VarBear

Go Developer Survey is out - 5,379 Go developers surveyed

Here are my 5 most important observations:

1. Go developers live in multi-language worlds.
Most also write Python, Rust, TypeScript, or Java. Friction appears when Go's idioms sharply diverge from familiar patterns.

2. Best practices are the 1st pain point.
Neither performance, nor syntax, developers want clearer, more opinionated direction on idiomatic Go.

3. Go's strength is boring reliability (That's the intentional branding)
Developers want stability, clarity, and fewer surprises. The survey reinforces that bias.

4. The standard library is powerful, but under-explained.
Go succeeds because of stdlib + tooling, yet many devs feel they aren't using it to its full potential.

5. AI is best at reducing toil, not writing systems.
Boilerplate, tests, refactors, explanations => Yes.
Core logic and architecture => No. It's still a human territory.

Boring, done well, scales. That's Go's superpower and the survey confirms it.

Have a great week!
Aymen

Copilot tuned like a real maintainer, a viral bot running into auth reality, and Postgres 18 tricks that shave IO without new tables. If private AI, soft delete without the mess, and the bigger question of machines writing code are on your mind, the details are inside.

🤖 How I Taught GitHub Copilot Code Review to Think Like a Maintainer
🚨 OpenClaw - Former Moltbot, Former Clawdbot - Went Viral Overnight. Then Security Reality Hit.
🏠 Run a Private Personal AI with Clawdbot + DMR
🧠 Software engineering when machine writes the code
🗑️ The challenges of soft delete
🐘 Unconventional PostgreSQL Optimizations

Stay sharp and ship.

Until next time!
FAUN.dev() Team

OpenClaw just dropped the corporate mask. New name, new direction. Support now spans Twitch, Google Chat, plus toys like KIMI K2.5 and MiMo-V2-Flash. Still all in on self-hosting, maybe even more so.

Version 2026.1.29 patched a nasty auth bypass flaw (CVE-2026-25253). Gateway connection checks are tighter now. Better late than never!

Uncover how to level-up your GitHub Copilot and VS Code experience from an autocomplete assistant to an intelligent, agentic teammate that can navigate your codebase, execute tasks, reason across files and even manage your GitHub projects.

In Building with GitHub Copilot course, you're not just learning how to use GitHub Copilot. You're exploring a shift in how we write, reason about, and collaborate on code.

Get your copy today!

Vibe coding has made contributing to open source easier, but the high number of contributions to the AI agent framework goose has posed a challenge. An AI Code Review agent like Copilot can help review PRs, but tuning its feedback is crucial for reducing noise and increasing value. By providing clear instructions, prioritizing key areas, and giving project-specific context, Copilot can become a valuable asset for code reviews.

"Soft delete" sounds gentle. It isn't. Slapping a deleted_at column on every table pollutes queries, drags down migrations, and leaves tombstones all over production.

This post digs into saner options: PostgreSQL triggers, event archiving in the app layer, and CDC via WAL. Each separates the dead stuff from the living, keeping OLTP lean - but they come with their own setup costs and infra quirks.

Clawdbot just plugged into Docker Model Runner (DMR). That means you can now run your own OpenAI-compatible assistant, locally, on your hardware. No cloud. No per-token fees. No data leaking into the void!

PostgreSQL 18 now supports virtual generated columns, indexable expressions without burning storage. Perfect for standardizing queries in analytics-heavy pipelines.

Pair that with planner constraint exclusion (constraint_exclusion=on), and Postgres can dodge irrelevant table scans based on constraints. Less IO. Faster ad-hoc BI queries.

In 1968, computer scientists identified the "software crisis" - the existing methods of programming were struggling to handle the power of computers. Today, AI coding assistants are accelerating productivity, but concerns arise about understanding the code they generate, the implications for debugging, and how to preserve deep learning in an age of AI assistance. The Jevons Paradox is at play, with more efficient coding tools leading to more code and increased system complexity.

Take a break, and get a coffee! Warm your soul with a nice mug perfectly sized black ceramic mug.

Jan is an open source alternative to ChatGPT that runs 100% offline on your computer.

An extremely good ASCII art generator, based on machine learning

A modern, high-performance DNS scanner with a beautiful Terminal User Interface (TUI) built with Textual. This tool can scan millions of IP addresses to find working DNS servers with optional Slipstream proxy testing and automatic multi-platform client download.

A curated list of Python tools, libraries, and frameworks built with Rust.

Get 10X more out of Claude Code, Codex or any coding agent

Did you know that FoundationDB tests every change inside a fully deterministic cluster simulator? A feature called buggify randomly injects crashes, network failures, and disk errors, all replayable from a single seed, so rare distributed bugs can be reproduced exactly. Contributors must prove their code works in simulation, not just in unit tests.

"Types prove what code cannot do; tests prove what it sometimes does; neither verifies the problem is the one that matters. Tools are exact about the question you asked, so make the spec stricter than the linter."
SenseiOne

Most mid-level devs who stall are the ones who never build the quiet habit of reading other people’s PRs in detail and leaving specific, technical comments about tradeoffs, failure modes, and naming. If you do this consistently, you start seeing patterns in how things go wrong and how systems evolve before you write your own code, which changes what you propose, how you argue for it, and who starts pulling you into the harder problems.