In NumPy we use _Py_HashDouble for our floating point hashing needs and that seems like a pretty good candiate for the unstable API to me: #91744 (comment) and #101101 (comment).

However, the function is currently removed rather than made unstable (to my understanding). While it should probably be easy enough to vendor what we needed, I would prefer the unstable API tier, it gives a better sense of having the required identical hashing.

NumPy issue: numpy/numpy#25035

Originally reported by @ngoldbaum in #111418 (comment)

Linked PRs

• gh-111545: Add Include/cpython/pyhash.h header file #112063
• gh-111545: Add PyHash_Double() function #112095
• gh-111545: Add Py_HashPointer() function #112096
• gh-111545: Test PyHash_GetFuncDef() function #112098
• gh-111545: Add Py_HashDouble() function #112449
• [DO NOT MERGE][BENCHMARK] gh-111545: Benchmark PyHash_Double() #112476
• gh-111545: Mention PEP 456 in PyHash_GetFuncDef() doc #112647
• gh-111545: Add Py_HashDouble() function #113115