feat(sync): add canonical room media state

This commit is contained in:
KoalaDev
2026-08-20 17:27:08 +02:00
parent 1e41d98c2f
commit c064953cca
16 changed files with 1191 additions and 15 deletions
+21
View File
@@ -32,6 +32,27 @@ Ensures all peers are buffered and synchronized before resuming:
> **Network Transit Buffer Rule**: The orchestrator (`background.js`) must always use a timeout at least 500ms longer than the worker (`content.js`) to account for IPC and network transit time. Never align them exactly 1:1, as this will introduce a race condition on slow connections.
4. **Resume**: All peers call `play()` simultaneously.
## 3.1 Canonical Media State v1
The relay keeps one optional, in-memory canonical playback state per active room.
Accepted `PLAY`, `PAUSE`, and `SEEK` commands advance a server-owned revision.
Playing positions advance lazily from the server update time; paused positions do
not. Heartbeats remain observational and do not mutate canonical state.
`ROOM_DATA` materializes the playing position at snapshot creation and advertises
the optional `media-state-v1` capability. A joining/reconnecting client validates
the room/revision, respects Host Control solo mode and Episode Lobby, then sends an
internal `APPLY_CANONICAL_MEDIA_STATE` message to the existing content/video path.
That path reuses frame election, Netflix/Disney page-API seeks, native play/pause,
the 2-second drift tolerance, and programmatic-event suppression. The apply is
one-shot recovery: it creates no action history, notification, command ACK, or
relay media event.
Force Sync remains a two-phase ACK protocol. `PREPARE` is temporary choreography;
the matching `EXECUTE` commits its validated target to canonical state. Per-sender
`seq`, peer heartbeats, and the existing reconnect event queue remain separate.
Offline media-command compaction is intentionally deferred.
## 4. Episode Auto-Sync
Maintains continuous synchronized viewing when watching series:
1. **Detection**: `content.js` monitors the Media Session API for title changes.
+58 -2
View File
@@ -83,13 +83,67 @@ Payload:
"hostPeerId": "string or null",
"controlMode": "everyone | host-only",
"controllers": ["peerId"],
"capabilities": ["host-control", "co-host", "chat", "chat-v1"]
"mediaState": "canonical media state object or null",
"capabilities": ["host-control", "co-host", "chat", "chat-v1", "media-state-v1"]
}
```
`room_data` is sent to the joining socket. It is not the general broadcast used
for every later room update.
## Canonical Media State v1
Relays advertise this optional recovery primitive with `"media-state-v1"` in
`room_data.capabilities`. Each active room stores at most one state:
```json
{
"revision": 42,
"playbackState": "playing",
"currentTime": 1234.5,
"updatedAt": 1787234425123,
"updatedBy": "peer-id"
}
```
The internal `currentTime` is the media position at server-owned `updatedAt`.
Playing state advances lazily when a snapshot is requested; paused state stays
fixed. `revision`, `updatedAt`, and `updatedBy` are server-owned. Clients cannot
spoof them. The wire snapshot contains the already-projected `currentTime`,
`revision`, `playbackState`, and `updatedBy`, so clients never compare client and
server wall clocks.
Only accepted, sanitized room controls update canonical state:
- `play` uses its valid `currentTime`, or an existing effective canonical position.
- `pause` uses its valid `currentTime`, or freezes an existing effective position.
- `seek` prefers `targetTime` (with `currentTime` compatibility) and preserves the
established playback state.
- `force_sync_prepare` records only temporary coordination state. Its matching
`force_sync_execute` commits the prepared target as playing.
`peer_status` heartbeats are observations and never rewrite canonical intent.
Per-sender `seq` still orders commands from one sender; canonical `revision`
orders server-accepted room transitions. Neither replaces the other.
On join/reconnect, a capable extension applies a valid snapshot once through an
extension-internal recovery message. Existing seek/page-API and native-event
suppression prevent `play`, `pause`, or `seek` echoes. Pending recovery is scoped
to the room/revision in `chrome.storage.session`, waits for the selected media
target lifecycle, and is cleared on leave/switch. Intentional host-only guest
desync and an active Episode Lobby take precedence over snapshot recovery.
Compatibility is additive: new clients use old behavior with a relay that omits
the capability; old clients ignore the extra `room_data` field from a new relay.
A new relay canonicalizes every accepted legacy `play`, `pause`, `seek`, and
matching Force Sync command regardless of `join_room.clientCapabilities`, while
relaying the established event names, payloads, and order unchanged. This makes
server-first rollout safe: old clients populate recovery state without needing to
understand or acknowledge it, and new clients consume it only when the relay
advertises the capability. No protocol-version or minimum-version bump is
required. Offline `play`/`pause`/`seek` compaction is planned separately and is
not part of Media State v1.
## Ephemeral encrypted chat
Relays advertise chat support with `"chat-v1"` in `room_data.capabilities` and keep
@@ -253,7 +307,9 @@ them with the same sanitized relay envelope as other room events, including
### `force_sync_execute`
Payload includes `targetTime`. In `host-only` mode, only controllers may send it.
The current extension sends sequence/action metadata but no target; the relay uses
the validated target retained from the matching `force_sync_prepare`. In
`host-only` mode, only controllers may send it.
The relay also allows a matching initiator's execute event after that initiator
started the prepare step, even if their controller state changed before execute.