Sweet, @github released a new API for fetching stargazer info. Thanks @cassidoo and co. Swapped over Best of Go and it’s now chugging along (again). #golang bestofgo.dev/

Nice! @duckdb adds "DuckDB as a server" in v2.0 What a fantastic project. duckdb.org/2026/08/17/duckdb…

I shut down bestofgo.dev because @github restricts the APIs needed to pull data. Fun while it lasted; on to other projects. github.blog/changelog/2026-0…

These are kinda cool. I like taking photos and maybe AI can give some pics a different way to enjoy them. Tried the various prompts.

Okay, I think this opens opportunity for something useful: type FlagName[T any] string const verbose FlagName[bool] = "verbose" - defined once and reused - typed - s.Get(verbose) without specifying [bool] github.com/pressly/cli/issue…

Introducing protobuf.com a visual, interactive guide to Protocol Buffers.

The world needs a way off GitHub Actions. For the past three weeks, we’ve been building an open-source compatibility layer that runs Actions workflows natively on @buildkite without changing them. Buildkite powers CI for repos where scale is table stakes. Pipelines are dynamically programmable and parallel by default, with every job running in its own environment. Test suites that take hours - or days - on other tools can fan out and finish in minutes. Actions uses a different model: static workflows, with sequential steps sharing a mutable workspace. We now transparently compile that model into our Buildkite pipelines, so teams can keep their existing workflows and actions while gaining Buildkite’s speed at scale, control and reliability. It’s early and there are rough edges, but it's already proving useful across our design partners. If you want to start shifting workloads, give it a go: github.com/buildkite/buildki…

Interesting, there's still quite a bit of interest for improved @ClickHouseDB support in pressly/goose. github.com/pressly/goose/iss…

Crafting Interpreters is probably one of the best programming books I've ever read. It walks you through building a full language from scratch and quietly teaches: • lexing • parsing • tree-walk interpreters • bytecode and VMs • how to write C • memory management • garbage collection • hash tables • dynamically resizable arrays • NaN boxing • bit twiddling Written by Bob Nystrom (who works on Dart at Google). Read the whole thing here for free here: craftinginterpreters.com

Full post on the subject: brandur.org/fragments/pgtest…

Today, we’re open-sourcing the draft specification for DESIGN.md, so it can be used across any tool or platform. We’re also adding new capabilities. DESIGN.md lets you easily export and import your design rules from project to project. Instead of guessing intent, agents know exactly what a color is for and can even validate their choices against WCAG accessibility rules. Watch David East break down this shared visual language in action👇. New capabilities and links in 🧵

Braun audio system designed by Dieter Rams. The complete set includes the following P1, T1, C1, CD2/3, A1 and the pedestal. De Mauricio

We processed a trillion transactions to test TigerBeetle’s scale...

🤯BREAKING: Alibaba just proved that AI Coding isn't taking your job, it's just writing the legacy code that will keep you employed fixing it for the next decade. 🤣 Passing a coding test once is easy. Maintaining that code for 8 months without it exploding? Apparently, it’s nearly impossible for AI. Alibaba tested 18 AI agents on 100 real codebases over 233-day cycles. They didn't just look for "quick fixes"—they looked for long-term survival. The results were a bloodbath: 75% of models broke previously working code during maintenance. Only Claude Opus 4.5/4.6 maintained a >50% zero-regression rate. Every other model accumulated technical debt that compounded until the codebase collapsed. We’ve been using "snapshot" benchmarks like HumanEval that only ask "Does it work right now?" The new SWE-CI benchmark asks: "Does it still work after 8 months of evolution?" Most AI agents are "Quick-Fix Artists." They write brittle code that passes tests today but becomes a maintenance nightmare tomorrow. They aren't building software; they're building a house of cards. The narrative just got honest: Most models can write code. Almost none can maintain it.

hot take: AI doesn't kill meaningful work, it kills meaningless work disguised as meaningful. most knowledge work is reformatting data between systems. the people who actually create things — design, strategy, taste — become more powerful, not less

There’s people that use AI and then there’s people that KNOW how to use AI, effectively. You can tell. And there’s a world where this doesn’t mean one-shot vibes.

Old CDs. Good games.

Sell me a model like software: $X/year, runs locally, updates, no token tax. Unlimited usage. Predictable spend. Metering per token is like charging per keystroke, just saying.