7
(The current version of this document is available in the file
8
``doc/developers/HACKING.txt`` in the source tree, or at
9
http://doc.bazaar-vcs.org/bzr.dev/developers/HACKING.html)
15
Exploring the Bazaar Platform
16
=============================
18
Before making changes, it's a good idea to explore the work already
19
done by others. Perhaps the new feature or improvement you're looking
20
for is available in another plug-in already? If you find a bug,
21
perhaps someone else has already fixed it?
23
To answer these questions and more, take a moment to explore the
24
overall Bazaar Platform. Here are some links to browse:
26
* The Plugins page on the Wiki - http://bazaar-vcs.org/BzrPlugins
28
* The Bazaar product family on Launchpad - https://launchpad.net/bazaar
30
* Bug Tracker for the core product - https://bugs.launchpad.net/bzr/
32
* Blueprint Tracker for the core product - https://blueprints.launchpad.net/bzr/
34
If nothing else, perhaps you'll find inspiration in how other developers
35
have solved their challenges.
38
Planning and Discussing Changes
39
===============================
41
There is a very active community around Bazaar. Mostly we meet on IRC
42
(#bzr on irc.freenode.net) and on the mailing list. To join the Bazaar
43
community, see http://bazaar-vcs.org/BzrSupport.
45
If you are planning to make a change, it's a very good idea to mention it
46
on the IRC channel and/or on the mailing list. There are many advantages
47
to involving the community before you spend much time on a change.
50
* you get to build on the wisdom on others, saving time
52
* if others can direct you to similar code, it minimises the work to be done
54
* it assists everyone in coordinating direction, priorities and effort.
56
In summary, maximising the input from others typically minimises the
57
total effort required to get your changes merged. The community is
58
friendly, helpful and always keen to welcome newcomers.
61
Bazaar Development in a Nutshell
62
================================
64
Looking for a 10 minute introduction to submitting a change?
65
See http://bazaar-vcs.org/BzrGivingBack.
67
TODO: Merge that Wiki page into this document.
70
Understanding the Development Process
71
=====================================
73
The development team follows many best-practices including:
75
* a public roadmap and planning process in which anyone can participate
77
* time based milestones everyone can work towards and plan around
79
* extensive code review and feedback to contributors
81
* complete and rigorous test coverage on any code contributed
83
* automated validation that all tests still pass before code is merged
84
into the main code branch.
86
The key tools we use to enable these practices are:
88
* Launchpad - https://launchpad.net/
90
* Bazaar - http://bazaar-vcs.org/
92
* Bundle Buggy - http://bundlebuggy.aaronbentley.com/
94
* Patch Queue Manager - https://launchpad.net/pqm/
96
For further information, see http://bazaar-vcs.org/BzrDevelopment.
99
A Closer Look at the Merge & Review Process
100
===========================================
102
If you'd like to propose a change, please post to the
103
bazaar@lists.canonical.com list with a bundle, patch, or link to a
104
branch. Put '[PATCH]' or '[MERGE]' in the subject so Bundle Buggy
105
can pick it out, and explain the change in the email message text.
106
Remember to update the NEWS file as part of your change if it makes any
107
changes visible to users or plugin developers. Please include a diff
108
against mainline if you're giving a link to a branch.
110
You can generate a bundle like this::
112
bzr bundle > mybundle.patch
114
A .patch extension is recommended instead of .bundle as many mail clients
115
will send the latter as a binary file. If a bundle would be too long or your
116
mailer mangles whitespace (e.g. implicitly converts Unix newlines to DOS
117
newlines), use the merge-directive command instead like this::
119
bzr merge-directive http://bazaar-vcs.org http://example.org/my_branch > my_directive.patch
121
See the help for details on the arguments to merge-directive.
123
Please do **NOT** put [PATCH] or [MERGE] in the subject line if you don't
124
want it to be merged. If you want comments from developers rather than
125
to be merged, you can put '[RFC]' in the subject line.
127
Anyone is welcome to review code. There are broadly three gates for
130
* Doesn't reduce test coverage: if it adds new methods or commands,
131
there should be tests for them. There is a good test framework
132
and plenty of examples to crib from, but if you are having trouble
133
working out how to test something feel free to post a draft patch
136
* Doesn't reduce design clarity, such as by entangling objects
137
we're trying to separate. This is mostly something the more
138
experienced reviewers need to help check.
