Fixes

doc/source/whatsnew/v0.24.0.rst

@@ -1326,7 +1326,6 @@ Deprecations
 - Passing an integer to :meth:`Series.fillna` and :meth:`DataFrame.fillna` with ``timedelta64[ns]`` dtypes is deprecated, will raise ``TypeError`` in a future version.  Use ``obj.fillna(pd.Timedelta(...))`` instead (:issue:`24694`)
 - ``Series.cat.categorical``, ``Series.cat.name`` and ``Sersies.cat.index`` have been deprecated. Use the attributes on ``Series.cat`` or ``Series`` directly. (:issue:`24751`).
 - Passing a dtype without a precision like ``np.dtype('datetime64')`` or ``timedelta64`` to :class:`Index`, :class:`DatetimeIndex` and :class:`TimedeltaIndex` is now deprecated. Use the nanosecond-precision dtype instead (:issue:`24753`).
-- The functions :func:`pandas.to_datetime` and :func:`pandas.to_timedelta` have deprecated the ``box`` keyword. Use :attr:`Series.values` and :meth:`Timestamp.to_datetime64`/:meth:`Timedelta.to_timedelta64` instead to get an ndarray of values or ``numpy.timestamp64``/``numpy.timedelta64``, respectively (:issue:`24416`).
 
 .. _whatsnew_0240.deprecations.datetimelike_int_ops:
 

doc/source/whatsnew/v0.25.0.rst

@@ -79,6 +79,7 @@ Deprecations
 ~~~~~~~~~~~~
 
 - Deprecated the `M (months)` and `Y (year)` `units` parameter of :func: `pandas.to_timedelta`, :func: `pandas.Timedelta` and :func: `pandas.TimedeltaIndex` (:issue:`16344`)
+- The functions :func:`pandas.to_datetime` and :func:`pandas.to_timedelta` have deprecated the ``box`` keyword. Use :attr:`Series.values` and :meth:`Timestamp.to_datetime64`/:meth:`Timedelta.to_timedelta64` instead to get an ndarray of values or ``numpy.timestamp64``/``numpy.timedelta64``, respectively (:issue:`24416`)
 
 .. _whatsnew_0250.prior_deprecations:
 

pandas/core/indexes/datetimelike.py

@@ -300,11 +300,7 @@ def asobject(self):
         return self.astype(object)
 
     def _convert_tolerance(self, tolerance, target):
-        tolerance = to_timedelta(tolerance)
-        if isinstance(tolerance, ABCIndexClass):
-            tolerance = tolerance.to_numpy()
-        else:
-            tolerance = np.asarray(tolerance.to_timedelta64())
+        tolerance = np.asarray(to_timedelta(tolerance).to_numpy())
 
         if target.size != tolerance.size and tolerance.size > 1:
             raise ValueError('list-like tolerance size must match '

pandas/tests/indexes/datetimes/test_tools.py

@@ -1215,11 +1215,15 @@ def test_to_datetime_types(self, cache):
     def test_to_datetime_unprocessable_input(self, cache, box, klass):
         # GH 4928
         # GH 21864
-        result = to_datetime([1, '1'], errors='ignore', cache=cache, box=box)
+        with tm.assert_produces_warning(FutureWarning):
+            result = to_datetime([1, '1'], errors='ignore', cache=cache,
+                                 box=box)
+
         expected = klass(np.array([1, '1'], dtype='O'))
         tm.assert_equal(result, expected)
         msg = "invalid string coercion to datetime"
-        with pytest.raises(TypeError, match=msg):
+        with pytest.raises(TypeError, match=msg), \
+             tm.assert_produces_warning(FutureWarning):
             to_datetime([1, '1'], errors='raise', cache=cache, box=box)
 
     def test_to_datetime_other_datetime64_units(self):

pandas/tests/scalar/timedelta/test_timedelta.py

@@ -310,12 +310,12 @@ def test_iso_conversion(self):
         assert to_timedelta('P0DT0H0M1S') == expected
 
     def test_nat_converters(self):
-        result = to_timedelta('nat', box=False)
-        assert result.dtype.kind == 'm'
+        result = to_timedelta('nat').to_numpy()
+        assert result.dtype.kind == 'M'
         assert result.astype('int64') == iNaT
 
-        result = to_timedelta('nan', box=False)
-        assert result.dtype.kind == 'm'
+        result = to_timedelta('nan').to_numpy()
+        assert result.dtype.kind == 'M'
         assert result.astype('int64') == iNaT
 
     @pytest.mark.filterwarnings("ignore:M and Y units are deprecated")