6
The Importance of Testing
7
=========================
9
Reliability is a critical success factor for any Version Control System.
10
We want Bazaar to be highly reliable across multiple platforms while
11
evolving over time to meet the needs of its community.
13
In a nutshell, this is what we expect and encourage:
15
* New functionality should have test cases. Preferably write the
16
test before writing the code.
18
In general, you can test at either the command-line level or the
19
internal API level. See `Writing tests`_ below for more detail.
21
* Try to practice Test-Driven Development: before fixing a bug, write a
22
test case so that it does not regress. Similarly for adding a new
23
feature: write a test case for a small version of the new feature before
24
starting on the code itself. Check the test fails on the old code, then
25
add the feature or fix and check it passes.
27
By doing these things, the Bazaar team gets increased confidence that
28
changes do what they claim to do, whether provided by the core team or
29
by community members. Equally importantly, we can be surer that changes
30
down the track do not break new features or bug fixes that you are
33
As of September 2009, Bazaar ships with a test suite containing over
34
23,000 tests and growing. We are proud of it and want to remain so. As
35
community members, we all benefit from it. Would you trust version control
36
on your project to a product *without* a test suite like Bazaar has?
39
Running the Test Suite
40
======================
42
Currently, bzr selftest is used to invoke tests.
43
You can provide a pattern argument to run a subset. For example,
44
to run just the blackbox tests, run::
46
./bzr selftest -v blackbox
48
To skip a particular test (or set of tests), use the --exclude option
49
(shorthand -x) like so::
51
./bzr selftest -v -x blackbox
53
To ensure that all tests are being run and succeeding, you can use the
54
--strict option which will fail if there are any missing features or known
57
./bzr selftest --strict
59
To list tests without running them, use the --list-only option like so::
61
./bzr selftest --list-only
63
This option can be combined with other selftest options (like -x) and
64
filter patterns to understand their effect.
66
Once you understand how to create a list of tests, you can use the --load-list
67
option to run only a restricted set of tests that you kept in a file, one test
68
id by line. Keep in mind that this will never be sufficient to validate your
69
modifications, you still need to run the full test suite for that, but using it
70
can help in some cases (like running only the failed tests for some time)::
72
./bzr selftest -- load-list my_failing_tests
74
This option can also be combined with other selftest options, including
75
patterns. It has some drawbacks though, the list can become out of date pretty
76
quick when doing Test Driven Development.
78
To address this concern, there is another way to run a restricted set of tests:
79
the --starting-with option will run only the tests whose name starts with the
80
specified string. It will also avoid loading the other tests and as a
81
consequence starts running your tests quicker::
83
./bzr selftest --starting-with bzrlib.blackbox
85
This option can be combined with all the other selftest options including
86
--load-list. The later is rarely used but allows to run a subset of a list of
87
failing tests for example.
90
Test suite debug flags
91
----------------------
93
Similar to the global ``-Dfoo`` debug options, bzr selftest accepts
94
``-E=foo`` debug flags. These flags are:
96
:allow_debug: do *not* clear the global debug flags when running a test.
97
This can provide useful logging to help debug test failures when used
98
with e.g. ``bzr -Dhpss selftest -E=allow_debug``
104
Where should I put a new test?
105
------------------------------
107
Bzrlib's tests are organised by the type of test. Most of the tests in
108
bzr's test suite belong to one of these categories:
111
- Blackbox (UI) tests
112
- Per-implementation tests
115
A quick description of these test types and where they belong in bzrlib's
116
source follows. Not all tests fall neatly into one of these categories;
117
in those cases use your judgement.
123
Unit tests make up the bulk of our test suite. These are tests that are
124
focused on exercising a single, specific unit of the code as directly
125
as possible. Each unit test is generally fairly short and runs very
128
They are found in ``bzrlib/tests/test_*.py``. So in general tests should
129
be placed in a file named test_FOO.py where FOO is the logical thing under
132
For example, tests for merge3 in bzrlib belong in bzrlib/tests/test_merge3.py.
133
See bzrlib/tests/test_sampler.py for a template test script.
