slozier

PEP 528 -- Change Windows console encoding to UTF-8

1

https://docs.python.org/3/whatsnew/3.6.html#pep-528-change-windows-console-encoding-to-utf-8 https://www.python.org/dev/peps/pep-0528/

Support modifier spans in re

1

For example: ```py assert re.match('(?i)g(?-i:v)r', "GVR") is None ``` Modifier spans are used by the `string` module in 3.6. See https://docs.python.org/3/whatsnew/3.6.html#re

PEP 468 – Preserving the order of **kwargs in a function

2

https://docs.python.org/3/whatsnew/3.6.html#pep-468-preserving-keyword-argument-order https://peps.python.org/pep-0468/

Add support for e format in _struct

5

See https://docs.python.org/3/whatsnew/3.6.html#struct https://github.com/IronLanguages/ironpython3/pull/1457 adds support to .NET 6.0 (via `System.Half`).

https://docs.python.org/3/whatsnew/3.6.html#pep-495-local-time-disambiguation https://peps.python.org/pep-0495/

Multiple targets could match in write function of io.BytesIO subclass

6

For example: ```Python import io class Test(io.BytesIO): pass x = Test() x.write(b"") ``` results in a > Multiple targets could match: write(object), write(IBufferProtocol) I'm guessing it's because the object overload...

For example: ```py from ctypes import cast cast(1, None) ``` raises an exception and crashes the CLR. This is causing `ctypes.test_cast` to fail on Linux.

The `PythonOps.TryInvokeLengthHint` implementation is not quite correct. For example, we do not handle the following scenarios: - ~ensure `__len__` and `__length_hint__` are not negative~ - `__len__` can return an index...

See https://docs.python.org/3/whatsnew/3.5.html#other-language-changes For example: ```py assert b"\x80".decode("utf-8", "backslashreplace") == "\\x80" ``` Causes failures in 3.6 in `test_fileinput`.

See https://docs.python.org/3/whatsnew/3.6.html#encodings Used by `uuid` in 3.6 - causes `test_uuid` to fail.