🎨 implement diff within extensionarray

Author: Marco Gorelli

doc/source/whatsnew/v1.0.0.rst

@@ -558,7 +558,7 @@ Default dtype of empty :class:`pandas.Series`
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 Initialising an empty :class:`pandas.Series` without specifying a dtype will raise a `DeprecationWarning` now
-(:issue:`17261`). The default dtype will change from ``float64`` to ``object`` in future releases so that it is
+(:issue:`17261`). The default dtype will change from ``float64`` to ``Object`` in future releases so that it is
 consistent with the behaviour of :class:`DataFrame` and :class:`Index`.
 
 *pandas 1.0.0*

pandas/core/arrays/base.py

@@ -24,7 +24,7 @@
 from pandas.core.dtypes.missing import isna
 
 from pandas.core import ops
-from pandas.core.algorithms import _factorize_array, unique
+from pandas.core.algorithms import _factorize_array, diff, unique
 from pandas.core.missing import backfill_1d, pad_1d
 from pandas.core.sorting import nargsort
 
@@ -516,6 +516,9 @@ def argsort(
         result = nargsort(self, kind=kind, ascending=ascending, na_position="last")
         return result
 
+    def diff(self, n_periods):
+        return type(self)._from_sequence(diff(self, n_periods), dtype=self.dtype)
+
     def fillna(self, value=None, method=None, limit=None):
         """
         Fill NA/NaN values using the specified method.

pandas/core/series.py

@@ -2316,13 +2316,11 @@ def diff(self, periods=1) -> "Series":
         5    NaN
         dtype: float64
         """
-        result = algorithms.diff(com.values_from_object(self), periods)
         if is_extension_array_dtype(self.dtype) and not is_datetime64tz_dtype(
             self.dtype
         ):
-            return self._constructor(
-                result, index=self.index, dtype=self.dtype
-            ).__finalize__(self)
+            return self.values.diff(n_periods=periods)
+        result = algorithms.diff(com.values_from_object(self), periods)
         return self._constructor(result, index=self.index).__finalize__(self)
 
     def autocorr(self, lag=1) -> float: