Build-Time Baseline (2026-05-08)

Build-Time Baseline (2026-05-08)

Phase 0 baseline for the workspace reorg. See 2026-05-08-workspace-reorg-design.md.

Measurement methodology

For incremental scenarios:

1. Touch the target file: touch <path>
2. Time the check: time cargo check -p <crate> --quiet
3. The reported real time is the recorded baseline.

For cold scenarios (captured opportunistically — full cargo clean runs take 30+ min on a 40GB target/, so we don’t run them as a routine measurement):

• A clean baseline run is captured at the start of each phase if the phase’s acceptance criterion would otherwise be unverifiable.
• Otherwise, incremental measurements are sufficient to detect 30%+ wins.

For per-crate compile time inside any build, append --timings (Cargo writes target/cargo-timings/cargo-timings-*.html).

Recorded baselines

Scenario	Command	Real time
L0 leaf check (cached)	cargo check -p vox-orchestrator-types	0.36s
L0 leaf check (cached)	cargo check -p vox-db-types	0.63s
Orchestrator incremental (touch lib.rs)	cargo check -p vox-orchestrator	5.59s
Orchestrator incremental (touch mcp_tools/dispatch.rs)	cargo check -p vox-orchestrator	5.06s
CLI incremental (touch vox-cli/src/lib.rs)	cargo check -p vox-cli	26.76s

Cached L0 leaf checks bottoming out near 0.5s confirm those crates are already near-floor — wins on them will be marginal. The real targets are the 5–27s incremental rebuilds.

Targets (Phase 9 acceptance)

Scenario	Today	Target
Orchestrator incremental after touching mcp_tools/ file	5.06s	≤ 1.5s (only vox-orchestrator-mcp rebuilds)
Orchestrator incremental after touching coordinator code	5.59s	≤ 3s (slimmed coordinator)
CLI incremental	26.76s	≤ 10s (vox-cli-thin path)
L0 leaf clean (will be measured per phase)	TBD	≤ 5s (post-hack-split, only vox-hack-core)

Build-time log

Each phase appends a row to build-time-log.md with post-phase measurements, comparing against this baseline.