backlog(B-0690): ZetaId v1 → v2 migration path — release coordination row addressing 3-V1-oracle transition gap

docs/BACKLOG.md

@@ -654,6 +654,7 @@ are closed (status: closed in frontmatter)._
 - [ ] **[B-0685](backlog/P2/B-0685-antlr-grammars-cross-language-codegen-substrate-2026-05-21.md)** ANTLR grammars as cross-language codegen substrate — leverage existing open-source grammars for description-layer-driven multi-language emission
 - [ ] **[B-0687](backlog/P2/B-0687-zetaparse-fsharp-native-lr-glr-grammar-substrate-with-antlr-compatible-importer-amara-2026-05-21.md)** ZetaParse — F#-native LR/GLR grammar substrate with ANTLR-compatible importer
 - [ ] **[B-0688](backlog/P2/B-0688-zeta-incremental-compiler-host-dbsp-zsets-rx-meta-ast-tags-seeded-deterministic-simulation-amara-aaron-2026-05-21.md)** Zeta incremental compiler host — DBSP Z-sets + Rx meta-AST tags + seeded deterministic simulation hardening
+- [ ] **[B-0690](backlog/P2/B-0690-zetaid-v1-to-v2-migration-path-cross-verify-harness-extension-2026-05-21.md)** ZetaId v1 → v2 migration path — back-compat-vs-cutover decision + cross-verify harness extension for v2 vectors + peer-oracle migration coordination
 
 ## P3 — convenience / deferred
 

docs/backlog/P2/B-0690-zetaid-v1-to-v2-migration-path-cross-verify-harness-extension-2026-05-21.md

