50, bg-accent-rose/10, bg-surface-*/N,
etc.) with explicit rgba values. Tailwind can't decompose var() for
opacity, causing fallback to currentColor. Fixes 16 files across
error boxes, table borders, surface tints, and input separators.
Consolidated two competing scroll effects into one to fix an effect
ordering bug where prevMessagesLength was updated before the initial
scroll check could read it. Channel switch now resets tracking state
so the snap-to-bottom fires reliably via requestAnimationFrame.
Increased message list bottom padding to clear the glass input bubble.
Tailwind can't decompose CSS variables for opacity modifiers, so
border-border-hard/50 fell back to white currentColor. Replaced with
border-white/[0.06] on image attachments, code blocks, and action
toolbar. Also applied frosted glass material to message action toolbar.
Replace @apply border-none with @apply border-0 on global * selector.
border-none sets border-style: none, which silently kills all Tailwind
border utilities (they only set width/color, relying on preflight's
border-style: solid). border-0 sets border-width: 0 instead, preserving
the solid style so border-r/border-l/border-b utilities render correctly.
- Add border-r border-border-hard to ChannelSidebar (both DM and server views)
- Add border-l border-border-hard to MemberSidebar and ActivityPanel
- Replace shadow-header with border-b border-border-hard on sidebar headers
- Remove duplicate mt-1 on reaction container (parent gap-1 already provides 4px)
Channel sidebar: subtle bg-surface-elevated highlight with rounded-[6px],
active channel left pill, rose unread dot on right, smaller #/voice icons
(18px), 10px horizontal padding, 24px voice user avatars at 36px indent.
Reaction bubbles: inline styles for background/border colors to bypass
Tailwind opacity modifier failures on hex CSS variables. Both mine (mint
tint) and non-mine states now render visible pill containers reliably.
Jannis Braun