From 45742b0f0ecdb9a371955bddb633be2496e22bc9 Mon Sep 17 00:00:00 2001 From: Claude Date: Fri, 21 Aug 2026 10:12:30 +0000 Subject: [PATCH] docs(r2il): supersede the "no V4" verdict per the operator ruling The V3-vs-V4 verdict in the plan ("V3 is sufficient. No V4") and its restatement in the section 22.5 stop condition are superseded by the operator ruling of 2026-08-21: V4 is minted as an additive sibling tier for R2IL's 100%-coverage requirement and other special needs, with V3 continuing unchanged for everything it already carries. Both sites get a supersession block rather than an edit, per the append-only convention. What is retracted is the conclusion, not the evidence: no V3 mechanism was falsified, and the mint is not a repair of a V3 defect. The blocks also name what the ruling leaves open (layout, home crate, the full special-needs set, the routing rule, and whether the 0xC4 mint lands as V3 or V4) so a later session does not read the ruling as settling more than it did. --- .claude/plans/r2il-behavioral-ir-v1.md | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/.claude/plans/r2il-behavioral-ir-v1.md b/.claude/plans/r2il-behavioral-ir-v1.md index 27c22c679164c5..249f1b97e0f9b8 100644 --- a/.claude/plans/r2il-behavioral-ir-v1.md +++ b/.claude/plans/r2il-behavioral-ir-v1.md @@ -112,6 +112,23 @@ 1. **Verdict: V3 is sufficient. No V4. The hypothesis stands un-falsified** pending the corpus measurements (which gate only the per-route LAYOUT choices, not the physical grammar). + + > **⊘ SUPERSEDED — OPERATOR RULING, 2026-08-21.** *"V4 ist für R2IL 100% + > coverage und andere Special needs / V3 wie bisher."* V4 IS minted, as an + > **additive sibling tier**, not a replacement: V3 continues unchanged for + > everything it already carries; V4 is the tier for R2IL's 100%-coverage + > requirement and other special needs. What survives from the line above is + > its *evidence*: no V3 mechanism was falsified, and the mint is not a + > repair of a V3 defect. What is retracted is the *conclusion* — "no V4" was + > a session verdict, and a session verdict does not outrank an operator + > ruling. Board entry: lance-graph `.claude/board/EPIPHANIES.md` + > `E-V4-IS-THE-100-PERCENT-TIER-V3-UNCHANGED-1`. Open under the ruling and + > NOT decided here: V4's layout, its home crate, the full special-needs set, + > the V3-vs-V4 routing rule, and whether the `0xC4` mint (O5/PR 3) lands as + > V3 or V4. Corroboration that the label was already in circulation: + > lance-graph `docs/architecture/ARC-B-OWNERSHIP-AND-ADDRESSING-REASSESSMENT.md` + > §7 already writes the external path as `binary → Ghidra/SLEIGH/P-code → + > r2sleigh → V4 R2IL` (2026-08-19). 1. **Honesty note**: network.rs's "byte-parity vs real Tesseract" is designed-for (oracle named) but in-repo tests are synthetic round-trips against pre-registered values. R2IL routes should meet the same standard @@ -395,6 +412,9 @@ direct consumption already succeeds without it. - §22.1: direct r2il/r2ssa consumption solves the upstream seam — YES (43 s). - §22.4: V3 represents R2IL via routes+overflow — YES per audit; corpus gates layout only. - §22.5: no V4; variable arity already routed upstream (SsaGraph descriptor shape). + **⊘ SUPERSEDED 2026-08-21 (operator ruling):** the "no V4" half is retracted — + see the ⊘ block under the §22.5 verdict above. The "variable arity already + routed upstream" half is unaffected and stands. ## Open items