Developer’s Guide

This page contains information useful for developing pebl (adding code to it), not for using it. For instructions on using pebl, consult the Tutorial.

Setting up a development environment for Pebl

1. Checkout pebl code from the subversion repository:

   svn co http://pebl-project.googlecode.com/svn/trunk pebl

2. Install development-related dependencies. Install nose (easy_install nose) for unit tests and sphinx (easy_install sphinx) for creating documentation.

3. Install pebl in development mode. This allows you to make changes to the source code and have it reflected in the “installed” version of pebl immediately.:

   sudo python setup.py build develop

You can now modify pebl source code and use the changes simply by importing pebl. If you modify C or C++ code, you must run python setup.py build again.

Coding guildelines

1. Style guidelines

   We follow the python community standard PEP 8 for coding style although in-project consistency is more important that strict compliance with PEP 8.

2. Using pebl.config

   Use the pebl.config module to parameterize your code. Configuration should be read during object initialization (in __init__) even if it is not used until later. Any changes to configuration parameters after object intialization should not affect your object.

3. Testing

   We’re not religious about following strict unit testing methodologies but ALL critical code should be accompanied by tests (unit tests are preferred but doctests are also accepted). Run the unit tests by running nosetests -v pebl.test from any directory. Run unit tests frequently during development and especially before commiting any code. You can place test to be run manually in pebl/test.manual but automated unit tests are strongly preferred.

Bug Reports

Use the Issues tab on the Google code site for all bug reports and proposed enhancements. When closing a bug, include the subversion revison that resolves the bug in the comments.

Committing code

Before you can commit code, I will have to you to the developers list in Google code. Please email me for this.

Thanks dude

And finally, thank you for considering contributing code to Pebl. I envision pebl to be a community resource that can serve as a good base for further research into Bayesian networks. Feel free to contact me with any questions or comments (abhikshah@gmail.com).