140
* Improves bugs, features, speed, or code simplicity.
142
Code that goes in should pass all three. The core developers take care
143
to keep the code quality high and understandable while recognising that
144
perfect is sometimes the enemy of good. (It is easy for reviews to make
145
people notice other things which should be fixed but those things should
146
not hold up the original fix being accepted. New things can easily be
147
recorded in the Bug Tracker instead.)
149
Anyone can "vote" on the mailing list. Core developers can also vote using
150
Bundle Buggy. Here are the voting codes and their explanations.
152
:approve: Reviewer wants this submission merged.
153
:tweak: Reviewer wants this submission merged with small changes. (No
155
:abstain: Reviewer does not intend to vote on this patch.
156
:resubmit: Please make changes and resubmit for review.
157
:reject: Reviewer doesn't want this kind of change merged.
158
:comment: Not really a vote. Reviewer just wants to comment, for now.
160
If a change gets two approvals from core reviewers, and no rejections,
161
then it's OK to come in. Any of the core developers can bring it into the
162
bzr.dev trunk and backport it to maintenance branches if required. The
163
Release Manager will merge the change into the branch for a pending
164
release, if any. As a guideline, core developers usually merge their own
165
changes and volunteer to merge other contributions if they were the second
166
reviewer to agree to a change.
168
To track the progress of proposed changes, use Bundle Buggy. See
169
http://bundlebuggy.aaronbentley.com/help for a link to all the
170
outstanding merge requests together with an explanation of the columns.
171
Bundle Buggy will also mail you a link to track just your change.
174
Preparing a Sandbox for Making Changes to Bazaar
175
================================================
177
Bazaar supports many ways of organising your work. See
178
http://bazaar-vcs.org/SharedRepositoryLayouts for a summary of the
179
popular alternatives.
181
Of course, the best choice for you will depend on numerous factors:
182
the number of changes you may be making, the complexity of the changes, etc.
183
As a starting suggestion though:
185
* create a local copy of the main development branch (bzr.dev) by using
188
bzr branch http://bazaar-vcs.org/bzr/bzr.dev/ bzr.dev
190
* keep your copy of bzr.dev prestine (by not developing in it) and keep
191
it up to date (by using bzr pull)
193
* create a new branch off your local bzr.dev copy for each issue
194
(bug or feature) you are working on.
196
This approach makes it easy to go back and make any required changes
197
after a code review. Resubmitting the change is then simple with no
198
risk of accidentially including edits related to other issues you may
199
be working on. After the changes for an issue are accepted and merged,
200
the associated branch can be deleted or archived as you wish.
203
Navigating the Code Base
204
========================
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.
473
Interface implementation testing and test scenarios
474
---------------------------------------------------
476
There are several cases in Bazaar of multiple implementations of a common
477
conceptual interface. ("Conceptual" because
478
it's not necessary for all the implementations to share a base class,
479
though they often do.) Examples include transports and the working tree,
480
branch and repository classes.
482
In these cases we want to make sure that every implementation correctly
483
fulfils the interface requirements. For example, every Transport should
484
support the ``has()`` and ``get()`` and ``clone()`` methods. We have a
485
sub-suite of tests in ``test_transport_implementations``. (Most
486
per-implementation tests are in submodules of ``bzrlib.tests``, but not
487
the transport tests at the moment.)
489
These tests are repeated for each registered Transport, by generating a
490
new TestCase instance for the cross product of test methods and transport
491
implementations. As each test runs, it has ``transport_class`` and
492
``transport_server`` set to the class it should test. Most tests don't
493
access these directly, but rather use ``self.get_transport`` which returns
494
a transport of the appropriate type.
496
The goal is to run per-implementation only tests that relate to that
497
particular interface. Sometimes we discover a bug elsewhere that happens
498
with only one particular transport. Once it's isolated, we can consider
499
whether a test should be added for that particular implementation,
500
or for all implementations of the interface.
502
The multiplication of tests for different implementations is normally
503
accomplished by overriding the ``test_suite`` function used to load
504
tests from a module. This function typically loads all the tests,
505
then applies a TestProviderAdapter to them, which generates a longer
506
suite containing all the test variations.
