Cargo Workspace Split

Status: Deferred — Phase 3 (trigger not yet met, but architecture is ready)

When to do this

jackin today is ~91K lines of production Rust (116 files in src/) — still below the ~150K threshold where a multi-crate Cargo workspace outweighs its overhead, but growing fast (was ~50K at original analysis — 82% growth). This item is not ready to execute. The trigger conditions are:

• LOC exceeds ~150K, or
• A sub-component needs external consumers for third-party agent manifest tooling, or
• Compile times on the CI check job exceed 5 minutes cold cache

Caution

The original claim “console/ has zero imports from runtime/ in production code” is FALSE. The re-analysis found 2 production-code dependencies: crate::runtime::register_agent_repo and crate::runtime::RepoError in console/manager/input/editor.rs. The Cargo workspace split will need these 2 sites addressed first — either through a trait abstraction or by moving the shared function to a common module.

Target 6-crate structure

Follows matklad’s virtual manifest + flat crates/ pattern:

jackin/
├── Cargo.toml             ← virtual workspace manifest (no [package])
├── crates/
│   ├── jackin-core/       ← Tier 0: workspace types, manifest, selector, paths, docker trait
│   ├── jackin-config/     ← Tier 1: TOML persistence (AppConfig, ConfigEditor)
│   ├── jackin-tui/        ← Tier 1: terminal output, animation, prompts
│   ├── jackin-runtime/    ← Tier 2: container bootstrap pipeline
│   ├── jackin-console/    ← Tier 3: workspace manager TUI + operator_env
│   └── jackin-shell/      ← ShellRunner (concrete subprocess impl)
├── src/                   ← thin binary (CLI dispatch)
└── validate/              ← jackin-validate binary

Dependency tiers (verified)

• Tier 0 (no cross-module deps): workspace/, manifest/, docker.rs, paths.rs, selector.rs
• Tier 1: config/, tui/, instance/, env_resolver/
• Tier 2: operator_env/, runtime/, repo/
• Tier 3: console/ — almost no import from runtime/ (2 call sites need fixing first)

Prerequisites before this phase

1. Phase 2 file splits — establish clean internal module boundaries
2. Phase 1.5 DRY extractions — reduce duplication that would otherwise be frozen into crate APIs
3. Fix the 2 console→runtime dependencies — register_agent_repo and RepoError must move to a shared module or through a trait

Related files

• Cargo.toml — target: convert to virtual workspace manifest
• src/console/manager/input/editor.rs — contains the 2 runtime dependencies that must be resolved
• src/runtime/launch.rs — the runtime entry point (7575L)