ArxLang

Arx is a multi-purpose compiler that aims to provide native list, tensor, and dataframe abstractions backed internally by IRx runtime support. It uses the power of LLVM to provide multi-architecture machine target code generation.

The language syntax is influenced by Python, C++, and YAML, featuring significant whitespace, static typing (planned), and a focus on data-oriented computing.

Quick Example

```
title: Quick average example
summary: Demonstrates a function with module and function docstrings.
```
fn average(x: f32, y: f32) -> f32:
  ```
  title: average
  summary: Returns the arithmetic mean of x and y.
  ```
  return (x + y) * 0.5

arx --show-llvm-ir examples/average.x

See the Getting Started guide for installation and more examples. For language details, see the Library Reference.

Architecture Boundary

Arx is the surface-language front end. It owns source syntax, lexing, parsing, CLI flow, tests, examples, and docs. IRx owns the AST model (astx), semantic analysis, lowering, and backend code generation. When a new feature needs AST or lowering support, that support should be added in IRx first and then consumed from Arx.

Key Features

• LLVM-powered – compiles to native machine code via LLVM
• Python-like syntax – indentation-based blocks, familiar keywords
• Lists, tensors, and dataframes – generic collections, Arrow-backed numeric tensors with compiler-known shapes, and Arrow-backed named-column DataFrames
• Multiple output modes – inspect tokens, AST, LLVM IR, or compile to object files

Project Status

Arx is currently a prototype built on the Kaleidoscope tutorial compiler. It supports functions, control flow (if/else, for), variables, and extern declarations. See the Roadmap for what’s planned next.

Arx Enhancement Proposals

Any change to the language syntax should be done using an Enhancement Proposal via the arx-proposals repository.