512
Some utilities are provided for generating variations of tests. This can
513
be used for per-implementation tests, or other cases where the same test
514
code needs to run several times on different scenarios.
516
The general approach is to define a class that provides test methods,
517
which depend on attributes of the test object being pre-set with the
518
values to which the test should be applied. The test suite should then
519
also provide a list of scenarios in which to run the tests.
521
Typically ``multiply_tests_from_modules`` should be called from the test
522
module's ``test_suite`` function.
525
Essential Domain Classes
526
########################
528
Introducing the Object Model
529
============================
531
The core domain objects within the bazaar model are:
541
Transports are explained below. See http://bazaar-vcs.org/Classes/
542
for an introduction to the other key classes.
547
The ``Transport`` layer handles access to local or remote directories.
548
Each Transport object acts like a logical connection to a particular
549
directory, and it allows various operations on files within it. You can
550
*clone* a transport to get a new Transport connected to a subdirectory or
553
Transports are not used for access to the working tree. At present
554
working trees are always local and they are accessed through the regular
555
Python file io mechanisms.
560
Transports work in URLs. Take note that URLs are by definition only
561
ASCII - the decision of how to encode a Unicode string into a URL must be
562
taken at a higher level, typically in the Store. (Note that Stores also
563
escape filenames which cannot be safely stored on all filesystems, but
564
this is a different level.)
566
The main reason for this is that it's not possible to safely roundtrip a
567
URL into Unicode and then back into the same URL. The URL standard
568
gives a way to represent non-ASCII bytes in ASCII (as %-escapes), but
569
doesn't say how those bytes represent non-ASCII characters. (They're not
570
guaranteed to be UTF-8 -- that is common but doesn't happen everywhere.)
572
For example if the user enters the url ``http://example/%e0`` there's no
573
way to tell whether that character represents "latin small letter a with
574
grave" in iso-8859-1, or "latin small letter r with acute" in iso-8859-2
575
or malformed UTF-8. So we can't convert their URL to Unicode reliably.
577
Equally problematic if we're given a url-like string containing non-ascii
578
characters (such as the accented a) we can't be sure how to convert that
579
to the correct URL, because we don't know what encoding the server expects
580
for those characters. (Although this is not totally reliable we might still
581
accept these and assume they should be put into UTF-8.)
583
A similar edge case is that the url ``http://foo/sweet%2Fsour`` contains
584
one directory component whose name is "sweet/sour". The escaped slash is
585
not a directory separator. If we try to convert URLs to regular Unicode
586
paths this information will be lost.
588
This implies that Transports must natively deal with URLs; for simplicity
589
they *only* deal with URLs and conversion of other strings to URLs is done
590
elsewhere. Information they return, such as from ``list_dir``, is also in
591
the form of URL components.
600
We have a commitment to 6 months API stability - any supported symbol in a
601
release of bzr MUST NOT be altered in any way that would result in
602
breaking existing code that uses it. That means that method names,
603
parameter ordering, parameter names, variable and attribute names etc must
604
not be changed without leaving a 'deprecated forwarder' behind. This even
605
applies to modules and classes.
607
If you wish to change the behaviour of a supported API in an incompatible
608
way, you need to change its name as well. For instance, if I add an optional keyword
609
parameter to branch.commit - that's fine. On the other hand, if I add a
610
keyword parameter to branch.commit which is a *required* transaction
611
object, I should rename the API - i.e. to 'branch.commit_transaction'.
613
When renaming such supported API's, be sure to leave a deprecated_method (or
614
_function or ...) behind which forwards to the new API. See the
615
bzrlib.symbol_versioning module for decorators that take care of the
616
details for you - such as updating the docstring, and issuing a warning
617
when the old api is used.
619
For unsupported API's, it does not hurt to follow this discipline, but it's
620
not required. Minimally though, please try to rename things so that
621
callers will at least get an AttributeError rather than weird results.
624
Coding Style Guidelines
625
=======================
627
Please write PEP-8__ compliant code.
629
One often-missed requirement is that the first line of docstrings
630
should be a self-contained one-sentence summary.
