From source fragments to publications

Grapharch separates source collection, logical documentation structure, and publication layout. The shared documentation KG connects all three.

Collect documentable things

Each collector handles one kind of input. Markdown-file collectors retain prose and frontmatter; RDF-file collectors retain graph assertions; SPARQL-file collectors document queries without executing them. Specialized documentors can generate fragments for ontology classes and other RDF constructs.

A documentation tree also exposes topics, ordered text fragments, and code examples. A topic has an identity in the KG; its source filename is provenance, not its meaning.

Assemble and query the graph

Collection writes through a bounded channel to the graph store. Independent collectors run concurrently. Embedded Oxigraph writes are coordinated, and a failed input cannot silently replace the last complete contribution.

Assembly validates a project revision. Scopes select content, and publication profiles arrange components. A resolved target plan pins the content and ordering that a renderer will consume.

Publish for the reader

A documentation website supports exploration across topics. A standards document needs publication metadata, references, conformance sections, and stable anchors. A PDF needs pagination, a table of contents, and reviewable document structure.

Those outputs can use the same logical fragments while applying different arrangements. Grapharch's own website build is a working example of this separation.

Follow implementation progress

The architecture, design goals, and implementation status describe current behavior and work that remains, including external stores and incremental editor previews.