20
* Add bug information to log output when available.
21
(Neil Martinsen-Burrell, Guillermo Gonzalez, #251729)
23
* ``bzr branch`` now takes a ``--bind`` option. This lets you
24
branch and bind all in one command. (Ian Clatworthy)
26
* ``bzr update`` now takes a ``--revision`` argument. This lets you
27
change the revision of the working tree to any revision in the
28
ancestry of the current or master branch. (Matthieu Moy, Mark Hammond,
31
* ``-Dbytes`` can now be used to display the total number of bytes
32
transferred for the current command. This information is always logged
33
to ``.bzr.log`` for later inspection. (John Arbash Meinel)
35
* The ``suppress_warnings`` configuration option has been introduced and
36
accept the ``format_deprecation`` value to disable the corresponding
37
warning for repositories. It can be set to in either ``bazaar.conf``,
38
``locations.conf`` or ``branch.conf``.
39
(Ted Gould, Matthew Fuller, Vincent Ladeuil)
44
* Always show a message if an OS error occurs while trying to run a
45
user-specified commit message editor.
46
(Martin Pool, #504842)
48
* ``bzr export dir`` now requests all file content as a record stream,
49
rather than requsting the file content one file-at-a-time. This can make
50
exporting over the network significantly faster (54min => 9min in one
51
case). (John Arbash Meinel, #343218)
53
* ``bzr serve`` no longer slowly leaks memory. The compiled
54
``bzrlib.bencode.Encoder()`` class was using ``__del__`` to cleanup and
55
free resources, and it should have been using ``__dealloc__``.
56
This will likely have an impact on any other process that is serving for
57
an extended period of time. (John Arbash Meinel, #494406)
59
* ``bzr switch -b`` can now create branches that are located using directory
60
services such as ``lp:``, even when the branch name doesn't contain a
61
'/'. (Neil Martinsen-Burrell, #495263)
63
* ``bzr unshelve`` has improved messages about what it is doing.
64
(Neil Martinsen-Burrell, #496917)
66
* Check for SIGINT (Ctrl-C) and other signals immediately if ``readdir``
67
returns ``EINTR`` by calling ``PyErr_CheckSignals``. This affected the
68
optional ``_readdir_pyx`` extension. (Andrew Bennetts, #495023)
70
* Fixed a side effect mutation of ``RemoteBzrDirFormat._network_name``
71
that caused some tests to fail when run in a non-default order.
72
Probably no user impact. (Martin Pool, #504102)
74
* Fixed ``ObjectNotLocked`` error in ``bzr cat -rbranch:../foo FILE``.
75
(Andrew Bennetts, #506274)
77
* FTP transports support Unicode paths by encoding/decoding them as utf8.
78
(Vincent Ladeuil, #472161)
80
* Give a clearer message if the lockdir disappears after being apparently
81
successfully taken. (Martin Pool, #498378)
23
83
* Listen to the SIGWINCH signal to update the terminal width.
24
84
(Vincent Ladeuil, #316357)
86
* The 2a format wasn't properly restarting autopacks when something
87
changed underneath it (like another autopack). Now concurrent
88
autopackers will properly succeed. (John Arbash Meinel, #495000)
90
* When operations update the working tree, all affected files should end
91
up with the same mtime. (eg. when versioning a generated file, if you
92
update the source and the generated file together, the generated file
93
should appear up-to-date.)
94
(John Arbash Meinel, Martin <gzlist>, #488724)
99
* Added ``add_cleanup`` and ``cleanup_now`` to ``bzrlib.command.Command``.
100
All the builtin commands now use ``add_cleanup`` rather than
101
``try``/``finally`` blocks where applicable as it is simpler and more
102
robust. (Andrew Bennetts)
104
* Push will now inform the user when they are trying to push to a foreign
105
VCS for which roundtripping is not supported, and will suggest them to
106
use dpush. (Jelmer Vernooij)
108
* The version of bzr being run is now written to the log file.
111
* Transport network activity indicator is shown more of the time when
112
Bazaar is doing network IO.
118
* There is a System Administrator's Guide in ``doc/en/admin-guide``,
119
including discussions of installation, relevant plugins, security and
120
backup. (Neil Martinsen-Burrell)
122
* The ``conflicts`` help topic has been renamed to ``conflict-types``.