139
Tests can be written for the UI or for individual areas of the library.
140
Choose whichever is appropriate: if adding a new command, or a new command
141
option, then you should be writing a UI test. If you are both adding UI
142
functionality and library functionality, you will want to write tests for
143
both the UI and the core behaviours. We call UI tests 'blackbox' tests
144
and they belong in ``bzrlib/tests/blackbox/*.py``.
146
When writing blackbox tests please honour the following conventions:
148
1. Place the tests for the command 'name' in
149
bzrlib/tests/blackbox/test_name.py. This makes it easy for developers
150
to locate the test script for a faulty command.
152
2. Use the 'self.run_bzr("name")' utility function to invoke the command
153
rather than running bzr in a subprocess or invoking the
154
cmd_object.run() method directly. This is a lot faster than
155
subprocesses and generates the same logging output as running it in a
156
subprocess (which invoking the method directly does not).
158
3. Only test the one command in a single test script. Use the bzrlib
159
library when setting up tests and when evaluating the side-effects of
160
the command. We do this so that the library api has continual pressure
161
on it to be as functional as the command line in a simple manner, and
162
to isolate knock-on effects throughout the blackbox test suite when a
163
command changes its name or signature. Ideally only the tests for a
164
given command are affected when a given command is changed.
166
4. If you have a test which does actually require running bzr in a
167
subprocess you can use ``run_bzr_subprocess``. By default the spawned
168
process will not load plugins unless ``--allow-plugins`` is supplied.
171
Per-implementation tests
172
~~~~~~~~~~~~~~~~~~~~~~~~
174
Per-implementation tests are tests that are defined once and then run
175
against multiple implementations of an interface. For example,
176
``per_transport.py`` defines tests that all Transport implementations
177
(local filesystem, HTTP, and so on) must pass. They are found in
178
``bzrlib/tests/per_*/*.py``, and ``bzrlib/tests/per_*.py``.
180
These are really a sub-category of unit tests, but an important one.
182
Along the same lines are tests for extension modules. We generally have
183
both a pure-python and a compiled implementation for each module. As such,
184
we want to run the same tests against both implementations. These can
185
generally be found in ``bzrlib/tests/*__*.py`` since extension modules are
186
usually prefixed with an underscore. Since there are only two
187
implementations, we have a helper function
188
``bzrlib.tests.permute_for_extension``, which can simplify the
189
``load_tests`` implementation.
195
We make selective use of doctests__. In general they should provide
196
*examples* within the API documentation which can incidentally be tested. We
197
don't try to test every important case using doctests |--| regular Python
198
tests are generally a better solution. That is, we just use doctests to
199
make our documentation testable, rather than as a way to make tests.
201
Most of these are in ``bzrlib/doc/api``. More additions are welcome.
203
__ http://docs.python.org/lib/module-doctest.html
209
``bzrlib/tests/script.py`` allows users to write tests in a syntax very close to a shell session,
210
using a restricted and limited set of commands that should be enough to mimic
211
most of the behaviours.
213
A script is a set of commands, each command is composed of:
215
* one mandatory command line,
216
* one optional set of input lines to feed the command,
217
* one optional set of output expected lines,
218
* one optional set of error expected lines.
220
Input, output and error lines can be specified in any order.
222
Except for the expected output, all lines start with a special
223
string (based on their origin when used under a Unix shell):
225
* '$ ' for the command,
227
* nothing for output,
230
Comments can be added anywhere, they start with '#' and end with
233
The execution stops as soon as an expected output or an expected error is not
236
When no output is specified, any ouput from the command is accepted
237
and execution continue.
239
If an error occurs and no expected error is specified, the execution stops.
241
An error is defined by a returned status different from zero, not by the
242
presence of text on the error stream.
244
The matching is done on a full string comparison basis unless '...' is used, in
245
which case expected output/errors can be less precise.
