feat: universal bitstream streaming — any input → any output
New crate: engine/ds-stream/
- Binary protocol: 16-byte header, typed frame/input enums
- Frame types: Pixels, Delta, Audio, Signal, Neural (0x01-0x43)
- Input types: Pointer, Key, Scroll, Gamepad, MIDI, BCI (0x01-0x90)
- WebSocket relay server (tokio + tungstenite)
- Source → receivers: frame broadcast
- Receivers → source: input routing
- Codec: encode/decode, XOR delta compression, RLE, convenience builders
- 17 unit tests, all passing
Streaming modes (stream-source.html):
1. Pixel mode: raw RGBA framebuffer (~28 MB/s)
2. Delta mode: XOR + RLE compression (~1-9 MB/s, 70-95% savings)
3. Signal mode: compact JSON signal diffs (~2 KB/s, 12000x reduction)
4. Neural mode: procedural SDF pixel generator (concept demo)
5. Audio channel: spring velocity→frequency synthesis
6. Multi-receiver: broadcast to all connected clients
Thin receiver client (stream-receiver.html, ~300 lines):
- Zero framework, zero build step
- Renders any incoming bitstream mode
- Local signal-diff renderer for signal mode
- AudioContext playback for audio frames
- Full input capture: click/drag, keyboard, scroll
- Per-channel bitstream bus visualization
DREAMSTACK.md: Phase 7 section with protocol spec
2026-02-25 10:29:44 -08:00
|
|
|
[package]
|
|
|
|
|
name = "ds-stream"
|
engine: v0.28–v0.50 milestone
ds-physics v0.50.0 (138 tests)
- v0.28: apply_body_torque, is_body_sleeping, get_body_angle
- v0.30: set_body_gravity, set_linear_damping, count_awake_bodies
- v0.40: joints (distance/pin), raycast, kinematic, time scale, world stats
- v0.50: point query, explosion, velocity/position set, contacts, gravity, collision groups
ds-stream v0.50.0 (201 tests)
- v0.28: BufferPool, PacketJitterBuffer, RttTracker
- v0.30: FrameRingBuffer, PacketLossDetector, ConnectionQuality
- v0.40: QualityAdapter, SourceMixer, FrameDeduplicator, BackpressureController, HeartbeatMonitor, CompressionTracker, FecEncoder, StreamSnapshot, AdaptivePriorityQueue
- v0.50: StreamCipher, ChannelMux/Demux, FramePacer, CongestionWindow, FlowController, ProtocolNegotiator, ReplayRecorder, BandwidthShaper
ds-stream-wasm v0.50.0 (111 tests)
- WASM bindings for all stream features
ds-screencast v0.50.0
- CLI: --jitter-buffer, --latency-window, --ring-buffer, --loss-threshold, --adaptive, --dedup, --backpressure, --heartbeat-ms, --fec, --encrypt-key, --channels, --pacing-ms, --max-bps, --replay-file
2026-03-11 12:47:56 -07:00
|
|
|
version = "0.50.0"
|
feat: universal bitstream streaming — any input → any output
New crate: engine/ds-stream/
- Binary protocol: 16-byte header, typed frame/input enums
- Frame types: Pixels, Delta, Audio, Signal, Neural (0x01-0x43)
- Input types: Pointer, Key, Scroll, Gamepad, MIDI, BCI (0x01-0x90)
- WebSocket relay server (tokio + tungstenite)
- Source → receivers: frame broadcast
- Receivers → source: input routing
- Codec: encode/decode, XOR delta compression, RLE, convenience builders
- 17 unit tests, all passing
Streaming modes (stream-source.html):
1. Pixel mode: raw RGBA framebuffer (~28 MB/s)
2. Delta mode: XOR + RLE compression (~1-9 MB/s, 70-95% savings)
3. Signal mode: compact JSON signal diffs (~2 KB/s, 12000x reduction)
4. Neural mode: procedural SDF pixel generator (concept demo)
5. Audio channel: spring velocity→frequency synthesis
6. Multi-receiver: broadcast to all connected clients
Thin receiver client (stream-receiver.html, ~300 lines):
- Zero framework, zero build step
- Renders any incoming bitstream mode
- Local signal-diff renderer for signal mode
- AudioContext playback for audio frames
- Full input capture: click/drag, keyboard, scroll
- Per-channel bitstream bus visualization
DREAMSTACK.md: Phase 7 section with protocol spec
2026-02-25 10:29:44 -08:00
|
|
|
edition.workspace = true
|
|
|
|
|
license.workspace = true
|
|
|
|
|
description = "Universal bitstream streaming — any input to any output"
|
|
|
|
|
|
|
|
|
|
[[bin]]
|
|
|
|
|
name = "ds-stream-relay"
|
|
|
|
|
path = "src/main.rs"
|
|
|
|
|
|
|
|
|
|
[lib]
|
|
|
|
|
path = "src/lib.rs"
|
|
|
|
|
|
|
|
|
|
[dependencies]
|
|
|
|
|
tokio = { version = "1", features = ["full"] }
|
|
|
|
|
tokio-tungstenite = "0.24"
|
|
|
|
|
futures-util = "0.3"
|
2026-02-26 10:07:47 -08:00
|
|
|
serde_json = "1"
|
feat: universal bitstream streaming — any input → any output
New crate: engine/ds-stream/
- Binary protocol: 16-byte header, typed frame/input enums
- Frame types: Pixels, Delta, Audio, Signal, Neural (0x01-0x43)
- Input types: Pointer, Key, Scroll, Gamepad, MIDI, BCI (0x01-0x90)
- WebSocket relay server (tokio + tungstenite)
- Source → receivers: frame broadcast
- Receivers → source: input routing
- Codec: encode/decode, XOR delta compression, RLE, convenience builders
- 17 unit tests, all passing
Streaming modes (stream-source.html):
1. Pixel mode: raw RGBA framebuffer (~28 MB/s)
2. Delta mode: XOR + RLE compression (~1-9 MB/s, 70-95% savings)
3. Signal mode: compact JSON signal diffs (~2 KB/s, 12000x reduction)
4. Neural mode: procedural SDF pixel generator (concept demo)
5. Audio channel: spring velocity→frequency synthesis
6. Multi-receiver: broadcast to all connected clients
Thin receiver client (stream-receiver.html, ~300 lines):
- Zero framework, zero build step
- Renders any incoming bitstream mode
- Local signal-diff renderer for signal mode
- AudioContext playback for audio frames
- Full input capture: click/drag, keyboard, scroll
- Per-channel bitstream bus visualization
DREAMSTACK.md: Phase 7 section with protocol spec
2026-02-25 10:29:44 -08:00
|
|
|
|
|
|
|
|
[dev-dependencies]
|