215
203
Navigating the Code Base
216
204
========================
218
.. Was at <http://bazaar-vcs.org/NewDeveloperIntroduction>
220
Some of the key files in this directory are:
223
The command you run to start Bazaar itself. This script is pretty
224
short and just does some checks then jumps into bzrlib.
227
This file covers a brief introduction to Bazaar and lists some of its
231
Summary of changes in each Bazaar release that can affect users or
235
Installs Bazaar system-wide or to your home directory. To perform
236
development work on Bazaar it is not required to run this file - you
237
can simply run the bzr command from the top level directory of your
238
development copy. Note: That if you run setup.py this will create a
239
'build' directory in your development branch. There's nothing wrong
240
with this but don't be confused by it. The build process puts a copy
241
of the main code base into this build directory, along with some other
242
files. You don't need to go in here for anything discussed in this
246
Possibly the most exciting folder of all, bzrlib holds the main code
247
base. This is where you will go to edit python files and contribute to
251
Holds documentation on a whole range of things on Bazaar from the
252
origination of ideas within the project to information on Bazaar
253
features and use cases. Within this directory there is a subdirectory
254
for each translation into a human language. All the documentation
255
is in the ReStructuredText markup language.
258
Documentation specifically targetted at Bazaar and plugin developers.
259
(Including this document.)
263
Automatically-generated API reference information is available at
264
<http://starship.python.net/crew/mwh/bzrlibapi/>.
265
(There is an experimental editable version at
266
<http://starship.python.net/crew/mwh/bzrlibapi-oe/>.)
267
See also the `Essential Domain Classes`_
268
section of this guide.
206
TODO: List and describe in one line the purpose of each directory
207
inside an installation of bzr.
209
TODO: Refer to a central location holding an up to date copy of the API
210
documentation generated by epydoc, e.g. something like
211
http://starship.python.net/crew/mwh/bzrlibapi/bzrlib.html.
217
The Importance of Testing
218
=========================
220
Reliability is a critical success factor for any Version Control System.
221
We want Bazaar to be highly reliable across multiple platforms while
222
evolving over time to meet the needs of its community.
224
In a nutshell, this is want we expect and encourage:
226
* New functionality should have test cases. Preferably write the
227
test before writing the code.
229
In general, you can test at either the command-line level or the
230
internal API level. See Writing tests below for more detail.
232
* Try to practice Test-Driven Development: before fixing a bug, write a
233
test case so that it does not regress. Similarly for adding a new
234
feature: write a test case for a small version of the new feature before
235
starting on the code itself. Check the test fails on the old code, then
236
add the feature or fix and check it passes.
238
By doing these things, the Bazaar team gets increased confidence that
239
changes do what they claim to do, whether provided by the core team or
240
by community members. Equally importantly, we can be surer that changes
241
down the track do not break new features or bug fixes that you are
244
As of May 2007, Bazaar ships with a test suite containing over 6000 tests
245
and growing. We are proud of it and want to remain so. As community
246
members, we all benefit from it. Would you trust version control on
247
your project to a product *without* a test suite like Bazaar has?
250
Running the Test Suite
251
======================
253
Currently, bzr selftest is used to invoke tests.
254
You can provide a pattern argument to run a subset. For example,
255
to run just the blackbox tests, run::
257
./bzr selftest -v blackbox
259
To skip a particular test (or set of tests), use the --exclude option
260
(shorthand -x) like so::
262
./bzr selftest -v -x blackbox
264
To ensure that all tests are being run and succeeding, you can use the
265
--strict option which will fail if there are any missing features or known
268
./bzr selftest --strict
270
To list tests without running them, use the --list-only option like so::
272
./bzr selftest --list-only
274
This option can be combined with other selftest options (like -x) and
275
filter patterns to understand their effect.
281
In general tests should be placed in a file named test_FOO.py where
282
FOO is the logical thing under test. That file should be placed in the
283
tests subdirectory under the package being tested.
285
For example, tests for merge3 in bzrlib belong in bzrlib/tests/test_merge3.py.
286
See bzrlib/tests/test_sampler.py for a template test script.
