Use crypt_r() when available instead of crypt() in the crypt module.

As a nice side effect: This also avoids a memory sanitizer flake as clang msan doesn't know about crypt's internal libc allocated buffer.

`set_child_watcher()` *sets* the watcher.

* bpo-32492: 2.5x speed up in namedtuple attribute access using C fast path

* Add News entry

* fixup! bpo-32492: 2.5x speed up in namedtuple attribute access using C fast path

* Check for tuple in the __get__ of the new descriptor and don't cache the descriptor itself

* Don't inherit from property. Implement GC methods to handle __doc__

* Add a test for the docstring substitution in descriptors

* Update NEWS entry to reflect time against 3.7 branch

* Simplify implementation with argument clinic, better error messages, only __new__

* Use positional-only parameters for the __new__

* Use PyTuple_GET_SIZE and PyTuple_GET_ITEM to tighter the implementation of tuplegetterdescr_get

* Implement __set__ to make tuplegetter a data descriptor

* Use Py_INCREF now that we inline PyTuple_GetItem

* Apply the valid_index() function, saving one test

* Move Py_None test out of the critical path.

bpo-35602: Make sure the transport is always closed in SelectorEventLoopUnixSockSendfileTests (GH-11338)

There is a race condition in SelectorEventLoopUnixSockSendfileTests that causes the prepare() method return a non connected server protocol, making the cleanup() method skips the correct handling of the transport. This commit makes prepare() always return a connected server protocol that can always be cleaned up correctly.

The problem affects _testWithTimeoutTriggeredSend in test_socket.py.

There is a race condition regarding signal delivery in test_signal_handling_args for
test_asyncio.test_events.KqueueEventLoopTests. The signal can be received at any moment outside the time window provided in the test. The fix is to wait for the signal to be received instead with a bigger timeout.

The attribute is still used in other modules.

This will be needed for other menu items.  Change outwin to call the function instead of updating the menu item directly.

Command line options for the xlc compiler behave differently from gcc and clang,
so skip this test case for now when xlc is the compiler.

Patch by aixtools (Michael Felt)

* Fix test_mktime on AIX by adding code to get mktime to behave the
  same way as it does on other *nix systems
* Fix test_pthread_getcpuclickid in AIX by adjusting the test case
  expectations when running on AIX in 32-bit mode

Patch by Michael Felt.

A few other changes make the code easier to follow.

This reverts commit 7cf3d8e2.

Due to regressions found with using Tk 8.6.9.1, build the python.org macOS installers with Tcl/Tk 8.6.8 as used in previous releases.



https://bugs.python.org/issue35402

https://bugs.python.org/issue35596

bpo-35585: Speed up enum by-value lookup

https://bugs.python.org/issue35579


https://bugs.python.org/issue35579

Depending on system config, a missing candidate compiler name may be
reported as the empty string rather than as None, so adjust the test
helper accordingly.

AIX allows a trailing slash on local file system paths, which isn't what we want
in http.server. Accordingly, check explicitly for this case in the server code,
rather than relying on the OS raising an exception.

Patch by Michael Felt.

This test case needs "signed short" bitfields, but the
IBM XLC compiler (on AIX) does not support this.

Skip the code and test when AIX and XLC are used.

Use __xlc__ as identifier to detect the XLC compiler.

* use platform.system() as runtime test, rather than sys.platform() build-time test
* IPv6 zone id support on AIX is limited to inet_pton6_zone(), so skip related
  getaddrinfo() and getnameinfo() tests as not supported

* Revert part of "bpo-35565:  Remove incorrect test and one that wasn't needed for the fix.

Fix also handling non-ascii default values.

defines utility function -> defines utility functions

Reported in https://mail.python.org/pipermail/docs/2018-December/038693.html