Added error message.

doc/library/scan.txt

@@ -390,7 +390,7 @@ ensured by the user. Otherwise, it will result in an error.
 Using the original Gibbs sampling example, with ``strict=True`` added to the
 ``scan()`` call:
 
-.. code-block:: python
+.. testcode:: scan1
 
     # Same OneStep as in original example.
     def OneStep(vsample) :
@@ -401,15 +401,39 @@ Using the original Gibbs sampling example, with ``strict=True`` added to the
                              dtype=theano.config.floatX)
 
     # The new scan, adding strict=True to the original call.
-    # Produces an error, because W, bvis and bhid are used by OneStep, but not
-    # passed explicitly.
     values, updates = theano.scan(OneStep,
                                   outputs_info=sample,
                                   n_steps=10,
                                   strict=True)
 
-The above will result in an error, indicating that ``OneStep`` relies on
-variables that are not passed as its explicit arguments.
+.. testoutput:: scan1
+
+    MissingInputError: An input of the graph, used to compute
+    InplaceDimShuffle{1,0}(<TensorType(float64, matrix)>), was not provided
+    and not given a value.Use the Theano flag exception_verbosity='high',for
+    more information on this error.
+
+The error indicates that ``OneStep`` relies on variables that are not passed
+as arguments explicitly. Here is the correct version, with the shared
+variables passed explicitly to ``OneStep`` and to scan: 
+
+.. testcode:: scan1
+
+    # OneStep, with explicit use of the shared variables (W, bvis, bhid)
+    def OneStep(vsample, W, bvis, bhid) :
+        hmean = T.nnet.sigmoid(theano.dot(vsample, W) + bhid)
+        hsample = trng.binomial(size=hmean.shape, n=1, p=hmean)
+        vmean = T.nnet.sigmoid(theano.dot(hsample, W.T) + bvis)
+        return trng.binomial(size=vsample.shape, n=1, p=vmean,
+                             dtype=theano.config.floatX)
+
+    # The new scan, adding strict=True to the original call, and passing
+    # expicitly W, bvis and bhid.
+    values, updates = theano.scan(OneStep,
+                                  outputs_info=sample,
+                                  non_sequences=[W, bvis, bhid],
+                                  n_steps=10,
+                                  strict=True)
 
 
 Multiple outputs, several taps values - Recurrent Neural Network with Scan