Each unit test tests one bit of functionality in the software. Unit tests are entirely automated and complete quickly. Unit tests for the entire system are gathered into one test suite, and may all be run in a single batch. The result of a unit test is simple: either it passes, or it doesn't. All this means you can test the entire system at any time without inconvenience, and quickly see what passes and what fails.
The Twisted development team adheres to the practice of Extreme Programming (XP), and the usage of unit tests is a cornerstone XP practice. Unit tests are a tool to give you increased confidence. You changed an algorithm -- did you break something? Run the unit tests. If a test fails, you know where to look, because each test covers only a small amount of code, and you know it has something to do with the changes you just made. If all the tests pass, you're good to go, and you don't need to second-guess yourself or worry that you just accidently broke someone else's program.
You don't have to write a test for every single method you write, only production methods that could possibly break.
-- Kent Beck, Extreme Programming Explained, p. 58.
$ Twisted/admin/runtests
You'll find that having something like this in your emacs init files is quite handy:
(defun runtests () (interactive) (compile "python /somepath/Twisted/admin/runtests")) (global-set-key [(alt t)] 'runtests)
Always always always be sure all the tests pass before committing any code. If someone else checks out code at the start of a development session and finds failing tests, they will not be happy and may decide to hunt you down.
Since this is a geographically dispersed team, the person who can help you get your code working probably isn't in the room with you. You may want to share your work in progress over the network, but you want to leave the main CVS tree in good working order. So use a branch, and merge your changes back in only after your problem is solved and all the unit tests pass again.
Please don't add new modules to Twisted without adding tests for them too. Otherwise we could change something which breaks your module and not find out until later, making it hard to know exactly what the change that broke it was, or until after a release, and nobody wants broken code in a release.
Tests go in Twisted/twisted/test/, and are named test_foo.py
,
where foo
is the name of the module or package being tested.
Extensive documentation on using the PyUnit framework for writing
unit tests can be found in the links
section below.
One deviation from the standard PyUnit documentation: To ensure
that any variations in test results are due to variations in the
code or environment and not the test process itself, Twisted ships
with its own, compatible, testing framework. That just
means that when you import the unittest module, you will from twisted.trial import unittest
instead of the
standard import unittest
.
As long as you have followed the module naming and placement
conventions, runtests
will be smart
enough to pick up any new tests you write.
Trial, the Twisted unit test framework, has some extensions which are designed to encourage developers to add new tests. One common situation is that a test exercises some optional functionality: maybe it depends upon certain external libraries being available, maybe it only works on certain operating systems. The important common factor is that nobody considers these limitations to be a bug.
To make it easy to test as much as possible, some tests may be skipped in
certain situations. Individual test cases can raise the
SkipTest
exception to indicate that they should be skipped, and
the remainder of the test is not run. In the summary (the very last thing
printed, at the bottom of the test output) the test is counted as a
skip
instead of a success
or fail
. This should be used
inside a conditional which looks for the necessary prerequisites:
def testSSHClient(self): if not ssh_path: raise unittest.SkipTest, "cannot find ssh, nothing to test" foo() # do actual test after the SkipTest
You can also set the .skip attribute on the method, with a string to indicate why the test is being skipped. This is convenient for temporarily turning off a test case, but it can also be set conditionally (by manipulating the class attributes after they've been defined):
def testThing(self): dotest() testThing.skip = "disabled locally"
class MyTestCase(unittest.TestCase): def testOne(self): ... def testThing(self): dotest() if not haveThing: MyTestCase.testThing.im_func.skip = "cannot test without Thing" # but testOne() will still run
Finally, you can turn off an entire TestCase at once by setting the .skip attribute on the class. If you organize your tests by the functionality they depend upon, this is a convenient way to disable just the tests which cannot be run.
