CHANGES
=======

1.4.5
-----

* Fix python\_requires according to setuptools 67.1.0's requirements (fixes #16)
* Fix README's build status link
* Remove test\_requirements.txt (use testing extra dependency instead)
* Remove usage of deprecated pytest API from unittest

1.4.4
-----

* Fix GH actions

1.4.3
-----

* Only remove the added hook
* Fix GH actions
* Add GH actions

1.4.2
-----

* add python 3.10
* fix ast.alias call in Python 3.10+

1.4.1
-----

* Remove import of \_pytest (fixes #12)
* change requirement from attr to attrs
* Add 'attr' dependency requirement

1.4.0
-----

*  Add atomicwrites to requirements
* Update code according to pytest 5.2.1
* Switch to PBR & Drop support for python version < 3.5

1.3.4
-----

* Bump version
* Update Travis YAML, drop 3.3-3.4
* Pass filename to ast.parse()
* Fix README
* Fix README
* Remove pytest-capturelog from deps
* Switched broken pypip.in badges to shields.io

1.3.3
-----

* Bump version

1.3.2
-----

* Fix argument passing for format\_explanation

1.3.1
-----

* Bump version
* Change API name to disable/enable\_message\_introspection

1.3.0
-----

* Bump version
* Add test to disable\_introspection feature
* Remove pypy from travis
* Allow disabling introspection

1.2.0
-----

* Bump version
* Development docs
* Add integration tests for messages
* Add DEVELOPMENT docs
* Added 'Paweł Adamczak' to AUTHORS
* Added 'Possible exit codes' section to docs (#2239)
* Move CHANGELOG entry for #2276 to 3.0.8
* Add AUTHORS & CHANGELOG
* Change ValueError to io.UnsupportedOperation in capture.py. Resolves issue #2276
* Mention the src layout as recommended practice
* Post 3.0.7 release handling
* Add more information to test-layout docs as discussed during PR
* Clarify that record\_xml\_property is experimental, not junitxml
* Include release 3.0.7 announce in index.rst
* Update regendoc
* Bump to version 3.0.7 and update CHANGELOG
* Fix typo
* Attempt to clarify the confusion regarding \_\_init\_\_ files and unique test names
* Fix trailing whitespace in terminal output
* assert.rst: typographical correction
* added in the correct alphabitcal order
* added in the correct alphabitcal order
* Allow py37-nightly to fail on Travis
* added link to changelog
* added to authors and changelog
* fixed internal error on unprintable raised AssertionErrors
* Improve docs for yield-fixture and with statement a bit
* junitxml: Fix double system-out tags per testcase
* Simplify travis.yml with tox environment variables
* Add note documenting #2257
* Correct typo
* Fix typo in docs "textures" -> "fixtures"
* Spelling fixes
* Python 3.6 invalid escape sequence deprecation fixes
* Improve pytest\_plugins docs
* Update changelog
* Fix docs
* Update docs as requested
* Safer sys.modules delete
* Mention next training event

