The OS X buildbots were failing at the second setrlimit() call with EPERM, as
if they were trying to raise the hard limit. The call should be keeping the
hard limit the same and raising the soft limit back to its original value, so
I don't understand the failure.

These were broken in 3aec776fc796 when they were converted
away from using support.run_unittest().  Oops :)

Initial patch by Felippe da Motta Raposo.

Before the argparse conversion, compileall would (sometimes) accept multiple
paths when -d was specified.  Afterward, it does not.  The corresponding check
in the original code claimed to prevent multiple *directories* from being
specified...but it didn't really work even to do that.  So this patch fixes
the regression by invoking the consenting adults rule: if you specify a
combination of arguments to compileall that produces files with inconsistent
destdirs (which you could do before), it is on you.

Patch by Jake Garver.

Thanks to Raúl Cumplido for the bug report and Thomas Kluyver for the
patch.

for importlib.util.resolve_name() but is needed.

Thanks to Martin Panter for the bug report.

statements search for __import__ in the global scope.

Thanks to Sergei Lebedev for finding the documentation bug.

with time.tzset().

with time.tzset().

with time.tzset().

Initialize package before calling find_spec() for __main__, so that we do not
incorrectly handle exceptions from __init__.py. When runpy is used from the
Python CLI, use an internal exception rather than ImportError, to avoid
catching unexpected exceptions.

Also remove exception message rewriting in _run_module_as_main(), because it
seems to be redundant with the _get_main_module_details() function.