The os.getcwdb() function now uses the UTF-8 encoding on Windows,
rather than the ANSI code page: see PEP 529 for the rationale. The
function is no longer deprecated on Windows.

os.getcwd() and os.getcwdb() now detect integer overflow on memory
allocations. On Unix, these functions properly report MemoryError on
memory allocation failure.

The sqlite3 module now raises TypeError, rather than ValueError, if
operation argument type is not str: execute(), executemany() and
calling a connection.

Add a fast-path to PyUnicode_Decode() for size equals to 0.

In development mode and in debug build, encoding and errors arguments
are now checked on string encoding and decoding operations. Examples:
open(), str.encode() and bytes.decode().

By default, for best performances, the errors argument is only
checked at the first encoding/decoding error, and the encoding
argument is sometimes ignored for empty strings.

Increase robustness of test_resource_tracker(): retry for 60 seconds.

* bpo-33972: Fix EmailMessage.iter_attachments raising AttributeError.

When certain malformed messages have content-type set to 'mulitpart/*' but
still have a single part body, iter_attachments can raise AttributeError. This
patch fixes it by returning a None value instead when the body is single part.

Use os.getgroups() rather than grp.getgrall() to get groups.
Rename also the test to test_chown_gid().

Python now gets the absolute path of the script filename specified on
the command line (ex: "python3 script.py"): the __file__ attribute of
the __main__ module, sys.argv[0] and sys.path[0] become an absolute
path, rather than a relative path.

* Add _Py_isabs() and _Py_abspath() functions.
* _PyConfig_Read() now tries to get the absolute path of
  run_filename, but keeps the relative path if _Py_abspath() fails.
* Reimplement os._getfullpathname() using _Py_abspath().
* Use _Py_isabs() in getpath.c.

test.pythoninfo now also logs pwd.getpwuid(os.getuid()) and
os.getgrouplist(). Extract also os.getrandom() test to run it first.

* The UTF-8 incremental decoders fails now fast if encounter
  a sequence that can't be handled by the error handler.
* The UTF-16 incremental decoders with the surrogatepass error
  handler decodes now a lone low surrogate with final=False.

eval() was being called an extra time without a filter for
deprecation warnings.

Remove sys.getcheckinterval() and sys.setcheckinterval() functions.
They were deprecated since Python 3.2. Use sys.getswitchinterval()
and sys.setswitchinterval() instead.

Remove also check_interval field of the PyInterpreterState structure.

* regrtest: Add --cleanup option to remove "test_python_*" directories
  of previous failed test jobs.
* Add "make cleantest" to run "python3 -m test --cleanup".

At the moment you can definitely use UDPLITE sockets on Linux systems, but it would be good if this support were formalized such that you can detect support at runtime easily.

At the moment, to make and use a UDPLITE socket requires something like the following code:

```
>>> import socket
>>> a = socket.socket(socket.AF_INET, socket.SOCK_DGRAM, 136)
>>> b = socket.socket(socket.AF_INET, socket.SOCK_DGRAM, 136)
>>> a.bind(('localhost', 44444))
>>> b.sendto(b'test'*256, ('localhost', 44444))
>>> b.setsockopt(136, 10, 16)
>>> b.sendto(b'test'*256, ('localhost', 44444))
>>> b.setsockopt(136, 10, 32)
>>> b.sendto(b'test'*256, ('localhost', 44444))
>>> b.setsockopt(136, 10, 64)
>>> b.sendto(b'test'*256, ('localhost', 44444))
```

If you look at this through Wireshark, you can see that the packets are different in that the checksums and checksum coverages change.

With the pull request that I am submitting momentarily, you could do the following code instead:

```
>>> import socket
>>> a = socket.socket(socket.AF_INET, socket.SOCK_DGRAM, socket.IPPROTO_UDPLITE)
>>> b = socket.socket(socket.AF_INET, socket.SOCK_DGRAM, socket.IPPROTO_UDPLITE)
>>> a.bind(('localhost', 44444))
>>> b.sendto(b'test'*256, ('localhost', 44444))
>>> b.set_send_checksum_coverage(16)
>>> b.sendto(b'test'*256, ('localhost', 44444))
>>> b.set_send_checksum_coverage(32)
>>> b.sendto(b'test'*256, ('localhost', 44444))
>>> b.set_send_checksum_coverage(64)
>>> b.sendto(b'test'*256, ('localhost', 44444))
```

One can also detect support for UDPLITE just by checking

```
>>> hasattr(socket, 'IPPROTO_UDPLITE')
```


https://bugs.python.org/issue37345

`_PyUnicode_Writer` is a relatively complex structure.  Initializing it is significant overhead when decoding short ASCII string.

When the Windows default event loop changed, `asyncio-policy.rst` was updated but `asyncio-eventloop.rst` was missed.

… as proposed in PEP 572; key is now evaluated before value.





https://bugs.python.org/issue35224

Rephrase and clarify that "the entire Python program exits when only daemon threads are left". This matches the documentation at https://docs.python.org/3/library/threading.html#thread-objects.

When the test is ran with `PYTHONWARNINGS=error` the environment variable is passed to the python interpreter used in `assert_python_ok` where `DeprecationWarning` from  `@asyncio.coroutine` is converted into an error. Ignore the `DeprecationWarning` in `assert_python_ok`.


https://bugs.python.org/issue37323

https://bugs.python.org/issue37364

test_gdb no longer fails if it gets an "unexpected" message on
stderr: it now ignores stderr. The purpose of test_gdb is to test
that python-gdb.py commands work as expected, not to test gdb.