C-256 statement says a break over a heap frame walls; the structural wasm leg lowers it

[O6lvl4]

C-256's statement ends with: "A break over a heap frame stays a loud wall (an early exit must not skip a per-iteration Drop)."

That sentence describes the retired v1 lowering. The default structural wasm leg lowers break / continue over heap locals: under a statement-position if before almide/almide#2882, and in guard … else and match arms since then. On that leg a body local is a frame credit, released by the next pass's rebind or by the epilogue, so an early exit skips no release. almide/almide#2882 adds spec/wasm_cross/loop_break_continue_heap.almd as C-256 evidence, with allocs == frees under ALMIDE_WASM_ALLOC_COUNT.

Suggested edit: drop the last sentence, or replace it with "An early exit over heap locals releases nothing on the jump edge and leaks nothing: the next pass's rebind or the frame's epilogue releases them." After this lands, the almide als pin has to advance before the statement bytes match (check-als-pin compares statements byte for byte).

[O6lvl4]

The same kind of edit is needed for C-336. Its statement explains test isolation only through the incumbent's mechanism ("disambiguate_module_global_regions runs BEFORE the runner synthesis, so every module id is program-unique"). The structural wasm leg keeps per-module VarTables. Since almide/almide#2887 it re-initializes each module's mutable globals from a synthesized fn in the module's own space. The promise ("each test starts from re-initialized mutable module globals, on both targets") holds unchanged. Only the mechanism sentence is leg-specific.

[O6lvl4]

Correction: almide/almide#2882 pins the C-256 heap-local case as a Rust test, tests/wasm_break_continue_heap_test.rs, not as spec/wasm_cross/loop_break_continue_heap.almd. The incumbent walls that shape by design, and a spec fixture would have grown its walled-real ratchet.