BUSBuilding Understanding Standard
Conformance / Registry

Registry

Self-declared, publicly verifiable. Not a certification programme: no fees, no mark, no approval. The site renders what implementers submit and does not editorialise — and it renders failures in the same type as passes, deliberately.

busref 0.2.0

BUS v0.2 · 2026-08-04 · The BUS-1 editor (busref and busjs share this author)

Targets: package · importer · exporter · roundTrip   Profile: core-intent

53 pass · 0 fail · 2 not applicable · 1 reported

Failed tests

  • None reported.

Caveats

  • Group RT was additionally run as a cross-implementation chain — see crossImplementation — between busref (Python) and busjs (TypeScript). Both implementations share an author; an independent-party implementation remains the outstanding test.
  • The full profile is not claimed: timeSeries is deferred to BUS-2 and busref declares it absent.
  • EX-5 and EX-6 are product-surface tests and are reported not-applicable for a library.
  • Normalized from the published v0.2 statement (spec.bustandard.org/v0.2/conformance/busref-statement.json, vendored at src/vendor/busref-statement.json) to the registry statement schema: implementation.product renamed to name, result to outcome, detail to notes, not-applicable to notApplicable, the by-group results object flattened to one list, and the restated grades and summary counts dropped because the suite holds the grades and the registry recomputes the counts. No result was added, removed, or altered; tools/validate_registry.py re-derives this normalization from the vendored statement on every CI run and fails if the two diverge. The date is the v0.2 publication date. The run environment was not recorded by the v0.2 run and is not invented here.

Statement as submitted: busref-0.2.0.json

busjs 0.1.0

BUS v0.2 · 2026-07-30 · The BUS-1 editor (busref and busjs share this author)

Targets: package   Profile: core-intent

17 pass · 0 fail · 0 not applicable

Failed tests

  • None reported.

Caveats

  • busjs and busref share an author. Their agreement proves the specification is implementable twice; it does not prove it is interpretable independently. An independent-party implementation remains the outstanding test.
  • Derived from the cross-implementation artifacts, not from a full suite run by busjs itself. The only per-test evidence those artifacts hold is group PK: busref ran the full group, schemas included, on the package busjs exported in the A -> busref(py) -> B -> busjs(ts) -> C chain, and every check passed. The importer, exporter and roundTrip targets are therefore not claimed here, even though all thirteen documents were canonically identical through busjs six times over — that is chain-level evidence, and this registry does not convert chain-level evidence into per-test results.
  • The full profile is not claimed: timeSeries is deferred to BUS-2 under v0.2 and the exported package declares it absent. The chain also carried knowledge and presentation documents byte-identically, but only the core-intent profile is claimed, matching the judge's own statement.
  • The date is the file date of the conformance artifacts the entry is derived from. The run environment was not recorded in those artifacts and is not invented here.

Statement as submitted: busjs-0.1.0.json

Submitting a statement

  1. Run the suite for the targets you claim and write the results as a statement JSON validating against the published schema, https://spec.bustandard.org/v0.3/bus-conformance-statement.schema.json. A statement MUST answer every test of every group its targets name — fail with notes is a complete answer, and this page renders it in the same type as a pass.
  2. Open a pull request adding the file to the registry/ directory of the site repository. CI (tools/validate_registry.py) validates the shape against the schema, checks that every claimed target's test groups are answered in full against the suite of the busVersion you declare, and re-runs every claim it can check against this repository's published artifacts.
  3. It is published verbatim beside this page, failures included. A statement with failures and clear notes is more useful than a clean one with none.