ID as a form of primitive obsession

Primitive obsession: the act of utilizing language primitives like strings, integers, arrays, and such to represent complex domain objects. A common example of this is representing money with ints or floats. Primitive obsession tends to result in the spread of logic that could be cohesively contained in one file across the application, repetition of conversion logic through flows, and many strange hacks that slow comprehension.

A common issue I find across many organizations that is quietly hindering them is the utilization of identifiers inside of domain logic. Somewhat recently it hit me that this is a form of primitive obsession. Identifiers are strings or integers and they represent much larger domain objects, which qualifies them as a form of primitive obsession. Code that utilizes identifiers is noticeably harder to refactor and usually requires a lot of database accesses.

How to spot ID obsession

Imagine you are working in Rails or an equivalent framework. There is an interface layer (controllers, workers, etc) that takes external requests and converts them into domain actions. That conversion layer is the only place that identifiers should be handled. The identifiers should be converted to domain objects at that point and all domain logic will utilize those domain objects.

So, if you happen to see a method on a model instance that utilizes an ID, then that is incorrect. Instead, pass a model instance to that method.

If you have a service object, that service object should only be handling models and not identifiers.