1.1.0
-----

* --override-ini now correctly overrides some fundamental options like "python\_files"
* Ignore errors raised from descriptors when collecting fixtures
* Remove an unneeded \`except KeyboardInterrupt\`
* Test for \`pytest.mark.xfail\` with non-Python Item
* Update authors/history
* Do not asssume \`Item.obj\` in 'skipping' plugin
* extract a \_handle\_skip method, secure PY2 branch
* Replace 'raise StopIteration' usages in the code by 'return's in accordance to PEP-479
* fixing code-style, keep flake8 happy
* Allow to skip unittests if --pdb active
* Fix reference
* Discourage users from using this all the time
* Fix release date for 3.0.6
* Add py37-nightly to travis
* Add py36 identifier
* bump version to next dev
* take off author\_email after pypi rejects
* fix copy+paste error, its supposed to be 3.0.6
* rerun regendoc with correct install
* generate the release announcement
* fix the xfail docstring typo at the actual docstring + regendoc
* regendoc
* bump version
* fix devpi test for doctesting env
* update changelog as suggested
* remove unused import
* comment out compatproperty deprecations
* Assert statements of the pytester plugin again benefit from assertion rewriting
* Fix pytester internal plugin to work correctly with latest versions of zope.interface
* Ignore flycheck files
* rewrite: support for osx
* Drop support for 2.6, add 3.6
* Fix marker example on "linux" platform
* Revert "Test dessert is working"
* Test dessert is working
* Remove mention of --assert=reinterp in documentation
* Refactor plugin specs handling into an isolated function
* Consider plugins loaded by PYTEST\_PLUGINS for assertion rewrite
* Fail assert\_outcomes() on missing terminal report Currently if the terminal report of testdir.runpytest() is missing, assert\_outcomes() on its output fails because parseoutcomes() returns an unexpected value (None). It's better to fail parseoutcomes() directly
* Ensure the LICENSE is included in built wheels
* rewrite: support for osx
* Add py36 to test on AppVeyor
* Switch monkeypatch fixture to yield syntax
* Fix broken links
* Fix CHANGELOG.rst
* Update AUTHORS, CHANGELOG
* 'xfail' markers without a condition no longer rely on the underlying \`Item\` deriving from \`PyobjMixin\`
* Fix typos
* Sort issues and user references in CHANGELOG
* Move module error compatibility code to \_pytest.compat
* Test Python 3.6 on Travis CI
* Tests: Check for ModuleNotFoundError on Python 3.6+
* Fixed #2148 - parse directory names properly when args contains ::
* Allow trial environments to fail on CI for now
* Update AUTHORS and CHANGELOG.rst
* Improve error message when pytest.warns fail
* Fix wrong fixture name. Closes #2143
* Explicitly add setuptools to install\_requires
* Avoid importing asyncio directly because that in turn initializes logging (#8)
* Add CHANGELOG entry for #2129
* Move compat tests to a single file using testdir
* Use inspect to properly detect generators. Fixes #2129
* Mention that Python also adds CWD to sys.path using python -m
* Also update yield teardown for Issue #687
* fix #2118 - rework Node.\_getcustomclass and Node compat properties
* Add release dates to CHANGELOG entry
* Bump version to 3.0.6.dev0
* Regendocs for 3.0.5
* Add release announcement for 3.0.5
* Finalize CHANGELOG for 3.0.5
* Bump version to 3.0.5
* Use a simple \`\`+-\`\` ASCII string in the string representation of pytest.approx In Python 2
* Fix minor typo
* Fix typo in docstring of register\_assert\_rewrite
* Remove hook proxy cache
* Fix false-positive assert rewrite warnings when using 'pytest\_plugins'
* Add warning for incorrect passing args to \`-o\`
* Remove unused import
* Remove support code for earlier Python 3 version in Source.compile
* minor: fix grammar with help for --setup-{only,show}
* Improved description of functionality for Issue #687
* Add \`type\` validation
* Use "inc" instead of "func" in the snipped on README and doc index
* Improve CHANGELOG entry for #2034
* Show name argment in compatproperty deprecation message
* update CHANGELOG and AUTHORS
* Add compatproperty deprecation warning
* Remove setuptools pin now that upstream has been fixed
* Add documentation building note
* Add test case for #595
* Add \`:ref:\` targets to \`recwarn.rst\`
* Pin setuptools to < 29 because of AppVeyor failures
* Clarify test discovery docs
* Add hint of Issue #478 to error text
* Only execute "coveralls" toxenv on master once
* Fix error in approx's repr with complex numbers
* Show an error if --confcutdir is not a valid directory
* Remove an accidental double space
* Fix test\_excinfo\_getstatement that broke because of whitespace changes
* Fix flake8 E305 and E306 errors
* Don't fail if imp can't find the source for a .pyc file. #2038
* Fix spelling mistakes
* Docs: Added pytest promotional talk in Russian
* Only install pypy on AppVeyor for "pypy" tox-env
* Use one job for each tox env on AppVeyor
* Use a wrapper script to bypass check-manifest if not under git
* Allow failure of pypy on AppVeyor
* Allow failure of py35-trial on Travis
* Bump version to 3.0.5.dev
* add pygments dependency because of rst-lint
* Check README.rst with rst-lint
* Add some recursive-exclude related to hypothesis and freeze
* Add keywords to setup.py as suggested by pyroma
* Add "check-manifest" to linting and remove unused scripts from root
* Regen doc for 3.0.4
* Version bump to 3.0.4, CHANGELOG, announcement
* Fix the stubborn test about cyclic references left by pytest.raises
* added test (thanks @nicoddemus) and added links in Changelog
* Fix memory leak with pytest.raises by using weakref
* Clean up unittest TestCase objects after tests are complete (#1649)
* Change version to be in normal form according to PEP 440
* Properly handle exceptions in multiprocessing tasks
* Fix teardown error message in generated xUnit XML
* Update "Next Open Trainings"
* Report teardown output on test failure
* Restore pexpect tests and bypass isalive/wait on macOS. Ref #2022
* Also use flush where wait was called unconditionally
* Extract child flush as a staticmethod
* Re-enable docstring testing of \_pytest modules on CI
* Fix false-positive warnings from assertion rewrite hook
* Fix cmdline help message for custom options with two or more metavars
* Disable py35-trial while #1989 is not fixed
* Remove example of obsolete pytest.main usage with string
* docs: remove mention of string args to main
* Fix link in CHANGELOG for #1853
* Handle import errors with non-ascii messages when importing plugins
* Update fixture.rst
* Mention small doc fixes don't need tests/changelog entries in PR template
* Add note about not monkey-patching builtins (#1986)
* Fix #1981, improve ini-options help text
* Display traceback from Import errors using pytest's short representation
* Display full traceback from Import errors when collecting test modules
* Fix test about pytest.skip message being used at global level
* Improve error message when using pytest.skip at module level
* post 3.0.3 release changes
* Include release-3.0.3 into the announce toctree
* Regendoc for version 3.0.3
* Prepare for 3.0.3 release
* Use hypothesis >= 3.5.2
* Add a summary on how to skip all tests in a module in different situations
* Add a note about pytest.skip not being allowed at module level
* Vendor pluggy 0.4.0
* Disable pypy on AppVeyor until #1963 gets fixed
* Pin hypothesis to 3.5.0 because 3.5.1 breaks the test suite
* Remove implementation of \`\_\_getslice\_\_\`
* Explicitly set  to None to have consistent behavior in Python 2 and Python 3
* Fix \`DeprecationWarnings\` found when running py.test in Python 2.7 with the -3 flag
* Explain a bad scope value to the user
* added a test
* added to changelog and authors
* AttributeError chaining bug #1944 fix
* fixup! Merge pkg\_resources workaround
* better doc for #1890 based on #1932
* Fixed assertion rewriting for plugins in development mode
* Remove duplicated for Metafunc members on docs
* Fix link to PDF docs
* docs: remove unused helper script
* docs: remove the need for special scripts
* Mention register\_assert\_rewrite explicitly
* Fix keyword docs
* Fix reference docs for monkeypatch
* The "ids" argument to "parametrize" again accepts unicode strings in Python 2
* Send only minor/major release announces to TIP mailing list
* Set version to 3.0.3.dev0
* Fix release 3.0.2 release announcement list of authors
* Fix use of deprecated getfuncargvalue method in the internal doctest plugin
* Fix version typo in announce for 3.0.2
* Run regendoc for 3.0.2 release
* Fix reportingdemo call to pytest
* Changes for 3.0.2 release
* spelling
* integrating review commets of @nicoddemus
* adding corresponding test, authors and changelog
* Add CHANGELOG entry for #1888
* Fix issue where pytest\_plugins as string was marking wrong modules for rewrite
* avoid tearDown and cleanup for unittest debugging
* pdbcls ipython example with tab completion
* docs warnings as errors and initial page similar to README
* Remove example of "monkeypatch" used in session scope
* Update AUTHORS and CHANGELOG
* Fix UnicodeEncodeError when string comparison with unicode has failed
* Add buffer attribute to stdin stub
* Moved import pkg\_resources to else clause
* Improve error message when passing non-string ids to pytest.mark.parametrize
* Updates for new patch version 3.0.2
* Changes for 3.0.1 release
* Fix internal error when parametrizing using and empty list of ids()
* mend
* Added #1853 to changelog as requested on PR
* Added myself as AUTHOR as requested on PR
* Import pkg\_resources only when necessary
* Delete unreferenced setup.cfg from docs
* Fix code which guesses parametrized scope based on arguments
* Update AUTHORS and CHANGELOG for #1837
* Add acceptance test for module loader
* Fix documentation
* Fix importer call
* Remove duplicated name
* mark plugin: move the unclean marked parameter extraction
* Update HOWTORELEASE based on the 3.0.0 release
* minor: s/no pkg\_resources/pkg\_resources/
* Fix spelling: s/outside a/outside of a/
* Remove duplicate entry for #717 in CHANGELOG
* py.test team -> Pytest team :)
* Fix typo in 3.0.0 release announcement
* Add 3.0.0 release announcement to index.rst
* Fix regression when using importorskip at module level
* Bump version to 3.0.1.dev
* Remove duplicate CHANGELOG messages
* remove dot in pytest
* add missing @matthiasha link
* Mention doc update in CHANGELOG
* Update AUTHORS
* Documentation update for rootdir discovery
* Add release announcement
* Finalize CHANGELOG
* Bump version to 3.0.0
* Run regendoc for 3.0 release
* Use "pytest" instead of "py.test" on trial environments
* Tweak CHANGELOG: move "change" entry to the proper place
* Remove obsolete docstring
* Fix conflicts in CHANGELOG
* Move internal \_is\_unittest\_unexpected\_success\_a\_failure to "compat" module
* Update CHANGELOG entries
* Quick fix for tests in config depended on the current directory
* unittest's unexpectedSuccess should work as non-strict xpass
* Enable py35-trial testenv on Windows
* Fix sys.version\_info expression in xfail marker
* Fix py3 xfail expression evaluation and parametrize strict
* Fix broken test in test\_junitxml
* Use a better xfail reason
* Fix broken test in test\_skipping and add one for strict xfail
* Fix test in test\_junitxml and add one for strict
* Support [tool:pytest] in setup.cfg files
* Revert all invocation-fixtures code
* Deprecate --resultlog cmdline option
* Update test\_junitxml.py to interpret XPASS as passed
* Add CHANGELOG entry for #1809
* Improve display of continuation lines with multiline errors
* Fix off-by-one error with lines from request.node.warn
* Pass exitstatus to pytest\_terminal\_summary hook
* Set wasxfail only for xpass w/o strict and else set longrepr
* Extend test to verify longrepr in stdout
* Treat unittest.expectedFailure pass as a failure
* Update unittest test to expect failure for an unexpected success
* Add thanks as requested
* Fix #1798 to include errors in total tests in junit xml output
* Add CHANGELOG entry
* Update documentation to describe expected rootdir behaviour
* Fix determining rootdir from common\_ancestor
* Change outcome to 'passed' for xfail unless it's strict
* Add originalname attribute to Function
* Improve TestReport.sections docs a bit
* Add capstdout and capstderr attrs to TestReport
* Add longreprtext property to TestReport objects
* Add more interpreter versions to multipython example
* Make parametrize example deterministic
* Fix missing print()
* Regendoc after more fixes on features branch
* Rewrite asserts in test-modules loaded very early in the startup
* Sort fixture names when a fixture lookup error occurs
* add changelog entry
* fix deselect tests to match reason removal
* Run regendoc again
* Sort yml items to get same results for regendoc runs
* Pass list of params to pytest.main() in docs
* Run regen-docs for pytest 3.0
* Strip invocation-scope suffix when displaying fixture lookup error
* invocation-scoped fixtures show up once with --fixtures
* Fix broken test in py3k
* Highlight the path of file location in error report
* terminal: dont pretend to know the deselection reason
* Add 'invocation' scope option to fixture docstring
* Fix deprecated directive in docstring
* Set version to 3.0.0.dev1
* Freeze docs: PyInstaller hook and wording
* Use PyInstaller for freeze test env
* Point doc links in README to docs.pytest.org
* Fix pdf links in the documentation
* Sort link refs in CHANGELOG
* fix keep-duplicates help line
* Remove some comments and improved changelog
* Add deprecation module to centralize deprecation messages and bits of code
* Fixed scope override inside metafunc.parametrize. Fixes #634
* feature: default behavior now is to ignore duplicate paths specified from the command line. Use --keep-duplicates to retain duplicate paths
* Fixed collection of classes with custom \`\`\_\_new\_\_\`\` method
* Cleanups for #1763
* drop parenthesis around GH issue number
* Fix #1210 display msg for early calls to exit
* fixed conflicts
* added a test for when the indirect is just a string
* Add new target links in CHANGELOG
* Followup to #1718: style/formatting
* Thanks to Tom Viner for his guidance during EuroPython2016 sprint
* Thanking myself in CHANGELOG
* Add changelog entry for this bugfix branch
* Add myself to the AUTHORS
* Adjust test involving FixtureLookupErrorRepr
* Start FixtureLookupErrorRepr with an 'E'
* updated changelog, resolve #1539
* Test case for overriding autouse fixtures
* added myself to authors
* changed error message for unused parametrize name
* Changes variable name so it better describes what it does now
* Cleaner implementation of #1749
* Simplify test in test\_doctest\_report\_none\_or\_only\_first\_failure
* Explain why thanks yourself in the CHANGELOG
* adds the versionadded flag in docs (#1749)
* using @pytest.mark.parametrize instead of calling one test many times from another as suggested by @nicoddemus in pr #1754
* adds a bit of doctest hint on why the key and value getters are separate functions
* Refactors test to be in its own class, as discussed with @nicoddemus
* fixes changelog, I think I got it
* Removed bug reference in changelog because it makes the linting fail
* Separate the option keys and value to avoid importing "doctest" (and tested things like "logging") for argument parsing (fixes #1749)
* Choose the doctest output format in case of failure, still work in progress as a few checks fail (related to #1749)
* Implements an option to choose the doctest output format in case of failure. (fixes #1749)
* Add changes to changelog
* Uppercase first word in docstrings. Change to an imperative form. Add name to authors
* Log LineMatcher output in a stream instead of stderr
* Make monkeypatch invocation-scoped
* Add more tests for invocation scoped fixtures
* Small documentation improvements
* skipping plugin: remove python2.5 compat code
* record --exitfirst change in changelog
* add test for --maxfail=NUM overiding -x
* implement exitfirst as store\_const option
* update changelog
* argument parsing: always warn for string types
* optparse compatibility - add float and complex
* add backwards compatibility policy
* Document the re-writing of plugins
* setup\_\* and teardown\_\* functions argument now optional
* Remove assertion reinterpretation
* Make assert in test\_str\_args\_deprecated more resilient
* Added confcutdir in testing/test\_conftest.py::test\_conftest\_import\_order and   testing/python/fixture.py::TestAutouseManagement::()::   test\_class\_function\_parametrization\_finalization to avoid problems with   abandoned conftest.py files in /tmp dir.   Fixes #1536
* Fixup things after rebase
* Document limitations for yield-tests in nose
* Remove duplicated changelog entry and formatting fix
* Deprecate support for passing command-line as string to pytest.main()
* funcarg\_prefix\_warning: remove newline
* Introduce pytest.register\_assert\_rewrite()
* Avoid rewrite warning for inline runs
* Enable re-writing of setuptools-installed plugins
* Improve overall CHANGELOG formatting and consistency for 3.0
* De-404 links in changelog
* Fix 2.10 -> 3.0 versions in docs
* Add punctuation to funcarg\_prefix\_warning
* Add tests for custom pdb class
* Replace all usages of "pytest\_funcarg\_\_" for @pytest.fixture
* Deprecate "pytest\_funcarg\_\_" prefix to declare fixtures
* Deprecate yield tests
* Add test to ensure capsys and capfd error out when using "getfixturevalue" inside a test
* Improve docs
* Add CHANGELOG for invocation-scoped fixtures
* Allow passing a custom Pdb subclass via --pdbcls
* Add acceptance test for invocation-scoped fixtures
* Implement invocation-scoped fixtures
* Add documentation for "invocation" scoped fixture
* split most fixture related code into own plugin
* Clean up changelog
* fix duplicate target in changelog
* merge from master to features
* \* implemented changes recommended by nicoddemus
* Fixing link to issue and creating testcase that shows that it finds the line in the stderr lines
* Making conftest import failures easier to debug
* Split AppVeyor test runs in multiple jobs to avoid timeout issues
* Fix links and removed 404 links from talks.rst
* Fix PEP-8
* Implement --setup-show cli flag
* Merged branch features into features
* Add changelog to requirements for pytest-dev plugins
* remove unsupported python versions from code/source xfail
* Document the interaction of autouse scopes
* docs: no longer include python 3.0-3.2 in the index page
* fix typo and remove python3.2 from readme
* drop python 3.0-3.2 support code from setup.py
* catched -> caught
* mark tests: use better name of the test for #568
* Fix CHANGELOG: obestwalter appearing twice due to mergig separate PRs
* Remove commented out code
* Fix bad merge in CHANGELOG
* Add issue and obestwalter to CHANGELOG
* Improve formatting in CHANGELOG
* Move the freezing function from genscript.py to a new module freeze\_support.py
* Remove deprecated cmd options
* \* added missing link to the referenced issue
* \* removed tailing whitespaces
* Add a test for indirect use of tuple in the assert that should not cause a warning
* Update changelog
* Warn about asserts on tuples (fixes #1562)
* \* now showing pytest warnings summary by default. \* added \`\`--disable-pytest-warnings\` flag to let users disable the warnings summary. \* extended/changed unit tests for the changes in the pytest core
* Improve of the test output for logical expression with brackets.   Fixes #925
* add --override-ini option to overrides ini values
* rephrase changes in CHANGELOG.rst
* add changes in CHANGELOG.rst
* add xfailing test for issue #568
* Add feature proposal to changelog
* Change version in issues section to pytest 3.0 only
* Move document to proposals subfolder
* Add changelog entry for #1564
* Renamed the pdb module and changed unit tests accordingly
* Update changelog for new fixture hooks
* Remove all py.test-X\* entry points
* Improve commenting for setupplan unittest
* Add docs for setuponly and setupplan options
* Move setupplan and setuponly options to their respective modules
* Switch setuponly and setupplan options to a hook-based implementation
* Renamed pytest pdb to debugging which conflicts with python pdb. Combining multiple imports the "import pdb" imports the pytest module as opposed to the python debugger
* Remove note on scoping
* Fix rst bullet point lists
* Add yielded values to function example
* Add alternative approach that uses wrappers
* Add proposed solution using a module function
* Add issues section to proposal doc
* Don't ignore ImportError with setuptools plugins
* Simplify Argument.\_\_repr\_\_
* Update CHANGELOG with #607 and add version 3.0.0.dev1
* Raise CollectError if pytest.skip() is called during collection
* update sprint page to be past tense
* rm global header
* remove links to funding campaign
* Fixes #1503 no longer collapse false explanations
* Start proposal for parametrize with fixtures
* issue1625, name getfuncargvalue to getfixturevalue
* Fix fixture parameter display when ids is a function
* Fix capturing with --setup-only/--setup-plan
* Trivial spelling fix in runtest\_setup.py
* added check for already existing option names to OptionGroup.addoption()
* Use correct links in changelog
* Add to changelog and authors
* Add a test for handling of dynamic requests for fixtures from other fixtures
* Implement --setup-plan option
* Fix the tests (#3)
* merge from master again
* cache.rst: Fix wrong command used
* Fix .format string failures on python 2.6
* merge master into features
* Add tests to make sure expected entry points exist (#1629)
* Fix the tests
* Add printing of fixture dependencies
* Disable capturing for setuponly output
* Refactor logging of fixtures
* Add test for setuponly option
* Add test for failing assertion
* Factor setuponly code out of runtestprotocol()
* fix changelog links
* issue 1553: Include terminal newlines in diffs
* Implement working version of --setuponly
* CONTRIBUTING.rst: spelling fixes
* Update CHANGELOG and add Oliver Bestwalter to AUTHORS
* Introduce pytest command as recommended entry point Fixes #1629
* help the user in the rare case this assertion actually fails
* Exit pytest on collection error (without executing tests)
* Return explicit None from rewrite hook's find\_module
* Remove print left by accident
* Issue 460: getfuncargvalue fixture w. params err
* Update repository transfer instructions
* tests: getexecutable: call \`--version\` on all Pythons
* Update AUTHORS and CHANGELOG
* conftest files now use assertion rewriting
* Ignore hidden files (.testmondata) in test\_pytest\_collect\_file
* Fix test name typo
* Added versionchanged directives
* Updated documentation
* pytest.raises accept cutom message only when used as context manager
* Add disabled() method to capsys and capfd
* CR fixes
* Add CHANGLOG entry
* pytest.raises accpets custom message
* Simplify default pytest\_runtestloop
* Cut a dead test helper function
* Added taschini to AUTHORS and #1597 to CHANGELOG.rst
* Incorporated feedback (#1597). Fixed problem caused in a test on Windows by file left open by PyPy and not immediately garbage collected
* Ensure that a module within a namespace package can be found by --pyargs
* Change format for test function locations
* Fixed issue shadowing error when missing argument on teardown\_method
* Fix py26 by using indices in str.format
* Add changelog entry for new cli flag
* Implement show\_fixtures\_per\_test and add cli flag
* Implement tests for --fixtures-per-test
* Remove dead code and simplify code in call\_fixture\_func
* Remove most content from yieldfixture as it is now deprecated
* Update CHANGELOG (yield statements in normal fixtures)
* Update docs in prol of using yield statements
* Remove yieldctx variable being passed around
* Fix custom name for yield\_fixtures
* Make normal fixtures work with "yield"
* Use same python which invoked tox for "doctesting" env
* [exception handling] Fix case the current working directory (CWD) gets deleted during testing
* Docs: config.option is deprecated
* Use same python which invoked tox for "doctesting" env
* List contributors alphabetically in AUTHORS
* Mention #1580 in the CHANGELOG and add Thomas Grainger to AUTHORS
* Don't mark empty doctest files as skipped, fixes #1578
* Fix rst-lint CHANGELOG error
* fix another copy&paste error in the announcement list
* finish up the version bump in \_\_init\_\_ i forgot
* merge master to features
* prepare cangelog for the next bugfix release
* fix minor typos
* add release announcement
* alter changelog header
* regendoc and comment out a python2.7 example as per #1573
* bump regendoc python version to 3.5
* update version
* ignore hypothesis temp folder
* Remove "experimental" status from hookwrapper
* Implement a test for showfixtures to show fixtures with same name
* Update changelog with --fixtures fix
* Change \_showfixtures\_main to loop over fixturedefs
* merge master
* better message for empty argument skip
* remove the old empty argument triggers unlike the marker based one its not composable
* add missing link to changelog
* fix #510 by adding a describing skip marker on empty parameterize
* Fix shell argument split in win32
* merge from master again
* issue 1544: norecursedirs build & dist dirs
* unittest.UnitTestCase: Allow \_\_test\_\_ for methods
* fix typo
* Move comment in tox.ini due to recent bug in pip
* Convert readthedocs link for their .org -> .io migration for hosted projects
* merge next chunk from master and fix changelog linting issue
* Fixes Issue 1549
* Fix broken tests
* Remove unnecessary dependency
* bump version
* Fix merge issues (missing code)
* Remove erroneous code ignoring assertions with messages
* document reason for #1540
* add xfailing test for #1540
* Fix typo in doc
* pytest\_make\_parametrize\_id receive config object
* Added hook to plugin docs and new CHANGELOG record
* Added test\_pytest\_make\_parametrize\_id
* Added pytest\_make\_parametrize\_id hook
* Raise CollectError if import test module fails
* Update sprint 2016
* Update sprint2016.rst
* Add changelog
* Use py.builtin.callable
* Use a callable \_\_tracebackhide\_\_ for filtering
* Fix tests for python 2.6
* Filter selectively with \_\_tracebackhide\_\_
* Fix TracebackItem documentation in pytest.code
* Add test test\_absolute\_win32\_path
* issue 1496 - xfail with condition keyword
* Fix a small issue about shlex.split not working well with win32 path
* doc: Use ascii chars for file tree
* Perform a "unicode aware" check for maximum recursion depth error
* Hudson -> Jenkins
* Add self to AUTHORS
* Implement ExceptionInfo.match() to match regexp on str(exception)
* Add .hypothesis to .gitignore and try an older version of Hypothesis for 2.6
* Fix the changelog and dependencies for tox
* Add Hypothesis test for \_idval and fix bug it found
* Fix test\_escaped\_parametrized\_names\_xml
* Always lstrip() keyword expression
* Fix a test\_unicode\_idval\_python2 (now test\_unicode\_idval) and associated string handling on Python 3
* Adapt get\_issues.py script for GitHub (instead of Bitbucket)
* Comment a workaround for #1485
* Added thanks and PR links in changlog
* checking first there are duplciates ids before changing to unique names
* Mention Pytest::Framework PyPI classifier on docs
* added entry to CHANGELOG
* improved idmaker name selection in case of duplicate ids
* Add strict parameter to xfail marker doc
* Fix pytest.mark.skip mark when used in strict mode
* Escape both bytes and unicode strings for "ids" in Metafunc.parametrize
* Add 'Minor doc fixes' to CHANGELOG
* Document --full-trace option and KeyboardInterrupt
* Replace --fulltrace with --full-trace in doc
* Added changlog entries
* updated parametrize documentation
* Replace --traceconfig with --trace-config in doc
* Fix minor typo in 'writing plugins' doc
* Fix 'test grouping by fixture instances' doc example
* Add short explanation and link to yield fixtures in fixture doc
* No longer refer to the 'yield fixture mechanism' as experimental (doc)
* Add Jenkins xUnit Plugin link to doc
* Fix minor mistake in test discovery doc
* Fix minor mistake in usage doc (pkg instead of pypkg)
* Update authors & changelog
* Fix issue #138
* added test for unique names when recievieng identical ids in parametrize
* added test for None in paramtrized ids list
* added test for None in idmaker
* allow None to be passed in ids list for specific test and recieve idmaker name for test
* Bump version to 2.9.2.dev1
* Regendocs
* Add announce doc
* Finalize CHANGELOG for 2.9.1
* Bump version to 2.9.1
* Update talks page
* junitxml: add properties node in testsuite level
* added TODO comment to remove the parametrized workaround
* Fix CHANGELOG merge conflicts
* Fix some typos in the documentation
* Fix a few stylistic issues
* added CHANGELOG and AUTHORS entry
* allow parametrized nodes to be specified from command line
* Add "thanks" line to the CHANGELOG
* doc: fix argument order for pytest\_load\_initial\_conftests
* Fix Decimal() and \_\_ne\_\_() errors
* Discuss alternative float comparison algorithms
* Fix 'check' in markdown for github templates
* Properly handle inf, nan, and built-in numeric types
* Use testscollected to make xdist happy
* Copy over name and path attributes to the re-raised exception
* Allow custom fixture names for fixtures
* Improve error message when a plugin fails to import
* Fix sys.version\_info errors
* Fix trailing whitespace errors
* Fix zero-length field name error in python2.6
* Add approx() to the CHANGELOG
* Add approx() to the Sphinx docs
* Reduce the default absolute error threshold to 1e-12
* Use the plus/minus unicode symbol in the repr string
* Make a few stylistic improvements
* Add a convenient and correct way to compare floats
* When a regex pattern contains bytes instead of a string use escape\_encode to turn it into a string before further processing. Thanks @nicoddemus for the review and tips!
* Add test when using native strings with non-ascii chars
* Fix decoding issue while formatting SyntaxErrors during collection
* pytest.fail with non-ascii characters raises an internal pytest error
* Mention that pytest\_addoption may be used from conftest files
* issue469: junit parsing nodeid, add method test
* Add versionadded directive to doctest\_namespace section
* Give proper credit for PR #1428
* issue469: add fix junit double colon split issue
* Reformat versions and header levels in CHANGELOG file
* Minor README tweaks (markup, phrasing)
* Correct indentation in documentation
* Add a new doctest\_namespace fixture
* Add 2.9.1.dev1 to CHANGELOG
* Add 2.10.0.dev1 to CHANGELOG
* Bump version to 2.10.0.dev1
* Bump version to 2.9.1.dev1
* Add company name to flub in sprint attendees
* Fix reference to \_CallOutcome in docs
* Fix typo in CHANGELOG
* Regendocs
* Release announcement for 2.9.0
* Finalize CHANGELOG
* Bump version to 2.9.0
* junit: allow multiple properties with same name
* Add company name to @hackebrot in sprint attendees
* Update changelog with traceback style docs fix
* Add auto to tb styles (new default since v2.6)
* Fix a typo in the docstring for mark.MarkGenerator
* Simplify IndexError test
* catch IndexError exceptions when getting exception source location
* Add test for unconditional skip with reason
* Fix skip/xfail markers in parametrized arguments
* add some more people, and preliminary company affiliations
* Add Raphael Pierzina to AUTHORS
* Add reference to some builtin fixtures in "contents"
* Correct JUnit test invocation example
* fix typo
* Display collect progress only when in a terminal
* Code review suggestions
* Move sections around (from easier to hardest) and added howto for repo transfers
* Add issue and PR GitHub templates
* Add full MIT license text to the docs section
* Add link to license page to index.rst
* Add a license section to the docs toc
* Add a reference to LICENSE file to README
* Wrap lines at 79 characters
* Insert copyright holders and year as mentioned in README
* Grab official MIT license text from opensource.org
* LastFailed now creates .cache only when needed. Fixes #1342
* Fix rst syntax
* Fix strict xfail: it should behave exactly like xfail when a test fails
* Fix bug in strict xfail: test was not being actually called
* Small typo in CHANGELOG
* Add CHANGELOG entry for #1355
* Add strict option to xfail, making tests which XPASS to actually fail the suite
* Add bool type to addini
* update release process to ensure correct usage of base branches and propper finalization
* update participants
* re-introduce topics
* some layout refinements
* adapt the header text based on a discussion with bruno
* give the header contrast
* Fix tox doc task
* add links for the funding campaign in the header and sidebar
* fix formatting utf-8 error explanation
* fix issue #1366 by showing a note on the --fulltrace option
* remove superflous file
* fix names once again
* Also fix Brianna's name
* Fix names properly
* Fix CHANGELOG entry to new format
* Fix #1290: Py3.5's @ operator/assertion rewriting
* fix brianna's name once again
* fix EUR->USD
* add a link to lieve indiegogo campaign in the pytest sprint page
* - preliminary sprint page, not yet linked through the header - also removed plugins\_index call from tox.ini "doc" env because it's not there
* correct merge misstake in appveyor.yml
* merge latest master into features as well
* merge from master
* Add expected exceptions to 'DID NOT RAISE' msg
* Test with pypy and enable coveralls in AppVeyor
* fixed name
* a few refinements
* sprint funding draft
* Add CHANGELOG and docs for #1345
* Add explicit flag names for failed first and last failed
* Dummy change to test readthedocs hook
* Fix version at the top of the CHANGELOG
* Add CHANGELOG entry for py.code merge
* Remove unused import
* mark test\_comments as xfail on pypy
* Drop assertionnew and assertionold from \_pytest.\_code
* Moved py.code code over to py.test
* begin 2.8.8 cycle
* finalize changelog for 2.8.7
* Prepare 2.8.7 release
* add requests dependency to tox.ini to ensure all monkeypatch tests run
* monkeypatch: unnest handling code
* changelog
* work around python 2/3 difference by using str(exception)
* remove unused import
* pass trough annotated exceptions
* fix issue 1338
* reformat monkeypatch core plugin/its tests
* Bump version to 2.8.7.dev1
* Use a full url to pytest logo on README
* 2.8.6 release: version, CHANGELOG
* Add CHANGELOG and AUTHORS entry for #1334
* Add captured stdout to jUnit report on setup error
* Fix practise -> practice typo in documentation
* Add a few missing hooks to the docs
* Fix flakes
* Change \`input\` to \`test\_input\` in docs for clarity
* Always show full comparison output if on CI
* update docs plugin.rst typo
* Add CHANGELOG entry for #1223
* Fixes bug with stdout/stderr capture on pdb
* Fix decode error in Python 2.7 when docstrings contain a non-ascii character
* Correct platform name osx -> darwin
* Add CHANGELOG.rst to MANIFEST and small format fix
* Rename "flakes" testenv to "linting" as requested in review
* pytest warnings emitted during \`\`pytest\_terminal\_summary\`\` are now properly displayed
* Changelog now in rst format (2.9.0 and onward) and add rst-lint check
* Update copyright in README
* Complement #1255 by adding tests and docs
* make --doctest-glob multi-allowed
* Make monkeypatch calls O(1)
* Rename "BugFixes/Adjustments" to just "Bug Fixes" as commented elsewhere
* Add CHANGELOG entry for ALLOW\_BYTES doctest option
* Add docs for ALLOW\_BYTES doctest option
* Implement ALLOW\_BYTES doctest option
* Small typo on README
* Small wording on README
* Add CHANGELOG entry for #900 fix
* Make monkeypatch differentiate ImportError sources
* Updated out of date link
* Add logo and update readme
* Separate 2.9.0 CHANGELOG into sections
* Fix trailing white-space
* Give proper thanks for the PR in the CHANGELOG
* Apply readline workaround during initial conftest loading
* Add badge of supported python versions to README
* \* Moved workaround to its own function, mostly for the sake of adding   a more descriptive docstring (the workaround itself is just to import   readline earlier).  I removed the conditional on targetfd from the   workaround since it doesn't really matter
* Make sure readline has been imported before duping any stdio handles--otherwise pyreadline fails to connect to the correct handle for the console's stdout/in
* Remove "index" entry from overview.rst
* doc typo
* junitxml tests: extend with extra items
* implement review comments for #1266
* Mention #1226 in the CHANGELOG
* Remove references to python pre-2.6 from docs
* Remove obsolete "oldinterpret" module
* Remove code related to support python <= 2.5
* fixes #1259 - take finalized nodes out of the mapping
* Remove experimental status from pytest\_exception\_interact
* Add 2.8.6.dev1 to CHANGELOG
* Bump master to 2.8.6.dev1
* Remove plugins\_index generation from HOWTORELEASE
* Restore regen-doc result
* Regenerate 2.8.5 docs
* Prepare for 2.8.5: bump version, CHANGELOG, announce
* Tyding up the CHANGELOG
* Optimize appveyor build
* fixed meesage
* added docstrign to inection collection test
* added myself to authors, added changelog entry
* added line comparison that is pytest-sugar agnostic
* added verification that test actually passed
* nit: fixed newline
* added test to verify injection
* Fix CHANGELOG merge
* Remove plugins\_index from the docs
* Use https for images in readme
* add a reminder of the features branch merge to HOWTORELEASE.rst
* Update python.py
* Add "no -rP" case to test\_pass\_output\_reporting
* Add tests to test\_terminal.py for -rp and -rP
* Fix deprecated\_call regression introduced in 2.8.4
* update test\_recwarn to new style
* Update AUTHORS and CHANGELOG
* Add -rp and -rP options to report passing tests
* add changelog entry
* handle duplicate test ids via collection and xdist each reporting
* add xfailing test for double test id failure
* Junitxml: correct node reporter attribute names
* junitxml: correct docstring of make\_properties\_node
* junitxml: follow Bruno's attribute/method naming hint from the review
* junitxml: fix python3 compat of the tests
* finalize nodereporters by throwing away the intermediate xml nodes, fixes issue #1074
* reintroduce junitxml report order and debug cleanups
* junitxml: intermediate, move testcase generation to NodeReporter
* junitxml: keep track of custom property insert order
* junitxml: use node.warn to ensure fslocations
* junitxml: introduce nodereporter and track durations/properties there
* junitxml: remove debug print
* junitxml restrucutre stat generation - use node tags
* junitxml: simplify the api used for testing junitml
* junitxml: simplify tests by creating a api wrapper
* junitxml: yapf clean the tests
* junitxml: reverse the if/else logic for failure appending
* yapf junitxml
* bump to 2.8.5.dev
* correct copy&paste error in the release announcement version number
* finish release announcement
* doc regen for release 2.8.4
* prepare release 2.8.4 - changelog updates + version bump
* adopt review comment of #1186
* fix #1185 - ensure MANIFEST.in exactly matches what should go to a sdist
* Add a note about how DeprecationWarning and PendingDeprecationWarning are treated differently
* Fix same inconsistency in next example
* Fix inconsistency in skipif example
* Fix pastebin when captured output contains non-ascii characters
* Fix typo: previosuly --> previously
* Improve description on how pytest starts test collection in goodpractises
* Remove promise about documenting how to create a zipped pytest
* Replace "--assertmode=off" by "--assert=plain"
* Removed incorrect note about genscript requiring wheels
* Remove reference to distutils
* Add #1198 fix to CHANGELOG
* Decode urlopen response in pastebin
* Fix genscript deprecation version and document reasons for such
* Fix formatting errors
* Merged virtual env into the Tox section
* Fixed some formatting
* Removed setuptools/genscript session
* Move deprecated genscript method to the bottom of the document
* Moved "conventions for Python test discovery to the top"
* Remove finalize\_options override from goodpractices
* Update xdist link; BitBucket -> GitHub
* Fix spelling: explicitely --> explicitly
* Fix spelling mistake in #1207
* Fix terminal output if no tests were run
* Don't collect classes with truthy \_\_getattr\_\_
* Fix getting line number with nasty \_\_getattr\_\_
* fix #1198 - decoding monkeypatched data to unicode
* fix #1198 - removed docoding the result
* fixed #1198 issue by encoding the unicode parameters to bytes and decoding the bytes response to unicode
* Fixed docs
* deprecated\_call now uses monkey patching strategy to capture warnings
* Restored 2.7 implentation of deprecated\_call
* Small changelog formatting fix
* Fix formatting in HOWTORELEASE.rst
* Mention pytest\_enter\_pdb in the docs
* Pass pytest's config object to pytest\_enter\_pdb
* Mention pytest\_enter\_pdb in the docs
* bump to 2.8.4.dev
* refer more explicit to osx el cap in the changelogs
* update authors list for 2.8.3 and add a log command to the howto
* correct drop in version in announcement
* run regendoc
* release announcements for 2.8.3
* Changelog groming and release preps for 2.8.3
* Add 'Improve your testing with Pytest and Mock' video from PyCon SG 2015
* Change pytest ignore invocation as suggested by @nicoddemus
* Give credit where it is due
* Fix unittest.skip decorator test and separate the fix into a different, self-doc function
* Suggested edits by Bruno
* Added test for issue #1169
* Fixes #1169
* Add pytest invocation along with result
* Add example tests directory structure
* Start doc section on \`\`--ignore\`\` in pytestcollection.rst
* Use pytest-runner for setuptools integration
* Fix typo in parametrize.rst
* Updated entries on CHANGELOG and AUTHORS. Ref: #1157
* Color highlight on the name of the failed tests
* Fix link in FAQ about parametrizing tests
* Update usage.rst
* Update CONTRIBUTING.rst
* Remove commented out extrainit
* Update plugins\_index.rst
* Make pytest-version mandatory
* Add link to unittest2pytest
* Handle EPERM when writing rewritten bytecode
* Fix situation where a traceback entry "path" returns a str object
* collection: Prevent non-function callables from being collected
* Use a string for pytest\_plugins in docs
* bump version
* Travis
* Require new emport in dependencies
* Automatic pypi deployment
* When activating dessert, take precedence over other meta\_path entries
* Test dessert is working
* Add 3.5 to Travis
* Fix tox config
* Python 3.4+ compatibility
* [doc] Fixed indentation
* issue1035 Override inspect.isclass for python 2.6
* strengthen the ini assertion
* changelog entry
* all tests pass
* an ugly patch to fix all but the most important part =/
* unit tests of Config.fromdictargs. currently failing
* issue1035 add test for classes setting \_\_getattr\_\_
* Add empty 2.8.3.dev section to CHANGELOG
* Fix contents.rst
* start 2.8.3 development
* Add 2.8.2 release annoucement
* Update plugin\_index
* Prepare 2.8.2 release
* typo fix, changed position of context manager notes
* Fix problems when mixing autouse fixtures and doctest modules
* Add credit for pytest.mark.skip to the CHANGELOG
* TestXFail also shouldnt explicitly inherit from object
* Add unconditional skip entry to CHANGELOG
* Update default reason to "unconditional skip"
* Update reason in test to prevent confusing with test\_no\_reason
* Add myself to AUTHORS
* Test that "unconditional skip" is the default reason if none given
* Remove redundant comments
* Dont explicitly inherit from object
* newlines
* First argument in pytest.mark.skip is a reason
* Check no reason displayed if none specified
* Fix failing test
* classes inherit from object
* Update skippings tests for better coverage
* Update skipping.rst with correct version marker
* Add a test\_skip\_class test
* Spelling and grammar fixes
* Delete trailing whitespace
* Update docs with new skip marker
* Fix case where skip is assigned to as an attribute directly
* Work towards test coverage of mark.skip
* Remove incorrect use of pytest.mark.skip
* Add MarkEvaluator for skip
* Add a reminder on HOWTORELEASE to update plugins\_index
* plugins\_index was still generating ".txt" files, so it was effectively being ignored
* Move "note" to the end of the main text
* pyflakes fix
* Add a note about usage of the context manager
* Detect dynamic code explicitly in filter\_traceback
* Move imports outside \_escape\_bytes as suggested in review
* Use a variable for the skipif marker as suggested by @nicoddemus
* Implement skipif marker for slow test based on pytest.config
* Remove pytest\_runtest\_setup from conftest.py
* Edit examples again to use hookimpl
* Correct hook examples in docs
* fix docstring for cache fixture regarding \`\`/\`\` on keys
* add entries for #1085 and #1087 to the CHANGELOG
* Fix internal error when filtering tracebacks where one entry was generated by an exec() statement
* Fix SystemError when using unicode\_escape on Python 3
* Fix encoding errors for parametrized tests with unicode parameters in py2
* fix flaky get\_user fallback tests
* remove unclear todo that irritates in the online docs
* start 2.8.2 development
* avoid pullrequest picture and fix bugfix/feature target info. I could not get the pullrequest picture to be included correctly in both latex and html sphinx builders due to the "include" directive from doc/en/contributing.rst
* hide some parts from generating latexpdf
* - fix a flaky test on py35-xdist by calling   importlib.invalidate\_caches()
* Add missing nodeid on pytest\_logwarning call in addhook
* Show known environment variables in py.test --help
* Remove print left by accident
* Moved more detailed options on how to disable a plugin to plugins.rst
* Check \`deprecated\_call\` specific to deprecation
* \`deprecated\_call\` detects pending warnings again
* simplify by removing the single-call "exclude\_pytest\_names" function
* add changelog entry for fix #1073
* seems like pypy's callable builtin calls \_\_getattr\_\_ so we do the check later
* fix issue #1073 -- shortcut plugin hook lookup if the attrname is not prefixed with pytest\_
* add changelog note for fix #704
* Vendor pluggy-0.3.1
* Test py27-xdist and py35 environments on AppVeyor
* Test py35 on AppVeyor
* Update travis with new tox environments
* Use py35 as preferred py3 interpreter for special test environments
* Update CHANGELOG
