docs(physics): close Slice I evidence and roadmap

Record the accepted flat-authoritative collision cutover, production parsed-graph removal, exact rollbacks, zero-residency connected gates, and RDP measurement boundary. Reconcile architecture, milestones, roadmap, issue ledger, agent instructions, and the active Slice J handoff.

Validated by the existing zero-warning Release build, 8,413 passing Release tests with five pre-existing skips, 201 focused closeout tests, and both exact-binary connected routes.

Co-authored-by: Codex <codex@openai.com>
This commit is contained in:
Erik 2026-07-25 18:44:43 +02:00
parent 82f8d4f82e
commit afebbe3eca
11 changed files with 257 additions and 71 deletions

View file

@ -2,11 +2,11 @@
**Date:** 2026-07-24
**Status:** Slices AH are complete. Slice I6's flat-authoritative production
cutover is committed and awaits the final user collision/visual gate before
parsed-graph deletion. Slices FL were explicitly approved 2026-07-24. Slice
J0's dependency-boundary preparation is complete without moving gameplay
owners. J1 remains behind the I6/I7 user collision/visual closeout.
**Status:** Slices AI are complete. Flat-authoritative collision, zero-
allocation resolve scratch, parsed-graph removal, connected correctness/
lifetime acceptance, and Slice-I closeout landed 2026-07-25. Slices FL were
explicitly approved 2026-07-24. Slice J0's dependency-boundary preparation is
complete without moving gameplay owners; J1 is the active execution point.
**Scope:** Reconcile and sequence the existing Modern Pipeline (`MP`) and
Linux/headless (`LH`) tracks using the 2026-07-24 connected performance audit.
@ -1240,8 +1240,12 @@ validated immutable flat collision records: synthetic and installed-DAT
source identity, exact float bits, deterministic pre-order, and corruption
tripwires pass without changing production traversal. I3 now packages all
flat collision/topology records through a shared typed mmap source: two
complete 2,232,170-key bakes are byte-identical with zero failures. I4 flat
traversal shadow implementation is the current execution point.
complete 2,232,170-key bakes are byte-identical with zero failures. I4I6 then
proved exact flat traversal, strict dual publication, and flat-authoritative
gameplay before removing every production parsed collision graph. Slice I7
passes 8,413 Release tests / 5 skips plus both exact-binary connected routes;
every stable checkpoint reports zero retained parsed collision graphs. J1 is
now the current execution point.
Evidence:
[`../research/2026-07-25-slice-g5-production-profile.md`](../research/2026-07-25-slice-g5-production-profile.md)
and
@ -1251,6 +1255,8 @@ and
and
[`../research/2026-07-25-slice-i3-prepared-collision-package.md`](../research/2026-07-25-slice-i3-prepared-collision-package.md)
and
[`../research/2026-07-25-slice-i7-closeout.md`](../research/2026-07-25-slice-i7-closeout.md)
and
[`2026-07-25-modern-runtime-slice-i.md`](2026-07-25-modern-runtime-slice-i.md).
The intended order is therefore:
@ -1264,8 +1270,8 @@ honest metrics + committed baselines (A — exit criteria block C)
-> incremental render scene (F)
-> delta GPU submission (G)
-> residual frame cleanup (H-a, H-b, H-c)
-> flat collision assets (I — CURRENT at I4)
-> presentation-independent runtime (J — §0.3 authorization)
-> flat collision assets (I — COMPLETE)
-> presentation-independent runtime (J — CURRENT at J1)
-> Linux/headless/multi-session (K)
-> evidence-gated GPU jobs (L)
```