This might help fix some of the failures on Windows box(es). It doesn't hurt
either way and ensure the tests are a little more self contained (ie have
less assumptions).
to listen on in network-oriented tests has been refined in an effort to
facilitate running multiple instances of the entire regression test suite
in parallel without issue. test_support.bind_port() has been fixed such
that it will always return a unique port -- which wasn't always the case
with the previous implementation, especially if socket options had been
set that affected address reuse (i.e. SO_REUSEADDR, SO_REUSEPORT). The
new implementation of bind_port() will actually raise an exception if it
is passed an AF_INET/SOCK_STREAM socket with either the SO_REUSEADDR or
SO_REUSEPORT socket option set. Furthermore, if available, bind_port()
will set the SO_EXCLUSIVEADDRUSE option on the socket it's been passed.
This currently only applies to Windows. This option prevents any other
sockets from binding to the host/port we've bound to, thus removing the
possibility of the 'non-deterministic' behaviour, as Microsoft puts it,
that occurs when a second SOCK_STREAM socket binds and accepts to a
host/port that's already been bound by another socket. The optional
preferred port parameter to bind_port() has been removed. Under no
circumstances should tests be hard coding ports!
test_support.find_unused_port() has also been introduced, which will pass
a temporary socket object to bind_port() in order to obtain an unused port.
The temporary socket object is then closed and deleted, and the port is
returned. This method should only be used for obtaining an unused port
in order to pass to an external program (i.e. the -accept [port] argument
to openssl's s_server mode) or as a parameter to a server-oriented class
that doesn't give you direct access to the underlying socket used.
Finally, test_support.HOST has been introduced, which should be used for
the host argument of any relevant socket calls (i.e. bind and connect).
The following tests were updated to following the new conventions:
test_socket, test_smtplib, test_asyncore, test_ssl, test_httplib,
test_poplib, test_ftplib, test_telnetlib, test_socketserver,
test_asynchat and test_socket_ssl.
It is now possible for multiple instances of the regression test suite to
run in parallel without issue.
suite as a side-effect of importing the module.
- in test_capi, a thread tried to import other modules
- re.compile() imported sre_parse again on every call.
close() will now raise an IOError if any operations on the file object
are currently in progress in other threads.
Most code was written by Antoine Pitrou (pitrou). Additional testing,
documentation and test suite cleanup done by me (gregory.p.smith).
Fixes issue 815646 and 595601 (as well as many other bugs and
references to this problem dating back to the dawn of Python).
HandlerBException is ignored, and fix one such problem, where it was thrown
during the __del__ method of the previous Popen object.
We may want to find a better way of printing verbose information so it's not
spammy when the test passes.
calls threading.currentThread.
The correction somewhat improves the code, but it was close.
Many thanks to the "with" construct, which turns python code into C calls.
I wonder if it is not better to sys.settrace(None) just after
running the __main__ module and before finalization.
When cls is an ABCMeta, every call to isinstance(x, cls)
records type(x) in the cls._abc_cache of cls_abc_negative_cache.
So we clear these caches at the end of the test.
inspect.isabstract() is not the correct test for all ABCs, because there is no @abstractmethod in io.py (why?)
isinstance(cls, ABCMeta) would be more exact, but it fails with an infinite recursion.
So I used a hack to determine whether a class is an ABCMeta.
The true correction would be to turn cls._abc_cache &co into a WeakSet, as py3k does.
But classic classes are not weak referenceable...
Of course, this change should not be merged into the py3k branch.
The moved tests use a local server rather than going out to external servers.
Accepts patch from issue2429.
Contributed by Jerry Seutter & Michael Foord (fuzzyman) at PyCon 2008.
reliable, but I'm not convinced it is the right solution. We need
to determine if this causes the test to hang on any platforms or do
other bad things.
Even if it gets the test to pass reliably, it might be that we want
to fix this in socket. The socket returned from accept() is different
on different platforms (inheriting attributes or not) and we might
want to ensure that the attributes (at least blocking) is the same
across all platforms.
up after it was joined had a traceback pointing to that thread's (deleted)
target attribute, while the test was trying to check that the target was
destroyed. Big thanks to Antoine Pitrou for diagnosing the race and pointing
out sys.exc_clear() to kill the exception early. This fixes issue 2496.
Neal Norwitz
Gregory P. Smith
Jerry Seutter
Trent Nelson
Amaury Forgeot d'Arc
Jeffrey Yasskin
Brett Cannon
Georg Brandl
Benjamin Peterson
Martin v. Löwis
Christian Heimes
Eric Smith