For example, in NumPy, pyrefly reports
{
"kind": "attr",
"name": "numpy._core.multiarray.array",
"n_typable": 1,
"n_typed": 0,
"n_any": 0,
"n_untyped": 1,
"location": {
"line": 36,
"column": 49
}
},for the fully typed array function (so not an attr): https://github.com/numpy/numpy/blob/b832a09cf2a169c833dd2371e7c07aa00b293242/numpy/_core/multiarray.pyi#L458-L517
This seems to be related to the fact that this array function is not directly defined in the corresponding .py; it instead re-exports it (from a C extension module).
For example, in NumPy, pyrefly reports
{ "kind": "attr", "name": "numpy._core.multiarray.array", "n_typable": 1, "n_typed": 0, "n_any": 0, "n_untyped": 1, "location": { "line": 36, "column": 49 } },for the fully typed
arrayfunction (so not anattr): https://github.com/numpy/numpy/blob/b832a09cf2a169c833dd2371e7c07aa00b293242/numpy/_core/multiarray.pyi#L458-L517This seems to be related to the fact that this
arrayfunction is not directly defined in the corresponding.py; it instead re-exports it (from a C extension module).