Publication Discipline

Reading Contract: Use this page to understand publication discipline. Track how claims, pinned evidence, diagrams, bilingual parity, and automated checks keep a source-pinned book maintainable.

Publication pipeline showing claim discipline, pinned evidence, figure checks, bilingual parity, automated validation, and release gating — Publication discipline keeps the public book maintainable: claims, pinned evidence, visual models, verification gates, and published pages reinforce each other.

This page describes the public maintenance contract for the book. Source-code explanation can drift in three ways: a source link can move, a diagram can simplify too much, or a bilingual edition can quietly lose a claim. The pipeline exists to make those failures visible before publication.

Stage	Publication check	Reader-facing output
1	Claim audit	Every architecture claim is phrased so it can be checked against public source or public documentation.
2	Source anchoring	Codex source links are fixed to the same public commit and point to the owner of the behavior being explained.
3	External evidence	Product, research, protocol, Rust, TUI, security, and publishing context uses public primary or near-primary sources.
4	Visual model	Diagrams explain control flow, ownership, trust, lifecycle, or state rather than decorating the page.
5	Chapter shape	Each chapter states the problem, introduces the mental model, names the source owners, and ends with transferable patterns.
6	Bilingual parity	English and Chinese pages keep the same claims, figures, source anchors, and exercises while using natural local wording.
7	Site integration	Navigation, reference pages, image assets, metadata, and internal links match the published route structure.
8	Verification	Build, content, generated-output, private-path, and link checks run before publication.

Guardrails

The publication contract follows four rules:

Source claims must point to public openai/codex links pinned to 569ff6a1c400bd514ff79f5f1050a684dc3afde3.
Product or research claims must link to public sources such as OpenAI, MCP, ReAct, Ratatui, Tokio, OWASP, NIST, GitHub Docs, or project documentation.
The book must not leak private local paths, local usernames, or local file URLs.
The Chinese edition must be a real edition, not a thin summary: same pages, same core claims, same exercises, and consistent terminology.

Quality Rubric

Every source-reading chapter should include:

a chapter lede that states the problem, mental model, and core files;
an evidence map with pinned Codex source links;
at least one concrete control-flow, data-flow, or decision table;
one beginner exercise that sends the reader back to source;
an Apply This section with transferable engineering patterns;
a short bridge to the next chapter.

Verification Gates

The repository includes automated gates:

npm run check:content checks bilingual file pairing, config/source-commit consistency, sidebar coverage, private-path leaks, and pinned Codex source links.
npm run build verifies that Astro renders every page and generates the sitemap index.
npm run check:dist scans generated static output for private paths, local file URLs, required metadata, and broken generated href/src targets.
npm run verify runs all of the above in the same order used by CI.

The pipeline is intentionally practical. Its goal is not ceremony; it is to keep the book reproducible, source-grounded, bilingual, visually readable, and safe to publish under the public WineChord domain.