264
191
<http://starship.python.net/crew/mwh/bzrlibapi/>.
265
192
(There is an experimental editable version at
266
193
<http://starship.python.net/crew/mwh/bzrlibapi-oe/>.)
267
See also the `Essential Domain Classes`_
268
section of this guide.
274
The Importance of Testing
275
=========================
277
Reliability is a critical success factor for any Version Control System.
278
We want Bazaar to be highly reliable across multiple platforms while
279
evolving over time to meet the needs of its community.
281
In a nutshell, this is want we expect and encourage:
283
* New functionality should have test cases. Preferably write the
284
test before writing the code.
286
In general, you can test at either the command-line level or the
287
internal API level. See Writing tests below for more detail.
289
* Try to practice Test-Driven Development: before fixing a bug, write a
290
test case so that it does not regress. Similarly for adding a new
291
feature: write a test case for a small version of the new feature before
292
starting on the code itself. Check the test fails on the old code, then
293
add the feature or fix and check it passes.
295
By doing these things, the Bazaar team gets increased confidence that
296
changes do what they claim to do, whether provided by the core team or
297
by community members. Equally importantly, we can be surer that changes
298
down the track do not break new features or bug fixes that you are
301
As of May 2008, Bazaar ships with a test suite containing over 12000 tests
302
and growing. We are proud of it and want to remain so. As community
303
members, we all benefit from it. Would you trust version control on
304
your project to a product *without* a test suite like Bazaar has?
307
Running the Test Suite
308
======================
310
Currently, bzr selftest is used to invoke tests.
311
You can provide a pattern argument to run a subset. For example,
312
to run just the blackbox tests, run::
314
./bzr selftest -v blackbox
316
To skip a particular test (or set of tests), use the --exclude option
317
(shorthand -x) like so::
319
./bzr selftest -v -x blackbox
321
To ensure that all tests are being run and succeeding, you can use the
322
--strict option which will fail if there are any missing features or known
325
./bzr selftest --strict
327
To list tests without running them, use the --list-only option like so::
329
./bzr selftest --list-only
331
This option can be combined with other selftest options (like -x) and
332
filter patterns to understand their effect.
334
Once you understand how to create a list of tests, you can use the --load-list
335
option to run only a restricted set of tests that you kept in a file, one test
336
id by line. Keep in mind that this will never be sufficient to validate your
337
modifications, you still need to run the full test suite for that, but using it
338
can help in some cases (like running only the failed tests for some time)::
340
./bzr selftest -- load-list my_failing_tests
342
This option can also be combined with other selftest options, including
343
patterns. It has some drawbacks though, the list can become out of date pretty
344
quick when doing Test Driven Development.
346
To address this concern, there is another way to run a restricted set of tests:
347
the --starting-with option will run only the tests whose name starts with the
348
specified string. It will also avoid loading the other tests and as a
349
consequence starts running your tests quicker::
351
./bzr selftest --starting-with bzrlib.blackbox
353
This option can be combined with all the other selftest options including
354
--load-list. The later is rarely used but allows to run a subset of a list of
355
failing tests for example.
357
Test suite debug flags
358
----------------------
360
Similar to the global ``-Dfoo`` debug options, bzr selftest accepts
361
``-E=foo`` debug flags. These flags are:
363
:allow_debug: do *not* clear the global debug flags when running a test.
364
This can provide useful logging to help debug test failures when used
365
with e.g. ``bzr -Dhpss selftest -E=allow_debug``
371
In general tests should be placed in a file named test_FOO.py where
372
FOO is the logical thing under test. That file should be placed in the
373
tests subdirectory under the package being tested.
375
For example, tests for merge3 in bzrlib belong in bzrlib/tests/test_merge3.py.
376
See bzrlib/tests/test_sampler.py for a template test script.
378
Tests can be written for the UI or for individual areas of the library.
379
Choose whichever is appropriate: if adding a new command, or a new command
380
option, then you should be writing a UI test. If you are both adding UI
381
functionality and library functionality, you will want to write tests for
382
both the UI and the core behaviours. We call UI tests 'blackbox' tests
383
and they are found in ``bzrlib/tests/blackbox/*.py``.