* Add note on docs about record\_xml\_property not working with xdist
* Fix xml generation when used with pytest-xdist
* Clarify documentation
* Reorganize "Contents" page
* Small format fixes in HOWTORELEASE
* Mark sources correctly in parametrize.rst
* Fix test\_time in parametrize doc
* Regenerate fixture list doc with cache docstring
* Docs: Fixtures page - clarify usage of module-level pytestmark
* Add docs on how to disable cache provider
* Remove duplicate line
* (w)warnings -> (w)pytest-warnings in "-r chars" help
* Add docstring to cache fixture
* Update CHANGELOG
* escape bytes when creating ids for parametrized values
* Write failing test for parametrized tests with unmarshable parameters
* skip chmod using cache access warning tests on windows
* Don't hardcode 'std' for captured stdout/stderr
* add an acceptance test for cache write errors
* generate docs on python3.4
* Added minor documentation change
* update changelog
* Minor typo in CHANGELOG
* handle access errors when writing cache files silently as pytest warning, fixes #1039
* Fix git checkout command in CONTRIBUTING.rst
* fix erroring doc example
* update regendoc, substitutes PYTHON\_PREFIX
* fix documentation gegeneration, kill Makefile
* remove nonsense line
* fix changelog
* re-add 2.8.x changelog so that MASTER can be merged into features wrt to the changelog
* also change pytest version to target 2.9.0
* start features branch
* start bugfix changelog on master
* fix another occurence of "macro" to "major"
* fix naming
* update according to discussion on pytest-dev and in PR
* removed mutation of global state, changed filter addition in WarningsRecorder
* Fixes #653 use deprecated\_call as context\_manager
* Update changelog for merged PR
* Add \_\_eq\_\_ method to assertion comparison example
* adapt contributing to new versioning and to python3.5 as latest
* Mention fix for issue 766 in changelog
* Remove references to distutils as per Issue 766
* contributing: Add instructions from PyCon
* Make tmpdir resiliant to user ids that do not exist
* #877 - reencoding causes typeerrors on python2
* reencode non-ascii python2 assertion reprs, fixes #877
* Bump version to 2.8.0
* White list rm for cxfreeze env
* Fix install cx\_freeze
* Fix import
* Fix flaky failure about unused shutil
* Use temporary directory for cx\_freeze testing
* junitxml: move custom properties to properties element
* re-vendor pluggy to 0.3.1 release and also fix the README
* update changelog/author info, bump version number
* Various Python 3.5 documentation mentions
* revert setuptools\_scm addition introduced with f22d14b105553e99946ccdbe8c47bc450863d843
* remove \_\_init\_\_.py as advised by ronny
* Reorganize tests
* revert importing test module behaviour to prepend by default
* Use temproot as a fallback if the current user couldn't be obtained
* avoid oldstyle funcarg usage
* some more doc refinements
* refine command line option naming and docs
* test\_get\_user should execute on windows only
* Extract get\_user logic into a separate function
* Make tmpdir more resilient in case environment variables required by getpass are missing
* improve and integrate docs
* Use \_\_code\_\_.co\_argcount
* Add tests for argument formatting
* Fix lookup of special values
* Fix checking for args
* Simplify
* Use inspect.signature instead of inspect.getargspec
* fix flakes problem, configure devpi upload to always upload wheels and sdist
* fix a few issues with pytest-2.7 branch and bump version number - importorskip: skip a test if we have a minversion but cannot parse   version numbers due to pkg\_resources not present - make runner tests work with latest xdist
* restrucure pytest.main.wrap\_session to allow for non-testrun wraps
* Remove unused run\_with\_env script
* Use pip and tox from py27 in appveyor
* Use python "3.5" on travis
* Ignore vendored\_packages folder for coverage purposes
* monkeypatch: ensure the internal \_savesyspath attribute always exists
* more pep8 fixes
* minor flake8 fixes
* update docs on the json usage of builtin cache
* restrucure pytest.main.wrap\_session to allow for non-testrun wraps
* minor doc reference fixes
* create the previously missing cache fixture
* use flake8 in the flakes testenv
* s/--cache/--show-cache/
* merge the pytest-cache plugin into core
* Add note in changelog about issue 960
* Added testcase and help for report option
* Added fix for specifying every option for reporting
* Fix --target parameter for pip install
* Keep \*dist-info directory on vendored\_packages
* Create \`\_pytest.vendored\_packages\` and \_pytest.\_pluggy
* Document single-method usage
* turn the pytest\_logwarning hook historic
* Make doctest SKIP option register doctests as SKIPPED rather than PASSED
* Add CHANGELOG entry about renaming "warnings" to "pytest-warnings"
* replay initial warnings when terminalreporter is loaded
* Add docstring
* Fix overriding of fixtures with parametrization
* Adjust phrasing in monkeypatch.undo docstring
* Add notes to docstring of \`monkeypatch.undo()\`
* Show a few diff lines when truncating string diffs
* Add note that using marks on fixtures is not supported
* Rename "warnings" to "pytest-warnings" in terminal output
* Fix regression caused by changing the signature for parse\_known\_args
* Add warning about implementing pytest\_addoption from conftest files
* Mention cookie-cutter in the docs and fix broken links to <extplugin>
* Paths after normal options are now properly used to discover rootdir and ini files
* Improve docs and using warning system for record\_xml\_property fixture
* Fix getdoctarget to ignore comment lines
* Always report error about parametrize data that doesn't correspond to fixtures in test functions
* Add warning of "preliminary feature" to record\_xml\_property
* Correcting implementation based on pull request feed back
* Fix forked\_run\_report in pytest.xdist 1.13
* Add support to record custom properties on xml output
* Using packages from pypi
* Force to use xdist >= 1.13
* Make testsfailed and testscollected public Session attributes
* return non-zero exit code if no tests are collected
* terminalwriter: extract plugin printing logic and add positive unittests
* terinalwriter: use dash between plugin name and version
* adapt plugin printing
* Fix coverage
* Add docs and CHANGELOG for ALLOW\_UNICODE option
* Add ALLOW\_UNICODE doctest option
* Fix accidental inversion in skip example
* Add myself to authors
* Update changelog for #926
* Don't skip fixtures that are substrings of params
* Use "union" merge strategy for CHANGELOG
* Reintroduce hasplugin to PytestPluginManager
* update changelog
* use local isclass
* check nose.istest \_\_test\_\_ attr issue526
* Change cx\_freeze example to not use regendoc
* Replaced \_\_multicall\_\_ examples in docs by hookwrapper
* add test for @nose.tools.istest
* Added more test cases
* Move enum import and adjust comments
* More style changes
* Fix PR image link
* Clean up type comparisons
* Update CHANGELOG
* Add CHANGELOG entry
* Alert user about other parametrize spellings
* Generate parametrize IDs for enum/re/class objects
* Give credit where it is due in CHANGELOG :smile:
* Use testdir fixture in test\_double\_test to ensure controlled environment for execution
* Warnings always raised in WarningsRecorder
* Updated CHANGELOG and AUTHORS
* Converted documentation from txt to rst
* Move WarningsChecker import in test to a local import to workaround flakes issue
* Added makefile changes for docs
* Updated documentation from txt to rst
* Update docs
* Some refactorings after code review
* fix docs folder name in HOWTORELEASE
* Update docs to represent reality
* Update test for error in 'indirect' parameter
* Add test for bad value in argument 'indirect'
* Update CHANGELOG and AUTHORS
* Rewrite test using @pytest.fixture
* Support for python 2.6
* Request #714: Apply indirect=True on particular argnames
* Added \`warns\` to assert warnings are thrown
* Added myself to AUTHORS
* Use smtp.gmail.com server for examples and fixes examples for python 3
* Fix regendoc section in fixture.txt
* Use pytest-of-$user as base directory for tmpdir\_factory
* special-case \_pytest.\_\_init\_\_ in genscript to avoid a python3 bug
* Delete release.txt from docs, merge extra comments to HOWTORELEASE.rst
* Document and test stacking of parametrize
* fix version import
* minor tweaks to CONTRIBUTING
* Mention fix for issue #683
* Test file run twice fails if it contains marked class #683
* Mention Contributors team by name in CONTRIBUTING
* Add changelog entry for capture improvement
* fail if snap is removed in the future
* reset capture even if readouterr throws
* changed contents -> content but still #753 did not solved, might be issues with regendoc
* Update changelog
* Issue #751 - fix
* Issue #751 - test
* Deprecate and warn about --genscript, fixes #730
* setuptools not present issue deprecating gentest
* use setuptools\_scm to determine the version
* Re-enable Python 3.5 testing on Travis
* Workaround for cmdexec bug on Windows
* Workaround for cmdexec bug on Windows
* Use monkeypatch for setting modules in tests
* importorskip: Allow non-integer version strings
* Interpret strings to "plugins" arg in pytest.main() as module names
* Avoid detecting conftest files upwards from setup.cfg/pytest.ini/tox.ini files by default
* Fix basepython for tox envs "doctesting" and "coveralls"
* Fixed typo in CHANGELOG: "testdirs" -> "testpaths"
* Fix flakes
* Make pastebin use \_pytest.config.create\_terminal\_writer
* Fix autouse fixtures defined in a TestCase subclass
* remove merge artifact
* Fix flakes
* Report correct reason when using multiple skip/xfail markers
* Consider --color option in more places which deal with TerminalWriters
* Allow py35 to fail on Travis until it is properly supported (fix2)
* Allow py35 to fail on Travis until it is properly supported (fix2)
* Allow py35 to fail on Travis until it is properly supported (fix)
* Allow py35 to fail on Travis until it is properly supported
* Fix broken links
* Style fix
* Match .travis.yml env list to tox envs
* No more failures
* OK, try running flakes with 2.7
* Try running flakes tests with Python 3.4
* Fix AST rewriting with starred expressions in function calls
* Move Interrupted exception class out of Session
* No Starred node type on Python 2
* generify
* simplify + fix
* a test
* nigh
* try isntall 35 on tox
* allow faillure on 35
* Ast Call signature changed on 3.5
* appveyor scripts
* appveyor integration
* Use functools.partial name explicitly and simplify the code a bit as asked in review
* Support for tests created with functools.partial
* Use a subdirectory in the TEMP directory to speed up tmpdir creation
* Make sure marks in subclasses don't change marks in superclasses
* Style fix
* Update sidebar links from Bitbucket to GitHub
* Update CONTRIBUTING documentation with GitHub instructions
* Fix flakes
* Use a subdirectory in the TEMP directory to speed up tmpdir creation
* Match .travis.yml env list to tox envs
* No more failures
* OK, try running flakes with 2.7
* Try running flakes tests with Python 3.4
* Fix AST rewriting with starred expressions in function calls
* Move Interrupted exception class out of Session
* No Starred node type on Python 2
* generify
* simplify + fix
* a test
* nigh
* try isntall 35 on tox
* allow faillure on 35
* Ast Call signature changed on 3.5
* Mention pytest-runner in the docs
* Pinning mock module to < 1.1 fo py26
* Consider session fixtures for doctest docstrings in modules
* Reword CHANGELOG a bit
* --fixtures only shows fixtures from first file
* Add support for PEP302 get\_data API
* Print inifile and rootdir when there's usage errors
* Add new testdirs ini option
* Pinning mock module to < 1.1 fo py26
* Add syntax highlighting to missing snippets in the documentation and other fixes
* configure a plain "devpi upload" to create sdist and universal wheel
* Fix typo + grammar
* Fix flaky test\_timing\_function in Windows
* Reword entry in the changelog to show case the new status bar colors better
* Add small thanks note to Eric Siegerman
* Add \`file\` and \`line\` attributes to junit-xml output
* Add CHANGELOG and AUTHORS entries
* PEP8ify parametrized-test parameters
* Default color is now yellow
* If there are warnings, make the status bar yellow
* Add missing "error" status to the list
* Refactor slightly
* Add tests for the empty-key special case
* test\_summary\_stats() now prints its parameter values
* Factor out build\_summary\_stats\_line(), and add tests
* Fix plugins\_index script and update plugins\_index.txt
* start 2.7.3 CHANGELOG
* Test that deprecated\_call keeps internal warnings structures intact
* Update AUTHORS and CHANGELOG
* Correct test\_deprecated\_call\_preserves test
* deprecated\_call dit not revert functions back to original
* Reintroduce get\_plugin\_manager() for backward-compatibility
* try to fix coveralls env
* check webhook again
* small fix to test webhook
* finish pytest-2.7.2 release
* Return an empty list directly
* Check conftest.py is not loaded with --confcutdir
* Add changelog entry
* Add a --noconftest option
* Fixed typo
* bump version to 2.7, depend on py-1.4.29
* change spelling, thanks Maciej Maciaszek
* Use SVGs for all README badges
* Add coveragerc file for coverage options
* Added workaround to still show the diff on containers with unsortable elements
* Fix issue where pytest.raises() doesn't always return Exception instance in py26
* Adjust IRC notifications to use #pytest
* unittest.SkipTest now reports original function location
* Skipif marker report now refers to the skipped function
* Use os.path.isdir instead of os.path.exists
* Allow failure of py27-subprocess tox-env while on pytest-2.7
* split travis jobs
* use travis containers
* Remove duplicated author from AUTHORS
* Add docs for new tests
* Update CHANGELOG and AUTHORS
* Test creating directory for junit-xml and resultlog
* Automatically create directory for results
* backport fix for #713
* Sort AUTHORS names
* Mention issue 713 in the test's docstring
* Handle reports that don't have a reprcrash
* Remove remnants from pytest
* Sync with pytest/master
* Use shields.io badge for coveralls
* Use pypi version badge from shields.io
* appveyor badge
* appveyor scripts
* appveyor integration
* pass coveralls env
* pass coveralls env
* add travis badge
* Remove hgignore and hgtags files
* split travis jobs
* use travis containers
* Update sidebar links from Bitbucket to GitHub
* Update CONTRIBUTING documentation with GitHub instructions
* Change links from bitbucket to GitHub
* Merged in RonnyPfannschmidt/pytest/regendoc-upgrade (pull request #303)
* use regendoc normalization and regenerate docs
* make regendoc a direct requirement
* sort AUTHORS and add Janne
* add Janne\`s changes to CHANGELOG
* Merged in jpvanhal/pytest/include-setup-teardown-duration-in-junitxml (pull request #287)
* backport fixed issue735
* Merged in issue735 (pull request #302)
* Close branch issue735
* use NameConstant node when it exists (fixes #735)
* port typo fix
* fix typo
* fix issue741: make running output from testdir.run copy/pasteable
* Merged in nicoddemus/pytest/issue-741-pytester-output (pull request #293)
* add changelog for latest merge
* Merged in The-Compiler/pytest (pull request #298)
* Close branch pytest-2.7
* Merged in flub/pytest (pull request #299)
* Use platform.python\_version() to show Python version number
* Merged in flub/pytest (pull request #297)
* add in missing finalizer to doc example
* Ignore emacs' .dir-locals.el file
* Fix monkeypatch.setenv with string and raising=False
* Add how to upload a wheel and tag the release
* Added tag 2.7.1 for changeset edc1d080bab5
* Regenerate docs using 2.7.1
* Fix rest formatting and set final version number
* Remove unused link
* More updates/fixes to release process
* Set 2.7.1 version number
* remove adopt pytest note
* remove adopt message
* Make "running" output from testdir.run copy/pastable
* another try at ignoring cx\_freeze during other tox envs
* try to ignore cx\_freeze from regular pytest runs and see if travis is happy
* make default depend on just published py/pluggy releases
* Merged in hpk42/pytest-patches/plug30 (pull request #291)
* use new pluggy api (now at 0.3.0) for adding hookcall monitoring and reraise real keyboard interrupts during inline pytest runs to allow for better stopping of the pytest tests
* Merged in hpk42/pytest-patches/pluggy1 (pull request #290)
* - make sure sub pytest runs use the same basetemp - depend on pluggy < 0.3
* adapt to pluggy naming, rename pytest.hookspec\_opts to pytest.hookspec,s ame with hookimpl\_opts
* Merged default into pluggy1
* merge
* - some more adaptation to most recent pluggy API - avoid using pluggin underscore api - show pluggy version in header
* merge default
* Update to pytest revision f78c26828b71
* adapt for current API changes of pluggy
* now that we are going to have wheels, py source code might be not be installed and the resulting genscript is useless
* Merged in famousgarkin/pytest/famousgarkin/monkeypatchchdir-docstring-punctuation-1430453576841 (pull request #288)
* merging pytest-2.7
* Doc typo
* Include setup and teardown in junitxml test durations
* monkeypatch.chdir docstring punctuation
* Use current regendocs location
* Some more tweaks of how to release
* Some small corrections to release process
* Prepare docs for bugfix release
* Fix collapse false to look at unescaped braces only
* adapt pytest to pluggy's decoratorclass branch
* adapt pytest to using pluggy (current master)
* Merged in hpk42/pytest-patches/testrefactor (pull request #284)
* use runpytest() instead of runpytest\_inprocess if a test can run as subprocess as well
* merge
* streamline pytester API majorly:
* - refine lsof checking - make runpytest() create an inline testing process instead of   a subprocess one
* - make API between runpytest() and inline\_run() more similar - shift a number of tests to become inline\_run() tests
* Merged in hpk42/pytest-patches/plugtestfix (pull request #283)
* make test suite more tolerable against xdist causing warnings itself (which it does currently)
* Merged in hpk42/pytest-patches/more\_plugin (pull request #282)
* simplify load\_setuptools\_entrypoints and refine comments/docstrings
* deprecate and warn about \_\_multicall\_\_ usage in hooks, refine docs about hook ordering, make hookwrappers respect tryfirst/trylast
* actually revert back to using older simpler method for subset hook calling.  It is slightly more inefficient but easier to implement and read
* merge contribution fix
* Merged in contributing-fix (pull request #281)
* Close branch contributing-fix
* re-add tryfirst/trylast marker documentation, mark it as to be removed
* move consider\_setuptools\_entrypoints to core pluginmanager
* fix some doc strings
* introduce a new subset\_hook\_caller instead of remove make\_hook\_caller and adapat and refine conftest/global plugin management accordingly
* specialize make\_hook\_caller to work with a subset of the registered plugins
* ensure proper get\_name references
* simplify plugins bookkeeping further, refine API
* simplify addition of method and scanning of plugins
* simplify tracing mechanics by simply going through an indirection
* avoid direct circular reference between config and pluginmanager
* fix issue732: make sure removed plugins remove all hook callers
* make pytest\_plugin\_registered a historic hook
* Streamline data structures
* properly perform hook calls with extra methods
* introduce historic hook spec which will memorize calls to a hook in order to call them on later registered plugins
* split plugin documentation into "using" and "writing plugins", referencing each other.  Also add tryfirst/trylast examples
* add documented hookimpl\_opts and hookspec\_opts decorators so that one doesn't have to use pytest.mark or function-attribute setting anymore
* remove \_do\_register indirection between PluginManager and PytestPluginManager
* also incrementally remove plugins from hook callers
* incrementally update hook call lists instead of regenerating the whole list on each registered plugin
* allow to register plugins with hooks that are only added later
* Merged in hpk42/pytest-patches/plugin\_no\_pytest (pull request #278)
* Remove duplicated step in CONTRIBUTING
* remove useless check
* minimize HookCaller attributes: avoid passing in hookrelay to HookCallers
* remove some redundancy when parsing import spec
* streamline and document handling of builtin module special casing
* merge default
* merge cxfreeze fix
* Merged in nicoddemus/pytest/cx\_freeze\_ubuntu (pull request #280)
* Merged in hpk42/pytest-patches/reintroduce\_pytest\_fixture (pull request #279)
* Fix py27-cxfreeze tox environment
* accomodate Floris' comments. (The reason was i just reinstanted the old code :)
* reintroduced \_pytest fixture of the pytester plugin which is used at least by pytest-xdist
* reshuffle pluginmanager methods and add some docstrings
* shuffle PluginManager method order to first have the public API and then the internal
* remove shutdown logic from PluginManager and add a add\_cleanup() API for the already existing cleanup logic of the config object. This simplifies lifecycle management as we don't keep two layers of shutdown functions and also simplifies the pluginmanager interface
* slight cleanup of plugin register() functionality
* merge conftest management into PytestPluginManager
* avoid undocumented special casing of "pytest\_" prefix
* simplify exception capturing
* move bookkeeping of conftest plugins in core pluginmanager to PytestPluginManager
* avoid prepend to register api as it's redundant wrt to hooks
* minimize HookRelay to become a pure container, refactor initialization and tests of plugin management to be a bit better split between pytest and pytest-independent bits
* remove redundant py check as our setup.py excludes py <=1.4 already
* Merged in hpk42/pytest-patches/nodrone (pull request #277)
* strike drone badge as it doesn't make sense on PYPI (where the README is rendered)
* Merged in hpk42/pytest-patches/prefer\_installed (pull request #275)
* pytester: add method \`\`TmpTestdir.delete\_loaded\_modules()\`\` , and call it from \`\`inline\_run()\`\` to allow temporary modules to be reloaded
* fix regendoc repository location on requirements-docs.txt
* merge fix of issue553 on pytest-2.7
* Merged in getsourcelines-error-issue-553-pytest2.7 (pull request #273)
* Close branch getsourcelines-error-issue-553-pytest2.7
* change test module importing behaviour to append to sys.path instead of prepending.  This better allows to run test modules against installated versions of a package even if the package under test has the same import root.  In this example::
* Merged in hpk42/pytest-patches/systemexit (pull request #274)
* backport Y->y fix from floris
* fixed regression to 2.6.4 which surfaced e.g. in lost stdout capture printing when tests raised SystemExit
* Handle inspect.getsourcelines failures in FixtureLookupError
* Use capital Y as the tests look for that
* Merged in issue660 (pull request #268)
* Close branch issue660
* Add wheel support in the changelog
* Do imports at the head of the file
* Some docstrings for the pytester plugin
* Rename function and add nice docstring
* Make argpase a dependency for python 3.0 and 3.1 too
* Test for setuptools 0.7.2, turns out there's no 0.7 release on pypi
* Improve version test (use pkg\_resources to compare versions). Also log failures to stderr
* Add support for building proper wheels (universal and proper dependency evnironment markers for argparse/colorama if setuptools is new-ish)
* add me as author of PR
* merge pytest-2.7 branch
* shift default to 2.8.0.dev
* add changelog entries
* Merged in release-checklist (pull request #266)
* Close branch release-checklist
* Merged in yield-experimental-docs (pull request #267)
* Close branch yield-experimental-docs
* Brought back discussion session
* Fixed straightforward spelling
* Removed "discussion" session
* Reviewed wording about yield being a "new" feature
* Removed note about yield fixtures being experimental
* more streamlining of release checklist
* - avoid setting of versions and targets in conf.py and Makefile   as discussed on pytest-dev
* fix issue660: properly report fixture scope mismatches independent from fixture argument ordering
* refine release process according to Bruno's feedback
* add a release checklist
* bump versions to dev, including changing the doc-install target to "dev" instead of "latest"
* Added tag 2.7.0 for changeset 7ed701fa2fb5
* bump verrsion to python2.7, fix a too precise test for windows, regen docs
* revert verbosity
* correct config
* add verbosity to the tests
* fix pep257
* merge almarklein/default
* merge pfctdayelise/issue463
* merge with default
* update changelog
* Change docstring style
* Raise specific MarkerError rather than generic ValueError
* duplicate assignment
* merge
* fix issue435: make reload() work when assert rewriting is active. Thanks Daniel Hahler
* Change string format syntax from {} to {0} for py2.6
* Use hasattr instead of try/except
* #463
* address reviewer comments
* Storing sys.last\_traceback: test, docs and changelog
* allow postmortem debugging on failed test
* Mention requirement for licensing information
* add a project someone reported a while ago
* remove default-verbose running
* Merged in parametrized-fixture-override (pull request #257)
* Close branch parametrized-fixture-override
* Merged in blueyed/pytest/strip-docstrings-from-fixtures (pull request #260)
* Merged in link-update (pull request #261)
* Close branch link-update
* update website sidebar links for repo move
* Strip docstrings in output with \`--fixtures\`
* Fix \`reload()\` with modules handled via \`python\_files\`
* update regendocs
* add docs target
* use make develop in contribution guide
* merge with upstream
* fix typo
* Merged in eks/pytest/issue616 (pull request #258)
* make loop more readable
* Add comments
* document fixture override techniques
* support override of the parametrized fixture on the test level
* allow to override parametrized fixtures with non-parametrized ones and vice versa
* Merged in plugincompat-move (pull request #256)
* Close branch plugincompat-move
* moved pytest-plugs to plugincompat
* mention "pytestdotorg" twitter account
* Close branch issue616
* Merged in issue616 (pull request #252)
* a few more fixes
* make "Adopt pytest" note more prominent, fix link from plugins page
* merge dead branch
* close branch
* Merged in update-issues (pull request #255)
* another update for get\_issues.py to account for repository move
* Merged in docs\_community (pull request #254)
* Close branch docs\_community
* fix typos, clarify git/hg
* switch to push to latest docs, because otherwise people get misled to old repos which don't exist anymore
* change links to go to the new pytest-dev bitbucket team's repo location
* added documentationo on the new pytest-dev teams on bitbucket and github.  See https://pytest.org/latest/contributing.html (tentative) Thanks to Anatoly for pushing and initial work on this
* Merged contributing-community into default
* move conftest visibility tests and their setup into a class, accomodates @nicoddemus 's comment
* fix issue650: introduce new --doctest-ignore-import-errors option courtesy of Charles Cloud
* Merged in cpcloud/pytest/ignore-doctest-import-errors (pull request #243)
* fix issue616 - conftest visibility fixes.  This is achieved by refactoring how nodeid's are constructed.  They now are always relative to the "common rootdir" of a test run which is determined by finding a common ancestor of all testrun arguments
* Add test for command line usage
* change "install" targets to use pytest-dev user so that Brianna, Ronny, me and others can push to pytest.org
* Merged in LucasC/pytest (pull request #247)
* Merged in tush/pytest/junit-verbose-failures (pull request #240)
* merge
* merge in the 2.6 maintenance branch
* Merged in msabramo/pytest/remove\_pdbpp\_xfail (pull request #250)
* Remove xfail\_if\_pdbpp\_installed from test\_pdb.py
* Remove pdbpp xfails; don't seem necessary anymore
* Fixing description of --dist=load in the --help documentation for the xdist plugin
* Merged in jeffwidman/pytest/jeffwidman/fix-faq-grammar-and-spelling-1423641159173 (pull request #246)
* Changed back to "parametrized" to match the API
* Fix: FAQ grammar and spelling
* Merged in davehunt/pytest/env-addopts (pull request #241)
* Added documentation for PYTEST\_ADDOPTS environment variable, updated CHANGELOG and AUTHORS
* Merged in pfctdayelise/pytest\_new/adopt-pytest-docs (pull request #244)
* #676
* Add option to ignore import errors in doctests
* Use monkeypatch to set the PYTEST\_ADDOPTS environment variable in the test
* Use shlex to split the arguments from PYTEST\_ADDOPTS
* Support setting configuration using the PYTEST\_ADDOPTS environment variable
* Failure message in junit xml report now are more informative
* Merged in nicoddemus/pytest/pastebin-xdist (pull request #239)
* Attempting to patch terminal only if terminalreporter is available
* Merged in okin/pytest/okin/only-one-import-on-one-line-if-this-shou-1421239108222 (pull request #238)
* Only one import on one line. If this should show good practices, the code should not set a bad example ;)
* fix issue 655: crude workarounds around python2/3 exception leaks
* Cleaning up the docstrings in monkeypatch.py
* Revert pypi badge again as the svg broke
* Merged in techtonik/pytest/techtonik/readmerst-fix-link-to-pypi-1416719991514 (pull request #234)
* README.rst fix link to PyPI
* add a py.test placeholder setup that errors out when trying to get installed (this is registered at pypi.python.org already)
* Mention fix for issue 615 in changelog
* Merged in tomviner/pytest/format\_boolop\_percent6 (pull request #231)
* fix py26 compatibility
* fix for issue615: \_format\_boolop must escape %
* bump version to 2.7 on default
* remove "universal wheel" again until the py26/py27 difference (argparse) does not play a role anymore
* Merged in alex\_gaynor/pytest-1/alex\_gaynor/mark-pytest-as-a-universal-wheel-1414169551518 (pull request #230)
* Mark pytest as a universal wheel
* fix changelog -- we should have all changes in trunk that are in 2.6.4
* Added tag 2.6.4 for changeset f03b6de8325f
* backport pastebin fix
* regen docs
* backport fixed issue620 (doc for genscript)
* Added explanation for the binary blob in runtests.py
* Merged in nicoddemus/pytest/fix-pastebin (pull request #228)
* Fixing --pastebin option by using a POST request instead of a XMLRPC call
* closing branch
* Mention issue351/PR161 in changelog
* Merged in pfctdayelise/pytest/issue351 (pull request #161)
* Merged in nicoddemus/pytest/python-classes-glob (pull request #225)
* added changelog entry about glob-patterns in python\_classes and python\_functions
* checking that option contains glob characters before calling fnmatch
* Fix assertion.rewrite on read-only filesystem
* added support for glob-style patterns to python\_classes and python\_functions config options
* Merged in tomviner/pytest/assert\_percent\_test (pull request #223)
* Sync with py.test 3896:1d2fa202e54c
* add changelog note for tom's PR about parametrized markers
* Merged in tomviner/pytest (pull request #222)
* Merged in tomviner/pytest/some\_spelling\_fixes (pull request #224)
* fix some spelling mistakes
* test for issue615: compound assert with percent
* fix issue552: note about marking single callables
* Adding "auto" to help for "--tb" option
* Adding "auto" to help for "--tb" option
* remove unused import
* remove unused import
* fix changelog
* fix changelog
* Merged in HolgerPeters/pytest (pull request #221)
* Merged in HolgerPeters/pytest (pull request #221)
* link fix: Use restructured text :doc: link instead of html link
* link fix: Use restructured text :doc: link instead of html link
* add changelog entry for new hookwrapper mechanism
* add changelog entry for new hookwrapper mechanism
* some docs and refined semantics for wrappers
* some docs and refined semantics for wrappers
* re-scan methods during plugin register and unregister and not during hook calling anymore.  Simplify register/getplugin api of PluginManager
* re-scan methods during plugin register and unregister and not during hook calling anymore.  Simplify register/getplugin api of PluginManager
* improve docstring, remove unused custom Exception
* improve docstring, remove unused custom Exception
* remove all occurences of "\_\_multicall\_\_" on hook impls in pytest/\*. also simplify pytest\_runtest\_markereport hook in \_pytest/skipping.py while touching the code anyway
* remove all occurences of "\_\_multicall\_\_" on hook impls in pytest/\*. also simplify pytest\_runtest\_markereport hook in \_pytest/skipping.py while touching the code anyway
* Add a doctest for module docstrings
* Add a doctest for module docstrings
* Add documentation for doctest flags and remove dead code
* Add documentation for doctest flags and remove dead code
* Add configuration option for doctest flags
* Add configuration option for doctest flags
* - refactor wrapped call support to also accomodate   pytest.mark.hookwrapper - introduce a CallOutcome class to hold the result/excinfo status of   calling a function. - rename add\_method\_controller to add\_method\_wrapper
* - refactor wrapped call support to also accomodate   pytest.mark.hookwrapper - introduce a CallOutcome class to hold the result/excinfo status of   calling a function. - rename add\_method\_controller to add\_method\_wrapper
* Document the ids keyword for fixture parametrisation
* Functional tests for id function
* merge pytest  default
* merge pytest  default
* fix add\_method\_controller to deal properly in the event of exceptions. add a docstring as well
* fix add\_method\_controller to deal properly in the event of exceptions. add a docstring as well
* open up pytest-2.6 branch in case we want to release a 2.6.4 before 2.7.0 happens
* Merged in bubenkoff/pytest/test\_for\_issue\_604 (pull request #220)
* Merged in bubenkoff/pytest/test\_for\_issue\_604 (pull request #220)
* merge with default
* merge with default
* Escape % character in the assertion message. closes #604
* Escape % character in the assertion message. closes #604
* docs for "pytest\_addhooks" hook.  Thanks Bruno Oliveira
* docs for "pytest\_addhooks" hook.  Thanks Bruno Oliveira
* cleanup core collection of python methods and remove unncessary cache
* cleanup core collection of python methods and remove unncessary cache
* Merged in issue557 (pull request #218)
* Merged in issue557 (pull request #218)
* Close branch issue557
* Close branch issue557
* simplify internal pytester machinery
* simplify internal pytester machinery
* fix issue557: with "-k" we only allow the old style "-" for negation at the beginning of strings and even that is deprecated.  Use "not" instead. This should allow to pick parametrized tests where "-" appeared in the parameter
* fix issue557: with "-k" we only allow the old style "-" for negation at the beginning of strings and even that is deprecated.  Use "not" instead. This should allow to pick parametrized tests where "-" appeared in the parameter
* add some changelog entries
* add some changelog entries
* Merged in nicoddemus/pytest/defer-hook-example (pull request #216)
* Merged in nicoddemus/pytest/defer-hook-example (pull request #216)
* Merged in nicoddemus/pytest (pull request #214)
* Merged in nicoddemus/pytest (pull request #214)
* Added test for ValueError in custom assert message with % sign (issue #604) https://bitbucket.org/hpk42/pytest/issue/604/valueerror-unsupported-format-character-in
* Added test for ValueError in custom assert message with % sign (issue #604) https://bitbucket.org/hpk42/pytest/issue/604/valueerror-unsupported-format-character-in
* Adding docs on how to properly add new hooks and using them in 3rd party plugins
* Adding docs on how to properly add new hooks and using them in 3rd party plugins
* simplify method to record calls
* simplify method to record calls
* factor out a small "wrapping" helper
* factor out a small "wrapping" helper
* Merged in dgilcrease/change-defaults-from-a-tuple-to-a-list-1412285169250 (pull request #1)
* Merged in dgilcrease/change-defaults-from-a-tuple-to-a-list-1412285169250 (pull request #1)
* Close branch dgilcrease/change-defaults-from-a-tuple-to-a-list-1412285169250
* Close branch dgilcrease/change-defaults-from-a-tuple-to-a-list-1412285169250
* change the defaults from a tuple to a list so I can use config.addinivalue\_line("python\_files", "...") to append new options in my own plugins
* change the defaults from a tuple to a list so I can use config.addinivalue\_line("python\_files", "...") to append new options in my own plugins
* change the defaults from a tuple to a list so I can use config.addinivalue\_line("norecursedirs", "...") to append new options in my own plugins
* change the defaults from a tuple to a list so I can use config.addinivalue\_line("norecursedirs", "...") to append new options in my own plugins
* remove overhead for tracing of hook calls and remove some old unused code
* remove overhead for tracing of hook calls and remove some old unused code
* optimize argument slicing when calling plugin hooks
* optimize argument slicing when calling plugin hooks
* simplify \_scan\_plugin implementation and store argnames on HookCaller
* simplify \_scan\_plugin implementation and store argnames on HookCaller
* call scanning of plugins directly, code is shifted from helpconfig.py to core.py
* call scanning of plugins directly, code is shifted from helpconfig.py to core.py
* refine internal management of plugins and conftest files
* refine internal management of plugins and conftest files
* removed outdated japanese docs from source tree
* removed outdated japanese docs from source tree
* added plugins\_index page generation to tox
* added plugins\_index page generation to tox
* Fixed minor typo in plugins.txt
* Fixed minor typo in plugins.txt
* Merged in bubenkoff/pytest/better-diff-on-verbose-2 (pull request #213)
* Merged in bubenkoff/pytest/better-diff-on-verbose-2 (pull request #213)
* Improve assertion failure reporting on iterables, by using ndiff and pprint
* Improve assertion failure reporting on iterables, by using ndiff and pprint
* bump version to 2.6.4.dev
* add release announce
* Added tag 2.6.3 for changeset 2967aa416a4f
* fix doc release version
* regen and fix some docs (tox -e regen) bump versions, depend on already released py-1.4.25
* add changelog: check xfail/skip also with non-python function test items. Thanks Floris Bruynooghe
* rename orgname to pytest-dev
* Let xfail work on non-python Items
* mention Wolfang wrt pytest\_enter\_pdb
* Merged in wosc/pytest (pull request #204)
* Merged in nicoddemus/pytest (pull request #203)
* Merged in flub/pytest (pull request #207)
* Merged in davidszotten/pytest/stop\_leaking\_fds (pull request #206)
* stop leaking file descriptors
* Merged in davidszotten/pytest/dontreadfrominput-encoding (pull request #205)
* add \`encoding\` attr to DontReadFromInput
* Introduce pytest\_enter\_pdb hook
* fix issue575: xunit-xml reporting collection errors as failures
* fix issue584: fix py3 syntax errors for example/special.txt. Thanks Bruno Oliveira
* fix link to pylib
* add badges to the readme
* Merged in conftest-nodeid (pull request #202)
* Close branch conftest-nodeid
* fix conftest related fixture visibility issue: when running with a CWD outside a test package pytest would get fixture discovery wrong. Thanks to Wolfgang Schnerring for figuring out a reproducable example
* fix issue589: fix bad interaction with numpy and others when showing exceptions.  check for precise "maximum recursion depth exceed" exception instead of presuming any RuntimeError is that one (implemented in py dep) Thanks Charles Cloud for analysing the issue
* fix a broken link
* fix issue587: claim only python2.6 compat in README/pypi project page, thanks sscarwell
* Improve pytest.raises examples
* Remove jython from tested python versions
* detalize the plugin development proposal
* Mention PR #194 in the changelog
* Merged in uweschmitt/pytest/default (pull request #194)
* Use py3k compatible .\_\_getattr\_\_() code
* fix issue582: fix setuptools example, thanks Laszlo Papp and Ronny Pfannschmidt
* mention XXX regendoc for release process
* regen docs for 2.6.2
* Added tag 2.6.2 for changeset a4f25c5e6498
* prepare 2.6.2, release announcement, also add HOWTORELEASE.rst
* strike python2.5 from test code cc @flub
* Merged in nocoddemus/pytest/cx\_freeze-support (pull request #189)
* merged in flub/pytest (PR #195)
* fix issue572 - python3 compat of tmpdir example in docs
* remove debugging turd
* improvements to rewrite cache invalidation
* Improve test by also matching expected output
* Merged in flub/pytest (pull request #196)
* Updating plugins\_index
* Fix doctest tox run
* Show user assertion messages and instrospection together
* Added changelog entry for freeze\_includes()
* merging with default
* Improve the docstring further
* Mention why no universal wheel in changelog
* Merged in sontek/pytest/fix\_universal (pull request #198)
* Removed marking pytest as universal, py26 and py27 are not compatible
* plugin organization proposal
* Mention doc fix in changelog
* Explain why this is important
* added smoke test for bug fixed in 3716:dc080608b6d1
* [doc] fix requests monkeypatch example
* Escape newlines in repr for assertion rewriting
* Depend on newer version of py
* standard lib modules no longer required in freeze\_includes() and updated docs
* Merging with default
* Moved freeze\_includes() to genscript
* hopefully final fix for strange infinite recursion bug
* hopefully final fix for strange infinite recursion bug
* actually regen fixture docs with python3.4 instead of python2.7 (doh)
* fix issue561 example adapted to python3
* fix doc config for latex
* Added tag 2.6.1 for changeset a4f9639702ba
* finalize pytest-2.6.1 release, regen docs
* even better fix as replacement for last commit which was wrong
* better fix as replacement for last commit
* fixed strange infinite recursion bug
* fix issue555: just add "errors" attribute to internal Capture stream
* merge PR192, streamline a bit
* updated CHANGELOG and trace error message as requested in review
* Fixed assertionrewrite.\_read\_pyc to handle corrupted pyc files properly
* Mention change in -v output in changelog
* put some imports back to function-level and streamline py2/py3 compat in one place
* Backed out changeset e57017ad86ce -- logging should not be imported
* Simply show the node ID for verbose output
* Minor fixed in test\_capture
* Removing py.std usage from \_pytest
* guarding scripts with \_\_main\_\_ and doc changes
* fixed line endings for test\_doctest.txt
* Docs about cx\_freeze support and minor adjustments
* first implementation and tox environment for cx-freeze support
* reorder talks, more talks from pelme and others
* add flub and pelme talks at EuroPython2014, remove one past training note
* Merged in nicoddemus/pytest/cx\_freeze-docs (pull request #188)
* Improved the text a little
* Adding blurb about using pytest runner and cx\_freeze
* Merged in fix\_initial\_parsing (pull request #186)
* add a comment for why we only consider .py files when removing @
* speedup @ replacement for the massive lists from pytest-bdd :)
* fix issue547 2.6 regression: capsys/capfd now work again when output capturing ("-s") is disabled
* fix issue544 by only removing "@NUM" at the end of a part (parts are separated by "::") and if the part has an .py extension
* fix conftest detection if commandline arguments contain "::" syntax
* add changelog entry and refactor unittest.mock.patch fix a bit
* Merged in ticosax/pytest/mock-unittest-252 (pull request #184)
* address issue170 by merging David Mohr's PR on adding "raises" as an optional argument to pytest.mark.xfail
* Merged in squisher/pytest/xfail-cause (pull request #183)
* unittest.mock from stdlib should come last
* Handle also unittest.mock
* Tiny example update for clarification
* Directly pass multiple parameters with mark.parametrize()
* Update documentation
* isinstance() on exception value instead of comparing types, consolidate tests
* Injection of fixture doesn't work when decorated with unittest.mock.patch
* Add mark.xfail argument raises so that unexpected exceptions show up as test failures
* Added tag 2.6.0 for changeset 88af949b9611
* merge
* some notes about upcoming trainings
* Added tag 2.6.0 for changeset 60725b17a9d1
* fix release announce
* Updating plugins\_index
* Add test case for using different modules
* bump doc version
* mention next open trainings at EP2014 and in Freiburg in November 2014
* regen docs for pytest-2.6
* mention py34 compatibility
* bump version, preliminary announcement
* pytest depends on the freshly released 1.4.22 to function properly
* remove extra parentheses - stylistic code change
* avoid importing old assertion interpretation code by default (fixes #537)
* Merged in wooparadog/pytest (pull request #163)
* Merged in alex\_gaynor/pytest/alex\_gaynor/marked-pytest-as-being-a-universal-wheel-1405267754279 (pull request #179)
* with pypy this tests leaks one FD but it's probably a residual effect of the testing machinery and Pypy's lazy GC
* Marked pytest as being a universal wheel
* Merged in bmcorser/pytest-1/bmcorser/scratching-an-itch-love-pytest-1404943207227 (pull request #178)
* Scratching an itch. Love py.test!
* Updating plugins\_index
* Merged in msabramo/pytest/norecursedirs\_add\_star\_dot\_egg (pull request #177)
* Add \*.egg to default for norecursedirs
* Merged in msabramo/pytest/fix\_norecursedirs\_doc\_typos (pull request #176)
* doc/en/customize.txt: Fix norecursedirs typo
* fix ordering of import line of last commit
* avoid importing "py.test" (an old alias module for "pytest")
* fix issue364: shorten and enhance tracebacks representation by default. The new "--tb=auto" option (default) will only display long tracebacks for the first and last entry.  You can get the old behaviour of printing all entries as long entries with "--tb=long".  Also short entries by default are now printed very similarly to "--tb=native" ones
* Merged in c\_henz/pytest/explicit-ini-filename (pull request #175)
* colors as used on the live web page