632
__ http://www.python.org/peps/pep-0008.html
638
* Imports should be done at the top-level of the file, unless there is
639
a strong reason to have them lazily loaded when a particular
640
function runs. Import statements have a cost, so try to make sure
641
they don't run inside hot functions.
643
* Module names should always be given fully-qualified,
644
i.e. ``bzrlib.hashcache`` not just ``hashcache``.
650
Functions, methods or members that are "private" to bzrlib are given
651
a leading underscore prefix. Names without a leading underscore are
652
public not just across modules but to programmers using bzrlib as an
653
API. As a consequence, a leading underscore is appropriate for names
654
exposed across modules but that are not to be exposed to bzrlib API
657
We prefer class names to be concatenated capital words (``TestCase``)
658
and variables, methods and functions to be lowercase words joined by
659
underscores (``revision_id``, ``get_revision``).
661
For the purposes of naming some names are treated as single compound
662
words: "filename", "revno".
664
Consider naming classes as nouns and functions/methods as verbs.
666
Try to avoid using abbreviations in names, because there can be
667
inconsistency if other people use the full name.
673
``revision_id`` not ``rev_id`` or ``revid``
675
Functions that transform one thing to another should be named ``x_to_y``
676
(not ``x2y`` as occurs in some old code.)
682
Python destructors (``__del__``) work differently to those of other
683
languages. In particular, bear in mind that destructors may be called
684
immediately when the object apparently becomes unreferenced, or at some
685
later time, or possibly never at all. Therefore we have restrictions on
686
what can be done inside them.
688
0. Never use a __del__ method without asking Martin/Robert first.
690
1. Never rely on a ``__del__`` method running. If there is code that
691
must run, do it from a ``finally`` block instead.
693
2. Never ``import`` from inside a ``__del__`` method, or you may crash the
696
3. In some places we raise a warning from the destructor if the object
697
has not been cleaned up or closed. This is considered OK: the warning
698
may not catch every case but it's still useful sometimes.
704
In some places we have variables which point to callables that construct
705
new instances. That is to say, they can be used a lot like class objects,
706
but they shouldn't be *named* like classes:
708
> I think that things named FooBar should create instances of FooBar when
709
> called. Its plain confusing for them to do otherwise. When we have
710
> something that is going to be used as a class - that is, checked for via
711
> isinstance or other such idioms, them I would call it foo_class, so that
712
> it is clear that a callable is not sufficient. If it is only used as a
713
> factory, then yes, foo_factory is what I would use.
719
Several places in Bazaar use (or will use) a registry, which is a
720
mapping from names to objects or classes. The registry allows for
721
loading in registered code only when it's needed, and keeping
722
associated information such as a help string or description.
728
To make startup time faster, we use the ``bzrlib.lazy_import`` module to
729
delay importing modules until they are actually used. ``lazy_import`` uses
730
the same syntax as regular python imports. So to import a few modules in a
733
from bzrlib.lazy_import import lazy_import
734
lazy_import(globals(), """
743
revision as _mod_revision,
745
import bzrlib.transport
749
At this point, all of these exist as a ``ImportReplacer`` object, ready to
750
be imported once a member is accessed. Also, when importing a module into
751
the local namespace, which is likely to clash with variable names, it is
752
recommended to prefix it as ``_mod_<module>``. This makes it clearer that
753
the variable is a module, and these object should be hidden anyway, since
754
they shouldn't be imported into other namespaces.
757
Modules versus Members
758
~~~~~~~~~~~~~~~~~~~~~~
760
While it is possible for ``lazy_import()`` to import members of a module
761
when using the ``from module import member`` syntax, it is recommended to
762
only use that syntax to load sub modules ``from module import submodule``.
763
This is because variables and classes can frequently be used without
764
needing a sub-member for example::
766
lazy_import(globals(), """
767
from module import MyClass
771
return isinstance(x, MyClass)
773
This will incorrectly fail, because ``MyClass`` is a ``ImportReplacer``
774
object, rather than the real class.
777
Passing to Other Variables
778
~~~~~~~~~~~~~~~~~~~~~~~~~~
780
It also is incorrect to assign ``ImportReplacer`` objects to other variables.
