Replace os.popen() with subprocess.check_output() in the platform module:
* platform.uname() (its _syscmd_ver() helper function) now redirects
stderr to DEVNULL.
* Remove platform.DEV_NULL.
* _syscmd_uname() and _syscmd_file() no longer catch AttributeError.
The "except AttributeError:" was only needed in Python 2, when
os.popen() was not always available. In Python 3,
subprocess.check_output() is always available.
Join 3 pools in these tests:
* test.test_multiprocessing_spawn.WithProcessesTestPool.test_context
* test.test_multiprocessing_spawn.WithProcessesTestPool.test_traceback
When running test_bdb.py as a script, `import test_module` would be
importing the existing Lib/test/test_modules.py instead of the
tempcwd/test_module.py module which was dynamically created by
test_bdb.py itself.
* posixpath.expanduser() now returns the input path unchanged if
the HOME environment variable is not set and pwd.getpwuid() raises
KeyError (the current user identifier doesn't exist in the password
database).
* Add test_no_home_directory() to test_site.
platform.libc_ver() now uses os.confstr('CS_GNU_LIBC_VERSION') if
available and the *executable* parameter is not set. The default
value of the libc_ver() *executable* parameter becomes None.
Quick benchmark on Fedora 29:
python3 -m perf command ./python -S -c 'import platform; platform.libc_ver()'
94.9 ms +- 4.3 ms -> 33.2 ms +- 1.4 ms: 2.86x faster (-65%)
On Travis CI, FTP tests of test_urllib2net randomly fail with "425
Security: Bad IP connecting".
test.pythoninfo now also logs TRAVIS environment variable.
Only platform._syscmd_file() uses subprocess. Move subprocess import
inside this function to reduce the number of imports at Python
startup.
Remove also warnings import which is no longer needed.
Modify asyncio tests to utilize the certificates from the test directory
instead of its own set, as they are the same and with each update they had
to be updated as well.
* Replace testInheritFlags() with two tests:
testInheritFlagsBlocking() and testInheritFlagsTimeout()
to test different default socket timeout. Moreover, the test now
checks sock.gettimeout() rather than a functional test on recv().
* Replace time.time() with time.monotonic()
* Add socket_setdefaulttimeout() context manager to restore the
default timeout when the test completes.
* Remove testConnect(): accept() wasn't blocking and testAccept()
already tests non-blocking accept().
* Remove accept() functional test from testInitNonBlocking():
already tested by testAccept()
* Rewrite testSetBlocking() with a new assert_sock_timeout() method
* Use addCleanup() and context manager to close sockets
* Replace assertTrue(x < y) with assertLess(x, y)
testAccept() and testRecv() of test_socket.NonBlockingTCPTests have a
race condition: time.sleep() is used as a weak synchronization
primitive and the tests fail randomly on slow buildbots.
Use a reliable threading.Event to fix these tests.
Other changes:
* Replace send() with sendall()
* Expect specific BlockingIOError rather than generic OSError
* Add a timeout to select() in testAccept() and testRecv()
* Use addCleanup() to close sockets
* Use assertRaises()
Remove platform.popen() function, it was deprecated since Python 3.3:
use os.popen() instead.
Rename also the "Removed" section to "API and Feature Removals"
of What's New in Python 3.8.
Fix memory leak in PyUnicode_EncodeLocale() and
PyUnicode_EncodeFSDefault() on error handling.
Changes:
* Fix unicode_encode_locale() error handling
* Fix test_codecs.LocaleCodecTest
Fix WithThreadsTestPool.test_wrapped_exception()
of test_multiprocessing_fork: join the pool.
WithThreadsTestPool.test_del_pool() is now also decorated
with @support.reap_threads.