1.0.1
-----

* bump version
* Silence dessert.rewrite logger to errors-only
* Cleaner implementation of early handling of the '-c' command line switch
* Implement the "-c" command line switch that allows to explicitly specifiy the config file to load
* Re-add accidentally removed changelog items
* fix flakes issues
* Add test for inter-dependent fixtures
* Updated plugins index
* Merged in zyegfryed/pytest/zyegfryed/fixed-typo-noticed-by-dcramer-httpstwitt-1401780587159 (pull request #172)
* Fixed typo noticed by @dcramer (https://twitter.com/zeeg/status/473676721128886272)
* fix typo, thanks @dcramer
* also add favicon
* incorporate the new pytest logo
* fix test on Python 3.2
* assert reinterpretation: try mangling attributes that look like private class vars (fixes #514)
* Updated plugins\_index and updated python 3 to 3.4
* Add Python 3.4 to test matrix

1.0.0
-----

* Fix travis build
* README
* Initial version
* Makefile
* License and README
* improve example for pytest integration with "python setup.py test" which now has a generic "-a" or "--pytest-args" option where you can pass additional options as a quoted string.  Thanks Trevor Bekolay
* Merged in tbekolay/pytest/tbekolay/better-setuptools-integration-in-goodpra-1400520902689 (pull request #171)
* Better setuptools integration in goodpractices
* Remove unnecessary parts
* Sync with pytest 3612:9a891617d5a2
* Sync with pytest 3612:9a891617d5a2
* Initial version
* fix issue516: tell in getting-started about current dependencies
* fix issue512 : document dependencies in getting-started doc
* Merged in jurko/pytest/setup cleanup (pull request #169)
* - restore compatibility to old getvalueorskip behaviour - introduce a better NOTSET representation to improve docs
* Merged in jurko/pytest/python 3.1 fix (pull request #168)
* use py.builtin.callable instead of raw callable in \_pytest/runner.py
* update the list of officially supported Python versions in setup.py
* stylistic setup.py code cleanup
* Merged in msabramo/pytest/require\_argparse\_for\_py3\_lt\_32 (pull request #166)
* remove the idea that nose support should be moved to a plugin
* fix issue512: show "<notset>" for arguments which might not be set in monkeypatch plugin.  Improves output in documentation
* setup.py: Require argparse for PY3 < 3.2
* Do not list python 2.5 as supported in the docs
* Add test for skipped case with utf-8 message
* Use processced skipreason for generating skip Junit node
* Also replace \`report.longrepr\` with \`bin\_xml\_escape\`ed skipreason
* issue351: Add ability to specify parametrize ids as a callable, to generate custom test ids. + tests, docs
* Cache exception raised in fixtures according to their scope
* Changelog for issue 475
* Improve error message if pytest.raises is used wrongly
* issue504: verbose output displays node IDs for each test
* Merged in rouge8/pytest/issue499 (pull request #157)
* Interal link to node ID explanation
* fix issue499: document selecting tests by node ID
* Update changelog with last change
* Change XPASS colour to be yellow rather then red
* Merged in hpk42/pytest-hpk/nose\_test\_attr (pull request #154)
* fix tests to properly fail on failed collectiosn (which was hiding an error) and also implement \_\_test\_\_=False for test functions properly
* fix wrong merge
* fix typo in changelog
* merge default
* support nose-style \`\`\_\_test\_\_\`\` attribute on modules, classes and functions, including unittest-style Classes.  If set to True, the test will not be collected
* Merged in hpk42/pytest-hpk/issue473 (pull request #152)
* fix issue473: work around mock putting an unbound method into a class dict when double-patching
* Merged in hpk42/pytest-hpk/issue498 (pull request #151)
* fix issue439: clarify that capsys/capfd capture output during test execution, not test setup
* fix issue498: if a fixture finalizer fails, make sure that the fixture is still invalidated
* Merged in jurko/pytest/jurko/tox\_usage (pull request #150)
* make runtox.py not import external modules unless run as a script
* stylistic runtox.py code cleanup
* fix runtox.py failure when 'tox' is not available on the current system path
* bump version to 2.6.0.dev1 because i think we are going for a 2.6.0 release next and not just a 2.5 maintenance one
* fix issue443: fix skip examples to use proper comparison.  Thanks Alex Groenholm
* add some changelog entries
* Merged in jurko/pytest/fix\_capfd\_fixture\_docstring (pull request #149)
* Merged in msabramo/pytest/makepyfile\_utf8 (pull request #134)
* correct a capfd fixture docstring typo
* Merged in jurko/pytest/break\_ExceptionInfo\_reference\_cycles (pull request #144)
* Merged in hpk42/pytest-hpk/conftest-clean (pull request #148)
* Merged in hpk42/pytest-hpk/issue486 (pull request #147)
* cleanup internal conftest handling and avoid the strange None entry in the conftest cache. (There is basically no reason to ask for conftestmodules without specifying a path.)
* fix test on py33, thanks msabramo1
* fix issue486: better reporting and handling of early conftest loading failures
* add a test for robustness of capturing when a test closes FD1/2
* Merged in msabramo/pytest/add\_drone\_io\_to\_README\_rst (pull request #143)
* Mention fix for issue 453 in changelog
* Merged in msabramo/pytest/remove\_unused\_import (pull request #146)
* test\_capture.py: More PEP8. Remove unused \`cap\` in \`with\`
* Escape newlines in result from assertrepr hook
* Style fixes for pep8
* testing/test\_capture.py: Remove unused \`import tempfile\`
* README.rst: Remove drone.io build badge; keep link
* break reference cycles caused by storing local reference to exception info
* fix issue493: don't run tests in doc directory with \`\`python setup.py test\`\` (use tox -e doctesting for that)
* fix issue492: avoid leak in test\_writeorg.  Thanks Marc Abramowitz
* fix issue492: avoid leak in test\_writeorg
* README.rst: Add drone.io link and build badge image
* fix issue496: add pytest-sugar github repo override and regen index pages, upload it
* fix issue436: improved finding of initial conftest files from command line arguments by using the result of parse\_known\_args rather than the previous flaky heuristics.  Thanks Marc Abramowitz for tests and initial fixing approaches in this area
* add changelog entry for some PRs: improvements to pytest's own test-suite leakage detection, courtesy of PRs from Marc Abramowitz
* Merged in msabramo/pytest/test\_writeorg\_close\_tempfile (pull request #136)
* Merged in msabramo/pytest/refactor\_LsofFdLeakChecker (pull request #138)
* Merged in msabramo/pytest/doc\_plugins\_pytest\_sugar (pull request #141)
* Update entry for pytest-sugar; GitHub icon instead of BitBucket, correct description
* doc/en/plugins\_index/index.txt: Update pytest-sugar
* Remove cast of fd to int and sorting
* testing/conftest.py: Reintialize config.\_openfiles for each test
* Improve LsofFdLeakChecker; more reliable and useful leak checking
* testing/conftest.py: Refactor lsof fd leak checking
* merge main
* remove dupped\_stdout logic and related changes, also simplify pytest\_runtest\_\* calls to not use a contextlib with-decorator anymore
* remove non-documented per-conftest capturing option and simplify/refactor all code accordingly.  Also make capturing more robust against tests closing FD1/2 and against pdb.set\_trace() calls
* introduce resume/suspend functionality for FDCapture and SysCapture, fixing problems with early bailouts (from argparse's parse() function e.g.) that wrote to stdout
* avoid some redundancy by using SysCapture from FDCapture for manipulating sys.std{out,in,err}
* rename StdCaptureBase to MultiCapture
* test\_writeorg: Close a tempfile
* Simplify tests; don't use u literal not supported in py32
* Make makepyfile accept UTF-8 so a few cookie tests in test\_assertrewrite.py don't need to be dedented
* fix toxenv spec for py33-pexpect, thanks Daniel Grana
* fix issue483: trial/py33 works now properly.  Thanks Daniel Grana for PR
* Merged in dangra/pytest (pull request #132)
* add py33-trial tox environment
* add changelog entry: fix issue412: messing with stdout/stderr FD-level streams is now captured without crashes
* fix issue412 and other encoding issues.  Streamline dupfile() into a new more thoughtful safe\_text\_dupfile helper
* cleanup and refine issue412 test (still failing on py33)
* merge
* Merged in msabramo/pytest/sys\_meta\_path\_remove\_hook\_only\_if\_present (pull request #133)
* test\_sys\_meta\_path\_munged: Simplify with makepyfile
* Only try to remove hook from sys.meta\_path if it's present
* simplify some capturing tests
* simplify capturing funcarg handling
* simplify reset/stop\_capturing and fix capturing wrt to capturing simple os.write() calls
* unify and normalize Sys/FD Capturing classes \* \* \* more unification
* remove unused "suspend/resume" on capturing, some formatting cleanup
* no need for im\_func and it does not exists in python3
* condition for python3 of the xfailing test for 412
* merge in current default
* xfailing test for issue 412
* Merged in nloadholtes/pytest/nloadholtes/slight-change-to-the-wording-for-readabi-1395890493571 (pull request #131)
* Slight change to the wording for readability
* extend test
* add changelog: merge PR123: improved integration with mock.patch decorator on tests
* Merged in cgilling/pytest (pull request #123)
* Merged in msabramo/pytest/help\_show\_args\_with\_equals\_instead\_of\_space (pull request #130)
* \_pytest/config.py: In --help, show args with \`=\` instead of space
* use sys.modules.get correctly and reference DEFAULT with respect to it
* change try/except to sys.module.get and a conditional
* fix issue472: clarify that \`\`pytest.config.getvalue()\`\` cannot work if it's triggered ahead of command line parsing
* fix issue490: include pytest\_load\_initial\_conftests in documentation and improve docstring
* tentatively fix py33 and py25 compat
* fix issue #479: properly handle nose/unittest(2) SkipTest exceptions during collection/loading of test modules.  Thanks to Marc Schlaich for the complete PR
* Merged in schlamar/pytest (pull request #129)
* Merged in nicoddemus/pytest/integrate-plugin-index-docs (pull request #120)
* Removed unnecessary iteration in nose.pytest\_runtest\_makereport
* Fixed pyflakes errors
* Fixed race condition with SkipTest when module not in sys.modules on collection
* Merged in jurko/pytest/doc\_typo\_correction (pull request #122)
* Merged in graingert/pytest/graingert/pep8-good-practices-code-1394196858258 (pull request #125)
* simplify pdb disabling of capturing, also accomodate the new semantics that capturing is always on during a test session
* - turn on capturing before early conftest loading and make terminal writer   use the original stream
* implement a new hook type: hook wrappers using a "yield" to distinguish between working at the front and at the end of a hook call chain. The idea is to make it easier for a plugin to "wrap" a certain hook call and use context managers, in particular allow a major cleanup of capturing
* remove externally setting and dealing with "item.outerr" from capturing in favor of a direct interface for adding reporting sections to items. \* \* \* refactor makereport implementation to avoid recursion with \_\_multicall\_\_
* remove unneccessary indirections and options
* add changelog for warning system, bump internal version
* shrink and merge the somewhat obscure and undocumented internal hinting system with the new warnings one
* warn if instances are callable and have a test name
* introduce warning system with this API:
* pep8 good practices code
* Merged in jurko/pytest/pdb\_doc\_update (pull request #118)
* Fix to work properly when @patch is used with new not equal to DEFAULT
* fix documentation typo
* Using github and bitbucket icons on plugins\_index page
* Improved text on plugins\_index
* plugins\_index no longer has the package names hard-coded
* reword PDB usage documentation
* make all 'PDB' doc references be recognized as links
* correct inconsistent comment wording
* Added tag 2.5.2 for changeset 421d3b4d150d
* make it clear that this xfail is an unimplemented feature, nothing more
* add release announcement, bump version to 2.5.2, add links to plugins index, regenerate doc examples
* require py>=1.2.20
* refine skipif to use direct booleans, to help with flakes
* refactor lsof checking and fix an lsof leak in pypy
* fix flakes failures
* add changelog entry about issue429, adapt README
* Fix assertrepr for mojibake
* Avoid wasted string concatenation and improve english
* simplify loop which turns direct funcarg parametrization to indirect
* allow positional args to tox invocation
* setupstate.addfinalizer(): fix docstring and remove related unit test not covering functional reality
* Merged in hpk42/pytest-capsimple/capsimple1 (pull request #115)
* remove "mixed" capturing mode which is not used by pytest
* remove "StdCapture\*.call" classmethod because pytest does not use it
* remove now parameter because pytest only used now==False everywhere
* Merged in RonnyPfannschmidt/pytest/multi-usageerror (pull request #113)
* merge from default
* Merged in jurko/pytest/remove\_github\_references (pull request #112)
* remove references to the no longer used github mirror
* Redo the Capture integration propperly
* Contribution guide: added pull request button image
* Contribution guide: removed confusion regarding git
* Contribution guide: added "what is pull request" section
* address issue416: clarify docs as to conftest.py loading semantics
* add Jurko to authors/contributors file
* remove github links
* merge with mainline
* remove the github contribution section
* add Daniel Greensfeld 3-part blog series about pytest
* have travis use the devpi index to get the pylib dependency
* Merged in jurko/pytest/document\_ExceptionInfo\_ref\_cycle (pull request #109)
* speak about "pytest" rather than "py.test". Thanks Jurko
* refine contributing text in several places
* reword note on explicitly clearing local references to pytest.raises() results
* add explicit branch creation note
* rename, refine and link to new contributing doc from several places
* document explicitly clearing local references to pytest.raises() results
* add Piotr Banaszkiewicz
* merge pbanaszkiewicz/contributiondocs
* mark encoding test as xfail also on py2
* backing out Ronny's PR because it was merged too early (still has failing tests)
* stop exposing capsys/capfd.capture
* add notes on the copied pylib version
* capture tests: move imports and declarations to the top
* kill ancient capture devnullpath, os.devnull exists since py 2.4
* small cleanp
* kill the str magic of Encodedfile
* move imports and declarations to the top
* simplify StdCaptureFD snapshot reading
* rewrite all \_pytest.capture uses of py.io to \_pytest.capture
* rewrite all testing uses of py.io to \_pytest.capture
* initial code import for capture transfer
* xfailing test for captire encoding issues with binary stdio
* fix issue413: exceptions with unicode attributes are now printed correctly also on python2 and with pytest-xdist runs. (the fix requires py-1.4.20)
* Contribution guide: moved tox inst. instructions up
* Merged in pelme/pytest/py2\_pkg\_skip (pull request #107)
* fixed issue428: Skip test for packages without \_\_init\_\_.py on Python 2
* add test: '--markers' listing info from plugins in current folder
* remove superflous line
* Addressed contribution guide issues: virtualenv, links, Github mirror, RST rendering
* Moved contribution guide to the rootdir/CONTRIBUTING.txt
* Include py version in the terminal output
* New enthusiastic contribution guide based on Audreyr's cookiecutter-pypackage
* fixed circular imports by reverting a few py.test -> pytest substitions
* fix issue425: mention at end of "py.test -h" that --markers and --fixtures work according to specified test path (or current dir)
* Merged in derdon/pytest/no-p-option (pull request #102)
* fix handling MarkDecorators called with a single positional plus keyword args
* added docs about the \`no:\` syntax for the -p option
* fix doc typo
* fix name
* added changelog: fixed docs and code to use "pytest" instead of "py.test" everywhere.  Thanks Jurko Gspodnetic for the complete PR
* Merged in jurko/pytest (pull request #100)
* update README.rst docs - pytest is compatible with Python 2.5 instead of 2.4
* fix comment typos
* replace py.test module references with pytest
* trim trailing spaces
* plugins\_index: status images now link to actual tox output
* Updated plugins\_index.txt after changes in plugin\_index.py
* Improvements in plugins\_index.txt
* Removing test\_plugins\_index
* remove build status that shows on pypi -- doesn't make sense because it shows the current trunk, not the released version.  And start a new doc/en/status.txt
* Merged in lakka/pytest//minor-doc-fix-in-skippingtxt-also-submi-1387492852421 (pull request #94)
* doc: fix desc for \`parametrize\`
* correct documentation typo
* Fix minor typo in special.txt
* Minor doc fix in skipping.txt.  Also submitted at Github before I realised that this was the master repo.  Will close over there
* Updated error message to be more helpful
* fix issue409 -- better interoperate with cx\_freeze by not trying to import from collections.abc which causes problems for py27/cx\_freeze
* updated plugins\_index for pytest 2.5.1 release
* plugins\_index.py and test run under py33
* Formatted plugins\_index files to use a 80 char width
* Fixed test\_plugins\_index to work on python 2.6
* just use "sans-serif" as the default font, thankfully recommended by hynek
* change to the non-serif version of "Deja Vu"
* don't use guidea as font as it appears to give troubles to windows/chrome users, see also here: http://stackoverflow.com/questions/11487427/is-there-any-font-smoothing-in-google-chrome
* a few minor fixes and removing google+ stuff
* Added tag 2.5.1 for changeset 039d543d1ca0
* fix links and add github link
* use gudea/gudea as header/body fonts
* last change to release announce
* regen docs for 2.5.1, add links for coverage reports
* adding a release announcement and some doc fixes
* pypy has >21K tests now
* fix issue407: fix addoption docstring to point to argparse instead of optparse. Thanks Daniel D. Wright
* remove unused var (fixes flakes tests)
* Fix docstring
* Still print this, but use py2/py3 compat syntax
* amend CHANGELOG with credits and issue mentioning
* fix issue403 : allow same-name parametrized functions within a collector
* fix py32 failures and remove random print from commit accident
* fix issue405 -- xfail the plugin generation test as it is not supposed to run as part of the pytest core tests and only runs on specific environments
* merge
* Allow parameterised fixtures to give paramemter IDs
* Do not use py2-only print, breaks py3 testruns
* doc: Replaced header font
* doc: Moved G+ button
* doc: Fixed small res sidebar font color
* doc: Use native scrollbars on webkit
* merge
* fix py25 mention
* fix issue404 by more strict junitxml escape
* Updated plugins\_index to use pytest 2.5.0
* doc: Removed unused template
* doc: Don't use italic font for internal references
* doc: Use green color for links
* doc: Moved link colors into variables
* doc: Don't show not existing logo
* doc: Use different font combination
* doc: Moved font family attributes into variables
* doc: Adjusted TOC sidebar section
* CHANGELOG: Simplified section headings
* CHANGELOG: Removed trailing whitespace
* doc: Adjusted sidebar link
* doc: Added "Useful Links" section to all sidebars
* doc: Removed overwritten searchbox template
* doc: Moved PDF documentation link into links.html
* doc: Moved searchbox into searchbox.html file
* doc: Moved links section to its own file
* doc: Removed unused themes
* doc: Added proper sidebar to index page
* doc: Activate "Flask" theme
* doc: Imported Flask Sphinx Styles
* .gitignore: Removed \*.html rule
* some doc fixes
* Added tag 2.5.0 for changeset a064ad64d167
* fix typo
* remove py25 from automated testing
* don't declare py25 as supported anymore
* remove unusued import
* fix expicit assert messages for Python2.6: it turns out python2.6 instantiates the AssertionError differently for tuples.  Test and fix to neutralize it
* Updated plugins\_index.txt with latest plugins
* regen docs and bump version to 2.5.0
* preparing 2.5.0 release announcement
* add changelog: fix issue319 - correctly show unicode in assertion errors.  Many thanks to Floris Bruynooghe for the complete PR.  Also means we depend on py>=1.4.19 now
* Merged in paylogic/pytest/override-fixture-via-parametrization (pull request #92)
* close issue240 - rework "good practises" document and discuss discuss the two common test directory layouts in more detail, including an explicit note on how it interacts with PEP420-namespace packages
* Paratrization overrides existing fixtures
* clarify that pytest.mark.parametrize() takes a list of argvalues and not just arbitrary iterators.  Addresses issue122
* address issue122 -- explode "params" into a list in fixture function decorators
* some minor internal cleanup
* fix changelog
* speed up reorder for large higher-than-function-scoped parametrizations
* make bench.py accept an optional script name and add a slow "manyparam" test
* speed up a test
* backout allowing @pytest.fixture in front of pytest\_funcarg\_\_NAME functions. It was introduced because of pylint warnings and it's probably better to go for a pylint-pytest plugin that avoids also other warnings/issues
* Merged in msabramo/pytest/color\_option (pull request #91)
* Remove u'' literal in test\_color\_{yes,no} for Python 3.2 compat
* add changelog entry: PR90: add --color=yes|no|auto option to force terminal coloring mode ("auto" is default).  Thanks Marc Abramowitz
* Merged in msabramo/pytest/color\_option (pull request #90)
* Remove superfluous \`monkeypatch\` arg to test\_color\_yes
* remove unused cache argument for re-ordering items
* fix issue396 -- properly sort tests using class-scoped parametrization
* refactor sorting wrt class-scopes.  This fixes issue396 and also simplifies the internal sorting algorithm a bit
* add a skip benchmark file (from issue400)
* radically simplify eq/neq with nodes by just using Pythons builtin "is" relationship. The need for comparing two separately instantiated nodes seems to be historic (related to an already-gone mode of pytest-xdist which would re-collect nodes) and not actually needed anymore
* refactor internal finalization mechanics such that all fixture arguments in a test invocation will have a corresponding FixtureDef instance. also fixes issue246 (again)
* Assert 'test session starts' in output for test\_color\_{yes,no}
* Add option: --color=(yes/no/auto)
* remove unused line
* refactor and document parametrized sorting code
* reopen #246 -- it turns out parametrized finalization ordering is not fully fixed -- i modified the test and marked it xfail for now
* allow to use pytest.fixture decorator on old-style pytest\_funcarg\_\_NAME definitions
* Merged in paylogic/pytest/parametrize-hashable (pull request #89)
* implement index-based mechanizm for collection of parametrized tests
* fix reporting for @mock'd test functions
* fix importorskip test
* Add test for unicode assertion descriptions
* Respect unicode in AssertionError argument
* fix issue275 - allow usefixtures and autouse fixtures for running doctest text files
* add a note to the documentation that pytest does not mimick nose try to perform "import isolation".  Addresses issue268
* fix issue377 by clarifying in the nose-compat docs that pytest does not duplicate the unittest-API into the "plain" namespace
* python2.4 is not really tested or supported anymore
* apply doc changes from just backed out changeset
* Backed out changeset 73b1eed8ce09
* clarify that python\_functions does not apply to unittest.TestCase classes and their methods.  Addresses issue284
* Ensure the long descriptions and formatting preserve unicode correctly
* Correct comment
* fix issue357 - special case "-k" expressions to allow for filtering with simple strings that are not valid python expressions. Examples: "-k 1.3" matches all tests parametrized with 1.3. "-k None" filters all tests that have "None" in their name and conversely "-k 'not None'". Previously these examples would raise syntax errors
* simplify basedir isolation
* remove an old duplicate marker and use recent pytest mechanism for parametrization
* fix ordering of finalizers of parametrized interdependent fixtures. This fixes issue246 as reported.  Thanks Ralph Schmitt for the precise failure example
* fixed version comparison in pytest.importskip(modname, minverstring)
* remove \_fixturestack attribute now that we have a proper request->subrequest->subrequest chain
* avoid maintaining a fixturestack
* fix py25 compat
* refactor internal FixtureRequest handling to avoid monkeypatching. One of the positive user-facing effects is that the "request" object can now be used in closures
* addresses issue246: add a test for module/function scope that shows that finalizer ordering is wrong
* move two fixture test modules into bigger testing/python/fixture.py
* some amendments/fixes to changelog
* Fixup changelog
* Merge
* Re-raise the first exception instead of the last
* fix issue221 - handle importing of namespace-package with no \_\_init\_\_.py properly. (This is a commit after the fix -- the original issue steps for failure cannot be reproduced anymore)
* removing pexpect from general dependencies because it doesn't install on windows anymore.  Instead to specific configurations for pexpect on py27 and py33 which only call the tests that need it
* add py33-xdist to tox testing
* remove testing of xdist+genscript -- doesn't really make sense because for installing pytest-xdist you need pytest installed which defeats the purpose of genscript
* fix ordering when mock.patch or other standard decorator-wrappings are used with test methods.  This fixues issue346.  Thanks to Ronny Pfannschmidt and Donald Stufft for helping to isolate it
* Ensure all finalizations are run when one fails
* internally make varnames() deal with classes's \_\_init\_\_, although it's not needed by pytest itself atm.  Also fix caching.  Fixes issue376
* don't hide an ImportError when importing a plugin produces one. fixes issue375
* fix issue380 by making --resultlog only rely on longrepr instead of the "reprcrash" attribute which only exists sometimes
* add missing importorskip
* xfail a test on pypy that checks wrong encoding/ascii (pypy does not error out). fixes issue385
* fix issue384 by removing the trial support code
* fix pexpect-3.0 compatibility for pytest's own tests. (fixes issue386)
* fix issues link
* a script to read bitbucket pytest issues
* Merged in hsoft/pytest (pull request #81)
* Added test for previous crash on failed import fix
* Merged in paylogic/pytest/fix-fixturedef-merge (pull request #86)
* correctly check for fixturedef when  merging
* Merged in nicoddemus/pytest/plugins-index-status-images (pull request #85)
* Merged in paylogic/pytest/ignore-autocomplete-on-darwin (pull request #84)
* Merged in paylogic/pytest/multi-level-fixture-deps-override (pull request #83)
* Merged in paylogic/pytest/python32-test-fix (pull request #82)
* Adding status images in plugins\_index
* ignore argcomplete on darwin
* support python32
* use deepest fixturedef for fixture closure
* initial
* Fix TypeError crash on failed imports under py3.3
* merge, add changelog entry
* Merged in jameslan/pytest/multi-marks (pull request #79)
* Support multiple marks for individual parametrized argument set
* add german pytest talk (pyconde2013)
* change metadata
* fix tests on py32/py33
* fix unicode handling with junitxml, fixes issue368
* fix unicode handling with new monkeypatch.setattr(import\_path, value) API.  Thanks Rob Dennis.  Fixes issue371
* fix typo, thank avanderneuth
* fix changelog
* pytestconfig is now session-config as it is the same object during the whole test run.  Fixes issue370
* Merged in nicoddemus/pytest/plugins-index-adjustments (pull request #77)
* Correct hyperlink and target
* adding column for download counts
* using a single column for name and version
* moved test\_plugins\_index.py
* moved plugins\_index to its own directory
* fix flakes issues and make --flakes run part of tox runs
* Merged in nicoddemus/pytest/plugins-index (pull request #76)
* fixed 'PyPI' spelling
* included test for plugins\_index and added command line options
* added plugins\_index documentation and generated plugins\_index file
* adding first version of plugins\_index script
* avoid one surprising case of marker malfunction/confusion::
* add changelog entry: refine pytest / pkg\_resources interactions: The AssertionRewritingHook PEP302 compliant loader now registers itself with setuptools/pkg\_resources properly so that the pkg\_resources.resource\_stream method works properly.  Fixes issue366. Thanks for the investigations and full PR to Jason R. Coombs
* merge
* Fix bytes/string mismatch in test on Python 3
* Implement suggestions by HPK
* Register the AssertionRewritingHook loader with pkg\_resources; fixes #366
* Adding test capturing #366 where an error occurs when package resources are loaded from the test package
* make "--runxfail" turn imperative pytest.xfail calls into no ops (it already did neutralize pytest.mark.xfail markers)
* bump version to 2.3.3.dev1
* add test for detecting coding cookie with CRLF newlines
* fix detection of the coding cookie when it is on the second line of the file (fixes #330)
* fix coding cookie detection logic
* Added tag 2.4.2 for changeset 8d051f89184b
* bump version of docs
* finalize release announcement, changelog
* add pluginmanager.do\_configure(config) as a link to config.do\_configure() for plugin-compatibility add some more plugins to plugin-test.sh
* bump version to 2.4.2, regen docs
* remove fd-fixing attempt at startup of pytest.  It's not clear it's actually needed and it's not nice to still do FD-dupping when "-s" is specified
* fix python2.5 issues
* fix argcomplete-test to use sys.argv[0] if it looks like a py.test executable
* fix issue365 and depend on a newer py versions which uses colorama for coloring instead of its own ctypes hacks
* always dupfile if os.dup is available
* more fixes regarding marking, in particular plugins should use add\_marker/get\_marker now
* fix issue354: avoid tmpdir fixture to create too long filenames especially when parametrization is used
* simplify the implementation of NodeKeywords getting rid of \_\_ descriptors appearing there
* add setup.cfg for building sphinx\_docs
* allow test items to not be associated with a test function
* remove unused imports (using "importchecker" project)
* Added tag 2.4.1 for changeset 8828c924acae
* avoid "IOError: Bad Filedescriptor" on pytest shutdown by not closing the internal dupped stdout (fix is slightly hand-wavy but work)
* reference CHANGELOG
* 2.4.1 release preps
* fix regression reported by dstufft: regression when a 1-tuple ("arg",) is used for specifying parametrization (the values of the parametrization were passed nested in a tuple)
* adapt changelog
* merge
* Merged in dirn/pytest/dirn/adjust-syntax-for-parametrize-documentat-1380671670976 (pull request #72)
* Adjust syntax for parametrize documentation
* add to CHANGELOG
* complete\_dotted: fix for #361, filecompleter on dot files had differing behaviour from bash
* fix typos (thanks @faassen)
* strike wrong yield\_fixture paragraph
* Allow unicode strings in parser.add\_argument()
* bump version to dev again, new CHANGELOG section for 2.4.1.dev
* release announcement, bump version numbers, some test adapatations because of random win32 test failures
* Added tag 2.4.0 for changeset af860de70cc3
* bump version to 2.4.0
* fix usage docs for "-k" (addresses issue357)
* bump version, some windows test fixes, prevent logging from raising exceptions at the end (finally), add py25 to tox.ini
* disallow yield in non-yield-fixtures for now.  This is an incompataibility but we want to prepare for possibly merging fixture and yield\_fixture some day
* strike keyword argument in favour of new pytest.yield\_fixture decorator
* add missing entry to CHANGELOG
* fix issue358 -- introduce new pytest\_load\_initial\_conftests hook and make capturing initialization use it, relying on a new (somewhat internal) parser.parse\_known\_args() method
* some more separation of core pluginmanager from pytest specific functionality. Idea is to have the PluginManager be re-useable from other projects at some point
* shift pytest\_configure/unconfigure/addoption/namespace hook calling to config object
* localize some argcomplete-related functionality
* avoid creation of file in os.getcwd() cc avanderneut
* introduce pluginmanager.ensure\_teardown() which allows
* remove very likely unused pytest\_plugin\_unregister hook (pytest itself and all plugins i know don't use it)
* refine fromdictargs to avoid an uncessary re-setup of the pluginmanager
* simplify Config constructor
* move FILE\_OR\_DIR constant out
* add terminalreporter.section|line methods to print extra information
* is actually a new feature, the syntax: pytest.mark.parametrize("arg1,arg2", ...)
* re-order CHANGELOG and group it into "features" and "bugs" and "known incompatibilities"
* bump version
* merge monkeypatch.replace into monkeypatch.setattr, also support monkeypatch.delattr
* fix some tests wrt to expecting output now that pytest does no introduce unwanted "buffering" on "-s" calls
* remove gittip mention for now
* rework docs to demonstrate and discuss current yield syntax in more depth
* don't manipulate FDs at all if output capturing is turned off
* introduce yieldctx=True in the @pytest.fixture decorator. Refactor tests and docs
* fix issue355: junitxml generates name="pytest" tag
* argcomplete\_win: skip testing of argcomplete on windows
* fix issue333: fix a case of bad unittest/pytest hook interaction
* output errors for all failures of specific collection
* only force tox to upgrade
* Merged in paylogic/pytest/parametrize-fails-when-values-are (pull request #68)
* parametrize fails when values are unhashable - tests
* initial
* make "import pdb ; pdb.set\_trace()" work natively wrt capturing (no "-s" needed anymore), turning \`\`pytest.set\_trace()\`\` into a mere shortcut
* add nose.SkipTest for python2.6 -- which apparently is a subclass from python2.7 on. addresses issue236
* fix <py27 tests with nose
* fix issue352: fix reference to py.path.local description in tmpdir documentation (generally to be found at http://pytest.org/latest/tmpdir.html )
* - fix issue181: --pdb now also works on collect errors.  This was   implemented by a slight internal refactoring and the introduction   of a new hook \`\`pytest\_exception\_interact\`\` hook
* xfail some tests for doctest support if pdbpp is installed
* another way how bash reacts when no argcomplete is there?
* show more info if the test fails
* fix issue169: respect --tb=style with setup/teardown errors as well
* Merged in mcmtroffaes/pytest/fix-nose-docs (pull request #67)
* Fix instructions for nose users
* refs issue290 -- move and refactor the test the string/int-id parametrization test (Which xfails)
* Merged in markon/pytest (pull request #66)
* merge pull request #27: correctly handle nose.SkipTest during collection.  Thanks Antonio Cuni, Ronny Pfannschmidt.  I did a few tweaks to the test and the activation (depending on if unittest is imported at all)
* test marked as
* fix manifest
* Merged in bubenkoff/pytest/fix-broken-tests (pull request #65)
* Fix @parametrize when using an integer and strings as parameters in a test accepting a parameter and a fixture as arguments
* fix broken python3 and python2.5 tests
* initial
* Merged in bubenkoff/pytest/overriden-fixture-finalizer (pull request #64)
* tests for fixture finalizers
* ignores
* strike distribute dep
* add a note about how a lightweight but more powerful function-mocker could be done (compared to standard mock)
* mention pytest\_collection\_modifyitems in plugin page
* remove automatic tox-testing of py25
* monkeypatch.replace() now only accepts a string.  Improved error handling and docs thanks to suggestions from flub, pelme, schmir, ronny
* a new monkeypatch.replace(target, value) call which derives the monkeypatch location from target (can be class/module/function or string which is taken as importable python path) examples:
* Merged in anthon\_van\_der\_neut/pytest\_argcomplete (pull request #63)
* Merged in pelme/pytest/quiet-color-summary (pull request #61)
* argcomplete: FastFileCompleter that doesn't call bash in subprocess, strip prefix dir
* Added color to the quite mode summary. Also changed the output format slightly to match the output of the standard summary
* ref #322 cleanup all teardown calling to only happen when setup succeeded. don't use autouse fixtures for now because it would cause a proliferation and overhead for the execution of every test.  Rather introduce a node.addfinalizer(fin) to attach a finalizer to the respective node and call it from node.setup() functions if the setup phase succeeded (i.e. there is no setup function or it finished successfully)
* fix issue322: tearDownClass is not run if setUpClass failed. Thanks Mathieu Agopian for fixing.  The patch moves handling setUpClass into a new autofixture. (XXX impl-decide if rather adding addfinalizer() API to node's would have a similar effect)
* refs #322: setUpClass and tearDownClass as autouse fixture and finalizer
* merge
* fix issue317: assertion rewriter support for the is\_package method
* mention fix for issue 336 in changelog
* Fix issue 336: autouse fixtures in plugins work again
* add changelog entry for anthon's hynek-fication of options, and change the docs and tests to use the new style
* Merged in anthon\_van\_der\_neut/pytest/opt-drop-non-hyphened-long-options (pull request #57)
* changes to addoption() for hyphenated long-options
* Merged in anthon\_van\_der\_neut/pytest/opt-drop-non-hyphened-long-options (pull request #56)
* removed two superfluous parser arguments
* drop help for long options if longer versions with hyphens are available
* Close issue 279: improved assertrepr\_compare
* fix issue305 - ignore any problems in writing a pyc file, but print out a trace
* fix collection imports for python2.5
* refs #279: sequence assertions can also deal with (Mutable)Sequence instances
* merge
* fix some py33 issues introduced with rev 2985
* fix Mathieu's name
* add changelog entry for fixed issue 335
* merge
* refs #335: clarify that the exception info returned by pytest.raises is a py.code.ExceptionInfo()
* Merged hpk42/pytest into default
* fixes #335: document ExceptionInfo returned by pytest.raises
* small mod to test BND
* plugin versions are displayed now
* no funcargs for setup functions, rather use autouse fixtures
* remove an old issue
* remove an entry
* Merged in anthon\_van\_der\_neut/pytest/argcomplete (pull request #51)
* fix issue334: don't recommend distribute but setuptools everywhere, also remove implicit distribute\_setup support from setup.py
* add license note to README
* fix issue338: honor --tb style for setup/teardown errors as well.  Thanks Maho. (this is just changing the CHANGELOG entry because the bug was already fixed earlier)
* monkeypatch for os.environment changes
* fix for tests running subprocesses of py.test after test\_argcomplete (which all still ran with argcompletion enabled) -> fail
* Merged in anthon\_van\_der\_neut/pytest/argcomplete (pull request #50)
* minor adjustment, added test for positional argument completion
* Fixes for argcomplete
* be more liberal with respect to lsof checks because jenkins keeps some files open
* make genscript provide information as to compatibility (now that argparse is a dependency on python2.6)
* merge doc changes
* Merged in anthon\_van\_der\_neut/pytest/tox\_reference (pull request #47)
* also update testrun.org in 日本語
* updated tox to live on testrun.org (the old links are still working on codespeak.net, but those docs are outdated)
* add changelog: integrate option tab-completion when argcomplete is used.  Thanks Anthon van der Neut for the PR.  This also lets pytest use argparse instead of optparse
* Merged in anthon\_van\_der\_neut/pytest/argparse (pull request #46)
* auto change %default -> %(default)s in help parameter string (on retrieval) added code for warnings on optparse arguments (type, help), which can be easily switched on with TYPE\_WARN = True in config.py
* moving from optparse to argparse. Major difficulty is that argparse does not have Option objects -> added class Argument Needed explicit call of MyOptionParser.format\_epilog as argparse does not have that. The parse\_arg epilog argument wraps the text, which is not the same (could be handled with a special formatter)
* mention github and bitbucket
* stick to virtualenv<1.10 for now because it breaks python2.5
* bump version
* SO-17664702: call fixture finalizers even if the fixture function partially failed (finalizers would not always be called before)
* paint last line red if "failures" or "errors" occured, attribute theuni
* some python2.5/3.3 fixes of Brianna's parametrize improvements
* merge better parametrize error messages, thanks Brianna Laugher
* Merged in pfctdayelise/pytest (pull request #38)
* add python testing training
* fix link to kotti
* send IRC notifications to pytest-dev
* correct travis build status image url
* fixes for py25 in test\_skipping
* change copyright years
* Update README.rst
* fix issue320 - fix class scope for fixtures when mixed with module-level functions.  Thanks Anatloy Bubenkoff
* fix docs wrt norecursedirs, thanks @mgax
* re #320 fallback to test scope if the class-scoped fixture is used in non-class-based test function
* Merged in bubenkoff/pytest/329-skipif-requires-expression-as-a-string (pull request #43)
* re #329 add test for skipif failure when you pass boolean without the reason. add emphasize to the docs
* Always check for both ENOENT and ENOTDIR
* Solve fixture ordering when loading plugins from conftest
* add KAsia to authors
* add kasia to changelog
* merge
* Merged in katarzynaanna/pytest (pull request #42)
* merge
* Merged in bubenkoff/pytest/travis-integration (pull request #41)
* improved reporting
* move to development doc target
* add travis integration, fixes for py25 and py27 no pyc tox env
* Mention issue 300 in changelog
* remove unnecessary print
* merge with upstream
* Merged in ctheune/pytest-greenbar-1/ctheune/typo-1372873724648 (pull request #1)
* Typo
* Compatibility with my spinal cord reflexes: colorize last summary line
* Support working in a local virtualenv
* - add my ep2013 talk to talks page - add "talks/blogs" to the navigation side bar
* fix issue323 - parametrize() of many module-scoped params
* some internal renaming to make more sense of the sorting algo, no semantical changes
* mention added support for setUpModule/tearDownModule detection, thanks Brian Okken
* change how the test is called
* add test\_unittest\_style\_setup\_teardown() to test setUpModule() and tearDownModule()
* support unittest setUpModule/tearDownModule
* added some endorsements, not quite properly layouted
* make sessionfinish hooks execute with the same cwd-context as at session start (helps fix plugin behaviour which write output files with relative path such as pytest-cov)
* Merged in embray/pytest (pull request #37)
* reindent a few of the blockquotes in these tests
* adding ref targets on recwarn
* fix issue316 - properly reference collection hooks in docs
* A couple of improvements to parametrize - When not specifying ids, let None and bools use their native string form (like str, int, float) rather than obfuscated form used for objects - When specifying ids, explicitly raise a ValueError if a different number of ids are specified compared to the test cases - Add tests for both these items
* Adds a test for and fixes #112.  If attempting to write to the \_\_pycache\_\_ directory raises a permission error \_write\_pyc() should just return False to prevent any further write attempts
* allow to specify parametrize inputs as a comma-separated string add Wouter to changelog and to authors
* use \_\_dict\_\_ not func\_dict for Python 3 compatibility
* Merged in w00t0r/pytest-fixes (pull request #35)
* Issue 306: Used a set for the extra\_keywords, and used listchain for parent iteration
* Issue 306: Use the names of all the parents in the chain for matching, except the Instance objects
* Added new test to check on matching markers to full test names, which was possible before. Also adjusted check on number of deselected tests
* Added a test to check there is no matching on magic values
* - add Brianna (@pfctdayelise ) to changelog and contributors - fix some broken tests on py32/py33 (related to issue308 merge) - re-format docstrings -
* Merged in pfctdayelise/pytest (pull request #36)
* Added lost space
* add holger's gittip account, would also like to add ronny's
* Merged hpk42/pytest into default
* issue #308
* Fixed issue #306: Keywords and markers are now matched in a defined way. Also applied some pep8 formatting while fixing
* ? pull/merge
* issue #308
* added changelog for improved doctest counting
* zero to many doctests from module instead of one
* Merged hpk42/pytest into default
* issue #308 first attempt, mark individual parametrize test instances with other marks (like xfail)
* fix issue307 - use yaml.safe\_load instead of yaml.load, thanks Mark Eichin
* add Jaap Broekhuizen for junitxml gen
* fix junitxml generation when test output contains control characters, addressing issue267
* Merged in jaapz/pytest-xml-escape-control-chars (pull request #32)
* Implemented a test for xml control character fail
* mention --tb style change in changelog
* honor --tb style for setup/teardown errors as well
* add maho as contributor
* Merged in maho/pytest (pull request #31)
* #299 - polishing
* Fix pytest.py permissions
* Fix junitxml generation when using special characters in parametrized tests
* bump version
* document context fixtures, also improve plugin docs
* enhance index page, fix announcement index
* support boolean condition expressions in skipif/xfail change documentation to prefer it over string expressions
* don't use indexservers anymore
* Removed tag 1.4.14
* Removed tag 1.4.14
* fix issue245 by depending on py-1.4.14 which fixes py.io.dupfile to not assume file.mode is present
*   implemented as context managers.  Thanks Andreas Pelme,   ladimir Keleshev. fix issue245 by depending on the released py-1.4.14 which fixes py.io.dupfile to work with files with no mode. Thanks Jason R. Coombs
* Added tag 1.4.14 for changeset b93ac0cdae02
* #299
* allow fixture functions to be implemented as context managers:
* bump version
* change version
* Added tag 2.3.5 for changeset fc3a793e87ec
* release 2.3.5 packaging
* fix recursion within import hook and source.decode in particular
* never consider a fixture function for test function collection
* Treat frozenset as a set
* Ingore rope auto-generated files
* Minor style cleanup
* allow re-running of a test item (as exercised by the pytest-rerunfailures plugins) by re-initializing and removing request/funcargs information in runtestprotocol() - which is a slightly odd place to add funcarg-related functionality but it allows all pytest\_runtest\_setup/teardown hooks to properly see a valid request/funcarg content on test items
* Load conftest files in the correct order initially
* pdb plugin: move entering pdb into a toplevel function
* charify pdb visible stack end finding by turning it into a function
* turn the postmortem traceback selection to a function
* add to changelog: put captured stdout/stderr into junitxml output even for passing tests (thanks Adam Goucher)
* merge
* slightly improve -k help string cosmetic change to test\_nose.py
* Merged in adamgoucher/pytest (pull request #29)
* move pdb plugin post morten traceback selection to a own function
* stdout/stderr now captured by junitxml
* fix reference
* doc fix: 'x' should be '-x' so it is not interpreted as a filename
* - add release announce
* Attempt to improve detailed failure reporting
* fix Issue 265 - integrate nose setup/teardown with setupstate
* add a test for issue 14 that will xfail on python < 2.7
* fix issue 271 - dont write junitxml on slave nodes
* fix Issue 274 - dont fail when doctest does not know the example location
* disable assertion rewriting on CPython 2.6.0 because of bugs (fixes #280)
* added changelog entry for getfixture() for doctests
* Merged in witsch/pytest/doctest-fixtures (pull request #25)
* rename \`get\_fixture\` to \`getfixture\` to better match the current API style
* remove debugging left-overs
* update the documentation regarding the \`get\_fixture\` helper
* also provide \`get\_fixture\` helper for module level doctests
* test \`get\_fixture\` helper for doctests
* don't expose the \`FixtureRequest\` object itself in doctests.  in most cases \`get\_fixture\` is sufficient, and you can always call \`get\_fixture('request')\` anyway
* python 2.4 compatibility
* (antocuni, ronny around): import directly from \_pytest.runner to avoid the usage of @property
* correctly handle nose.SkipTest during collection
* Add texinfo build target to doc/\*/Makefile
* in the default Python 2 case, manually check the source is ASCII (fixes #269)
* Use py.builtin.\_basestring
* Mention fix of issue 266 in changelog
* Allow MarkEvaluator expressions to be unicode
* fix issue 251 - report a skip instead of ignoring classes with init
* fix typo in link
* update changelog
* fix issue134 - print the collect errors that prevent running specified test items
* unify logic for error exit on test failures
* fix py32 incompatible syntax
* make dev pytest depend on installing from pypi.testrun.org
* issue250 unicode/str mixes in parametrization names and values now works
* fix --genscript option to generate standalone scripts that also work with python3.3 (importer ordering)
* fix issue257 assertion-triggered compilation of source ending in a comment line doesn't blow up in python2.5 (fixed through py>=1.4.13.dev6)
* merge
* strike python3.1 tox testing, 3.2 and 3.3 is enough
* fix issue 260 - don't use nose specials on plain unittest cases
* small line length fix in nose plugin call optional
* fix parametrized testid to provide for uniqueness
* py3 fixes
* strip old comment and hack
* add note on leipzig course in june 2013
* pass fixture request object (and convenience shortcut to get fixtures) into doctest files
* merge
* add some bits to ISSUES
* ensure OutcomeExceptions like skip/fail have initialized exception attributes
* don't test on py24 for now because tox/virtualenv-1.8 does not support python2.4
* add Brian Okken's blog post as a tutorial
* Add isolation plugin as a feature
* display the repr of some global names (fixes #171)
* adapt locations of ML to new @python.org  location
* fix pylib links
* fix/enhance example
* adding an example on how to do interact with the list of collected tests once before any tests are run
* add sentry
* pocoo no longer has a pastebin service, so this section title is incorrect
* fixed versioning, thanks Arfrever
* fix test after ronny's pytest-debug improvements
* when informations gets truncated, mention use of "-vv" to see it
* improve PYTEST\_DEBUG tracing output
* allow to specify prefixes starting with "\_" when customizing python\_functions test discovery. (thanks Graham Horler)
