Merge pull request #226 from nouiz/remove_deprecated

theano/compile/mode.py

@@ -312,15 +312,9 @@ class Mode(object):
 # string as the key
 FAST_COMPILE = Mode('py', 'fast_compile')
 FAST_RUN = Mode('c|py', 'fast_run')
-FAST_RUN_NOGC = Mode("c|py_nogc", 'fast_run')
-STABILIZE = Mode("c|py", OPT_STABILIZE)
 
-# The strings 'FAST_RUN_NOGC' and 'STABILIZE' are deprecated,
-# the modes in question should be built when needed.
 predefined_modes = {'FAST_COMPILE': FAST_COMPILE,
                     'FAST_RUN': FAST_RUN,
-                    'FAST_RUN_NOGC':FAST_RUN_NOGC,
-                    'STABILIZE': STABILIZE,
                     }
 
 instanciated_default_mode=None
@@ -355,17 +349,6 @@ def get_mode(orig_string):
             from profilemode import ProfileMode,prof_mode_instance_to_print
             ret = eval(string+'(linker=config.linker, optimizer=config.optimizer)')
     elif predefined_modes.has_key(string):
-        # 'FAST_RUN_NOGC' and 'STABILIZE' are deprecated
-        if string == 'FAST_RUN_NOGC':
-            warnings.warn("Using the string 'FAST_RUN_NOGC' as a mode is "
-                    "deprecated, you should use the object "
-                    "Mode(linker='c|py_nogc') instead.",
-                    stacklevel=5)
-        elif string == 'STABILIZE':
-            warnings.warn("Using the string 'STABILIZE' as a mode is "
-                    "deprecated, you should use the object "
-                    "Mode(optimizer='stabilize') instead.",
-                    stacklevel=5)
         ret = predefined_modes[string]
     else:
         raise Exception("No predefined mode exist for string: %s"%string)

theano/configdefaults.py

@@ -26,6 +26,11 @@ AddConfigVar('cast_policy',
                ),
         )
 
+# python 2.* define int / int to return int and int // int to return int.
+# python 3* define int / int to return float and int // int to return int.
+# numpy 1.6.1 behaves as python 2.*. I think we should not change it faster
+# than numpy. When we will do the transition, we should create an int_warn
+# and floatX_warn option.
 AddConfigVar('int_division',
         "What to do when one computes x / y, where both x and y are of "
         "integer types",

theano/tensor/basic.py

@@ -4175,10 +4175,6 @@ class Join(Op):
         outputs = [output_maker(bcastable)]
 
         node = Apply(self, inputs, outputs)
-        if python_any(not x.type.broadcastable[0] for x in orig):
-            node.tag.shape_zero = None
-        else:
-            node.tag.shape_zero = len(orig)
         return node
 
     def perform(self, node, axis_and_tensors, out_):
@@ -4225,17 +4221,6 @@ class Join(Op):
                 [slice(None)] * (n_dims - axis - 1)] \
                 for k in xrange(len(sizes_along_axis))]
 
-    def vec_length(self, node):
-        """Guess the length of a Join Variable"""
-        assert isinstance(node.owner.op, Join)
-        if node.ndim != 1:
-            raise TypeError('argument must be symbolic vector')
-        if node.owner.tag.shape_zero is None:
-            raise ValueError("could not determine vector length")
-        else:
-            return node.owner.tag.shape_zero
-
-
     def infer_shape(self, node, ishapes):
         # ishapes[0] contains the size of the axis on which we join
         # Join op should get at least one input to join
@@ -4427,11 +4412,6 @@ def get_vector_length(v):
         return 1
     if isinstance(v, gof.Constant) and v.type.ndim == 1:
         return len(v.data)
-    if v.owner and isinstance(v.owner.op, Join):
-        try:
-            return join.vec_length(v)
-        except ValueError:
-            pass
     if v.owner and isinstance(v.owner.op, theano.tensor.opt.MakeVector):
         return len(v.owner.inputs)
     if v.owner and isinstance(v.owner.op, Shape):

theano/tensor/tensor_grad.py

@@ -259,9 +259,8 @@ def grad(cost, wrt, g_cost=None, consider_constant=None, warn_type=False,
     :return: symbolic expression of gradient of `cost` with respect to `wrt`.
              If an element of `wrt` is not differentiable with respect
              to the output, then a zero variable is returned.
-             If `wrt` is a list/tuple, longer then 1, a list will be returned.
-             DEPRECATION: In Theano 0.5, grad will return an object of the same
-             type as `wrt`: a list/tuple or TensorVariable in all case.
+             It returns an object of same type as `wrt`: a list/tuple
+             or TensorVariable in all cases.
 
     This function is a wrapper around the more general function
     `theano.gradient.grad_sources_inputs``.