class SSLTestCase(unittest.TestCase): ... class TCPTestCase(unittest.TestCase): ... if not haveSSL: SSLTestCase.skip = "cannot test without SSL support" # but TCPTestCase will still run
Two good practices which arise from the XP
development process are
sometimes at odds with each other:
These two goals will sometimes conflict. The unit tests that are written first, before any implementation has been done, are certain to fail. We want developers to commit their code frequently, for reliability and to improve coordination between multiple people working on the same problem together. While the code is being written, other developers (those not involved in the new feature) should not have to pay attention to failures in the new code. We should not dilute our well-indoctrinated Failing Test Horror Syndrome by crying wolf when an incomplete module has not yet started passing its unit tests. To do so would either teach the module author to put off writing or committing their unit tests until after all the functionality is working, or it would teach the other developers to ignore failing test cases. Both are bad things.
.todo
is intended to solve this problem. When a developer first
starts writing the unit tests for functionality that has not yet been
implemented, they can set the .todo
attribute on the test
methods that are expected to fail. These methods will still be run, but
their failure will not be counted the same as normal failures: they will go
into an expected failures
category. Developers should learn to treat
this category as a second-priority queue, behind actual test failures.
As the developer implements the feature, the tests will eventually start
passing. This is surprising: after all those tests are marked as being
expected to fail. The .todo tests which nevertheless pass are put into a
unexpected success
category. The developer should remove the .todo
tag from these tests. At that point, they become normal tests, and their
failure is once again cause for immediate action by the entire development
team.
The life cycle of a test is thus:
.todo
. Test fails: expected failure.
unexpected success.
.todo
tag is removed. Test passes. success.
failure. Developers spring into action.
success.
Any test which remains marked with .todo
for too long should
be examined. Either it represents functionality which nobody is working on,
or the test is broken in some fashion and needs to be fixed.
Please add a test-case-name
tag to the source file that is
covered by your new test. This is a comment at the beginning of the file
which looks like one of the following:
# -*- test-case-name: twisted.test.test_defer -*-
or
#!/usr/bin/python # -*- test-case-name: twisted.test.test_defer -*-
This format is understood by emacs to mark File Variables
. The
intention is to accept test-case-name
anywhere emacs would on
the first or second line of the file (but not in the File
Variables:
block that emacs accepts at the end of the file). If you
need to define other emacs file variables, you can either put them in the
File Variables:
block or use a semicolon-separated list of
variable definitions:
# -*- test-case-name: twisted.test.test_defer; fill-column: 75; -*-
If the code is exercised by multiple test cases, those may be marked by using a comma-separated list of tests, as follows: (NOTE: not all tools can handle this yet.. trial --testmodule does, though)
# -*- test-case-name: twisted.test.test_defer,twisted.test.test_tcp -*-
The test-case-name
tag will allow trial
--testmodule twisted/dir/myfile.py
to determine which test cases need
to be run to exercise the code in myfile.py
. Several tools (as
well as twisted-dev.el
's F9 command) use this to automatically
run the right tests.
The standard Python unittest
framework, from which Trial is
derived, is ideal for testing code with a fairly linear flow of control.
Twisted is an asynchronous networking framework which provides a clean,
sensible way to establish functions that are run in response to events (like
timers and incoming data), which creates a highly non-linear flow of control.
Trial has a few extensions which help to test this kind of code. This section
provides some hints on how to use these extensions and how to best structure
your tests.
Trial runs the entire test suite (over one thousand tests) in a single process, with a single reactor. Therefore it is important that your test leave the reactor in the same state as it found it. Leftover timers may expire during somebody else's unsuspecting test. Leftover connection attempts may complete (and fail) during a later test. These lead to intermittent failures that wander from test to test and are very time-consuming to track down.
Your test is responsible for cleaning up after itself. The
tearDown
method is an ideal place for this cleanup code: it is
always run regardless of whether your test passes or fails (like a bare
except
clause in a try-except construct). Exceptions in tearDown
are flagged as errors and flunk the test.
