> On 6 Sep 2019, at 17:11, Martin Buchholz <marti...@google.com> wrote: > > Bigger picture: all of this timeout-fiddling infrastructure should be in a > test library. It's not easy! > > test/jdk/java/util/concurrent/tck effectively has its own test library, for > historical reasons.
Agreed. It is not an easy task. I'd address this concern of having library level support for measuring timeouts and asserting threads blocking in a separate bug. For now, I'm afraid I have to reinvent the wheel.