Ideologists & theorists don't need to face the consequences of their failures. They live far away from it. As engineers, when we fail, it has a direct impact on our lives & career. We learn & adapt This is why, like many, I am very self-critical & I've started to appreciate it

JavaScript/TypeScript vs. Rust Side-by-side comparison of the same programmer logic error that caused recent Cloudflare outage & React breaches: Unsafe access patterns on nullable objects: obj! vs. obj.unwrap() Found in: 5% of all JS/TS code vs. 15% of all Rust code

Meet your customers where they are! - ClickHouse edition:

I'm hopeful that said "rules" will either be made consistent or abolished altogether. On the surface, it implies that AI code is responsibility of contributors & wont be treated differently, but without holding anyone accountable for failures, this will only lead to overloading

Why is AI surrounded by obnoxious duality? Linux Kernel: AI is fine in low level core code, no disclosure needed even GNOME: Blanket bans AI in high-level, user opt-in & manually installed extensions that aren't even part of it's core x.com/phoronix/statu…

Cascading Bloom Filters - they are amazing for storing vast amounts of "flag" data in a surprisingly compact & efficient size, serializable for offline usage with blazingly fast local queries & tiny delta patches for updates e.g. - track which urls/domains are definitely bad -

🥲

This is MOST UNDERRATED Safety feature of Fil-C. People don't realise the amount of bugs & vulnerabilities that are introduced due to unsafe FFI bindings, instead they focus on which side of the FFI to put the blame on rather than solving the issue at hand. You know like we're

Using Ray Tracing to speed up SQL queries by 18x over SOTA GPU Query Engines was not on my 2025 Bingo Card Mind boggling 🤯