On Tue, 19 Dec 2023 16:00:59 GMT, Raffaello Giulietti <rgiulie...@openjdk.org> wrote:
>> test/jdk/jdk/jfr/event/io/TestSerializationMisdeclarationEvent.java line 50: >> >>> 48: * @requires vm.hasJFR >>> 49: * @library /test/lib >>> 50: * @run junit/othervm >>> jdk.jfr.event.io.TestSerializationMisdeclarationEvent >> >> Is the use of "othervm" needed here because of the use of >> `jdk.jfr.consumer.RecordingStream`? Does `RecordingStream` do JVM wide state >> changes? I did a quick look at that class but couldn't come to a conclusion. > > Not sure, I have to check. All the other events run in othervm. While it may be possible in some case to not do that, it's much easier to analyse issues, if we are sure the JVM is fresh. For example, JFR may not generate bytecode if an event is disabled. The JFR tests have been hardened to be able to run in parallel, so they run pretty fast. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/17129#discussion_r1431727571