781
Because the replacer only knows about the original name, it is unable to
782
replace other variables. The ``ImportReplacer`` class will raise an
783
``IllegalUseOfScopeReplacer`` exception if it can figure out that this
784
happened. But it requires accessing a member more than once from the new
785
variable, so some bugs are not detected right away.
791
The null revision is the ancestor of all revisions. Its revno is 0, its
792
revision-id is ``null:``, and its tree is the empty tree. When referring
793
to the null revision, please use ``bzrlib.revision.NULL_REVISION``. Old
794
code sometimes uses ``None`` for the null revision, but this practice is
801
Processing Command Lines
802
------------------------
804
bzrlib has a standard framework for parsing command lines and calling
805
processing routines associated with various commands. See builtins.py
806
for numerous examples.
809
Standard Parameter Types
810
------------------------
812
There are some common requirements in the library: some parameters need to be
813
unicode safe, some need byte strings, and so on. At the moment we have
814
only codified one specific pattern: Parameters that need to be unicode
815
should be checked via ``bzrlib.osutils.safe_unicode``. This will coerce the
816
input into unicode in a consistent fashion, allowing trivial strings to be
817
used for programmer convenience, but not performing unpredictably in the
818
presence of different locales.
824
(The strategy described here is what we want to get to, but it's not
825
consistently followed in the code at the moment.)
827
bzrlib is intended to be a generically reusable library. It shouldn't
828
write messages to stdout or stderr, because some programs that use it
829
might want to display that information through a GUI or some other
832
We can distinguish two types of output from the library:
834
1. Structured data representing the progress or result of an
835
operation. For example, for a commit command this will be a list
836
of the modified files and the finally committed revision number
839
These should be exposed either through the return code or by calls
840
to a callback parameter.
842
A special case of this is progress indicators for long-lived
843
operations, where the caller should pass a ProgressBar object.
845
2. Unstructured log/debug messages, mostly for the benefit of the
846
developers or users trying to debug problems. This should always
847
be sent through ``bzrlib.trace`` and Python ``logging``, so that
848
it can be redirected by the client.
850
The distinction between the two is a bit subjective, but in general if
851
there is any chance that a library would want to see something as
852
structured data, we should make it so.
854
The policy about how output is presented in the text-mode client
855
should be only in the command-line tool.
862
Bazaar has online help for various topics through ``bzr help COMMAND`` or
863
equivalently ``bzr command -h``. We also have help on command options,
864
and on other help topics. (See ``help_topics.py``.)
866
As for python docstrings, the first paragraph should be a single-sentence
867
synopsis of the command.
869
The help for options should be one or more proper sentences, starting with
870
a capital letter and finishing with a full stop (period).
872
All help messages and documentation should have two spaces between
879
In general tests should be placed in a file named test_FOO.py where
880
FOO is the logical thing under test. That file should be placed in the
881
tests subdirectory under the package being tested.
883
For example, tests for merge3 in bzrlib belong in bzrlib/tests/test_merge3.py.
884
See bzrlib/tests/test_sampler.py for a template test script.
886
Tests can be written for the UI or for individual areas of the library.
887
Choose whichever is appropriate: if adding a new command, or a new command
888
option, then you should be writing a UI test. If you are both adding UI
889
functionality and library functionality, you will want to write tests for
890
both the UI and the core behaviours. We call UI tests 'blackbox' tests
891
and they are found in ``bzrlib/tests/blackbox/*.py``.
893
When writing blackbox tests please honour the following conventions:
895
1. Place the tests for the command 'name' in
896
bzrlib/tests/blackbox/test_name.py. This makes it easy for developers
897
to locate the test script for a faulty command.
899
2. Use the 'self.run_bzr("name")' utility function to invoke the command
900
rather than running bzr in a subprocess or invoking the
901
cmd_object.run() method directly. This is a lot faster than
902
subprocesses and generates the same logging output as running it in a
903
subprocess (which invoking the method directly does not).
905
3. Only test the one command in a single test script. Use the bzrlib
906
library when setting up tests and when evaluating the side-effects of
907
the command. We do this so that the library api has continual pressure
908
on it to be as functional as the command line in a simple manner, and
909
to isolate knock-on effects throughout the blackbox test suite when a
910
command changes its name or signature. Ideally only the tests for a
911
given command are affected when a given command is changed.
