Contributing Changes

PanDA components are developed based on GitHub Flow with small customization. The sequence is as follows:

  1. Creation of a branch or fork. In git, branches are lightweight things that are often temporary and may be deleted anytime. The master branch is protected so that only a couple of persons can push commits there.

  2. Adding commits to the branch. You can experiment with any changes since they don’t affect the master branch.

  3. Deployment of changes on a test instance. All changes must be verified before being merged to the master branch.

  4. Submission of a Pull Request which will initiate a discussion about your changes.

  5. Discussion and review on your code. Once a Pull Request has been opened, your changes are reviewed to check functionalities, the coding style, test results, and so on. Note that your code must follow PEP8.

  6. Merging changes. Once your changes are approved, they will be merged to the master branch. The branch can be deleted at this stage since commit logs are also merged.

How to install your local changes to your instances

You can install and test your changes locally before submitting pull requests. The following example shows how to install local changes to your own PanDA server.

$ # checkout the repository
$ git clone https://github.com/PanDAWMS/panda-server.git

$ # add changes in panda-server
$ cd panda-server
$ ...

$ # stop the PanDA server
$ /sbin/service httpd-pandasrv stop

$ # go to virtual env if necessary
$ . <venv_dir>/bin/activate

$ # make sdist and install it twice with different pip options since pip doesn't install it
$ # without those steps when the version number is incremented
$ cd panda-server
$ rm -rf dist
$ python setup.py sdist
$ pip install dist/p*.tar.gz --upgrade --force-reinstall --no-deps; pip install dist/p*.tar.gz --upgrade

Publishing a new version of panda-* package on PyPI

The procedures are as follows:

  1. Increment the version number, which is typically defined in panda-*/PandaPkgInfo.py.

  2. Push it to the master branch of the git repository.

  3. Make a new release on the master branch of the git repository in GitHub.

which automatically triggers a git action to publish the version on PyPI.