Piotr Kubowicz

A day on GitLab without a global incident becomes a rarity. Today: all your jobs start with an hour delay! Not to mention their status page acknowledged the problem 6 hours after first symptoms appeared.

Jeśli przespałeś ostatnie kilka lat i dalej uważasz, że GitLab CI ma sens komercyjnie albo dla open source – nie ma, zwłaszcza do integracji z AI ale nawet najprostszych automatyzacji łatwiej (i taniej!) użyć GitHub Actions pkubowicz.pl/5-powodow-czem…

Terminator: Rise of the Machines in your terminal – here's how OpenCode's Big Pickle model can ignore your instructions and edit files without you asking it to do so

Po regularnym używaniu do pracy z kodem mogę powiedzieć: europejski AI może być dobry, szybki i tani. Mistral Vibe i modele Devstral to ciekawa opcja. pkubowicz.pl/mistral-vibe-s…

Programmers and software developers lost the plot on naming their tools: The Hoover Dam is a dam, named after the president who commissioned it, not “Project Thunderfall” larr.net/p/namings.html

Krótki kod, który może ogromnie usprawnić reakcję na rate limiting: obsługa nagłówka Retry-After w bibliotece spring-kafka: pkubowicz.pl/spring-kafka-z…

"No consistency, no overarching plan. It’s like I'd asked 10 junior-mid developers to work on this codebase, with no Git access, locking them in a room without seeing what the other 9 were doing." albertofortin.com/writing/coding…

Sonar way: 3 days of requests timing out on production community.sonarsource.com/t/sonarqube-cl…

Stop writing that prompt and read how AI can slow you down (even if you aren't writing open source) johnwhiles.com/posts/mental-m…

@gradle Don't use lock files. It's a half-baked feature. The use is complicated. Many problems (like someone changing a version manually and not updating the lockfile) won't cause a build failure, but be silently accepted, affecting final application runtime. Buggy support in Renovate.

A nice update from our friends at #GitHub 🎉 #Dependabot support for #Gradle #lockfiles is now generally available! 👇 👇 👇 github.blog/changelog/2025…

5 Boring Things That Have a Bigger Impact Than AI Assistants on Dev Productivity codemanship.wordpress.com/2025/05/21/fiv… (news.ycombinator.com/item?id=440508…)

I'm tired of unpleasant surprises from the 'stable' feature, like the one here: #focus=Comments-27-11171756.0-0" target="_blank" rel="nofollow noopener">youtrack.jetbrains.com/issue/KTIJ-241…

So K2 is the default for Kotlin, but this does not mean you should use K2. K2 is officially still a work in progress, and using K2 means IDE won't allow some common refactorings/quick fixes (that work for Java or non-K2 Kotlin). I still keep K2 disabled.

Spring + Kotlin -> uwaga na wyjątki pkubowicz.pl/pulapki-transa…

ArchUnit vs Konsist - szczegółowe porównanie testów architektury dla Kotlina pkubowicz.pl/archunit-czy-k… #kotlin #architecturetests

Konsist czyli nowy gracz na rynku testów architektury dla Kotlina pkubowicz.pl/konsist-rzut-o…

jakarta.annotation i Optional.orElse nie działają z Kotlinem. Warto zabronić ich używania, na przykład używając ArchUnita pkubowicz.pl/archunit-prakt… #kotlin #archunit #architecturetests

Quartz to scheduler, którego nikt nie powinien używać ani polecać pkubowicz.pl/quartz-czyli-s…

Spring Boot 3.2.7 is the second release in a row that I am unable to use. Has a regression in Spring Data→Hibernate calls. 3.2.6 had broken spring-aop. I have been using Spring Boot for many years, but I don't remember such an awkward situation.