913
4. If you have a test which does actually require running bzr in a
914
subprocess you can use ``run_bzr_subprocess``. By default the spawned
915
process will not load plugins unless ``--allow-plugins`` is supplied.
921
We have a rich collection of tools to support writing tests. Please use
922
them in preference to ad-hoc solutions as they provide portability and
923
performance benefits.
928
The ``TreeBuilder`` interface allows the construction of arbitrary trees
929
with a declarative interface. A sample session might look like::
931
tree = self.make_branch_and_tree('path')
932
builder = TreeBuilder()
933
builder.start_tree(tree)
934
builder.build(['foo', "bar/", "bar/file"])
935
tree.commit('commit the tree')
936
builder.finish_tree()
938
Please see bzrlib.treebuilder for more details.
943
The ``BranchBuilder`` interface allows the creation of test branches in a
944
quick and easy manner. A sample session::
946
builder = BranchBuilder(self.get_transport().clone('relpath'))
947
builder.build_commit()
948
builder.build_commit()
949
builder.build_commit()
950
branch = builder.get_branch()
952
Please see bzrlib.branchbuilder for more details.
957
We make selective use of doctests__. In general they should provide
958
*examples* within the API documentation which can incidentally be tested. We
959
don't try to test every important case using doctests -- regular Python
960
tests are generally a better solution.
962
Most of these are in ``bzrlib/doc/api``. More additions are welcome.
964
__ http://docs.python.org/lib/module-doctest.html
969
Currently, bzr selftest is used to invoke tests.
970
You can provide a pattern argument to run a subset. For example,
971
to run just the blackbox tests, run::
973
./bzr selftest -v blackbox
975
To skip a particular test (or set of tests), use the --exclude option
976
(shorthand -x) like so::
978
./bzr selftest -v -x blackbox
980
To list tests without running them, use the --list-only option like so::
982
./bzr selftest --list-only
984
This option can be combined with other selftest options (like -x) and
985
filter patterns to understand their effect.
988
Handling Errors and Exceptions
989
==============================
991
Commands should return non-zero when they encounter circumstances that
992
the user should really pay attention to - which includes trivial shell
995
Recommended values are:
998
1. Conflicts in merge-like operations, or changes are present in
999
diff-like operations.
1000
2. Unrepresentable diff changes (i.e. binary files that we cannot show
1002
3. An error or exception has occurred.
1004
Errors are handled through Python exceptions. Exceptions should be defined
1005
inside bzrlib.errors, so that we can see the whole tree at a glance.
1007
We broadly classify errors as either being either internal or not,
1008
depending on whether ``internal_error`` is set or not. If we think it's our
1009
fault, we show a backtrace, an invitation to report the bug, and possibly
1010
other details. This is the default for errors that aren't specifically
1011
recognized as being caused by a user error. Otherwise we show a briefer
1012
message, unless -Derror was given.
1014
Many errors originate as "environmental errors" which are raised by Python
1015
or builtin libraries -- for example IOError. These are treated as being
1016
our fault, unless they're caught in a particular tight scope where we know
1017
that they indicate a user errors. For example if the repository format
1018
is not found, the user probably gave the wrong path or URL. But if one of
1019
the files inside the repository is not found, then it's our fault --
1020
either there's a bug in bzr, or something complicated has gone wrong in
1021
the environment that means one internal file was deleted.
1023
Many errors are defined in ``bzrlib/errors.py`` but it's OK for new errors
1024
to be added near the place where they are used.
1026
Exceptions are formatted for the user by conversion to a string
1027
(eventually calling their ``__str__`` method.) As a convenience the
1028
``._fmt`` member can be used as a template which will be mapped to the
1029
error's instance dict.
1031
New exception classes should be defined when callers might want to catch
1032
that exception specifically, or when it needs a substantially different
1035
Exception strings should start with a capital letter and should not have a
1036
final fullstop. If long, they may contain newlines to break the text.
1042
When you change bzrlib, please update the relevant documentation for the
1043
change you made: Changes to commands should update their help, and
1044
possibly end user tutorials; changes to the core library should be
1045
reflected in API documentation.
1050
If you make a user-visible change, please add a note to the NEWS file.
