BUG: don't try to deal with a dead cache referant (GH5216)

doc/source/release.rst

@@ -372,7 +372,7 @@ See :ref:`Internal Refactoring<whatsnew_0130.refactoring>`
  - Internal type checking is now done via a suite of generated classes, allowing ``isinstance(value, klass)``
    without having to directly import the klass, courtesy of @jtratner
  - Bug in Series update where the parent frame is not updating its cache based on
-   changes (:issue:`4080`) or types (:issue:`3217`), fillna (:issue:`3386`)
+   changes (:issue:`4080`, :issue:`5216`) or types (:issue:`3217`), fillna (:issue:`3386`)
  - Indexing with dtype conversions fixed (:issue:`4463`, :issue:`4204`)
  - Refactor ``Series.reindex`` to core/generic.py (:issue:`4604`, :issue:`4618`), allow ``method=`` in reindexing
    on a Series to work

pandas/core/generic.py

@@ -943,7 +943,13 @@ def _maybe_update_cacher(self, clear=False):
             if clear, then clear our cache """
         cacher = getattr(self,'_cacher',None)
         if cacher is not None:
-            cacher[1]()._maybe_cache_changed(cacher[0],self)
+            try:
+                cacher[1]()._maybe_cache_changed(cacher[0],self)
+            except:
+
+                # our referant is dead
+                del self._cacher
+
         if clear:
             self._clear_item_cache()
 

pandas/tests/test_indexing.py

@@ -1557,6 +1557,24 @@ def test_cache_updating(self):
         self.assert_("A+1" in panel.ix[0].columns)
         self.assert_("A+1" in panel.ix[1].columns)
 
+        # 5216
+        # make sure that we don't try to set a dead cache
+        a = np.random.rand(10, 3)
+        df = DataFrame(a, columns=['x', 'y', 'z'])
+        tuples = [(i, j) for i in range(5) for j in range(2)]
+        index = MultiIndex.from_tuples(tuples)
+        df.index = index
+
+        # setting via chained assignment
+        df.loc[0]['z'].iloc[0] = 1.
+        result = df.loc[(0,0),'z']
+        self.assert_(result == 1)
+
+        # correct setting
+        df.loc[(0,0),'z'] = 2
+        result = df.loc[(0,0),'z']
+        self.assert_(result == 2)
+
     def test_floating_index_doc_example(self):
 
         index = Index([1.5, 2, 3, 4.5, 5])