Merge p3yk branch with the trunk up to revision 45595. This breaks a fair

number of tests, all because of the codecs/_multibytecodecs issue described
here (it's not a Py3K issue, just something Py3K discovers):
http://mail.python.org/pipermail/python-dev/2006-April/064051.html

Hye-Shik Chang promised to look for a fix, so no need to fix it here. The
tests that are expected to break are:

test_codecencodings_cn
test_codecencodings_hk
test_codecencodings_jp
test_codecencodings_kr
test_codecencodings_tw
test_codecs
test_multibytecodec

This merge fixes an actual test failure (test_weakref) in this branch,
though, so I believe merging is the right thing to do anyway.

4 files changed, 42 insertions, 19 deletions

diff --git a/Lib/test/leakers/README.txt b/Lib/test/leakers/README.txt
index 69ee35a..beeee0e 100644
--- a/Lib/test/leakers/README.txt
+++ b/Lib/test/leakers/README.txt
@@ -5,6 +5,15 @@ the interpreter was built in debug mode.  If the total ref count
 doesn't increase, the bug has been fixed and the file should be removed
 from the repository.
 
+Note:  be careful to check for cyclic garbage.  Sometimes it may be helpful
+to define the leak function like:
+
+def leak():
+    def inner_leak():
+        # this is the function that leaks, but also creates cycles
+    inner_leak()
+    gc.collect() ; gc.collect() ; gc.collect()
+
 Here's an example interpreter session for test_gestalt which still leaks:
 
 >>> from test.leakers.test_gestalt import leak
@@ -17,3 +26,7 @@ Here's an example interpreter session for test_gestalt which still leaks:
 [28940 refs]
 >>> 
 
+Once the leak is fixed, the test case should be moved into an appropriate
+test (even if it was originally from the test suite).  This ensures the
+regression doesn't happen again.  And if it does, it should be easier
+to track down.
diff --git a/Lib/test/leakers/test_ctypes.py b/Lib/test/leakers/test_ctypes.py
new file mode 100644
index 0000000..0f9a2cd
--- /dev/null
+++ b/Lib/test/leakers/test_ctypes.py
@@ -0,0 +1,16 @@
+
+# Taken from Lib/ctypes/test/test_keeprefs.py, PointerToStructure.test().
+# When this leak is fixed, remember to remove from Misc/build.sh LEAKY_TESTS.
+
+from ctypes import Structure, c_int, POINTER
+import gc
+
+def leak_inner():
+    class POINT(Structure):
+        _fields_ = [("x", c_int)]
+    class RECT(Structure):
+        _fields_ = [("a", POINTER(POINT))]
+
+def leak():
+    leak_inner()
+    gc.collect()
diff --git a/Lib/test/leakers/test_selftype.py b/Lib/test/leakers/test_selftype.py
new file mode 100644
index 0000000..4207c32
--- /dev/null
+++ b/Lib/test/leakers/test_selftype.py
@@ -0,0 +1,13 @@
+# Reference cycles involving only the ob_type field are rather uncommon
+# but possible.  Inspired by SF bug 1469629.
+
+import gc
+
+def leak():
+    class T(type):
+        pass
+    class U(type):
+        __metaclass__ = T
+    U.__class__ = U
+    del U
+    gc.collect(); gc.collect(); gc.collect()
diff --git a/Lib/test/leakers/test_tee.py b/Lib/test/leakers/test_tee.py
deleted file mode 100644
index 4ce24ca..0000000
--- a/Lib/test/leakers/test_tee.py
+++ /dev/null
@@ -1,19 +0,0 @@
-
-# Test case taken from test_itertools
-# See http://mail.python.org/pipermail/python-dev/2005-November/058339.html
-
-from itertools import tee
-
-def leak():
-    def fib():
-        def yield_identity_forever(g):
-            while 1:
-                yield g
-        def _fib():
-            for i in yield_identity_forever(head):
-                yield i
-        head, tail, result = tee(_fib(), 3)
-        return result
-
-    x = fib()
-    x.next()