yon

KO | EN

gitlite — search

#javascript #ai #go #golang #android #rust #deep-learning #docker #macos #java #nodejs #awesome #security #openai

(★ 25)

An experimental language whose object system is a topos: places are pre-sheaves, values are content-addressed on the Leech lattice, and the type checker refuses to compile a false proof.

yon Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Yon

**Yon** is an experimental, topos-oriented programming language: a
compiled language whose semantics are drawn from category theory
(elementary topoi, Heyting algebras, directed type theory) and whose
runtime stores every value in a **content-addressed heap built on the
Leech lattice Λ₂₄**, with canonicalization under the Conway group Co₀.
Same content, same address: equality is one machine comparison,
deduplication is global, and the heap chain grows without tuning knobs.

The pipeline: OCaml frontend → custom MLIR "topos" dialect → LLVM →
native binary, linked against a self-contained C runtime (xleech allocator with the vendored mmgroup mathematical core).

## Status

**1.0 baseline**. The ground truth is the
regression suite: **112 examples + 2 cross-Space scenarios**, exit
codes identical across platforms.

Platforms: Linux x86-64 and macOS Apple Silicon; the regression suite runs green on both.
macOS Intel: untested (expected to work; reports welcome).

## Install

One line (macOS arm64, Linux x86-64):

```bash
curl --proto '=https' --tlsv1.2 -sSf https://yon-lang.org/install.sh | bash
```

It checks prerequisites (and prints what to install if something is missing),
builds the three stages from source, and drops a self-contained toolchain into
`~/.yon` with `yonc`, `yon-lsp`, `yon-fmt`, `yon-lint` and `yon-pkg` on your
`PATH`, plus the VS Code / Cursor extension. Re-runnable; pass `--check` (as
`| bash -s -- --check`) for a dry prerequisites report. For the manual,
step-by-step path, follow `INSTRUCTIONS.md`.

## Build

Follow **`INSTRUCTIONS.md`** — the single authoritative walkthrough
(runtime, OCaml frontend, MLIR dialect,
then the regression). The acceptance criterion on every platform:

```
cd regression && ./run_regression.sh
→ REGRESSION OK: 112 examples, identical to the baseline.
→ CROSS-SPACE OK: 2 scenarios
```

House rule: run the regression before every push. No silent
degradation, in code or in process.

## First program

```bash
toolchain/yonc examples/v1_control_flow.yon -o prog && ./prog
```

The book (23 chapters + appendices, every snippet compiled before it
was written) lives in `website/docs/book/` and at **yon-lang.org**.

## Packages

Go-style, registry-less: a package is a git repository with a
`yon.toml` at its root. In your manifest:

```toml
[dependencies]
geometria = { pkg = "user/geometria", version = "1.0" }  # any GitHub account
json      = { pkg = "json" }                             # bare name → yon-language/json
```

`pkg/yon-pkg install` resolves into `yon_modules/` and writes a
`yon.lock`. Official packages live under the `yon-language` organization;
anyone can publish under their own account.

## License

- **Compiler, frontend, MLIR dialect, toolchain: AGPLv3** (`LICENSE`).
  Anyone who modifies Yon or serves a modified Yon over a network must
  publish their changes.
- **Runtime: AGPLv3 + linking exception** (`runtime/LICENSE`).
  **Programs you write in Yon are entirely yours**, under any license
  you choose — the exception exists precisely so that using the
  language never encumbers your code.

Third-party: LLVM/MLIR (Apache-2.0 with LLVM exception),
mmgroup (BSD-2-Clause), menhir-generated parser (free of constraints
by explicit exception).

yon

yon Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

yon

yon Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README