1051
The description should be written to make sense to someone who's just
1052
a user of bzr, not a developer: new functions or classes shouldn't be
1053
mentioned, but new commands, changes in behaviour or fixed nontrivial
1054
bugs should be listed. See the existing entries for an idea of what
1057
Within each release, entries in the news file should have the most
1058
user-visible changes first. So the order should be approximately:
1060
* changes to existing behaviour - the highest priority because the
1061
user's existing knowledge is incorrect
1062
* new features - should be brought to their attention
1063
* bug fixes - may be of interest if the bug was affecting them, and
1064
should include the bug number if any
1065
* major documentation changes
1066
* changes to internal interfaces
1068
People who made significant contributions to each change are listed in
1069
parenthesis. This can include reporting bugs (particularly with good
1070
details or reproduction recipes), submitting patches, etc.
1075
The docstring of a command is used by ``bzr help`` to generate help output
1076
for the command. The list 'takes_options' attribute on a command is used by
1077
``bzr help`` to document the options for the command - the command
1078
docstring does not need to document them. Finally, the '_see_also'
1079
attribute on a command can be used to reference other related help topics.
1084
Functions, methods, classes and modules should have docstrings
1085
describing how they are used.
1087
The first line of the docstring should be a self-contained sentence.
1089
For the special case of Command classes, this acts as the user-visible
1090
documentation shown by the help command.
1092
The docstrings should be formatted as reStructuredText_ (like this
1093
document), suitable for processing using the epydoc_ tool into HTML
1096
.. _reStructuredText: http://docutils.sourceforge.net/rst.html
1097
.. _epydoc: http://epydoc.sourceforge.net/
1106
The copyright policy for bzr was recently made clear in this email (edited
1107
for grammatical correctness)::
1109
The attached patch cleans up the copyright and license statements in
1110
the bzr source. It also adds tests to help us remember to add them
1111
with the correct text.
1113
We had the problem that lots of our files were "Copyright Canonical
1114
Development Ltd" which is not a real company, and some other variations
1115
on this theme. Also, some files were missing the GPL statements.
1117
I want to be clear about the intent of this patch, since copyright can
1118
be a little controversial.
1120
1) The big motivation for this is not to shut out the community, but
1121
just to clean up all of the invalid copyright statements.
1123
2) It has been the general policy for bzr that we want a single
1124
copyright holder for all of the core code. This is following the model
1125
set by the FSF, which makes it easier to update the code to a new
1126
license in case problems are encountered. (For example, if we want to
1127
upgrade the project universally to GPL v3 it is much simpler if there is
1128
a single copyright holder). It also makes it clearer if copyright is
1129
ever debated, there is a single holder, which makes it easier to defend
1130
in court, etc. (I think the FSF position is that if you assign them
1131
copyright, they can defend it in court rather than you needing to, and
1132
I'm sure Canonical would do the same).
1133
As such, Canonical has requested copyright assignments from all of the
1136
3) If someone wants to add code and not attribute it to Canonical, there
1137
is a specific list of files that are excluded from this check. And the
1138
test failure indicates where that is, and how to update it.
1140
4) If anyone feels that I changed a copyright statement incorrectly, just
1141
let me know, and I'll be happy to correct it. Whenever you have large
1142
mechanical changes like this, it is possible to make some mistakes.
1144
Just to reiterate, this is a community project, and it is meant to stay
1145
that way. Core bzr code is copyright Canonical for legal reasons, and
1146
the tests are just there to help us maintain that.
1149
Miscellaneous Topics
1150
####################
1155
Bazaar has a few facilities to help debug problems by going into pdb_, the
1158
.. _pdb: http://docs.python.org/lib/debugger-commands.html
1160
If the ``BZR_PDB`` environment variable is set
1161
then bzr will go into pdb post-mortem mode when an unhandled exception
1164
If you send a SIGQUIT signal to bzr, which can be done by pressing
1165
Ctrl-\\ on Unix, bzr will go into the debugger immediately. You can
1166
continue execution by typing ``c``. This can be disabled if necessary
1167
by setting the environment variable ``BZR_SIGQUIT_PDB=0``.
1174
Integer identifier for a revision on the main line of a branch.
