To me, that article appears weak on the merits, resulting in what’s basically a big and complicated ad hominem argument.
If one wanted to position oneself in favor of a sustainable paradigm shift, I’d find this piece way more convincing: honnibal.dev/blog/ai-bubble
Are you confused by Gradle terminology? What's a module, a configuration, a build, a project? Unless you're deep into Gradle, they probably aren't what you think.
I wrote an article to organize all of that → ivan.canet.dev/blog/2026/03...
Wir müssen reden …
über Fun Facts
Hand aufs Hirn: Ist euch das Lachen (schon) vergangen? Mit Blick auf die Nachrichtenlage und in die Welt. Iran, Israel, USA, Russland, Ukraine, Pakistan, Afghanistan - 🧵1/8
(c) Clemens Porikys
I love it! ❤️
That's your bike? Did Louis convert you? 😃
Android testing at scale is either fast or easy, never both, right?
Integrating a modern DSL-based Kotlin test framework with class-based Roboelectric was considered impossible by genAI and others. Yet here it is: Roboelectric meets 🎈 TestBalloon
infix-de.github.io/testBalloon/...
Guidance I have been waiting for. 🎉👏❤️
Arguing about unit tests vs. integration tests sucks. It's dogma and I feel like a heretic.
This post discusses a different way to think about a test’s scope. I love it.
Good points in the article. Let's not get fooled by false dichotomies.
Note to self and other folks working on test infra: Make effective testing easy, regardless of scope. Simplify all kinds of high-value tests, so these will actually get written.
Signing up on LookupTracker. Adding ICFileMappingTracker.
Did you check this one?
youtrack.jetbrains.com/issue/IJPL-2...
Also, the introductory docs („why might I want to use this?“) are superb! 👏
I haven’t checked, but the more platforms to cover, the more it might call for a rewrite in Kotlin. 😉
If it’s JMH only, I’d agree. But you mentioned web…
Why not integrate with TestBalloon where suspend functions are readily available in tests everywhere?
genAI: "This PR reverts the attempted Robolectric integration due to fundamental architectural incompatibility between TestBalloon and Robolectric."
Human: "Hold my beer."
github.com/infix-de/tes...
Whoa! I'm just at Saoirse. A lot more to learn. 😆
Banner for KotlinConf 2026, announcing the talk "TestBalloon: Kotlin testing is easier (and more fun) than you think", with Oliver Okrongli and Bernd Prünster speaking
Grateful and happy to be appearing on stage at @kotlinconf.com in May! Looking forward to seeing you all in Munich.
Informative and fun – thanks a lot! ♥️
Tried with TestBalloon. Could look better, but works:
Gradle 9.3 added support for nested tests in reports.
TestBalloon 🎈 0.8.0 is out, the Kotlin-first test framework. Conciseness and flexibility meet JUnit*-like testing powers on all platforms: Scope-friendly, anti-boilerplate DSL, function calls create tests (parameterizable), nesting fixtures, all easily customized. ♥️ Yours? github.com/infix-de/tes...
Thankfully, that section is now gone with FIR replacing Gradle-based code generation. But there is still another one working around compiler bugs. And one `gradle.taskGraph.whenReady`, which is not that much better, but puts TestBalloon in the lead (running nested tests individually with std tasks).
I guess the entire experience is always in the context of the world as it exists.
Actual use case regarding source sets (otherwise the same problem): github.com/infix-de/tes...
An example: Try to configure a task in a way that depends on another task *not* being there.
Implicit ordering. Schrödingers tasks. We cannot depend on what may be there or not there.
Old school AI. 🙄
Yeah, but power grids are usually not that vulnerable here.
Seems like they could have taken care of insisting on sufficient network redundancy long ago. Having a single point of failure is definitely not state of the art.
