1
============================
2
Guidelines for modifying bzr
3
============================
7
(The current version of this document is available in the file ``HACKING``
8
in the source tree, or at http://doc.bazaar-vcs.org/current/hacking.html)
13
* New functionality should have test cases. Preferably write the
14
test before writing the code.
16
In general, you can test at either the command-line level or the
17
internal API level. See Writing Tests below for more detail.
19
* Try to practice Test-Driven Development. before fixing a bug, write a
20
test case so that it does not regress. Similarly for adding a new
21
feature: write a test case for a small version of the new feature before
22
starting on the code itself. Check the test fails on the old code, then
23
add the feature or fix and check it passes.
25
* Exceptions should be defined inside bzrlib.errors, so that we can
26
see the whole tree at a glance.
28
* Imports should be done at the top-level of the file, unless there is
29
a strong reason to have them lazily loaded when a particular
30
function runs. Import statements have a cost, so try to make sure
31
they don't run inside hot functions.
33
* Module names should always be given fully-qualified,
34
i.e. ``bzrlib.hashcache`` not just ``hashcache``.
36
* Commands should return non-zero when they encounter circumstances that
37
the user should really pay attention to - which includes trivial shell
40
Recommended values are
42
1. Conflicts in merge-like operations, or changes are present in
44
2. Unrepresentable diff changes (i.e. binary files that we cannot show
46
3. An error or exception has occurred.
51
We have a commitment to 6 months API stability - any supported symbol in a
52
release of bzr MUST NOT be altered in any way that would result in
53
breaking existing code that uses it. That means that method names,
54
parameter ordering, parameter names, variable and attribute names etc must
55
not be changed without leaving a 'deprecated forwarder' behind. This even
56
applies to modules and classes.
58
If you wish to change the behaviour of a supported API in an incompatible
59
way, you need to change its name as well. For instance, if I add an optional keyword
60
parameter to branch.commit - that's fine. On the other hand, if I add a
61
keyword parameter to branch.commit which is a *required* transaction
62
object, I should rename the API - i.e. to 'branch.commit_transaction'.
64
When renaming such supported API's, be sure to leave a deprecated_method (or
65
_function or ...) behind which forwards to the new API. See the
66
bzrlib.symbol_versioning module for decorators that take care of the
67
details for you - such as updating the docstring, and issuing a warning
68
when the old api is used.
70
For unsupported API's, it does not hurt to follow this discipline, but it's
71
not required. Minimally though, please try to rename things so that
72
callers will at least get an AttributeError rather than weird results.
75
Standard parameter types
76
------------------------
78
There are some common requirements in the library: some parameters need to be
79
unicode safe, some need byte strings, and so on. At the moment we have
80
only codified one specific pattern: Parameters that need to be unicode
81
should be checked via ``bzrlib.osutils.safe_unicode``. This will coerce the
82
input into unicode in a consistent fashion, allowing trivial strings to be
83
used for programmer convenience, but not performing unpredictably in the
84
presence of different locales.
90
The copyright policy for bzr was recently made clear in this email (edited
91
for grammatical correctness)::
93
The attached patch cleans up the copyright and license statements in
94
the bzr source. It also adds tests to help us remember to add them
95
with the correct text.
97
We had the problem that lots of our files were "Copyright Canonical
98
Development Ltd" which is not a real company, and some other variations
99
on this theme. Also, some files were missing the GPL statements.
101
I want to be clear about the intent of this patch, since copyright can
102
be a little controversial.
104
1) The big motivation for this is not to shut out the community, but
105
just to clean up all of the invalid copyright statements.
107
2) It has been the general policy for bzr that we want a single
108
copyright holder for all of the core code. This is following the model
109
set by the FSF, which makes it easier to update the code to a new
110
license in case problems are encountered. (For example, if we want to
111
upgrade the project universally to GPL v3 it is much simpler if there is
112
a single copyright holder). It also makes it clearer if copyright is
113
ever debated, there is a single holder, which makes it easier to defend
114
in court, etc. (I think the FSF position is that if you assign them
115
copyright, they can defend it in court rather than you needing to, and
116
I'm sure Canonical would do the same).
117
As such, Canonical has requested copyright assignments from all of the
120
3) If someone wants to add code and not attribute it to Canonical, there
121
is a specific list of files that are excluded from this check. And the
122
test failure indicates where that is, and how to update it.
