summaryrefslogtreecommitdiffstats
path: root/lib/subunit/NEWS
blob: 713d272bfaa9186cffc1b8e56096043d2f5adf54 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
---------------------
subunit release notes
---------------------

NEXT (In development)
---------------------

IMPROVEMENTS
~~~~~~~~~~~~

* Perl module now correctly outputs "failure" instead of "fail".  (Stewart Smith)

* Shell functions now output timestamps. (Stewart Smith)

BUG FIXES
~~~~~~~~~

* Add 'subunit --no-xfail', which will omit expected failures from the subunit
  stream. (John Arbash Meinel, #623642)

* Add 'subunit -F/--only-genuine-failures' which sets all of '--no-skips',
  '--no-xfail', '--no-passthrough, '--no-success', and gives you just the
  failure stream. (John Arbash Meinel)

0.0.7
-----

The Subunit Python test runner ``python -m subunit.run`` can now report the
test ids and also filter via a test id list file thanks to improvements in
``testtools.run``. See the testtools manual, or testrepository - a major
user of such functionality.

Additionally the protocol now has a keyword uxsuccess for Unexpected Success
reporting. Older parsers will report tests with this status code as 'lost
connection'.

IMPROVEMENTS
~~~~~~~~~~~~

* Add ``TimeCollapsingDecorator`` which collapses multiple sequential time()
  calls into just the first and last. (Jonathan Lange)

* Add ``TagCollapsingDecorator`` which collapses many tags() calls into one
  where possible. (Jonathan Lange, Robert Collins)

* Force flush of writes to stdout in c/tests/test_child.
  (Jelmer Vernooij, #687611)

* Provisional Python 3.x support.
  (Robert Collins, Tres Seaver, Martin[gz], #666819)

* ``subunit.chunked.Decoder`` Python class takes a new ``strict`` option,
  which defaults to ``True``. When ``False``, the ``Decoder`` will accept
  incorrect input that is still unambiguous. i.e. subunit will not barf if
  a \r is missing from the input. (Martin Pool)

* ``subunit-filter`` preserves the relative ordering of ``time:`` statements,
  so you can now use filtered streams to gather data about how long it takes
  to run a test. (Jonathan Lange, #716554)

* ``subunit-ls`` now handles a stream with time: instructions that start
  partway through the stream (which may lead to strange times) more gracefully.
  (Robert Collins, #785954)

* ``subunit-ls`` should handle the new test outcomes in Python2.7 better.
  (Robert Collins, #785953)

* ``TestResultFilter`` now collapses sequential calls to time().
  (Jonathan Lange, #567150)

* ``TestResultDecorator.tags()`` now actually works, and is no longer a buggy
  copy/paste of ``TestResultDecorator.time()``. (Jonathan Lange, #681828)

* ``TestResultFilter`` now supports a ``fixup_expected_failures``
  argument. (Jelmer Vernooij, #755241)

* The ``subunit.run`` Python module supports ``-l`` and ``--load-list`` as
  per ``testtools.run``. This required a dependency bump due to a small
  API change in ``testtools``. (Robert Collins)

* The help for subunit-filter was confusing about the behaviour of ``-f`` /
  ``--no-failure``. (Robert Collins, #703392)

* The Python2.7 / testtools addUnexpectedSuccess API is now supported. This
  required adding a new status code to the protocol. (Robert Collins, #654474)

CHANGES
~~~~~~~

* testtools 0.9.11 or newer is new needed (due to the Python 3 support).
  (Robert Collins)

0.0.6
-----

This release of subunit fixes a number of unicode related bugs. This depends on
testtools 0.9.4 and will not function without it. Thanks to Tres Seaver there
is also an optional native setup.py file for use with easy_install and the
like.

BUG FIXES
~~~~~~~~~

* Be consistent about delivering unicode content to testtools StringException
  class which has become (appropriately) conservative. (Robert Collins)

* Fix incorrect reference to subunit_test_failf in c/README.
  (Brad Hards, #524341)

* Fix incorrect ordering of tags method parameters in TestResultDecorator. This
  is purely cosmetic as the parameters are passed down with no interpretation.
  (Robert Collins, #537611)

* Old style tracebacks with no encoding info are now treated as UTF8 rather
  than some-random-codec-like-ascii. (Robert Collins)

* On windows, ProtocolTestCase and TestProtocolClient will set their streams to
  binary mode by calling into msvcrt; this avoids having their input or output
  mangled by the default line ending translation on that platform.
  (Robert Collins, Martin [gz], #579296)

IMPROVEMENTS
~~~~~~~~~~~~

* Subunit now has a setup.py for python deployments that are not using
  distribution packages. (Tres Seaver, #538181)

* Subunit now supports test discovery by building on the testtools support for
  it. You can take advantage of it with "python -m subunit.run discover [path]"
  and see "python -m subunit.run discover --help" for more options.

* Subunit now uses the improved unicode support in testtools when outputting
  non-details based test information; this should consistently UTF8 encode such
  strings.

* The Python TestProtocolClient now flushes output on startTest and stopTest.
  (Martin [gz]).


