setuptools_scm is a setuptools plugin that uses git metadata to manage python package versions
https://github.com/pypa/setuptools_scm
It also replaces the use of MANIFEST.in, by including in the soruce distribution all the
files that are under version control (using `git ls-files`).
An extra plugin 'setuptools_scm_git_archive' also adds support for git archives,
e.g the .zip files automatically generated on Github.
https://github.com/Changaco/setuptools_scm_git_archive
For example, one could do: `pip install https://github.com/greginvm/pyclipper/archive/1.0.4.zip`
Note that installing from git archives currently works only for tagged commits, because
of limitations of git archive format itself.
However, for non-tagged commits one can still install from the git repo:
`pip install git+https://github.com/greginvm/pyclipper`
* add matthew-brett/multibuild as git submodule
* add .travis.yml file
* add config.sh file with run_tests function to invoke py.test command
print python version and number of bits (32 vs 64) before running the tests,
just to be sure...
* test_pyclipper.py: define 'integer_types' as (int, long) on PY2 and int on PY3
* setup.cfg: default to sdist --formats=zip as it's more portable
* .travis.yml: add deploy section to upload sdist and wheel to Github Releases on tags
* .travis.yml: commented out 'deploy' stage; greginvm needs to replace it using his authentication details
* appveyor: run tests and build wheels on Windows Python 2.7, 3.4 and 3.5 (32 and 64 bit)
We use Olivier Grisel (ogrisel) batch script to configure environment
variables to use the correct MSVC C++ compilers when building the
extension module.
https://github.com/ogrisel/python-appveyor-demo
* .appveyor.yml: add deploy section for uploading wheels to Github Releases on tags
* .appveyor: comment out deploy section; to be replaced by greginvm's auth_token
* .travis.yml: add Python 3.3 (Linux and OS X) to the matrix
* .appveyor.yml: add Windows Python 3.3 (32 and 64 bit) to the matrix
Removed test which tested the behavior of SCALING_FACTOR and added tests which test whether the appropriate warnings are raised when SCALING_FACTOR is set to a value different from 1.
Having e.g. Cython installed in a virtualenv inside the project root make pytest also find the Cython test cases. These take a lot of time and also fail without some additional libraries being installed.