1175
Revision 0 is always the null revision; others are 1-based
1176
indexes into the branch's revision history.
1179
Unicode and Encoding Support
1180
============================
1182
This section discusses various techniques that Bazaar uses to handle
1183
characters that are outside the ASCII set.
1188
When a ``Command`` object is created, it is given a member variable
1189
accessible by ``self.outf``. This is a file-like object, which is bound to
1190
``sys.stdout``, and should be used to write information to the screen,
1191
rather than directly writing to ``sys.stdout`` or calling ``print``.
1192
This file has the ability to translate Unicode objects into the correct
1193
representation, based on the console encoding. Also, the class attribute
1194
``encoding_type`` will effect how unprintable characters will be
1195
handled. This parameter can take one of 3 values:
1198
Unprintable characters will be represented with a suitable replacement
1199
marker (typically '?'), and no exception will be raised. This is for
1200
any command which generates text for the user to review, rather than
1201
for automated processing.
1202
For example: ``bzr log`` should not fail if one of the entries has text
1203
that cannot be displayed.
1206
Attempting to print an unprintable character will cause a UnicodeError.
1207
This is for commands that are intended more as scripting support, rather
1208
than plain user review.
1209
For exampl: ``bzr ls`` is designed to be used with shell scripting. One
1210
use would be ``bzr ls --null --unknows | xargs -0 rm``. If ``bzr``
1211
printed a filename with a '?', the wrong file could be deleted. (At the
1212
very least, the correct file would not be deleted). An error is used to
1213
indicate that the requested action could not be performed.
1216
Do not attempt to automatically convert Unicode strings. This is used
1217
for commands that must handle conversion themselves.
1218
For example: ``bzr diff`` needs to translate Unicode paths, but should
1219
not change the exact text of the contents of the files.
1222
``bzrlib.urlutils.unescape_for_display``
1223
----------------------------------------
1225
Because Transports work in URLs (as defined earlier), printing the raw URL
1226
to the user is usually less than optimal. Characters outside the standard
1227
set are printed as escapes, rather than the real character, and local
1228
paths would be printed as ``file://`` urls. The function
1229
``unescape_for_display`` attempts to unescape a URL, such that anything
1230
that cannot be printed in the current encoding stays an escaped URL, but
1231
valid characters are generated where possible.
1237
The ``bzrlib.osutils`` module has many useful helper functions, including
1238
some more portable variants of functions in the standard library.
1240
In particular, don't use ``shutil.rmtree`` unless it's acceptable for it
1241
to fail on Windows if some files are readonly or still open elsewhere.
1242
Use ``bzrlib.osutils.rmtree`` instead.
1248
We write some extensions in C using pyrex. We design these to work in
1251
* User with no C compiler
1252
* User with C compiler
1255
The recommended way to install bzr is to have a C compiler so that the
1256
extensions can be built, but if no C compiler is present, the pure python
1257
versions we supply will work, though more slowly.
1259
For developers we recommend that pyrex be installed, so that the C
1260
extensions can be changed if needed.
1262
For the C extensions, the extension module should always match the
1263
original python one in all respects (modulo speed). This should be
1264
maintained over time.
1266
To create an extension, add rules to setup.py for building it with pyrex,
1267
and with distutils. Now start with an empty .pyx file. At the top add
1268
"include 'yourmodule.py'". This will import the contents of foo.py into this
1269
file at build time - remember that only one module will be loaded at
1270
runtime. Now you can subclass classes, or replace functions, and only your
1271
changes need to be present in the .pyx file.
1273
Note that pyrex does not support all 2.4 programming idioms, so some
1274
syntax changes may be required. I.e.
1276
- 'from foo import (bar, gam)' needs to change to not use the brackets.
1277
- 'import foo.bar as bar' needs to be 'import foo.bar; bar = foo.bar'
1279
If the changes are too dramatic, consider
1280
maintaining the python code twice - once in the .pyx, and once in the .py,
1281
and no longer including the .py file.
1284
Making Installers for OS Windows
1285
================================
1286
To build a win32 installer, see the instructions on the wiki page:
1287
http://bazaar-vcs.org/BzrWin32Installer
1291
vim: ft=rst tw=74 ai
added
removed
doc/developers/HACKING.txt
