Jul 15, 2014
Join Mark, Greg, and Peter for some more absurd type level rants, whilst we weren’t exactly planning on diving once again into a big argument around types and functional programming, somehow we fell afoul of one almost immediately even before we got around to introducing the episode.
The argument starts with an argument over whether one can do “functional programming” in Java, and the difference between “functional programming” and “a functional programming language”. In this argument Greg claims that memory allocated for a class instance constitutes the state of a class — even if the class structure holds no state, and that ‘functions’ don’t allocate memory when defined.
The main discussion for the evening relates to an Ars Technica article about how much information you should pass in a function: a rich object structure or individual arguments.
During the discussion, Mark suggested that simply using
String as the type of the argument is probably a bad
design as you may want to restrict the context of the arguments,
such as using a specific
Password type, this
ensued with an argument over deep type hierarchies (do I smell a
theme here?) to which Mark countered this by (badly) describing a
way of using parametric typing (and phantom typing) to improve the
contract of passwords in the API.
Unfortunately the discussion got sidetracked with factories and PK7 implementations being tightly woven into the which probably detracted from any useful discussion.
Escaping away from type theory, Peter brings us back to the real worldwith a discussion on Web Fragments and JBoss Hell…