Skip to content

language-ecosystem.md

Latest commit

 

History

History
98 lines (79 loc) · 5.19 KB

language-ecosystem.md

File metadata and controls

98 lines (79 loc) · 5.19 KB
Raw

AIOS Language Ecosystem: Technical Deep Dive

Parent document: architecture.md, development-plan.md Related: agents.md — Agent SDK and runtime adapters, posix.md — C/C++ toolchain and POSIX layer, browser.md — Browser Kit (QuickJS/WASM in reference browser) Scope: Rust, Python, TypeScript, WASM — how each runs on AIOS, what's needed, when it arrives

1. Overview

AIOS supports four first-class development languages through a unified capability model. Every language runtime enforces the same security boundaries — the language is an implementation detail, the capability set is the security boundary.

flowchart TD
    Manifest["`Agent Manifest (manifest.toml)
*Declares: runtime, capabilities, schedule*`"]

    subgraph Adapter["RuntimeAdapter trait"]
        Rust["`Rust
*NativeRuntime
native ELF*`"]
        Python["`Python
*PythonRuntime
RustPython embedded*`"]
        TS["`TypeScript
*TypeScriptRuntime
QuickJS-ng embedded*`"]
        WASM["`WASM
*WasmRuntime
wasmtime AOT-compiled*`"]
    end

    CapSys["`AIOS Capability System (8 layers)
*Intent, Capability, Behavior, Zone, ...*`"]
    Kernel["`AIOS Kernel
*31 syscalls, IPC*`"]

    Manifest --> Adapter --> CapSys --> Kernel
Loading

When Each Language Arrives

Language Introduced Tooling Complete Self-Hosting on AIOS
Rust Phase 0 (kernel) Phase 17 (SDK) Phase 23+ (needs rustc + LLVM)
Python Phase 17 Phase 17 Phase 17 (RustPython ships with OS)
TypeScript Phase 17 Phase 17 Phase 17 (QuickJS-ng ships with OS)
WASM Phase 17 (agents) Phase 17 + 30 (browser) N/A (compile on host, deploy .wasm)
C/C++ Phase 23 Phase 23f Phase 23f (clang builds on AIOS)
Linux binaries Phase 36 Phase 36 Whatever runs on Linux

Document Map

Document Sections Content
This file §1 Overview, language arrival timeline, document navigation
runtimes.md §2-§5 Per-runtime deep dives: Rust (native), Python (RustPython), TypeScript (QuickJS-ng), WASM (wasmtime)
integration.md §6-§8 Dependency chain, implementation work list, key architectural decisions, RuntimeAdapter trait
operations.md §9-§12 Runtime interoperability (WIT/Component Model), observability & debugging, supply chain security, resource isolation
ai.md §13-§14 AI-driven runtime optimization (AIRS Runtime Advisor, learned scheduling, lifetime allocation, GC tuning, anomaly detection), future directions

Implementation Order

Phase Deliverable Languages Affected
13 Agent framework + Rust SDK Rust
16 Multi-language SDK + developer experience Rust, Python, TypeScript, WASM
21 Performance optimization + AI-driven tuning All runtimes
22 POSIX + BSD userland + C/C++ toolchain C/C++, CPython, Node.js, Rust (self-hosting)
30 Browser (Servo + SpiderMonkey) WASM (browser path), JavaScript
35 Linux binary compatibility All Linux-compatible languages

Cross-Reference Index

Section Sub-file Topic
§1 This file Overview and language timeline
§2 runtimes Rust — native ELF agents
§3 runtimes Python — RustPython interpreter
§4 runtimes TypeScript — QuickJS-ng engine
§5 runtimes WebAssembly — wasmtime + WAMR alternative
§6 integration Dependency chain and phase timeline
§7 integration Per-language implementation work
§8 integration Language selection, embedded vs system runtimes, QuickJS-ng vs Boa
§9 operations WASM Component Model, WIT, cross-runtime communication
§10 operations Logging integration, source maps, profiling
§11 operations Dependency integrity, curated registry, interpreter-in-WASM
§12 operations Per-runtime resource budgets, memory pressure response
§13 ai AIRS Runtime Advisor, learned scheduling (ALPS), lifetime allocation (LLAMA), GC tuning, anomaly detection, capability minimization
§14 ai Boa, StarlingMonkey, WIT manifests, MSWasm, AI code migration, pipeline fusion