* Remove check for "\_" prefix on python functions (use python\_functions)
* Fix a broken link to pytest-twisted
* fix bug where using capsys with pytest.set\_trace() in a test function would break when looking at capsys.readouterr()
* improve docstring for metafunc.parametrize()
* reanme README.txt to README.rst
* move long description into README
* bump version, fix -k option help
* Added tag 2.3.4 for changeset ef299e57f242
* fix version number, final fixes
* bump version, add announcement, regen docs
* don't run long-args test on windows because it can't work
* adapt changelog entry about autouse fixtures and yield
* make yielded tests participate in the autouse protocol
* fix autouse invocation (off-by-one error), relates to issue in moinmoin test suite
* modernize tmpdir fixture (use request.node in tmpdir fixture, use @pytest.fixture)
* getting rid of redundant "active" attribute
* fix issue226 - LIFO ordering for fixture-added teardowns
* add a note about yield tests at least in the CHANGELOG
* fix typo (thanks Thomas Waldmann)
* add example for accessing test result information from fixture
* fix issue224 - invocations with >256 char arguments now work
* Fix typos in a document
* switch to pushing docs to dev, amend markers example which needs the dev candidate
* allow to pass expressions to "-k" option, just like with the "-m" option
* allow to dynamically define markers (e.g. during pytest\_collection\_modifyitems)
* add an example for postprocessing a test failure
* fix misleading typo
* - add a Package/dir level setup example - make tox.ini's doc/regen use pytest release instead of dev version
* prolong workaround for jython AST bug http://bugs.jython.org/issue1497 to make pytest work for post-2.5.1 jython versions
* mention that jython-2.5.1 works
* fix typo
* add release announce for 2.3.3
* Added tag 2.3.3 for changeset 7fe44182c434
* fix py31 compat, amend setup.py long description
* bump to 2.3.3, add release announce
* fix issue127 improve pytest\_addoption docs, add new config.getoption(name) method for consistency
* fix issue217 - to support @mock.patch with pytest funcarg-fixtures, also split out python integration tests into python/integration.py and fix nose/mark tests
* amend changelog entries
* fix issue219 - add trove classifiers for py24-py33
* addresses issue209 - avoid error messages from pip on python2.4 related to file, however, never be imported with this interpreter
* add a xfailing test for issue 199
* update changelog
* merge pull request
* test call\_optional not calling non-callable functions
* fix issue209 - depend on pylib dev version which again supports python2.4
* fix issue148 - recognize @unittest.skip on classes, avoid setup/teardown
* fix issue215 - refactor test\_python.py into multiple files:
* remove unused code
* extended - fix issue214 - ignore attribute-access errors with objects in test modules that can blow up (for example flask's request object)
* merge
* fix issue 214 - gracefully handle proxy objects that look like fixtures
* remove issue that doesn't make sense anymore
* fix wrong reference in basic fixture example, thanks for reporting! (closes #212)
* re-allow to parametrize with values that don't support \_\_eq\_\_ (closes issue213)
* i think "helps you write better programs" fits better than "makes"
* fix wrong document version on pytest.org (closes #210)
* remove unused code
* add release announce
* Added tag 2.3.2 for changeset 8738b828dec5
* bump version to 2.3.2, regen docs and changelog
* fix issue208 and fix issue29 - avoid long pauses in traceback printing by using the new getstatementrange() code of the py lib which uses AST-parsing rather than the previous heuristic which had O(n^2) complexity (with n = len(sourcelines))
* fix issue127 - improve pytest\_addoption and related documentation
* improve support for trial a bit more: don't run trial's empty TestCase.runTest()  method
* fix unittest emulation: TestCase.runTest is now ignored if there are test\* methods
* make sure ihook uses a node's fspath - important for hooks e.g. during a Module's collect to pick up conftest.py files residing in the same dir
* fix teardown-ordering for parametrized setups/teardowns
* fix issue206 - unset PYTHONDONTWRITEBYTECODE in assertrewrite test
* - fix test\_nose.py by being more tolerant about the error message   (differs between py32 and py33, thanks Arfrever) - use pypi again now that py is released
* fix issue205 - nested conftest to pickup pycollect\_makemodule - relates to the two reports of a failing doc/en/example/py2py3
* mention twisted with external plugins
* properly handle non-existent PYTHONPATH
* allow to run self-tests with "python setup.py test" for pytest tests itself
* add tox.ini to distribution
* Added tag 2.3.1 for changeset acf0e1477fb1
* some more fixes
* prepare a 2.3.1
* improve --markers output
* fix issue203 - fixture functions with a scope=function should have a "self" that points to the actual instance with which the test functions run
* some doc refinements
* proper version number (2.3.1.dev\*)
* skip pexpect using tests on freebsd
* start new dev cycle
* Added tag 2.3.0 for changeset c27a60097767
* final touches
* fixing links for 2.3 release, and fixing a windows32 failure on py3
* make usefixtures appear in py.test --markers output
* improve automatic id generation for parametrized tests
* some more finalization of docs
* avoid recursing into "ja" japanese examples
* fix trial tests
* refine docs, fix a marker/keywords bit, and add a test that request.keywords points to node.keywords
* remove .markers attribute which was added in development and after 2.2.4 so never released.  Rather extend keywords to also exist on nodes. Assigning to node.keywords will make the value appear on all subchildren's keywords
* many doc improvements and fixes
* fixing the fix of the last commit
* fix issue198 - detection of fixtures from conftest.py files in deeper nested dir structures with certain invocations
* fix and test --fixtures location information
* improve --fixtures output with per-plugin grouping  and hiding underscore names in non-verbose mode, re-introduce --funcargs for compatibiliy
* remove unused code
* simplify/integrate fixturemapper into FixtureManager also fix jstests test failures
* typographic fixes, little simplification
* remove outdated IMPL.txt and move up-to-date doc bits to FixtureMapper class
* refine parsefactories interface, fix two\_classes test originally reported by Alex Okrushko, also add a few more tests to make sure autouse-fixtures are properly distinguished
* strike another use of getfuncargnames() and rename FixtureDef.fixturenames to "argnames" because it's really just the fixture function arguments
* make factorydeflist immutable by using an index
* use FixtureInfo from FixtureRequest
* use fixturemapper/fixtureinfo from Function objects
* implement fixture information stored on the parentnode of functions to be reused by metafunc mechanics and Function setup
* add plan for better fixture implementation, an xfailing test and a slight refactoring of Metafunc tests/creation
* improve docs further, refine unittest docs, rename \`\`autoactive\`\` to \`\`autouse\`\` to better match \`\`@pytest.mark.usefixtures\`\` naming
* nose.py: don't try to call setup if it's not callable
* make tmpdir fixture always return a realpath()ed tmpdir and make a note about it in the changed test.  Currently, i don't see a reason why this is a bad idea (tm)
* fix output of --fixtures for @pytest.fixture defined functions
* try to move docs to a more releasable state, also refine release announce and a few coding bits
* Fix test for windows
* somewhat simplify pytest\_generate\_tests example
* remove support for @pytest.fixture on classes, to be reserved for future use: Fixture-classes could offer setup/teardown/addoption/configure methods and provide higher level support.  Preliminary allowing it to work on classes may make introducing it harder
* remove pytest.setup usage
* - fix doc references, refactor fixtures docs to more quickly start   with examples instead of big text blobgs - also silence -q and -qq reporting some more
* - allow to use fixtures directly, i.e. without () - also allow scope to be determined by a dynamic function
* allow metafunc.parametrize(scope=...) calls to override the scope of a Fixture function definition.  This is useful for cases where you want to dynamically set scope and parametrization for a fixture instead of statically declaring it on the fixture function
* fix issue197 - in case a function is parametrized with zero arguments, skip it during setup
* Use updated names
* implement pytest.mark.usefixtures and ini-file usefixtures setting and also refine fixture docs a bit - fixtures.txt should now mostly reflect the current state of the implementation
* bump version
* make the default non-error pass simpler and faster, refine error reporting by presenting "fixture" tracebacks
* rename a number of internal and externally visible variables to use the fixture name rather than funcargs.  Introduce .funcargnames compatibility attribute for backward compat
* internally unify setup and fixture code, making setup a shortcut to fixture(autoactive=True)
* rename pytest.factory usages into pytest.fixture ones
* merge factories/funcargs and setup functions into the new "fixture" document
* avoid pyc file issues by parametrizing the test instead of rewriting conftest.py files
* add py33 to tox.ini, report pypy-1.9 as working as well
* fixes to against python3.3
* remove unneccessary internal \_\_request\_\_ funcarg
* move Item.applymarker to Node, and defer to it from Funcargrequest.applymarker
* remove print, pass python32
* ensure proper calling of finalizers in case of parametrization on classes
* fixes issue 156: monkeypatch class level descriptors
* back out accidental changes introduced by last patch
* add an xfail-ed test for a potential future "enabled" parameter to setup functions
* avoid double-instantiation of PluginManager in case of the "python pytest.py" or -m pytest invocation
* - make request.funcargnames carry the closure of all used funcargs
* backout, the \_memoizedcall change worked only due to a local effect
* get rid of \_memoizedcall - we dont really need it anymore
* make sure setups are called ahead of the funcarg factories of the test function
* merge
* fix issue 191 - add support for runTest method of unittest.TestCase subclasses
* extend --help to tell about --markers and --funcargs
* don't call nose' setup methods if they are marked with pytest.setup
* improve the parametrization scenario example to sort by id, rather than by file-order, see also: http://stackoverflow.com/questions/12521924/pytest-running-scenarios-in-the-correct-order-in-the-class
* refine internal test support for unicode-related bits (used by a test in pytest-pep8)
* allow factory/setup-markers on classes, using their respective \_\_init\_\_ methods which can use the funcarg mechanism
* fix bug introduced with last checkin
* implement full @pytest.setup function unittest.TestCase interaction
* remove distinction of new versus old funcarg factories
* - add request.node which maps to the collection node as specified by the scope. - remove request.markers which is now available via request.node.markers
* fix @funcarg to @factory
* introduce a new "markers" attribute to nodes and the request object. It is a dynamic class making holdin
* drops special testcontext object in favour of "old" request object, simplifying communication and code for the 2.2-2.3 transition. also modify docs and examples
* merge
* fix issue 188 - ensure sys.exc\_info on py2 is clear before calling into a test
* modify detection of factories located in plugins, allowing pytest's own test functions to access plugin defined funcargs even if they use internal machinery instead of a full test run
* fix issue 182: testdir.inprocess\_run now considers passed plugins
* adapt the junit xml escaping test to my escaping changes
* fix issue185 monkeypatching time.time does not cause pytest to fail
* merge
* use py3 compatible print syntax
* remove usage of exception module, which is gone in py3.3
* add talk from brianna and me from 2012
* fix issue 179 - propperly show the dependency chain of factories on setup failure
* correctly have the test for issue #[C179 actually fail
* add xfailing test for issue 179
* exchange the rawcode factory marker check with a more robust and specific instance check as advised by holger
* fix issue 176: raises(AssertionError) now catches builtin AssertionError as well
* ignore magic callables with no sane code in factory/setup discovery
* fix issue #178 and extend the failure escape test
* Merged in pfctdayelise/pytest (pull request #17)
* expand  list of projects based on URLs from holger
* fix issue172 so that @pytest.setup marked setup\_module/function... functions are not called twice.  Also fix ordering to that broader scoped setup functions are executed first
* fix/update some docs to work with @pytest.factory instead of pytest\_funcarg\_\_ naming
* fix typos
* Fix URL for waskr project
* also improve missing funcarg error for setup functions
* improve error representation for missing factory definitions in recursive funcarg reconstruction
* fix/consolidate --junitxml=path construction with relative pathes
* majorly refine funcargs docs and rename "resources.txt" to "funcargs.txt" so that existing web links will eventually land at this new page when pytest is released.  Also integrated the detailed reasoning and update setup function docs to reflect latest discussions and feedback gathered on py-dev mailing list
* move pytest.mark.factory/setup to pytest.factory/setup, as per flub 's suggestion
* refine documentation, move setup to own "setup" page and provide some more examples. move old setup\_module/... to xunit\_old page
* reshuffle docs, try to get a bit closer to release-relevant documentation
* - rename @funcarg to @factory - introduce a "testcontext" object for new-style funcargs and setup methods - New-style funcargs and setup methods cannot use the "request" object anymore
* reorder internal layout so that funcarg-related functionality is in python.py
* mark a test as xfailing on python2.5
* - enhance ordering of tests using parametrized resources - introduce a refined way to perform finalization for setup functions   which does not use cached\_setup() anymore
* introduce a funcargcall object, holding meta information
* introduce a SetupCall, holding meta information and setup calling state
* minimize active parametrized non-function scoped resources  by - re-ordering at collection time - modifying setup/teardown
* introduce @pytest.mark.setup decorated function, extend newexamples.txt and draft a V4 resources API doc
* allow funcarg factories to receive funcargs
* clarify and add to sort-by-session-scoped parametrized resources example
* implement a scope/parametrized examples using the so-far new features also fix a bug with scoping/parametrization
* allow registration of "funcarg" marked factories
* implement funcarg factory scope marker and ScopeMismatch detection
* extend Metafunc and write a pytest\_generate\_tests hook on the funcarg manager which discovers factories
* discover funcarg factories independently from request/Function items
* move funcarg factory to a new FuncargManager object at session level
* re-introduce the old 2.2.4 FuncargRequest implementation as it is a better base for implementing the new funcarg/setup api. Also Un-optimize funcargnames discovery for now
* V3 draft of resource api
* ci
* DOC: typo in doc/en/goodpractices.txt ("pytest" -> "PyTest")
* v2 of resources API draft
* V1 of the resources API draft
* put automatic funcarg\_ API to Py\*objects only, refine internal subclassing and initialisation logic
* add a little example on how to group test execution by parametrized resource
* Fix typo in terminal help text
* cleanup test a bit
* Fix extension of of cached re-written file
* merge
* exit with errno instead of always signalling success, thanks John Anderson
* merge
* fix issue165 - fix broken links in documentation, also point to stackoverflow from FAQ and contact page
* test and implement showing verbose assert repr for py.test -vv
* fix issue151 - heuristcally lookup conftest files on all command line arguments, not just the first existing dir/file
* remove unused code
* fix issue139 - make it possible to access funcargs from pytest\_runtest\_setup
* better name for the oejskit-compatibility-class
* mid-scale refactoring to make request API available directly on items
* fix problem with unicode in writing failure representations to terminal, thanks ThomasWaldmann
* fix skip/xfail confusion, reported and discussed on http://stackoverflow.com/questions/11105828/in-py-test-when-i-explicitly-skip-a-test-that-is-marked-as-xfail-how-can-i-get
* ignore .cache
* fix internal test setup failure
* fix some pep8 issues, more to go ... is there a tool that helps with pep8-ifying?
* some refinements to reporting and hook order
* add header info: always report 3rd party plugins in test runs
* fix issue160 a failing setup of an xfail-marked tests should be reported as xfail (not xpass)
* fix faq once more to get rid of the strange "missed" bit
* (correction-commit for wrong previous changelog message) fix issue159 -- improve http://pytest.org/latest/faq.html especially with respect to the "magic" history, also mention pytest-django, trial and unittest integration
* fix issue129 - improve http://pytest.org/latest/faq.html especially with respect to the "magic" history, also mention pytest-django, trial and unittest integration
* change pluginmanager.register API to raise ValueError if the plugin object or the name is already registered
* mention pep302 in docstring
* fix comment handling
* fix ReST errors, increment doc-version and push to pytest.org
* Don't use deprecated API in example
* also set release
* use doc-versions that can increment separately from the pytest version
* added an example on how to do python2/python3 customized test collection
* extend marker section with a platform example
* added Japanese translation for 2.2.4 (79a5b776a6f3)
* fixed to find the CHANGELOG's path
* added "docs/en" directory and moved
* simplify activate\_funcargs
* fix issue128 - show captured output when capsys/capfd are in use
* fix py2py3 example tests
* remove pyc file
* draft example for skipping py2 and py3 only tests on a per-module level
* update examples with 2.2.4 version, ReST fixes
* Added tag 2.2.4 for changeset ad9fe504a371
* upgrade inlined distribute\_setup.py
* fix issue 144 - wrong classname in junitxml
* skip test for pyo on pypy since pypy doesnt do pyo
* minor fix to docs/usage
* fix help string for --paste
* use higher difference on timing
* fix issue143 - call unconfigure/sessionfinish always when configure/sessionstart where called
* fix wrong release version
* add release announcement
* require py-1.4.8
* bump version to next release
* hande the trial todo class by using repr
* bump version
* switch pastebin to bpaste.net, fixes #141
* add the fix for issue 140 to CHANGELOG
* strip bound wrappers of class setup/tardown, fixes #140
* prepend the assertion rewriting hook, so as not to break when builtin import is explicitly on sys.meta\_path
* escape the % operator in string formatting
* use non-hacky dynamic package import method
* don't use octal syntax, since its not py2/py3 compatible
* fix a import strange loop that affects pypy test appsupport on python2.5
* write down some thoughts on parametrization - still not completely clear what the next stage should look like ..
* merge
* word change in the docs fixes #130
* try to better handle @unittest.expectedFailure decorator
* remove unused import
* bump version number to dev version
* walk through issues
* add acomment so highlighting won\'t fail
* document integration with setuptools/distribute test command and tests\_require
* junitxml: use a exclusive match on the legal ranges of xml for binary escaping, fixes issue 126
* mention 2.2.3 in 2.2.2 release announce
* Added tag 2.2.3 for changeset 3c11c5c9776f
* release 2.2.3 to fix package contents (2.2.2 contained too many files)
* Added tag 2.2.2 for changeset 92b916483c1e
* regen docs / examples for 2.2.2
* preparing release 2.2.2
* no longer check if indirect metafunc.parametrize params are funcarg names
* readd CHANGELOG, since i accidentially killed it with mq
* more quit collectonly shows only files and the number of tests in them
* create session.items before preparse, so collectonly wont break on preparse fail, fixes #115
* fix issue118 - typo
* add CHANGELOG for last commit
* fix issue177 - actually perform session scope finalization
* kwarg support for reports, so xdist can deserialized extended reports
* fix link to pdf from contents page
* remove nonsennse part of commit related to "mp" shortcut. I wonder if introducing "mp" as a shortcut to monkeypatch is a good idea, actually :)
* bump version, mention "mp" also in the docs and changelog
* update documentation for the new monkeypatch.chdir method
* add monkeypatch.chdir method
* make sure calling undo a second time doesn't change sys.path
* test that a second undo doesn't change sys.path
* only run test\_unittest.py for the twisted/trial config
* fix trial test failure and simplify todo->xfail conversion
* fix unittest/marker integration
* bump version
* work around an apparent python2.4/python2.5 bug with subprocess.Popen, causing jenkins failures.  Apparently "os.environ.popitem(name, None)" is not the same as::
* fix issue106: allow parametrize to be applied per-class/per-module
* internally keep multiple applications of the same markers as separate entities such that the new iter() API can iterate over pytest.mark function attributes, getting all such applications.  See added example for more info
* fix wrongly committed line
* fix Jenkins test failures
* fix typo, thanks jurko
* fix issue102: report more useful errors and hints for when a test directory was renamed and some pyc/\_\_pycache\_\_ remain
* fix spacing
* propogate current PYTHONPATH
* use an exception more consistent across python versions
* fix issue101: wrong args to unittest.TestCase test function now produce better output
* use newer distribute\_setup.py
* fix test to actually mean something useful (thanks Jurko)
* robustify monkeypatch
* add 2.2.1 announce to index
* Added tag 2.2.1 for changeset 3da8cec6c532
* mention pytest-xdist-1.8
* regen docs, prepare release 2.2.1
* remove debugging-changes to tox.ini
* fix unorderable types as reported by Ralf Schmitt
* bump version depend on new pylib
* fix issue99 - internalerror with --resultlog now produce better output. the fix depends on another change in the py lib which unifies the output for native and non-native traceback formatting styles
* Documentation cleanup - corrected typos & minor stylistic changes
* yay! now that we have perfect teardowns we don't need some ugly internal hooks anymore
* fix issue93 - avoid "delayed" teardowns for distributed testing by simplifying handling of teardowns
* take the skip property of unittest cases and functions into account
* use py.xml for generating the junitxml files
* simplify the loop in Node.listchain
* add new "parametrize refinement" issue critical for 2.2 series also cleanup issues a bit and make a note on xUnit/funcargs collaboration
* some updates, add fslayout idea
* fix docstring for setup.py
* improve parametrize() docs
* fix makeinstall
* isolate test example run
* fix typo in alfredo's name
* Added tag 2.2.0 for changeset 152271036933
* small fix to release announcement
* improve release announcement, shift and fix examples a bit. Bump version to 2.2.0
* another try to properly fix durations sorting (still producing sometimes failing tests, apparently when two durations of a test report are identical)
* finally fixing a bug that resulted in sometimes-failing duration tests (doh)
* remove a hack that isn't needed because runtestprotocol now memorizes pending teardowns and thus dist-testing has "exact" teardowns as well
* fix compat with testcases from trial-11.1.0
* fix issue90 - perform teardown after its actual test function/item.  This is implemented by modifying the runtestprotocol to remember "pending" teardowns and call them before the setup of the next item
* add padding to durations in rep.when
* introduce metafunc.parametrize() and @pytest.mark.parametrize with examples. deprecate metafunc.addcall()
* add changelog entry for issue87
* fix issue87 - pastebin option now works with python3
* fix issue89 apply Daniel Nouri's patch to doctest/--pdb interaction
* test and fix pastebin xmlrpc import name missmatch, fixes #87
* fix issue50 (add a reference to the already implemented -m) and improve release annoucnement and changelog
* fix test
* introduce a new -m mark\_expression option
* add ini-file "markers" option and a cmdline option "--markers" to show defined markers.  Add "skipif", "xfail" etc. to the set of builtin markers shown with the --markers option
* add a method to the config object to dynamically add a value to an (line-type) ini-value
* improve mark.txt document and add new regristration/markers features. (welcome to documentation driven development)
* skip pexpect tests on darwin
* fix formatting
* try to avoid timing/race condition
* avoid race condition in test, fix doc link
* simplify durations output, no percentage, no "remaining" bits
* fix duration option in case of collection errors
* fix py3 compat
* separate out the duration tests
* make --durations also show the execution times of setup/teardown calls. This requires a slight incompatibility - pytest\_runtest\_logreport now sees setup/teardown reports even if the tests passed
* introduce --durations=N showing slowest test executions
* skip the symlink test on windows, win32/py32 does not support it without privs
* use os.symlink to make things work on windows/py32
* fix py3 failure
* refine lsof/FD leakage testing and rework test setup and some of pytest own tests.   Note that the actual diff to non-test code is small.  Also remove some redundant tests (introduced by a copy-paste-error apparently in test\_mark.py)
* don't remove symlinks from temporary directory path - should help with some standard OSX setups
* fix FD leakage during pytest's own test run and add "--lsof" option to tox default test runs
* fix some of my typos, thanks Arfrever
* proper tests for issue74, thanks Arfrever
* fix issue74 - propperly filter out missfit names in \_tryconvertpyarg
* fix issue83 - add a link to already generated list of funcargs
* Added tag 2.1.3 for changeset 12a05d59249f
* add 2.1.3 release announce to index, add floris to thanks
* bump to version 2.1.3, add release announce, regen docs (with make regen)
* Mention issue #77 in changelog
* Look up the pytest\_assertrepr\_compare hook for each test item
* cast boolean thing to int to make py3.3 happy
* put the explanation generating code in the conditional fail body (fixes #79)
* add a reportchar for reporting errors, fixes #76
* correctly handle zero length cmdline arguments
* fix issue75 - test failure on jython
* correct base version number
* fix issue67 - junitxml now contains correct durations. thanks ronny
* bump version
* Added tag 2.1.2 for changeset 5864412c6f3c
* add release announcement for 2.1.2
* make call durations part of the test report
* make pip a bit more prominent now that it works on py3
* going for 2.1.2 bug fix release
* try to make test suite pass on jython 2.5.1 again
* quick review of issues
* don't try assertion rewriting on jython for now
* relax error string matching
* rewrite file newlines when the python parser is picky
* merge, bump version
* Capitalised start of headlines, added -ing to a few headlines
* Add FIXME
* Add acceptance test for new --pyargs behavior
* Improve --pyargs
* every boolop operand must have it's own format context (fixes #69)
* clear instead of deleting temporary assertion variables
* only use the last part of the module name in the filename (fixes #68)
* use different caches for optimized and unoptimized code (fixes #66)
* add heading for next version
* fix announcement
* bump to 2.1.1, regen examples, add release announcement
* jython-2.5.2 has a core bug preventing pytest to run :(
* adding issue numbers to the CHANGELOG
* merge
* bump version number, refine goodpractises wrt to importing test modules
* factor out win32 checks
* make test skipping more precise to fix a py32 test failure
* same as 6e94b1809f67: ENOTDIR is ENOENT on windows
* windows kicks up a ENOENT when a part of the path is not a dir
* support pytest.set\_trace in collection
* \_make\_rewritten\_pyc doesn't need to return anything
* only try to create the \_\_pycache\_\_ dir (not a tree to it) fixes #60
* now fix py3..
* use binary mode
* add a newline for window's sake
* escape '%' in specialized comparison explanations (fixes #63)
* add ability to build pytest man page with "cd doc ; make man".  Is to be found in doc/\_build/man/pytest.1 afterwards
* add some debugging tracing to assertion rewriting to understand where failures (specifically issue60) come from
* enhance debug tracing: print trace tags at the end of message and forget about "prefix". Always log to "pytestdebug.log" if "--debug" option is given. also move related code to pytest\_helpconfig plugin
* add a test for vararg call
* fix assertion rewriting on calls with a double-star arg
* merge heads
* respect sys.dont\_write\_bytecode and PYTHONDONTWRITEBYTECODE
* put systemout/systemerr to correct xml location
* fix assertion rewriting in read-only directories (refs #60)
* fix #59: provide better Jenkins stdout and stderr sections
* add changelog note about fixing large boolops
* merge heads
* add benjamin's post to docs, up version, open changelog
* roll test\_long\_chain in with other boolop tests
* Fix a typo in assertion rewriting
* update assertion option names
* add experimental "+1" button
* switching back to pytest.org self-hosting for docs - there is too much issues where i needed/need to poke people :/
* Added tag 2.1.0 for changeset e5e1746a197f
* mention pypi package name
* try use distribute always, update version
* needs pypi.testrun.org for now to pull py lib
* only invoke distribute's use\_setuptools when there is no setuptools installed
* customize pyc tag based on implementation
* on windows, rename is not atomic, so utilize exclusive access to the file
* finalize 2.1.0 version numbering
* adding release announcement
* fix issue 35 - provide download link and improved PDF version
* bump version
* report keyboardintterupt even if inteerrupted during sessionstartup
* add a note about hook partial finalization issues
* make test name shorter, so its testdir path isn't too long on windows
* use py.builtin.exec\_
* merge heads
* simplify rewrite-on-import
* reshuffle start page as per gutworth feedback
* fix a doc link
* releax py requirement to allow readthedocs installing pytest dev
* don't import py
* rearrange and streamline documentation navigation to better work with readthedocs and also with PDF generation
* show release level info for pypy
* fix issue53: nose-style setup now called with the correct ordering
* up pytest version to 2.1.0.dev8, depend on py-1.4.4.dev2
* insure moving pyc files around is atomic
* fix pytest-xdist breakage
* rename and simplify the assert option: cmdline usage is now: --assert=rewrite/reinterp/plain there is no conflict detection (don't think that's neccessary)
* fix python2.5 compatibility
* test files are rewritten in a subprocess
* simplify
* we want second resolution on mtime
* adjust for new option
* don't try to remove pycs twice
* update assert docs
* rename --assertmode choices to be more explicit
* rewrite test modules on import
* use a plain old list for queuing
* rewrite with proper short-circuting on boolean operators (fixes #57)
* fix the rewriter on relative imports (fixes #58)
* refine and streamline Floris example for an assert\_reprcompare hook
* Add pytest\_assertrepr\_compare() docs
* some fixes and clarifications to assert docs
* merge heads
* note condition for introspection happening
* fix getting-started which claimed you need to avoid side effect in asserts
* fix name
* put explanation simplification in format\_explanation so everyone can benefit
* treat local as a black box
* some tweaks to allow pypy apptests to use newinterpret
* fix spacing
* account for quotes in error messages
* fix assertion introspection on python 3.2+
* back out pytest\_configure\_funcargs hook for now
* add issue49 fix to CHANGELOG
* fix issue49 - avoid confusing errors when initialization goes wrong
* fix issue48 - test and fix typo in MarkInfo repr
* update authors to reflect more current situation
* - properly include \_pytest.assertion in distribution - import assertion only at import-test module time
* merge Benjamin's assertion-rewrite branch:  all assertion related code is now part of py.test core distribution - the builtin assertion plugin to be precise. See doc/assert.txt for details on how what has been improved
* fix a buffering issue that i think/hope only occurs during internal tests
* describe how assert rewriting interacts with cross test imports
* return to the old scheme of rewriting test modules from \_importtestmodule
* fix timing float comparison
* add Mozilla QA people to pytest users
* typo
* s/debugging/introspection/
* add another normpath in the junitxml tests
* hopefully final win32 fix for the junitxml path expansion
* bump pylib required
* dont wrap comparisation paths in py.path.local for the junitxml tests, since missing $HOME causes issues else
* also apply normpath to junitxml file path
* fix issue43 - better tracebacks for unexpected exceptions in doctests
* fix issue47 - fix time-per-test timing output for junitxml
* a few more sentences
* update changelog
* use os.path.expanduser/expandvars on the junitxml path for convience, fixes #44
* can use non-underscored addoption
* versionadded and versionchanged for asserts
* write an explicit raise if the assertion fails
* doc updates for new assertion debugging
* fix help for --assertmode
* beef up --assertmode help
* test that tests get rewritten
* remove module before/after import hooks
* new way to rewrite tests: do it all during fs collection
* stuff contents of pytest\_collection hook into perform\_collect
* remove after\_initial\_collect hook
* refactor common config/session protocol code for main() functions
* expose Session on pytest namespace
* add a hook called after the inital fs collection
* add a hook called when a Module is successfully created
* do configure hooks here, too
* call configure hooks in reparseconfig
* cause configure hooks to be called
* a less ugly way to detect if assert rewriting is enabled
* move \_setupstate into session
* correctly initialize and shutdown sessions
* these tests should cause pytest\_configure to be called
* correct attribute name
* add some tracing in the assert plugin
* introduce --assertmode option
* this comment was moved away
* forgot to util module
* fix grammar
* fix comment
* refactor explanation formatting things into their own module
* import assertion code from pylib
* create the \_pytest/assertion package
* expand try/except/finally which py2.4 does't like
* account py3 range objects
* account for py3 dict.values
* use py.builtin.exec\_
* handle comparison results which raise when asked for their truth value
* test that python loads our fake pycs
* refactor writing the fake pyc into its own function
* add a way to disable assertion rewriting for a module
* fix sentence
* give initial imports a reasonable lineno
* improve this test
* in the common case, the and operation isn't needed
* a less silly way to check comparison results
* use assertion rewriting on test files
* unconditionally override lineno and col\_offset on generated ast
* add hooks before and after a module is imported
* remove unneeded list
* correctly handle multiple asserts
* small refactoring
* place assertion imports after \_\_future\_\_ statements and docstrings
* new assertion debugger which rewrites asserts before they are run
* bumping version and adding changelog entry for configure funcargs
* introduce the pytest\_configure\_funcargs hook for better control on funcarg instanciation/configuration
* fix link to pypy tests
* Added tag 2.0.3 for changeset 363e5a5a59c8
* unbump version to retag
* bump version
* regen examples to use 2.0.3 version number
* Added tag 2.0.3 for changeset 49f11dbff725
* add release announcement
* Added tag 2.0.3 for changeset c777dcad1665
* Added tag 2.0.3 for changeset 2ef82d82daac
* bump version
* tentatively use internal list for cleanups at unconfigure time - this helps reporting with partially executed pytest\_configure() hooks
* fix issue38 - nicer tracebacks on sessionstart/configure (and other internal/custom hook failures)
* Update changelog
* Prevent null-characters from appearing in junitxml's output
* Allow unicode characters in testdir.makepyfile()
* shift version string to \_pytest directory
* fix missing reason/name information for skipped tests
* add @classmethod although it's not strictly neccessary for basic cases
* offer a semi-internal method to create a config object in subprocesses (helps pytest-xdist plugin to fix issue34)
* fix issue33 - no collection error for classes prefixed "test" deriving from object
* don't import stuff at genscript import time but rather when it is used
* speed up skipping
* fix issue link
* Added tag 2.0.2 for changeset 84e5c54b7244
* fix install location
* bump to release version, regenerate docs
* simplify \_locationline helper
* fix slightly wrong verbose output for non subclasses on windows
* half the overhead for calling a test function by introducing some caching
* fix issue25 --pdb and win32/python encodings cause a crash in certain situations
* mention that one might need to use the pypi.testrun.org repository
* fix issue27 - --collectonly and -k keyword selection now work together. internally, collectonly and terminal reporting has been unified
* re-introduce pytest.\_fillfuncargs - it's actually used by oejskit, added a test documenting this
* add changelog entry about unittest change, bump version
* unittest plugin: prune \_\_unittest marked modules from traces
* avoid this test on pypy because syntax errors on pypy-1.4.1 are not precise it seems
* actually don't expose unused \_fillfuncargs
* don't expose \_fillfuncargs (no clue why it ever was exposed)
* avoid deprecation warnings for our internal accesses
* improve and clarify skipping docs
* fix and improve error reporting for parametrizing funcargs (originally reported by antlong)
* incorporate typo/grammar fixes from Laura and respond to a number of issues she raised in comments. Also fixed links and some other bits and pieces
* fix issue30 (the second time) put module globals into namespace for xfail and skipif expressions
* fix issue30 - better handling and reporting of errors in xfail expressions
* fix issue 28 - setup\_method now works with pytest\_generate\_tests
* fix help string
* add "issues" to layout of web page
* Fix mistakes in monkeypatch doc example
* doc typo fixes from Victor Garcia, thanks!
* Fix pytest\_assertrepr\_compare on python3 (issue24)
* fix typo, thanks antocuni
* some doc fixes and improvements to parametrized test examples, thanks ccxCZ for review and suggestions
* up version, commit 2.0.1 annoucnement as sent out
* Added tag 2.0.1 for changeset e4497c2aed35
* regen docs with examples
* doc fixes
* accept a left out "()" for ids on command line for better compatibility with pytest.vim
* fix the last committed laxation of a test
* laxer test for also passing it with pypy
* add release announcement
* bumping version to pytest-2.0.1, regen docs and examples
* fix error reporting issue when a "pyc" file has no relating "py"
* fix test related to "not in"
* Improve the "not in" assertion output
* refine and unify initial capturing - now works also if the logging module is already used from an early-loaded conftest.py file (prior to option parsing)
* fix a pypy related regression - re-allow self.NAME style collection tree customization
* fix regression with yield-based tests (hopefully)
* fix typo (thanks derdon)
* fix issue12 - show plugin versions with "--version" and "--traceconfig" and also document how to add extra information to reporting test header
* fix issue10 - numpy arrays should now work better in assertion expressions (or any other objects which have an exception-raising \_\_nonzero\_\_ method ...)
* pypy doesn't neccessarily honour -OO it seems, let's not test assertions there
* use pypi.testrun.org so that py>1.4.0 gets picked up correctly
* fix issue14 : it was actually issue14 instead of issue8 that was fixed with the older https://bitbucket.org/hpk42/pytest/changeset/1c3eb86502b3
* fix issue17 by requiring an update to pylib which helps to fix it
* fix issue15 - tests for python3/nose-1.0 combo work now
* remove same-conftest.py detection - does more harm than good (see mail from Ralf Schmitt on py-dev)
* duplicate word
* small improvements, add assertion improvement to CHANGELOG
* Add "not in" to detailed explanations
* check docstring at test time instead of runtime, improve and test warning on assertion turned off (thanks FND for reporting)
* fix issue11 doc typo (thanks Eduardo)
* rather named the new hook cmdline\_preparse
* make getvalueorskip() be hidden in skip-reporting. also bump version
* introduce a pytest\_cmdline\_processargs hook to modify/add dynamically to command line arguments
* bump version and fix changelog issue reference
* fix issue7 - assert failure inside doctest doesn't prettyprint
* fix hasplugin() method / test failures
* fix issue9 wrong XPass with failing setup/teardown function of xfail marked test
* fix issue8 : avoid errors caused by logging module wanting to close already closed streams
* refine plugin registration, allow new "-p no:NAME" way to prevent/undo plugin registration
* add some docs and new projects
* make pytest test suite pypy ready
* bumping version to a dev version, run tests by using python PyPI by default
* regenerating examples
* need double colon here
* Added tag 2.0.0 for changeset e9e127acd6f0
* fix trove classifier
* last changes, preparing 2.0.0
* some small pre-release updates
* adding anto's projects
* fix some more trial/unittest related bits, particularly allow todo/skip items, now we can run a large fraction of twisted's own test suite, mostly not those that depend on the exact Failure() semantics (e.g. frame objects not being around after gc.collect() but py.test kills them only slightly later anyway)
* tons and tons of refinements and additions to docs
* remove invalid comment
* don't try to load conf.py
* express filter as a listcomp
* simplify pluginlist computation
* introduce new discovery mechanism XXX experiment with using it before introducing it or wait for feature request
* [mq]: doc
* also accept non-pytrace pytest.fail() call in setup/teardown methods
* allow setup\_method/teardown\_method to be mixed into unittest cases, reshuffle tests a bit
* teach trial support code to throw separate errors/failures for setup/call/teardown
* Finish the test
* A test for trial
* fix #6 : allow skip/xfail/pdb with trial by hacking the raw exception info out from trial
* python3 fixes
* some cajoling to get pytest.py to be found when it's not on path
* tw is unused here
* allow setup\_class in unittest test cases
* improve docs
* fix the py version check
* depend on py, not pylib distro
* fix #128 show tracebacks for all failures and errors that haven't beed PDB-debugged
* refine unittest support to also work with twisted trial test cases better by introducing a slightly hackish way to report a failure upstream
* nice-fy error reporting of self-tests
* fix bug on windows
* refine initialization and collection reporting, introduce a progress bar
* add a way to mark hooks as "tryfirst" or "trylast" to influence its position in a hook chain. Use 'tryfirst' for capturing hooks so they can start capturing as early as possible, including when conftests add output in runtest\_setup hooks
* refine tmpdir handling and docs - clear tmpdir specified with --basetemp - remove config.mktmp and config.getbasetemp methods
* merging and refining examples,  also refining skipping documentation
* fix: mark.\* objects are now immutable as long as they are not an attribute on a function, enables usage like this::
* add ability to use scope="class" in request.cached\_setup() calls
* refine release announcement
* better deal with importing conftest.py with --doctest-modules and re-enable default of "--doctest-modules" even if issued at root level
* fix bare "py.test" runs without a directory by not defaulting to --doctest-modules which will virtually import everything
* add some missing files
* refine docs and docstrings, fix some small bits here and there while doing that
* streamline docs, especially use "import pytest" and "pytest.\*" in python code examples instead of "import py" and "py.test.\*"
* bump version number
* fix doctest IDs, also fix tree traversal and remove dead code
* run doctests in .txt/.rst files directly specified on command line irrespective of "test\*.txt" pattern
* bump version and comment out ignore-testclass-if-unittest-module-feature
* some pep8 fixes
* perform represenation of short paths at test execution site
* fix skip reporting over distributed testing.  if we have a "skip" report rep.longrepr will now be a 3-tuple (path, lineno, message)
* some fixes to make cross linux/windows remote testing work again
* revert benjamin's change: script could be py.test.exe so we cannot just return "python,script".  When was the actual problem occuring?
* also un-nest test directory
* flat is better than nested (cont'd): - pytest.py is new module, making "python -m pytest" work always - \_pytest/\*.py now contains core.py, hookspec and the plugins, no sub packages
* internally use pytest.\* instead of \`\`py.test.\*\`\` in many places. make sub namespace names 'collect' and 'cmdline' available on pytest directly
* run subprocess py.test scripts with the python version we're testing on
* add coding for py3
* adapt to simplified tox indexserver definition
* adapt to new tox indexserver syntax
* install dependency from pytest distribution, not prior
* avoid parsing of path objects when pytest.main(path) is called
* bump version
* refine initilization: read config also from a "pytest.ini" file if exists and revert earlier commandline option and group ordering change
* remove duplicate code, normalize relative path names to fix windows running tests
* probably the last major internal cleanup action: rename collection to session which now is the root collection node.  This means that session, collection and config objects have a more defined relationship (previously there was no way to get from a collection node or even from a runtest hook to the session object which was strange)
* bump version
* fix test, bump version
* fix bug showing up on windows
* some python3 related fixes
* PYTHONDONTWRITEBYTECODE might not be set
* replace with list comp
* some fixes for --pyargs situations and the docs, remove wrongly added test
* introduce an option that avoids discovery of classes other than unittest.TestCase in modules importing unittest
* introduce new --testpkg importpath option, add more meat to draft release announcement
* show return values of hooks more explicitely
* test and fix tracing indentation in case of exceptions
* show traces for all hook invocations not just "path/node" based ones
* some more improvements and updates to docs, add release announcements
* majorly refactor collection process - get rid of py.test.collect.Directory alltogether. - introduce direct node.nodeid attribute - remove now superflous attributes on collect and test reports
* add indent facility to tracing
* implement and document new invocation mechanisms, see doc/usage.txt also rename pytest.\_core to pytest.main for convenience
* introduce a minimal tag-based tracer, to be extended if needed, strike pytest\_trace hook
* update issues
* reverse options ordering in Parser class instead of on PluginManager
* add "linelist" type for ini-files
* document and refine py.test.fail helper and strike superflous ExceptionFailure class refine builtin organisation and start a new doc
* some more refinements to docs
* introduce norecursedirs config option, remove recfilter()
* remove pytest\_report\_iteminfo hook, i strongly guess nobody needs or uses it
* remove imperative xfail, this test passes
* massive documentation refinements
* majorly changing the unittest compatibility code, calling TestCase(name)(result)
* allow unregistration by name
* rename addargs to addopts, make adding of opts configurable
* some test fixes and refinements
* slightly simplify collection node init
* fix tests by using less likely existing import names
* simplify session object and rename some more hooks, not exposed/released yet
* streamline some hook docs and option handling, remove cruft bits, fix doc links
* update issues
* bump version, add a bit to changelog
* introduce "-q" option which decreases verbosity and basically leads to a unittest/nosetest-style "." output add it in an ini file like this:
* remove restdoc plugin which now lives as pytest-restdoc on bitbucket, and be easily included in a project now (like PyPy which still needs it)
* fix --help output for ini-options
* allow modules/conftest files specify dotted import paths for loading plugins
* show pytest.\_\_version\_\_ not pylib
* remove feature deprecated prior even to 1.0
* change status
* merge trunk
* bump version
* remove superflous collect\_by\_name, and improve some docs
* get option settings from ini-file.  make getting configuration options from conftest.py only an internal feature
* add and document new parser.addini(name, description) method to describe ini-values. Also document the parser object with its public methods
* document "setup.py test" to use genscript'ed version
* remove old ways to set option defaults, relying on global setup.cfg or tox.ini files now. revamp py.test --help-config
* introduce reading of setup.cfg / ini-style configuration files rename internal config.Error to pytest.UsageError
* make examples less nested for now
* link to PyYAML parser
* a new example for doing non-python extensions
* close branch "trunk"
* changing default branch name to "default" instead of historic "trunk"
* add a genscript target
* add --lsof self-testing option
* fix capturing: properly finalize self-testing of py.test, reducing number of open files during the test run
* re-introduce compatibility attributes on collection nodes to keep compatible with code like::
* allow unittest test functions to work with the "pytestmark" mechanism by refactoring mark/keyword handling and initialization
* unify collection for finding items and for finding initial nodes
* also check for stderr, add changelog entry
* fix issue93 - hide output of code in early-loaded conftest files
* add example for catching exceptions, simplify install doc
* bump version
* fix broken tests / last checkin
* xpass tests don't cause non-zero exit codes
* fix some doc bits
* streamline tox ini
* make safer filenames
* bump version, fix readme
* skip tests that want to invoke py.test without proper installation
* also fix py31 tox.ini entry
* reconfig tox.ini to care use testrun.org for installation of sdist
