Two unit tests [1] in maven-core override std/out and std/in. It looks like the provider did not receive NOOP command after timeout which is a subject to shutdown of forked JVM. That's maybe related to the commit you found. I will investigate more because the main Thread must have some deadlock; otherwise there is no other explanation because the unreceved NOOP should kill JVM. The JVM is still alive.
[1] ThreadOutputMuxerTest, BuilderCommonTest On Thu, Oct 8, 2015 at 8:44 PM, Kristian Rosenvold < kristian.rosenv...@gmail.com> wrote: > I identified the commit that caused the breakage, > cb97ba70cb9ebe685f8f2a06e87b538795b5dd9b. > > > I dont have further time to find the bug right now, but I suspect this > should be enough to find the problem. > > Kristian > > > 2015-10-08 19:09 GMT+02:00 Kristian Rosenvold < > kristian.rosenv...@gmail.com>: > > I created https://issues.apache.org/jira/browse/SUREFIRE-1182 to track > > this issue. > > > > I analyzed the two thread dumps, and it would appear that the > > interesting bits are in the main process; the forked process appears > > to be completely idle. But why are there two streampumper threads in > > the main process ? > > > > We'll continue in the jira. I'm definitely -1 on this release until we > > can sort this out. > > > > Kristian > > > > > > 2015-10-08 18:54 GMT+02:00 Kristian Rosenvold < > kristian.rosenv...@gmail.com>: > >> 2015-10-08 16:55 GMT+02:00 Tibor Digana <tibordig...@apache.org>: > >>> @Kristian > >>> Can we switch to other mail thread "surefire 2.19 hangs". > >> > >> I dont necessarily think this is a good idea. We are discussing a > >> candidate for a release, and I believe this is the appropriate place > >> to discuss these matters, since it's really a go/nogo matter for the > >> release. > >> > >> Kristian > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org > For additional commands, e-mail: dev-h...@maven.apache.org > > -- Cheers Tibor
