1
=======================
2
Guide to Testing Bazaar
3
=======================
10
The Importance of Testing
11
=========================
13
Reliability is a critical success factor for any Version Control System.
14
We want Bazaar to be highly reliable across multiple platforms while
15
evolving over time to meet the needs of its community.
17
In a nutshell, this is what we expect and encourage:
19
* New functionality should have test cases. Preferably write the
20
test before writing the code.
22
In general, you can test at either the command-line level or the
23
internal API level. See `Writing tests`_ below for more detail.
25
* Try to practice Test-Driven Development: before fixing a bug, write a
26
test case so that it does not regress. Similarly for adding a new
27
feature: write a test case for a small version of the new feature before
28
starting on the code itself. Check the test fails on the old code, then
29
add the feature or fix and check it passes.
31
By doing these things, the Bazaar team gets increased confidence that
32
changes do what they claim to do, whether provided by the core team or
33
by community members. Equally importantly, we can be surer that changes
34
down the track do not break new features or bug fixes that you are
37
As of May 2008, Bazaar ships with a test suite containing over 12000 tests
38
and growing. We are proud of it and want to remain so. As community
39
members, we all benefit from it. Would you trust version control on
40
your project to a product *without* a test suite like Bazaar has?
43
Running the Test Suite
44
======================
46
Currently, bzr selftest is used to invoke tests.
47
You can provide a pattern argument to run a subset. For example,
48
to run just the blackbox tests, run::
50
./bzr selftest -v blackbox
52
To skip a particular test (or set of tests), use the --exclude option
53
(shorthand -x) like so::
55
./bzr selftest -v -x blackbox
57
To ensure that all tests are being run and succeeding, you can use the
58
--strict option which will fail if there are any missing features or known
61
./bzr selftest --strict
63
To list tests without running them, use the --list-only option like so::
65
./bzr selftest --list-only
67
This option can be combined with other selftest options (like -x) and
68
filter patterns to understand their effect.
70
Once you understand how to create a list of tests, you can use the --load-list
71
option to run only a restricted set of tests that you kept in a file, one test
72
id by line. Keep in mind that this will never be sufficient to validate your
73
modifications, you still need to run the full test suite for that, but using it
74
can help in some cases (like running only the failed tests for some time)::
76
./bzr selftest -- load-list my_failing_tests
78
This option can also be combined with other selftest options, including
79
patterns. It has some drawbacks though, the list can become out of date pretty
80
quick when doing Test Driven Development.
82
To address this concern, there is another way to run a restricted set of tests:
83
the --starting-with option will run only the tests whose name starts with the
84
specified string. It will also avoid loading the other tests and as a
85
consequence starts running your tests quicker::
87
./bzr selftest --starting-with bzrlib.blackbox
89
This option can be combined with all the other selftest options including
90
--load-list. The later is rarely used but allows to run a subset of a list of
91
failing tests for example.
94
Test suite debug flags
95
----------------------
97
Similar to the global ``-Dfoo`` debug options, bzr selftest accepts
98
``-E=foo`` debug flags. These flags are:
100
:allow_debug: do *not* clear the global debug flags when running a test.
101
This can provide useful logging to help debug test failures when used
102
with e.g. ``bzr -Dhpss selftest -E=allow_debug``
108
Where should I put a new test?
109
------------------------------
111
Bzrlib's tests are organised by the type of test. Most of the tests in
112
bzr's test suite belong to one of these categories:
115
- Blackbox (UI) tests
116
- Per-implementation tests
119
A quick description of these test types and where they belong in bzrlib's
120
source follows. Not all tests fall neatly into one of these categories;
121
in those cases use your judgement.
127
Unit tests make up the bulk of our test suite. These are tests that are
128
focused on exercising a single, specific unit of the code as directly
129
as possible. Each unit test is generally fairly short and runs very
132
They are found in ``bzrlib/tests/test_*.py``. So in general tests should
133
be placed in a file named test_FOO.py where FOO is the logical thing under
136
For example, tests for merge3 in bzrlib belong in bzrlib/tests/test_merge3.py.
137
See bzrlib/tests/test_sampler.py for a template test script.
143
Tests can be written for the UI or for individual areas of the library.
