fix the printing of BadOptimization when we have CudaNdarray.

theano/compile/debugmode.py

@@ -165,11 +165,19 @@ class BadOptimization(DebugModeError):
             print >> sio, "  New Value: ", str(self.new_r_val)[:val_str_len_limit], '...'
         else:
             print >> sio, "  New Value: ", str(self.new_r_val)
-            
-        print >> sio, "  Max Abs Diff: ", numpy.max(numpy.absolute(self.new_r_val-self.old_r_val))
-        print >> sio, "  Mean Abs Diff: ", numpy.mean(numpy.absolute(self.new_r_val-self.old_r_val))
-        print >> sio, "  Median Abs Diff: ", numpy.median(numpy.absolute(self.new_r_val-self.old_r_val))
-        print >> sio, "  Std Abs Diff: ", numpy.std(numpy.absolute(self.new_r_val-self.old_r_val))
+
+        try:
+            ov=numpy.asarray(self.old_r_val)
+            nv=numpy.asarray(self.new_r_val)
+            ssio = StringIO()
+            print >> ssio, "  Max Abs Diff: ", numpy.max(numpy.absolute(nv-ov))
+            print >> ssio, "  Mean Abs Diff: ", numpy.mean(numpy.absolute(nv-ov))
+            print >> ssio, "  Median Abs Diff: ", numpy.median(numpy.absolute(nv-ov))
+            print >> ssio, "  Std Abs Diff: ", numpy.std(numpy.absolute(nv-ov))
+            # only if all succeeds to we add anything to sio
+            print >> sio, ssio.getvalue()                                    
+        except:
+            pass
 
         print >> sio, "  Reason: ", str(self.reason)
         print >> sio, "  Old Graph:"