Use convert_to_tensor to do type/shape check instead of make_tensor_p…

…roto,

as convert_to_tensor handles a Tensor in addition to other Python values.
Change: 119078171

tensorflow/python/kernel_tests/concat_op_test.py

@@ -139,18 +139,23 @@ def testRandom(self):
   def testInvalidConcatDimTypeAndShape(self):
     a = tf.Variable(tf.constant(1.0, shape=[1]))
     b = tf.Variable(tf.constant(2.0, shape=[1]))
-    with self.assertRaises(TypeError):
+    with self.assertRaises(ValueError):
       tf.concat(a, b)
     with self.assertRaises(TypeError):
       tf.concat(4.2, 1)
-    with self.assertRaises(TypeError):
+    with self.assertRaises(ValueError):
       tf.concat(a, 1)
     with self.assertRaises(TypeError):
       tf.concat(a, [a, b])
     with self.assertRaises(ValueError):
       tf.concat([3], [a, b])
     with self.assertRaises(ValueError):
       tf.concat(0, [])
+    # An integer tensor for shape dim should throw no error.
+    tf.concat(tf.constant(0, shape=[]), 1)
+    # A non-scalar tensor for shape should throw ValueError.
+    with self.assertRaises(ValueError):
+      tf.concat(tf.constant(0, shape=[1]), 1)
 
   def _testGradientsSimple(self, use_gpu):
     with self.test_session(use_gpu=use_gpu):

tensorflow/python/ops/array_ops.py

@@ -319,12 +319,16 @@ def concat(concat_dim, values, name="concat"):
     values = [values]
   # TODO(mrry): Change to return values?
   if len(values) == 1:  # Degenerate case of one tensor.
-    # Make a throwaway call to make_tensor_proto to make sure
-    # that concat_dim is of the correct type.
-    # TODO(keveman): Extract the type and shape checks out of make_tensor_proto
-    # in to a standalone function.
-    tensor_util.make_tensor_proto(concat_dim, dtype=dtypes.int32, shape=[])
-    return identity(values[0], name=name)
+    # Make a throwaway call to convert_to_tensor to make sure
+    # that concat_dim is of the correct type, and make sure that
+    # the returned tensor is a scalar.
+    # TODO(keveman): Implement a standalone type and shape checker.
+    with ops.name_scope(name) as scope:
+      ops.convert_to_tensor(concat_dim,
+                            name="concat_dim",
+                            dtype=dtypes.int32).get_shape(
+                            ).assert_is_compatible_with(tensor_shape.scalar())
+      return identity(values[0], name=scope)
   return gen_array_ops._concat(concat_dim=concat_dim,
                                values=values,
                                name=name)