BUG: Raise when casting infinity to int by dsaxton · Pull Request #28475 · pandas-dev/pandas (original) (raw)

Raises a ValueError when attempting to cast a Float64Index with infinite values to a (non-ExtensionArray) integer dtype. Currently this returns a garbage value:

idx                                                               
# Float64Index([1.0, 2.0, inf], dtype='float64')
idx.astype(int)                                                   
# Int64Index([1, 2, -9223372036854775808], dtype='int64')