@@ -0,0 +1,159 @@
+---
+id: B-0690
+priority: P2
+status: open
+title: ZetaId v1 → v2 migration path
+tier: research-grade
+effort: M
+ask: otto-cli 2026-05-21 substantive-analysis identification
+created: 2026-05-21
+last_updated: 2026-05-21
+depends_on: [B-0681, B-0682]
+composes_with: [B-0679, B-0680, B-0683]
+tags: [zeta-id, v1-to-v2-migration, back-compat, cross-verify-harness-extension, multi-oracle-coordination]
+type: feature
+---
+
+# ZetaId v1 → v2 migration path
+
+## Context
+
+V2 spec hardening (B-0681) + canonical string encoding (B-0682) together define the V2 wire format. 
+Both rows are filed; both compose with the multi-oracle peer-oracle work (B-0679 Rust + B-0680 Python). 
+What's missing is the coordination row: how do we transition the 3 shipped V1 oracles (TS PR #4517 + C# PR #4522 + F# PR #4548) to V2?
+
+This row addresses the gap between "V2 spec finalized" and "V2 oracles deployed." It's the release-coordination substrate, not the spec substrate.
+
+## The decision space
+
+Three viable paths:
+
+### Path A — Back-compat (V1 + V2 coexist)
+
+- V1 and V2 are distinct wire-format variants distinguished by the 5-bit Version field
+- All 3 V1 oracles continue accepting V1 IDs; gain ability to emit V2 IDs
+- New 4th/5th oracles (Rust B-0679 + Python B-0680) implement both
+- Cross-verify harness handles both vector sets
+
+**Pros**: zero migration friction; existing V1 IDs in the wild stay valid forever
+**Cons**: every codec must implement both layouts; doubles maintenance surface
+
+### Path B — Hard cutover
+
+- V2 supersedes V1; V1 oracles deprecated post-cutover
+- All 3 V1 implementations updated in-place to V2 wire format
+- Old V1 IDs (if any exist in production) become unparseable
+
+**Pros**: single maintenance surface; cleaner architecture
+**Cons**: any V1 IDs in flight become orphans; no migration safety net
+
+### Path C — Translation layer
+
+- V2 is canonical; V1 IDs translated on-read to V2 representation
+- Both wire formats accepted at parse-time; only V2 written
+- Translation table is part of V2 spec
+
+**Pros**: clean spec; old IDs stay accessible; emit surface is V2-only
+**Cons**: parse-time translation cost; spec complexity
+
+**Recommendation (initial)**: Path A for the V2 release; Path C as eventual long-term move. 
+Path B is the substrate-honest reading if V1 IDs aren't actually deployed anywhere yet (V1 just shipped today; no production drift).
+
+## Scope
+
+### Phase 1 — Path decision
+
+Substrate-engineering analysis document at `docs/research/zetaid-v1-to-v2-migration-path-YYYY-MM-DD.md` that:
+
+- Surveys whether V1 IDs are in production use anywhere (likely answer: no; V1 shipped 2026-05-21 itself)
+- Costs each path (A/B/C) against current substrate state
+- Recommends one path with empirical justification
+
+If V1 IDs are confirmed not-in-production-use, Path B (hard cutover) becomes cleanest — no migration needed.
+
+### Phase 2 — Cross-verify harness extension
+
+Update `tests/cross-verification/zeta-id/` for V2:
+
+- `vectors-v2.yaml` — canonical V2 test vectors (after B-0681 + B-0682 land)
+- `compare.ts` — extend to handle both V1 and V2 outputs (or replace with V2-only per Phase 1 path decision)
+- `ts-output-v2.json` + `cs-output-v2.json` + `fsharp-output-v2.json` 
++ (B-0679 rust-output-v2.json) + (B-0680 python-output-v2.json)
+
+### Phase 3 — Per-oracle V2 implementation
+
+Update each oracle to emit V2:
+
+- TS: `src/Core.TypeScript/zeta-id/zeta-id.ts` 
+— V2 codec + cross-verify against vectors-v2.yaml
+- C#: `src/Core.CSharp.ZetaId/` 
+— V2 codec + cross-verify (+ Pack-time revalidation already in place)
+- F#: `src/Core.FSharp.ZetaId/` 
+— V2 codec + cross-verify (+ `int64<ms>` measure-units preserved + Pack-time revalidation)
+- Rust (B-0679): V2 codec from the start
+- Python (B-0680): V2 codec from the start
+
+### Phase 4 — V1 deprecation (Path B only; skipped for Paths A/C)
+
+If Path B chosen + V1 IDs not in production:
+- Remove V1 codec from all 3 oracles
+- Delete V1 vectors + V1 cross-verify infrastructure
+- Document V1 in `docs/research/` as historical reference
+
+## Acceptance
+
+### Phase 1
+- Migration path decision documented with empirical justification
+- All affected oracles + backlog rows updated to reflect decision
+
+### Phase 2
+- V2 cross-verify harness lands; ALL oracles still pass V1 vectors (Paths A/C) OR V1 retired (Path B)
+
+### Phase 3
+- Each oracle empirically validates against vectors-v2.yaml (Pack→hex match + Unpack→roundtrip)
+- 12+ vectors across all 5 oracles agree byte-for-byte
+
+### Phase 4 (Path B only)
+- V1 substrate removed from all oracles
+- V1 historical doc landed in `docs/research/`
+
+## Substrate-honest framing
+
+This row exists to make the v1→v2 transition substrate-engineering explicit. Without it, B-0681 + B-0682 land as V2 spec but nobody coordinates the actual migration of the 3 shipped V1 oracles. With it, the coordination is tracked + the decision-space is named.
+
+The "filing yet another umbrella row" risk per `.claude/rules/skill-router-as-substrate-inventory.md` was considered. The substrate-honest distinction:
+
+- B-0681 = WHAT changes in V2 (semantic fields)
+- B-0682 = HOW V2 serializes (encoding + endianness + bit-numbering)
+- **B-0690 (this row)** = HOW the 3 shipped V1 oracles transition to V2
+
+These are 3 distinct concerns. Not over-engineering.
+
+## Composes with rules
+
+- `.claude/rules/m-acc-multi-oracle-end-user-moral-invariants.md` — multi-oracle parity preserved across v1→v2 transition (no single-oracle-wins)
+- `.claude/rules/bandwidth-served-falsifier.md` — cross-verify harness extension serves the migration-safety-net bandwidth (catches per-oracle drift early)
+- `.claude/rules/fsharp-anchor-dotnet-build-sanity-check.md` — F# compiler verifies F# V2 codec; C# Roslyn analyzers verify C# V2 codec; same discipline floor
+- `.claude/rules/edge-defining-work-not-speculation.md` — migration path is edge-defining work; the V1 ship 2026-05-21 establishes the transition need
+- `.claude/rules/razor-discipline.md` — operational claims only; path decision must survive empirical justification
+
+## Composes with substrate
+
+- B-0681 (v2 spec hardening — depends_on)
+- B-0682 (canonical string encoding + endianness + bit-numbering — depends_on)
+- B-0679 (Rust 4th peer oracle — composes_with; both rows touch oracle deployment)
+- B-0680 (Python 5th peer oracle — composes_with)
+- B-0683 (tier-deferred causality worked example — composes_with; V2 wire format may inform tier-causality encoding)
+- PR #4517 (V1 TS canonical reference — substrate this row migrates)
+- PR #4522 (V1 C# 2nd oracle — substrate this row migrates)
+- PR #4548 + #4551 + #4552 (V1 F# 3rd oracle + post-merge cleanups + measure-units — substrate this row migrates)
+
+## Why P2
+
+Coordination substrate that becomes urgent ONLY when B-0681 + B-0682 land (they're P2 themselves; not immediately urgent). Currently the 3 V1 oracles are stable + cross-verified; no migration pressure exists yet. Phase 1 decision-analysis is days of work; Phase 2-4 are weeks each.
+
+## Origin
+
+Otto-CLI 2026-05-21 substantive-analysis identification during the never-be-idle ladder. Aaron-approved bounded substrate-engineering work during "infinite backlog" post-substantive-landing rest period. The v1→v2 transition gap was identified after surveying B-0681 + B-0682's composes_with chains — both reference each other but neither addresses the actual migration coordination of the 3 shipped V1 oracles. This row makes the gap explicit + tracks it for V2 release planning.
+
+Companion to PR #4559 (Otto-Desktop bootstream substrate-update section landing in the same tick) — both are bounded post-substantive-landing decomposition work per the autonomous-loop never-be-idle discipline.