At sufficient scale everything fails all the time, and the only real feature is being able to watch it happen.

_{— me, to a 🦆}

I work on the metric backend of Checkmk, its data backend for OpenTelemetry metrics that lets it monitor applications, not just infrastructure.
Before that I was at SAP, building the CI/CD infrastructure for HANA (their in-memory database): a graph-based task execution framework running on ~2000 compute nodes spread across multiple cloud providers and regions.

since 2021            csshW: broadcast your keystrokes to all your SSH sessions (successor: cssh-rs)
PyConDE '25            Rustifying Python: what's worth moving to Rust, and what will go wrong

2026-08-27 15:21 CEST  Checkmk/checkmk            graph designer: derive the sum-metric…
(yesterday)            Checkmk/checkmk            _{467 commits}

2026-08-26 13:59 CEST  Checkmk/checkmk            graph designer: keep the stored…
(wednesday)            Checkmk/checkmk            _{467 commits}

2026-08-24 14:50 CEST  whmade/cssh-rs             Remove intermediate e2e connect-blocked…
(monday)               whmade/cssh-rs             _{565 commits · 116 pull requests · 145 issues}

2026-08-22 09:38 CEST  whme/bg3-name-your-summons docs: point the install step at the…
(last week)            whme/bg3-name-your-summons _{111 commits · 100 pull requests · 56 issues}

_{Last 30 days}

Lua 52.2% (14.4k) · TypeScript 11.0% (3k) · PowerShell 9.5% (2.6k) · Vue 9.5% (2.6k) · Rust 8.7% (2.4k) · Python 7.8% (2.2k) · Other 1.3% (360)

_{All time}

JavaScript 25.0% (71.3k) · Python 24.5% (69.7k) · TypeScript 14.3% (40.7k) · Rust 13.2% (37.7k) · Vue 10.9% (31k) · Lua 5.1% (14.4k) · Other 7.1% (20.3k)

_{Last updated 2026-08-28 09:04 CEST.}

_{Everything above refreshes every few hours via a small workflow. Here's how it's computed.
The 🦆 is maintained by hand.}