288
Tests can be written for the UI or for individual areas of the library.
289
Choose whichever is appropriate: if adding a new command, or a new command
290
option, then you should be writing a UI test. If you are both adding UI
291
functionality and library functionality, you will want to write tests for
292
both the UI and the core behaviours. We call UI tests 'blackbox' tests
293
and they are found in ``bzrlib/tests/blackbox/*.py``.
295
When writing blackbox tests please honour the following conventions:
297
1. Place the tests for the command 'name' in
298
bzrlib/tests/blackbox/test_name.py. This makes it easy for developers
299
to locate the test script for a faulty command.
301
2. Use the 'self.run_bzr("name")' utility function to invoke the command
302
rather than running bzr in a subprocess or invoking the
303
cmd_object.run() method directly. This is a lot faster than
304
subprocesses and generates the same logging output as running it in a
305
subprocess (which invoking the method directly does not).
307
3. Only test the one command in a single test script. Use the bzrlib
308
library when setting up tests and when evaluating the side-effects of
309
the command. We do this so that the library api has continual pressure
310
on it to be as functional as the command line in a simple manner, and
311
to isolate knock-on effects throughout the blackbox test suite when a
312
command changes its name or signature. Ideally only the tests for a
313
given command are affected when a given command is changed.
315
4. If you have a test which does actually require running bzr in a
316
subprocess you can use ``run_bzr_subprocess``. By default the spawned
317
process will not load plugins unless ``--allow-plugins`` is supplied.
323
We make selective use of doctests__. In general they should provide
324
*examples* within the API documentation which can incidentally be tested. We
325
don't try to test every important case using doctests -- regular Python
326
tests are generally a better solution.
328
Most of these are in ``bzrlib/doc/api``. More additions are welcome.
330
__ http://docs.python.org/lib/module-doctest.html
333
Skipping tests and test requirements
334
------------------------------------
336
In our enhancements to unittest we allow for some addition results beyond
337
just success or failure.
339
If a test can't be run, it can say that it's skipped. This is typically
340
used in parameterized tests - for example if a transport doesn't support
341
setting permissions, we'll skip the tests that relating to that. ::
344
return self.branch_format.initialize(repo.bzrdir)
345
except errors.UninitializableFormat:
346
raise tests.TestSkipped('Uninitializable branch format')
348
Raising TestSkipped is a good idea when you want to make it clear that the
349
test was not run, rather than just returning which makes it look as if it
352
Several different cases are distinguished:
355
Generic skip; the only type that was present up to bzr 0.18.
358
The test doesn't apply to the parameters with which it was run.
359
This is typically used when the test is being applied to all
360
implementations of an interface, but some aspects of the interface
361
are optional and not present in particular concrete
362
implementations. (Some tests that should raise this currently
363
either silently return or raise TestSkipped.) Another option is
364
to use more precise parameterization to avoid generating the test
368
**(Not implemented yet)**
369
The test can't be run because of an inherent limitation of the
370
environment, such as not having symlinks or not supporting
374
The test can't be run because a dependency (typically a Python
375
library) is not available in the test environment. These
376
are in general things that the person running the test could fix
377
by installing the library. It's OK if some of these occur when
378
an end user runs the tests or if we're specifically testing in a
379
limited environment, but a full test should never see them.
382
The test exists but is known to fail, for example because the
383
code to fix it hasn't been run yet. Raising this allows
384
you to distinguish these failures from the ones that are not
385
expected to fail. This could be conditionally raised if something
386
is broken on some platforms but not on others.
388
We plan to support three modes for running the test suite to control the
389
interpretation of these results. Strict mode is for use in situations
390
like merges to the mainline and releases where we want to make sure that
391
everything that can be tested has been tested. Lax mode is for use by
392
developers who want to temporarily tolerate some known failures. The
393
default behaviour is obtained by ``bzr selftest`` with no options, and
394
also (if possible) by running under another unittest harness.
