Romain Manni-Bucau

@rmannibucau.bsky.social

ASF/Java/.NET/BigData/Js LinkedIn: http://bit.ly/2GIh7Bb ASF: https://s.apache.org/rmannibucau Javaccino founder (java/.NET/K8s service) .NET Blog: https://dotnetbirdie.github.io Java/Js Blog: https://rmannibucau.github.io Book: http://bit.ly/4dVRmvh

Encountered a weird NoSuchMethodError when invoking an outdated dependency. Decide to look into its source code. Let’s get the list of dependencies … It uses Gradle as build tool 😱 With .kts files 😓 Mixes Java and Kotlin in source code 🤬 FFS (╯°□°)╯︵ ┻━┻

Today is the first time an article I wrote by hand was flagged as "clearly AI-generated" by someone... and it hurt. It took me several hours to write an article such as "What's new in Java 27" (6-8h approx). 🧵1/2

I keep seeing AI cost charts that compare tokens to salaries and quietly skip the rest of the SDLC. This piece argues the real break-even point is review, repair, and delivery overhead. That is where the human bill comes back. buff.ly/eiJ7drT

Bild

You have too much (LLM) tokens you want to consume in a single session? Ask it to implement a cloudflare key based workflow providing it a HAR....it will think forever and just give you wrong insights....not sure why the original analyzis is so wrong without full guidance...it is a simple HAR

Des événements comme Devoxx France ne se construisent pas seuls ! Alors merci à tous ceux qui ont rendu cette édition 2026 aussi exceptionnelle, nos Sponsors 2026 : www.devoxx.fr/sponsors/spo... Envie de faire partie de l'aventure Devoxx France 2027 ? 👉 sponsor@devoxx.fr

Bild

Phillip Kruger pointed me to a small Quarkus Agent MCP test thread, and it says more than most benchmark decks. My takeaway: skills earn their keep when they cut wrong turns, retries, and stale framework guesses. buff.ly/G2ck1iJ

Bild