* fix example good -> bad (thanks frank)
* fix CHANGELOG
* use pyfuncitem name for tmpdir in order to take generative test id into account
* add an example for testing from an app if a test run is ongoing
* turn most funcargrequest attributes into properties and document them
* fix genscript by copying the new implementation from the genscript package
* alias function keywords to funcarg request keywords
* fixing jython specs for testing
* use testrun indexserver, remove hudson artifact handling
* adding a small bench script to see where time is spend in the hook architecture
* some fixes to packaging and urls
* merge install and basic usage into getting-started.txt add an assert1.txt
* major refinements to documentation and examples
* refine and document conftest loading and handling
* remove all deprecated functionality and tests
* select tests by call-id, add and refine documentation around it
* removed unnccessary indirections in the PluginManager, also fixed a bug in \_core.varnames(), which probably considerably speeds up hook calls
* assert py.\_\_version\_\_ to be 2.0 at least to avoid other weird errors
* remove superflous directory
* shift config initialization to own "config" plugin
* merge \_pytest into pytester self-testing plugin
* merge keyword into mark plugin
* remove non-working pylint plugin
* merge config, pluginmanager, main into one file
* move pytest/collect.py to pytest/plugin/session.py - approaching total py.test pluginizations ..
* avoid early import of doctest
* various documentation related refinements
* regen docs, add a specific test\_collectonly.py example
* advance customization docs, enhance docstrings, add more reference object docs
* regen and extend examples a bit with regendoc.py
* rename last test files
* internally switch to pytest.plugin.NAME instead of pytest.plugin.pytest\_NAME
* start documenting hooks, improve hookspec docstrings a bit
* move and consolidate some more plugin docs
* fix format of examples so that ronny's regendoc detects it
* start reorganizing docs, write more docs, shift plugin docs, to proper documentation, use sphinx, remove old docs ... work in progress
* remove some more cruft
* fix dep
* fix tox.ini and dependencies, various fixes all around, tests pass
* move config to \_config
* move session.py and collect.py to a unified pytest\_session.py plugin
* remove pylib things and move things to new pytest namespace
* merge parseopt into config module
* merge conftesthandle into config.py
* fix and test a unbound local in \_diff\_text of the assertion plugin
* two fixes for Jython
* skip attribute tests on <(2,6)
* add to assertion related changelog
* Show final value first when explaining an attribute
* fix changelog
* fix issue126 : introduce py.test.set\_trace() to allow dropping to interactive debugging even when py.test is configured to capture output. If you like you can override pdb.set\_trace by default like this:
* remove unccessary code from pdb plugin
* fix the annoying interaction between "pdb.set\_trace()" and --pdb. The problem is that pdb raises BdbQuit on exit, which is then caught by --pdb, showing an unwanted pdb prompt.  Fix it by making --pdb to ignore BdbQuit
* fix tests to avoid pyc-caching and skip python2.4 which doesn't support "python -m" on packages
* remove unused args
* fix issue123 - new "python -m py.test" invocation
* fix issue124 - make test reporting more resilient against tests changing FD 1
* refine printing of exceptions via the pluginmanager. if there is no pytest\_internalerror() hook acknowledging receival we print the exception to sys.stderr.  This helps to see issues when there are failures in TerminalReporter initialization
* introduce py.builtin.any
* use repr() to print extra / differing values in assertion comparison failures and guard against failures in detail-representations
* remove config.getinitialnodes() method that was only used for testing method after the refactoring
* fix issue 109 - sibling conftest.py files shall not be loaded. also simplify / refine tests a bit
* small simplification and shuffling of python tests, no content change
* fix a problem and make a note about pytest\_nose calling setup/teardown functions
* some fixes after the merge
* merge heads
* to better match the naming of the corresponding AST (and in case we want to add more customizations later) rename pytest\_assert\_binrepr -> pytest\_assertrepr\_compare rename binrepr -> reprcompare
* refactor assert interpretation to invoke a simple callable and let the assertion plugin handle the hook invocation and its multi-results and also pass in an (optional) test config object to the hook. Add and refactor also a few tests
* merge Floris branch and skip interpret-tests on python2.4
* Truncate the text passed to difflib where possible
* fix python3 issues, add py32 environment
* fix python3 bugs
* fix tox.ini invocation
* refine reporting a bit, show only "dots" for distributed testing
* fix a collection bug  where a::b::c could not be resolved properly if there are multiple 'b' nodes
* rework session instantiation and exitstatus handling
* make "tools-on-path" the default and add new random fnmatch-matching method
* implement pytest\_runtest\_logstart(nodeid, location) hook factor out a NodeInfo helper, and streamline terminal printing a bit
* shift reporting info generation away from terminal reporting time, simplify code. also get rid of redundant 'shortrepr' on collect/test reports and rename reportinfo to "location" in some places
* clean up and simplify startup test protocols and objects introduce some new experimental hooks pytest\_runtest\_mainloop to better integrate distributed testing
* shift all python related testing functioanlity to a dedicated
* Fix bug when the right list was longer then the left
* Make pytest\_assert\_binrepr work on python3 too
* Rename pytest\_assert\_compare to pytest\_assert\_binrepr
* Don't load py.test.config inside py.\_code.\_assertionnew
* Move all tests to test\_pytest\_assertion
* Add specialised explanations to the demo
* Add set comparison
* refactor all collection related logic - drop all pickling support (for now) - perform collection completely ahead of test running (no iterativity) - introduce new collection related hooks - shift all keyword-selection code to pytest\_keyword plugin - simplify session object - besides: fix issue88
* Added tag 1.3.4 for changeset 90fffd35373e
* recreated plugin docs
* Added tag 1.3.4 for changeset 79ef63777051
* finalize release announce and changelog
* some small doc fixes
* introduce py.builtin.\_sysex as alias for the special exceptions, fixes #115
* preliminary release announcement
* simplify and fix installation instructions particularly for windows (fixes #111) and bump version to 1.3.4
* Split the tests between the core and plugin
* fixing test for python2.4 (thanks ronny)
* applied ronny's patch, fixes #116
* implement and naively test the native traceback style
* patch from flub to allow callable objects as hook implementations
* Don't import difflib and pprint up-front
* Initial patch as sent to py-dev
* fix issue116 : --doctestmodules also works in the presence of \_\_init\_\_.py files, done by fixing the underlyingly used path.pyimport()
* Added a test and fix for nose compatible setup/teardown functions so that even less errors are ignored
* Added a test and fix for nose compatible setup/teardown functions that are partials, and so errors are not ignored
* Whitespace normalisation inside funcs in test\_pytest\_nose.py
* Whitespace normalisation between funcs in test\_pytest\_nose.py
* Added a test and fix for nose compatible setup/teardown functions that contain a variable
* committed a xfailing test for sibling conftests
* bumping version number to 1.3.4a1
* test and fix for apipkg (also available in apipkg default branch)
* add 1.3.3 release "announcement"
* Added tag 1.3.3 for changeset c59d3fa8681a
* bump version, prepare 1.3.3
* another whitespace-correction commit
* fixes issue113 - assertion represenation issue
* fix bug on writing out objects with terminalwriter on windows
* avoid loading conftest files which are exactly the same content as a previously loaded conftest file
* fix windows32 terminal coloring
* remove trailing whitespace everywhere
* update issues
* fix test on python2.4
* fix terminal dimension detection to work with stdout
* updating some feature descriptions
* Added tag 1.3.2 for changeset 3bff44b188a7
* re-add exclude pattern
* don't run too-long-filename test
* finalizing docs
* ship distribute\_setup.py version 0.6.13
* fix windows homedir detection
* using improved versioing
* bold summary line
* fixing a doc reference, removing development dependency
* progressing towards 1.3.2, adding announcement, regen docs
* reintroduce --junit - i think it is actually useful
* small fix
* split out pytest-xdist related reporting to the plugin
* fix test for python2.7
* adjust tox.ini
* use the new sdistfile option
* skip sdist on hudson
* updating tox.ini to new format
* some minor compatibility issues wrt to the just released python2.7
* fix doc env
* improve testing of docs
* refine and extend custom error reporting particularly for collection-related errors
* enable some more tests
* don't use dist-testing with jython to uncomplicate the testing matter
* remove the --junitxmlprefix feature - it's kind of YAGNI i guess - i introduced it after 1.3.1 but don't need it anymore and thus it's not going to be there for 1.3.2
* refine header message
* update again
* update tox.ini according to tox progress
* another correction
* updated tox.ini
* apply patch from Jakub wrt fixing resultlog/xdist combo
* use the new envbindir subst
* py.test-1.3.1 does not provide py.test proper for jython, only py.test-jython (py.test-1.3.2 will provide py.test even for jython installs)
* actually run only "testing" tests
* add jython env
* make initial conftest finding ignore "--" arguments
* (ARGH) of windows/hudson/multi-config combo produces too long filenames, so use the global temp dir
* use confcutdir
* add a tox.ini file
* refine bestrelpath to return "." for X.bestrelpath(X) and refine its docstring
* install plain py.test script also for jython
* adjust changelog, add fixed issue92
* name CollectOnlyReporter's output attribute "\_tw" for consistency with TerminalReporter
* merge heads
* remove uneeded exec
* refine py.process.cmdexec handling wrt unicode on all python versions
* correct expected message
* a fix from maciej who claims this fixes issues on some systems. passes all tests so i just apply it
* remove tox.ini for now
* refine reporting with --pdb some more
* fix --pdb to not drop interactive on xfailed tests
* adding a link to Floris' new post
* improve comment
* fix assertion interpretation when the operator is \*\*
* defer a number of other compiles to frame.eval (patch from Amaury on trunk/pypy fork, thanks)
* defer compilation to frame.eval (pypy overrides frame.eval and has its own compilation of source code to bytecode)
* Interpret assignments while examining asserts corrects
* use new --junitprefix option for tox reporting
* introduce a new --junitprefix option to influence xml reporting. also internally avoid some redundant code
* fix issue104 properly xml-escape names in junitxml files
* make py.test.raises as-VAR be an ExceptionInfo object but only initialize it after the block is finished
* merge Ronny's changes, add some documentation and changelog entries
* less imports at function level, add a CHANGELOG entry - i guess there are not many win32/python2.4 users anymore these days
* make terminal tests pass on win32/python2.4 and update tox.ini
* support using py.test.raises in context manager style
* capture a concrete idea for easing platform-specific testing
* introduce a new request.applymarker() function and refactor internally to allow for dynamically adding keywords to test items
* fix tox.ini
* some adjustments to make py.test --basetemp=XYZ work where XYZ is a subdir the checkout which contains a conftest.py
* add ignore\_errors to local.remove()
* remove py.\_path.gateway.\* - its weird and unused
* fix py.code.compile to generate unique filenames
* fix python3 failure by making a relative import work
* adding a tox file and a note in changelog
* don't depend on (and don't actually use anymore) testing/\_\_init\_\_.py
* don't print empty lines with junitxml file printing
* add a CHANGELOG entry for ronny's changes
* add a rootdir param to py.path.local.mkdtemp
* use tempdir.mkdtmp instead of mktmp + repeated tries for making tmpdirs
* add kwarg support to py.errpr.checked\_call
* fix pyimport() bug on directories
* fix name of fedora package, thanks thm
* fix issue 57 - make --looponfail work with xpassing tests
* add Meme's pytest-cov plugin to the plugin index page
* Added tag 1.3.1 for changeset 8b8e7c25a13c
* update plugin docs, restructure release announcement a bit
* Added tag 1.3.1 for changeset d5eacf390af7
* update release announcement
* fix for py3 exception printing logic
* fix issue102 by introducing a --maxfailures=NUM option also print an informative line about "stopped/interrupted" test runs near the end
* merge changes
* internal test runs: do inline\_run() without io-capturing as this nested capturing can leave open FDs which breaks larger test runs.  also introduce an internal option "--lsof" for checking the number of file descriptors
* adding xfail example
* py-1.3.1 release prep and version bumping
* when --runxfail is supplied also show tracebacks when running a test that calls py.test.xfail
* \* improve and test --tb=short reporting \* show --tb=short tracebacks for importing test modules
* refine and structure CHANGELOG
* terser reporting header
* fix for python3 - class.\_\_dict\_\_ is now a dict\_proxy which doesn't have setdefault() anymore
* update docs: mention that py.test.xfail is there
* fix issue89 - allow py.test.mark decorators to be used with classes (if you are using >=python2.6) also allow to have multiple markers applied at class level and test and fix a bug with chained skip/xfail decorators: if any of the conditions is true a test will be skipped/xfailed with a explanation which condition evaluated to true
* note that issue99 is also fixed
* fix issue94 make reporting more robust against bogus source code (and internally be more careful when presenting unexpected byte sequences) also make py.code.Source accept a list of lines directly
* unify handling of reportcharacters across resultlog/junitxml plugins
* fix issue91 introduce new py.test.xfail(reason) helper to imperatively mark a test as expected to fail. Can be used from within setup and test functions. This is useful especially for parametrized tests when certain configurations are expected-to-fail.  In this case the declarative approach with the @py.test.mark.xfail cannot be used as it would mark all configurations as xfail
* fix wrong test invocation
* fix and test "-rP" option to show xpass-test ids
* revert 1735 - fix issue95 differently: just shift the offending zlib import (and others) to happen when they are actually needed
* fix issue95 - treat a failing pytest\_genscript import as non-critical, give a hint
* a crucial close() to prevent too-many-open-files
* some special handling of stdin capturing, unification, un-xfail the win32 test
* for now don't test close(0) on windows - it hangs there
* simplify and unify FDCapture API and usage: \* FDCapture now takes care through the 'patchsys' option to   also set sys.stdin/out/err - setfiles/unsetfiles methods removed -   i doubt anybody uses this outside of py.test's own old usage. \* stdin also goes through FDCapture now
* fix issue98 - xdist documentation wrongly told to set pytest\_option\_X where it is only option\_X that is correct
* deal gracefully with invalid file descriptors - don't capture the particular stream
* - try to fix the nightly failures by refining internal capturing mechanism   and adding tests, including a "lsof" test for making sure the number of   open file descriptors does not increase. - also move a py.io related logging test to testing/io
* fix test to account for earlier capfd skipping (on jython)
* fix issue96 - make capturing more resilient against KeyboardInterrupt
* tentative fix to py3's dependency on a filename->module->\_\_loader\_\_ chain for executing inspect.findsource ..
* fix test to work on jython and cpy
* adding three x-failing tests for issue88, issue93 and related issues
* fix test to account for jython python file ending
* merging again - seems i am using mercurial wrong or in some confused way ..
* remove duplicate and done issues
* merge with issue-commits
* bump version to 1.3.1a1 for now
* remove code.new() function and store lines directly into linecache.cache instead. This avoids the need for custom code objects, improving compatibility for jython and pypy-c
* allow to run py.test.cmdline.main() multiple times
* avoid helper functions showing up in py.test tracebacks
* add an issue about py.test.config deprecation
* fix init-check to work also on pypy-c (armin around)
* update ISSUES
* fix some ReST issues
* fix typo
* Added tag 1.3.0 for changeset eafd3c256e87
* fix interpreter compat
* updating docs
* update release announcement
* deprecate --report option in favour of a new shorter and easier to remember -r option: this takes a string argument consisting of any combination of 'xsfX' Those letters basically correspond to the letters you see during terminal reporting
* add python3 classifier
* new --runxfail option to ignore xfail markers on functions
* add unit-tests for xfail and refine xfail handling and reporting
* add a terminalreporter.testid method
* add new parameters: xfail(run=False) will not run expected-to-fail tests xfail(reason=True) will report the specified reason
* some internal fixes regarding the new required hook-finding prefix
* refine and test new hook registration, now it is called "pytest\_addhooks" similar to pytest\_addoption and raises on bogus input
* rename pytest\_ignore\_collect\_path to pytest\_ignore\_collect before release
* update changelog, install info and bum version to 1.3.0 rather than 1.2.2 because of the added features
* update implementation ISSUES, add one for session/refinements/a collection crash
* fixing and adding to CHANGELOG
* add close method to DontReadFromInput so multiprocessing can close it
* introduce new pytest\_pycollect\_makemodule(path, parent) hook for allowing customization of the Module collection object for a matching test module
* introduce a new pytest\_ignore\_collect\_path(path, config) hook - returning a true value will prevent considering the path for collection The hook is called for both files and directory paths
* refine to allow more characters for instance reprs like it was before
* introduce new py.io.saferepr for printing the 'repr' of an object safely and without consuming too much space
* fix a py3k related skip - py.io.TextIO on py3k should probably not allow to write bytes to it
* make py.io.ansi\_print and py.io.get\_terminal\_width() directly available
* expose py.code.\_reinterpret functions so that pypy and internal uses don't need to go through internal implementation imports
* be more robust about bad std stream encodings
* remove the unused return value of fnmatch\_lines
* some py3 encoding fixes
* \* various jython related fixes. \* more care for print-errors including unicode-encoding related errors
* fix some py3 issues, particularly for 3.1.2 which has truncate(0) not change the file position so that a seek(0) is needed in addition
* \* rather expose internal exceptions under py.test.ACTION.Exception   with ACTION being skip, fail, exit, raises. \* move and refine test\_outcome.py tests into runner tests
* \* properly expose and document runtest-protocol related Exceptions   and move all definitions to the runner plugin for now
* strike unused (buggy) keyword param
* refining changelog + draft release announcement
* refine/fix isimportable-logic and ensure that 'tmpdir' has a python-importable name
* (fixes issue83) don't try to import conftest from an invalid package path, refine path.pyimport() logic
* (fixes issue85) correctly write non-ascii test output to junitxml files, refine some internal methods for it
* fix unicode issues (port of pypy/py repo changeset r72526 by Armin)
* factor out session main loop so that distribute testing can make use of it
* make test source syntax valid
* fetch code object compatibly
* add a helper to get a function's code
* merge main
* provide encoding to dupfile() for py3
* this should test and fix the same issue that was committed in the pypy svn-repo as r72534
* another couple of checks on jython, still some problems
* a couple of more mostly jython-related fixes
* update distribute
* fixes for testrun on jython
* fix version numbers
* introduce an experimental approach for allowing dynamic addition of hooks from plugin. Plugins may register new hooks by implementing the new
* use taskkill cmdline for jython/win32 but skip test on jython because it does not return a subprocess PID
* merge in fixes
* robustify a check to not randomly fail
* robustify check
* a couple of more fixes/refinements for getting py.test to run better on jython/win32
* refining the win32 checks some further
* fix check to work when there is no jython
* refine win32 checks to also work on top of jython/win32
* add capturelog generated plugin text
* fix issue86 - point to pytest-xdist plugin for looponfailing feature
* fix issue78 - strike superflous "import sys" that cause unbound local var errors
* add links to new plugins
* actually skip doc tests if pygments is not available
* fix typo
* remove cover and xmlresult plugins as they are not easily installable and lead to frustration (thanks again to prologic for getting back on it)
* fix docs to not point to a downloadable plugin if the plugin is external (thanks to prologic for feedbacking on this confusion)
* note down two issues after having helped prologic - also related to earlier discussions with ronny and others
* Added tag 1.2.1 for changeset c143a8c8840a
* going for the 1.2.1 release
* fix a pdb problem when dropping to a "raises" related failure
* add some issues
* if a funcarg is misspelled/missing, hint at using "--funcargs"
* nice-ify --funcarg reporting, show paths instead of useless python reprs
* fix docstring
* actually look into all non-dot subdirs for conftest.py files - recursive walk would be too heavy for large source trees but first-level subdirs are fine IMO.  Note that prior to py.test 1.0 doing this "look-ahead" was not easily doable because it was hard to avoid global state in conftest.py, this is not true anymore - so i feel ok telling people to cleanup their conftest files if they get problems (you can imagine people doing all kinds of things at global conftest.py module scope, can't you?)
* fix this test
* refined usage and options for "py.cleanup":     py.cleanup     # remove "\*.pyc" and "\*$py.class" (jython) files     py.cleanup -e .swp -e .cache # also remove files with these extensions     py.cleanup -s  # remove "build" and "dist" directory next to setup.py files     py.cleanup -d  # also remove empty directories     py.cleanup -a  # synonym for "-s -d -e 'pip-log.txt'"     py.cleanup -n  # dry run, only show what would be removed
* add a --funcargs option showing available funcargs
* fix a test
* show a short and nice traceback for funcarg lookup errors
* check and load test\*/conftest.py early from anchors - this makes it a bit more convenient to have command line options available from a root directory of a project that does not directly contain a conftest.py
* fix title to not mention distributed programming which is separated out into execnet
* refine setup ordering some more - test and avoid a problem with funcarg setups where the surrounding setup\_module would fail, but the funcarg setup still be called (which might assume that setup\_module has been called so would raise a confusing error)
* again addresses issue78 : we now call teardown also if setup raised a Skipped exception. I also made sure, setup\_module/class will only be called once - before they'd be call again and again if they raise an error or a skip - for each test in their scope
* install pygments for tests
* closes #67 new super-short traceback-printing option: "--tb=line" will print a single line for each failing (python) test indicating its filename, lineno and the failure value
* fix issue78 - now python-level teardown functions are now called even if the setup failed. Important detail: if the setup raises a Skipped exception, teardown will not be called.  This helps to avoid breaking setup\_module/class that performs a skip - it would otherwise internally be considered as a "successful" setup in order to have teardown called later.  I guess it also makes sense to treat Skip specially because it is unlikely a teardown should be called if a Skip was raised on setup
* fail doc generation if pygments is not installed
* upgrade apipkg.py to fix a potential recursive import issue
* fix doc links, bump to dev version
* better default for bogus terminal getdimensions() call, fixes issue63
* Added tag 1.2.0 for changeset 4fc5212f7626
* add report\_header\_info to release announce and remove some ignores/excludes from MANIFEST.in for a less verbose installation experience
* make sure we get an absolute path when writing the genscript file
* add terminal plugin to overview page again
* some finalizing docs bit, regen plugin docs
* turn this into a black-box test
* always directly use basename for tracebacks, independently from code.path fixes issue77 although i guess it was already fixed before
* pushing towards 1.2.0
* refine excludepath handling to treat entries with no path as matching
* fix python2.4 issue
* move rsync reporting out
* refine tests and refine code to deal with new xdist semantics
* fix deprecation warnings
* rename logxml plugin to junitxml
* kill unused code
* fix test\_importall to not stop on skipped plugins and fix the uncovered failure of genscript: standalone.py template is now safely importable
* get rid of the funccollector node, which nice-ifies names of funcarg-generated tests nodes, also test and fix one anomaly wrt to funcarg setups and instance uniqueness
* remove superflous building of a dict, preserve order for nodes that have identical file:lineno
* another wish
* some issue soritng related to the 1.2 series
* reduce usage of the global py.test.config which maybe should die or become less global at some point (along with py.test.ensuretemp)
* move down py/impl/XYZ to py/\_XYZ
* flatten test directory hierarchy and merge smaller into larger files
* remove dist-testing and looponfail code from core.  there remain some (pytest\_runner particularly) tests that test both plain and dist modes which cannot be easily dis-entangled. food for thought
* remove figleaf which now lives "outside"
* fix reqs2 to point to released execnet
* introduce a new pytest\_report\_header(hook) hook to add additional test-run relevant information to the header of a test report
* remove the PickleChannel dependency for looponfail
* adding a second requirements file which uses execnet-1.0.2
* test and fix looponfailing wrt to a bug introduced with the cmdline/session startup cleanup
* refine classname normalization for junit-xml
* fix sessionstart/sessionfinish handling at the slave side, set "session.nodeid" to id of the slave and make sure "final" teardown failures are reported nicely.  fixes issue66
* fix rest syntax error (thanks fijal)
* refine rsyncing and internal dir/transferal handling: don't transfer roots in a popen- no-chdir situation and only use one py.\_pydir everywhere
* fix and test bug: dist-testing now works again without execnet/pylib installed remotely.  fixes issue65
* add to changelog, remove docstring
* (experimental) allow cmdline arguments to deep-point to a test, also remove virtually redundant session.getinitialitems() calls
* fix some "import py" test issues, and prevent "genscript" script from having dist-options
* porting latest apipkg
* avoid dependency on directory ordering
* add potential feature from py-dev discussion
* update issues, version numbers
* relax a test to pass on jython and fix install docs to include genscript standalone usage
* fix python3 issues, add missing plugin docs
* re-arrange "py.test -h" command line option grouping and update some plugin docs
* bumping version number: 1.2.0a1
* disable default inclusion of figleaf plugin because it caused test failures wrt to capturing/logging interaction.  pytest\_figleaf should anyway better become its own externally living plugin
* vastly simplify and cleanup collection initialization by internally introducing a RootCollector. Note that the internal node methods \_fromtrail and \_totrail are shifted to the still internal config.\_rootcol.fromtrail/totrail
* enable doctest plugin by default, add a --doctest-glob option and some documentation, regen plugin docs
* enhance figleaf setup, enabled by default now (requires --figleaf). Generalize internal ability to show "hints" at the end of "-h"
* higher timeout to accomodate slower execution environments
* streamlined plugin loading: order is now setuptools, ENV, commandline and setuptools entry point names are turned to canonical namees ("pytest\_\*")
* slightly refine invocation of py.test: use the py lib that we got invoked with, does away with the need to not-chdir some tests
* remove/refine some doc strings.  create popen-files with absolute paths
* fix standalone script generation on windows, make tests not do a chdir() so that distributed testing discovers the transferred lib
* fix some failures introduced by the last commit, document new "pytestconfig" funcarg
* run py.\* tools through "-c import py ; py.cmdline.py\*" by default and introduce --tools-on-path to force discovery of tools from PATH
* internal: always use scripts found in the environment
* introduce --confcutdir option to early-inhibit lookup of conftest files above a certain directory
* fix xml generation for skipped collections of tests
* fixing invocation
* adding a logxml plugin and a --xml=path option generating a junit-xml style result log.  The xml result log can be parsed nicely by hudson. Initial code was based on Ross Lawley's pytest\_xmlresult plugin
* move standalone script to become a plugin offering "--genscript", adjust paths accordingly and add CHANGELOG entry
* make inspect.getsource work for standalone py.test by implementing a get\_source method on our DictImporter
* fix some standalone-script running issues: \* standalone can run standalone tests \* exception handling is more careful with assuming valid filenames \* bits here and there
* refine tests to cache single-script and make standalone work with distributed testing
* fix accidentally broken pylib's own conftest.py
* a first go at testing schmir's generate\_standalone\_pytest script and making it work on python2.4-3.1 + jython
* deprecate direct definition of Directory, Module, ... in conftest.py's, add some pytest collect related tests + some refinements
* refine deprecations, move some over to test\_deprecated\_api
* fix aimed at passing jstests functional tests: allow to have \_fillfuncargs() called even for non-pycollect-object test-items
* streamline svn test setup a bit, clear caches on setup-restore, hopefully will erase random failures with test\_export
* deprecate use of 'disabled'
* update ISSUES some more, introduce duration to RunResult and a failing dist-testing termination test
* skip tests using 'capfd' funcarg but not having os.dup. cleanup issues and regen plugin docs
* generalize hook calling from collection nodes but stop short of allowing general hooks in python test modules. It'd be easily possible (a 1-line change) but considering it i refrained from it because the collector API is a bit too low level
* only consider matching conftest plugins for discovering hooks related to collection nodes
* fix capturing to be more careful during teardown when a setup never happened (due to e.g. an error in user-provided runtest\_setup code)
* streamline some tests and overall reduce py.test.ensuretemp usage, note down issue about deprecation 
* some debug info aimed at helping to find out about a randomly failing test\_export test setup issue
* make looponfailing a bit more robust against relative imports and changed directories - needs more work, probably
* some testing hygene: move \_reparse testing functionality to actual test support code, un-xfail a now passing test, reduce direct py.test.config usage aiming for deprecation
* fixup funcargs docs
* cleanup py.test.\* namespace, docstrings for improved pydoc and interactive usage. use new apipkg \_\_onfirstaccess\_\_ feature to initialize the py.test namespace with the default plugins.  This, besides other good implications, means that you can now type:  pydoc py.test   or help(py.test)
* simplify Config initialization
* robustiy some randomly failing tests
* remove/reduce internal global state: py.\_com.registry is now fully contained and always instantiated from the py.test PluginManager class
* simplify pluginmanager, move plugin validation code to plugin, remove unused code
* remove defaultconfest.py and make PluginManager directly do early initialization of default plugins
* always import defaultconftest by python import path.  strike some redundant code
* add script to generate standalone py.test
* another fix for windows
* fix typo
* fixing windows32 svn-testing issues
* windows fixes and print funcargs for keyboardinterrupt traces
* try to fix windows path issue
* fixing and cleaning up some tests
* introduce --ignore option to ignore paths during collection
* temporarily adding files for playing with hudson
* rather use newest execnet always
* adding pip requirements file
* fixes to various tests, related to execnet automatic ID generation and other bits. also lowering the version as "1.1.1post1" for now.  1.1.2 is still a bit off
* capitialize I
* cleanup bin-script creation, fix docs, add FAQ entry about py.test --version
* create version/interpreter differentiated py.test$VER for cpython, jython, pypy-c's, prepare 1.1.2 release
* add some issues
* adding a link to the tutorial page
* some doc fixes
* fix links, partially thanks to fijal
* add note related to issue64
* note down dist-testing/execnet issue
* another issue
* adding a warning note, see also issue64
* update some small issues
* skip tests properly
* 2.7's TextIO requires unicode
* fix keyword calling
* starting an ISSUES.txt with a conftest issue
* Added tag 1.1.1 for changeset 319187fcda66
* adjustments and fixes to test run, distribution files. thanks thm
* by default flush log writes to files
* don't consider setuptools plugins if it is not installed
* small refinements/precision regarding execnet checks
* introduce plugin discovery through setuptools "pytest11" entrypoints and refine execnet dependency handling.  Prepare 1.1 release
* some forgotten doc fixes
* fixing docs, adding draft announcement
* fix a flaky test
* add % as allowed char, condense CHANGELOG
* be a bit more helpful by default regarding --report settings
* move CHANGELOG back to root level, add entries
* fix compatibility issue with svnwc.update and put CHANGELOG to rootlevel
* adapt to new execnet.Group code (since execnet-1.0.0b4), strike superflous code
* reintroduce py.test.cmdline.main() (alias for py.cmdline.pytest()) resolves issue #61
* a bit of padding under the logo
* a few internal test related fixes as to run on a osx/no-execnet situation
* improve deprecation, start changelog
* fix a bug with svnwc.listdir() not accepting a checker(versioned=...)
* Added tag 1.1.0 for changeset 60c44bdbf093
* fix up install docs and plugin docs for the final release have CHANGELOG be a file containing links instead of a symlink beause it causes issues with pip-install on some systems
* largely improve and reshuffle docs, heading strongly towards a 1.1.0
* make py lib a self-contained directory again - move and merge \_py/ bits back to py/ - fixes all around
* adding more alternatives as asked for by bluebird75
* update and fix docs for installation - rework installation - add a new FAQ entry related to issue58 Windows/setuptools/multiprocess - strike api/source references
* fix jython issue, flexibilize sysexec params
* trying a bit harder to get a realpath for the py lib because execnet-rsync does not support working with links
* fix windows32 issues, introduce a simplistic path.samefile for it, fix tests
* rewrite nose-optional-call check, fixes python2.4 compat
* move examples to doc directory
* some release preps and cleanups - update setup.py for release - use distributes\_setup on python3 - remove unncessary package\_data - remove execnet example
* moving py/bin to rootlevel bin/ and fixing tests
* remove pyrest and \_py/rest before first 1.1. release
* use new apipkg version
* remove unnecessary builtin directory in favour of a single file
* fix a test-import issue occuring when there is a second 'testing' directory in PYTHONPATH or so
* fix three python3 issues
* resolves issue #59 resolves issue #48
* can't use .format() on jython :(
* merge trunk
* hack around Jython's incorrect AST heriachy
* enabling assertions with jython, fixing one .format occurence to provide the setting for http://paste.pocoo.org/show/147361/
* first round of fixing jython compatibility issues, marking some tests as xfail-on-jython
* using apipkg 1.0b2 snapshot version - adjusting/cleaning up some impl-detail accesses
* merged ronny's nose-compatibility hacks, i.e. nosestyle setup\_module() and setup() functions are supported. added a few notes to changelog and documentation about it
* fix "py.cleanup -d" - add test and check to only remove empty dirs (!)
* fix bug: a false xfail expression would erranonously report XPASS on failures
* remove deprecated parser.addgroup usage in favour of getgroup
* removing some py.execnet references and moving scripts to execnet repo
* support nose style argument-free setup/teardown functions
* nose plugin wont call setup functions that arent made for it
* better tests for the nose plugin, support module level teardown
* nosetest plugin now supports fallback to module level setup
* refine naming, API and docs for py.test.mark mechanism - now contained in pytest\_mark plugin
* use new marking idioms, simplify generalized skipping implementation
* extend and refine test marking - allow to mark tests via a "pytestmark" name at class/module level. - make combined positional args of marker calls available via an \_args argument
* don't visit '\_' attributes on python objects for calling hooks
* cleanup: move creation of python colitems to a default pytest\_pycollect\_makeitem hook impl
* player nicer for missing parent Module objects for a collected function (bug triggered by oejskit)
* flush looponfail output to get around line-buffering
* reshuffle/refine option grouping, introduce "terminal reporting options"
* deprecate addgroup / allow ordering of option groups
* add --report cmdline option, shift refined xfailed and skipped reporting to skipping plugin
* streamline pluginmanager api and test/beautify printing of plugins with --trace
* fix formatting of session log output
* - make importorskip static at py.test.importorskip because it's   used for conditional plugin loading - fix case where xfail is defined at module/class level - fixes and improvements to docs, correct links to plugins - use new skip facilities here and there
* generalize skipping - rename pytest\_xfail to pytest\_skip - dynamic "skipif" and "xfail" decorators - move most skipping code to the plugin
* resolve issue 54 triggered by @haypo's issue and patch the process.cmdexec function now always uses subprocess under the hood. Also fixed some 3k related encoding issues
* adding the console-runtest helper as discussed on py-dev
* introduce "-d" to py.cleanup
* fix some tests after the py/\_py split
* resolves #59 - robustify unittest collection
* ignore more dirs and files
* forgot to commit the verbatim copy of apipkg in \_py/apipkg.py
* \* remove unused py.\_thread namespace, rewrite the one usage \* remove unused py/test/web directory
* rewrote the initpkg mechanism and moved py lib implementation files to
* \* use the MIT license for the py lib \* bump version to prospective 1.1.0b1 \* strike some unused code from initpkg
* \* don't add distributed command line options when 'execnet' is not   installed, report a nice message
* remove py.execnet, substitute py.execnet usages with "execnet" ones
* adding internal repr for debugging adding an example for generating multi-args/multi python tests
* fix typo, add ronny to authors, normalize email addresses
* internally rename "provider" to "factory" to be consistent with documentation
* fix cached\_setup to deal properly for test\_functions with multiple args.  closes #50
* simplify serializer tests
* the check\_sequence name is more specific
* enhance the serializer tests
* add separate test for the serializer bigint fail
* support floats
* don't need to import py
* use default argument
* clean up unused compatibility code
* test cross version serialization by launching subprocesses; much cleaner!
* merging jarko'S fixes, resolves issue #45, resolves issue #46
* Add a simple (hopefully) cross-python marshaller
* update docstring
* add changelog entry for last commit
* allow a path to explicity given for py.lookup
* visit() now returns paths in depth-first order.  fixes issue #47
* \* allowing arbitrary keys for xspecs but adding some sanity checks to xspec-parsing and makegateway. \* fixing a python3 IO issue - we need to retain sys.stdout/stdin   references to keep the underlying byte stream open
* (micke, pedronis)
* Corrected the constructed system path value (broken by the env.cmd, env.sh & env.py file move in 4abc620bb044)
* Fixed a typo in error.py causing it to fail on Windows
* compiling AST to code is new in python 2.6
* \* various cleanups and detailed doc string for gateway\_base module \* remove old multi-file-send mechanism/tests now that   only gateway\_base is send to the other side. \* adding some (c) notices where i am pretty sure about them
* added another funcarg example i had lying around
* fix a bug with funcarg setup and remove XXX comment because "scope=module" now would work but leaving it as session for now
* some doc about the experiemntal pytest\_gwmanage\_newgateway hook. and use process-scope for execnet test funcargs because of weird setup/teardown issues when running distributedly itself
* \* move gateway management code to py/test/dist because it's not clear   how generally useful it is. \* provide pytest\_dist\_makegateway(txspec) hook so that plugins   can add their own interpretation/keywords
* ups, forgot to add a neccessary file
* \* moving execnet tests to funcarg-style, some cleanup \* slight refinement to FAQ license topic
* relicense to LGPL, add an FAQ entry reasoning about it
* monkeypatch, doc, apiwarn, deprecation fixes
* some fixes to support Jython better
* regen setup.py and docs so that "python3 setup.py build" maybe works if setuptools does
* move test files out of py lib proper \* separate all tests from plugins \* simplify implicit inclusion of plugins under test \* have test\_initpkg perform direct checks instead of yielding tests \* fix example tests for 3k
* execnet cleanup/refinements: avoid creating a shell for each subprocess \* introduce HostNotFound, raised for Socket and SshGateways \* factored out basic tests, cleaned up existing tests \* removed sshgateway identity argument which was deprecated in 1.0
* seems like compile is way slower than just parser.suite so we try to see if it's available (only jython doesn't have it)
* \* refactor some setup/teardown/ensuretemp usages to use funcargs \* introduce monkeypatch.syspath\_prepend for safe monkey patching of module import path \* fix monkeypatch naming
* get py.test to run at least basically on top of jython \* allow and document calling of monkeypatch.undo() from a test \* default to 'sys' on platforms there no 'os.dup' is available \* use "compile" to perform "tryparsing" checks
* \* delete or text files to hacking/ directory. \* split license file into authors and license file, minor fixes. \* minor unicode fixes
* \* fixing lots of remaining 3k compatibility issues, mostly with py.test itself
* introduce py.builtin.\_tryimport to try importing modules in a row, fix a few places
* make xmlgen a single file + a single file test instead of a whole directory
* reviewing, refactoring, porting xml/html object/tree generation to work with 3k
* various 3k related fixes and cleanups removal of virtually unused py/rest/rst.py helpers
* fix remaining execnet 3k issues until all tests pass
* remove so-far superflous \_getcode and pickle from builtins, some streamlining
* I think this is supposed to be immutable
* fix xfail
* add a helper to get a function's dictionary
* give code objects a filename in the replacement execfile
* \* fix various remaining 3k issues until test\_gateway.py passes with python3 py/bin/py.test \* we now wait on gateway initialization until we got a byte back after   we sent the bootstrap
* \* simplify stdout/stderr handling and modules and for now remove support   for directly stdout/stderr directly on remote\_exec
* filter out and test exception printing
* \* more tests and fixes for cross-python compatibility \* use byte-buffer files if available for io \* shift receivelock to gateway object \* kill dead code
* \* simplify lock acquiration for received messages, review code \* try to fix seldomly occuring race condition with setcallback/receive and closing of channel
* \* make Gateway interface more asymetric: remote\_\* methods   and  cleanup/atexit handling now live exclusively with the "InitiatingGateway"
* \* use py.builtin.\_getimself instead of getattr(..., '\*self\*') everywhere \* fix logging to work with 3k, implement buffering manually \* fix unicode capturing issue - re-introduce EncodedFile for <3K file writes
* \* refactor gateway code and tests to live in fewer files, remove some lock usage \* move text files to a new "hacking" directory
* enable assertion reinterpretation on 3k
* \* simplify and refactor path tests to use funcargs instead of the layered xunit-setup mess \* port py/path to pass 3.11 \* added py.builtin.\_totext, \_tobytes, \_isbytes, \_istext helpers
* add test for conversion to string
* convert argument to string
* fix module name reuse in execnet
* no unbound methods in py3
* get rid of usage of the new module
* use correct attribute to find the instance of a bound method
* only use cmp() in 2.x
* fix generators on python 3
* add py.builtin.callable
* use py.builtin.execfile()
* add py.builtin.execfile to \_\_init\_\_.py
* add a py.builtin.execfile helper
* use py.builtin.builtins instead of import test
* remove usage of the new module
* handle Queue renaming
* DeprecationWarning is in the builtin namespace
* use print function
* fix typos in converting test\_oldmagic
* guard against tests trying to import this
* fix syntax for py3
* fix some broken things from syntax conversion
* make print write each argument individually
* fix need for py import
* make all syntax compatible with 3.1 and 2.5
* allow file to be compiled on 2.5
* add a test which checks the syntax of the pylib on various python versions
* \* fix some syntax and 3k issues for py/path and py/process, tests only partially working \* have py.process.cmdexec return unicode/text (for now) \* rename py.builtin.basestring to \_basestring
* fix the rest of py/code tests on python 3
* fix tests involving Queue
* fix py/io classes and tests to pass 3.1 introduce py.builtin.\_totext helper to make a 2k=unicode / 3k=str object, allow a string as data
* merge the benjamins and my changes, accidentally caused a new remote head
* replace iteritems() with items()
* fix interpreting is/is not/in/not in
* fixing builtin tests and print\_ builtin
* only test View on 2.x
* new except syntax
* move the old assertion reinterpreting implementation to \_assertionold.py
* remove magic directories from install
* implement assert debugging with builtin AST
* make the patched compile() work with AST
* fix location of magic AssertionError
* temporary checking towards 3.1 compatibility introduced some helpers to py.builtin namespace which need some review after things begin to work more nicely
* \* add print\_, exec\_ and \_reraise helpers for 2-3 compatible code \* consolidate builtins implementation to be compatible with >=2.3
* simplify broken-repr test for python2.4
* deprecate py.compat.doctest|subprocess|textwrap|... (and for now pass through Python stdlib provided modules)
* deprecate py.magic.autopath() and finally remove py/magic directory
* \* deprecate py.magic.invoke/revoke in favour of   the new py.code.patch\_builtins, py.code.unpatch\_builtins
* merge 1.0.x branch to trunk, fix doc link
* fix tests to not fail if pyc-file-writing is disabled
* switching release branch back to 1.0.x versioning
* Added tag 1.0.2 for changeset 4816e8b80602
* finalizing doc bits for 1.0.2
* merging 1.0.x branch
* some fixes and small doc updates, aiming to finalize 1.0.2
* \* using a MANIFEST.in file instead of our own generated MANIFEST file \* port the test\_install refactoring from trunk, move to bin-for-dist
* consolidate py/code files into code.py, simplify SafeRepr code and docs
* move and rename html rest helper to rest directory - finally remove py/misc completely
* simplify caching class
* death to "misc" directories.  moved most files out of py/misc, either to a private attic or to other places in the lib
* simplifying errno error class creation and introduce a py.error.checked\_call helper that creates a proper errno-specific exception instead of OSErrors.  use it from py.path.local
* add delattr/delenv/delitem methods and tests, enhance terminalwriter tests
* consolidate py/log into fewer files, remove one old approach, sketch simplified API
* moved laura's utestconvert script to a more visible place
* \* introduce py.io.TextIO and py.io.StringIO to help with 3k transition and to clarify   intentions when doing "in-memory" files. Replace most usages of StringIO
* fix autopath bug introduced with path refactoring
* consolidate svn path implementations and tests into files named after the package namespaces
* - strike lots of basically unused code around local path implementation.   and tweak things a bit to make py.path.local at least importable on 3k
* move localpath implementation to a single file, simplify unix/posix difference and fix a bit
* kill/replace some execnet debug code bump version to "trunk" on trunk add "py" to rsyncdirs
* remerge from 1.0.x after changelog addition change
* adding a link to changelog, hope that mercurial versions links correctly
* also merging the tag from 1.0.x branch
* merging 1.0.1 branch
* Added tag 1.0.1 for changeset 6bd221981ac9
* finalize release announcement
* enable nose by default
* [mq]: 101-prep
* simplify approach to encoding of sys.stdout/stderr files, encode unicode strings and pass-through non-unicode strings
* added a pytest\_helpconfig plugin which groups --version and the new "--help-config" option. rename options and configuration names. streamlines docs
* shift install test, fix example test, add 1.0.1 release announce
* [mq]: 101doc
* fix --version test, set version to 1.0.x rather than 1.0.1 for now, regen setup.py/MANIFEST