144
Choose whichever is appropriate: if adding a new command, or a new command
145
option, then you should be writing a UI test. If you are both adding UI
146
functionality and library functionality, you will want to write tests for
147
both the UI and the core behaviours. We call UI tests 'blackbox' tests
148
and they belong in ``bzrlib/tests/blackbox/*.py``.
150
When writing blackbox tests please honour the following conventions:
152
1. Place the tests for the command 'name' in
153
bzrlib/tests/blackbox/test_name.py. This makes it easy for developers
154
to locate the test script for a faulty command.
156
2. Use the 'self.run_bzr("name")' utility function to invoke the command
157
rather than running bzr in a subprocess or invoking the
158
cmd_object.run() method directly. This is a lot faster than
159
subprocesses and generates the same logging output as running it in a
160
subprocess (which invoking the method directly does not).
162
3. Only test the one command in a single test script. Use the bzrlib
163
library when setting up tests and when evaluating the side-effects of
164
the command. We do this so that the library api has continual pressure
165
on it to be as functional as the command line in a simple manner, and
166
to isolate knock-on effects throughout the blackbox test suite when a
167
command changes its name or signature. Ideally only the tests for a
168
given command are affected when a given command is changed.
170
4. If you have a test which does actually require running bzr in a
171
subprocess you can use ``run_bzr_subprocess``. By default the spawned
172
process will not load plugins unless ``--allow-plugins`` is supplied.
175
Per-implementation tests
176
~~~~~~~~~~~~~~~~~~~~~~~~
178
Per-implementation tests are tests that are defined once and then run
179
against multiple implementations of an interface. For example,
180
``test_transport_implementations.py`` defines tests that all Transport
181
implementations (local filesystem, HTTP, and so on) must pass.
183
They are found in ``bzrlib/tests/*_implementations/test_*.py``,
184
``bzrlib/tests/per_*/*.py``, and
185
``bzrlib/tests/test_*_implementations.py``.
187
These are really a sub-category of unit tests, but an important one.
193
We make selective use of doctests__. In general they should provide
194
*examples* within the API documentation which can incidentally be tested. We
195
don't try to test every important case using doctests |--| regular Python
196
tests are generally a better solution. That is, we just use doctests to
197
make our documentation testable, rather than as a way to make tests.
199
Most of these are in ``bzrlib/doc/api``. More additions are welcome.
201
__ http://docs.python.org/lib/module-doctest.html
212
In our enhancements to unittest we allow for some addition results beyond
213
just success or failure.
215
If a test can't be run, it can say that it's skipped by raising a special
216
exception. This is typically used in parameterized tests |--| for example
217
if a transport doesn't support setting permissions, we'll skip the tests
218
that relating to that. ::
221
return self.branch_format.initialize(repo.bzrdir)
222
except errors.UninitializableFormat:
223
raise tests.TestSkipped('Uninitializable branch format')
225
Raising TestSkipped is a good idea when you want to make it clear that the
226
test was not run, rather than just returning which makes it look as if it
229
Several different cases are distinguished:
232
Generic skip; the only type that was present up to bzr 0.18.
235
The test doesn't apply to the parameters with which it was run.
236
This is typically used when the test is being applied to all
237
implementations of an interface, but some aspects of the interface
238
are optional and not present in particular concrete
239
implementations. (Some tests that should raise this currently
240
either silently return or raise TestSkipped.) Another option is
241
to use more precise parameterization to avoid generating the test
245
The test can't be run because a dependency (typically a Python
246
library) is not available in the test environment. These
247
are in general things that the person running the test could fix
248
by installing the library. It's OK if some of these occur when
249
an end user runs the tests or if we're specifically testing in a
250
limited environment, but a full test should never see them.
252
See `Test feature dependencies`_ below.
255
The test exists but is known to fail, for example this might be
256
appropriate to raise if you've committed a test for a bug but not
257
the fix for it, or if something works on Unix but not on Windows.
259
Raising this allows you to distinguish these failures from the
260
ones that are not expected to fail. If the test would fail
261
because of something we don't expect or intend to fix,
262
KnownFailure is not appropriate, and TestNotApplicable might be
265
KnownFailure should be used with care as we don't want a
266
proliferation of quietly broken tests.
268
We plan to support three modes for running the test suite to control the
269
interpretation of these results. Strict mode is for use in situations
270
like merges to the mainline and releases where we want to make sure that
271
everything that can be tested has been tested. Lax mode is for use by
272
developers who want to temporarily tolerate some known failures. The
273
default behaviour is obtained by ``bzr selftest`` with no options, and
274
also (if possible) by running under another unittest harness.
