The Ferrum Programming Language

Note: This language is a work in progress. The designs and implementations are unstable.

Another Programming Language?

Ferrum is meant to be a simple programming language, batteries-included, for general software development.

The syntax is simplified, the APIs are more constrained, and there are built-in standard solutions for many common problems.

Details of the language are not yet well documented. Sorry about that!

Some Key Features

• Runtime panics are only possible from within risk functions.
  • This makes it very easy to find problematic call-stacks when debugging runtime panics.
• Simple basic types such as Integer, Decimal, and String.
  • Ferrum's compiler will ensure that the most optimal types are used under-the-hood.
  • Type conversion (ie. Integer to Decimal) will happen implicitly whenever possible.
  • Integer supports ints of any size (if no constraints are set), and will fallback to a big-int if necessary
• Compile-time type constraints.
  • For example, String<$Length=3> can implicitly convert to String<$MinLength=2, $MaxLength=5>, but not the other way.
  • This encourages developers to do input validation as soon as possible, allowing the compiler to optimize as much as possible.
• No need for async/await.
  • Functions that should be async, will be. These functions will be awaited when called. There are separate mechanisms to deal with background tasks or awaiting multiple tasks.
• Compile-time evaluation.
  • Code that can be calculated at compile-time, will be.
• ? for optionals, ! for results.
  • ?String is the type reference for an optional string
  • !Integer is the type reference for a result of either an int or some error
  • !<Bool, String> is the type reference for a result of either a bool or a string error
• No lifetime syntax.
  • Ferrum uses Rust's ownership model, but doesn't rely on Rust's lifetime syntax. Instead, sane defaults are used.
  • If multiple references are passed into a function, and a reference is returned from the function, then you can optionally specify a lifetime for the return.

    // The ' lifetime is optional here, but allows the msg reference to not be tied to a, b, and return-type
    fn largest(a: &'String, b: &'String, msg: &String): &'String
        print(msg)
        return if a > b then a else b
    ;
    

• $ for shared ownership.
  • The compiler will do it's best to choose the best implementation for shared ownership. This may be a shared static-reference, reference-counting, or garbage collected.
  • The compiler will determine if atomic synchronized types are necessary or not.

    const shared_name_1: $String = $"Adam"        // Create shared-ownership object
    const shared_name_2: $String = $shared_name_1 // Immutably share ownership
    const shared_name_3: $String = $shared_name_1 // Immutably share ownership
    
    mut shared_name_1: $mut String = $mut "Adam"        // Create mutable shared-ownership object
    mut shared_name_2: $mut String = $mut shared_name_1 // Mutably share ownership
    const shared_name_3: $String = $shared_name_1       // Immutably share ownership
    

• #[] for sets, #{} for maps.
  • Sets are a list of unique items: #[1, 2, 3, 3] == #[1, 2, 3] // true
  • Maps are a map of unique keys to values: #{ "one": 1, "one": 1 } == #{ "one": 1 } // true

Demo

Hello World

use fe::print

fn main()
    print("Hello, World!")
;

For more examples, see examples