124
4) If anyone feels that I changed a copyright statement incorrectly, just
125
let me know, and I'll be happy to correct it. Whenever you have large
126
mechanical changes like this, it is possible to make some mistakes.
128
Just to reiterate, this is a community project, and it is meant to stay
129
that way. Core bzr code is copyright Canonical for legal reasons, and
130
the tests are just there to help us maintain that.
136
If you change the behaviour of a command, please update its docstring
137
in bzrlib/commands.py. This is displayed by the 'bzr help' command.
142
If you make a user-visible change, please add a note to the NEWS file.
143
The description should be written to make sense to someone who's just
144
a user of bzr, not a developer: new functions or classes shouldn't be
145
mentioned, but new commands, changes in behaviour or fixed nontrivial
146
bugs should be listed. See the existing entries for an idea of what
149
Within each release, entries in the news file should have the most
150
user-visible changes first. So the order should be approximately:
152
* changes to existing behaviour - the highest priority because the
153
user's existing knowledge is incorrect
154
* new features - should be brought to their attention
155
* bug fixes - may be of interest if the bug was affecting them, and
156
should include the bug number if any
157
* major documentation changes
158
* changes to internal interfaces
160
People who made significant contributions to each change are listed in
161
parenthesis. This can include reporting bugs (particularly with good
162
details or reproduction recipes), submitting patches, etc.
167
Functions, methods, classes and modules should have docstrings
168
describing how they are used.
170
The first line of the docstring should be a self-contained sentence.
172
For the special case of Command classes, this acts as the user-visible
173
documentation shown by the help command.
175
The docstrings should be formatted as reStructuredText_ (like this
176
document), suitable for processing using the epydoc_ tool into HTML
179
.. _reStructuredText: http://docutils.sourceforge.net/rst.html
180
.. _epydoc: http://epydoc.sourceforge.net/
187
Please write PEP-8__ compliant code.
189
One often-missed requirement is that the first line of docstrings
190
should be a self-contained one-sentence summary.
192
__ http://www.python.org/peps/pep-0008.html
199
Functions, methods or members that are in some sense "private" are given
200
a leading underscore prefix. This is just a hint that code outside the
201
implementation should probably not use that interface.
203
We prefer class names to be concatenated capital words (``TestCase``)
204
and variables, methods and functions to be lowercase words joined by
205
underscores (``revision_id``, ``get_revision``).
207
For the purposes of naming some names are treated as single compound
208
words: "filename", "revno".
210
Consider naming classes as nouns and functions/methods as verbs.
212
Try to avoid using abbreviations in names, because there can be
213
inconsistency if other people use the full name.
219
``revision_id`` not ``rev_id`` or ``revid``
221
Functions that transform one thing to another should be named ``x_to_y``
222
(not ``x2y`` as occurs in some old code.)
228
Python destructors (``__del__``) work differently to those of other
229
languages. In particular, bear in mind that destructors may be called
230
immediately when the object apparently becomes unreferenced, or at some
231
later time, or possibly never at all. Therefore we have restrictions on
232
what can be done inside them.
234
0. Never use a __del__ method without asking Martin/Robert first.
236
1. Never rely on a ``__del__`` method running. If there is code that
237
must run, do it from a ``finally`` block instead.
239
2. Never ``import`` from inside a ``__del__`` method, or you may crash the
242
3. In some places we raise a warning from the destructor if the object
243
has not been cleaned up or closed. This is considered OK: the warning
244
may not catch every case but it's still useful sometimes.
250
In some places we have variables which point to callables that construct
251
new instances. That is to say, they can be used a lot like class objects,
252
but they shouldn't be *named* like classes:
254
> I think that things named FooBar should create instances of FooBar when
255
> called. Its plain confusing for them to do otherwise. When we have
256
> something that is going to be used as a class - that is, checked for via
257
> isinstance or other such idioms, them I would call it foo_class, so that
258
> it is clear that a callable is not sufficient. If it is only used as a
259
> factory, then yes, foo_factory is what I would use.
265
Several places in Bazaar use (or will use) a registry, which is a
266
mapping from names to objects or classes. The registry allows for
267
loading in registered code only when it's needed, and keeping
268
associated information such as a help string or description.