276
======================= ======= ======= ========
277
result strict default lax
278
======================= ======= ======= ========
279
TestSkipped pass pass pass
280
TestNotApplicable pass pass pass
281
UnavailableFeature fail pass pass
282
KnownFailure fail pass pass
283
======================= ======= ======= ========
286
Test feature dependencies
287
-------------------------
289
Writing tests that require a feature
290
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
292
Rather than manually checking the environment in each test, a test class
293
can declare its dependence on some test features. The feature objects are
294
checked only once for each run of the whole test suite.
296
(For historical reasons, as of May 2007 many cases that should depend on
297
features currently raise TestSkipped.)
301
class TestStrace(TestCaseWithTransport):
303
_test_needs_features = [StraceFeature]
305
This means all tests in this class need the feature. If the feature is
306
not available the test will be skipped using UnavailableFeature.
308
Individual tests can also require a feature using the ``requireFeature``
311
self.requireFeature(StraceFeature)
313
Features already defined in bzrlib.tests include:
318
- UnicodeFilenameFeature,
319
- FTPServerFeature, and
320
- CaseInsensitiveFilesystemFeature.
323
Defining a new feature that tests can require
324
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
326
New features for use with ``_test_needs_features`` or ``requireFeature``
327
are defined by subclassing ``bzrlib.tests.Feature`` and overriding the
328
``_probe`` and ``feature_name`` methods. For example::
330
class _SymlinkFeature(Feature):
333
return osutils.has_symlinks()
335
def feature_name(self):
338
SymlinkFeature = _SymlinkFeature()
341
Testing exceptions and errors
342
-----------------------------
344
It's important to test handling of errors and exceptions. Because this
345
code is often not hit in ad-hoc testing it can often have hidden bugs --
346
it's particularly common to get NameError because the exception code
347
references a variable that has since been renamed.
349
.. TODO: Something about how to provoke errors in the right way?
351
In general we want to test errors at two levels:
353
1. A test in ``test_errors.py`` checking that when the exception object is
354
constructed with known parameters it produces an expected string form.
355
This guards against mistakes in writing the format string, or in the
356
``str`` representations of its parameters. There should be one for
357
each exception class.
359
2. Tests that when an api is called in a particular situation, it raises
360
an error of the expected class. You should typically use
361
``assertRaises``, which in the Bazaar test suite returns the exception
362
object to allow you to examine its parameters.
364
In some cases blackbox tests will also want to check error reporting. But
365
it can be difficult to provoke every error through the commandline
366
interface, so those tests are only done as needed |--| eg in response to a
367
particular bug or if the error is reported in an unusual way(?) Blackbox
368
tests should mostly be testing how the command-line interface works, so
369
should only test errors if there is something particular to the cli in how
370
they're displayed or handled.
376
The Python ``warnings`` module is used to indicate a non-fatal code
377
problem. Code that's expected to raise a warning can be tested through
380
The test suite can be run with ``-Werror`` to check no unexpected errors
383
However, warnings should be used with discretion. It's not an appropriate
384
way to give messages to the user, because the warning is normally shown
385
only once per source line that causes the problem. You should also think
386
about whether the warning is serious enought that it should be visible to
387
users who may not be able to fix it.
390
Interface implementation testing and test scenarios
391
---------------------------------------------------
393
There are several cases in Bazaar of multiple implementations of a common
394
conceptual interface. ("Conceptual" because it's not necessary for all
395
the implementations to share a base class, though they often do.)
396
Examples include transports and the working tree, branch and repository
399
In these cases we want to make sure that every implementation correctly
400
fulfils the interface requirements. For example, every Transport should
401
support the ``has()`` and ``get()`` and ``clone()`` methods. We have a
402
sub-suite of tests in ``test_transport_implementations``. (Most
403
per-implementation tests are in submodules of ``bzrlib.tests``, but not
404
the transport tests at the moment.)
406
These tests are repeated for each registered Transport, by generating a
407
new TestCase instance for the cross product of test methods and transport
408
implementations. As each test runs, it has ``transport_class`` and
409
``transport_server`` set to the class it should test. Most tests don't
410
access these directly, but rather use ``self.get_transport`` which returns
411
a transport of the appropriate type.
