Skip to content

ApplePy vs PyO3

ApplePy takes direct inspiration from PyO3, adapting its patterns for Swift. Here's how they compare.

Macro Comparison

Concept PyO3 (Rust) ApplePy (Swift)
Functions #[pyfunction] @PyFunction
Classes #[pyclass] @PyClass
Methods #[pymethods] @PyMethod
Static methods #[staticmethod] @PyStaticMethod
Properties #[getter] / #[setter] @PyProperty
Module #[pymodule] @PyModule
Enums Class-based @PyEnum → IntEnum

Side by Side

import ApplePy
@preconcurrency import ApplePyFFI

@PyFunction
func add(a: Int, b: Int) -> Int {
    a + b
}

@PyModule("mylib", functions: [add])
func mylib() {}

use pyo3::prelude::*;

#[pyfunction]
fn add(a: i64, b: i64) -> i64 {
    a + b
}

#[pymodule]
fn mylib(m: &Bound<'_, PyModule>) -> PyResult<()> {
    m.add_function(wrap_pyfunction!(add, m)?)?;
    Ok(())
}

Memory Safety

Aspect PyO3 ApplePy
GIL enforcement 'py lifetime (compile-time) GILGuard: ~Copyable (compile-time)
Object ownership Bound<'py, T> PythonHandle
Concurrency Send + ?Sync bounds @PythonActor global actor
Ref counting Automatic via Py<T> PyBridge ARC ↔ refcount

Build Tooling

Aspect PyO3 ApplePy
Build tool maturin applepy CLI
Scaffold maturin new applepy new
Dev install maturin develop applepy develop
Wheel build maturin build applepy build
Publish maturin publish applepy publish
Package manager Cargo Swift Package Manager

Key Differences

Platform

PyO3 is cross-platform (Linux, macOS, Windows). ApplePy works anywhere Swift runs — on macOS, it additionally unlocks Apple frameworks (CoreML, Metal, NaturalLanguage, Security, etc.) that no other Python interop can access.

Language

Rust has zero-cost abstractions and no runtime. Swift has ARC and a rich standard library. Both have excellent type systems and compile-time safety.

Ecosystem

PyO3 has a mature ecosystem with 15+ type conversion crates. ApplePy provides built-in conversions for primitives, collections, tuples, and optionals, with the entire Apple SDK as its "ecosystem".