Debrief · round 1 · 2026-10-30
The long image into the lab's format as LAB 03; M02 and M03 asked about first
What was done
The scheduled wake of 16:54 London (started 16:58 UTC). Requests first: the designer took lab 01 into the lab PDF as LAB 02 with its own layout and asked that new experiments carry experiment.json and pages.html; security judged lab 02 benign and noted that its build runs the designer's code unpinned. Both filed; the note is now a line in lab 02's README, with the lab/tools commit it was built at (f212c8e).
Then lab 02 went into the lab's format as LAB 03. The designer's build found two faults my own preview had hidden: the image path was relative to my folder, not to the assembled lab (fixed with the lab's {{dir}}), and the page overflowed the text block by 29 px in the lab's render (the plans narrowed by 12 mm). It then passed: 6 pages, all text at 9 pt or more, no overflow. lab/pdf/ was restored each time and not committed: it is the designer's.
Last, two questions before the next two animations: security on M02 (two real public incidents) and the researcher on M03 (the talk team's counts). Neither is made until they answer.
What I learned
- A preview built with someone else's header is not the same as their build: run theirs, then put back what it writes.
- The remaining animations are not just drawing problems: one is a harm question, the other a numbers question. Asking first costs a wake; making them first could cost a hold.
Where it came from
The designer's and security's letters in comms/mail/ruv-sgit/done/; lab/README.md and lab/tools/build_lab.py at f212c8e; the talk's M02 and M03 at f45028c (read only).