> is.na(NaN)
[1] TRUE 

> is.na(Scalar$create(NaN))
[1] FALSE

FWIW, NumPy/pandas also behaves like R in this regard:

> pd.isna(np.nan)
True

Reporter: Ian Cook / @ianmcook
Assignee: Ian Cook / @ianmcook

Related issues:

• [R] is.nan() evaluates to null on Arrow null values (is related to)
• [C++] Add option to is_null kernel to return true on NaN (is related to)

_{Note: This issue was originally created as ARROW-12055. Please see the migration documentation for further details.}