Developer guide

Index

• Developer guide
  • git repository
  • Running tests
  • Adding a new model
  • Publishing the documentation
  • Uploading a new version of pyAnno to PyPI

git repository

pyAnno is hosted on GitHub at https://github.com/enthought/uchicago-pyanno.

The recommended workflow to extend and improve pyAnno is the following:

1. Fork the project on GitHub
2. Implement your changes on your repository
3. Submit a pull request on GitHub

Running tests

Even though the tests are written using unittest, we recommend the use of nosetests to execute the library test suites. From the root directory of the pyanno project, type

$ nosetests -v pyanno/test

Note

The tests that verify some of the models’ functionality (e.g., the estimation of the model parameters), are stochastic. This has the advantage that they test different, general scenarios at each round, but occasionally leads to test failures. If one of the test fails, please run it a second time. If the failures are consistent, please report a bug using the issue tracker.

Adding a new model

This is a checklist of things to do when implementing a new model in pyAnno:

1. Write a new model implementation as a subclass of pyanno.abstract_model.AbstractModel, implementing the full interface.
2. Add an import statement in pyanno.models
3. Add some documentation about the class in the models page.

To add the model to the UI:

1. Write a new subclass of pyanno.ui.model_view.PyannoModelView, the traitsui graphical view of your model parameters, and of pyanno.ui.model_view.NewModelDialog, which creates a dialog requesting the parameters to create a new instance of the model. See the classes defined in pyanno.ui.model_a_view for reference.
2. Add the name of the model, and a reference to the class and model view in model_name, _model_name_to_class, and _model_class_to_view, at the beginning of the pyanno.ui.model_data_view.ModelDataView.

Publishing the documentation

The documentation is based on Sphinx. It can be found in pyanno/docs.

After editing the documentation, you should compile it, revise the result, and publish it online:

1. Check out the gh-pages branch from the git repository in a new directory, DOCPATH.

2. Enter the directory pyanno/docs and edit the BUILDDIR variable in the local Makfile to DOCPATH.

3. Type

   $ make html

   Make sure the the pyanno package is in the PYTHONPATH, or Sphinx will fail to generate the API documentation.

4. Enter DOCPATH, check that the documentation has been correctly generated, and push the branch back to GitHub:

   $ git add *
   $ git commit -m"DOC: describe your changes here"
   $ git push origin gh-pages

The new documentation will be published within seconds.

Uploading a new version of pyAnno to PyPI

1. Revise the setup.py file and update the version number

2. Run the tests and correct any bug

3. Push to PyPI:

   $ python setup.py register bdist_egg bdist_wininst sdist upload