DesignNotes.txt

Explanation of decisions and design
===================================

- Used Gradle (dependency management, build and test) and a mix of Java and Kotlin for this solution. 
  Java is my most familiar language and I used Kotlin where Java has limitations - for tail-recursion 
  optimisation and for the ZIP stream function.

- Implemented the Collatz series with tail-recursion in Kotlin which optimises tail-recursive functions.
  Used BigInteger for the input, output and intermediate term values. This allows arbitrarily large 
  series tests without risk of stack overflow or number out of range problems.

- I chose Spark Java (sparkjava.com) as the HTTP server framework for building the REST API Controller. 
  It is a lightweight framework/library which can be dropped into the application. I prefer this over 
  heavier enterprise frameworks such as Spring Boot where possible (although I'm happy to use 
  Spring Boot when required). Spark has low memory usage , see: 
  https://craftsmen.nl/memory-usage-6-popular-rest-server-frameworks-compared/ 
  and has performance comparable to it's underlying Jetty server. 
  If required, the Controller implementation can be swapped out with another server implementation 
  without affecting the rest of the implementation.
  
- I chose the REST Assured test framework for integration testing of the REST API. The syntax is 
  declarative and concise and the tests can be automated as plain JUnit tests and run
  with Gradle.
  
- Collatz term calculations are delegated to a Thread Pool Executor and I chose a synchronised 
  Map for caching results as this allows safe multi-threaded access.