[pypy-svn] r69487 - in pypy/trunk/pypy/interpreter: . test

arigo at codespeak.net arigo at codespeak.net
Sat Nov 21 12:11:27 CET 2009


Author: arigo
Date: Sat Nov 21 12:11:26 2009
New Revision: 69487

Modified:
   pypy/trunk/pypy/interpreter/executioncontext.py
   pypy/trunk/pypy/interpreter/test/test_executioncontext.py
Log:
Improve the test, and see it failing.  For now, no fix; I am
trying to fix this in the branch/faster-raise, because it requires
a bit deeper changes.


Modified: pypy/trunk/pypy/interpreter/executioncontext.py
==============================================================================
--- pypy/trunk/pypy/interpreter/executioncontext.py	(original)
+++ pypy/trunk/pypy/interpreter/executioncontext.py	Sat Nov 21 12:11:26 2009
@@ -71,6 +71,8 @@
     def _init_frame_chain(self):
         # 'some_frame' points to any frame from this thread's frame stack
         # (although in general it should point to the top one).
+        # XXX not true: some_frame must point to a frame from which we can
+        # reach the top frame by following the chain of f_forward
         self.some_frame = None
         self.framestackdepth = 0
 

Modified: pypy/trunk/pypy/interpreter/test/test_executioncontext.py
==============================================================================
--- pypy/trunk/pypy/interpreter/test/test_executioncontext.py	(original)
+++ pypy/trunk/pypy/interpreter/test/test_executioncontext.py	Sat Nov 21 12:11:26 2009
@@ -575,7 +575,7 @@
         assert frame.f_back_some is None
         assert ec.gettopframe() is None
 
-
+    @py.test.mark.xfail
     def test_frame_chain_jitted_forced(self):
 
         ec = self.EC()
@@ -593,8 +593,10 @@
         # recursive enter/leave seen by the jit
         frame3 = self.Frame(ec)
         ec._chain(frame3)
+        assert ec.gettopframe() is frame3
         res = frame3.force_f_back()
         assert res is frame2
+        assert ec.gettopframe() is frame3
 
         assert frame3.f_back() is frame2
         ec._unchain(frame3)



More information about the Pypy-commit mailing list