Evalution model

Clojure’s evaluation model differs from the more ‘standard’ programming languages in that it has two phases: read phase and evalution phase. For example to process (+ 1 2) Clojure first reads this text and converts it to a list data strucutre. This is then passed to the evaluator which looks up the function correspondind to + and applies it to 1 and 2.

Languages that process code in this way are called homoiconic. They allow us to reason about the code as a set of data structures that are manipulated programmatically.

Compiled languages vs Clojure

Programming languages that are compiled or interpreted need to convert our code as text to machine instructions. They do that by converting our code into an abstract syntax tree (AST) which is a data structure that represents our program. In many programming languages AST is inaccessible within the programming language.

In Clojure this AST is a tree structured using Clojure lists, whose nodes are values. This allows us to change the AST in our program. This also means the evaluator doesn’t care where its inputs comes from, it can come from the reader of we can supply it directly using eval.

1(def addition-list (list + 1 2))
2(eval addition-list)
3; => 3

At the end, this means we can manipulate the code after it has been read by the reader allowing us to do many interesting things. Code that does these manipulations is called macro.