* use item's own fspath when doing progress reporting, fixes #31
* merge 1.0.x changes back to trunk
* fixing svn status on incomplete files issue #32
* add a --version option to print the pylib version
* [mq]: monkey
* adding a 5 LOC plugin for capturing and ignoring the output of test function calls
* [mq]: flexcom
* \* add pytest\_nose plugin \* have unittest functions always receive a fresh instance
* simplify internal plugin dispatching code, rename parts of the py.\_com plugin helpers
* turn some tests from skipped to xfail strike dead code, small refinements to xfail exception reporting
* merging 1.0 branch
* Test\* classes with an \_\_init\_\_ method are not collected anymore
* nicer error message for non-collectable test files / items fixes issue #27
* [mq]: rename
* more terse reporting of collection errors / tracebacks, bump version number aiming for 1.0.1
* \* fix capturing and unicode printing in tests \* introduce "\_encoding" to py/io/terminalwriter writing \* beautify a few \_\_repr\_\_ for better internal debugging
* update various doc bits, fix typos
* merge 1.0.x branch
* Added tag 1.0.0 for changeset 7acde360d94b
* regen
* Added tag 1.0.0 for changeset 5ea0cdf7854c
* \* rename "rep" to "report" in reporting hooks \* refine docs \* bump version data \* improve announcement
* fix bestrelpath related tests and docstrings
* make pylint plugin not bail out early, still does nothing useful, though
* showing relative paths by default, improving xfail output a bit, bumping version info to 1.0.0
* shift options and tests to terminal plugin
* rename iocapture to capture, add/remove some missing files
* rename pocoo to "pastebin" plugin, write documentation, implement whole-session-pasting
* merge 1.0.x branch
* Added tag 1.0.0b9 for changeset e2a60653cb49
* fixes for python 2.4
* regen manifest, improve docs generation
* fix/work around some corner cases for dist-testing
* handle final teardown properly, add a new experimental hook for it
* introduce new "Error" outcome and group setup/teardown and collection failures into that category. Report them separately
* \* reworked capturing to only capture once per runtest cycle \* added readouterr() method to py.io capturing helpers
* fix a svn-1.6 issue
* properly handle test items that get locally collected but cannot be collected on the remote side (often due to platform reasons)
* simplify py.test.mark API, add more plugin docs
* improve assert re-inteprpretation for comparisons
* enable capturing during collect
* \* reworked per-test output capturing into the pytest\_iocapture.py plugin
* targetting a b9
* remove not used files
* merge 1.0.x branch
* Added tag 1.0.0b8 for changeset 0eaa0fdf2ba0
* update docs, leave out internal plugins
* Added tag 1.0.0b8 for changeset c63f35c266cb
* regen links and add oejskit place holder text
* Added tag 1.0.0b8 for changeset 86f1e1b6e49b
* regen MANIFEST, update changelog
* applying samuele's patch fixing a regression resolves issue #26
* adding a test and some support code for creating an "sdist" package, unpacking in a virtualenv and importing and verifying the version
* \* refine keyboardinterrupt handling for --dist \* generate tests for various options
* poll for child process status after exit procedure
* merge 1.0.x branch
* regen plugin docs and links
* improve unittest documentation, enable plugin by default
* new cleaner layout, many improvements to docs
* introduced pytest\_keyboardinterrupt hook removed optional argument from pytest\_sessionfinish hook
* \* move some test\_pytest\_terminal tests to become functional/acceptance tests
* split off tests from pytest\_terminal reporter plugin
* resolves issue 25
* merge 1.0.x changes
* trying to tweak setup and py version attributes to not cause merging conflicts
* regen setup/manifest
* \* add a script for generating plugin docs \* improve generation of docs \* refine plugins docs
* adding a new execnet example
* merge trunk changes
* fix test
* fix logging interaction issue
* merging 1.0.x changes
* fix figleaf plugin import
* \* streamlining plugin docstrings \* better organisation of example directory, also does not break   whole-test run anymore
* merging some old changes (still struggling with mercurial a bit)
* \* refix handling of partial setup failures \* shuffle / consolidate related tests \* re-gen setup.py
* merging 1.0.x branch
* adding example scripts from tutorial
* \* refined funcarg docs and CHANGELOG \* fixed funcarg setup and error-raising issue
* extending and fixing docs about - test config, funcargs - talks and tutorials
* fixing setup.py in trunk
* merge 1.0.x changes
* also aadd a changelog entry
* (radomir, holger) tweak doctest reporting for docstrings from python modules
* trying to get together a working 1.0.0b7 without ez\_setup.py
* merge 1.0.x branch
* Added tag 1.0.0b6 for changeset 2cc0507f117f
* fix typo
* re-introduce py.test.mark and move py.test.xfail to py.test.mark.xfail
* re-adding py.test.mark as documented (!) by adding pytest\_keyword plugin
* improve docs some further, cleaner release announcement
* doc refinements introduce '\_\_channelexec\_\_' + docs
* merge 1.0.x changes / fixed setup.py
* tagging 1.0.0b5
* merge 1.0.x release branch changes
* another attempt at a full tarball
* Removed tag 1.0.0b4
* tag revision 1.0.0b4 for upload
* Added tag 1.0.0b4 for changeset 8cd6eb91eba3
* Removed tag 1.0.0b4
* Added tag 1.0.0b4 for changeset 1c7aaa8c61f3
* add py.test version to verbose reporting correctly regen setup
* resolves issue 18 multiprocessing py.test co-existence add fileno() method and test DontReadFromInput redirection some more
* resolve issue 23
* regen setup.py, already bump trove classifier
* add test and fix keyword recogniation, thanks Andreas Kloeckner
* addresses issue #22 allow test functions to have no func\_code
* perform state setup finalization earlier
* remove scope argument from request.addfinalizer
* bumping version to "1.0.x" for now, regen setup, opening 1.0.x branch
* remove TODO file (mostly done, superflous or in the issue tracker), remove keyword subsitution in text files. regen setup with hg info instead of svn
* Added tag 1.0.0b3 for changeset 52c6d9e78777
* b3, fix (likelY) windows-path-comparison related issue at plugin registration link to PyPI
* snapshot 1.0.0b2
* fixing some release/apigen related issues
* fixing docs, adding newest ez\_setup.py, fixing a few win32 failures, fixing iocapture plugin
* shift deprecated\_call helper into pytest\_recwarn plugin
* \* introduce and document new pytest\_namespace hook \* remove py.test.mark helper \* move xfail to work directly on py.test namespace, simplified
* bah - i guess i rather keep using setuptools for 1.0 because of the issue of installing non-setuptools mediated windows scripts on previously installed with setuptools ones
* \* refine collect hooks and docs, remove pytest\_collect\_recurse \* write and extend extension docs
* \* refine and rename pycollect related hooks \* refine runtest/test function call protocol
* \* remove redundant call\_next\_provider method, thus generalizing Request object \* refine according documentation
* striking config from Node signature
* remove unused function
* rename api to hookspec and use direct modules instead of class
* consolidate extension docs into a single file
* fix threading io-redirection tests
* \* fix too-many-open files issue with fd based io-capturing \* workaround 2.6 issue with pickling of objects from global conftest.py files
* document and refine/rename item \*\*runtest\*\* hook invocations capture output separately for fixture and actual test run
* \* rename, cleanup and document runtest hooks \* factor runner code into pytest\_runner plugin \* cleanup setupstate handling
* restart system calls on SIGCHLD signals (needed for OS X) this only works in python >= 2.6
* adapt to new plugin architecture and misc cleanups: - exit with non-zero value on import errors - remove unused imports - make --twisted-logging a boolean option
* fix generator tests
* fix typos in comments
* fix a bug that collectonly reporting did not show internal errors (thanks ronny)
* introduce Function.\_genid, also used for more correct equality implementation
* shift tests and remove superflous files
* integrate plugin hook checking directly when registering remove plugintester plugin, all functionality now in testdir
* \* rename testrunstart/finish to sessionstart and pass it an argument \* simplify pyexecnetcleanup plugin
* \* introduce pytest\_pdb: plugin handling --pdb invocation \* killing some unused/unneccessary hooks
* refactor basic\_run\_report a bit to allow for immediate PDB runs and clearer code
* make setupevent tests of pytest\_runner.py work
* derive Exit from KeyboardInterrupt to simplify exception catching
* merging pedronis-branch changes to resolve issue 15 for now. This introduces py.test.collect.\_fillfuncargs which can operate on Function like objects
* addresses issue15
* minimal changes to expose fillfuncargs
* regen setup.py, striking some unused headers
* add a very basic test and include \_pytest dependency. thanks pedronis
* introduce and document "session" scope for finalization helpers
* fixed examples according to new plugin and setup mechanisms
* \* xspec: define str(xspec) to give back the original specification string \* fix a test invocation
* rename sorter to reprec (report-recorder)
* update todo to make some sense again shift samuele in license file
* \* streamline hookrecording and global py.\_com.comregistry patching \* this fixes a reporting bug where events from inner test runs would   mix with the outer test run
* \* create funcarg Request object only once per function run setup \* add getfuncargvalue() for retrieving arbitrary funcargs from a provider
* plugin cleanups \* make pytest\_eventlog.py work again by adding a hack to the registry, rename \* disable resultdb hook plugin, it needs merging with resultlog \* add some docstrings, streamline bits
* remove superflous registry.call\_firstresult and NONEASFIRSTRESULT logic
* adding not implemented test to not forget about it
* nicer docstring for monkeypatch funcarg plugin
* rename recorder functionality, unify interfaces
* merge js-infra-cleanups changes
* sorting should be an option collector->child interface
* move \_getparent to Node and make it public
* updating plugin docs a bit
* \* refactor plugin support to work directly with   modules, no classes required anymore. \* call funcarg hook if defined on class
* get to a workable state for cached\_setup() and docs, move msot related code to SetupState class
* imported patch usegentest
* small documentation refinements
* adding exampels as per blog post
* apply and generalize patch from Ronny regarding dumb terminals, add doc note
* streamlining pocoo plugin
* addcall now takes direct funcargs values alternatively
* simplify example code
* refining examples and updating docs with actual output from examples
* implement funcargs according to docs, streamline docs
* merge samuele's metainfo() branch also fix a test after broken rename
* updating docs wrt funcargs and deprecations - still need updated implementation
* - rename metainfo to reportinfo for clarity
* test through the now public entry point
* introduce a pytest\_collect\_metainfo hook
* - expose py.code.getfslineno and use it in pycollect
* test fix, .py vs .pyc
* - implement a general getfslineno helper in code/source.py with tests
* avoid superflous methods
* bring back nicer headlines for failure tests
* add a remark about distinction of "generating" versus providing funcargs
* catch duplicate \_id values
* adding \_id parameter to addcall, refinements
* rename hook, rename addfuncarg to addcall interface, forget about combinations of funcargs
* merging the new function generators, addresses issue 2 - introduce a new pytest\_genfuncruns hook for generating tests with multiple funcargs - new and extended docs: doc/test/funcargs.txt - factor all funcargs related code into py/test/funcargs.py - remove request.maketempdir call (you can use request.config.mktemp)
* don't print py lib revision always
* care for threading module being set to None already during finalization
* better docstring
* test reporting of itemstart events some more
* better name for hook
* fixing a windows issues
* fixes issue 6 by reverting back to issuing attributeerror
* rename "api" into "hook" in most places
* fix issue 7 (delay in test writing)
* fix docstring
* provide basic metainfo because the terminal reporter needs it for printing progress
* provide testdir.spawn\_pytest for pexpect mediated interaction tests, kill code, yay
* remove unused option once again fix resultdb plugin
* \* performing funcarg setup during setup-phase \* make item.runtest() be a clean function and have   the caller deal with the deprecated invocation
* \* kill some code and indirections \* have metainfo() directly provide fileinfo + message
* greatly refine deprecaction warnings and use new "recwarn" plugin for it
* a new plugin for asserting that warnings have been shown to a user
* fix import issue
* fix missing import skip problem
* extend py.cleanup to supply a list of extensions to clean
* fixing compat test for 2.6
* due to "conftest.py" already being loaded when collecting directories we cannot skip here
* add an example for lazy "per-directory" setup
* better error messages as discussed here: http://codespeak.net/pipermail/py-dev/2009q2/001113.html
* refine error message for inconsistently failing asserts
* fix test for preparsed -p
* fixes #3 added tests and fix for early parsing of "-p" option
* more or less fixes #4 along with changeset e07b15140498
* ref addresses 4
* remove unused import, fix failing test mode
* merging from py-trunk
* [svn r64263] fix failing test on windows
* [svn r64259] porting afa's 64232 and 64253 to trunk
* Deleted duplicate declaration of 'pytest\_item\_runtest\_finished()'
* \* don't version generated .html files \* schedule \_findpy for removal sometime
* [svn r64091] adding missing file
* [svn r64077] sync from hg-trunk
* fix funcarg usage in execnet, regen setup
* adding a bit to changelog although that's almost hopeless
* merging funcargs docs and branch
* finalized docs and funcarg attributes, i think
* bringing docs and funcargs in sync
* fix reporting failure
* avoid .swp files
* remove print, shift code
* funcargs work mostly according to the documentation
* fix a bug
* implement request object as per docs
* move tests to test\_funcargs.py and start with new request object tests
* fix note on funcargs
* refining docs
* fixing ReST typo
* refining docs further
* refinements
* streamline
* refining formulations
* section on xUnit style setups, better linking, some streamlining
* [svn r64029] \* move test docs into their own subdir \* don't use sphinx for now, put the support code to attic \* remove some old docs
* small updates
* update docs
* [svn r64028] use pygments for syntax-coloring python code and console
* [svn r64027] make parser.getgroup create a group if it doesn't exist
* use pygments for sourcecode highlightning
* getgroup can also create groups now
* [svn r64007] port 64001-64004 to trunk
* [svn r64000] can't bear the setuptools problems anymore.  trying to switch away from it
* whitespace changes
* [svn r63991] adding ignore file
* [svn r63990] \* adding XXXed funcargs docs, moving release announcements to own subdir
* removing old file
* remove outdated info
* [svn r63938] Changed methods pyevent\_\* to pytest\_\* due to the changed plugin-API
* [svn r63932] \* Fixed broken setup\_method()-mechanism. \* Added --twisted-logging option. \* Added example test-script to the plugin-test
* [svn r63917] remove now unused code
* [svn r63916] KISS: remove interaction of warnings with plugin mechanism, it's not needed
* [svn r63914] py.log.\_apiwarn is fine enough because it's something of an py lib internal api deprecation mechanism atm
* [svn r63913] make py.log.APIWARN available
* [svn r63911] striking unneeded call\_each from pluginmanager
* [svn r63910] removing old mentionings of "bus". docstrings
* [svn r63908] first step in only allowing keyword arguments to plugin calls
* [svn r63905] \* have pytest pluginmanager do multicalls for registration \* more renaming
* [svn r63902] bugfix
* [svn r63899] more consistent naming
* [svn r63896] more renames, killing redundant code
* [svn r63894] grand renaming on plugin-related mostly internal objects
* [svn r63890] group hooks
* [svn r63889] renaming/streamlining missing event usage
* [svn r63888] finally.  the event concept is basically gone. now we only have plugin hooks aka plugin calls
* [svn r63887] move rest of events
* [svn r63886] another few events
* [svn r63885] moving two more events
* [svn r63883] \* moving many more events to become ordinary plugin hook calls. \* bit hackish because the code for handling the old events is   also still there ..
* [svn r63874] remove redundant code
* [svn r63869] \* moving execnet events to become api plugin calls. \* defining Execnet hooks in an explicit API
* [svn r63868] move towards having funcarg setup be part of setup()
* [svn r63846] fix bug
* [svn r63845] introduce new \_pytest plugin that allows to selectively record plugin calls and do assertions about them
* [svn r63822] simplifying and strictifying the internal calls to plugins using the new "api" mechanism
* [svn r63821] Added comment due to the breaking of the setup\_method() mechanism
* [svn r63808] \* refinements/renames to new PluginAPI \* have pytest\_runner start to use it, passes the main test
* [svn r63806] adding a MultiAPI helper for managing plugin APIs
* [svn r63805] first step towards having a dedicated test-run plugin, some more basics missing
* [svn r63797] introduce a clean "guardedcall" interface for performing calls with exception catching and stdout/stderr capturing
* [svn r63796] Initial import for pytest\_twisted
* [svn r63795] Initial import
* [svn r63786] rename CollectionReport to CollectReport, remove XXX
* [svn r63785] further renaming, streamlining the testing machinery and helpers
* [svn r63780] some renaming
* [svn r63686] fixing docstring, removing dead code
* [svn r63685] internal renaming of variable name
* [svn r63683] a small script to perform svn-hotsyncing
* [svn r63630] some more code to prepare for substituting the fixture/runtest code, using more plugins
* [svn r63628] \* shuffle SetupState and fixture handling into runner.py \* introduce a itemsetupreport and new setupitem/teardownitem methods. \* more tests
* [svn r63601] remove dead code
* [svn r63600] merge remaining content of event.py into runner.py
* [svn r63598] cleanup of imports and shuffling of plugin method
* [svn r63597] remove unused code. fix errors
* [svn r63593] have plugin method run the test item and report about it
* [svn r63588] shift running of test item to be done through a plugin method
* [svn r63584] killing more code, simplifying running of tests
* [svn r63580] \* removing usage of py.magic.patch \* removing obsolete greenlet.py
* [svn r63579] step one in simplifying runtest() collect() semantics
* [svn r63575] killing yet more test "events" and redundant code
* [svn r63571] kill another few test "events"
* [svn r63566] the InternalException event class bites the dust
* [svn r63563] the TestrunFinish event class bites the dust
* [svn r63562] fixing \_com regarding pyevents as well
* [svn r63560] beginning to review/polish test events \* pyevent() now receives args and kwargs as simple arguments \* refactoring event handling in tests
* [svn r63553] \* allow doc to be tested distributedly \* be more eager to cleanup execnet gateway after tests
* [svn r63552] adding processing of pytest\_restdoc plugin back for now
* [svn r63551] keep style.css around here
* [svn r63550] \* move hook and event API definitions to py/test/plugin/api.py \* make "py.test" work on "doc" directory again
* [svn r63549] Attempted to create a theme based on the previous pylib pages. This initial version looks worse than either the old pages or the default sphinx output. Can revert back to the default theme by referting the change to conf.py Also added a new make target to force all html to be rebuilt. Sphinx doesn't detect changes to stylesheets in themes
* [svn r63548] fixing lots of little issues with the docs
* [svn r63547] renaming all event consumers to use the "\_\_" convention that is also used for funcargs
* [svn r63546] small updates, regen setup.py
* [svn r63545] fixes to toctree updated headings so the outline works
* [svn r63544] fix typo in my name
* [svn r63543] some import-related skips, test suite passes again
* [svn r63542] moving pygreen to greenlet contrib dir
* [svn r63540] fixing shallow test bugs
* [svn r63539] moving py/doc to doc/ and removing some more files
* [svn r63536] refining docs some more. removing outdated or what-i-consider-unfitting docs
* [svn r63530] committing some old text about state management
* [svn r63529] remove idea here for a new "fs" namespace. 1.0.0 release announce draft some fixes and streamlines here and there
* [svn r63523] adding the pylib logo (from the webpage)
* [svn r63522] ReST fixes, Copyright fixes
* [svn r63521] fix py.std docs and simplify its implementation (which is a 5 liner, anyway)
* [svn r63520] \* removing versioning of generated \_build directory. \* removing an outdated note regarding py.std
* [svn r63519] Initial Sphinx-ification of the documentation
* [svn r63473] make passed, skipped and failed all default to 'False' for consistency when the attributes are not found
* [svn r63472] a new plugin which writes a simplified result log to either JSON or a SQLite database
* [svn r63462] fix syntax error (ehem)
* [svn r63461] a few fixes, now figleaf writes files correctly. also changed figleaf plugin to include only the files of the current working dir
* [svn r63430] First cut of pylint plugin. Not in working state
* [svn r63429] pytest\_figleaf now filters out all coverage for files outside of the tested package
* [svn r63428] added happy path tests for figleaf plugin
* [svn r63427] added figleaf plugin to pylib
* [svn r63360] bumping to first beta
* [svn r63359] change funcargs naming to use \_\_
* [svn r63355] bump version
* [svn r63353] better command option handling
* [svn r63346] update docs
* [svn r63345] better naming
* [svn r63341] regen setup
* [svn r63340] remove greenlet from py lib
* [svn r63339] undo rev 63000 so that there is only one method now for funcargs
* [svn r63338] rename pyfuncarg to funcarg
* [svn r63337] provide more info for the pyfuncarg failing lookup improve docstring
* [svn r63319] \* fixing verbose reporting to work the "old" way for non-dist settings \* bump version number \* comment out greenlet C-Extension for now
* [svn r63318] fix typo
* [svn r63302] avoiding a (what i think mostly) mac python setup issue
* [svn r63301] don't open the anyway unused stdin, this can (at last on python 2.4 on Mac) leave open pipes in the process
* [svn r63300] \* refining pyfuncarg setup, now there is explicit registration! \* porting monkeypatch and pytester funcargs to the new method \* fixing a kind-of-a-bug with MultiCalls
* [svn r63253] fix typo
* [svn r63251] fix reference
* [svn r63244] update the main entry doc page
* [svn r63242] fails on windows, unncessary to test anyway
* [svn r63241] regen setup, fix gensetup script
* [svn r63238] rather going for some alpha releases until i am sure that packaging works well
* [svn r63235] bumping version to 1.0.0b1
* [svn r63234] fix gensetup, regen setup
* [svn r63232] nicer repr for gateway.\_rinfo() informatio about remote location
* [svn r63229] don'T generate apigen information for now
* [svn r63224] regen setup
* [svn r63223] cleanup contact page
* [svn r63222] refactor test-features doc, make references to others
* [svn r63219] fixing tests to care for underlying symlinked working directories
* [svn r63218] temporary fix for the issue that rsync does not synchronize executable bits yet
* [svn r63217] \* remove obsolete doc \* regen setup.py \* update README.txt
* [svn r63216] 72: the code fragement -> fragment 114: You can use a string speficiation -> specification 147: Examples fo -> of
* [svn r63215] updating test distribution docs
* [svn r63214] re-gen setup.py with up-to-date list of files
* [svn r63213] fix typo
* [svn r63212] completely revamp the execnet web page
* [svn r63210] \* implement "--dist=each" to distribute each to each available node \* improved node-management and nice showing of results \* streamline command line options
* [svn r63209] polish command line options for distributed testing
* [svn r63208] moving example to top-level dir
* [svn r63203] add a method that allows to send an item to multiple nodes
* [svn r63202] prepare for allowing for items to be sent to multiple hosts
* [svn r63201] allow to specify "3\*" for host specs
* [svn r63200] new docs about testing refactoring of documentation new entry page
* [svn r63197] \* don't think "showskipsummary" is a useful option \* some typo fixes
* [svn r63195] \* fix rsyncdir usages all around \* rename looponfailing to looponfail
* [svn r63194] more option streamlining
* [svn r63193] normalize towards 'iocapture' option
* [svn r63192] rename dsession to dist
* [svn r63191] \* remove "--exec" \* cleanup of options
* [svn r63190] no resultlog plugin by default have custom options come last in "py.test -h"
* [svn r63188] allow reversed retrieval of methods
* [svn r63187] tweaks to cmdline options
* [svn r63178] remove wrong assert
* [svn r63177] add a "-p" option to allow to add plugins from the command line
* [svn r63171] avoid random writes to tempdir
* [svn r63164] rebuild setup.py
* [svn r63163] fixing tests, better verbose output for dist-runs
* [svn r63162] all tests pass again, output on test node setup and initialization is now much nicer
* [svn r63160] rename masterslave -> txnode
* [svn r63159] re-group methods
* [svn r63158] UI improvements
* [svn r63157] remove old events, some more renamings
* [svn r63155] comment cleanup
* [svn r63154] use "node" instead of "host" everywhere
* [svn r63153] refactoring almost complete, apart from testnodeready info
* [svn r63151] fix various bits, many tests pass modulo dsession host/node handling
* [svn r63150] fix hostmanager to work with new xspecs
* [svn r63149] always have a default chdir
* [svn r63148] add a \_\_repr\_\_ for xspecs
* [svn r63147] get rid of old method for specifying remote execution places
* [svn r63145] add support for "chdir"
* [svn r63144] have socket and ssh gateways
* [svn r63141] makegateway works for plain popen and popen+python-version
* [svn r63139] make \_rinfo() cache results by default
* [svn r63137] a helper some basic remote info
* [svn r63136] better grouping of gateway public API
* [svn r63135] adding a new gateway specification method
* [svn r63110] Typo in a (unused) constant
* [svn r63107] cleanup event handling of dsession
* [svn r63106] rename host -> gateway in a couple of places
* [svn r63105] remove non-existing event
* [svn r63104] \* introduce rsync events \* only rsync once if a gateway is specified multiply
* [svn r63103] samefile is only available on posix
* [svn r63102] add a "samefile" method to local path objects
* [svn r63091] report some more info on dist-setup shift comments
* [svn r63089] factor out MultiChannel and MultiGateway helpers
* [svn r63073] make sure that generated test names are always unique
* [svn r63061] dont print too much info on non-debug
* [svn r63060] execnet really needs to improve its bootstrapping
* [svn r63056] try to ignore build directory (which shouldn't be there at all) for rsyncing
* [svn r63054] several windows fixes, test suite passes now remotely
* [svn r63053] forbid too small windows width for now
* [svn r63052] some fixes for win32
* [svn r63051] be very careful when we cannot unpickle an colitem because its parent can't collect the same way as on the sending side. (due to platform skips etc.)
* [svn r63050] \* disabled classes or modules will now lead to a skip during setup of the colitem \* more graceful handling when "ssh" is not present
* [svn r63048] fix windows issues
* [svn r63046] change a few comments
* [svn r63040] try harder to record and auto-exit gateways after test runs
* [svn r63034] fixing some tests to work across python revisions
* [svn r63028] allow py.test --exec=python2.4 -n 3 to work and fix some bugs from doing so
* [svn r63027] new API:
* [svn r63024] don't close\_fds by default
* [svn r63020] allow to specify python executable in gatewayspecs, fix a few tests
* [svn r63019] some simple ssh tests, always transfer py lib as rsyncroot
* [svn r63018] rename "--hosts" to "--gateways" to make naming more consistent with py.execnet
* [svn r63017] rename colitem.\_config to colitem.config - it's an official attribute
* [svn r63016] \* fix setattr on apimodules \* higher timeout
* [svn r63014] streamlining multichannel interface, fixing test work with -n 3
* [svn r63013] adding a failing test for --dist-each
* [svn r63000] \* do an as lightweight config.mktemp() as possible \* avoid writing lock files if lock\_timeout is 0 or None
* [svn r62999] this test can easily pass
* [svn r62998] tweak another place to not create random subdirs
* [svn r62997] some more tests, seems like temp test dirs are now more contained when doing distributed testing
* [svn r62995] try to contain session test tempdirs in one parent
* [svn r62994] introducing internal MultiGateway class
* [svn r62993] \* moving ensuretemp to config object \* adding --basetemp option \* added/rewrote some tests
* [svn r62987] add a "setenv" helper for setting a value in the environment
* [svn r62986] fix tests that got broken
* [svn r62985] rewrite and fix a skipped test
* [svn r62984] allow for host to go down if it didn't go up yet
* [svn r62983] forgot to add traceback cutting tests long time ago, i guess
* [svn r62982] report basic configuration errors more gracefully to the user
* [svn r62981] allowing conftest to set default values for options
* [svn r62978] \* introduce --hosts and --rsyncdirs optiosn \* re-sort option groups, disable some options for now \* add docstrings to execnet gatewaymanage \* streamline tests a bit \* unify debugging and tracing
* [svn r62977] actually make use of option group description
* [svn r62975] - use "popen" instead of localhost if you want popen-gateways - have a default joinpath
* [svn r62974] don't allow "\_" in py.test.mark attributes
* [svn r62972] Oops, accidentally checked in a change, now rolling back (added 'apigen' to enabled plugins, which fails if the apigen package is not on the PYTHONPATH)
* [svn r62971] Re-added 'gendoc.py' script to build the py lib documentation, now using the externalized apigen package. The package should be on the PYTHONPATH, if not the script warns
* [svn r62967] fix mock object
* [svn r62857] Always substract the last column from the Win32 terminal console: otherwise the last char wraps and the \n causes an empty line
* [svn r62835] Add ANSI colouring to the Win32 console. This gives a nice display for py.test, and during pypy translation
* [svn r62832] unbound res on special files
* [svn r62754] Seems w3 have updated their HTML
* [svn r62738] quick fix of setup.py
* [svn r62662] introduce another hook "pytest\_item\_makereport" to allow a correct implementation of the xfail plugin: previously --pdb and -x would incorrrectly terminate if seeing a marked-to-fail test
* [svn r62633] fix path bugs, don't feel like writing a test for this, currently
* [svn r62617] renaming doc, small fix
* [svn r62615] implement a use case for avoiding recursion into a directory
* [svn r62614] implementing \_\_call\_\_.exclude\_other\_results() to allow plugin hooks to exclude results from other plugins
* [svn r62613] introduced a pytest\_collect\_directory hook
* [svn r62612] moved the attic somewhere darker
* [svn r62611] yay, the global setupstate died.  I guess what kept me from doing that was all kinds of tests breaking.  but they are cleaned up now. so there was no problem. nice
* [svn r62610] finally fixed a bug related to the global SetupState for test functions.  streamlined  testdir.inline\_run functions.  well killed most of them
* [svn r62492] removing apigen bits from py lib, they are now in svn/apigen
* [svn r62491] fix a typo
* [svn r62487] merging hostmanage branch: \* cleanup of the way distributed/remote sessions are setup up \* simplified config pickling \* configs are now more correctly wired on receival at remote sides \* introduced py.\_\_.execnet.gwmanage helps managing calls to multiple hosts \* grouping all pickling related tests in test\_pickle.py   and showcasing a nice pyfunc\_call hack
* [svn r62430] honour unitttest function default values for pyfuncarg protocol
* [svn r62352] add a note about how to specify plugins
* [svn r62348] add support fo setting command line options from PYTEST\_OPTION\_NAME environment vars add a first bit of documentation
* [svn r62333] remove apigen here, it is now at svn/apigen/trunk/pytest\_apigen.py
* [svn r62292] use prefix "do\_" for configure/unconfigure calls to plugins in addition to addoption
* [svn r62290] slightly sanitizing initialization of serialised config objects
* [svn r62288] better error handling
* [svn r62287] By default cut traceback such that py lib code does not appear test tracebacks
* [svn r62282] refine a test
* [svn r62281] allowing a plugin to see its registration and unregistration events
* [svn r62256] getting rid of "--session", you can do it by creating a plugin and calling setsessionclass from "pytest\_configure"
* [svn r62252] renaming "py.test.keywords" to "py.test.mark"
* [svn r62245] fixed more wrong renames, fine-tuning
* [svn r62244] fixed wrongly renamed methods
* [svn r62211] merge 60797:HEAD of pytestplugin branch: this merge contains:
* [svn r61975] subprocess is in stdlib since 2.4 for backwards compatibility we use compat.subprocess if needed
* [svn r61965] correct help string
* [svn r61961] improve documentation about "-k" keyword handling
* [svn r61954] removed of using some deprecated modules/functions from stdlib. used subprocess instead (the module is included to stdlib since 2.4)
* [svn r61864] remove unnecessary bracket from the help message
* [svn r61296] a failing test
* [svn r60977] add a few notes that i discussed with Brian when i was in Seattle
* [svn r60976] a few notes regarding improving remote tracebacks for py.execnet
* [svn r60975] a bit of planning for 1.0
* [svn r60974] some reformulations, less "magic" mentionings
* [svn r60277] triggered by https://codespeak.net/issue/py-dev/issue66 \* check for 1.5 messages for non-versioned resources \* modify non-recursive commit test to account for subversion   behaviour changes (non-recursive commit on directory   will now commit modified files in it, but will not recurse   or add subdirs)
* [svn r60142] - experimental support to let decorators used on tests specify a saner order by attaching the original function (we have been using something like this for quite a while at OE) - make the explicit names test more paranoid
* [svn r60141] refactor things such that all fallbacks to inspect.get/findsource happen in helpers in source.py
* [svn r60140] simplification, Source will invoke inspect.getsource on its own
* [svn r60139] some minimal testing for Code.source/fullsource
* [svn r60137] support for explicitly named generative tests with tests and doc addition
* [svn r59871] yet-another-except when trying to get source
* [svn r59728] fix broken import exposed by a pypy test (bah)
* [svn r59615] given the format and the fact that external timeout logic may kill a py.test emitting a resultlog switching to line buffering seems a good idea
* [svn r59464] Fix for a corner case: when the arguments are 'del'-eted from the local scope.  This can also occur when using Psyco because f\_locals is then empty
* [svn r59344] adding an item to consider
* [svn r59268] An improvement to pdb (which I should also propose as a CPython patch): in post\_portem, the "up" command isn't normally able to walk up past generator frames.  This fixes it
* [svn r58888] Catch all possible find\_source errors
* [svn r58775] only check for Recursion if we have a RuntimeError
* [svn r58699] show traceback before jumping into pdb
* [svn r58690] \* add to TODO, changelog \* regen setup
* [svn r58576] \* fix test to not rely on time.time() resolution \* strike some superflous \_\_init\_\_ code
* [svn r58517] clarify/better english for the cross-python version namespaces
* [svn r58516] fix typo
* [svn r58386] adding info for debian and rpm systems
* [svn r58385] \* move cross-platform process kill functionality and move it to   py.process.kill(pid) \* simplify test\_auth.py \* use new functionality from some tests
* [svn r58384] adding a test for unimplemented recording feature
* [svn r58375] some updates on the filesystem front
* [svn r58360] factor import of xml into one method to deal more gracefully with implementations that don't have expat
* [svn r58350] use optparse
* [svn r58340] this seems to need to be this way, otherwise a py.lookup from PATH will be used or a failure will occur instead
* [svn r58319] be more grateful on tests that disappear during a looponfailing session
* [svn r58318] (iko, pedronis)
* [svn r58316] (iko, pedronis)
* [svn r58311] remove dead code
* [svn r58309] fix a test, add a comment (pysvnwcrevert cannot be invoked generically)
* [svn r58308] \* de-generalize conditional skips and only care   nicely for common tedious causes of skipping:   import a module and checking it has a certain   version.  usage example:
* [svn r58300] adding option to specify namespace for ifraises execution
* [svn r58298] move and update TODO file
* [svn r58297] add a new way of conditionally skipping a test:
* [svn r58288] update TODOs, docstring
* [svn r58287] add a test for \_\_version\_\_ attr
* [svn r58286] also contain URL
* [svn r58285] add docstring and version info
* [svn r58284] make sure that "test" prefix is enough
* [svn r58225] remove outdated comment
* [svn r58224] remove redundant py import, thanks getxsick
* [svn r58211] allow to specify filenames that need to be preserved
* [svn r58190] adding arigo/hacks/svnwcrevert as a py.svnwcrevert script on py lib trunk
* [svn r58113] typo
* [svn r58112] helper to add the required option
* [svn r58111] make the tests runnable also under pypy
* [svn r58108] A hyperthreaded cpu core should only count as 1, although it is present as 2 entries in /proc/cpuinfo.  Counting it as 2 is misleading because it is \*by far\* not as efficient as two independent cores
* [svn r58107] first step in moving experimental filelog session into contrib
* [svn r58095] (pedronis, arigo)
* [svn r58094] (pedronis, arigo) Partial revert: don't print traceback failures after InternalExceptions
* [svn r58093] (pedronis, arigo) Show tracebacks for the failures even when py.test is interrupted by Ctrl-C
* [svn r58092] Ensure that the path is printed before the 1st test of a module starts running (even when -v is not specified)
* [svn r58091] Minor refactoring of a few methods.  The point is to add a couple of hooks for my conftest.  This is a bit hard to test, and I don't even know if it's worth a comment, as I'm ready to fix my conftest when needed
* [svn r58090] Add a hack with a comment that explains why I think it is a hack. I fear I don't care enough about --tb=short to do the "proper" fix as documented in the comment
* [svn r58037] adding small sysinfo script for retrieving remote host information through execnet
* [svn r58035] allow for passing in ssh\_config files, deprecat "identity" argument
* [svn r58034] introduce APIWARN helper allow for subscriptions to produced warnings use it from some places
* [svn r58030] check for different output according to cpython version
* [svn r58028] reverting 58009 - webcheck is actually used although not directly tested
* [svn r58027] skip specific test instead of all pexpect ones
* [svn r58026] avoiding an issue with older pexpect versions
* [svn r58021] regen setup
* [svn r58020] streamline pdb pexpect test
* [svn r58017] fixing two more glitches
* [svn r58013] bah, forgot that helper uses fnmatch-style matching, not regular expression
* [svn r58012] fix imports so that tests pass against pypy-dist
* [svn r58010] some support for old naming because many sites import ReST checker from this conftest
* [svn r58009] move old webcheck things into contrib
* [svn r57979] fix 2.3 compat
* [svn r57977] (iko, pedronis)
* [svn r57971] (iko, pedronis)
* [svn r57834] try to accomodate the fact that some conftests like the Prolog-test ones subclass Module but do not actually have an underlying python file. Nowadays, they should subclass "py.test.collect.File"
* [svn r57833] simplify test setup
* [svn r57831] small test cleanups while prepping for pycon uk slides
* [svn r57790] Reintroduce the --tb option.  Add test
* [svn r57772] tweak
* [svn r57771] adding Guido's py\_unittest unittest TestCase collector, adapting to py/trunk API and simplifing it a bit
* [svn r57762] \* settle on using suptest for supporting test runs \* fix a bug in getmodpath \* remove redundant functions/files
* [svn r57758] unify test support, remove basetest.py
* [svn r57757] adapt output checking test for slightly different output
* [svn r57756] \* create new contrib directory \* move py/green to contrib/pygreen,   fix tests and code to pass
* [svn r57755] regen setup, readme for trunk
* [svn r57754] \* introduce py.test.collect.File (File for py or non-py files) \* introduce py.test.collect.Collector.collect\_by\_name and   special case it for Directories to allow specifying   files that would otherwise be ignored because of filters. \* fix py/doc/conftest to work with new API \* refactor py/doc/test\_conftest.py to use suptest helper \* avoid old APIs in some more places
* [svn r57744] Merging collection-API refactoring branch: \* collectors now have a collect() method \* items have a runtest() method \* deprecated old run/join of collectors and   run/execute methods of items
* [svn r57707] review todo.txt and future/\* items, merge into one TODO.txt file
* [svn r57689] mentioning "-U" switch
* [svn r57609] Keep comments in the same style accross the file
* [svn r57595] i've heart somewhere that running tests before checking in is a good idea (aka bah!)
* [svn r57594] \* adding tracing to dsession and master/slave communication   (enable with --tracedir) \* factor slave loop into a class \* add comment to pickling
* [svn r57591] flush true by default
* [svn r57587] merging doc and tool changes from release branch svn merge -r 57540:HEAD release/0.9.x
* [svn r57565] also introduce config.maketrace(name, flush=False) which returns either a nulltracer or opens a log in the tracedir and returns an object that you can call with args to print into the file
* [svn r57564] adding an option for setting a tracedirectory so that components can write log files, depending on what they get from config.gettracedir()
* [svn r57558] should have come with 57557 - extend "acceptance" test machinery
* [svn r57557] add lookup in filenames and test machinery
* [svn r57556] remove special Function \_\_repr\_\_
* [svn r57554] fix case where argument to AssertionError has broken \_\_repr\_\_
* [svn r57552] regen setup.py
* [svn r57551] normalizing py.which cmdline script
* [svn r57550] \* fixing syntax error \* removing setup.cfg
* [svn r57548] \* introduce py.process.ForkedFunc object (previously lived at   py.io and before that py.test) \* avoid deprecated py/misc/terminal\_helper.py \* starting changelog for 1.0
* [svn r57546] \* channels now also provide makefile(mode) with mode = 'r'   for providing file-like read/readline/close methods. \* added and refined crash and finalization tests
* [svn r57544] still notify callbacks if we can't send to the other side anymore
* [svn r57540] merge changes from release branch back [svn merge -r 57430:HEAD ../release/0.9.x/ .]
* [svn r57538] re-apply few fixes to make experimental gateway path basically work again
* [svn r57536] get the py.path fixes of 0.9.x release branch back to trunk
* [svn r57535] extending hacks for a safe representation of objects to also work for broken \_\_repr\_\_s on newstyle classes
* [svn r57529] preparing experiments with "dev" tags and pypi
* [svn r57481] fix a few things, seems to work again
* [svn r57480] fix changed namespace
* [svn r57474] Fixed bug reported by Martijn Faassen - when the XML output of 'svn st --xml' didn't contain author information, the code raised an exception
* [svn r57468] regen setup for trunk
* [svn r57464] removing print statement (there is something to be said about doctests :)
* [svn r57455] fixing for changed output
* [svn r57449] Merging https://codespeak.net/svn/py/branch/guido-svnwc-xml-status with trunk (revisions 56843:57448)
* [svn r57445] retain the behaviour of printing if we insert to sys.path
* [svn r57442] robustify. this can also happen to get exectued while being in another version of the py lib
* [svn r57440] various fixes for python2.6
* [svn r57439] account for changed location
* [svn r57429] merging/porting the release branch changes and cleanups to trunk
* [svn r57358] move more scripts to build directory
* [svn r57346] shift talk from py/doc to extradoc/talk dir
* [svn r57331] port 57330 (which accidentally went to the branch)
* [svn r57325] for noew skip apigen tests that fail
* [svn r57324] remove superflous helper attribute
* [svn r57323] port 57174 from branch: ignore errors when removing dirs in teardowns
* [svn r57322] port 57306 from branch: check for actually needed binaries
* [svn r57321] merging the event branch: \* moving in test, misc, code, io directories and   py/\_\_init\_\_.py \* py/bin/\_find.py does not print to stderr anymore \* a few fixes to conftest files in other dirs some more fixes and adjustments pending
* [svn r57320] remove test, misc, doc, io, and code directories that are to come from the event branch with the next commit
* [svn r57228] fix always-true assertion
* [svn r56835] Refactored the wcpath.status() method, moved the parsing of the regexp to the WCStatus class
* [svn r56749] porting fixes from dist to trunk
* [svn r56743] port path-fix 56724
* [svn r56722] porting the windows-execnet 56717 and 56718 fixes to other active branches
* [svn r56721] some more upcoming notes
* [svn r56712] add a note about the needed FILE:LINENO summary
* [svn r56663] some more planning and notes for 1.0
* [svn r56636] note regarding collection/runtest ambiguity
* [svn r56635] a few notes for release planning
* [svn r56619] sic
* [svn r56618] skip this test for a more explicit reason
* [svn r56513] extend doc string
* [svn r56512] avoid infinite recursion issue with svn 1.5, thanks chris lamb
* [svn r56510] svn 1.3 is in common use these days and seems to pass the test
* [svn r56509] small helper script for creating py/test2 hierarchy
* [svn r56304] almost any version works
* [svn r56286] note that --showlocals doesn't work on trunk
* [svn r56285] sanitize doctesting a bit.  this also fixes the problem that you could not have "import" working in doctests previously.  thanks nshepperd
* [svn r56284] removing unadvertised new names and adjusting the test to show how you can put custom messages otherwise
* [svn r56231] add a small script that tells where an import of a python module package would come from. py.which is meant to be the equivalent of "which" in unix
* [svn r55053] also allow using pdb in failures of the collection process
* [svn r54623] A test for 54622 Fix for running tests
* [svn r54622] "How this could ever worked..."
* [svn r52707] Allow some mocking
* [svn r52483] Skipping apigen test again
* [svn r52481] Python2.3 has no 'sorted()' yet
* [svn r52001] Removed (outdated, and perhaps a bit too verbose) document about svn auth, in favour of a short note and code example in the existing path.txt doc in docs
* [svn r52000] Merging the 'guido-auth-svn' branch back into the trunk. This means there's a new class py.path.SvnAuth of which instances store user credentials and auth config, and can be passed to py.path.svnurl and py.path.svnwc objects to control SVN authentication behaviour
* [svn r51329] rename special \_\_package\_\_ to \_\_pkg\_\_ because python 2.6 needs the former (thanks to Ralf Schmitt)
* [svn r51292] Just another except, in order to represent source when it's not there
* [svn r51285] Fix \_docgen.py documentation building
* [svn r51132] a todo/note on skip's and speed (inspired by exarkun on #pypy)
* [svn r51087] add a note that skip-reporting is broken after the branch-merge
* [svn r51077] bump version number
* [svn r51058] fix ReST
* [svn r51057] review, refactor and write up for some 1.0 related TODOs
* [svn r51056] fijal: don't use py.test.skip directly in a conftest.py - this will get called very early during import time. This lead to windows not executing any py/ tests at all
* [svn r51034] merging in fijal's reporter-merge branch into the trunk, still needs refactoring as far as i am concernced
* [svn r51033] merging of svn+ssh://codespeak.net/svn/py/branch/reporter-merge/py/test/session.py revisions 47584 to 51032:
* [svn r51029] make sure the artificial GeneratorExit claims to come from the exceptions module
* [svn r50755] Port from reporter-merge branch
* [svn r50645] Fix for "utestconvert.py < input"
* [svn r50606] don't fail for links to html files where we don'T have a .txt
* [svn r50576] This patch by Thiemo Seufer adds support for mips and mipsel
* [svn r49974] Restore the following behavior: py.test exits with an exit code 1 when there are failures
* [svn r49423] (naro) Fix for list.append() obviously, has no single test
* [svn r48347] make status on working copy objects deal with replaced files
* [svn r47584] Implement very sophisticated algorith for -k TestClass.test to work
* [svn r47277] Argh. Fix -d -x
* [svn r46794] embarassing typo
* [svn r46772] add a custom pdb.Pdb subclass that has the ability to properly list the lines generated by py.code.Source. Very useful in PyPy debugging
* [svn r46771] remove nonsense \_\_init\_\_.py line -- belongs to the next checkin
* [svn r46770] add py.builtin.GeneratorExit, to make it possible to write 2.3 and 2.4 compatible code. The exception is never raised there
* [svn r46692] Added 'export()' method to py.path.svnurl
* [svn r46011] Branch merge of session cleanups
* [svn r46010] Merge from branch - a lock for files, prevents segfaults of cpython
* [svn r45994] Adding support for locking and non-recursive commits (a somewhat obscure feature, but was easy to implement) to svnwc. Not tested on SVN 1.2 (anyone?)
* [svn r45907] Fix number
* [svn r45901] Do not use os.tmpfile() as it will fail on Windows unless You are Administrator
* [svn r45671] Change lines for code.txt not to fail on cpy2.5 Problem is described in tests (skipped) anyway, hence there is no point in failing code snippet in documentation
* [svn r45655] Kill unnecessary sanity check
* [svn r45649] Two possible skip reasons
* [svn r45648] Add possibility to specify reason for skips
* [svn r45647] use repr() rather than direct return
* [svn r45646] Rewrite it a bit not to rely on exact formatting (rather checking whether information is there)
* [svn r45550] Skip those tests on windows
* [svn r45549] fix a bug that servemain() wouldn't actually try to join threads at exit because of a (hidden) attribute error
