Guard rail tools¶

Guard rail tools safeguard the development of Renaissance.Py. Each of them encodes a rule that the repository has agreed on, checks it automatically, and fails the build when the rule is broken, so that a rule stays true instead of slowly eroding.

They are kept in the tools directory, can all be run locally before a pull request is opened, and are all run again in CI.

Available tools¶

1. Layering check (tools/check_layering.py): forbids imports that cross a layer boundary, which keeps the core independent of the parser bindings and the recipes.
2. Lint and type-check budget (tools/lint_budget.py): counts the ruff and pyright issues per kind and fails when a change introduces a new one.
3. Image policy check (tools/check_doc_images_policy.py): enforces the naming and the location of documentation image assets.

Adding a guard rail¶

A new guard rail is worth adding when a rule is stated in the documentation, is easy to break by accident, and can be checked mechanically. Give it a script in tools, tests in test/tools, a page in this section, an entry in the list above, and a step in the appropriate workflow in .github/workflows.