396
======================= ======= ======= ========
397
result strict default lax
398
======================= ======= ======= ========
399
TestSkipped pass pass pass
400
TestNotApplicable pass pass pass
401
TestPlatformLimit pass pass pass
402
TestDependencyMissing fail pass pass
403
KnownFailure fail pass pass
404
======================= ======= ======= ========
407
Test feature dependencies
408
-------------------------
410
Rather than manually checking the environment in each test, a test class
411
can declare its dependence on some test features. The feature objects are
412
checked only once for each run of the whole test suite.
414
For historical reasons, as of May 2007 many cases that should depend on
415
features currently raise TestSkipped.)
419
class TestStrace(TestCaseWithTransport):
421
_test_needs_features = [StraceFeature]
423
This means all tests in this class need the feature. The feature itself
424
should provide a ``_probe`` method which is called once to determine if
427
These should generally be equivalent to either TestDependencyMissing or
428
sometimes TestPlatformLimit.
434
Known failures are when a test exists but we know it currently doesn't
435
work, allowing the test suite to still pass. These should be used with
436
care, we don't want a proliferation of quietly broken tests. It might be
437
appropriate to use them if you've committed a test for a bug but not the
438
fix for it, or if something works on Unix but not on Windows.
441
Testing exceptions and errors
442
-----------------------------
444
It's important to test handling of errors and exceptions. Because this
445
code is often not hit in ad-hoc testing it can often have hidden bugs --
446
it's particularly common to get NameError because the exception code
447
references a variable that has since been renamed.
449
.. TODO: Something about how to provoke errors in the right way?
451
In general we want to test errors at two levels:
453
1. A test in ``test_errors.py`` checking that when the exception object is
454
constructed with known parameters it produces an expected string form.
455
This guards against mistakes in writing the format string, or in the
456
``str`` representations of its parameters. There should be one for
457
each exception class.
459
2. Tests that when an api is called in a particular situation, it raises
460
an error of the expected class. You should typically use
461
``assertRaises``, which in the Bazaar test suite returns the exception
462
object to allow you to examine its parameters.
464
In some cases blackbox tests will also want to check error reporting. But
465
it can be difficult to provoke every error through the commandline
466
interface, so those tests are only done as needed -- eg in response to a
467
particular bug or if the error is reported in an unusual way(?) Blackbox
468
tests should mostly be testing how the command-line interface works, so
469
should only test errors if there is something particular to the cli in how
470
they're displayed or handled.
476
The Python ``warnings`` module is used to indicate a non-fatal code
477
problem. Code that's expected to raise a warning can be tested through
480
The test suite can be run with ``-Werror`` to check no unexpected errors
483
However, warnings should be used with discretion. It's not an appropriate
484
way to give messages to the user, because the warning is normally shown
485
only once per source line that causes the problem. You should also think
486
about whether the warning is serious enought that it should be visible to
487
users who may not be able to fix it.
490
Interface implementation testing and test scenarios
491
---------------------------------------------------
493
There are several cases in Bazaar of multiple implementations of a common
494
conceptual interface. ("Conceptual" because
495
it's not necessary for all the implementations to share a base class,
496
though they often do.) Examples include transports and the working tree,
497
branch and repository classes.
499
In these cases we want to make sure that every implementation correctly
500
fulfils the interface requirements. For example, every Transport should
501
support the ``has()`` and ``get()`` and ``clone()`` methods. We have a
502
sub-suite of tests in ``test_transport_implementations``. (Most
503
per-implementation tests are in submodules of ``bzrlib.tests``, but not
504
the transport tests at the moment.)
506
These tests are repeated for each registered Transport, by generating a
507
new TestCase instance for the cross product of test methods and transport
508
implementations. As each test runs, it has ``transport_class`` and
509
``transport_server`` set to the class it should test. Most tests don't
510
access these directly, but rather use ``self.get_transport`` which returns
511
a transport of the appropriate type.
513
The goal is to run per-implementation only tests that relate to that
514
particular interface. Sometimes we discover a bug elsewhere that happens
515
with only one particular transport. Once it's isolated, we can consider
516
whether a test should be added for that particular implementation,
517
or for all implementations of the interface.
519
The multiplication of tests for different implementations is normally
520
accomplished by overriding the ``test_suite`` function used to load
521
tests from a module. This function typically loads all the tests,
522
then applies a TestProviderAdapter to them, which generates a longer
523
suite containing all the test variations.
