Change log for tl.testing¶
0.6 (unreleased)¶
- Fixed PyPI link in Sphinx template.
- For absolute imports, use the Python language feature rather than
manuel
. Moved the logic for cairo mocks used when building the docs to the Sphinx configuration. - Applied package layout conventions according to tl.pkg.
0.5 (2012-01-04)¶
- Added
tl.testing.thread
providing thread classes that handle or report exceptions as well as a thread-aware test case class that provides reporting on threads left behind by tests and some conveniences for using threads in test code. - Dropped compatibility with Python 2.5 by introducing a context manager for joining threads.
- Added
tl.testing.doctest
providing aDocFileSuite
factory which is compatible to that of standard doctest but uses pypi:manuel and allows for some manuel-related additional configuration. Always require manuel, removed thecairo
extra requirements. - Fixed #5: Shield test runs that are the subject of this package’s tests from options passed to runs of this package’s test suite itself.
- Changed this package’s own tests to use the stock
unittest
test runner instead of the Zope test runner. Since this caused changes in doc-test counting, use mocks to make clearer whether certain tests are executed. Requires the pypi:mock package. - Updated code and development environment to newer versions of packages: no
longer use deprecated
zope.testing.doctest
, build with more recent tl.buildout_gtk. - Keep buildout.cfg under version control instead of base.cfg, expect overriding to happen in local.cfg.
- Updated to using Sphinx 1.1 for building the documentation, require graphviz to create class inheritance diagrams, require sphinxcontrib-bitbucket to generate links to bitbucket issues and sphinxcontrib-cheeseshop to generate links to packages’ PyPI pages. Use a global table of contents in the sidebar. Added the technically detailed doc-tests to the Sphinx docs.
- Temporarily ignore import errors for the cairo C module to allow the full API docs to be built at Read the Docs.
0.4 (2009-09-14)¶
- Cairo testing:
- Updated to using manuel 1.0.0b3, restoring Python 2.5 compatibility.
- Allowed for text in addition to the expression in figure captions, as well as multi-line captions.
- Interpret an option for excluding one or more rectangular areas of a surface from comparison with the expected image.
0.3 (2009-07-23)¶
- Cairo testing:
- Updated to using manuel 1.0.0b2, implying the requirement of Python 2.6 and a refactoring of the DocFileSuite that loses encoding support.
- Improved the behaviour in the case that a result cannot be written to an image file.
- Write the test result to an image file also in the case that the expected image cannot be loaded.
- Disallowed absolute image paths as they would only work on Unix-like systems because of the requirement to use / as the path separator.
- Fixed DocFileSuite to accept multiple test file names, renamed the
manuel_object
parameter to justmanuel
, only include the doc test, footnote and cairo Manuels by default. - Require a cairo.ImageSurface instead of a Surface as the value of an example’s Python expression.
- Report ImageSurface format mismatches, avoid false mismatches with FORMAT_RGB24 surfaces.
- Organisation:
- Moved the project to the Mercurial version control system.
- Build documentation using Sphinx.
- Improved documentation, in particular added doc strings to the code.
0.2.2 (2009-06-24)¶
- Updated to using the manuel 1.0.0b1 API.
- Made the cairo test runner recognize the CAIRO_TEST_RESULTS environment variable as a place to dump failed test results as png images.
0.2.1 (2009-06-09)¶
- Use new
manuel.codeblock
functionality in cairo image doc-testing. - Simplified the cairo testing code a little due to a bug fix in manuel.
0.2 (2009-05-15)¶
- Added a manuel test runner that compares cairo surfaces to images included as ReST figures in doc test files.
0.1 (2008-11-11)¶
initial release