ReSpec compatibility and migration

ReSpec is an explicit reference for Grapharch's specification features and migration behavior. Its documentation, source modules, tests, and published outputs provide acceptance examples. Grapharch does not require ReSpec at runtime.

The source review on 2026-10-24 used upstream version 37.4.1, revision 4b35fa2b43891577773e50968e700e6145d12baf. The upstream license notice is retained with this reference inventory. No upstream JavaScript is bundled or executed by the Grapharch renderer. Retain source attribution and applicable notices when adding reference fixtures or incorporating upstream material.

Publication authority and rendering engine are separate choices

A target selects a renderer and a profile. ga:publisher selects a reusable ga:Authority: name, short name, logo, homepage, and accent color. W3C, EKGF, OMG, EDMA, and other authorities use the same mechanism. An authority change must not require a renderer fork or rewriting document bodies. Actual status and process text remain explicit graph content, independent of visual branding.

A publisher-neutral specification style and faithful W3C output are both required use cases. W3C output retains the selected W3C authority; it is not forcibly rebranded. Existing Grapharch CSS is an initial interpretation, not yet an exact W3C output profile.

Feature map

Statuses describe tested Grapharch behavior, not a claim of ReSpec equivalence. All upstream source links below are pinned to the reviewed revision.

ReSpec feature/reference	Grapharch equivalent	Current evidence and gaps
`w3c/headers.js`: title, subtitle, versions, status, date, editors, logos, other links	Target metadata plus `ga:Authority`	Authored and published-document fixtures render these fields. Branding replacement is tested. Editor/author/former-editor roles, ordered metadata rows, disclosure details, ORCID links, contributor notes, and explicit person ordering are supported; full header-link coverage, labels, and all status combinations remain incomplete.
`core/structure.js`, `tests/spec/core/structure-spec.js`	Profile placements, numbered outline, ToC, native aliases	Nested headings, appendices, abstract/status front matter, and native anchor resolution are tested. `maxTocLevel`, `noTOC`, unusual introductory sections, and complete numbering parity remain open.
`core/dfn.js`, `core/link-to-dfn.js`, `core/dfn-panel.js`	Identified terms and graph references	Published `dfn` resources and their native markup are collected. Alias/plural resolution, automatic linking, cross-spec indexes, and definition panels are not implemented.
`core/biblio.js`, `core/render-biblio.js`	Bibliographic resources and reference components	Published reference lists are preserved, including native anchors. ReSpec citation syntax, `localBiblio`, bibliography lookup, and automatic normative/informative classification are not implemented.
`core/examples.js`, `core/issues-notes.js`, `core/figures.js`	Example/figure/annotation resources and fragments	Published examples are queryable and displayed; RDF script examples become escaped source text. Complete labels, figure assets, issue/note boxes, captions, and indexes remain open.
`core/data-include.js`, `core/markdown.js`	Specialized collectors, source contributions, normalized fragments	Independent Markdown files work. ReSpec include/transformation syntax and dependency reconciliation are not implemented.
`w3c/conformance.js`, `w3c/group.js`, `w3c/linter-rules/`	Authority/process metadata, constraints, conformance components	Authored status and requirement text can be published. W3C policy templates, group lookup, legal boilerplate generation, and publication-rule parity are not implemented.
`core/webidl.js`, `core/cddl.js`, syntax highlighting	Specialized construct collectors/documentors and code fragments	Plain source blocks are supported. WebIDL/CDDL semantics and HTML syntax-highlighting parity are not implemented.
`core/exporter.js`, `tools/respec2html.js`	Static artifacts from pinned graph plans	HTML/CSS and embedded Typst/PDF publishing work. Native published HTML is converted into Typst for paired specification PDFs. Both outputs use the same plan; a PDF download icon and diagnostic addenda are supported.
`respecConfig`, `preProcess`, `postProcess`, `data-transform`	Declarative migration adapters and explicit conversion plugins	Published HTML can be collected without executing JavaScript. Import of authoring-time JavaScript configuration or hooks is not implemented. Unsupported executable hooks must be reported, not silently dropped or evaluated.

Real-standard acceptance fixtures

The primary fixture is the published RDF 1.2 N-Triples Working Draft of 23 October 2026. examples/w3c-rdf12-n-triples/source.html is a pinned reference input. Its digest, source URL, and attribution are recorded beside it. Integration tests run offline against that fixture. They collect content into the shared KG, assemble a target, export and restore the graph, delete the source, and render from the restored KG. They check publication metadata, section/term/example/reference resources, native anchors, and representative grammar, tables, and source blocks.

The specialized published-document collector is an initial migration path and a way to populate behavioral tests. The RDF 1.2 source contains a broken #must-signal-errors reference. Fail-fast rendering rejects it; explicit keep-going rendering records a KG diagnostic, includes an HTML/PDF addendum, and the CLI exits unsuccessfully after producing the preview. It does not substitute for importing a ReSpec authoring repository. Source markup is retained as provenance; rendered sections are independently identified fragments, not an opaque copy of the original page.

Exact-output acceptance contract

Exact W3C compatibility must be established against a named, versioned output profile and pinned inputs. A passing content test is insufficient. Acceptance requires:

Equivalent content, section numbering, document order, normative/informative distinctions, definitions, references, examples, tables, and code.
Preserved public anchors and working internal/external links, with deterministic identity and explicit reports for unresolved references.
Equivalent header metadata, authority assets, status/process text, and notices.
Pinned CSS/fonts/assets and browser/viewport settings, with reviewed visual comparisons for desktop, narrow-screen, and print layouts.
Equivalent supported interactions and accessibility behavior, with every deliberate difference documented.
Reproduction after source removal and dataset restore, with no implicit fetches of mutable upstream content during rendering.

Neither byte-identical HTML nor pixel-identical W3C output is claimed today. Golden visual comparisons, complete style/asset capture, and authoring-config migration are remaining acceptance work. The DPROD repository also references an EKGF ReSpec build (fork HEAD observed as b35286e4e5bd3b0499dda41cb83ba751f0753e7c on 2026-10-24); its fork-specific behavior needs a separate audit before a DPROD replacement can be called compatible.

The examples/dprod/ showcase instead collects native ontology and SHACL RDF and a Markdown worked example from a pinned DPROD commit. It generates 53 term fragments and publishes HTML and PDF from the resulting KG. This tests native DPROD inputs; it does not establish compatibility with the EKGF ReSpec fork or replace the complete DPROD specification.