17 releases (6 breaking)

new 0.6.0 Aug 31, 2024
0.5.0 Jul 30, 2024
0.4.0 Jun 29, 2024
0.1.0 Mar 29, 2024
0.0.5 Jul 7, 2023

#345 in Programming languages

Download history 5/week @ 2024-05-20 147/week @ 2024-05-27 33/week @ 2024-06-03 11/week @ 2024-06-10 139/week @ 2024-06-24 18/week @ 2024-07-01 377/week @ 2024-07-29 3/week @ 2024-08-12 191/week @ 2024-08-26

196 downloads per month

MIT/Apache

470KB
16K SLoC

The Air Programming Language

The Air programming language is carefully designed to solve programming problems once and for all.

It is an experimental proof-of-concept project and is still in the very early stages of development.

Goals

The Air language seeks to solve programming problems once and for all. It should be able to

  • express any describable information, such as requirements and implementations, problems and solutions, propositions and proofs.
  • provide any information about the language and the program itself.
  • implement any theoretically possible information processing requirement, such as implementing requirements, answering questions, and proving propositions.
  • use information about the language and the program itself to perform property proofs and performance optimizations, achieving the best properties and optimal performance.
  • provides stable syntax and semantics, allowing users to learn the programming language once and for all.

Non-Goals

  • No design choices are taken for granted, and language features are not copied from other languages without review.
  • Suboptimal designs are not chosen to accommodate user habits.
  • Solutions that only solve most but not all problems are not satisfactory.
  • Impossible tasks are not attempted to be implemented.
  • The language is not constantly updated to implement more requirements.

Design

  • Decouple syntax from semantics, making syntax available as a general data exchange format.
  • Build a concise semantic core and provide rich initial context.
  • Allow functions to access context, which means that control statements are just functions that can access context.
  • Implement a universal logical framework based on computability theory, replacing type systems based on type theory.
  • Implement a universal problem framework based on reverse computation theory, used to express any describable requirement or problem, replacing interface/trait systems.
  • Implement a universal algorithm framework based on complexity theory, attempting to achieve artificial general intelligence.

Demo

"A demo of implementing a C-like for loop function" @ ; ! [
    c_for = function ! {
        input_name : .args,
        context_name : .ctx,
        context_access : .mutable,
        input_mode : id,
        prelude : prelude ! .,
        body : ; ! [
            [init, condition, next, body] = args,
            ctx | form ! ; ! [
                .&init,
                .&condition while [
                    .&body,
                    .&next,
                ],
            ],
        ],
    },
    c_for [[i = 1, sum = 0], i <= 10, i = i + 1, sum = sum + i],
    sum
]

Dependencies

~4–13MB
~164K SLoC