385
When writing blackbox tests please honour the following conventions:
387
1. Place the tests for the command 'name' in
388
bzrlib/tests/blackbox/test_name.py. This makes it easy for developers
389
to locate the test script for a faulty command.
391
2. Use the 'self.run_bzr("name")' utility function to invoke the command
392
rather than running bzr in a subprocess or invoking the
393
cmd_object.run() method directly. This is a lot faster than
394
subprocesses and generates the same logging output as running it in a
395
subprocess (which invoking the method directly does not).
397
3. Only test the one command in a single test script. Use the bzrlib
398
library when setting up tests and when evaluating the side-effects of
399
the command. We do this so that the library api has continual pressure
400
on it to be as functional as the command line in a simple manner, and
401
to isolate knock-on effects throughout the blackbox test suite when a
402
command changes its name or signature. Ideally only the tests for a
403
given command are affected when a given command is changed.
405
4. If you have a test which does actually require running bzr in a
406
subprocess you can use ``run_bzr_subprocess``. By default the spawned
407
process will not load plugins unless ``--allow-plugins`` is supplied.
413
We make selective use of doctests__. In general they should provide
414
*examples* within the API documentation which can incidentally be tested. We
415
don't try to test every important case using doctests -- regular Python
416
tests are generally a better solution.
418
Most of these are in ``bzrlib/doc/api``. More additions are welcome.
420
__ http://docs.python.org/lib/module-doctest.html
423
Skipping tests and test requirements
424
------------------------------------
426
In our enhancements to unittest we allow for some addition results beyond
427
just success or failure.
429
If a test can't be run, it can say that it's skipped. This is typically
430
used in parameterized tests - for example if a transport doesn't support
431
setting permissions, we'll skip the tests that relating to that. ::
434
return self.branch_format.initialize(repo.bzrdir)
435
except errors.UninitializableFormat:
436
raise tests.TestSkipped('Uninitializable branch format')
438
Raising TestSkipped is a good idea when you want to make it clear that the
439
test was not run, rather than just returning which makes it look as if it
442
Several different cases are distinguished:
445
Generic skip; the only type that was present up to bzr 0.18.
448
The test doesn't apply to the parameters with which it was run.
449
This is typically used when the test is being applied to all
450
implementations of an interface, but some aspects of the interface
451
are optional and not present in particular concrete
452
implementations. (Some tests that should raise this currently
453
either silently return or raise TestSkipped.) Another option is
454
to use more precise parameterization to avoid generating the test
458
**(Not implemented yet)**
459
The test can't be run because of an inherent limitation of the
460
environment, such as not having symlinks or not supporting
464
The test can't be run because a dependency (typically a Python
465
library) is not available in the test environment. These
466
are in general things that the person running the test could fix
467
by installing the library. It's OK if some of these occur when
468
an end user runs the tests or if we're specifically testing in a
469
limited environment, but a full test should never see them.
472
The test exists but is known to fail, for example because the
473
code to fix it hasn't been run yet. Raising this allows
474
you to distinguish these failures from the ones that are not
475
expected to fail. This could be conditionally raised if something
476
is broken on some platforms but not on others.
478
If the test would fail because of something we don't expect or
479
intend to fix, KnownFailure is not appropriate, and
480
TestNotApplicable might be better.
482
We plan to support three modes for running the test suite to control the
483
interpretation of these results. Strict mode is for use in situations
484
like merges to the mainline and releases where we want to make sure that
485
everything that can be tested has been tested. Lax mode is for use by
486
developers who want to temporarily tolerate some known failures. The
487
default behaviour is obtained by ``bzr selftest`` with no options, and
488
also (if possible) by running under another unittest harness.
490
======================= ======= ======= ========
491
result strict default lax
492
======================= ======= ======= ========
493
TestSkipped pass pass pass
494
TestNotApplicable pass pass pass
495
TestPlatformLimit pass pass pass
496
TestDependencyMissing fail pass pass
497
KnownFailure fail pass pass
498
======================= ======= ======= ========
501
Test feature dependencies
502
-------------------------
504
Rather than manually checking the environment in each test, a test class
505
can declare its dependence on some test features. The feature objects are
506
checked only once for each run of the whole test suite.
