Development Guidelines

1. Languages & Build Tools

• Primary Language: Kotlin (built with Gradle using Kotlin DSL).
• Secondary Language: Python (restricted to small scripts or lightweight web services).
• Fallback: Follow canonical language guidelines (e.g., Pkl formatter for Pkl). Match surrounding code style.

2. Code Quality & Formatting

• Formatters: Kotlin must use ktfmt (GoogleStyle). Python must use black.
• Enforcement: GitHub Actions CI pipelines must fail on unformatted code.
• Documentation: Write self-documenting code. Use KDoc for Kotlin.

3. Testing

• Requirement: Unit tests are mandatory for all components.
• Enforcement: GitHub Actions CI pipelines will automatically execute tests on all pull requests. Builds will fail if tests do not pass.

4. Repository Structure & Execution

• Required Files:
  • README.md: Must contain repository intent and comprehensive usage instructions.
  • ATTRIBUTIONS.md: Must list third-party libraries and author information.
• Portability: Projects must be entirely self-contained.
• Docker: Runnable projects must have a GitHub Actions CI pipeline that automatically produces a Docker image.

5. Git & Contribution Workflow

• Versioning: All releases and tags must strictly adhere to Semantic Versioning (SemVer).
• Commits: Strictly follow Conventional Commits. GitHub Actions CI linting will fail on invalid messages.
• Branching: Main repositories are restricted to develop, main/master, and release branches.
• Pull Requests (Stabilized Repos):
  • All contributions must be PRs originating from a fork and targeting the develop branch.
  • Use the mandatory PR template.
• Code Review Requirements: Pull requests require at least one approval from an authorized reviewer before merging is permitted.
• Issues: Report via GitHub issues strictly using the issue template.

6. Grid'5000 Deployments

• Tooling: Python deployment scripts utilizing EnOSLib are absolutely required.
• Reproducibility: Zero hidden information is permitted. Anyone must be able to reproduce the results exactly using the provided deployment files.

7. Generative AI

• Policy: Discouraged. Only use AI tools if you possess a deep understanding of the generated output and assume complete responsibility for the code.