Y#
Y# logo
Back to home

Y# Updates

The official changelog. Every release is documented — performance, features, fixes.

  1. v8.0.4

    « Oyster Shell »patchJune 6, 2026

    • Windows installer v8.0.4

      New stable Windows x64 build: faster startup, refreshed signature, and the full toolchain (oys, yo, formatter, LSP) bundled.

    • Runtime fixes

      Print/PrintLine stabilization on very small floats, fix for a yo package manager crash on a corrupted cache, and better handling of Windows paths with spaces.

    • Richer diagnostics

      Compiler error messages now include causes, faulty examples and inline suggested fixes.

    • Extended stdlib

      New file/console I/O APIs (CursorPos, ClearScreen, FileReadLines) plus updated std/server, std/web and std/ai helpers.

  2. v8.0.1

    « Oyster Shell »majorJune 2, 2026

    • Y# v8 compiler

      Complete pipeline: Lexer → Parser → Typeck → HIR → MIR → Codegen. 38 AST node types, Hindley-Milner inference.

    • Multiple backends

      Native C (.exe via GCC), Game (C++ ECS, .ysg) and GPU (SPIR-V 1.6 .spv) all active.

    • Polymorphic Print

      Print and PrintLine now accept every primitive type with compile-time dispatch.

    • yo package manager

      yo install / remove / publish. Deterministic resolution via oy.toml.

  3. v7.4.0

    minorApril 15, 2026

    • MIR Optimizer

      Const folding, LICM, SIMD auto-vectorization, block reordering and aggressive inlining.

    • Actor model

      Actors with State<T>, On(Event) and Reply — lock-free concurrency.

  4. v7.2.1

    patchMarch 21, 2026

    • Parser fixes

      Nested comments /* /* */ */, floats with exponent, multi-line strings.

    • Lexer performance

      logos tokenizer 2.3× faster on large files (> 10k lines).

  5. v7.0.0

    « Pearl »majorJanuary 10, 2026

    • Native ECS

      Component, Entity, System as first-class citizens. Built-in templates: Transform, RigidBody, Mesh, Camera.

    • Tensor & Autodiff

      Tensor<Float, [N, M]> with differentiable Function for GPU-compiled ML.

    • Standard library

      std/core, std/ai, std/game, std/server, std/web — essential modules ready to use.