256
251
Holds documentation on a whole range of things on Bazaar from the
257
252
origination of ideas within the project to information on Bazaar
258
253
features and use cases. Within this directory there is a subdirectory
259
for each translation into a human language. All the documentation
254
for each translation into a human language. All the documentation
260
255
is in the ReStructuredText markup language.
263
Documentation specifically targeted at Bazaar and plugin developers.
258
Documentation specifically targetted at Bazaar and plugin developers.
264
259
(Including this document.)
268
Automatically-generated API reference information is available at
269
<http://starship.python.net/crew/mwh/bzrlibapi/>.
271
See also the `Bazaar Architectural Overview
272
<http://doc.bazaar-vcs.org/developers/overview.html>`_.
275
The Code Review Process
276
#######################
278
All code changes coming in to Bazaar are reviewed by someone else.
279
Normally changes by core contributors are reviewed by one other core
280
developer, and changes from other people are reviewed by two core
281
developers. Use intelligent discretion if the patch is trivial.
283
Good reviews do take time. They also regularly require a solid
284
understanding of the overall code base. In practice, this means a small
285
number of people often have a large review burden - with knowledge comes
286
responsibility. No one likes their merge requests sitting in a queue going
287
nowhere, so reviewing sooner rather than later is strongly encouraged.
296
Please put a "cover letter" on your merge request explaining:
298
* the reason **why** you're making this change
300
* **how** this change achieves this purpose
302
* anything else you may have fixed in passing
304
* anything significant that you thought of doing, such as a more
305
extensive fix or a different approach, but didn't or couldn't do now
307
A good cover letter makes reviewers' lives easier because they can decide
308
from the letter whether they agree with the purpose and approach, and then
309
assess whether the patch actually does what the cover letter says.
310
Explaining any "drive-by fixes" or roads not taken may also avoid queries
311
from the reviewer. All in all this should give faster and better reviews.
312
Sometimes writing the cover letter helps the submitter realize something
313
else they need to do. The size of the cover letter should be proportional
314
to the size and complexity of the patch.
317
Reviewing proposed changes
318
==========================
320
Anyone is welcome to review code, and reply to the thread with their
323
The simplest way to review a proposed change is to just read the patch on
324
the list or in Bundle Buggy. For more complex changes it may be useful
325
to make a new working tree or branch from trunk, and merge the proposed
326
change into it, so you can experiment with the code or look at a wider
329
There are three main requirements for code to get in:
331
* Doesn't reduce test coverage: if it adds new methods or commands,
332
there should be tests for them. There is a good test framework
333
and plenty of examples to crib from, but if you are having trouble
334
working out how to test something feel free to post a draft patch
337
* Doesn't reduce design clarity, such as by entangling objects
338
we're trying to separate. This is mostly something the more
339
experienced reviewers need to help check.
341
* Improves bugs, features, speed, or code simplicity.
343
Code that goes in should not degrade any of these aspects. Patches are
344
welcome that only cleanup the code without changing the external
345
behaviour. The core developers take care to keep the code quality high
346
and understandable while recognising that perfect is sometimes the enemy
349
It is easy for reviews to make people notice other things which should be
350
fixed but those things should not hold up the original fix being accepted.
351
New things can easily be recorded in the Bug Tracker instead.
353
It's normally much easier to review several smaller patches than one large
354
one. You might want to use ``bzr-loom`` to maintain threads of related
355
work, or submit a preparatory patch that will make your "real" change
359
Checklist for reviewers
360
=======================
362
* Do you understand what the code's doing and why?
364
* Will it perform reasonably for large inputs, both in memory size and
365
run time? Are there some scenarios where performance should be
368
* Is it tested, and are the tests at the right level? Are there both
369
blackbox (command-line level) and API-oriented tests?
371
* If this change will be visible to end users or API users, is it
372
appropriately documented in NEWS?
374
* Does it meet the coding standards below?
376
* If it changes the user-visible behaviour, does it update the help
377
strings and user documentation?
379
* If it adds a new major concept or standard practice, does it update the
380
developer documentation?
382
* (your ideas here...)
388
From May 2009 on, we prefer people to propose code reviews through
391
* <https://launchpad.net/+tour/code-review>
393
* <https://help.launchpad.net/Code/Review>
395
Anyone can propose or comment on a merge proposal just by creating a
398
There are two ways to create a new merge proposal: through the web
399
interface or by email.
402
Proposing a merge through the web
403
---------------------------------
405
To create the proposal through the web, first push your branch to Launchpad.
