Last updated on

Announcing noir-metrics v0.2.0

noir-metrics v0.2.0 is now live on crates.io.

This release is less about adding flashy new metrics and more about making the existing ones reliable, testable, and stable enough to build tooling on top.

What changed in v0.2.0

1. Stabilization via mutation testing (cargo-mutants)

The core focus of v0.2.0 was stabilization.

noir-metrics is a line-based analyzer with heuristics (tests, comments, brace depth, etc.). That kind of logic is exactly where bugs hide: a single += turning into -= can silently skew results.

To harden this, development was guided heavily by cargo-mutants:

  • run mutation testing against the codebase,
  • identify missed mutants (places where tests didn’t detect wrong behavior),
  • add targeted tests (and snapshots) until remaining survivors were either fixed or proven equivalent,
  • iterate until all non-equivalent mutants are caught.

This gives much higher confidence that metrics output is not “accidentally correct” for a narrow set of fixtures.

2. Snapshot testing with insta

To prevent fragile “wall of asserts” tests (and to catch subtle regressions), v0.2.0 adds snapshot testing with insta for metrics structures and CLI JSON output.

Snapshots are especially useful here because they:

  • lock down the full structure of the report,
  • make changes explicit and reviewable,
  • reduce boilerplate when new fields get added later.

3. Deterministic output for stable tooling

To support snapshots and downstream tooling, v0.2.0 ensures deterministic ordering of discovered .nr files. This avoids “same project, different output ordering” issues across machines and CI runs.

4. Clearer CLI output format

The CLI now documents output format through a single flag:

  • --format human (default)
  • --format json

Backwards compatibility is preserved:

  • --json still works as a hidden alias for --format json, but --format json is the preferred interface going forward.

Installing and running v0.2.0

From crates.io:

cargo install noir-metrics

Run it in any directory containing a Nargo.toml:

# Human-readable output (default)
noir-metrics .

# JSON output to stdout
noir-metrics . --format json

# JSON output to a file
noir-metrics . --format json --output metrics.json

Why this release matters

Mutorium Labs’ longer-term goal is to bring coverage and mutation testing tooling to the Noir ecosystem.

For that to work, the ecosystem needs building blocks that are:

  • machine-readable (JSON output with schema versioning),
  • stable over time (deterministic output),
  • well-tested (including mutation testing),
  • and pleasant to integrate (CLI and library usage).

v0.2.0 is about getting those foundations right.

A note on heuristics (and what’s next)

noir-metrics still intentionally does not parse Noir’s full AST. It remains a fast, line-based analyzer with Noir-aware heuristics that are cheap enough to run in CI and useful enough for audits and tooling.

Future directions (not necessarily in this order):

  • control-flow and complexity indicators (if, loops, match counts),
  • more Noir-specific signals (constraint/assert density),
  • include/exclude configuration for real-world repos,
  • deeper integration ideas (eventual nargo metrics subcommand built on top of this crate).

References