* [svn r45548] - refactoring cleanup mechanics into its own class - have setDaemon(1) for the receiverThread as   otherwise on python2.5 atexit will not be invoked   (the receiver thread apparently blocks it)
* [svn r45547] remove global
* [svn r45545] Kill lock
* [svn r45541] don't use globals when we might be called from destructors
* [svn r45539] merge the execnet lessthreads branch (using the branch'es history):
* [svn r45538] not exported
* [svn r45535] Avoid hacks when creating exception name
* [svn r45525] be a bit more impersonal
* [svn r45520] Add some rough ideas about pylib plannign
* [svn r45519] Ignore non-existing files
* [svn r45518] Print out communicate atomically
* [svn r45484] A comment why this will not work on different python version
* [svn r45483] Fix 2.5 issue when str(excinfo.type) produced different thing than expected
* [svn r45295] Fix rsession tests
* [svn r45294] Complain loudly when there is pypy without stackless
* [svn r44655] be consistent in super usage
* [svn r44648] PyCollectorMixin is intented to override Collector behavior, if we want to use super it really needs to subclass from it to get mros to that effect
* [svn r44248] Fixed problem with calling .remove() on wcpaths of non-versioned files
* [svn r43575] Whack CSS a bit to have dots which does not go under the navbar
* [svn r43299] Be more keyword-friendly for cheesshop
* [svn r42273] Added a file about what I would like to have in apigen
* [svn r42215] Copy with SVN history and restore test
* [svn r42194] Add my notes (this time for real)
* [svn r42193] revert my changes
* [svn r42191] Move several issues from merlinux tracker
* [svn r42188] Revive test/compat.py again, seems to work fine for me. Can be used by patching sys.modules["unittest"] in the conftest.py file
* [svn r42103] added some issues/features previously tracked elsewhere
* [svn r42092] add rlcompleter2 / pdbplus issues
* [svn r42089] some more infos/issues put to the TODO file
* [svn r42085] adding some issues from the merlinux tracker (and removing them there)
* [svn r41860] \* kill start\_on \* add keyword\_oneshot flag, which indicates that -k is only one shot   than all tests are run \* simplify code a bit
* [svn r41855] Add a bit hackish option which allows to start from selected keyword test, doesn't work with rsession yet
* [svn r41655] Added list of possible exceptions for callables
* [svn r41620] Re-added anchors to the (full) source files, and using them from the API documentation
* [svn r41606] Rather strange hack to make sure that queue is not empty. Needed nowadays for passing test\_session in js/examples/console
* [svn r41480] Move the FunctionMixin to collect.py, as an attempt to avoid circular imports
* [svn r41224] Added failing test, which I would suppose to pass
* [svn r41080] Added support for filtering listdir() calls in the SourcePageBuilder, using it to filter out hidden files and the 'build' subdir of greenlet
* [svn r40982] Possible problem-point (exploded at least once there)
* [svn r40943] Regenerate js
* [svn r40834] skip test for XXX'ed download urls if version contains "alpha"
* [svn r40832] fixing syntax error
* [svn r40831] bump version number, fixes
* [svn r40753] Make MAX\_COUNTER small enough. It's an obscure hack, but at least works
* [svn r40739] Make FAILED TO LOAD MODULE clickable
* [svn r40738] Count also FAILED TO LOAD MODULE failures in web interface
* [svn r40737] A fix that allows (for command line reporter) to report failure in case when FAILED TO LOAD MODULE
* [svn r40720] Tracing defaults to false
* [svn r40719] Argh. There should be post-commit hook which warns about every import pdb in your code
* [svn r40706] test/\_\_init\_\_.py is needed as well (for determining test paths etc.)
* [svn r40705] move greenlet support objects to the "py/green" directory
* [svn r40702] (pedronis, arigo)
* [svn r40002] Re-enable url check test
* [svn r40001] Encoding (let's suppose UTF-8 will not hurt)
* [svn r39995] Possibly controversial checkin. Kill a hack for explicit isinstance(). We've got just two methods, so doing full getmro doesn't seem to make too much sense
* [svn r39994] fix rest
* [svn r39983] (arigo, fijal) -
* [svn r39982] introduce gateway.\_send and have all places route their sending of Messages (or None's) through that method
* [svn r39981] Some random notes about future pylib direction
* [svn r39975] Fix imports
* [svn r39974] Add some comments
* [svn r39973] Add a (stolen almost directly from arigo/hack) network layer based on top of greenlets. Needs some tweaking, so it's not exposed as a py.xxx, but rather py.\_\_.net
* [svn r39655] add a (somewhat clunky) way to kill processes by PID both on windows and unix-ish systems
* [svn r39340] A fix and a test. Thanks Marian Shubert for reporting problem
* [svn r39106] Plus sign works fine for me, added to ALLOWED\_CHARS in svncommon
* [svn r38969] Fixed ReST
* [svn r38967] spelling mistake
* [svn r38840] fix links to release branch
* [svn r38831] be more careful to not work with svnwc() paths when checking for mtimes
* [svn r38828] Try harder to get the logic right to avoid regenerating html files for unmodified txts
* [svn r38824] skip these tests on win32
* [svn r38821] hum, only do the precompiled greenlet if we are invoked with install
* [svn r38819] skip apigen on win32
* [svn r38818] add release 0.9.0 announcement link to index.txt
* [svn r38817] fix typo, make the promise less definite (it's ages old)
* [svn r38816] tried to fix sentences and singular/plural
* [svn r38815] Be at least consistent within one file
* [svn r38812] remove the exe for now and warn about missing greenlet support for win32
* [svn r38810] being yet more ignorant about build problems of c-extensions
* [svn r38808] trying to be more ignorant about missing abilities to remove c-extension modules (they might be in site-packages) and adding a hack to make sure that the greenlet module gets build ahead of the setup install
* [svn r38805] be more careful about win32 supporting claims
* [svn r38801] fixing the download page for prospected links
* [svn r38799] bumping version to 0.9.0 (trying :)
* [svn r38797] fixing the too global revision caching
* [svn r38782] here as well
* [svn r38781] the target docpath needs to be determined per project
* [svn r38779] Made that the package revision is now shown in the title rather than each file's individual revision
* [svn r38777] Fixed api and source link role relative paths
* [svn r38776] Removed duplicate links
* [svn r38775] use relative paths, and compute them late
* [svn r38773] Using ensure() better (thanks, hpk ;)
* [svn r38772] Make sure target dir exists before writing files to it
* [svn r38770] Removed option --apigen\_relpath, added options --docpath and --apigenpath, both of which are not only used for building the navigation, but also to actually tell py.test where docs should be generated. Also cleaned up and fixed the situation regarding relative links in the navigation menu and stylesheet/js paths
* [svn r38766] naming consistency
* [svn r38763] overriding the new a\_docref and a\_apigenref Page helper methods, getting rid of update\_menubar\_links
* [svn r38762] slight sanitizing and allow to change sep in one place
* [svn r38761] split Page methods for menubar generation allowing to override A link generation for docs and for apigen pages
* [svn r38760] fix ehem syntax ehem error
* [svn r38755] fix reporting of duplicate remote host rsyncs and adding an XXX test (not easy to test ssh hosts because it modifies remote state)
* [svn r38750] mention command line completion
* [svn r38749] Regenerated js
* [svn r38747] hum, a last-minute namespace change: i missed that fijal didn't move py.test.Item|Function to the py.test.collect namespace at the time.  a pypy fix will follow shortly after i merged the dist
* [svn r38745] striking another XXX, mentioning limited windows path support
* [svn r38744] \* adding a section about conftest.py files (removing an XXX) \* removing dead links
* [svn r38743] fix and remove unneeded external references remote test-distributed.txt which had lots of not actual links and IMO would need a higher-level introduction to become understandable
* [svn r38741] doing r38739 in the right place (dist will get recopied from trunk)
* [svn r38726] shuffle -> sample
* [svn r38713] Use self.source wherever applicable
* [svn r38712] Avoid py.test.config.\_reparse([])
* [svn r38711] Avoid infinte recursion when rsyncing (well finite - OSError file too long)
* [svn r38710] Invoke the skip rather than have it
* [svn r38709] kill dead code
* [svn r38708] document.location is not in DOM. We need to implemented some stuff in js backend to be able to test this
* [svn r38706] privatizing config.conftest to config.\_conftest as well
* [svn r38701] Fixed line heights in Konqueror
* [svn r38700] Fixed paths, docs are now no longer placed in a sub dir
* [svn r38685] Partial rewrite of fijal's pdb/apigen text
* [svn r38683] Add some information about pdb + apigen
* [svn r38659] mention that we think that py lib works on 2.3, 2.4 and 2.5
* [svn r38653] Skipping some tests on win32, for some reason svnwc.info().created\_rev seems to fail (need to look into it in more detail later)
* [svn r38652] Fixed rendering problems in IE, updated todo.txt
* [svn r38650] Added env var for creating relative links to the py lib docs (in the nav bar), added script to generate py lib's docs and api docs in a subdir in one go
* [svn r38649] Adding \_findpy call to not have to set PYTHONPATH to find the py lib
* [svn r38626] some XXX to TODO, and adding a part about proper linking
* [svn r38621] \* fix for linking problems: move future.txt one level up \* be slightly more defensive about apigen\_relpath
* [svn r38601] pointing to the just uploaded py-0.9.0-beta packages
* [svn r38598] fixing download url and fix makepyrelease script
* [svn r38597] Raise an ImportError if there is no way to build a C extension module, e.g. on top of pypy-c.  Importing e.g. py.magic.greenlet then gives an exception that is saner to catch than ValueError
* [svn r38595] streamlining info a bit
* [svn r38588] or maybe this version for google? :)
* [svn r38586] use merlinux.eu (did you know it exists? :)
* [svn r38585] remove the gmbh
* [svn r38584] cleaning up menu bar: "index" instead of doc, removing pseudo "home" page
* [svn r38582] refining the release announcement/finalizing from my site fijal, cfbolz, guido\_w: are you ok with it?  (your name is under the release announcement)
* [svn r38579] library code should not use magic timeouts, testing code may use a global TESTTIMEOUT
* [svn r38578] Made that the docstring isn't replaced by the function info anymore, but the func info is appended, some small style issues
* [svn r38577] Test for explicit Failed
* [svn r38576] typo!
* [svn r38574] Special case outcome.Failed - reuse that excinfo
* [svn r38573] This tests was completely broken when pypy is there
* [svn r38572] Move a bit into using high-level traceback interface instead of low-level one
* [svn r38571] Using wcpath.info().created\_rev (last changed revision) rather than wcpath.status().rev (current wc rev) as revision, added revision for source and namespace pages
* [svn r38568] improving r38560 a bit
* [svn r38560] Compare collectors by id rather than by name. It needs to be rethought a bit
* [svn r38559] Give info about tests in case of failure/interruption
* [svn r38549] also privatizing Config.initdirect (which relates to remote reprs and such)
* [svn r38548] carefully privatizing Config.make\_repr and Config.merge\_repr by doing a whole-pylib replace and reviewing the diff in detail
* [svn r38546] Typo
* [svn r38542] Fixed help string for --apigenrelpath
* [svn r38534] some more little issues
* [svn r38533] adding a way to modify the "apigen relative path" from the command line, unifying conftest and confrest
* [svn r38530] fix
* [svn r38524] add link to pyfuse (which is the newer approach, I think) and fix the link target
* [svn r38518] fix references
* [svn r38516] majorly refactor future chapter, mentioning APIgen and other more current ideas
* [svn r38514] rewrite/refactor why\_py (it was quite old)
* [svn r38508] skip signal detection tests for python < 2.4 (2.4 and 2.5 work apparently)
* [svn r38506] config.getvalue\_pathlist should also honour option values (to be uniform with getvalue semantics)
* [svn r38505] clarify that config.getvalue will first lookup in options
* [svn r38504] fixes for python2.3
* [svn r38496] don't print the test-mode, that needs more thought how to do it now
* [svn r38484] make config.option override any conftest provided value
* [svn r38480] unneeded import
* [svn r38474] make io capturing configurable per e.g
* [svn r38473] rewrote the future section of py test
* [svn r38472] clarifications regarding dist testing
* [svn r38470] try to have py.process.cmdexec run on top of PyPy (with its currentl limitations)
* [svn r38467] make string comparison lowercase-insensitive for windows
* [svn r38460] removing a very-likely-obsolete test and otherwise fixing files and tests to do py.test.skip()s only during setup
* [svn r38459] two functions with the same name, use the latter one
* [svn r38458] better docstring for top level object
* [svn r38456] konqueror issues
* [svn r38455] various merging of startup documentation and rewriting/refactoring information and references.  clarifying release-structure a bit
* [svn r38452] bumping version number to 0.9.0-beta
* [svn r38451] a draft release announcement
* [svn r38449] removed DONE issues, added one more (sources page) and comments
* [svn r38448] shorter title
* [svn r38447] privatize internal regular expressions
* [svn r38441] increase hard-coded timeouts a bit (waitclose() will not neccessarily wait that long, it returns as soon as the channel is closed so it's not as bad as outright sleeping)
* [svn r38440] Made that the blue links are colored from the stylesheet
* [svn r38439] Fixed a list of things suggested by hpk: changed method order in class pages, changed page titles, added links to the api and source index from the nav bar (also in py/doc html), changed function views, made it possible to remove an item from the navigation, changed header 'properties' to 'class attributes and properties', removed duplicate stack traces (in a somewhat unsatisfying way, needs revisiting later I think)
* [svn r38438] Rolling back r38422 and part of r38397 because it made that a lot of stuff was printed to stdout/err
* [svn r38427] significantly reducing the trailing "Exception in Thread" messages by having threads die more silently.  Most of these threads and tracebacks are anyway on some remote machine, and we need to take care that errors are routed to the place of invocation (e.g. the remote\_exec() invoker should get remote exceptions, if the remote thread goes out of scope, the resulting exception will not be seen likely, anyway)
* [svn r38422] Argh. Forgotten to check that in
* [svn r38408] my current list of things (not totally sorted in prio order) that i'd like for 0.9
* [svn r38398] for setup/teardown pairs the correct idiom is
* [svn r38397] Kill PidInfo and make boxing optional for distributed testing. This should make dist testing on windows possible
* [svn r38396] Capturing stdout and stderr when running the update\_website tests to not mess up --rest output
* [svn r38393] Made (displayed) paths to source files relative whenever possible
* [svn r38392] looponfailing tricked me into committing too early (syntax error, bah)
* [svn r38391] check more exactly if we are working from a default rsync\_root (checking for rsync\_root == topdir is not safe as the topdir may be explicitely contained in the rsync\_root list)
* [svn r38390] refining the logic: for inplace-localhosts ("optimized localhost") we do not add any relative path
* [svn r38389] completing the picture: now if you don't have rsync\_roots specified, the config.topdir is transfered but it is transferred to the "remotepath.join(topdir.basename)" (not actual code) to avoid random such rsyncs to destroy/affect remote filesystem state
* [svn r38388] Removed stuff that's done, added a couple more items to do
* [svn r38386] Fixed some more nasty problems after running py.test --apigen on pypy: problem recognizing """foo "bar" baz""" as a single line string (hack unfortunately), unicode problems. Also added some code to debug the tokenizer: you can now run the color.py script with a Python file as arg to tokenize the file
* [svn r38378] snapshot commit for: each rsync root is now copied to remote-topdir.join(root.basename)
* [svn r38370] Hrmph, probably the nastiest case of 'debug print' I ran into so far, bringing apigen almost to a crawl... Also small optimization making that 'normal' multi-line strings are resolved before those using \
* [svn r38362] Fixed stupidity in the last checkin: only matching multi-line strings using slashes if the slash is actually at the end of the line... :(
* [svn r38352] made localhost inplace handling safer (and more redundant, there is an additional flag now, and host.gw\_remotepath is None for localhost-inplace hosts)
* [svn r38315] fixing failing apigen test and adding a string equality assertion helper (if we need to have such exact string-comparison tests, then let's at least make it convenient to discover the problem quickly)
* [svn r38310] assertion to be sure about keywords
* [svn r38302] Kill \_dist\_import\_pypy
* [svn r38262] Fixed support for tokenizing multi-line strings that use \ to span lines
* [svn r38249] It seems that there was somehow different idea in mind, but I actually do not get it
* [svn r38246] Fixed problems with line height in sources, added missing commas to seperate base classes, added all Python (2.5) keywords
* [svn r38221] Reverting previous changes after a discussion with hpk
* [svn r38220] moved -f and --exec in the documentation as well
* [svn r38219] looponfailing and exec are not so experimental
* [svn r38218] i give up for now, somehow the local change in the FileBox affects --dist testing in ways i can't fully understand at the moment (i would think it only affects --boxed test runs). So we are back to share tempdirs when running boxed (causing failures elswhere).  We care later, i guess
* [svn r38217] Moved the -S option to a new conftest.py in the execnet dir to avoid nasty option re-definition problems
* [svn r38216] Adjusting the conftest-for-socketgateway example so it takes care of the -S option (like the default one does), this caused some test failure
* [svn r38215] Gotta love the XHTML validity checks... ;)
* [svn r38214] be a bit nicer: don't pollute /tmp with a new tempdir for each boxed test run.  Also fix the counting of boxes (how did it ever work like this, anyway)
* [svn r38213] Some small cleanups, made fonts a bit smaller and more consistent
* [svn r38212] have each boxed test run use its own tempdir
* [svn r38208] make test safer against race conditiation (dist testing)
* [svn r38207] revert 38182 and 38184 changes to ensuretemp() because getpid() is different per thread (and if e.g. run via execnet in threads ... it creates new tempdirectories all the time) let's consider this sometime else
* [svn r38203] rename getpymodule/getpycodeobj to "\_" methods (which can build C modules on the fly) it's not clear they are still useful this way and they are easy to confuse with pyimport()
* [svn r38201] fixing the windows conftest example to work with the new rsync interface
* [svn r38198] almost cosmetic
* [svn r38197] slight reporting fix
* [svn r38196] reducing the callback-indirections for rsyncing and improving the reporting for localhosts non-rsyncs
* [svn r38192] streamlining localhost optimization handling, and simplifying the test a bit
* [svn r38184] avoid globals and add a comment
* [svn r38183] Made that py.execnet.Channel (referring to py.\_\_.execnet.channel.Channel) is picked up by the apigen doc generator, fixed some bugs that made that that didn't work previously
* [svn r38182] Made that ensuretemp() creates a new dir after forks
* [svn r38181] delete is now an option per-host
* [svn r38179] Allowing to rsync also when tests are failing using a --ignorefail switch, and added a --help switch that shows a short help message
* [svn r38176] make delete a per-host option (internally it is anyway)
* [svn r38175] Fix
* [svn r38174] Flag instead of new method and a test
* [svn r38173] \* Minor semantics change, now host and host: are the same (we don't   want to rsync to home dir usually) \* Make a flag rsync\_flag in hostinfo which tells whether to rsync   or no, semantics are that in case of localhost we do not rsync   and in case of localhost: we do
* [svn r38172] Add a method which sends only if there are available targets
* [svn r38165] Update this as well
* [svn r38164] Move source as an rsync initialisation argument rather than as a send parameter
* [svn r38151] Small change in the text to make it a bit more understandable
* [svn r38150] Fixed test that failed after the last change in apigen (removing the stacks from the callable info boxes)
* [svn r38148] Made that stacks are built on seperate pages instead of inline in the function information, to avoid having > 5MB pages..
* [svn r38144] Fixed a problem with the linker which on building api/source docs found matches in its own source code, moved some elements in the apigen docs out of a header where they shouldn't have been (making the font too large)
* [svn r38143] Removed some confusing output to stdout/stderr (output of failing tests, which are part of a test that tests whether a set of tests ran by a nested py.test process fails... so the failing output is correct)
* [svn r38142] Fixed problem with indentation in the results of a doctest
* [svn r38140] Empty literal blocks are not allowed: removing them
* [svn r38125] run unboxed by default, split/improve tests for it
* [svn r38124] No, this is needed for C-c to work
* [svn r38123] skip path.txt svn tests if not run via --checkremote
* [svn r38122] fix parsing of extra lines (to be executed)
* [svn r38121] Same issue. Race conditions in file write. Probably we should not use the same file output for multiple tests (or to put data directory outside pylib itself)
* [svn r38120] Make sure that those test run on same machine when run. (This is broader issue - we probably need to do some exclusive lock on files given to process\_rest\_file and such)
* [svn r38119] avoid clashing the sample\* prefix which is used by a listdir() test
* [svn r38118] there is no gurantee anymore that tests run in order
* [svn r38117] call "python py.test" instead of "py.test" (does not rely on +x flag)
* [svn r38116] fixing svn tests for -d runs (which usually is a dependency problem)
* [svn r38115] Call python directly
* [svn r38114] Kill fatals() to be considered in future
* [svn r38113] Make fatal() attribute which makes boxed tests catch SystemExit, while in-process let it slide
* [svn r38112] This time don't catch SystemExit. Increases stability
* [svn r38111] remove test order dependency
* [svn r38110] i don't think one needs to use svnwc's anymore, at least it works for me, i think
* [svn r38109] bailing out the optimize\_localhost approach to consider it differently
* [svn r38108] skip this test if we have no .svn dir
* [svn r38107] \* Have optimise\_localhost flag and a test for it (but not turning it on   by default, this will be done later) \* two other occurences of possibly-failing channel.send
* [svn r38106] making tests independent of --dist implications
* [svn r38105] Make errors bright red and bold, easier to spot
* [svn r38104] fixing this so "py.test -s" passes on this as well
* [svn r38103] Hum. This test was just broken. Fix it and add additional checks
* [svn r38102] Skip this test not to interfere with py.test.config.boxed
* [svn r38101] oops, this will be implied anyway
* [svn r38100] Slave is always boxed as well (different level of skips)
* [svn r38099] \* RSession cannot work with nocapture, LSession can \* RSession is always boxed
* [svn r38098] A test and typo discovered by that test
* [svn r38097] streamline boxed test configuration: "config.option.boxed"  indicates now boxed tests and RSession.fixoptions takes care to honour dist\_boxed accordingly.  So you can do
* [svn r38096] typo
* [svn r38095] getremoteerror does not raise!
* [svn r38091] Increase verbosity in case of error. The thing is that we usually don't get remote error, so let's print it
* [svn r38083] saner tempdir using
* [svn r38082] Fix reporting (and a test for that)
* [svn r38079] Typo in some help string, and made fonts of apigen a bit smaller somewhere
* [svn r38078] some fixes and test skips for win32 (otherwise all tests pass for me on win32)
* [svn r38077] Script to run "py.test --apigen" on the py lib (well, can be used on other projects too) and rsync the results to some remote host/path (by default codespeak.net, currently to some dir in my home directory, when it's tested better in practice files will go to the website directory)
* [svn r38076] Typo
* [svn r38075] Adding some print
* [svn r38074] fix typo
* [svn r38070] Removed broken anchor from link
* [svn r38068] Grmbl, the failing svn tests turned out to be my fault: copy/paste bug. (when copying the comment, I must have copied along some lines of code... ):
* [svn r38051] Fix an option -s contradics -d
* [svn r38044] Now just writing dots when running tests, decreased font size a little of the class docstring and link
* [svn r38043] Changed the way the lazy linking is handled: instead of building the pages and replacing the links all in memory, pages are now written to disk with temporary hrefs, which are replaced afterwards
* [svn r38040] My favorite lazy-lexical-scope-binding error. Still thinking how to make test for that. Basically - callbacks with lexical scope variables are usually (depends of coz) called with variables after the for loop has finished
* [svn r38039] Fix a test
* [svn r38038] more precise rsyncing
* [svn r38037] not showing rsync-items by default, shows that something is odd regarding the messages, i think (read the diff, it's a simple change, i guess we actually want to generate an Event also for the web reporter at some point, i think, but we could live with this for the merge IMO)
* [svn r38036] Some small code cleanups (moved more HTML generation code to html.py)
* [svn r38035] be more precise when reporting about hosts (enumerate them)
* [svn r38034] i think the rsync roots should be shown absolute
* [svn r38033] Fix tests once again
* [svn r38032] simplified testing machinery a bit (you know just have to add an example and can immediately use it from tests without adding boilerplate anywhere)
* [svn r38031] refactor rest of tests
* [svn r38030] Refactor this test to use new testing service
* [svn r38029] Move funcxxxspec out of pylib itself
* [svn r38028] make rsync configurable regarding its output (verbose option and overridable method)
* [svn r38027] more precise reporting to the user
* [svn r38025] Fix a bit web reporter (still not perfect)
* [svn r38024] avoid that hostmanage.py tests interfere with the real world (through writing or creating pytestcache-XXX files), semantic change: upon gateway initialization the remote path is not automatically created (this will be done later by rsync anyway)
* [svn r38023] Report HostRSyncing only in case where rsyncing is really performed
* [svn r38022] be a bit more verbose before rsync
* [svn r38021] Simple refactoring to be a bit more verbose when starting
* [svn r38005] Made that source snippets are now shown in two tables, to allow selecting the source without the line numbers, decreased font size a bit, made that the Page class can be passed in from the outside (build() method) to more easily allow customization of the layout and pages
* [svn r38000] Fix web reporter
* [svn r37994] Imrpovement of host rsync reporting. Additionally it increased stability for no particular reason
* [svn r37992] have real definitions (yes, the empty lines make a difference). rest is strange
* [svn r37991] remove colons and change the subsecion title
* [svn r37990] Fix reporting (and a test for that)
* [svn r37989] add an overview over py.tests command-line options
* [svn r37988] adding a bit to the test (but the code is already correct)
* [svn r37987] split the py.test docs into implementation and customization docs and normal user docs
* [svn r37986] Add a repr if we have str Add delete keyword to parent
* [svn r37985] make sure that host.initgateway() will always have the other side chdir()ed to home, thus generalizing 37971 with respect to platform support and making it independent from gateway implementation details
* [svn r37984] Moved all the apigen information into one document and adjusted to match the current situation
* [svn r37971] This was there for a reason that we don't want to have different pylib loaded (there was even test for that, apparently killed)
* [svn r37960] refactored the tests and added tests and code for disallowing to send() twice without adding new targets
* [svn r37950] \* add and document dist\_rsync\_ignore option to ignore   files and directories for rsyncing
* [svn r37949] fix test for nocapturing output runs (and --pdb)
* [svn r37948] fix for win32
* [svn r37947] added \_gethomedir() helper to get at homedirectory
* [svn r37946] setup/teardown hosts correspond
* [svn r37945] make the hostmanager get at the hosts itself (or you can pass it in a custom list of hosts)
* [svn r37938] rename report.py to repevent.py (report events) also to ease the completion ambiguity a bit (and to avoid having to read "reporter" and "report" next to each other)
* [svn r37936] minor cleanups
* [svn r37935] remove done\_dict and according experimental code for re-scheduling (i guess) fijal: in the diff you'll find a XXX fijal, i changed the meaning of a test, wasn't sure about it. can you check?
* [svn r37931] Made that the \_\_init\_\_ method is the first method displayed
* [svn r37930] Docstrings
* [svn r37928] Updated test (oops)
* [svn r37927] Some docstrings
* [svn r37926] Small visual tweaks
* [svn r37922] Made that properties (class attributes) are displayed, fixed the links in the menubar
* [svn r37919] use a less ambigous test package name
* [svn r37916] dist\_rsync\_roots: py lib itself only needs itself
* [svn r37915] sanity checks
* [svn r37914] cleaner test setup
* [svn r37913] Some more cleanups in HTML generation, fixed support for docstrings in namespaces, in order to do this I had to change the way objects are retrieved: instead of getting them from the DSA I now walk the package tree, small change in apigen.py: to allow re-using the get\_documentable\_items function I split it up in a generic and a specific part
* [svn r37912] Made some small changes to the initpkg mechanism so docstrings are, if provided in the exportdefs, copied to the namespaces (from whereever they come), and added docstrings to all exposed namespaces (except for \_thread for now)
* [svn r37909] a much much much larger refactoring than i originally intended at this point:
* [svn r37908] fixing test
* [svn r37901] make\_numbered\_dir(): in a fork() situation, only the last process should remove the .lock, otherwise the other processes run the risk of seeing their temporary dir disappear.  For now we remove the .lock in the parent only (i.e. we assume that the children finish before the parent)
* [svn r37899] Try to print the exception before dying
* [svn r37898] provide a high-level helper for getting at a pathlist specified in a conftest (and the paths can be relative to the conftest.py file they are contained in)
* [svn r37886] hum, i thought i had refactored this option name already ... also the wrong name does not even cause a test failure. odd
* [svn r37885] adding docstrings, stripping non-used names from import
* [svn r37884] Few missing Skipped
* [svn r37881] Some other stuff goes private
* [svn r37880] removing public API from test.config (only one rather internal usage, anyway)
* [svn r37879] s/tryiter/\_tryiter/
* [svn r37877] Intermediate checkin for some privatising of attributes
* [svn r37876] s/get\_collector\_trail/\_get\_collector\_trail/
* [svn r37872] StdCaptureFD and StdCapture now try to take care of stdin in a mostly uniform way
* [svn r37866] small cleanup of ground std\* descriptors
* [svn r37865] more descriptive skip msg
* [svn r37864] Add a (failing) test about version URL available
* [svn r37863] Ooops, didn't meant to check that in
* [svn r37862] Move Skipped/Failed/Passed out of public namespace
* [svn r37861] Skip compat
* [svn r37859] separating tests into two files
* [svn r37855] a first rough list for py.test namespace cleanup
* [svn r37850] Split methods a bit to smaller parts
* [svn r37838] Privatize attributes of rsync.py
* [svn r37837] A bit of cleanup of the HTML generation: some of the layout related stuff is moved to the classes in the H namespace. This hopefully (when done, more can definitely be done later) seperates the HTML generation better from the presentation-related code
* [svn r37835] some updates to TODO items
* [svn r37834] reworked execnet intro to more explicitely state the main (high-level) features
* [svn r37833] Oops, was using py.path.local('.') instead of py.magic.autopath().dirpath() to find the scripts to copy..
* [svn r37831] improving more docstrings
* [svn r37830] better RSync docstring
* [svn r37824] fix typo
* [svn r37822] added lots of docstrings, general cleanup
* [svn r37815] path.txt: showing a bit more for string methods update TODO
* [svn r37812] Using style.css from the doc directory for apigen now, with some overrides in a seperate stylesheet. Cleaned up the stylesheet and script handling a bit: stylesheets and scripts are now defined in the LayoutPage class (where they're easily customizable). Also removed some old unused code and moved the H class (html related) stuff to its own file. The idea is that this will become the place where all the layout is defined
* [svn r37790] fix Ssh remote sanitzing of FD 1 and 0 and factor out the code in a function
* [svn r37789] small updates and streamlining of docs
* [svn r37788] guido, unless i am missing something there seem to be two problems with your making Channel's public: you didn't import it (py.\_\_.\* is not lazy) and it fails the apigen tests even after fixing it
* [svn r37787] more precision at shutdown: first remove from the sendqueue, then trigger the queue to shutdown. (This is an attempt to get rid of the still appearing "exception in thread" messages)
* [svn r37786] improve docstring
* [svn r37785] privatizing some more attributes
* [svn r37784] privatizing some attributes
* [svn r37783] puh, it took me quite a while to find out why ssh-stdout-confusion tests involving "os.write(1, ...)" were passing on earlier revisions: in 35888 a test name got fixed, but when you specify "-S somehost" then this will hang, so i am skipping the test for now, this never worked
* [svn r37782] Changed the checking in resolve\_linkrole() so that instead of the apigen results (the directory with HTML files) it uses the py lib object tree and source tree to find out whether links are valid
* [svn r37781] Added 'execnet.channel.Channel' object to the to-be-documented items
* [svn r37780] simplifying code a bit (but test\_confusion\* for Ssh still fails)
* [svn r37779] Fix and a test for disappearing files
* [svn r37776] Fixed some problems with the relative link checker (was using the wrong base path for checks), fixed links in the documents
* [svn r37774] Adding support for checking generated API links (for link roles)
* [svn r37773] typo
* [svn r37772] have both capturings have the same done/reset semantics (should also fix a buildcmodule related problem, e.g. for greenlets)
* [svn r37770] remove very old try at doing svn bindings
* [svn r37769] minor clarifi
* [svn r37767] Fixed Windows issues in doctests
* [svn r37766] unifying non-FD and FD capturing some more (could be more, but at least the APIs start to feel the same)
* [svn r37765] Was still getting filenames for source files from code objects, so adding more defensiveness, and made that the 'capture' object is passed over to the builder instances (to help debugging, currently not used)
* [svn r37752] Added some more code examples
* [svn r37751] fix two other places that used capturing (although the greenlet fix is not really related, but i first saw it now on win32)
* [svn r37745] fixing and adding to docstring
* [svn r37743] ah, there was yet another version of capturing hiding
* [svn r37742] fix reference
* [svn r37741] monster checking for
* [svn r37739] Made that only the first two lines of a source file are examined for the encoding
* [svn r37738] Fixed problem probably caused by removing some comment or something..
* [svn r37736] cosmetic docstring change
* [svn r37733] Removed two rather useless methods that just delegated to methods on the underlying object, in favour of calling the underlying object's methods directly (py.io.OutErrCapture.writeorg\*)
* [svn r37729] Added some code to py.io.FDCapture and py.io.OutErrCapture to allow writing to the original (patched) file descriptor. Also made that the capturing object is passed to apigen.py's build() function (from rsession.py), which uses the new methods to print progress information
* [svn r37723] Add a test
* [svn r37721] Fix. !Tests needed!
* [svn r37720] Kill dead import
* [svn r37703] adding references to path and code documentation
* [svn r37701] note on removal of compat.TestCase
* [svn r37700] refer to experimental gateway path code for substantiating future directions
* [svn r37699] removing py.\_dist export (only accessed from setup.py and py/bin/\_makepyrelease.py)
* [svn r37698] waiting even a bit longer for loop-on-failing
* [svn r37696] removing py.test.compat.TestCase - it's small code but nobody uses it i think and there is no documentation
* [svn r37693] Some considerations regarding the experimental Gateway/Remote Path implementation code. Also making it "py.\_\_." importable but not advertising it for 0.9 at all. It's an interesting feature for 1.0 :)
* [svn r37674] Added document 'code.txt' that describes py.code, added docstrings to py.code public items
* [svn r37664] some streamlining, additions and "de-marketing" :) and adding some XXX
* [svn r37661] fixing this script (IMO it was broken if you provided arguments)
* [svn r37647] make the example independent from the installed python binaries
* [svn r37646] Wording, and missing attribute in the greenlet documentation
* [svn r37645] adding an example for execnet (passes doctest)
* [svn r37644] typo
* [svn r37643] stroke the historical notes, and streamlined various parts of execnet documentation, added some api references
* [svn r37642] issue40 resolved
* [svn r37641] removing py.path.extpy after all
* [svn r37640] Added first bits of a document about py.path
* [svn r37639] Not sure why this was here... :|
* [svn r37638] Re-enabled the tests on win32
* [svn r37637] Fixed Windows path problems for apigen; test\_apigen\_functional.py now works (couldn't test with the py lib itself, though, as it wants to build greenlets and i don't have a suitable c compiler installed)
* [svn r37631] Kill unnecessarry cache
* [svn r37624] Slightly hackish way to put all tracing calls as low as possible. This should kill unnecessary frames. Right now tracing is done only for py.test.Function objects, I don't know what to do with sth else, let's leave it alone
* [svn r37623] Get rid of last frame if possible (the frame containing start\_tracing should not be there, unless this is the last frame)
* [svn r37622] Don't use previous versions in future examples
* [svn r37621] all tests works for me
* [svn r37620] ensure 2.5 compatibility
* [svn r37616] All code snippets are colored now, also properly fixed hiding of \_\* names in the navigation
* [svn r37615] skip this test on windows, it should work but doesn't and it's not too important (apigen does not fully run on win32 because of LSession not working there)
* [svn r37610] Stack trace code snippets are now syntax colored
* [svn r37609] Oops, forgot to check this in: update to deal with change in py.xml.html
* [svn r37608] Added some code to make sure 'inline' elements aren't indented (gotta love 'ignorable' whitespace)
* [svn r37602] Fixed support for files ending on a comment rather than a newline, fixed some unicode() call so it's not done on objects that are already unicode
* [svn r37599] fixing a windows-error: i believe that most strings passed into htmlgen.deindent() actually have a normalized '\n' line-ending already (e.g. if they come from python docstrings).  So the caller now needs to specify otherwise
* [svn r37598] removing broken tkinter session support ahead of the 0.9 release (might be re-introduced later)
* [svn r37596] bah, waiting longer also in py-trunk
* [svn r37594] Fixed unicode issues in apigen/htmlgen.py and apigen/source, moved some methods out of HTMLDocument and added support for finding out the encoding of a Python file in order to accomplish that (both in source/html.py), fixed some minor issues (items with a name starting with \_ are now hidden from nav, fixed indentation issue in nav) in htmlgen.py
* [svn r37582] Updated mock-up function return value
* [svn r37579] Made that the package name can be set when using from\_dict() to fill a DocStorage
* [svn r37555] update
* [svn r37554] add XXX
* [svn r37553] Skip if boxed
* [svn r37545] Moved get\_star\_import\_tree out of the DocStorage class, created 'pkg\_to\_dict' function in docstorage.py and using that from apigen.py (to later simplify py.test integration, and to improve testability). Simplified 'deindent()' a bit, made the wrap\_page use the filenames from the stylesheeturl and the scripturls instead of having them hard-coded, some minor HTML/CSS changes
* [svn r37540] Nicer formatting of docstrings (de-indented and such), fixed problem getting to frame source (IOError that popped up when building the py lib's api docs) in description.py
* [svn r37535] update on that
* [svn r37534] Some minor visual modifications for stack traces, fixed check for py.code.Source object filenames
* [svn r37533] Make distribution working with custom conftests at exact levels. This is very explicit - dist\_rsyncroots works only in that conftest directory
* [svn r37531] Removing try/except to make debugging a little bit easier
* [svn r37502] unify dist\_\* options and give dist\_rsync\_roots more precision: you now specify relative paths (relative to the conftest.py where a dist\_rsync\_root setting resides) or absolute paths
* [svn r37501] rename for clarity
* [svn r37500] postponing generalizing of host specifications for 0.9
* [svn r37499] concreter error message
* [svn r37498] bah, my local config but i want to have it committed to get rid of the "M" state, ehem
* [svn r37494] the start of factoring out common and unified host handling within py.test
* [svn r37481] new attempt :-(
* [svn r37478] refactored/streamlined tests slightly i think svn testing needs some more refinements (not before 0.9)
* [svn r37477] use relative paths to try to make the py-lib both work in paths that have quotable characters \_and\_ with some versions of latex that don't like quotes in \input. let's see whether it works..
* [svn r37465] \* all Sessions now have fixoptions() \* added some tests for implied and conflicting options \* make all Session objects inherit from the base session \* small cleanups with respect to test setup \* separate tests a bit
* [svn r37463] test\_get\_initpkg\_star\_items() fails for me when run as part of the whole py lib tests, succeeds when ran as a single test. Pruning sys.modules fixes this (but i am not sure why the tracer depends on seeing the import happen)
* [svn r37462] allow the remote side to run in verbose mode
* [svn r37461] skip these tests on windows (some reliances on unix-paths and i don't think apigen can run on windows anyway at the moment)
* [svn r37460] these tests rely on exact formatting, so skipping them again (fail for me on pypy2)
* [svn r37459] fix socketgateway test to run on both win32 and linux
* [svn r37458] try to only rsync the py lib, not the dirpath() of it
* [svn r37457] explicitely shutdown the gateway for the remote session after each run
* [svn r37456] some small adjustments/cleanups regarding intiialisation
* [svn r37455] added a sample conftest.py to run windows tests remotely, note the module docstring
* [svn r37453] Changed test a bit to avoid issues on non-UNIX platforms
* [svn r37450] skip auto-import checks of directive.py because it requires docutils
* [svn r37449] fix failures in case of docutils not being installed (related to importing 'py.\_\_.rest.directive')
* [svn r37447] config.\_reparse() from current dir can drag in conftest.py files (happened to me because i had two py lib checkouts)
* [svn r37446] don't try this on windows
* [svn r37445] use suprocess module to start socketserver subprocess
* [svn r37444] a try at a wrapping loop\_socketserver.py that restarts the actual socketserver.py
* [svn r37443] slightly re-organizing RemoteTerminalSession for easier subclassing
* [svn r37442] update
* [svn r37439] Sorry, cfbolz, seems the print to stderr breaks some test... :|
* [svn r37438] Fiddle-fixed --rest, although it works now some cleanups would be nice... Cleaned up the tests a bit and re-enabled some (sucky tests are better than no tests at all, I guess... :| ), re-enabled a debug print after complaints from cfbolz ;) but this time it prints to stderr (to avoid ReST pollution)
* [svn r37437] Made stuff a bit more robust by converting non-strings to strings in some places
* [svn r37435] Removing debug print (ended up in the ReST)
* [svn r37434] Fixed bug in add() that made that the to-be-added item didn't get added (and yes, I've added a test now ): ), made some error message a bit clearer
* [svn r37433] Add a helper which returns value + path of a dir where value was found
* [svn r37432] Some updates on stuff I've worked on, fixed ReST
* [svn r37431] Fixed problem that made that the wrong package was imported when ran after test\_apigen\_example
* [svn r37430] Fixed test, was depending on dict order and values had changed
* [svn r37429] Ugh, revert
* [svn r37428] added a note about dist\_rsyncroots semantics
* [svn r37427] Kill the last occurencies of pkgdir
* [svn r37425] Fixed Code.\_\_ne\_\_() (wrong function sig, missing 'other' argument)
* [svn r37411] Fix a bit rest and test\_rest. Note that actually these tests are testing anything from time to time, need closer look
* [svn r37409] Fix the test
* [svn r37408] Kill two unnecessary tests now
* [svn r37407] Unskip two tests
* [svn r37395] Fixed some small issues in source nav, turned namespace\_tree into an instance var (finally! :), changed the api nav so it now shows the full tree with the current item unfolded + siblings of itself and all parents (+ children if it's a namespace item)
* [svn r37383] Added call stacks, made that method pages aren't built at all anymore, made that build\_namespace\_view isn't called for anything but namespaces anymore, made that parts of documents are hidden at first (with a link to 'open' them)
* [svn r37381] Add a shortcut
* [svn r37380] This is rather huge checkin of a rsession refactoring. Got rid of most of the pkgdir and changed to use get\_collector\_trail. Some tests need to be rewritten or killed, right now they're skips
* [svn r37355] Fixing test (not sure why it was in this state?)
* [svn r37354] Move tracing a bit down (performance, no change in behaviour)
* [svn r37348] Fallback
* [svn r37347] Enchance the test
* [svn r37346] Add \_\_ne\_\_
* [svn r37345] Remove unneeded import
* [svn r37343] Hours later and tons of debug print changes...
* [svn r37337] Replaced try/except with a hasattr (makes that unrelated errors are no longer obfuscated)
* [svn r37330] Replacing any non-Unix newlines with plain \n in doctests, this fixes some win32 problem
* [svn r37328] Fixed problem with Windows paths in tests when running on Linux (and fixed relpath() api to deal with line seps in a better way)
* [svn r37326] Fixed exec command on win32
* [svn r37325] Fixed problems with windows path seperators, fixed functional test exec command for win32
* [svn r37322] Turned out there was a function for this already, even... Thanks, fijal!
* [svn r37318] Rolling back previous change, instead applying what seems to have been the intention from the start: if '' is supplied as hostname from the server, use the IP address as reported by socket.getsockname()
* [svn r37317] Fixed execnet tests on Windows, for some reason it uses '0.0.0.0' as IP address if '' is given as hostname
* [svn r37315] Skipping test on Windows (no fcntl), removing stuff from try/finally body\n(too much crap in there, obfuscates errors)
* [svn r37314] Added docstring
* [svn r37306] Fixed link (now properly has an href)
* [svn r37305] Added some JavaScript to show and hide function information (source and such)
* [svn r37303] Made functional test use an absolute path to py.test
* [svn r37302] experimentally enable FD-based output capturing per-test
* [svn r37301] No longer that way (somebody should read that document once again)
* [svn r37300] Make clean complaints about lack of disthosts and call fixoptions if present
* [svn r37299] Removed stupid space
* [svn r37298] (fijal, hpk) - bugfix
* [svn r37297] Fixed dependency on order in the function argument formatting tests
* [svn r37294] seediff
* [svn r37293] Another aproximation :-]
* [svn r37292] SkippedTryiter should be reporter more like in terminal.py
* [svn r37291] seediff
* [svn r37290] \* allow only CAPITAL letters for short options \* documentation and test for it \* streamlined documentation a bit
* [svn r37289] todotodo
* [svn r37288] added some RSession related TODOs
* [svn r37287] don't eat exception, simply use finally
* [svn r37286] Pass a config object deeper in hierarchy
* [svn r37285] this file describing option is obsolete now
* [svn r37281] omit tracer test directory for imports (XXX example Filesystem trees should be created from code and not reside in SVN)
* [svn r37278] move files from branch to trunk (and thus complete the merge of the config branch into the trunk)
* [svn r37277] merging of file:///svn/py/branch/config/py/test/rsession/testing/test\_webjs.py revisions 36936 to 37267:
* [svn r37276] merging of file:///svn/py/branch/config/py/test/rsession/testing/test\_rsession.py revisions 36936 to 37267:
* [svn r37275] merging of file:///svn/py/branch/config/py/apigen/testing/test\_apigen\_functional.py revisions 36936 to 37267:
* [svn r37274] merging of file:///svn/py/branch/config/py/test/rsession/web.py revisions 36936 to 37267:
* [svn r37273] merging of file:///svn/py/branch/config/py/test/rsession/rsession.py revisions 36936 to 37267:
* [svn r37272] merging of file:///svn/py/branch/config/py/test/rsession/rest.py revisions 36936 to 37267:
* [svn r37271] merging of file:///svn/py/branch/config/py/doc/test\_conftest.py revisions 36936 to 37267:
* [svn r37270] merging of file:///svn/py/branch/config/py/doc/TODO.txt revisions 36936 to 37267:
* [svn r37269] merging of file:///svn/py/branch/config/py/test/config.py revisions 36936 to 37267:
* [svn r37268] merging of file:///svn/py/branch/config/py/\_\_init\_\_.py revisions 36936 to 37267:
* [svn r37266] Fixed a problem in the CallFrame code, it doesn't like py.code.Source objects (no test, sorry, didn't succeed in reproducing it from a test yet), some minor HTML changes
* [svn r37264] create the new development trunk