406
For example, a branch dealing with documentation belonging to the Launchpad
407
User mbp could be pushed as ::
409
bzr push lp:~mbp/bzr/doc
411
Then go to the branch's web page, which in this case would be
412
<https://code.launchpad.net/~mbp/bzr/doc>. You can simplify this step by just
417
You can then click "Propose for merging into another branch", and enter your
418
cover letter (see above) into the web form. Typically you'll want to merge
419
into ``~bzr/bzr/trunk`` which will be the default; you might also want to
420
nominate merging into a release branch for a bug fix. There is the option to
421
specify a specific reviewer or type of review, and you shouldn't normally
424
Submitting the form takes you to the new page about the merge proposal
425
containing the diff of the changes, comments by interested people, and
426
controls to comment or vote on the change.
428
Proposing a merge by mail
263
Automatically-generated API reference information is available at
264
<http://bazaar-vcs.org/BzrLib>. See also the `Essential Domain Classes`_
265
section of this guide.
271
The Importance of Testing
272
=========================
274
Reliability is a critical success factor for any Version Control System.
275
We want Bazaar to be highly reliable across multiple platforms while
276
evolving over time to meet the needs of its community.
278
In a nutshell, this is want we expect and encourage:
280
* New functionality should have test cases. Preferably write the
281
test before writing the code.
283
In general, you can test at either the command-line level or the
284
internal API level. See Writing tests below for more detail.
286
* Try to practice Test-Driven Development: before fixing a bug, write a
287
test case so that it does not regress. Similarly for adding a new
288
feature: write a test case for a small version of the new feature before
289
starting on the code itself. Check the test fails on the old code, then
290
add the feature or fix and check it passes.
292
By doing these things, the Bazaar team gets increased confidence that
293
changes do what they claim to do, whether provided by the core team or
294
by community members. Equally importantly, we can be surer that changes
295
down the track do not break new features or bug fixes that you are
298
As of May 2008, Bazaar ships with a test suite containing over 12000 tests
299
and growing. We are proud of it and want to remain so. As community
300
members, we all benefit from it. Would you trust version control on
301
your project to a product *without* a test suite like Bazaar has?
304
Running the Test Suite
305
======================
307
Currently, bzr selftest is used to invoke tests.
308
You can provide a pattern argument to run a subset. For example,
309
to run just the blackbox tests, run::
311
./bzr selftest -v blackbox
313
To skip a particular test (or set of tests), use the --exclude option
314
(shorthand -x) like so::
316
./bzr selftest -v -x blackbox
318
To ensure that all tests are being run and succeeding, you can use the
319
--strict option which will fail if there are any missing features or known
322
./bzr selftest --strict
324
To list tests without running them, use the --list-only option like so::
326
./bzr selftest --list-only
328
This option can be combined with other selftest options (like -x) and
329
filter patterns to understand their effect.
331
Once you understand how to create a list of tests, you can use the --load-list
332
option to run only a restricted set of tests that you kept in a file, one test
333
id by line. Keep in mind that this will never be sufficient to validate your
334
modifications, you still need to run the full test suite for that, but using it
335
can help in some cases (like running only the failed tests for some time)::
337
./bzr selftest -- load-list my_failing_tests
339
This option can also be combined with other selftest options, including
340
patterns. It has some drawbacks though, the list can become out of date pretty
341
quick when doing Test Driven Development.
343
To address this concern, there is another way to run a restricted set of tests:
344
the --starting-with option will run only the tests whose name starts with the
345
specified string. It will also avoid loading the other tests and as a
346
consequence starts running your tests quicker::
348
./bzr selftest --starting-with bzrlib.blackbox
350
This option can be combined with all the other selftest options including
351
--load-list. The later is rarely used but allows to run a subset of a list of
352
failing tests for example.
354
Test suite debug flags
355
----------------------
357
Similar to the global ``-Dfoo`` debug options, bzr selftest accepts
358
``-E=foo`` debug flags. These flags are:
360
:allow_debug: do *not* clear the global debug flags when running a test.
361
This can provide useful logging to help debug test failures when used
362
with e.g. ``bzr -Dhpss selftest -E=allow_debug``
368
In general tests should be placed in a file named test_FOO.py where
369
FOO is the logical thing under test. That file should be placed in the
370
tests subdirectory under the package being tested.
372
For example, tests for merge3 in bzrlib belong in bzrlib/tests/test_merge3.py.
373
See bzrlib/tests/test_sampler.py for a template test script.
375
Tests can be written for the UI or for individual areas of the library.
376
Choose whichever is appropriate: if adding a new command, or a new command
377
option, then you should be writing a UI test. If you are both adding UI
378
functionality and library functionality, you will want to write tests for
379
both the UI and the core behaviours. We call UI tests 'blackbox' tests
380
and they are found in ``bzrlib/tests/blackbox/*.py``.
