alint — clap-rs/clap

The canonical "well-curated Rust library workspace" — clap's entire workspace-metadata contract maps to ~12 alint rules.

Why this case study matters

clap is the canonical “well-curated Rust library workspace” — five published members (clap_builder, clap_derive, clap_complete, clap_mangen, clap_lex) plus the umbrella clap facade and clap_bench, all sharing one [workspace.package] block for license, edition, MSRV, repository, and include glob.

The structural surface is small but dense: ~30 distinct checks across CI workflows, pre-commit hooks, cargo-release config, and the workspace metadata itself. Roughly 65% maps directly to declarative alint rules, 5% shells out via command: to typos / cffconvert / cargo deny, and ~30% is deliberately out of alint’s scope (rustfmt, clippy, rustdoc, the Cargo-graph checks the Rust toolchain owns).

The fit is higher than rust-lang/rust because clap doesn’t carry its own custom AST-aware linter — the structural surface is exactly the “metadata pinning + manifest hygiene” layer alint targets.

Headline catch

clap’s entire workspace-metadata contract maps to ~12 alint rules. One Cargo.toml concentrates almost every structural fact about the workspace:

[workspace.package] license / edition / MSRV / repository / include
[workspace.lints.rust] + [workspace.lints.clippy] (~70 lints)
Per-member inheritance (license.workspace = true, edition.workspace = true, rust-version.workspace = true, repository.workspace = true, [lints] workspace = true)
Per-member crates.io discoverability fields (categories, keywords)
Per-member [package.metadata.docs.rs] rustdoc-args convention

Each of those 12 invariants becomes one TOML-path rule. Drop in the five bundled rulesets (oss-baseline, rust, monorepo/cargo-workspace, ci/github-actions, hygiene/no-tracked-artifacts) for the standard hygiene layer, and the 24-rule starter config replaces every structural assertion clap makes about its own workspace.

Where alint earns its keep here

clap is the complementary case study to kubernetes / rust-lang/rust: those are the “huge polyglot monorepo” stories. clap is the “small, disciplined, every-decision-deliberate” story. Together the three span the adoption spectrum.

Use it as the positive baseline for the launch positioning:

“clap is what a clean Rust library workspace looks like, and alint can describe its entire structural-validation surface in 24 lines of YAML. If your Rust workspace is messier than clap’s, alint is probably the cheapest way to surface that delta.”

The pitch is discoverability + onboarding: a contributor runs alint check once and gets the same structural verdict that 5 separate tools (committed, typos, cffconvert, cargo deny, pre-commit) plus the bundled rules cover today, in one declarative file with grep-able rule IDs.

clap also confirms two cross-file rule-kind candidates by demand: the per-crate metadata identity check (every clap_* member’s categories[0] resolves to the same value) is now absorbed into cross_file_value_equals (10 sources, past-saturation, v0.10 ship-target). The pre-release-replacements regex sanity gap is narrower (cargo-release-specific) and stays on the v0.10 design candidate list.

Future story angles

clap as the v0.10 “for_each_dir over {members}” reference config once the bundled monorepo/cargo-workspace selector reads the workspace members array directly — clap’s explicit select: "{clap_builder,clap_derive,...}" bracket expansion would collapse to select: "{members}" and survive future crate additions without manual edits.
The “small + disciplined” archetype — clap is the right reference repo for any contributor who wants to model a Rust library workspace’s structural contract end-to-end before scaling up.