Add documentation about packaging rules

docs/package_rules.md

@@ -0,0 +1,73 @@
+# Package rules
+
+`dbt-score` is able to search for rules in Python packages.
+
+The default namespaces which are searched are:
+
+- `dbt_score.rules`: This namespace contains the core rules, i.e. those packaged
+  within `dbt-score`.
+- `dbt_score_rules`: This implicit namespace package can be used for any custom
+  rule, either implemented by 3rd party packages or by yourself.
+
+Additionally, `dbt-score`
+[can be configured to search for rules in other namespaces](configuration.md/#main-configuration).
+
+## Packaging for a single project
+
+If you want to write custom rules applicable for your project only, it is
+recommended to bundle them directly within your dbt project.
+
+The only requirement for `dbt-score` to discover your custom rules is for those
+rules to be present and importable in your Python environment, which might vary
+depending on the way you use Python and virtual environments.
+
+The following project structure is usually observed:
+
+```
+my-dbt-project/
+├─ dbt_score_rules/
+│  ├─ my_project_rules.py
+├─ dbt_project.yml
+├─ models/
+├─ ...
+```
+
+For your convenience, `dbt-score` will inject the current working directory in
+the Python path, making `dbt_score_rules` discoverable here without any further
+configuration.
+
+## Packaging for multiple projects
+
+If you want to write custom rules which apply to multiple dbt projects within an
+organization, or even to be distributed to the public, the best way is to
+package them in a Python wheel.
+
+The only thing the wheel needs to do is to expose modules within a top-level
+`dbt_score_rules` (or any other namespace, as long as projects are configured to
+read from such an additional namespace).
+
+To avoid naming conflicts within the `dbt_score_rules` namespace, it is
+recommended to pick module names which match either an organization, or a
+specific project. For example, an hypothetical `dbtviz` project which makes use
+of dbt's `meta` to describe visualizations associated to models could have rules
+for this metadata saved in `dbt_score_rules.dbtviz`.
+
+## Debugging
+
+You can verify the list of rules discovered and configured by `dbt-score` by
+running:
+
+```shell
+dbt-score list
+dbt-score list --namespace dbt_score_rules.dbtviz  # filter by a given namespace
+```
+
+If your custom rules are not present, try to open a Python shell and import
+them:
+
+```python
+import dbt_score_rules.dbtviz
+```
+
+If it doesn't succeed, it means the rules are not properly installed or not
+within the Python path.

mkdocs.yml

@@ -23,6 +23,7 @@ nav:
   - Home: index.md
   - Get started: get_started.md
   - Create rules: create_rules.md
+  - Package rules: package_rules.md
   - Configuration: configuration.md
   - Rules:
       - rules/generic.md