382
When writing blackbox tests please honour the following conventions:
384
1. Place the tests for the command 'name' in
385
bzrlib/tests/blackbox/test_name.py. This makes it easy for developers
386
to locate the test script for a faulty command.
388
2. Use the 'self.run_bzr("name")' utility function to invoke the command
389
rather than running bzr in a subprocess or invoking the
390
cmd_object.run() method directly. This is a lot faster than
391
subprocesses and generates the same logging output as running it in a
392
subprocess (which invoking the method directly does not).
394
3. Only test the one command in a single test script. Use the bzrlib
395
library when setting up tests and when evaluating the side-effects of
396
the command. We do this so that the library api has continual pressure
397
on it to be as functional as the command line in a simple manner, and
398
to isolate knock-on effects throughout the blackbox test suite when a
399
command changes its name or signature. Ideally only the tests for a
400
given command are affected when a given command is changed.
402
4. If you have a test which does actually require running bzr in a
403
subprocess you can use ``run_bzr_subprocess``. By default the spawned
404
process will not load plugins unless ``--allow-plugins`` is supplied.
410
We make selective use of doctests__. In general they should provide
411
*examples* within the API documentation which can incidentally be tested. We
412
don't try to test every important case using doctests -- regular Python
413
tests are generally a better solution.
415
Most of these are in ``bzrlib/doc/api``. More additions are welcome.
417
__ http://docs.python.org/lib/module-doctest.html
420
Skipping tests and test requirements
421
------------------------------------
423
In our enhancements to unittest we allow for some addition results beyond
424
just success or failure.
426
If a test can't be run, it can say that it's skipped. This is typically
427
used in parameterized tests - for example if a transport doesn't support
428
setting permissions, we'll skip the tests that relating to that. ::
431
return self.branch_format.initialize(repo.bzrdir)
432
except errors.UninitializableFormat:
433
raise tests.TestSkipped('Uninitializable branch format')
435
Raising TestSkipped is a good idea when you want to make it clear that the
436
test was not run, rather than just returning which makes it look as if it
439
Several different cases are distinguished:
442
Generic skip; the only type that was present up to bzr 0.18.
445
The test doesn't apply to the parameters with which it was run.
446
This is typically used when the test is being applied to all
447
implementations of an interface, but some aspects of the interface
448
are optional and not present in particular concrete
449
implementations. (Some tests that should raise this currently
450
either silently return or raise TestSkipped.) Another option is
451
to use more precise parameterization to avoid generating the test
455
**(Not implemented yet)**
456
The test can't be run because of an inherent limitation of the
457
environment, such as not having symlinks or not supporting
461
The test can't be run because a dependency (typically a Python
462
library) is not available in the test environment. These
463
are in general things that the person running the test could fix
464
by installing the library. It's OK if some of these occur when
465
an end user runs the tests or if we're specifically testing in a
466
limited environment, but a full test should never see them.
469
The test exists but is known to fail, for example because the
470
code to fix it hasn't been run yet. Raising this allows
471
you to distinguish these failures from the ones that are not
472
expected to fail. This could be conditionally raised if something
473
is broken on some platforms but not on others.
475
We plan to support three modes for running the test suite to control the
476
interpretation of these results. Strict mode is for use in situations
477
like merges to the mainline and releases where we want to make sure that
478
everything that can be tested has been tested. Lax mode is for use by
479
developers who want to temporarily tolerate some known failures. The
480
default behaviour is obtained by ``bzr selftest`` with no options, and
481
also (if possible) by running under another unittest harness.
483
======================= ======= ======= ========
484
result strict default lax
485
======================= ======= ======= ========
486
TestSkipped pass pass pass
487
TestNotApplicable pass pass pass
488
TestPlatformLimit pass pass pass
489
TestDependencyMissing fail pass pass
490
KnownFailure fail pass pass
491
======================= ======= ======= ========
494
Test feature dependencies
429
495
-------------------------
431
To propose a merge by mail, send a bundle to ``merge@code.launchpad.net``.
433
You can generate a merge request like this::
435
bzr send -o bug-1234.diff
437
``bzr send`` can also send mail directly if you prefer; see the help.
497
Rather than manually checking the environment in each test, a test class
498
can declare its dependence on some test features. The feature objects are
499
checked only once for each run of the whole test suite.
501
For historical reasons, as of May 2007 many cases that should depend on
502
features currently raise TestSkipped.)