529
Some utilities are provided for generating variations of tests. This can
530
be used for per-implementation tests, or other cases where the same test
531
code needs to run several times on different scenarios.
533
The general approach is to define a class that provides test methods,
534
which depend on attributes of the test object being pre-set with the
535
values to which the test should be applied. The test suite should then
536
also provide a list of scenarios in which to run the tests.
538
Typically ``multiply_tests_from_modules`` should be called from the test
539
module's ``test_suite`` function.
271
542
Essential Domain Classes
618
914
being phased out.
621
Object string representations
622
=============================
624
Python prints objects using their ``__repr__`` method when they are
625
written to logs, exception tracebacks, or the debugger. We want
626
objects to have useful representations to help in determining what went
629
If you add a new class you should generally add a ``__repr__`` method
630
unless there is an adequate method in a parent class. There should be a
633
Representations should typically look like Python constructor syntax, but
634
they don't need to include every value in the object and they don't need
635
to be able to actually execute. They're to be read by humans, not
636
machines. Don't hardcode the classname in the format, so that we get the
637
correct value if the method is inherited by a subclass. If you're
638
printing attributes of the object, including strings, you should normally
639
use ``%r`` syntax (to call their repr in turn).
641
Try to avoid the representation becoming more than one or two lines long.
642
(But balance this against including useful information, and simplicity of
645
Because repr methods are often called when something has already gone
646
wrong, they should be written somewhat more defensively than most code.
647
The object may be half-initialized or in some other way in an illegal
648
state. The repr method shouldn't raise an exception, or it may hide the
649
(probably more useful) underlying exception.
654
return '%s(%r)' % (self.__class__.__name__,
661
A bare ``except`` statement will catch all exceptions, including ones that
662
really should terminate the program such as ``MemoryError`` and
663
``KeyboardInterrupt``. They should rarely be used unless the exception is
664
later re-raised. Even then, think about whether catching just
665
``Exception`` (which excludes system errors in Python2.5 and later) would
672
All code should be exercised by the test suite. See `Guide to Testing
673
Bazaar <testing.html>`_ for detailed information about writing tests.
682
We have a commitment to 6 months API stability - any supported symbol in a
683
release of bzr MUST NOT be altered in any way that would result in
684
breaking existing code that uses it. That means that method names,
685
parameter ordering, parameter names, variable and attribute names etc must
686
not be changed without leaving a 'deprecated forwarder' behind. This even
687
applies to modules and classes.
689
If you wish to change the behaviour of a supported API in an incompatible
690
way, you need to change its name as well. For instance, if I add an optional keyword
691
parameter to branch.commit - that's fine. On the other hand, if I add a
692
keyword parameter to branch.commit which is a *required* transaction
693
object, I should rename the API - i.e. to 'branch.commit_transaction'.
695
When renaming such supported API's, be sure to leave a deprecated_method (or
696
_function or ...) behind which forwards to the new API. See the
697
bzrlib.symbol_versioning module for decorators that take care of the
698
details for you - such as updating the docstring, and issuing a warning
699
when the old api is used.
701
For unsupported API's, it does not hurt to follow this discipline, but it's
702
not required. Minimally though, please try to rename things so that
703
callers will at least get an AttributeError rather than weird results.
706
Deprecation decorators
707
----------------------
709
``bzrlib.symbol_versioning`` provides decorators that can be attached to
710
methods, functions, and other interfaces to indicate that they should no
711
longer be used. For example::
713
@deprecated_method(deprecated_in((0, 1, 4)))
715
return self._new_foo()
717
To deprecate a static method you must call ``deprecated_function``
718
(**not** method), after the staticmethod call::
721
@deprecated_function(deprecated_in((0, 1, 4)))
722
def create_repository(base, shared=False, format=None):
724
When you deprecate an API, you should not just delete its tests, because
725
then we might introduce bugs in them. If the API is still present at all,
726
it should still work. The basic approach is to use
727
``TestCase.applyDeprecated`` which in one step checks that the API gives
728
the expected deprecation message, and also returns the real result from
729
the method, so that tests can keep running.