0.0.5
-----

BUG FIXES
~~~~~~~~~

* make check was failing if subunit wasn't installed due to a missing include
  path for the test program test_child.

* make distcheck was failing due to a missing $(top_srcdir) rune.

IMPROVEMENTS
~~~~~~~~~~~~

* New filter `subunit-notify` that will show a notification window with test 
  statistics when the test run finishes.

* subunit.run will now pipe its output to the command in the 
  SUBUNIT_FORMATTER environment variable, if set.

0.0.4
-----

BUG FIXES
~~~~~~~~~

* subunit2junitxml -f required a value, this is now fixed and -f acts as a
  boolean switch with no parameter.

* Building with autoconf 2.65 is now supported.


0.0.3
-----

  CHANGES:
  
    * License change, by unanimous agreement of contributors to BSD/Apache
      License Version 2.0. This makes Subunit compatible with more testing
      frameworks.

  IMPROVEMENTS:

    * CPPUnit is now directly supported: subunit builds a cppunit listener
      ``libcppunit-subunit``. 

    * In the python API ``addExpectedFailure`` and ``addUnexpectedSuccess``
      from python 2.7/3.1 are now supported. ``addExpectedFailure`` is
      serialised as ``xfail``, and ``addUnexpectedSuccess`` as ``success``.
      The ``ProtocolTestCase`` parser now calls outcomes using an extended
      API that permits attaching arbitrary MIME resources such as text files
      log entries and so on. This extended API is being developed with the
      Python testing community, and is in flux. ``TestResult`` objects that
      do not support the API will be detected and transparently downgraded
      back to the regular Python unittest API.

    * INSTALLDIRS can be set to control the perl MakeMaker 'INSTALLDIRS'
      viarable when installing.

    * Multipart test outcomes are tentatively supported; the exact protocol
      for them, both serialiser and object is not yet finalised. Testers and
      early adopters are sought. As part of this and also in an attempt to
      provider a more precise focus on the wire protocol and toolchain, 
      Subunit now depends on testtools (http://launchpad.net/testtools)
      release 0.9.0 or newer.

    * subunit2junitxml supports a new option, --forward which causes it
      to forward the raw subunit stream in a similar manner to tee. This
      is used with the -o option to both write a xml report and get some
      other subunit filter to process the stream.

    * The C library now has ``subunit_test_skip``.

  BUG FIXES:

    * Install progress_model.py correctly.

    * Non-gcc builds will no longer try to use gcc specific flags.
      (Thanks trondn-norbye)

  API CHANGES:

  INTERNALS:

0.0.2
-----

  CHANGES:

  IMPROVEMENTS:

    * A number of filters now support ``--no-passthrough`` to cause all
      non-subunit content to be discarded. This is useful when precise control
      over what is output is required - such as with subunit2junitxml.

    * A small perl parser is now included, and a new ``subunit-diff`` tool
      using that is included. (Jelmer Vernooij)

    * Subunit streams can now include optional, incremental lookahead
      information about progress. This allows reporters to make estimates
      about completion, when such information is available. See the README
      under ``progress`` for more details.

    * ``subunit-filter`` now supports regex filtering via ``--with`` and
      ``without`` options. (Martin Pool)

    * ``subunit2gtk`` has been added, a filter that shows a GTK summary of a
      test stream.

    * ``subunit2pyunit`` has a --progress flag which will cause the bzrlib
      test reporter to be used, which has a textual progress bar. This requires
      a recent bzrlib as a minor bugfix was required in bzrlib to support this.

    * ``subunit2junitxml`` has been added. This filter converts a subunit
      stream to a single JUnit style XML stream using the pyjunitxml
      python library.

    * The shell functions support skipping via ``subunit_skip_test`` now.

  BUG FIXES:

    * ``xfail`` outcomes are now passed to python TestResult's via
      addExpectedFailure if it is present on the TestResult. Python 2.6 and
      earlier which do not have this function will have ``xfail`` outcomes
      passed through as success outcomes as earlier versions of subunit did.

  API CHANGES:

    * tags are no longer passed around in python via the ``TestCase.tags``
      attribute. Instead ``TestResult.tags(new_tags, gone_tags)`` is called,
      and like in the protocol, if called while a test is active only applies
      to that test. (Robert Collins)

    * ``TestResultFilter`` takes a new optional constructor parameter 
      ``filter_predicate``.  (Martin Pool)

    * When a progress: directive is encountered in a subunit stream, the
      python bindings now call the ``progress(offset, whence)`` method on
      ``TestResult``.

    * When a time: directive is encountered in a subunit stream, the python
      bindings now call the ``time(seconds)`` method on ``TestResult``.

  INTERNALS:

    * (python) Added ``subunit.test_results.AutoTimingTestResultDecorator``. Most
      users of subunit will want to wrap their ``TestProtocolClient`` objects
      in this decorator to get test timing data for performance analysis.

    * (python) ExecTestCase supports passing arguments to test scripts.

    * (python) New helper ``subunit.test_results.HookedTestResultDecorator``
      which can be used to call some code on every event, without having to
      implement all the event methods.

    * (python) ``TestProtocolClient.time(a_datetime)`` has been added which
      causes a timestamp to be output to the stream.