506
class TestStrace(TestCaseWithTransport):
508
_test_needs_features = [StraceFeature]
510
This means all tests in this class need the feature. The feature itself
511
should provide a ``_probe`` method which is called once to determine if
514
These should generally be equivalent to either TestDependencyMissing or
515
sometimes TestPlatformLimit.
521
Known failures are when a test exists but we know it currently doesn't
522
work, allowing the test suite to still pass. These should be used with
523
care, we don't want a proliferation of quietly broken tests. It might be
524
appropriate to use them if you've committed a test for a bug but not the
525
fix for it, or if something works on Unix but not on Windows.
528
Testing exceptions and errors
529
-----------------------------
531
It's important to test handling of errors and exceptions. Because this
532
code is often not hit in ad-hoc testing it can often have hidden bugs --
533
it's particularly common to get NameError because the exception code
534
references a variable that has since been renamed.
536
.. TODO: Something about how to provoke errors in the right way?
538
In general we want to test errors at two levels:
540
1. A test in ``test_errors.py`` checking that when the exception object is
541
constructed with known parameters it produces an expected string form.
542
This guards against mistakes in writing the format string, or in the
543
``str`` representations of its parameters. There should be one for
544
each exception class.
546
2. Tests that when an api is called in a particular situation, it raises
547
an error of the expected class. You should typically use
548
``assertRaises``, which in the Bazaar test suite returns the exception
549
object to allow you to examine its parameters.
551
In some cases blackbox tests will also want to check error reporting. But
552
it can be difficult to provoke every error through the commandline
553
interface, so those tests are only done as needed -- eg in response to a
554
particular bug or if the error is reported in an unusual way(?) Blackbox
555
tests should mostly be testing how the command-line interface works, so
556
should only test errors if there is something particular to the cli in how
557
they're displayed or handled.
563
The Python ``warnings`` module is used to indicate a non-fatal code
564
problem. Code that's expected to raise a warning can be tested through
567
The test suite can be run with ``-Werror`` to check no unexpected errors
570
However, warnings should be used with discretion. It's not an appropriate
571
way to give messages to the user, because the warning is normally shown
572
only once per source line that causes the problem. You should also think
573
about whether the warning is serious enought that it should be visible to
574
users who may not be able to fix it.
577
Interface implementation testing and test scenarios
578
---------------------------------------------------
580
There are several cases in Bazaar of multiple implementations of a common
581
conceptual interface. ("Conceptual" because
582
it's not necessary for all the implementations to share a base class,
583
though they often do.) Examples include transports and the working tree,
584
branch and repository classes.
586
In these cases we want to make sure that every implementation correctly
587
fulfils the interface requirements. For example, every Transport should
588
support the ``has()`` and ``get()`` and ``clone()`` methods. We have a
589
sub-suite of tests in ``test_transport_implementations``. (Most
590
per-implementation tests are in submodules of ``bzrlib.tests``, but not
591
the transport tests at the moment.)
593
These tests are repeated for each registered Transport, by generating a
594
new TestCase instance for the cross product of test methods and transport
595
implementations. As each test runs, it has ``transport_class`` and
596
``transport_server`` set to the class it should test. Most tests don't
597
access these directly, but rather use ``self.get_transport`` which returns
598
a transport of the appropriate type.
600
The goal is to run per-implementation only tests that relate to that
601
particular interface. Sometimes we discover a bug elsewhere that happens
602
with only one particular transport. Once it's isolated, we can consider
603
whether a test should be added for that particular implementation,
604
or for all implementations of the interface.
606
The multiplication of tests for different implementations is normally
607
accomplished by overriding the ``test_suite`` function used to load
608
tests from a module. This function typically loads all the tests,
609
then applies a TestProviderAdapter to them, which generates a longer
610
suite containing all the test variations.
616
Some utilities are provided for generating variations of tests. This can
617
be used for per-implementation tests, or other cases where the same test
618
code needs to run several times on different scenarios.
620
The general approach is to define a class that provides test methods,
621
which depend on attributes of the test object being pre-set with the
622
values to which the test should be applied. The test suite should then
623
also provide a list of scenarios in which to run the tests.
625
Typically ``multiply_tests_from_modules`` should be called from the test
626
module's ``test_suite`` function.
629
Essential Domain Classes
630
########################
632
Introducing the Object Model
633
============================
635
The core domain objects within the bazaar model are:
645
Transports are explained below. See http://bazaar-vcs.org/Classes/
646
for an introduction to the other key classes.
651
The ``Transport`` layer handles access to local or remote directories.
652
Each Transport object acts like a logical connection to a particular
653
directory, and it allows various operations on files within it. You can
654
*clone* a transport to get a new Transport connected to a subdirectory or
657
Transports are not used for access to the working tree. At present
658
working trees are always local and they are accessed through the regular
659
Python file io mechanisms.