413
The goal is to run per-implementation only the tests that relate to that
414
particular interface. Sometimes we discover a bug elsewhere that happens
415
with only one particular transport. Once it's isolated, we can consider
416
whether a test should be added for that particular implementation,
417
or for all implementations of the interface.
419
The multiplication of tests for different implementations is normally
420
accomplished by overriding the ``load_tests`` function used to load tests
421
from a module. This function typically loads all the tests, then applies
422
a TestProviderAdapter to them, which generates a longer suite containing
423
all the test variations.
425
See also `Per-implementation tests`_ (above).
431
Some utilities are provided for generating variations of tests. This can
432
be used for per-implementation tests, or other cases where the same test
433
code needs to run several times on different scenarios.
435
The general approach is to define a class that provides test methods,
436
which depend on attributes of the test object being pre-set with the
437
values to which the test should be applied. The test suite should then
438
also provide a list of scenarios in which to run the tests.
440
Typically ``multiply_tests_from_modules`` should be called from the test
441
module's ``load_tests`` function.
447
We have a rich collection of tools to support writing tests. Please use
448
them in preference to ad-hoc solutions as they provide portability and
449
performance benefits.
452
TestCase and its subclasses
453
~~~~~~~~~~~~~~~~~~~~~~~~~~~
455
The ``bzrlib.tests`` module defines many TestCase classes to help you
459
A base TestCase that extends the Python standard library's
460
TestCase in several ways. It adds more assertion methods (e.g.
461
``assertContainsRe``), ``addCleanup``, and other features (see its API
462
docs for details). It also has a ``setUp`` that makes sure that
463
global state like registered hooks and loggers won't interfere with
464
your test. All tests should use this base class (whether directly or
467
TestCaseWithMemoryTransport
468
Extends TestCase and adds methods like ``get_transport``,
469
``make_branch`` and ``make_branch_builder``. The files created are
470
stored in a MemoryTransport that is discarded at the end of the test.
471
This class is good for tests that need to make branches or use
472
transports, but that don't require storing things on disk. All tests
473
that create bzrdirs should use this base class (either directly or via
474
a subclass) as it ensures that the test won't accidentally operate on
475
real branches in your filesystem.
478
Extends TestCaseWithMemoryTransport. For tests that really do need
479
files to be stored on disk, e.g. because a subprocess uses a file, or
480
for testing functionality that accesses the filesystem directly rather
481
than via the Transport layer (such as dirstate).
483
TestCaseWithTransport
484
Extends TestCaseInTempDir. Provides ``get_url`` and
485
``get_readonly_url`` facilities. Subclasses can control the
486
transports used by setting ``vfs_transport_factory``,
487
``transport_server`` and/or ``transport_readonly_server``.
490
See the API docs for more details.
496
When writing a test for a feature, it is often necessary to set up a
497
branch with a certain history. The ``BranchBuilder`` interface allows the
498
creation of test branches in a quick and easy manner. Here's a sample
501
builder = self.make_branch_builder('relpath')
502
builder.build_commit()
503
builder.build_commit()
504
builder.build_commit()
505
branch = builder.get_branch()
507
``make_branch_builder`` is a method of ``TestCaseWithMemoryTransport``.
509
Note that many current tests create test branches by inheriting from
510
``TestCaseWithTransport`` and using the ``make_branch_and_tree`` helper to
511
give them a ``WorkingTree`` that they can commit to. However, using the
512
newer ``make_branch_builder`` helper is preferred, because it can build
513
the changes in memory, rather than on disk. Tests that are explictly
514
testing how we work with disk objects should, of course, use a real
517
Please see bzrlib.branchbuilder for more details.
523
The ``TreeBuilder`` interface allows the construction of arbitrary trees
524
with a declarative interface. A sample session might look like::
526
tree = self.make_branch_and_tree('path')
527
builder = TreeBuilder()
528
builder.start_tree(tree)
529
builder.build(['foo', "bar/", "bar/file"])
530
tree.commit('commit the tree')
531
builder.finish_tree()
533
Usually a test will create a tree using ``make_branch_and_memory_tree`` (a
534
method of ``TestCaseWithMemoryTransport``) or ``make_branch_and_tree`` (a
535
method of ``TestCaseWithTransport``).
537
Please see bzrlib.treebuilder for more details.
540
.. |--| unicode:: U+2014
added
removed
doc/developers/testing.txt