731
Deprecation warnings will be suppressed for final releases, but not for
732
development versions or release candidates, or when running ``bzr
733
selftest``. This gives developers information about whether their code is
734
using deprecated functions, but avoids confusing users about things they
998
In general tests should be placed in a file named test_FOO.py where
999
FOO is the logical thing under test. That file should be placed in the
1000
tests subdirectory under the package being tested.
1002
For example, tests for merge3 in bzrlib belong in bzrlib/tests/test_merge3.py.
1003
See bzrlib/tests/test_sampler.py for a template test script.
1005
Tests can be written for the UI or for individual areas of the library.
1006
Choose whichever is appropriate: if adding a new command, or a new command
1007
option, then you should be writing a UI test. If you are both adding UI
1008
functionality and library functionality, you will want to write tests for
1009
both the UI and the core behaviours. We call UI tests 'blackbox' tests
1010
and they are found in ``bzrlib/tests/blackbox/*.py``.
1012
When writing blackbox tests please honour the following conventions:
1014
1. Place the tests for the command 'name' in
1015
bzrlib/tests/blackbox/test_name.py. This makes it easy for developers
1016
to locate the test script for a faulty command.
1018
2. Use the 'self.run_bzr("name")' utility function to invoke the command
1019
rather than running bzr in a subprocess or invoking the
1020
cmd_object.run() method directly. This is a lot faster than
1021
subprocesses and generates the same logging output as running it in a
1022
subprocess (which invoking the method directly does not).
1024
3. Only test the one command in a single test script. Use the bzrlib
1025
library when setting up tests and when evaluating the side-effects of
1026
the command. We do this so that the library api has continual pressure
1027
on it to be as functional as the command line in a simple manner, and
1028
to isolate knock-on effects throughout the blackbox test suite when a
1029
command changes its name or signature. Ideally only the tests for a
1030
given command are affected when a given command is changed.
1032
4. If you have a test which does actually require running bzr in a
1033
subprocess you can use ``run_bzr_subprocess``. By default the spawned
1034
process will not load plugins unless ``--allow-plugins`` is supplied.
1040
We have a rich collection of tools to support writing tests. Please use
1041
them in preference to ad-hoc solutions as they provide portability and
1042
performance benefits.
1047
The ``TreeBuilder`` interface allows the construction of arbitrary trees
1048
with a declarative interface. A sample session might look like::
1050
tree = self.make_branch_and_tree('path')
1051
builder = TreeBuilder()
1052
builder.start_tree(tree)
1053
builder.build(['foo', "bar/", "bar/file"])
1054
tree.commit('commit the tree')
1055
builder.finish_tree()
1057
Please see bzrlib.treebuilder for more details.
1062
The ``BranchBuilder`` interface allows the creation of test branches in a
1063
quick and easy manner. A sample session::
1065
builder = BranchBuilder(self.get_transport().clone('relpath'))
1066
builder.build_commit()
1067
builder.build_commit()
1068
builder.build_commit()
1069
branch = builder.get_branch()
1071
Please see bzrlib.branchbuilder for more details.
1076
We make selective use of doctests__. In general they should provide
1077
*examples* within the API documentation which can incidentally be tested. We
1078
don't try to test every important case using doctests -- regular Python
1079
tests are generally a better solution.
1081
Most of these are in ``bzrlib/doc/api``. More additions are welcome.
1083
__ http://docs.python.org/lib/module-doctest.html
1088
Currently, bzr selftest is used to invoke tests.
1089
You can provide a pattern argument to run a subset. For example,
1090
to run just the blackbox tests, run::
1092
./bzr selftest -v blackbox
1094
To skip a particular test (or set of tests), use the --exclude option
1095
(shorthand -x) like so::
1097
./bzr selftest -v -x blackbox
1099
To list tests without running them, use the --list-only option like so::
1101
./bzr selftest --list-only
1103
This option can be combined with other selftest options (like -x) and
1104
filter patterns to understand their effect.
816
1107
Handling Errors and Exceptions
817
1108
==============================
added
removed
doc/developers/HACKING.txt
