@@ 1103,6 1103,14 @@ def test_infer_freq(engine, tsh):
     )
 
 
+def test_empty_update(engine, tsh):
+    ts = genserie(datetime(2010, 1, 1), 'D', 11)
+    d1 = tsh.update(engine, ts, 'empty-update', 'Babar')
+    assert len(d1)
+    d2 = tsh.update(engine, ts, 'empty-update', 'Babar')
+    assert not len(d2)
+
+
 def test_point_deletion(engine, tsh):
     ts_begin = genserie(datetime(2010, 1, 1), 'D', 11)
     ts_begin.iloc[-1] = np.nan

          

@@ 119,10 119,12 @@ class mainsource:
         the new version.
 
         A series made of the changed points is returned.  If there was
-        no change, None is returned and no new version is created.
+        no change, an empty series is returned and no new version is
+        created.
 
-        New points are added, changed points are changed,
-        points with NaN are _erased_.
+        New points are added, changed points are changed, points with
+        NaN are dropped if `keepnans` is False (by default) or
+        _erased_ if True.
 
         The `author` is mandatory.
         The `metadata` dictionary allows to associate any metadata

          
@@ 131,10 133,6 @@ class mainsource:
         It is possible to force an `insertion_date`, which can only be
         higher than the previous `insertion_date`.
 
-        The `keepnans` flag will treat, if true, the Nans values as
-        erasures. If false, we do a dropna() call on the series before
-        the actual update.
-
         .. highlight:: python
         .. code-block:: python