This is a bit more appropriate, because the internal data type is not specialized into StaticTuple objects only. Partially because I didn't see a specific memory/speed tradeoff to caching the hash, and that accessing said hash was siginficantly faster than just calling PyObject_Hash().