The 3.1 compiler don't check for keyword assignments in all places.
Find another way to generate a SyntaxError in the tests.

Previously, these statements would raise something strange like
   TypeError: "'int' object is not iterable"
diff --git a/Lib/test/test_syntax.py b/Lib/test/test_syntax.py
index be783c7..2afbb61 100644
--- a/Lib/test/test_syntax.py
+++ b/Lib/test/test_syntax.py
@@ -478,32 +478,22 @@
 
 Corner-cases that used to fail to raise the correct error:
 
-    >>> def f(*, x=lambda __debug__:0): pass
+    >>> def f(*, x=lambda *:0): pass
     Traceback (most recent call last):
-    SyntaxError: assignment to keyword
+    SyntaxError: named arguments must follow bare *
 
-    >>> def f(*args:(lambda __debug__:0)): pass
+    >>> def f(*args:(lambda *:0)): pass
     Traceback (most recent call last):
-    SyntaxError: assignment to keyword
+    SyntaxError: named arguments must follow bare *
 
-    >>> def f(**kwargs:(lambda __debug__:0)): pass
+    >>> def f(**kwargs:(lambda *:0)): pass
     Traceback (most recent call last):
-    SyntaxError: assignment to keyword
+    SyntaxError: named arguments must follow bare *
 
     >>> with (lambda *:0): pass
     Traceback (most recent call last):
     SyntaxError: named arguments must follow bare *
 
-Corner-cases that used to crash:
-
-    >>> def f(**__debug__): pass
-    Traceback (most recent call last):
-    SyntaxError: assignment to keyword
-
-    >>> def f(*xx, __debug__): pass
-    Traceback (most recent call last):
-    SyntaxError: assignment to keyword
-
 """
 
 import re