We don't use debugger on Travis.

Importing unicodedata in sre_parse leads to failure in compilation.
unicodedata is unused during compilation, and is not compiled when this
file is imported. The error occurs when generating posix variables,
pprint is required. The dependency chain goes on like this:

sysconfig -> pprint -> re -> sre_compile -> sre_parse (this file)

This commits fixes compilation issues introduced by
2272cec13b53c405d86c45d404f035f201c0baef.
(Issue 30688, GH-5588)

Co-authored-by: Jonathan Eunice <jonathan.eunice@gmail.com>

Co-authored-by: Jake Davis <jcdavis@awedge.net>

fnmatch.translate() no longer produces patterns which contain set
operations.

Sets starting with '[' or containing '--', '&&', '~~' or '||' will
be interpreted differently in regular expressions in future versions.
Currently they emit warnings. fnmatch.translate() now avoids producing
patterns containing such sets by accident.

* use isinstance(..)  instead of type(..)
* use  '.. not in ..'  instead of 'not .. in ..	'

Fix bug in travis configuration where it did not run the tests when
a change includes both code and doc changes.

Fixes the documentation for `subprocess.check_output()` not mentioning that the encoding and errors parameters were added in 3.6.

It now reads: ...be aware that Python has no control over...

it was using our mocked listdir to check when the files were gone.

Fix a rare but potential pre-exec child process deadlock in subprocess on POSIX systems when marking file descriptors inheritable on exec in the child process.  This bug appears to have been introduced in 3.4 with the inheritable file descriptors support.

This also changes Python/fileutils.c `set_inheritable` to use the "slow" two `fcntl` syscall path instead of the "fast" single `ioctl` syscall path when asked to be async signal safe (by way of being asked not to raise exceptions).  `ioctl` is not a POSIX async-signal-safe approved function.

ref: http://pubs.opengroup.org/onlinepubs/9699919799/functions/V2_chap02.html

Documents bpo-15873

now longer -> no longer

`arg_name` and `element_index` are defined as `digit`+ instead of `integer`.

When `__getattr__` is implemented, attribute lookup will always fall back to that,
even if the initial failure comes from `__getattribute__` or a descriptor's `__get__`
method (including property methods).

Add new entries to the widget list.

Changed the definition of width and precision from "integer" to "digit+" in format mini-language doc.

Fix typos found by codespell in docs, docstrings, and comments.

Previously the module name was used, which broke relative imports when pdb was run against a plain module or submodule.

Modify RE examples in documentation to use raw strings to prevent DeprecationWarning.
Add text to REGEX HOWTO to highlight the deprecation.  Approved by Serhiy Storchaka.

* Make sure ``__spec__.loader`` matches ``__loader__`` for namespace packages.
* Make sure ``__spec__.origin` matches ``__file__`` for namespace packages.

https://bugs.python.org/issue32303
https://bugs.python.org/issue32305

This reverts commit 72a0d218.

The reverted commit had a few issues so it was unanimously decided
to undo it. See the bpo issue for details.

Clarify that the level argument is used to determine whether to
perform absolute or relative imports: 0 is absolute, while a positive number
is the number of parent directories to search relative to the current module.