Closure environement wrongly marked as cyclic (orc)

[arnetheduck]

Nim Version

2.2.4

Description

Same code as #25047

proc makeiter(v: sink string): iterator(): string =
  return iterator(): string =
    yield v

var iter = makeiter("test")()
for c in iter:
  echo "works ", c

Looking at the ORC code:

struct tyObject_Env_testitdotnim_makeiter___5Q3mvYA0XmpVYxu68uPX9bQ {
	RootObj Sup;
	NimStringV2 v0;
};
struct tyObject_Env_testitdotnim_colonanonymous___1qbA8qHJS9cP9cGQc9bC9b19cRQ {
	RootObj Sup;
	NI colonstate_;
	tyObject_Env_testitdotnim_makeiter___5Q3mvYA0XmpVYxu68uPX9bQ* colonup_;
};
...
N_LIB_PRIVATE N_NIMCALL(void, _ZN8makeiter10eqdestroy_E3varIN15colonanonymous_32Env_testitdotnim_colonanonymous_EE)(tyObject_Env_testitdotnim_colonanonymous___1qbA8qHJS9cP9cGQc9bC9b19cRQ* dest_p0) {
	{
		NIM_BOOL T3_;
		T3_ = (NIM_BOOL)0;
		T3_ = nimDecRefIsLastCyclicDyn((*dest_p0).colonup_);
		if (!T3_) goto LA4_;
		nimDestroyAndDispose((*dest_p0).colonup_);
	}
LA4_: ;
}
...

	tyObject_Env_testitdotnim_colonanonymous___1qbA8qHJS9cP9cGQc9bC9b19cRQ* colonenv__2;
....
	nimMarkCyclic(colonenv__2);

This particular environment does not contain any types that could be cyclic, suggesting that the cycle analysis is off - this would have a large impact on async/await that generates lots of these environments.

Current Output

Expected Output

Known Workarounds

No response

Additional Information

No response