Publication stack¶

Publication is a controlled projection of approved semantics, not a free-standing writing step.

Chain of custody¶

Enterprise publication chain from semantic source through projection, renderer and visual QA to approval

26.10.4 makes the chain profile-native:

L1 intended audience / decision / authorised use
  ↓
L2 semantic material available for publication
  ↓
L3 inclusion, omission, evidence and release rules
  ↓
L4 publication/document design
  ↓
L5 renderer and output implementation
  ↓
L6 rendered artefact + hashes + visual evidence

ProjectionManifest v3¶

The manifest binds the publication to both JAI lineage and content identity. It carries the relationship between:

intent;
semantic snapshot;
policy;
design;
realisation;
evidence;
semantic/projection/render hashes.

The production renderer validates the matching manifest before emitting the supported current-release artefact.

Why the projection layer matters¶

A single semantic source may be used to create an executive one-pager, ARB pack and technical design. Those documents should not all contain the same detail, but each must be able to demonstrate that mandatory content for its audience/use was included and that omissions were permitted rather than accidental.

Publication archetypes¶

The release ships thirteen first-class archetypes, including:

Executive One-Pager;
Decision Paper;
Architecture Review Board Pack;
Architecture Definition;
Solution Design / HLD;
Detailed Design / LLD;
Risk & Assurance Pack;
Implementation Specification;
Migration & Cutover Pack;
Operational Handover;
Roadmap;
Benefits Review;
Audit / Compliance Evidence Pack.

Output formats¶

The publication stack supports first-class enterprise outputs such as HTML, DOCX, PPTX and XLSX, with PDF/SVG/EA interchange used where the relevant renderer/profile applies.

For SABSA documentation, Markdown is the default. Explicit LaTeX source and DOCX requests use Pandoc; compiling the LaTeX PDF additionally needs XeLaTeX, TeX packages and fonts. Office rendering for visual QA is a separate optional dependency from DOCX generation. See sandbox and environment prerequisites for the dependency/access matrix and deployment notes for controlled build qualification.

Semantic, projection and render identity¶

Do not collapse the three hashes:

semantic_hash   = what the approved semantic source was
projection_hash = what this audience/use was supposed to receive
render_hash     = the exact emitted file

The separation is essential for detecting both content omission and post-render file changes.

Visual QA¶

Render-based visual QA checks the visible result after rendering. It complements, rather than replaces, semantic review and ProjectionManifest coverage checks.