125
* The User Reference is now presented as a series of topics.
126
Many of the included topics have link and format tweaks applied.
132
* Many test features were renamed from ``FooFeature`` to ``foo_feature``
133
to be consistent with instances being lower case and classes being
134
CamelCase. For the features that were more likely to be used, we added a
135
deprecation thunk, but not all. (John Arbash Meinel)
137
* The Branch hooks pre_change_branch_tip no longer masks exceptions raised
138
by plugins - the original exceptions are now preserved. (Robert Collins)
140
* The Transport ``Server.tearDown`` method is now renamed to
141
``stop_server`` and ``setUp`` to ``start_server`` for consistency with
142
our normal naming pattern, and to avoid confusion with Python's
143
``TestCase.tearDown``. (Martin Pool)
145
* ``WorkingTree.update`` implementations must now accept a ``revision``
151
* New helper osutils.UnicodeOrBytesToBytesWriter which encodes unicode
152
objects but passes str objects straight through. This is used for
153
selftest but may be useful for diff and other operations that generate
154
mixed output. (Robert Collins)
156
* New exception ``NoRoundtrippingSupport``, for use by foreign branch
157
plugins. (Jelmer Vernooij)
162
* ``bzrlib.tests.permute_for_extension`` is a helper that simplifies
163
running all tests in the current module, once against a pure python
164
implementation, and once against an extension (pyrex/C) implementation.
165
It can be used to dramatically simplify the implementation of
166
``load_tests``. (John Arbash Meinel)
168
* ``bzrlib.tests.TestCase`` now subclasses ``testtools.testcase.TestCase``.
169
This permits features in testtools such as getUniqueInteger and
170
getUniqueString to be used. Because of this, testtools version 0.9.2 or
171
newer is now a dependency to run bzr selftest. Running with versions of
172
testtools less than 0.9.2 will cause bzr to error while loading the test
173
suite. (Robert Collins)
175
* The test progress bar no longer distinguishes tests that 'errored' from
176
tests that 'failed' - they're all just failures.
41
179
bzr 2.0.4 (not released yet)
42
180
############################
183
:2.0.4: smooth sailing
47
185
Compatibility Breaks
48
186
********************
194
* ``bzr annotate`` on another branch with ``-r branch:...`` no longer
195
fails with an ``ObjectNotLocked`` error. (Andrew Bennetts, #496590)
197
* ``bzr export dir`` now requests all file content as a record stream,
198
rather than requsting the file content one file-at-a-time. This can make
199
exporting over the network significantly faster (54min => 9min in one
200
case). (John Arbash Meinel, #343218)
202
* ``bzr serve`` no longer slowly leaks memory. The compiled
203
``bzrlib.bencode.Encoder()`` class was using ``__del__`` to cleanup and
204
free resources, and it should have been using ``__dealloc__``.
205
This will likely have an impact on any other process that is serving for
206
an extended period of time. (John Arbash Meinel, #494406)
208
* Check for SIGINT (Ctrl-C) and other signals immediately if ``readdir``
209
returns ``EINTR`` by calling ``PyErr_CheckSignals``. This affected the
210
optional ``_readdir_pyx`` extension. (Andrew Bennetts, #495023)
212
* Give a clearer message if the lockdir disappears after being apparently
213
successfully taken. (Martin Pool, #498378)
215
* If we fail to open ``~/.bzr.log`` write a clear message to stderr rather
216
than using ``warning()``. The log file is opened before logging is set
217
up, and it leads to very confusing: 'no handlers for "bzr"' messages for
218
users, rather than something nicer.
219
(John Arbash Meinel, Barry Warsaw, #503886)
221
* Refuse to build with any Pyrex 0.9.4 release, as they have known bugs.
222
(Martin Pool, John Arbash Meinel, #449372)
224
* The 2a format wasn't properly restarting autopacks when something
225
changed underneath it (like another autopack). Now concurrent
226
autopackers will properly succeed. (John Arbash Meinel, #495000)
228
* ``_update_current_block`` no longer suppresses exceptions, so ^C at just
229
the right time will get propagated, rather than silently failing to move
230
the block pointer. (John Arbash Meinel, Gareth White, #495023)
added
removed
NEWS
