Before, calling figure() would produce a big blank area. This ensures
output is only returned if the figure has content.
Also added basic testing for pylabtools, for which we had none.
ported from numpy.testing.noseclasses, with the equivalent code
extracted into _numpy_testing_noseclasses in case numpy is not
installed.
removed some unused imports throughout
Move all libraries in external into subfolders and place a custom
__init__.py into each of them. This way the system installed ones are
tried first, before falling back to the bundled ones.
Now every distribution can properly depend on the upstream library.
(e.g. bundled libraries are not allowed in fedora)
Signed-off-by: Toshio Ernie Kuratomi <toshio@fedoraproject.org>
Signed-off-by: Thomas Spura <tomspur@fedoraproject.org>
* Moved the pprint attribute to PlainTextFormatter.
* Added DisplayFormatter.plain_text_only for disabling all but the
plain text formatting. This is now used by doctest_mode.
* Remove result_display hook.
* Fixed doctest_mode magic.
* Refactored inline matplotlib backend to use new display system.
* Added display formatter for matplotlib.figure.Figure objects.
so the display system will show the svg for them.
* Added display and display_png to the user's namespace in pylab.
* Cleaned up pylabtools.py
As best I can tell, we've been mis-specifying data_files for ages.
For some reason with a plain setup.py it worked fine, but using a
customized cmdclass for build_py, the error shows up.
The problem was that we were returning a list of triples for
data_files, and the spec clearly says they must be pairs:
http://docs.python.org/distutils/setupscript.html?highlight=data_files#installing-additional-files
This will give us an easy way for users to report platform details,
without trying to cram lots of data into the version string.
Now, the version information in release.py is simple and doesn't
attempt to read version control data at runtime. We expose sys_info()
as a top-level function so it's very easy to remember/direct for new
users.
Minor doc/copyright notices updates in the same files.
This uses the tool to record the commit data from our previous commit,
and ensures that auto-generated archives from Github, for example,
always carry identifying information.
Ideas and code taken from Matthew Brett's tools in nibabel.
This branch cleans up our trunk to make a 2to3 run be as painless as
possible, while keeping our code fully 2.6-compatible.
Closes gh-159 (pull request).