249
The following will succeeds only if 'bzr add' outputs 'adding file'::
254
If you want the command to succeed for any output, just use::
258
The following will stop with an error::
262
If you want it to succeed, use::
265
2> bzr: ERROR: unknown command "not-a-command"
267
You can use ellipsis (...) to replace any piece of text you don't want to be
270
$ bzr branch not-a-branch
271
2>bzr: ERROR: Not a branch...not-a-branch/".
273
This can be used to ignore entire lines too::
279
# And here we explain that surprising fourth line
286
You can check the content of a file with cat::
291
You can also check the existence of a file with cat, the following will fail if
292
the file doesn't exist::
296
The actual use of ScriptRunner within a TestCase looks something like
299
def test_unshelve_keep(self):
302
sr.run_script(self, '''
304
$ bzr shelve --all -m Foo
307
$ bzr unshelve --keep
324
In our enhancements to unittest we allow for some addition results beyond
325
just success or failure.
327
If a test can't be run, it can say that it's skipped by raising a special
328
exception. This is typically used in parameterized tests |--| for example
329
if a transport doesn't support setting permissions, we'll skip the tests
330
that relating to that. ::
333
return self.branch_format.initialize(repo.bzrdir)
334
except errors.UninitializableFormat:
335
raise tests.TestSkipped('Uninitializable branch format')
337
Raising TestSkipped is a good idea when you want to make it clear that the
338
test was not run, rather than just returning which makes it look as if it
341
Several different cases are distinguished:
344
Generic skip; the only type that was present up to bzr 0.18.
347
The test doesn't apply to the parameters with which it was run.
348
This is typically used when the test is being applied to all
349
implementations of an interface, but some aspects of the interface
350
are optional and not present in particular concrete
351
implementations. (Some tests that should raise this currently
352
either silently return or raise TestSkipped.) Another option is
353
to use more precise parameterization to avoid generating the test
357
The test can't be run because a dependency (typically a Python
358
library) is not available in the test environment. These
359
are in general things that the person running the test could fix
360
by installing the library. It's OK if some of these occur when
361
an end user runs the tests or if we're specifically testing in a
362
limited environment, but a full test should never see them.
364
See `Test feature dependencies`_ below.
367
The test exists but is known to fail, for example this might be
368
appropriate to raise if you've committed a test for a bug but not
369
the fix for it, or if something works on Unix but not on Windows.
371
Raising this allows you to distinguish these failures from the
372
ones that are not expected to fail. If the test would fail
373
because of something we don't expect or intend to fix,
374
KnownFailure is not appropriate, and TestNotApplicable might be
377
KnownFailure should be used with care as we don't want a
378
proliferation of quietly broken tests.
380
ModuleAvailableFeature
381
A helper for handling running tests based on whether a python
382
module is available. This can handle 3rd-party dependencies (is
383
``paramiko`` available?) as well as stdlib (``termios``) or
384
extension modules (``bzrlib._groupcompress_pyx``). You create a
385
new feature instance with::
387
MyModuleFeature = ModuleAvailableFeature('bzrlib.something')
390
def test_something(self):
391
self.requireFeature(MyModuleFeature)
392
something = MyModuleFeature.module
395
We plan to support three modes for running the test suite to control the
396
interpretation of these results. Strict mode is for use in situations
397
like merges to the mainline and releases where we want to make sure that
398
everything that can be tested has been tested. Lax mode is for use by
399
developers who want to temporarily tolerate some known failures. The
400
default behaviour is obtained by ``bzr selftest`` with no options, and
401
also (if possible) by running under another unittest harness.
403
======================= ======= ======= ========
404
result strict default lax
405
======================= ======= ======= ========
406
TestSkipped pass pass pass
407
TestNotApplicable pass pass pass
408
UnavailableFeature fail pass pass
409
KnownFailure fail pass pass
410
======================= ======= ======= ========
413
Test feature dependencies
414
-------------------------
416
Writing tests that require a feature
417
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
419
Rather than manually checking the environment in each test, a test class
420
can declare its dependence on some test features. The feature objects are
421
checked only once for each run of the whole test suite.
423
(For historical reasons, as of May 2007 many cases that should depend on
424
features currently raise TestSkipped.)