274
To make startup time faster, we use the ``bzrlib.lazy_import`` module to
275
delay importing modules until they are actually used. ``lazy_import`` uses
276
the same syntax as regular python imports. So to import a few modules in a
279
from bzrlib.lazy_import import lazy_import
280
lazy_import(globals(), """
289
revision as _mod_revision,
291
import bzrlib.transport
295
At this point, all of these exist as a ``ImportReplacer`` object, ready to
296
be imported once a member is accessed. Also, when importing a module into
297
the local namespace, which is likely to clash with variable names, it is
298
recommended to prefix it as ``_mod_<module>``. This makes it clean that
299
the variable is a module, and these object should be hidden anyway, since
300
they shouldn't be imported into other namespaces.
303
Modules versus Members
304
~~~~~~~~~~~~~~~~~~~~~~
306
While it is possible for ``lazy_import()`` to import members of a module
307
when using the ``from module import member`` syntax, it is recommended to
308
only use that syntax to load sub modules ``from module import submodule``.
309
This is because variables and classes can frequently be used without
310
needing a sub-member for example::
312
lazy_import(globals(), """
313
from module import MyClass
317
return isinstance(x, MyClass)
319
This will incorrectly fail, because ``MyClass`` is a ``ImportReplacer``
320
object, rather than the real class.
323
Passing to other variables
324
~~~~~~~~~~~~~~~~~~~~~~~~~~
326
It also is incorrect to assign ``ImportReplacer`` objects to other variables.
327
Because the replacer only knows about the original name, it is unable to
328
replace other variables. The ``ImportReplacer`` class will raise an
329
``IllegalUseOfScopeReplacer`` exception if it can figure out that this
330
happened. But it requires accessing a member more than once from the new
331
variable, so some bugs are not detected right away.
337
(The strategy described here is what we want to get to, but it's not
338
consistently followed in the code at the moment.)
340
bzrlib is intended to be a generically reusable library. It shouldn't
341
write messages to stdout or stderr, because some programs that use it
342
might want to display that information through a GUI or some other
345
We can distinguish two types of output from the library:
347
1. Structured data representing the progress or result of an
348
operation. For example, for a commit command this will be a list
349
of the modified files and the finally committed revision number
352
These should be exposed either through the return code or by calls
353
to a callback parameter.
355
A special case of this is progress indicators for long-lived
356
operations, where the caller should pass a ProgressBar object.
358
2. Unstructured log/debug messages, mostly for the benefit of the
359
developers or users trying to debug problems. This should always
360
be sent through ``bzrlib.trace`` and Python ``logging``, so that
361
it can be redirected by the client.
363
The distinction between the two is a bit subjective, but in general if
364
there is any chance that a library would want to see something as
365
structured data, we should make it so.
367
The policy about how output is presented in the text-mode client
368
should be only in the command-line tool.
374
In general tests should be placed in a file named test_FOO.py where
375
FOO is the logical thing under test. That file should be placed in the
376
tests subdirectory under the package being tested.
378
For example, tests for merge3 in bzrlib belong in bzrlib/tests/test_merge3.py.
379
See bzrlib/selftest/test_sampler.py for a template test script.
381
Tests can be written for the UI or for individual areas of the library.
382
Choose whichever is appropriate: if adding a new command, or a new command
383
option, then you should be writing a UI test. If you are both adding UI
384
functionality and library functionality, you will want to write tests for
385
both the UI and the core behaviours. We call UI tests 'blackbox' tests
386
and they are found in ``bzrlib/tests/blackbox/*.py``.
388
When writing blackbox tests please honour the following conventions:
390
1. Place the tests for the command 'name' in
391
bzrlib/tests/blackbox/test_name.py. This makes it easy for developers
392
to locate the test script for a faulty command.
394
2. Use the 'self.run_bzr("name")' utility function to invoke the command
395
rather than running bzr in a subprocess or invoking the
396
cmd_object.run() method directly. This is a lot faster than
397
subprocesses and generates the same logging output as running it in a
398
subprocess (which invoking the method directly does not).
400
3. Only test the one command in a single test script. Use the bzrlib
401
library when setting up tests and when evaluating the side-effects of
402
the command. We do this so that the library api has continual pressure
403
on it to be as functional as the command line in a simple manner, and
404
to isolate knock-on effects throughout the blackbox test suite when a
405
command changes its name or signature. Ideally only the tests for a
406
given command are affected when a given command is changed.
