Skip to content

SOURCES.md

Latest commit

 

History

History
81 lines (72 loc) · 4.74 KB

SOURCES.md

File metadata and controls

81 lines (72 loc) · 4.74 KB
Raw

Vendored specification files

These files are unmodified copies of Credential Engine's published CTDL and CTDL-ASN schema and JSON-LD context documents. They are the only source of domain, range, inverse, and identifier-coercion rules in this tool. Nothing is encoded from memory.

All four were retrieved on 2026-08-06 with plain HTTP GET.

File Source URL SHA-256
ctdl/schema.json https://credreg.net/ctdl/schema/encoding/json a2dd28cb08f9e5e0324fe155e7381c276d70e69cb95521dab120f042cc538776
ctdl/context.json https://credreg.net/ctdl/schema/context/json ddb6b4586c38df5fa43aa5f6a558de407dd7acdec9bc1329ce790e164bf40db5
ctdlasn/schema.json https://credreg.net/ctdlasn/schema/encoding/json 8bacfcec140c03e144903acd04eeacc3f466c2245ca2a1bb703a322981fda0b7
ctdlasn/context.json https://credreg.net/ctdlasn/schema/context/json 783a5a5e132deb0023f6a4ea5201b9b69e298e6c359c1fd8d3c595392f8e7e72

Two prose sources are cited in findings but not vendored (they are HTML pages; the exact sentences relied on are quoted in ctdl_validate/rules.py):

  • "About the CTID", https://credreg.net/ctdl/ctid (page dated 5/10/2024, retrieved 2026-08-06). Defines the CTID grammar and the CTID-based URI structure.
  • CTDL "All Schemas Handbook", https://credreg.net/ctdl/handbook (retrieved 2026-08-06). Defines blank node identifier scope and documents that competency frameworks and their competencies are published in the same JSON-LD graph.
  • RDF Schema 1.1, W3C Recommendation 25 February 2014, section 3.1, https://www.w3.org/TR/rdf11-schema/#ch_resource (retrieved 2026-08-22). "All things described by RDF are called resources, and are instances of the class rdfs:Resource ... This is the class of everything." Relied on to treat a schema:rangeIncludes naming only rdfs:Resource as constraining nothing, rather than as a range no CTDL class can satisfy. This is the one source that makes a check emit fewer findings rather than more, so it cites no finding of its own; see schema.UNIVERSAL_RANGE_TERMS.
  • CTDL QData schema encoding, https://credreg.net/qdata/schema/encoding/json (retrieved 2026-08-22). Not vendored and not read by the validator: qdata: properties reach the tool through the CTDL encoding above, which declares them without declaring the qdata: classes they range on. Consulted once, during the hand verification behind docs/findings/2026-08-21-registry-survey-at-scale.md, to check whether qdata:relevantDataSet is declared for ceterms:AggregateDataProfile in its own home encoding. It is not: both encodings declare the same 61-class domain and both omit it.

The extract subcommand cites four further sources, also quoted in ctdl_validate/rules.py rather than vendored. Nothing about the CTDL rules depends on them; they define the formats a page publishes in and the manners a fetcher observes.

  • RFC 9309, "Robots Exclusion Protocol", https://www.rfc-editor.org/rfc/rfc9309 (retrieved 2026-08-14). Sections 2.2.1 (the User-Agent line and the product token), 2.3.1.1 through 2.3.1.4 (obeying parseable rules; five redirects; 4xx unavailable; 5xx complete disallow), and 2.5 (the 500 KiB parsing limit). Crawl-delay is not in RFC 9309; where a site declares one, this tool takes the larger of that and its own interval, which is recorded in the rule citation itself.
  • HTML Living Standard, "Microdata", https://html.spec.whatwg.org/multipage/microdata.html (retrieved 2026-08-14). Sections 5.2.3 (names) and 5.2.4 (values) are implemented case for case; 5.2.5 (itemref) is deliberately not implemented and is reported.
  • RDFa Lite 1.1 (Second Edition), W3C Recommendation 17 March 2015, https://www.w3.org/TR/rdfa-lite/ (retrieved 2026-08-14). Defines the five attributes the RDFa reader implements, and by exclusion the RDFa 1.1 Core attributes it does not.
  • schema.org JSON-LD context, https://schema.org/docs/jsonldcontext.json (retrieved 2026-08-14). Declares {"@vocab": "http://schema.org/"}, which is what makes a bare term under a schema.org context resolvable. The CTDL context declares the schema prefix as https://schema.org/; both forms are normalized to the CTDL one so the two published files agree.

The crosswalk the extractor uses is not an additional source. It is read out of the four vendored files above, from their own owl:equivalentClass, owl:equivalentProperty, rdfs:subClassOf, and rdfs:subPropertyOf declarations, and tests/test_extract_crosswalk.py checks the index against the files rather than against a copy.

To refresh the vendored files, re-download the four URLs above, update the hashes and dates here, and re-run the test suite. A schema release can change which findings this tool emits; that is by design.