This is a study of how to use Clojure protocols and records to invert the dependency injection.
Explanation.
In prot.clj we define two protocols that we want to implement for our project.
In data.clj we define a data type as record, and also create an implementation of the first protocol.
In tools.clj we create implement the other protocol, Prot2 for the record type Data1.
In tools2.clj we create another implementation of the protocol Prot2 for the record type Data1.
In code.clj, we want to use the protocols for processing the data received as argument. The whole purpose of this is to be able to call the protocols without having to include the actual implementation in tools.clj. As we can see from the namespace declaration, the implementation is not included in the namespace requirements.
In core.clj is the entry point of the application. The implementation of Prot2 used by the the function in code.clj is chosen here. You can switch between the namespaces tools and tools2 in the require statement.
The whole point of this demonstration is to show, how to write code.clj without having to explicitly require/use either of the tools or tools2 namespaces.
This repository has been tested for three tools:
- Creating a JAR with
clj -T:build uberand running withjava -jar target/clj-prot-standalone.jar \"hello\" - Running the REPL from Emacs CIDER
- Running with
clj -M -m core \"Hello\"