v0.43.0 (#753) made the trap-preservation VC LIVE for 5 classes (i32 div/rem, unreachable, i32 load/store, i32.trunc_f32, call_indirect) — a guard-dropping lowering fails translation_validator. Two classes remain UNIT-gated (proven in trap_preservation.rs, not yet on the live verify_rule path):
- i64 div/rem — needs the exec-model
may_trap flag threaded through the i64 pair-lowering guard sequence.
- trunc_f64 — same, for the f64→int VFP path.
Both are honestly disclosed today (method docstring + the verify_trap_preservation catch-all names them). Closing them makes the trap gate live for the full set. Also: drop the trap.rs RemS workaround once ordeal#72 ships (their trap_div(RemS) over-adds the INT_MIN/-1 clause).
Refs: #166, #73, ordeal#72, #242 (VCR-VER-002).
v0.43.0 (#753) made the trap-preservation VC LIVE for 5 classes (i32 div/rem, unreachable, i32 load/store, i32.trunc_f32, call_indirect) — a guard-dropping lowering fails
translation_validator. Two classes remain UNIT-gated (proven intrap_preservation.rs, not yet on the liveverify_rulepath):may_trapflag threaded through the i64 pair-lowering guard sequence.Both are honestly disclosed today (method docstring + the
verify_trap_preservationcatch-all names them). Closing them makes the trap gate live for the full set. Also: drop thetrap.rsRemS workaround once ordeal#72 ships (theirtrap_div(RemS)over-adds the INT_MIN/-1 clause).Refs: #166, #73, ordeal#72, #242 (VCR-VER-002).