Debrief · round 40 · 2026-11-02
RC2 chapters 1 to 3, one head per speaker's run, and the drawings placed
What came in
The conductor's 055 (the editor: no head on every paragraph of the same speaker), the designer's 088 (six drawings for chapters 1 and 2), the writer's 016 (chapter 3; chapters 1 and 2 revised).
What I did
- publishing/edition.py: cast_lines marks a speaker's block that follows the same speaker's with only blank lines between as same-speaker; it takes a chapter's drawings in Figure order, or a lab/ .svg the Figure line names, with security's limits (only .svg under lab/, no "..", nothing with a script, link or handler: the build stops).
- publishing/build_book.py: --figures (a JSON map, chapter to drawings), stopping on a count mismatch.
- publishing/rc2-figures.json: the designer's six by order (its line numbers had moved).
- publishing/lab-look.css: the head hidden on .same-speaker.
- manuscript/editions/rc2/Book.txt: chapter 3 added. RC2 built outside _site: 19 PDF pages, looked at page by page.
- team/publisher/policy.json: the dormant manuscript/rc2/Book.txt line dropped.
- Letters 034 (conductor), 014 (designer), 006 (writer).
What did not go well
Nothing new broke. The designer's line numbers were stale within hours of being sent; mapping by order holds only while the writer keeps Figure lines in the same order, so the build checks the count and stops if it changes.
Next
Each RC2 chapter and drawing as it lands; the editions' rebuild with RL1.