508
For historical reasons, as of May 2007 many cases that should depend on
509
features currently raise TestSkipped.)
513
class TestStrace(TestCaseWithTransport):
515
_test_needs_features = [StraceFeature]
517
This means all tests in this class need the feature. The feature itself
518
should provide a ``_probe`` method which is called once to determine if
521
These should generally be equivalent to either TestDependencyMissing or
522
sometimes TestPlatformLimit.
528
Known failures are when a test exists but we know it currently doesn't
529
work, allowing the test suite to still pass. These should be used with
530
care, we don't want a proliferation of quietly broken tests. It might be
531
appropriate to use them if you've committed a test for a bug but not the
532
fix for it, or if something works on Unix but not on Windows.
535
Testing exceptions and errors
536
-----------------------------
538
It's important to test handling of errors and exceptions. Because this
539
code is often not hit in ad-hoc testing it can often have hidden bugs --
540
it's particularly common to get NameError because the exception code
541
references a variable that has since been renamed.
543
.. TODO: Something about how to provoke errors in the right way?
545
In general we want to test errors at two levels:
547
1. A test in ``test_errors.py`` checking that when the exception object is
548
constructed with known parameters it produces an expected string form.
549
This guards against mistakes in writing the format string, or in the
550
``str`` representations of its parameters. There should be one for
551
each exception class.
553
2. Tests that when an api is called in a particular situation, it raises
554
an error of the expected class. You should typically use
555
``assertRaises``, which in the Bazaar test suite returns the exception
556
object to allow you to examine its parameters.
558
In some cases blackbox tests will also want to check error reporting. But
559
it can be difficult to provoke every error through the commandline
560
interface, so those tests are only done as needed -- eg in response to a
561
particular bug or if the error is reported in an unusual way(?) Blackbox
562
tests should mostly be testing how the command-line interface works, so
563
should only test errors if there is something particular to the cli in how
564
they're displayed or handled.
570
The Python ``warnings`` module is used to indicate a non-fatal code
571
problem. Code that's expected to raise a warning can be tested through
574
The test suite can be run with ``-Werror`` to check no unexpected errors
577
However, warnings should be used with discretion. It's not an appropriate
578
way to give messages to the user, because the warning is normally shown
579
only once per source line that causes the problem. You should also think
580
about whether the warning is serious enought that it should be visible to
581
users who may not be able to fix it.
584
Interface implementation testing and test scenarios
585
---------------------------------------------------
587
There are several cases in Bazaar of multiple implementations of a common
588
conceptual interface. ("Conceptual" because
589
it's not necessary for all the implementations to share a base class,
590
though they often do.) Examples include transports and the working tree,
591
branch and repository classes.
593
In these cases we want to make sure that every implementation correctly
594
fulfils the interface requirements. For example, every Transport should
595
support the ``has()`` and ``get()`` and ``clone()`` methods. We have a
596
sub-suite of tests in ``test_transport_implementations``. (Most
597
per-implementation tests are in submodules of ``bzrlib.tests``, but not
598
the transport tests at the moment.)
600
These tests are repeated for each registered Transport, by generating a
601
new TestCase instance for the cross product of test methods and transport
602
implementations. As each test runs, it has ``transport_class`` and
603
``transport_server`` set to the class it should test. Most tests don't
604
access these directly, but rather use ``self.get_transport`` which returns
605
a transport of the appropriate type.
607
The goal is to run per-implementation only tests that relate to that
608
particular interface. Sometimes we discover a bug elsewhere that happens
609
with only one particular transport. Once it's isolated, we can consider
610
whether a test should be added for that particular implementation,
611
or for all implementations of the interface.
613
The multiplication of tests for different implementations is normally
614
accomplished by overriding the ``test_suite`` function used to load
615
tests from a module. This function typically loads all the tests,
616
then applies a TestProviderAdapter to them, which generates a longer
617
suite containing all the test variations.
623
Some utilities are provided for generating variations of tests. This can
624
be used for per-implementation tests, or other cases where the same test
625
code needs to run several times on different scenarios.
627
The general approach is to define a class that provides test methods,
628
which depend on attributes of the test object being pre-set with the
629
values to which the test should be applied. The test suite should then
630
also provide a list of scenarios in which to run the tests.
