Developer Daily News - 2026-04-05 | DevRef

More Updates

LinuxDatabase

Linux 7.0 Causes Major PostgreSQL Performance Regression

AWS engineers report that PostgreSQL performance has halved on Linux 7.0. A potential architectural change in the kernel might require a complex fix.

AWS engineer reports PostgreSQL perf halved by Linux 7.0, fix may not be easy

AIOpen Source

Goose: An Extensible AI Agent Beyond Code Completion

Developed by Block, Goose is an open-source agent capable of installing, executing, editing, and testing code across various LLM providers.

block/goose

AIOpen Source

Onyx: Advanced Open Source AI Chat Platform

A robust, open-source AI platform featuring advanced chat capabilities that integrate seamlessly with any Large Language Model.

onyx-dot-app/onyx

AILLM

Andrej Karpathy Shares 'Idea File' for LLM Wiki

AI expert Andrej Karpathy released a gist containing a structured idea file for an LLM Wiki, providing insights into conceptualizing AI knowledge.

LLM Wiki – example of an "idea file"

ToolsOpen Source

OpenScreen: Open Source Screen Studio Alternative

A free, watermark-free tool for creating professional software demos, offering a high-quality alternative to subscription-based Screen Studio.

siddharthvaddem/openscreen

AIMachine Learning

Simple Self-Distillation Improves LLM Code Generation

New research demonstrates that a surprisingly simple self-distillation technique can significantly boost the code generation quality of LLMs.

Embarrassingly simple self-distillation improves code generation

NeovimTools

Major Development Shift: Nvim-treesitter Project Archived

The highly popular Neovim plugin Nvim-treesitter has been archived, signaling a significant transition for the Neovim ecosystem and community.

Nvim-treesitter (13K+ Stars) is Archived

AIApple Silicon

MLX-VLM: Vision Language Models on Apple Silicon

A specialized package for inference and fine-tuning Vision Language Models (VLMs) directly on Mac hardware using Apple's MLX framework.

Blaizzy/mlx-vlm

WebAssemblyAI

TurboQuant-WASM: Vector Quantization in the Browser

Google's vector quantization technology ported to WebAssembly, enabling efficient, high-performance vector operations directly in web browsers.

Show HN: TurboQuant-WASM – Google's vector quantization in the browser

EducationHardware

Mvidia: Educational Game for Building a Virtual GPU

An interactive, browser-based game that teaches developers the fundamentals of GPU architecture by allowing them to build one from scratch.

Show HN: A game where you build a GPU

Developer Daily

Top Stories

Apple Approves Nvidia eGPU Drivers for Arm Macs

Microsoft Launches Multi-Agent Workflow Framework