440
662
-----------------
442
From <https://code.launchpad.net/bzr/+activereviews> you can see all
443
currently active reviews, and choose one to comment on. This page also
444
shows proposals that are now approved and should be merged by someone with
448
Reviews through Bundle Buggy
449
============================
451
The Bundle Buggy tool used up to May 2009 is still available as a review
454
Sending patches for review
455
--------------------------
457
If you'd like to propose a change, please post to the
458
bazaar@lists.canonical.com list with a bundle, patch, or link to a
459
branch. Put ``[PATCH]`` or ``[MERGE]`` in the subject so Bundle Buggy
460
can pick it out, and explain the change in the email message text.
461
Remember to update the NEWS file as part of your change if it makes any
462
changes visible to users or plugin developers. Please include a diff
463
against mainline if you're giving a link to a branch.
465
You can generate a merge request like this::
467
bzr send -o bug-1234.patch
469
A ``.patch`` extension is recommended instead of .bundle as many mail clients
470
will send the latter as a binary file.
472
``bzr send`` can also send mail directly if you prefer; see the help.
474
Please do **NOT** put [PATCH] or [MERGE] in the subject line if you don't
475
want it to be merged. If you want comments from developers rather than
476
to be merged, you can put ``[RFC]`` in the subject line.
478
If this change addresses a bug, please put the bug number in the subject
479
line too, in the form ``[#1]`` so that Bundle Buggy can recognize it.
481
If the change is intended for a particular release mark that in the
482
subject too, e.g. ``[1.6]``.
483
Anyone can "vote" on the mailing list by expressing an opinion. Core
484
developers can also vote using Bundle Buggy. Here are the voting codes and
487
:approve: Reviewer wants this submission merged.
488
:tweak: Reviewer wants this submission merged with small changes. (No
490
:abstain: Reviewer does not intend to vote on this patch.
491
:resubmit: Please make changes and resubmit for review.
492
:reject: Reviewer doesn't want this kind of change merged.
493
:comment: Not really a vote. Reviewer just wants to comment, for now.
495
If a change gets two approvals from core reviewers, and no rejections,
496
then it's OK to come in. Any of the core developers can bring it into the
497
bzr.dev trunk and backport it to maintenance branches if required. The
498
Release Manager will merge the change into the branch for a pending
499
release, if any. As a guideline, core developers usually merge their own
500
changes and volunteer to merge other contributions if they were the second
501
reviewer to agree to a change.
503
To track the progress of proposed changes, use Bundle Buggy. See
504
http://bundlebuggy.aaronbentley.com/help for a link to all the
505
outstanding merge requests together with an explanation of the columns.
506
Bundle Buggy will also mail you a link to track just your change.
664
Transports work in URLs. Take note that URLs are by definition only
665
ASCII - the decision of how to encode a Unicode string into a URL must be
666
taken at a higher level, typically in the Store. (Note that Stores also
667
escape filenames which cannot be safely stored on all filesystems, but
668
this is a different level.)
670
The main reason for this is that it's not possible to safely roundtrip a
671
URL into Unicode and then back into the same URL. The URL standard
672
gives a way to represent non-ASCII bytes in ASCII (as %-escapes), but
673
doesn't say how those bytes represent non-ASCII characters. (They're not
674
guaranteed to be UTF-8 -- that is common but doesn't happen everywhere.)
676
For example if the user enters the url ``http://example/%e0`` there's no
677
way to tell whether that character represents "latin small letter a with
678
grave" in iso-8859-1, or "latin small letter r with acute" in iso-8859-2
679
or malformed UTF-8. So we can't convert their URL to Unicode reliably.
681
Equally problematic if we're given a url-like string containing non-ascii
682
characters (such as the accented a) we can't be sure how to convert that
683
to the correct URL, because we don't know what encoding the server expects
684
for those characters. (Although this is not totally reliable we might still
685
accept these and assume they should be put into UTF-8.)
687
A similar edge case is that the url ``http://foo/sweet%2Fsour`` contains
688
one directory component whose name is "sweet/sour". The escaped slash is
689
not a directory separator. If we try to convert URLs to regular Unicode
690
paths this information will be lost.
692
This implies that Transports must natively deal with URLs; for simplicity
693
they *only* deal with URLs and conversion of other strings to URLs is done
694
elsewhere. Information they return, such as from ``list_dir``, is also in
695
the form of URL components.
508
698
Coding Style Guidelines
509
699
#######################
added
removed
doc/developers/HACKING.txt
