Fork-based Workflow For XGC Repositories

In order to keep the repositories for code development clean, normal users have only read privileges on the XGC repositories. These repositories contain only one branch called “master”. Code development is carried out in user forks. A fork is a separate repository, that is initially an exact copy of its parent repo. Note that each user can create only one fork of a repository. The fork created by the user will not be listed under the PU organization (PrincetonUniversity/[repo name]) but under the name of the user ([username]/[repo name]). When code modifications prepared in a fork are ready for merging, a pull request has to be created on the GitHub website in order to pull changes from a user fork into the parent repository. A fork can be used by only one developer or by a group of developers. Inside a fork, branches can be used to organize code development.

Code development shall be organized as separate features or projects. Each feature/project has a lead developer in whose personal fork the code development for the feature/project is carried out.

Contents

• Brief description of the workflow
• Use of forks
• Feature branch workflow inside a fork
• Management of Branches
• Pull Requests
• Policy for large files
• Automated Testing of Pull Requests
• XGC Coding Guidelines
  • Basics
  • Formatting
  • Function/Subroutine arguments
  • Function/Subroutine
  • Doxygen comments
  • Version control (git)
  • Hardcoding
  • Compatibility
  • Runtime inputs and preprocessor flags
  • Modularity
  • Source file length
  • Testing
  • Pull requests
• Git Cheat-Sheet