Vladimir Sitnikov

Well, Critical JNI seems to be able to sum 16 bytes at 60'000 ops/ms that translates to 1'000'000/60'000 == 16.7ns/op: https://stackoverflow.com/a/36309652/1261287

> Out of curiosity: is there a way to fix the config so that one can re-run the same config after a fix? Everybody asks it, and I never needed...

I have several jdks locally since some of the projects require Java 17, and some can't build with 17 :) The need to reproduce locally comes when you want set...

> especially considering that we plan to introduce a compiler plugin. I believe they complement each other. A compiler plugin can't contribute to IDE's autocomplete. In other words, the code...

> I don't mind the annotation approach https://guava.dev/releases/23.0/api/docs/com/google/common/annotations/VisibleForTesting.html 😉

It looks like the case for generating feature extractors for third-party code is pretty real: https://github.com/robstoll/atrium/issues/1421

I wonder if semething like jackon's mixin would be enough. Then the user could declare a dummy class and configure "theat its annotations as if they were specified at SQLException"

The fix makes ```kotlin expect(listOf(User("joe", "qwerty"), User("j", "qwerty1"))) { toHaveElementsAndAll { feature("login", { login }) .feature("size", { length }) .because("login should be longer than one letter") { toBeGreaterThan(1) } feature("password",...

> we should also provide a report option which allows to define if we show also successful expectations or not I guess that is yet another issue, and I think...

Yet another thing for `allWithMismatchClauseAndDecorator` would be to hide the top `createExplanatoryAssertionGroup` in case the list contains exactly one element. When the actual list has one element only, then there's...