applied fixes mentioned by Pascal L, abstractconv unit test passes without failure

theano/sandbox/gpuarray/tests/test_abstractconv.py

@@ -81,7 +81,6 @@ class TestConv2d(unittest.TestCase):
                         filter_flip=filter_flip,
                         input_shape=imshp,
                         filter_shape=kshp)
-        self.assertTrue(hasattr(c.tag, 'trace'))
         f_ref = theano.function([], c_ref, mode=mode)
         f = theano.function([], c, mode)
 
@@ -89,11 +88,13 @@ class TestConv2d(unittest.TestCase):
             assert any([isinstance(n.op, target_op) for n
                         in f.maker.fgraph.toposort()])
 
+        self.assertTrue(hasattr(f.maker.fgraph.outputs[0].tag, 'trace'))
         res_ref = numpy.array(f_ref())
         res = numpy.array(f())
         utt.assert_allclose(res_ref, res)
         if verify_grad:
-            utt.verify_grad(conv.AbstractConv2d(border_mode="valid", imshp=imshp, kshp=kshp,
+            utt.verify_grad(conv.AbstractConv2d(border_mode="valid",
+                                                imshp=imshp, kshp=kshp,
                                                 subsample=subsample),
                             [inputs_val, filters_val],
                             mode=mode)
@@ -125,7 +126,6 @@ class TestConv2d(unittest.TestCase):
                                             filter_flip=filter_flip,
                                             subsample=subsample,
                                             imshp=imshp, kshp=kshp)
-        self.assertTrue(hasattr(c.tag, 'trace'))
         c = c(inputs, output, filters_shape[-2:])
         c_ref = ref(inputs, output,
                     filters_shape,
@@ -133,6 +133,7 @@ class TestConv2d(unittest.TestCase):
                     subsample=subsample,
                     conv_mode=conv_mode)
         f = theano.function([], c, mode)
+        self.assertTrue(hasattr(f.maker.fgraph.outputs[0].tag, 'trace'))
         f_ref = theano.function([], c_ref, mode)
 
         if target_op is not None:
@@ -178,12 +179,12 @@ class TestConv2d(unittest.TestCase):
                                            subsample=subsample,
                                            filter_flip=filter_flip,
                                            imshp=imshp, kshp=kshp)
-        self.assertTrue(hasattr(c.tag, 'trace'))
         c = c(filters, output, inputs_shape[-2:])
         c_ref = ref(filters, output, inputs_shape,
                     border_mode=border_mode, subsample=subsample,
                     conv_mode=conv_mode)
         f = theano.function([], c, mode)
+        self.assertTrue(hasattr(f.maker.fgraph.outputs[0].tag, 'trace'))
         f_ref = theano.function([], c_ref, mode)
 
         if target_op is not None:

theano/tensor/nnet/opt.py

@@ -94,6 +94,8 @@ def local_abstractconv_gradweight_gemm(node):
 
     rval = CorrMM_gradWeights(border_mode=node.op.border_mode,
                               subsample=node.op.subsample)(img, topgrad, shape)
+    copy_stack_trace(node.outputs[0], rval)
+
     # need to flip the kernel if necessary
     if node.op.filter_flip:
         rval = rval[:, :, ::-1, ::-1]
@@ -181,6 +183,7 @@ def local_conv2d_gradweight_cpu(node):
                                   shape[0], shape[1], 1,
                                   shuffled_img.shape[4]),
                           dCdH=shuffled_topgrad)
+        copy_stack_trace(node.outputs[0], rval)
 
         rval = theano.tensor.addbroadcast(rval, 3)
         rval = rval.dimshuffle(0, 4, 1, 2)
@@ -253,6 +256,8 @@ def local_conv2d_gradweight_cpu(node):
                 kshp_logical_top_aligned=kshp_logical_top_aligned,
                 direction_hint='bprop weights')
     res = dw(img, filters)
+    copy_stack_trace(node.outputs[0], res)
+
     if node.op.border_mode == 'valid':
         res = res.dimshuffle((1, 0, 2, 3))
         res = res[:, :, ::-1, ::-1]
@@ -289,6 +294,7 @@ def local_conv2d_gradinputs_cpu(node):
                             d=(node.op.subsample[0], node.op.subsample[1], 1),
                             H=shuffled_topgrad,
                             RShape=(shape[0], shape[1], 1))
+        copy_stack_trace(node.outputs[0], rval)
         rval = theano.tensor.addbroadcast(rval, 3)
         rval = rval.dimshuffle(0, 4, 1, 2)
         rval = theano.tensor.patternbroadcast(rval,
@@ -343,6 +349,7 @@ def local_conv2d_gradinputs_cpu(node):
                  version=-1,
                  direction_hint='bprop inputs')
     din = din(topgrad, filters)
+    copy_stack_trace(node.outputs[0], din)
     din = theano.tensor.patternbroadcast(din, node.outputs[0].broadcastable)
     copy_stack_trace(node.outputs[0], din)
     return [din]