Document Type.may_share_memory and implement it in Scalar.

doc/extending/type.txt

@@ -138,6 +138,13 @@ default values.
         :return: the number of bytes taken by the object described by
             ``shape_info``.
 
+    .. method:: may_share_memory(a, b)
+
+        Optional. Only needed for DebugMode. Return True if the python
+        objects `a` and `b` could share memory. Return False
+        otherwise. It is used to debug when Ops didn't declare memory
+        aliaing between variables. Must be a static method.
+
 For each method, the *default* is what ``Type`` defines
 for you. So, if you create an instance of ``Type`` or an
 instance of a subclass of ``Type``, you

theano/scalar/basic.py

@@ -145,6 +145,13 @@ class Scalar(Type):
         self.dtype = dtype
         self.dtype_specs()  # error checking
 
+    @staticmethod
+    def may_share_memory(a, b):
+        # This class represent basic c type, represented in python
+        # with numpy.scalar. They are read only. So from python, they
+        # can never share memory.
+        return False
+
     def filter(self, data, strict=False, allow_downcast=None):
         py_type = self.dtype_specs()[0]
         if strict and not isinstance(data, py_type):