gave binary and unary bit ops proper gradients instead of none (caused failure)

theano/scalar/basic.py

@@ -1090,7 +1090,7 @@ class UnaryBitOp(UnaryScalarOp):
         return upcast_out(*input_types[0])
 
     def grad(self, inputs, output_gradients):
-        return [None]
+        return [inputs[0].zeros_like().astype(theano.config.floatX)]
 
 
 class BinaryBitOp(BinaryScalarOp):
@@ -1103,7 +1103,8 @@ class BinaryBitOp(BinaryScalarOp):
         return upcast_out(*input_types[0])
 
     def grad(self, inputs, output_gradients):
-        return [None, None]
+        a,b = inputs
+        return [a.zeros_like().astype(theano.config.floatX), b.zeros_like().astype(theano.config.floatX)]
 
 
 class OR(BinaryBitOp):

theano/tensor/elemwise.py

@@ -743,8 +743,6 @@ class Elemwise(Op):
             scalar_inputs = map(as_scalar, inputs)
             scalar_ograds = map(as_scalar, ograds)
             scalar_igrads = self.scalar_op.grad(scalar_inputs, scalar_ograds)
-            for igrad in scalar_igrads:
-                assert igrad is not None
 
         finally: