Files
rustytorch/crates/production/rtx-streaming/Cargo.toml
T
osobhandClaude Fable 5 c83e0fb22d
CI / Format Check (push) Failing after 7s
CI / Clippy Check (push) Failing after 7s
CI / Build (ubuntu-latest) (push) Failing after 7s
Performance Benchmarks / Run Benchmarks (push) Failing after 9s
CI / Build (macos-latest) (push) Failing after 11s
CI / Test (macos-latest) (push) Has been skipped
CI / Test (ubuntu-latest) (push) Has been skipped
CI / Python Bindings (maturin) (macos-latest) (push) Has been skipped
CI / Python Bindings (maturin) (ubuntu-latest) (push) Has been skipped
CI / WASM Build + Size Check (push) Has been skipped
CI / Distributed Training Tests (push) Has been skipped
CI / Build CPU-Only (Explicit) (push) Failing after 11s
CI / CI Success (push) Failing after 1s
Documentation / Build API Documentation (push) Failing after 6s
Documentation / Build User Guide (push) Successful in 7s
fix(streaming): wire worker control planes for real graceful shutdown
Follow-up to a0bf294, which tolerated dead control channels; this makes
them functional:

- AdaptiveProcessor: mpsc control channel (single consumer behind a
  mutex, broke on ANY message including Start) replaced with broadcast;
  all three workers (resource monitor, batch optimizer, pressure
  monitor) subscribe and exit only on ControlCommand::Stop
- EdgeComputingManager / MonitoringSystem: their 7 interval-loop workers
  now subscribe to the existing broadcast control channels and exit on
  Stop instead of looping forever
- stop() in all three: graceful join with 5s timeout, abort only as a
  backstop (previously unconditional abort mid-tick)
- benches: criterion needs async_tokio for Bencher::to_async — bench
  target now compiles (clippy --all-targets clean)

cargo test -p rtx-streaming: 55 lib + 8 integration + 6 aux, all green.

Co-Authored-By: Claude Fable 5 <[email protected]>
2026-07-10 17:04:35 -07:00

98 lines
2.0 KiB
TOML

[package]
name = "rtx-streaming"
version = "1.0.0"
edition = "2024"
rust-version = "1.92"
description = "Production-grade real-time model streaming with sub-millisecond latency"
license = "MIT OR Apache-2.0"
[dependencies]
# Core async runtime
tokio = { version = "1.40", features = ["full"] }
tokio-tungstenite = "0.23"
futures = "0.3"
async-trait = "0.1"
# WebSocket and networking
tungstenite = "0.23"
http = "1.1"
bytes = "1.8"
# gRPC support
tonic = "0.13"
prost = "0.13"
# High-performance data structures
dashmap = "6.1"
crossbeam = "0.8"
parking_lot = "0.12"
# Serialization
serde = { version = "1.0", features = ["derive"] }
serde_json = "1.0"
bincode = "1.3"
msgpack = { package = "rmp-serde", version = "1.3" }
# Metrics and monitoring
prometheus = { version = "0.13", optional = true }
tracing = "0.1"
tracing-subscriber = "0.3"
# Message queue support
rdkafka = { version = "0.36", features = ["tokio"] }
# ML inference integration — token generation delegates to the real engine
rtx-inference = { path = "../rtx-inference" }
# Additional async utilities
tokio-stream = "0.1"
async-stream = "0.3"
futures-util = "0.3"
pin-project-lite = "0.2"
# Additional concurrency
crossbeam-channel = "0.5"
flume = "0.11"
rayon = "1.8"
# Time handling
chrono = { version = "0.4", features = ["serde"] }
# Enhanced monitoring (simplified - commented for now)
# metrics = "0.23"
# Circuit breaker (implementing our own for now)
# circuit-breaker = "0.5"
# System utilities
once_cell = "1.19"
arc-swap = "1.7"
num_cpus = "1.16"
libc = "0.2"
# Utilities
uuid = { version = "1.10", features = ["v4", "serde"] }
thiserror = "2"
anyhow = "1.0"
rand = "0.8"
# Development dependencies
[dev-dependencies]
criterion = { version = "0.5", features = ["html_reports", "async_tokio"] }
tokio-test = "0.4"
proptest = "1.5"
approx = "0.5"
rtx-tensor = { path = "../../core/rtx-tensor" }
[[bench]]
name = "streaming_bench"
harness = false
[features]
default = ["websocket", "grpc"]
websocket = []
grpc = []
metrics = ["prometheus"]
[lints]
workspace = true