408
4. If you have a test which does actually require running bzr in a
409
subprocess you can use ``run_bzr_subprocess``. By default the spawned
410
process will not load plugins unless ``--allow-plugins`` is supplied.
416
We make selective use of doctests__. In general they should provide
417
*examples* within the API documentation which can incidentally be tested. We
418
don't try to test every important case using doctests -- regular Python
419
tests are generally a better solution.
421
Most of these are in ``bzrlib/doc/api``. More additions are welcome.
423
__ http://docs.python.org/lib/module-doctest.html
428
Currently, bzr selftest is used to invoke tests.
429
You can provide a pattern argument to run a subset. For example,
430
to run just the blackbox tests, run::
432
./bzr selftest -v blackbox
434
To skip a particular test (or set of tests), you need to use a negative
437
./bzr selftest '^(?!.*blackbox)'
440
Errors and exceptions
441
=====================
443
Errors are handled through Python exceptions.
445
We broadly classify errors as either being either internal or not,
446
depending on whether ``user_error`` is set or not. If we think it's our
447
fault, we show a backtrace, an invitation to report the bug, and possibly
448
other details. This is the default for errors that aren't specifically
449
recognized as being caused by a user error. Otherwise we show a briefer
450
message, unless -Derror was given.
452
Many errors originate as "environmental errors" which are raised by Python
453
or builtin libraries -- for example IOError. These are treated as being
454
our fault, unless they're caught in a particular tight scope where we know
455
that they indicate a user errors. For example if the repository format
456
is not found, the user probably gave the wrong path or URL. But if one of
457
the files inside the repository is not found, then it's our fault --
458
either there's a bug in bzr, or something complicated has gone wrong in
459
the environment that means one internal file was deleted.
461
Many errors are defined in ``bzrlib/errors.py`` but it's OK for new errors
462
to be added near the place where they are used.
464
Exceptions are formatted for the user by conversion to a string
465
(eventually calling their ``__str__`` method.) As a convenience the
466
``._fmt`` member can be used as a template which will be mapped to the
467
error's instance dict.
469
New exception classes should be defined when callers might want to catch
470
that exception specifically, or when it needs a substantially different
473
Exception strings should start with a capital letter and should not have a
474
final fullstop. If long, they may contain newlines to break the text.
482
Integer identifier for a revision on the main line of a branch.
483
Revision 0 is always the null revision; others are 1-based
484
indexes into the branch's revision history.
490
The ``Transport`` layer handles access to local or remote directories.
491
Each Transport object acts like a logical connection to a particular
492
directory, and it allows various operations on files within it. You can
493
*clone* a transport to get a new Transport connected to a subdirectory or
496
Transports are not used for access to the working tree. At present
497
working trees are always local and they are accessed through the regular
498
Python file io mechanisms.
503
Transports work in URLs. Take note that URLs are by definition only
504
ASCII - the decision of how to encode a Unicode string into a URL must be
505
taken at a higher level, typically in the Store. (Note that Stores also
506
escape filenames which cannot be safely stored on all filesystems, but
507
this is a different level.)
509
The main reason for this is that it's not possible to safely roundtrip a
510
URL into Unicode and then back into the same URL. The URL standard
511
gives a way to represent non-ASCII bytes in ASCII (as %-escapes), but
512
doesn't say how those bytes represent non-ASCII characters. (They're not
513
guaranteed to be UTF-8 -- that is common but doesn't happen everywhere.)
515
For example if the user enters the url ``http://example/%e0`` there's no
516
way to tell whether that character represents "latin small letter a with
517
grave" in iso-8859-1, or "latin small letter r with acute" in iso-8859-2
518
or malformed UTF-8. So we can't convert their URL to Unicode reliably.
520
Equally problematic if we're given a url-like string containing non-ascii
521
characters (such as the accented a) we can't be sure how to convert that
522
to the correct URL, because we don't know what encoding the server expects
523
for those characters. (Although this is not totally reliable we might still
524
accept these and assume they should be put into UTF-8.)
526
A similar edge case is that the url ``http://foo/sweet%2Fsour`` contains
527
one directory component whose name is "sweet/sour". The escaped slash is
528
not a directory separator. If we try to convert URLs to regular Unicode
529
paths this information will be lost.
