VUSL is a platform-neutral specification and Rust implementation for reviewing how meaning carries between native artifacts. A mapping review identifies the exact source and target snapshots, records what was preserved or lost, and cites evidence for the relationship. The Rust core inspects reviews offline.
Status: Experimental native v0.1, maintained by Hypler.
vusl-core 0.1.0-dev.1
and vusl-coverage 0.1.0-dev.1
are available on crates.io as prereleases. Coverage pins the exact core version.
The implementation roadmap tracks stable 1.0 work
and consumer qualification.
| Component | What it provides |
|---|---|
| Native specification | A bounded mapping-review format with scoped native references, term-level fidelity and loss, and evidence-qualified relationships. |
| Rust core | Offline inspection with typed results, canonical bytes, a stable document ID, and an assessment of declared required meaning. |
| Authoring schema | A JSON Schema 2020-12 structural check for authoring. The native inspector determines conformance. |
| Independent coverage | Checks whether a review includes every source term required by a caller-selected policy and keeps those terms marked as required. |
| Reference consumer | Redacted CLI diagnostics, versioned semantic results, and offline examples for saved revisions, exact mapping, and multi-edge review. |
Historical RUSL/D0a artifacts are retained for provenance. The native VUSL contract and its conformance suite follow the accepted architectural direction.
Use Rust 1.99.1 or newer and pin the experimental version in Cargo.toml.
[dependencies]
vusl-core = "=0.1.0-dev.1"Call vusl_core::inspect with the bytes of a native mapping review. The Rust
API and native profile are experimental and may change before 1.0. See the
crate README for an inspection example.
To apply a caller-selected required-term policy to a checked review, pin both crates to the same prerelease.
[dependencies]
vusl-core = "=0.1.0-dev.1"
vusl-coverage = "=0.1.0-dev.1"Inspect the mapping review using vusl_core::inspect, then pass the checked
review and your policy to vusl_coverage::evaluate. See the
coverage crate README for the policy inputs.
Select Rust 1.99.1 and provision the example workspace's locked dependencies using the validation guide. Run from the repository root.
cargo run --locked --offline \
--manifest-path examples/reference-consumer/Cargo.toml -- \
--scenario saved-preview-preserved --format textThis example checks whether a target preview preserves the terms required for a specific saved revision. The companion cases cover missing or downgraded terms, stale targets, and unresolved relationships.
To check your own inputs, use the strict inspection example with native reviews and a policy selected by your application.
The opt-in exact mapping API checks caller-selected source-to-target term pairs, exact endpoint revisions, and the expected evidence references for each pair. Its versioned JSON result identifies each failed obligation and distinguishes mismatches, missing or unexpected citations, declared loss, and uncertainty. Library lending and irrigation fixtures exercise the same rules. Native v0.1 review inputs and existing consumer APIs remain compatible. Callers can also require an exact evidence basis and citation set for the relationship itself. Strict reports also surface producer-required terms and disclosed loss outside host scope.
The consumer result guide also covers exact review comparisons that distinguish different citations from conflicting claims. Chain diagnostics locate missing required term links between edges that share an exact snapshot. Whole-chain preservation remains unresolved until an explicit composition rule establishes it. Callers own evidence trust, live freshness, target-loaded readback, receipts, reconciliation, and authorization. The offline route comparison shows when strict checks change a synthetic host’s review route. The saved-report checker can verify strict reports offline against the host's selected policy and the supplied native reviews. Its comparison evidence view shows both claims, exact citation overlap, and differences between cited revisions of the same record.
With producer scope coverage selected, paired review also identifies conflicting claims about required terms outside the host inventory and retains each producer's declarations and citations for follow-up. It also lists citation differences for those terms, including when the declarations agree. Producer scope disagreements distinguish omitted terms from present optional claims.
The offline explanation example joins producer notes to checked document identities and selected terms. It distinguishes missing claims from skipped checks while retaining the semantic result unchanged. Its paired mode separates literal explanation differences from typed claim conflicts and preserves each review's attribution.
A machine-readable help catalog provides fixed descriptions for strict result findings and caller follow-up tokens. CI checks it against the human documentation and result schema. The explanation example's opt-in help mode joins those descriptions to exact finding locations while preserving the semantic result.
The reference consumer also provides a conservative draft helper from an independently selected exact policy. Every required term starts unknown, with no asserted citations and an unresolved relationship. The draft keeps semantic preservation unresolved.
A runnable authoring walkthrough shows twenty synthetic cases, including omitted requirements, foreign evidence and native refusal. Native conformance and host-policy outcomes stay separate.
- Documentation index for the specification, implementation guides, and project records.
- Validation guide for toolchain setup and reproducible checks.
- Contributing, governance, security, and change history.
Licensed under Apache-2.0.
