infrastructure-database

name	infrastructure-database
description	How orm/ and repositories/ split responsibility in app/infrastructure/database/

Infrastructure / Database

Two sublayers, two jobs:

app/infrastructure/database/
├── orm/            ← thin Sequel models, no business logic
└── repositories/   ← maps ORM rows ↔ domain entities, owns DB queries

Responsibilities

Layer	Knows about	Returns	Allowed to contain
orm/	Sequel + schema	ORM model instances	Sequel::Model(:table), plugins, associations
repositories/	ORM + domain entities	Domain entities	Query construction, entity (re)building

Dependency direction

repositories/  →  orm/

• repositories/ requires / uses orm/ classes.
• orm/ must NOT import repositories/, services, controllers, or anything upstream.
• Callers (services, controllers, application orchestration) talk to repositories/. They must not reference Database::*Orm directly.

Why split

The ORM model knows the database. The Repository knows the domain. If callers were allowed to touch the ORM directly, the schema would leak into services and controllers, and swapping persistence would touch every caller.

See sublayer details in:

• ./orm/SKILL.md
• ./repositories/SKILL.md