531
This implies that Transports must natively deal with URLs; for simplicity
532
they *only* deal with URLs and conversion of other strings to URLs is done
533
elsewhere. Information they return, such as from ``list_dir``, is also in
534
the form of URL components.
537
Unicode and Encoding Support
538
============================
540
This section discusses various techniques that Bazaar uses to handle
541
characters that are outside the ASCII set.
546
When a ``Command`` object is created, it is given a member variable
547
accessible by ``self.outf``. This is a file-like object, which is bound to
548
``sys.stdout``, and should be used to write information to the screen,
549
rather than directly writing to ``sys.stdout`` or calling ``print``.
550
This file has the ability to translate Unicode objects into the correct
551
representation, based on the console encoding. Also, the class attribute
552
``encoding_type`` will effect how unprintable characters will be
553
handled. This parameter can take one of 3 values:
556
Unprintable characters will be represented with a suitable replacement
557
marker (typically '?'), and no exception will be raised. This is for
558
any command which generates text for the user to review, rather than
559
for automated processing.
560
For example: ``bzr log`` should not fail if one of the entries has text
561
that cannot be displayed.
564
Attempting to print an unprintable character will cause a UnicodeError.
565
This is for commands that are intended more as scripting support, rather
566
than plain user review.
567
For exampl: ``bzr ls`` is designed to be used with shell scripting. One
568
use would be ``bzr ls --null --unknows | xargs -0 rm``. If ``bzr``
569
printed a filename with a '?', the wrong file could be deleted. (At the
570
very least, the correct file would not be deleted). An error is used to
571
indicate that the requested action could not be performed.
574
Do not attempt to automatically convert Unicode strings. This is used
575
for commands that must handle conversion themselves.
576
For example: ``bzr diff`` needs to translate Unicode paths, but should
577
not change the exact text of the contents of the files.
580
``bzrlib.urlutils.unescape_for_display``
581
----------------------------------------
583
Because Transports work in URLs (as defined earlier), printing the raw URL
584
to the user is usually less than optimal. Characters outside the standard
585
set are printed as escapes, rather than the real character, and local
586
paths would be printed as ``file://`` urls. The function
587
``unescape_for_display`` attempts to unescape a URL, such that anything
588
that cannot be printed in the current encoding stays an escaped URL, but
589
valid characters are generated where possible.
595
If you'd like to propose a change, please post to the
596
bazaar@lists.canonical.com list with a patch, bzr changeset, or link to a
597
branch. Please put '[patch]' in the subject so we can pick them out, and
598
include some text explaining the change. Remember to put an update to the NEWS
599
file in your diff, if it makes any changes visible to users or plugin
600
developers. Please include a diff against mainline if you're giving a link to
603
Please indicate if you think the code is ready to merge, or if it's just a
604
draft or for discussion. If you want comments from many developers rather than
605
to be merged, you can put '[rfc]' in the subject lines.
607
Anyone is welcome to review code. There are broadly three gates for
610
* Doesn't reduce test coverage: if it adds new methods or commands,
611
there should be tests for them. There is a good test framework
612
and plenty of examples to crib from, but if you are having trouble
613
working out how to test something feel free to post a draft patch
616
* Doesn't reduce design clarity, such as by entangling objects
617
we're trying to separate. This is mostly something the more
618
experienced reviewers need to help check.
620
* Improves bugs, features, speed, or code simplicity.
622
Code that goes in should pass all three.
624
If you read a patch please reply and say so. We can use a numeric scale
625
of -1, -0, +0, +1, meaning respectively "really don't want it in current
626
form", "somewhat uncomfortable", "ok with me", and "please put it in".
627
Anyone can "vote". (It's not really voting, just a terse expression.)
629
If something gets say two +1 votes from core reviewers, and no
630
vetos, then it's OK to come in. Any of the core developers can bring it
631
into their integration branch, which I'll merge regularly. (If you do
632
so, please reply and say so.)
635
Making installers for OS Windows
636
================================
637
To build a win32 installer, see the instructions on the wiki page:
638
http://bazaar-vcs.org/BzrWin32Installer
641
:: vim: ft=rst tw=74 ai
added
removed
HACKING