632
Typically ``multiply_tests_from_modules`` should be called from the test
633
module's ``test_suite`` function.
636
Essential Domain Classes
637
########################
639
Introducing the Object Model
640
============================
642
The core domain objects within the bazaar model are:
652
Transports are explained below. See http://bazaar-vcs.org/Classes/
653
for an introduction to the other key classes.
658
The ``Transport`` layer handles access to local or remote directories.
659
Each Transport object acts like a logical connection to a particular
660
directory, and it allows various operations on files within it. You can
661
*clone* a transport to get a new Transport connected to a subdirectory or
664
Transports are not used for access to the working tree. At present
665
working trees are always local and they are accessed through the regular
666
Python file io mechanisms.
671
Transports work in URLs. Take note that URLs are by definition only
672
ASCII - the decision of how to encode a Unicode string into a URL must be
673
taken at a higher level, typically in the Store. (Note that Stores also
674
escape filenames which cannot be safely stored on all filesystems, but
675
this is a different level.)
677
The main reason for this is that it's not possible to safely roundtrip a
678
URL into Unicode and then back into the same URL. The URL standard
679
gives a way to represent non-ASCII bytes in ASCII (as %-escapes), but
680
doesn't say how those bytes represent non-ASCII characters. (They're not
681
guaranteed to be UTF-8 -- that is common but doesn't happen everywhere.)
683
For example if the user enters the url ``http://example/%e0`` there's no
684
way to tell whether that character represents "latin small letter a with
685
grave" in iso-8859-1, or "latin small letter r with acute" in iso-8859-2
686
or malformed UTF-8. So we can't convert their URL to Unicode reliably.
688
Equally problematic if we're given a url-like string containing non-ascii
689
characters (such as the accented a) we can't be sure how to convert that
690
to the correct URL, because we don't know what encoding the server expects
691
for those characters. (Although this is not totally reliable we might still
692
accept these and assume they should be put into UTF-8.)
694
A similar edge case is that the url ``http://foo/sweet%2Fsour`` contains
695
one directory component whose name is "sweet/sour". The escaped slash is
696
not a directory separator. If we try to convert URLs to regular Unicode
697
paths this information will be lost.
699
This implies that Transports must natively deal with URLs; for simplicity
700
they *only* deal with URLs and conversion of other strings to URLs is done
701
elsewhere. Information they return, such as from ``list_dir``, is also in
702
the form of URL components.
195
See also the `Bazaar Architectural Overview <../../developers/overview.html>`_.
198
The Code Review Process
199
#######################
201
All code changes coming in to Bazaar are reviewed by someone else.
202
Normally changes by core contributors are reviewed by one other core
203
developer, and changes from other people are reviewed by two core
204
developers. Use intelligent discretion if the patch is trivial.
206
Good reviews do take time. They also regularly require a solid
207
understanding of the overall code base. In practice, this means a small
208
number of people often have a large review burden - with knowledge comes
209
responsibility. No one like their merge requests sitting in a queue going
210
nowhere, so reviewing sooner rather than later is strongly encouraged.
214
Sending patches for review
215
==========================
217
If you'd like to propose a change, please post to the
218
bazaar@lists.canonical.com list with a bundle, patch, or link to a
219
branch. Put ``[PATCH]`` or ``[MERGE]`` in the subject so Bundle Buggy
220
can pick it out, and explain the change in the email message text.
221
Remember to update the NEWS file as part of your change if it makes any
222
changes visible to users or plugin developers. Please include a diff
223
against mainline if you're giving a link to a branch.
225
You can generate a merge request like this::
227
bzr send -o bug-1234.diff
229
A ``.diff`` extension is recommended instead of .bundle as many mail clients
230
will send the latter as a binary file.
232
``bzr send`` can also send mail directly if you prefer; see the help.
234
Please do **NOT** put [PATCH] or [MERGE] in the subject line if you don't
235
want it to be merged. If you want comments from developers rather than
236
to be merged, you can put ``[RFC]`` in the subject line.
238
If this change addresses a bug, please put the bug number in the subject
239
line too, in the form ``[#1]`` so that Bundle Buggy can recognize it.