428
class TestStrace(TestCaseWithTransport):
430
_test_needs_features = [StraceFeature]
432
This means all tests in this class need the feature. If the feature is
433
not available the test will be skipped using UnavailableFeature.
435
Individual tests can also require a feature using the ``requireFeature``
438
self.requireFeature(StraceFeature)
440
Features already defined in bzrlib.tests include:
445
- UnicodeFilenameFeature,
446
- FTPServerFeature, and
447
- CaseInsensitiveFilesystemFeature.
450
Defining a new feature that tests can require
451
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
453
New features for use with ``_test_needs_features`` or ``requireFeature``
454
are defined by subclassing ``bzrlib.tests.Feature`` and overriding the
455
``_probe`` and ``feature_name`` methods. For example::
457
class _SymlinkFeature(Feature):
460
return osutils.has_symlinks()
462
def feature_name(self):
465
SymlinkFeature = _SymlinkFeature()
468
Testing exceptions and errors
469
-----------------------------
471
It's important to test handling of errors and exceptions. Because this
472
code is often not hit in ad-hoc testing it can often have hidden bugs --
473
it's particularly common to get NameError because the exception code
474
references a variable that has since been renamed.
476
.. TODO: Something about how to provoke errors in the right way?
478
In general we want to test errors at two levels:
480
1. A test in ``test_errors.py`` checking that when the exception object is
481
constructed with known parameters it produces an expected string form.
482
This guards against mistakes in writing the format string, or in the
483
``str`` representations of its parameters. There should be one for
484
each exception class.
486
2. Tests that when an api is called in a particular situation, it raises
487
an error of the expected class. You should typically use
488
``assertRaises``, which in the Bazaar test suite returns the exception
489
object to allow you to examine its parameters.
491
In some cases blackbox tests will also want to check error reporting. But
492
it can be difficult to provoke every error through the commandline
493
interface, so those tests are only done as needed |--| eg in response to a
494
particular bug or if the error is reported in an unusual way(?) Blackbox
495
tests should mostly be testing how the command-line interface works, so
496
should only test errors if there is something particular to the cli in how
497
they're displayed or handled.
503
The Python ``warnings`` module is used to indicate a non-fatal code
504
problem. Code that's expected to raise a warning can be tested through
507
The test suite can be run with ``-Werror`` to check no unexpected errors
510
However, warnings should be used with discretion. It's not an appropriate
511
way to give messages to the user, because the warning is normally shown
512
only once per source line that causes the problem. You should also think
513
about whether the warning is serious enought that it should be visible to
514
users who may not be able to fix it.
517
Interface implementation testing and test scenarios
518
---------------------------------------------------
520
There are several cases in Bazaar of multiple implementations of a common
521
conceptual interface. ("Conceptual" because it's not necessary for all
522
the implementations to share a base class, though they often do.)
523
Examples include transports and the working tree, branch and repository
526
In these cases we want to make sure that every implementation correctly
527
fulfils the interface requirements. For example, every Transport should
528
support the ``has()`` and ``get()`` and ``clone()`` methods. We have a
529
sub-suite of tests in ``test_transport_implementations``. (Most
530
per-implementation tests are in submodules of ``bzrlib.tests``, but not
531
the transport tests at the moment.)
533
These tests are repeated for each registered Transport, by generating a
534
new TestCase instance for the cross product of test methods and transport
535
implementations. As each test runs, it has ``transport_class`` and
536
``transport_server`` set to the class it should test. Most tests don't
537
access these directly, but rather use ``self.get_transport`` which returns
538
a transport of the appropriate type.
540
The goal is to run per-implementation only the tests that relate to that
541
particular interface. Sometimes we discover a bug elsewhere that happens
542
with only one particular transport. Once it's isolated, we can consider
543
whether a test should be added for that particular implementation,
544
or for all implementations of the interface.
546
The multiplication of tests for different implementations is normally
547
accomplished by overriding the ``load_tests`` function used to load tests
548
from a module. This function typically loads all the tests, then applies
549
a TestProviderAdapter to them, which generates a longer suite containing
550
all the test variations.
