coro->cr_origin wasn't initialized if compute_cr_origin() failed in
PyCoro_New(), which would cause a crash during the coroutine's
deallocation.



https://bugs.python.org/issue35269

Source of T_STRING: https://github.com/python/cpython/blob/e42b705188271da108de42b55d9344642170aa2b/Python/structmember.c#L51

Source of PyUnicode_FromString
https://github.com/python/cpython/blob/master/Include/unicodeobject.h#L702


https://bugs.python.org/issue25438

[Issue 28401](https://bugs.python.org/issue28401): Don't attempt to import the stable API extensions, they are not supported in PyDEBUG builds (which don't implement that ABI).


https://bugs.python.org/issue28401

This missed PyErr_NoMemory() could cause a SystemError when calling
_symtable.symtable().

* The _PySys_EndInit() function now copies the
  config->module_search_path list, so config is longer modified when
  sys.path is updated.
* config->warnoptions list and config->xoptions dict are also copied
* test_embed: InitConfigTests now also tests
  main_config['module_search_path']
* Fix _Py_InitializeMainInterpreter(): don't use config->warnoptions
   but sys.warnoptions to decide if the warnings module should
   be imported at startup.

Remove an unnecessary "that":
... will execute that the body ... ->  ... will execute the body ...

PyTuple_Pack can fail and return NULL. If this happens, then PyType_FromSpecWithBases will incorrectly create a new type without bases. Also, it will crash on the Py_DECREF that follows. Also free members and type in error conditions.

"single" needs to be decrefed if PyList_Append() fails.

test_embed.InitConfigTests tests more configuration variables.

Changes:

* InitConfigTests tests more core configuration variables:

  * base_exec_prefix
  * base_prefix
  * exec_prefix
  * home
  * legacy_windows_fs_encoding
  * legacy_windows_stdio
  * module_search_path_env
  * prefix

* "_testembed init_from_config" tests more variables:

  * argv
  * warnoptions
  * xoptions

* InitConfigTests: add check_global_config(), check_core_config() and
  check_main_config() subfunctions to cleanup the code. Move also
  constants at the class level (ex: COPY_MAIN_CONFIG).
* Fix _PyCoreConfig_AsDict(): don't set stdio_encoding twice
* Use more macros in _PyCoreConfig_AsDict() and
  _PyMainInterpreterConfig_AsDict() to reduce code duplication.
* Other minor cleanups.

Remove _Py_REF_DEBUG_COMMA, DK_DEBUG_INCREF, and DK_DEBUG_DECREF.
Convert DK_INCREF and DK_DECREF to static inline functions.

* Fix _PyCoreConfig_SetGlobalConfig(): set also Py_FrozenFlag
* Fix _PyCoreConfig_AsDict(): export also xoptions
* Add _Py_GetGlobalVariablesAsDict() and _testcapi.get_global_config()
* test.pythoninfo: dump also global configuration variables
* _testembed now serializes global, core and main configurations
  using JSON to reuse _Py_GetGlobalVariablesAsDict(),
  _PyCoreConfig_AsDict() and _PyMainInterpreterConfig_AsDict(),
  rather than duplicating code.
* test_embed.InitConfigTests now test much more configuration
  variables

Discovered using clang's MemorySanitizer when it ran python3's
test_fstring test_misformed_unicode_character_name.

An msan build will fail by simply executing: ./python -c 'u"\N"'

* Fix _PyMainInterpreterConfig_Copy():
   copy 'install_signal_handlers' attribute
* Add _PyMainInterpreterConfig_AsDict()
* Add unit tests on the main interpreter configuration
  to test_embed.InitConfigTests
* test.pythoninfo: log also main_config

If tracemalloc is not tracing Python memory allocations,
_PyMem_DumpTraceback() now suggests to enable tracemalloc
to get the traceback where the memory block has been allocated.

* Move "GC" macros together:

  * PyObject_IS_GC()
  * _Py_AS_GC()
  * _PyObject_GC_IS_TRACKED()
  * _PyObject_GC_MAY_BE_TRACKED()

* Mark other GC macros as internal (#ifdef Py_BUILD_CORE):

  * _PyGCHead_NEXT(g), _PyGCHead_SET_NEXT(g, p)
  * _PyGCHead_PREV(g), _PyGCHead_SET_PREV(g, p)
  * _PyGCHead_FINALIZED(g), _PyGCHead_SET_FINALIZED(g)
  * _PyGC_FINALIZED(o), _PyGC_SET_FINALIZED(o)
  * _PyObject_GC_TRACK(o), _PyObject_GC_UNTRACK(o)
  * _PyGC_PREV_MASK_FINALIZED
  * _PyGC_PREV_MASK_COLLECTING
  * _PyGC_PREV_SHIFT
  * _PyGC_PREV_MASK

* Replace _PyGC_generation0 with _PyRuntime.gc.generation0
* _queuemodule.c: replace _PyObject_GC_UNTRACK()
  with with PyObject_GC_UnTrack()
* Document that  _PyObject_GC_TRACK() _PyObject_GC_UNTRACK() macros
  have been removed from the public C API.

Datetime macros like PyDate_Check() have two implementations, one using
the C API capsule and one using direct access to the datetime type
symbols defined in _datetimemodule.c. Since the direct access versions
of the macros are only used in _datetimemodule.c, they have been moved
out of "datetime.h" and into _datetimemodule.c.

The _PY_DATETIME_IMPL macro is currently necessary in order to avoid
both duplicate definitions of these macros in _datetimemodule.c and
unnecessary declarations of C API capsule-related macros and varibles in
datetime.h.
Co-Authored-By: Victor Stinner <vstinner@redhat.com>

Rename our new MEMORY_SANITIZER define to _Py_MEMORY_SANITIZER.
Project based C Preprocessor namespacing at its finest. :P

reproduciblity -> reproducibility
PyPA are the group -> PyPA is the group

clang's MemorySanitizer understand getc() but does not understand
getc_unlocked().  Workaround: Don't use it on msan builds.

The warnings module now suggests to enable tracemalloc if the source
is specified, tracemalloc module is available, but tracemalloc is not
tracing memory allocations.

Adds configure flags for msan and ubsan builds to make it easier to enable.
These also encode the detail that address sanitizer and memory sanitizer
should disable pymalloc.

Define MEMORY_SANITIZER when appropriate at build time and adds workarounds
to existing code to mark things as initialized where the sanitizer is otherwise unable to
determine that.  This lets our build succeed under the memory sanitizer.  not all tests
pass without sanitizer failures yet but we're in pretty good shape after this.

a asynchronous generator -> an asynchronous generator