- About Scala
- Documentation
- Code Examples
- Software
- Scala Developers
partest deadlocks
Tue, 2010-02-23, 20:33
Since r20962 partest is deadlocking on me. I was gathering evidence but
I see hudson ran into it immediately so it shouldn't be hard to repeat.
http://scala-webapps.epfl.ch/hudson/job/scala-checkin/2128/console
...
/localhome/hudson/workspace/scala-checkin/scala/test/files/jvm/daemon-actor-termination.scala [FAILED]
/localhome/hudson/workspace/scala-checkin/scala/test/files/jvm/replyablereactor.scala [TIMOUT]
/localhome/hudson/workspace/scala-checkin/scala/test/files/jvm/replyreactor.scala [TIMOUT]
Am compelled to reiterate that there is absolutely no reason that actors
commits should be allowed to have this kind of effect on the entire
development process. It is such a momentum killer to have to drop
everything and debug concurrency issues which have zero to do with my
own work. I am not opposed to dogfooding when it's done in a productive
way, but this is not productive.
Hi Paul,
Sorry for breaking the build; I forgot to update a couple of tests. The
timeout failures are gone now. I am looking at the daemon actor test
now, but this test at least returns quickly without timeout.
Philipp
Paul Phillips wrote:
> Since r20962 partest is deadlocking on me. I was gathering evidence but
> I see hudson ran into it immediately so it shouldn't be hard to repeat.
>
> http://scala-webapps.epfl.ch/hudson/job/scala-checkin/2128/console
> ...
>
> /localhome/hudson/workspace/scala-checkin/scala/test/files/jvm/daemon-actor-termination.scala [FAILED]
> /localhome/hudson/workspace/scala-checkin/scala/test/files/jvm/replyablereactor.scala [TIMOUT]
> /localhome/hudson/workspace/scala-checkin/scala/test/files/jvm/replyreactor.scala [TIMOUT]
>
> Am compelled to reiterate that there is absolutely no reason that actors
> commits should be allowed to have this kind of effect on the entire
> development process. It is such a momentum killer to have to drop
> everything and debug concurrency issues which have zero to do with my
> own work. I am not opposed to dogfooding when it's done in a productive
> way, but this is not productive.
>