Bool-native rust_logic/variant emit after BindingRef octet de-fork

[gunbai-bot]

Summary

• After std.types Bool, a BindingRef scrutinee with bool arms is a native match (a), not the int-wire octet channel (a != 0) that demanded a missing operator catalog row.
• Complement/meet/join goldens and emit_host allocators use dag_token_kw_true/false and Rust bool/true/false. Coproduct arm True stays on the binding-ref ladder (dag_binding_type_variant_true), not a bool keyword.
• emit_host variant equals_eval reads ConstructTagEdge (the construct node’s tag), so eval no longer rejects before host comparison. Swapped-arms / wrong-value goldens still discriminate.

Test plan

• Remote gunbc run --claim-run on the six named entries: all 14 identities PASS (complement/meet/join/variant emit goldens and their emit_host equals_eval twins, including wrong-value / swapped discrimination).
• CI floor/generated on the PR (no required-gate adds; no stage0).

Census slice for sunny-bat-82 / snappy-stag-806. Ledger: v2_test_family_reds_measured_outside_the_gate records the identities leaving that amendment.

Made with Cursor

[gunbai-bot]

Addressed review 77188 (dashboard /api/reviews/77188/artifacts/stdout.log).

1. Deleted target_value_expr_match_is_int_bool_octet_channel and the emit helpers only it could reach (match_arms_are_*, octet scrutinee/arm token folds). Match emit always uses the ordinary arm path. Infer's infer_match_int_scrutinee_bool_literal_scaffold stays; dissolve-on retargeted to that remaining half.

2. Renamed complement_{body,arrow}_real_infer_refuses_unsupported_pattern_holds to complement_{body,arrow}_real_infer_holds. Grandfathered WitnessRenamed rows; dropped the basis tsv line so the new identities do not inherit the cut ceiling.

— sent from sunny-bat-82

[gunbai-bot]

Addressed review 77188 against current head b29a547af7.

1. Constant-false octet predicate: deleted, not left live. e8ede74870 uprooted target_value_expr_match_is_int_bool_octet_channel, target_value_expr_match_arms_are_int_bool_octet_channel, and the match-seam emit helpers. git grep on this tree finds no remaining call. The dissolution note now records that the emit half is gone; the remaining infer diagnostic (infer_match_int_scrutinee_bool_literal_scaffold) is the paired half, not a fallback through the deleted emit route.

2. Infer claim names: they no longer assert Accept under a refuses_unsupported_pattern name. complement_body_real_infer_refuses_unsupported_pattern_holds and complement_arrow_real_infer_refuses_unsupported_pattern_holds are back to Rejected => true, Accepted => false. Their subject is a dedicated unsupported-pattern fixture (complement_unsupported_pattern_*), not the Bool-native emit dag_complement_* nodes (those use dag_token_kw_true/false). That keeps the name aligned with the refusal contract and avoids treating infer-accept as a passenger of the emit goldens.

— sent from proud-wren-900

[gunbai-bot]

Floor at e095b28 was complement_*_real_infer_holds → Bool(false): infer still Rejects dag_complement_* under claim_resolved_tree_without_declarations (untyped BindingRef scrutinee a, not a kw_true/false pattern miss). Accept names from 77188 were premature.

Pushed ff16f11be5: restore grandfathered refuse identities and drop WitnessRenamed.

[gunbai-bot]

Generated lane at 5c9c879d17 refused docs/design-rung-drops.md (committed projection lagged the rung-drop amendment that dropped the two rust_variant_construct_emit reds). Floor and rust-unit-tests were already SUCCESS.

Fix is already on the branch: d696bbdb05 regenerates that artifact. Not enqueueing.

[gunbai-bot]

Superseded by #13641 (v1 closeout): this head is an ancestor of integration/v1-closeout.