241
If the change is intended for a particular release mark that in the
242
subject too, e.g. ``[1.6]``.
248
Please put a "cover letter" on your merge request explaining:
250
* the reason **why** you're making this change
252
* **how** this change acheives this purpose
254
* anything else you may have fixed in passing
256
* anything significant that you thought of doing, such as a more
257
extensive fix or a different approach, but didn't or couldn't do now
259
A good cover letter makes reviewers' lives easier because they can decide
260
from the letter whether they agree with the purpose and approach, and then
261
assess whether the patch actually does what the cover letter says.
262
Explaining any "drive-by fixes" or roads not taken may also avoid queries
263
from the reviewer. All in all this should give faster and better reviews.
264
Sometimes writing the cover letter helps the submitter realize something
265
else they need to do. The size of the cover letter should be proportional
266
to the size and complexity of the patch.
269
Reviewing proposed changes
270
==========================
272
Anyone is welcome to review code, and reply to the thread with their
275
The simplest way to review a proposed change is to just read the patch on
276
the list or in Bundle Buggy. For more complex changes it may be useful
277
to make a new working tree or branch from trunk, and merge the proposed
278
change into it, so you can experiment with the code or look at a wider
281
There are three main requirements for code to get in:
283
* Doesn't reduce test coverage: if it adds new methods or commands,
284
there should be tests for them. There is a good test framework
285
and plenty of examples to crib from, but if you are having trouble
286
working out how to test something feel free to post a draft patch
289
* Doesn't reduce design clarity, such as by entangling objects
290
we're trying to separate. This is mostly something the more
291
experienced reviewers need to help check.
293
* Improves bugs, features, speed, or code simplicity.
295
Code that goes in should not degrade any of these aspects. Patches are
296
welcome that only cleanup the code without changing the external
297
behaviour. The core developers take care to keep the code quality high
298
and understandable while recognising that perfect is sometimes the enemy
301
It is easy for reviews to make people notice other things which should be
302
fixed but those things should not hold up the original fix being accepted.
303
New things can easily be recorded in the Bug Tracker instead.
305
It's normally much easier to review several smaller patches than one large
306
one. You might want to use ``bzr-loom`` to maintain threads of related
307
work, or submit a preparatory patch that will make your "real" change
311
Checklist for reviewers
312
=======================
314
* Do you understand what the code's doing and why?
316
* Will it perform reasonably for large inputs, both in memory size and
317
run time? Are there some scenarios where performance should be
320
* Is it tested, and are the tests at the right level? Are there both
321
blackbox (command-line level) and API-oriented tests?
323
* If this change will be visible to end users or API users, is it
324
appropriately documented in NEWS?
326
* Does it meet the coding standards below?
328
* If it changes the user-visible behaviour, does it update the help
329
strings and user documentation?
331
* If it adds a new major concept or standard practice, does it update the
332
developer documentation?
334
* (your ideas here...)
337
Bundle Buggy and review outcomes
338
================================
340
Anyone can "vote" on the mailing list by expressing an opinion. Core
341
developers can also vote using Bundle Buggy. Here are the voting codes and
344
:approve: Reviewer wants this submission merged.
345
:tweak: Reviewer wants this submission merged with small changes. (No
347
:abstain: Reviewer does not intend to vote on this patch.
348
:resubmit: Please make changes and resubmit for review.
349
:reject: Reviewer doesn't want this kind of change merged.
350
:comment: Not really a vote. Reviewer just wants to comment, for now.
352
If a change gets two approvals from core reviewers, and no rejections,
353
then it's OK to come in. Any of the core developers can bring it into the
354
bzr.dev trunk and backport it to maintenance branches if required. The
355
Release Manager will merge the change into the branch for a pending
356
release, if any. As a guideline, core developers usually merge their own
357
changes and volunteer to merge other contributions if they were the second
358
reviewer to agree to a change.
360
To track the progress of proposed changes, use Bundle Buggy. See
361
http://bundlebuggy.aaronbentley.com/help for a link to all the
362
outstanding merge requests together with an explanation of the columns.
363
Bundle Buggy will also mail you a link to track just your change.
705
365
Coding Style Guidelines
706
366
#######################