Peter Niederwieser

Kotlin is great for apps. However, using it to implement a foundational and language-agnostic library such as UncleJim is problematic: 1. Imposes additional runtime dependencies (`kotlin-runtime` and `kotlin-stdlib`) on all...

Just to be clear, the idea is to _always_ use insertion-ordered maps instead of unsorted maps, in order to provide a more deterministic user experience. I believe this is why...

Small maps will be the common case, but I need to maintain order in any case.

Thanks for all your work on this. I'll definitely give it a shot. > Preserving insertion order is a good idea. I'm glad you proposed it! Is it worth a...

I'm the creator of Spock (http://spockframework.org), and have faced similar issues when integrating with JUnit (which is currently the only means for running a Spock test). It would be awesome...

Implementation note: I'd prefer to guard against the unavailability of ServiceLoader as follows: ``` try { ServiceLoader serviceLoader = ServiceLoader.load(RunListener.class); for (RunListener listener : serviceLoader) fNotifier.addListener(listener); } catch (LinkageError ignored)...

Note that the original proposal is about "JUnit's ability to integrate with other testing tools". In other words, this isn't a feature for people writing JUnit tests, but for people...

Spock can't use JUnitCore since it isn't in a position to trigger test execution. Instead, Spock implements a @RunWith runner, hidden away in the spock.lang.Specification class that every test class...

@Tibor17 I can't follow your argumentation. My problem is that a @RunWith runner has a "per test class" scope, which means that it can't react to "per test suite" or...

For the purpose of this discussion, think of Spock as its own standalone test framework that implements an org.junit.runner.Runner to benefit from the IDE and build tool integration that exists...