3 Commits

Author	SHA1	Message	Date
Siavash Sameni	4471797edf	docs: update all PRDs and PROGRESS to current state (2026-04-13) ... Updated 6 PRDs with implementation status: - PRD-adaptive-quality: P2P quality done, bandwidth probing remains - PRD-protocol-analyzer: all 5 phases documented - PRD-relay-concurrency: DashMap + clone-before-send done - PRD-p2p-direct: P2P adaptive quality update - PRD-engine-dedup: all phases done - PROGRESS.md: test count 372+, 3 new change sections Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-13 16:40:56 +04:00
Siavash Sameni	a52b011fb5	feat(relay): replace global Mutex<RoomManager> with DashMap sharding ... Eliminates the single-lock bottleneck for media forwarding. Before: all participants across all rooms competed for one Mutex. Now rooms are stored in DashMap (64 internal shards with per-shard RwLocks). Changes: - RoomManager.rooms: HashMap → DashMap<String, Room> - Per-room quality tracking (qualities, current_tier moved into Room) - Arc<Mutex<RoomManager>> → Arc<RoomManager> everywhere - 20 .lock().await sites removed across room.rs, main.rs, federation.rs, ws.rs - federation forward_to_peers: clone peer list, release lock, then send - ACL uses std::sync::Mutex (rarely accessed, non-async) Concurrency improvement: - Before: 100 rooms × 10 people = 1000 tasks → 1 Mutex - After: distributed across 64 DashMap shards, ~15 tasks per shard avg - Rooms are fully independent — room A never blocks room B 314 tests passing, 0 regressions. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-13 12:17:57 +04:00
Siavash Sameni	2514151a89	docs: PRD for relay concurrency — per-room lock sharding ... Full analysis of relay lock contention with precise inventory of every lock acquisition in the hot path. Evaluates 4 design options: A) Per-room Arc<Mutex<Room>> (recommended — 100x improvement for multi-room) B) DashMap (good but less explicit) C) Channel-based fan-out (over-engineered for current scale) D) Snapshot-on-change via arc-swap (best perf, more complex) Phase 1: per-room locks, Phase 2: federation lock fix, Phase 3: quality tracking out of critical path. Estimated 1.5-2.5 days total. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-13 12:01:21 +04:00