Explanation
Background: how zdocs works and why it is built the way it is, for readers who want to contribute to zdocs rather than just using it.
Start with Architecture for the shape of the engine as a whole. The pages after it take one mechanism each and follow it end to end. The guidelines pages state the norms this project holds itself to, and Decisions is the historical record: every architectural decision, dated, with the situation that forced it.
- Architecture
- The registry, and what it derives
- The deploy tree
- Documents zdocs does not build
- From annotated C to a test report
- Engineering guidelines
- Documentation guidelines
- Decisions
- 0001. zdocs is a Zephyr module, included by name
- 0002. Acceptance tests live in a consumer repository, over one cumulative fixture
- 0003. The engine carries its own unit suite, in the engine repository
- 0004. The registry is the single source of truth for document declarations
- 0005. Remote documents are three kinds, not one
- 0006. The deploy tree is organised by builder, not by document
- 0007. Doxygen XML is engine-managed, and lives outside the servable tree
- 0008. The Doxygen tag file is named
doxygen.tag - 0009. Need types and links are a consumer-supplied role→name mapping
- 0010. The engine payload is split by tool, freeing
doc/for documentation - 0011. Documentation structure: Diátaxis, arc42-lite, and two documents
- 0012. Cross-document links in the HTML tree are relative