TODO: helper functions: TestCase.addPort, TestCase.addTimer
reactor.stop is considered very harmful, and should only be used by reactor-specific test cases which know how to restore the state that it kills. If you must use reactor.run, use reactor.crash to stop it instead of reactor.stop.
Trial tries to help insure that the reactor is clean after each test, but the reactor does not yet support an interface that would make this work properly. It can catch leftover timers, but not lingering sockets.
If your test creates a Deferred
and simply wants to verify
something about its result, use deferredResult
. It will wait for the
Deferred to fire and give you the result. If the Deferred runs the errback
handler instead, it will raise an exception so your test can fail. Note that
the only thing that will terminate a deferredResult
call is if the Deferred fires; in particular, timers which raise exceptions
will not cause it to return.
The preferred way to run a test that waits for something to happen (always
triggered by other things that you have done) is to use a while not self.done
loop that does reactor.iterate(0.1)
at the beginning of each pass. The
0.1
argument sets a limit on how long the reactor will wait to return
if there is nothing to do. 100 milliseconds is long enough to avoid spamming
the CPU while your timers wait to expire.
It is common for tests to establish some kind of fail-safe timeout that will terminate the test in case something unexpected has happened and none of the normal test-failure paths are followed. This timeout puts an upper bound on the time that a test can consume, and prevents the entire test suite from stalling because of a single test. This is especially important for the Twisted test suite, because it is run automatically by the buildbot whenever changes are committed to the CVS repository.
Trial tests indicate they have failed by raising a FailTest exception
(self.fail and friends are just wrappers around this raise
statement). Exceptions that are raised inside a
callRemote timer are caught and logged but otherwise ignored. Trial uses a
logging hook to notice when errors have been logged by the test that just
completed (so such errors will flunk the test), but this happens after the
fact: they will not be noticed by the main body of your test code. Therefore
callRemote timers can not be used directly to establish timeouts which
terminate and flunk the test.
The right way to implement this sort of timeout is to have a
self.done
flag, and a while loop which iterates the reactor
until it becomes true. Anything that causes the test to be finished (success
or failure) can set self.done to cause the loop to exit.
Most of the code in Twisted is run by the reactor as a result of socket activity. This is almost always started by Protocol.connectionMade or Protocol.dataReceived (because the output side goes through a buffer which queues data for transmission). Exceptions that are raised by code called in this way (by the reactor, through doRead or doWrite) are caught, logged, handed to connectionLost, and otherwise ignored.
This means that your Protocol's connectionLost method, if invoked because of an exception, must also set this self.done flag. Otherwise the test will not terminate.
Exceptions that are raised in a Deferred callback are turned into a Failure and stashed inside the Deferred. When an errback handler is attached, the Failure is given to it. If the Deferred goes out of scope while an error is still pending, the error is logged just like exceptions that happen in timers or protocol handlers. This will cause the current test to flunk (eventually), but it is not checked until after the test fails. So again, it is a good idea to add errbacks to your Deferreds that will terminate your test's main loop.
Here is a brief example that demonstrates a few of these techniques.
class MyTest(unittest.TestCase): def setUp(self): self.done = False self.failure = None def tearDown(self): self.server.stopListening() # TODO: also shut down client try: self.timeout.cancel() except (error.AlreadyCancelled, error.AlreadyCalled): pass def succeeded(self): self.done = True def failed(self, why): self.done = True self.failure = why def testServer(self): self.server = reactor.listenTCP(port, factory) self.client = reactor.connectTCP(port, factory) # you should give the factories a way to call our 'succeeded' or # 'failed' methods self.timeout = reactor.callLater(5, self.failed, "timeout") while not self.done: reactor.iterate(0.1) # we get here if the test is finished, for good or for bad if self.failure: self.fail(self.failure) # otherwise it probably passed. Cleanup will be done in tearDown()
unittest
module documentation, in the Python Library
Reference.