552
See also `Per-implementation tests`_ (above).
558
Some utilities are provided for generating variations of tests. This can
559
be used for per-implementation tests, or other cases where the same test
560
code needs to run several times on different scenarios.
562
The general approach is to define a class that provides test methods,
563
which depend on attributes of the test object being pre-set with the
564
values to which the test should be applied. The test suite should then
565
also provide a list of scenarios in which to run the tests.
567
Typically ``multiply_tests_from_modules`` should be called from the test
568
module's ``load_tests`` function.
574
We have a rich collection of tools to support writing tests. Please use
575
them in preference to ad-hoc solutions as they provide portability and
576
performance benefits.
579
TestCase and its subclasses
580
~~~~~~~~~~~~~~~~~~~~~~~~~~~
582
The ``bzrlib.tests`` module defines many TestCase classes to help you
586
A base TestCase that extends the Python standard library's
587
TestCase in several ways. It adds more assertion methods (e.g.
588
``assertContainsRe``), ``addCleanup``, and other features (see its API
589
docs for details). It also has a ``setUp`` that makes sure that
590
global state like registered hooks and loggers won't interfere with
591
your test. All tests should use this base class (whether directly or
594
TestCaseWithMemoryTransport
595
Extends TestCase and adds methods like ``get_transport``,
596
``make_branch`` and ``make_branch_builder``. The files created are
597
stored in a MemoryTransport that is discarded at the end of the test.
598
This class is good for tests that need to make branches or use
599
transports, but that don't require storing things on disk. All tests
600
that create bzrdirs should use this base class (either directly or via
601
a subclass) as it ensures that the test won't accidentally operate on
602
real branches in your filesystem.
605
Extends TestCaseWithMemoryTransport. For tests that really do need
606
files to be stored on disk, e.g. because a subprocess uses a file, or
607
for testing functionality that accesses the filesystem directly rather
608
than via the Transport layer (such as dirstate).
610
TestCaseWithTransport
611
Extends TestCaseInTempDir. Provides ``get_url`` and
612
``get_readonly_url`` facilities. Subclasses can control the
613
transports used by setting ``vfs_transport_factory``,
614
``transport_server`` and/or ``transport_readonly_server``.
617
See the API docs for more details.
623
When writing a test for a feature, it is often necessary to set up a
624
branch with a certain history. The ``BranchBuilder`` interface allows the
625
creation of test branches in a quick and easy manner. Here's a sample
628
builder = self.make_branch_builder('relpath')
629
builder.build_commit()
630
builder.build_commit()
631
builder.build_commit()
632
branch = builder.get_branch()
634
``make_branch_builder`` is a method of ``TestCaseWithMemoryTransport``.
636
Note that many current tests create test branches by inheriting from
637
``TestCaseWithTransport`` and using the ``make_branch_and_tree`` helper to
638
give them a ``WorkingTree`` that they can commit to. However, using the
639
newer ``make_branch_builder`` helper is preferred, because it can build
640
the changes in memory, rather than on disk. Tests that are explictly
641
testing how we work with disk objects should, of course, use a real
644
Please see bzrlib.branchbuilder for more details.
646
If you're going to examine the commit timestamps e.g. in a test for log
647
output, you should set the timestamp on the tree, rather than using fuzzy
654
The ``TreeBuilder`` interface allows the construction of arbitrary trees
655
with a declarative interface. A sample session might look like::
657
tree = self.make_branch_and_tree('path')
658
builder = TreeBuilder()
659
builder.start_tree(tree)
660
builder.build(['foo', "bar/", "bar/file"])
661
tree.commit('commit the tree')
662
builder.finish_tree()
664
Usually a test will create a tree using ``make_branch_and_memory_tree`` (a
665
method of ``TestCaseWithMemoryTransport``) or ``make_branch_and_tree`` (a
666
method of ``TestCaseWithTransport``).
668
Please see bzrlib.treebuilder for more details.
671
.. |--| unicode:: U+2014
added
removed
doc/developers/testing.txt
