POSIX semaphores (among which FreeBSD < 8).

sched.scheduler class constructor's timefunc and delayfunct parameters are now optional.
scheduler.enter and scheduler.enterabs methods gained a new kwargs parameter.

Patch contributed by Matt Mulsow.

the Solaris issue is solved.

But add assertion on the maximum character in various encoders: UTF-7, UTF-8,
wide character (wchar_t*, Py_UNICODE*), unicode-escape, raw-unicode-escape.

Fix also unicode_encode_ucs1() for backslashreplace error handler: Python is
now always "wide".

 * Replace PyUnicode_FromUnicode(NULL, 0) by PyUnicode_New(0, 0)
 * Replce PyUnicode_FromUnicode(str, len) by PyUnicode_FromWideChar(str, len)
 * Replace Py_UNICODE by wchar_t
 * posix_putenv() uses PyUnicode_FromFormat() to create the string, instead
   of PyUnicode_FromUnicode() + _snwprintf()

 * Add tests for PyUnicode_EncodeDecimal() and PyUnicode_TransformDecimalToASCII()
 * Remove the unused "e" variable in replace()

 * unicode_ready() has a simpler API
 * try to reuse unicode_empty and latin1_char singleton everywhere
 * Fix a reference leak in _PyUnicode_TranslateCharmap()
 * PyUnicode_InternInPlace() doesn't try to get a singleton anymore, to avoid
   having to handle a failure

It tests indirectly PyUnicode_Copy(): ensure that the string is a copy.

USe PyUnicode_Copy() in fixup()

Add tests for PyUnicode_EncodeDecimal() and
PyUnicode_TransformDecimalToASCII().

make _multiprocessing.win32.WaitForMultipleObjects interruptible when
the wait_flag parameter is false.  Patch by sbt.

This almost catches up with pre-PEP 393 performance, when decoding needed
only one pass.

raised when the wrapped raw file is non-blocking and the write would block.
Previous code assumed that the raw write() would raise BlockingIOError, but
RawIOBase.write() is defined to returned None when the call would block.
Patch by sbt.

raised when the wrapped raw file is non-blocking and the write would block.
Previous code assumed that the raw write() would raise BlockingIOError, but
RawIOBase.write() is defined to returned None when the call would block.
Patch by sbt.

Dump the wchar_t that we are going to decode and dump the locale

character is bigger than U+10FFFF and locale.localeconv() dumps the string
before decoding it.

Temporary hack to debug the issue #13441.

And PyUnicode_GetSize() => PyUnicode_GetLength()

using an assertion

PyUnicode_GET_SIZE() => PyUnicode_GET_LENGTH()