- this was an internal implementation detail for iterparse
- this has been changed to use a new private method instead
- XMLPullParser.close docs are now more explicit about not
  returning a root element and instead direct users towards
  read_events
- also added missing docstrings and clarified some details
  related to exactly *when* events are consumed from the
  internal queue

(Initial patch by Stefan Behnel)

Issues such as iterating and containment on Enum classes, why Enum member
instances do not show up on other Enum members, and how __new__ should
be used.

bytes until end of data.

bytes until end of data.

Issue #19016

and can't bootstrap the parser.

1. Make it work when invoked directly from the command-line. It was failing
   due to a couple of stale function/class usages in the __main__ section.
2. Close the parsed file in the parse() function after opening it.

inspect.getmembers and inspect.classify_class_attrs now search the metaclass
mro for types.DynamicClassAttributes (what use to be called
enum._RouteClassAttributeToGetattr); in part this means that these two
functions no longer rely solely on dir().

Besides now returning more accurate information, these improvements also
allow a more helpful help() on Enum classes.

strings in error report.

Also slight code reorg for PEP 8 guidelines.

Issue #18978: Update docs to reflect explicitly the ability to set the attribute at the class level.

- be clear finalizers survive automatically
- update for PEP 442 __del__ changes
- mention module cleanup changes and weakref.finalize in What's New

Since all versions since 2.2 are under the same licensing terms,
this saves the release manager from touching the two files for
very new minor release.

(should fix the failures on OpenIndiana)