PanDA Pilot - Architecture 2

Contributions

1. Check the TODO.md and STYLEGUIDE.md files.

2. Fork the PanDAWMS/pilot2 repository into your private account as origin. Clone it and set the PanDAWMS/pilot2 repository as upstream.

3. Make new code contributions only to a new branch in your repository, push to origin and make a pull request into upstream. Depending on the type of contribution this should go against either upstream/master, upstream/main-dev or upstream/hotfix.

Verifying code correctness

Do not submit code that does not conform to the project standards. We use PEP8 and Flake verification, with everything enabled at a maximum line length of 160 characters and McCabe complexity 12:

flake8 pilot.py pilot/

For Python 2.6 you need to install flake8<3.0.0, which can miss a few things. Check the output of TravisCI to verify if you have to use this old version.

Running the pilot

The pilot is a dependency-less Python application and relies on /usr/bin/env python. The minimum pilot can be called like:

./pilot.py -d -q <QUEUE_NAME>

where QUEUE_NAME correspond to the ATLAS PandaQueue as defined in AGIS. This will launch the default generic workflow with lifetime default lifetime of 10 seconds (i.e., too short to do anything).

The -d argument changes the logger to produce debug output.

Running the testcases

The test cases are implemented as standard Python unittests under directory pilot/test/. They can be discovered and executed automatically:

unit2 -v

Building and viewing docs

1. Install sphinx into your environment by pip or other means with all the necessary requirements.

2. Navigate into ./doc in your fork and run make html.

3. Open _build/html/index.html with your browser.

Automate documentation to your module

Add the following lines somewhere in the documentation:

.. automodule:: your.module
    :members:

For more info, visit http://sphinx-doc.org