Features
Functional programming
Functional programming is a programming pradigm where functions are first-class values, which means functions can be passed as arguments to other functions. That allows programmers to express programs by combining functions into more complex ones.
(def (foo func) (func 34))
(def (bar num) (+ num 8))
(print (foo bar)) ; Prints "42" on terminal.
In such paradigm, it is encouraged to use pure functions, which have no side effects as they just compute their results from given arguments but must not do anything else. And the counterpart is called impure functions, which have certain side effects. Therefore, Cloe distinguishes these 2 types of functions in order to prevent impure functions from being called in context where only pure function calls are permitted, and vice versa.
Immutable data
As some other functional programming languages, Cloe prevents variables from being mutated because that makes programs more predictable and functions easier to test while they might be redefined shadowing previous ones.
(let x ["foo" "baz"])
(let y (insert x 2 "bar"))
(print x) ; -> ["foo" "baz"]
(print y) ; -> ["foo" "bar" "baz"]
Lazy evaluation
Lazy evaluation is a strategy to evaluate and run programs in functional programming. In most programming languages, pieces of programs are evaluated in an eager way; operations there come in order and function calls are evaluated just when called. However, in lazy evaluation, calling a function does not mean executing it right away but stores in memory a pair of the function and its arguments which may or may not be evaluated into its result later.
Adopting such program evaluation strategy, Cloe gains capability to blend data which become available at different time. For instance, Cloe can get incoming HTTP requests which a server receives from its launch to its termination in a list all at once. Or, it can express an endless stdin byte stream as an infinite list of bytes and process it just like other lists using common list manipulation functions, such as map and reduce.
Parallelism & concurrency
Parallelism and concurrency have been complicated challenges in a lot of programming languages for many years. Cloe provides implicit parallelism and concurrency while some parts of your programs may need to be made more parallel manually. Therefore, programmers do not usually have to care about them as much as other programming languages.
Parallelism
Top-level expressions in Cloe are evaluated parallelly, which is how it parallelizes components of programs automatically. However, it may not be enough or even excessive.
For that cases, Cloe provides 2 primitive functions of par and seq with which you can control parallelism of your programs. The former par evaluates given arguments in parallel while the latter seq does that sequentially.
Concurrency
When it comes to concurrency, software built on top of Cloe is always concurrent thanks to its host language, Go.
Reactiveness
Given implicit parallelism, programs written in Cloe are fully reactive. The output of the following program:
(print 123)
(print 456)
can be both
123
456
or
456
123
as the top-level expressions, (print 123) and (print 456) are evaluated at once and their side effects can happen out of order.