Traditionally, domain classes are too often:
Too much abstraction can easily make it hard to reason about what the program does from a process/use case level.
A “Person” can be endlessly many things in various contexts and doesn’t like being trapped in any domain class.
The impedance mismatch between fluent business processes and rigid class-based data abstractions pollutes our mental model about our domain. The atomic attribute building blocks of molecules allow us to model complex and specific data structures that exactly match our mental model in any context.
GraphQL filled the impedance mismatch gap by delivering flexible data structures with a more intuitive query language. The fact that the industry has adopted it so widely speaks volume of the strong need and value of a flexible data layer.
Molecule fills the same gap by offering a similar intuitive way of handling flexible data structures.
Instead of a separate query language, Molecule offers you to use your domain Data Model attributes directly as your custom query language with plain Scala.
Molecules can match the exact data needs of business processes, from case to case.
A “domain model” can be seen semantically as 3 layers, each supporting the layer above:
Attributes are the slowest changing part of a domain model since the intrinsic definitions of say an “atomic” firstName attribute will rarely, if ever, change. We define those attributes in a Data model.
In the layer above, groups of attributes form entity data-structures, or “molecules”, that are tailored to the exact needs of business processes of the top layer.
Molecules allows this more fluid and flexible “Molecular Domain Modeling” that can be tailored to the exact requirements of high-level processes in each context/use case.