* Always return bytes from _HackedGetData.get_data().

Ensure the imp.load_source shim always returns bytes by reopening the file in
binary mode if needed. Hash-based pycs have to receive the source code in bytes.

It's tempting to change imp.get_suffixes() to always return 'rb' as a mode, but
that breaks some stdlib tests and likely 3rdparty code, too.

Stop using key=lambda.  This behavior is same to C version encoder.

The multiprocessing module now uses the monotonic clock
time.monotonic() instead of the system clock time.time() to implement
timeouts.

* converted bytes methods: expandtabs, ljust, rjust, center, zfill
* updated char_convertor to properly set the C default value

The line is speaking of a list of commands, not the list command.

Most of the change involves fixing up the test suite, which previously made
the assumption that there wouldn't be a new line if the input didn't end in
one.

Contributed by Ammar Askar.

tarfile._Stream has two buffer for compressed and uncompressed data.
Those buffers are not aligned so unnecessary bytes slicing happens
for every reading chunks.

This commit bypass compressed buffering.

In this benchmark [1], user time become 250ms from 300ms.

[1]: https://bugs.python.org/msg320763

Per the recommendation in our Developer's Guide:
  https://devguide.python.org/documenting/#paragraph-level-markup

subprocess.Popen now copies the startupinfo argument to leave it
unchanged: it will modify the copy, so that the same STARTUPINFO
object can be used multiple times.

Add subprocess.STARTUPINFO.copy() method.

It was moved from bitbucket to GitHub.

This commit fixes the -timedelta overfllow issue not documented properly.
Signed-off-by: Farhaan Bukhsh <farhaan.bukhsh@gmail.com>

Result of function interpreter_requires_environment() depends on os.environ.
This was not covered by the tests, leading to fail when PYTHONHOME was set.

Fix test_forkserver_sigkill() of test_multiprocessing_forkserver:
give more time to the first child process to complete, double the
sleep in the parent process.

Reduce also the child process sleep from 1000 ms to 500 ms, to not change
the total duration of the test.

During buffered read, use a list followed by join instead of extending a bytes object.
This is how it was done before but changed in commit b506dc32.

Without tp_clear, GC can't break cyclic reference.
It will cause memory leak when cyclic reference is
created intentionally.

PyErr_Print() will not return when the exception is a SystemExit, so
decref the __main__ module object in that case.

The Opera Browser was using a outdated command line invocation that resulted in an incorrect URL being opened in the browser when requested using the webbrowser module.

* Correct the arguments passed to the Opera Browser when opening a new URL.

When hunting memory leaks using -R 3:3, test_imap_unordered() of
test_multiprocessing leaks randomly a few memory blocks. It is a
false alarm: when testing using -R 3:20 for example, no leak is
detected.

Modify test_imap_unordered() to be closer to test_imap():

* Only test 10 numbers instead of 1000: it's a pool of 4 processes, so
  10 is enough to test at least one number per process
* Use chunksize=100 instead of chunksize=53 to mimick test_imap()

 The line-length limit is not needed because the pages appear in a separate app rather
 than on a browser tab.  It can also interact badly with the DPI setting.

Using OptiPNG 0.7.7.
Used command is: `find . -name '*.png' | xargs optipng -o7`

They can be exposed when some C API calls fail due to lack of
memory.

* Failed Py_BuildValue() could cause an assertion error in the
  following TextIOWrapper.tell().
* input_chunk could be decrefed twice in TextIOWrapper.seek()
  after failed Py_BuildValue().
* initvalue could leak in StringIO.__getstate__() after failed
  PyDict_Copy().