Question about behavior of `ReturnToVTPool()` and `Reset()`

Using this proto:

message Parent {
  option (vtproto.mempool) = true;
  repeated Child children = 1;
  Child one = 2;
}

message Child {
  option (vtproto.mempool) = true;
  uint32 field = 1;
}

When calling ReturnToVTPool() on Parent it calls ResetVT on all children and then calls m.Reset()

func (m *Parent) ResetVT() {
	for _, mm := range m.Children {
		mm.ResetVT()
	}
	m.One.ReturnToVTPool()
	m.Reset()
}

However m.Reset() allocates a new object and overwrites the existing object entirely:

func (x *Parent) Reset() {
	*x = Parent{}

This nils out all fields on the parent throwing away the slice for the GC to handle. Am I missing something? Is there some way to put back into the pool, call ResetVT() but not call Reset()?

Start by tracing the generated Parent.ResetVT and ReturnToVTPool methods for the proto shown, then compare their behavior with Parent.Reset and the child pooling calls. Determine the intended reset and slice-retention semantics, and define completion as an agreed behavior backed by a regression test or documented explanation.