Debt Paydown

docs/coercion-design.md

@@ -160,15 +160,15 @@ string interpolation, container syntax, indentation, lint/formatting rules).
 Type coercion is data-driven; the full backend is broader.
 
 **Transitional state:** This document describes the target architecture.
-The compiler is currently in a transitional identity model: checkpoint
-and inhabitant keys are strings (`dag_name`, `algebra`), not declaration
-edges. M4/Lane 1 is dissolving string identity into structural
-declaration-backed facts; M4/Lane 2 is deleting `Node.name`. The
-coercion algorithm described here is designed for the post-M4 world
-where identity flows from declarations, but the initial implementation
-(M5-early, via E0c `TypeRendering`) will use the transitional string
-keys until Lane 1 completes. The algorithm shape is the same; only the
-key resolution mechanism changes.
+The compiler is still partially transitional on some identity surfaces:
+checkpoint and inhabitant keys are strings (`dag_name`, `algebra`), not
+declaration edges. The v3 `Node.name` deletion is already landed; the
+remaining transition is Lane 1's string-keyed coercion identity. The
+coercion algorithm described here is designed for the post-transition
+world where identity flows from declarations, but the initial
+implementation (M5-early, via E0c `TypeRendering`) will keep the
+transitional string keys until Lane 1 completes. The algorithm shape is
+the same; only the key resolution mechanism changes.
 
 ## The Coercion Model
 

docs/perf/clone-elimination.md

@@ -244,9 +244,10 @@ still TRUE — it's just not urgent anymore. The session got
 the urgent perf win from fact-flow, not clones. Clone
 elimination remains valuable for:
 
-- **Node.name deletion (M2)** — String clones are real heap
-  allocations. Still worth eliminating. But no longer on the
-  critical path for perf.
+- **Node.name deletion (v3 code: landed)** — the generic v3
+  `Node.name` carrier is gone. Residual String clones now come from
+  other carriers (`Declaration.name`, `BindNode.name`) and any stale
+  doc references, not from an in-flight node-identity scaffold.
 - **Stable binding identity** (unblocks last-use clone elision,
   Stream B Layer 1) — worth doing for modeling correctness.
 - **The exclusion categories** (callables, TCO, match-bound,
@@ -287,7 +288,7 @@ during reviews. The tactical answer is:
 | `.clone()` sites | 13,724 | Unchanged — clones weren't the problem |
 | merge_envs bug | Fixed | 68× speedup on reconcile |
 | Skip data-only analysis | Landed | 82 of 143 files skip CX/ownership |
-| authored_name_at fallback | Mostly eliminated | ~8 parser sites remain |
+| authored_name_at fallback | v3 eliminated | Remaining mentions here are historical; active parser sites are v2-only |
 | InternTable threading | Landed | Unblocks registry migration |
 
 ## What's next
@@ -298,9 +299,10 @@ during reviews. The tactical answer is:
 
 1. **Audit for other re-derivation hotspots.** Apply Rules 2
    and 3 above. Find the next merge_envs.
-2. **M2 Node.name deletion continues** — for modeling, not
-   perf. String clones are worth eliminating but no longer
-   critical.
+2. **Keep clone work focused on live carriers** — `Node.name`
+   itself is no longer active v3 debt, so follow-on clone cleanup
+   should target remaining live String surfaces rather than a
+   completed node-identity migration.
 3. **M1 Step 3 — thread existing authorities, don't invent
    new ones.** `SubValueRelation` (std/induction.dag),
    `TerminationProof` (std/termination.dag), and

docs/phase-plan-2026-04-18.md

@@ -115,7 +115,7 @@ If Python forces a `CleanEmissionContract` SHAPE change (not just a new instanti
 
 ### Brief — Lens-name-filter dissolution (#518 follow-up)
 
-**Status:** dispatchable now (independent of in-flight items).
+**Status:** ✅ Cleared on PR #548.
 
 **Scope:** broader migration of `ArrowBody::Pending` writes from anonymous sites (`lower.rs:872` nested Arrow, `infer.rs:1884` variant constructor, `infer.rs:2807` operator fallback) to `NoBody` where appropriate. Once migrated, `lens_structural_resolution` predicate drops the `name: Some(_)` filter and becomes purely structural.
 
@@ -129,7 +129,7 @@ If Python forces a `CleanEmissionContract` SHAPE change (not just a new instanti
 - Site 5 (operator fallback) — explicit classification: NoBody if "no body by construction"; documented Pending-with-rationale otherwise
 - Lens predicate drops `name: Some(_)` filter where safe
 - New regression tests: anonymous arrows previously silent-via-name-filter now silent-via-NoBody
-- ROADMAP entry added and closed in the same PR
+- ROADMAP closure note landed in the same PR (`ROADMAP.md` notes the `#518` follow-up as cleared on PR #548)
 
 **STOP-AND-ESCALATE rule:** if site 5 forces a NEW substrate variant (neither Pending nor NoBody fits), escalate — don't introduce variants inside an anonymous dispatch.
 
@@ -250,7 +250,7 @@ No migration needed for any row above — §3 briefs or ROADMAP's existing stanc
 |---|---|---|
 | Mutual-recursion planner vs `is_first` alignment | #519 ChatGPT non-blocking | Lane 3 Stage 3a.1 follow-up — §3 combined XS brief lands this |
 | Variant-payload field-access general model | #518 (multi-field) + #519 chat A (single-field) | New "infra debt" subsection OR Lane 1 follow-ups — needs a dedicated hygiene chat |
-| Lens-name-filter dissolution (#518 sites 2/4/5) | #518 broader migration | §3 brief handles; ROADMAP row added when that PR opens |
+| Lens-name-filter dissolution (#518 sites 2/4/5) | #518 broader migration | ✅ Cleared on PR #548 |
 | compiler.dag v2-path carryover (§6 Q3) | `hand_maintained_src` references `src/v2/stage0/src`, `cli_run.rs`, `v2_interpreter.rs` — v2 paths in a doc that should eventually key off v3 source | Lane 3 Stage 3c prerequisite OR downgrade to "open question until 3c starts" — needs director pre-clearance before migration |
 
 **Dispatch decision:** §3 briefs cover 2 rows — the combined XS brief lands the planner alignment; the lens-name-filter brief lands that dissolution. The **variant-payload field-access general model** row needs a dedicated ROADMAP-hygiene chat (not folded into the XS brief — it won't get attention inside an XS dispatch). The **compiler.dag v2-path carryover** row needs director-chat pre-clearance first (is it a real 3c blocker, or a v2→v3 bridging detail that self-resolves?) before ROADMAP migration.

docs/single-emitter-design.md

@@ -225,7 +225,7 @@ assumption that the target is foreign. That assumption must not exist.
 |---|---|---|---|
 | LanguageSpec modeling | Track 2 | Presentation facts must be modeled as data before the emitter can read them | LS-4 (borrow model) in design |
 | Core table dissolution | Track 7 | String-keyed maps must become node-keyed before the emitter can use structural identity | Partially done |
-| Node.name deletion | Track 3 | String identity must dissolve before homomorphism discovery works | ~15 n.name reads remaining |
+| Node.name deletion | Track 3 | String identity must dissolve before homomorphism discovery works | Landed in v3 code; only stale docs remain |
 | CX gate | KF-1 | CX must be able to prove bounds on emission functions | 421 violations remaining (M1) |
 
 ### What can move without dependencies