Fixed tests when on_shape_error = 'raise'

It is expected that some implementations of 'infer_shape' raise a ShapeError to indicate that they do not know how to compute the shape. In such a situation, we should silently fall back to the default implementation, regardless of the value of 'on_shape_error' (whose name may be misleading for developers, since we want to catch all exceptions *except* ShapeError). Also, when a bad shape was inferred, we should not raise a ShapeError (which may be caught silently), but rather something that indicates that there is a serious bug somewhere, like an AssertionError.

Fixed tests when on_shape_error = 'raise'
73d4fe6c · Olivier Delalleau · 5c0887dd · 73d4fe6c
--- a/theano/tensor/opt.py
+++ b/theano/tensor/opt.py
@@ -786,7 +786,7 @@ class ShapeFeature(object):
            self.shape_of[r] = s
        else:
            if r.ndim != len(s):
-                raise ShapeError(
+                raise AssertionError(
                        "Something inferred a shape with %d dimensions "
                        "for a variable with %d dimensions." % (
                        len(s), r.ndim))
@@ -908,8 +908,6 @@ class ShapeFeature(object):
            o_shapes = shape_infer(node,
                                   [self.shape_of[r] for r in node.inputs])
        except ShapeError:
-            if config.on_shape_error == "raise":
-                raise
            o_shapes = self.default_infer_shape(node, [self.shape_of[r] for
                                                       r in node.inputs])
        except NotImplementedError, e: