Displaying #maven-dev/2017-03-21.log:

Tue Mar 21 08:45:31 2017  mbenson:Joined the channel
Tue Mar 21 18:25:14 2017  hboutemy:Joined the channel
Tue Mar 21 19:45:11 2017  tibor_:Joined the channel
Tue Mar 21 21:46:11 2017  Michael-O:Joined the channel
Tue Mar 21 21:52:22 2017  tibor_:Michael-O: Hi Michael
Tue Mar 21 21:53:34 2017  Michael-O:tibor_: reading through the mails. What a chaos. Guillaume is testing the 8-day-old SUREFIRE-1342 branch, right?
Tue Mar 21 21:54:30 2017  tibor_:In real I am not sure if I should totally trust to what he is testing.
Tue Mar 21 21:55:08 2017  tibor_:It is difficult to talk with him because the gap between us is in days.
Tue Mar 21 21:57:17 2017  Michael-O:He should join the IRC chat
Tue Mar 21 21:57:23 2017  Michael-O:IT would testing way easier
Tue Mar 21 21:57:29 2017  Michael-O:email doesn't scale here
Tue Mar 21 21:57:58 2017  tibor_:I will ask him to join us in IRC again.
Tue Mar 21 22:00:06 2017  Michael-O:I have talked to our admin about it.
Tue Mar 21 22:00:15 2017  Michael-O:The first question he asked me: why don't you use sockets?
Tue Mar 21 22:00:18 2017  Michael-O:Funny, isn't it?
Tue Mar 21 22:01:24 2017  tibor_:It's obvious. Sockets are obviously a good choice.
Tue Mar 21 22:01:47 2017  tibor_:In next version we can make it.
Tue Mar 21 22:03:11 2017  tibor_:We are just waiting for Guillaume Boué however he made also a lot of testing. Interesting is that the bad result was with OpenJdk and not the OracleJdk.
Tue Mar 21 22:03:44 2017  stephenc:I did ask about sockets two or three weeks ago...
Tue Mar 21 22:04:15 2017  tibor_:An what you want to say with that :)
Tue Mar 21 22:04:30 2017  stephenc:(I always thought it was using sockets anyway... surprised to find it was stdio instead)
Tue Mar 21 22:06:12 2017  tibor_:Stephen, we wanted to find root cause, or better we wanted to understand it. Let's suppose the same won't happen with sockets.
Tue Mar 21 22:07:10 2017  tibor_:As Michael told me bad result was only with one installation of freebsd, and later Michael said that some HDD issue.
Tue Mar 21 22:07:25 2017  Michael-O:Joined the channel
Tue Mar 21 22:07:48 2017  Michael-O:tibor_: sorry, I was kicked.
Tue Mar 21 22:07:52 2017  Michael-O:Did you write something
Tue Mar 21 22:07:53 2017  Michael-O:?
Tue Mar 21 22:07:59 2017  tibor_:I will repeat.
Tue Mar 21 22:08:32 2017  tibor_:Stephen, we wanted to find root cause, or better we wanted to understand it. Let's suppose the same won't happen with sockets.
Tue Mar 21 22:08:41 2017  tibor_:As Michael told me bad result was only with one installation of freebsd, and later Michael said that some HDD issue.
Tue Mar 21 22:09:08 2017  tibor_:Stephen was surprised stdio was in use.
Tue Mar 21 22:09:57 2017  Michael-O:tibor_: I checked the hard disk and changed to another physical raid volume. There was no difference.
Tue Mar 21 22:10:01 2017  Michael-O:So it was not the disk.
Tue Mar 21 22:10:27 2017  Michael-O:But if Guillaume can reproduce failures, this isn't just my "broken" setup
Tue Mar 21 22:11:07 2017  tibor_:What is bad is the fact that he could not find dump file.
Tue Mar 21 22:11:18 2017  Michael-O:I don't have any too
Tue Mar 21 22:11:21 2017  Michael-O:zero of them
Tue Mar 21 22:11:33 2017  tibor_:That's not good.
Tue Mar 21 22:11:57 2017  tibor_:Some ITs won't pass then.
Tue Mar 21 22:13:09 2017  Michael-O:Another improvement: I started fiddling with freebsd-issue and consider /bin/sh -c absolutely superfluous.
Tue Mar 21 22:13:20 2017  tibor_:The test Surefire141PluggableProvidersIT has dependency to broken provider which throws RuntimeException. This is intended and dump file should handle it.
Tue Mar 21 22:13:22 2017  Michael-O:Just java -jar ... is fully sufficient
Tue Mar 21 22:13:37 2017  Michael-O:Someone swallows the expception?
Tue Mar 21 22:14:38 2017  tibor_:Nevertheless /bin/sh -c, the test did run, but somehow yes exception swallowed.
Tue Mar 21 22:15:59 2017  Michael-O:It is likely a thread.
Tue Mar 21 22:16:35 2017  tibor_:It would be nice to run the test from target in CLI and debug ForkedBooter. There are three places the dump file is written.
Tue Mar 21 22:17:25 2017  tibor_:What you mean that it's likely a thread?
Tue Mar 21 22:18:09 2017  Michael-O:Exceptions from threads aren't captured in the master thread
Tue Mar 21 22:18:18 2017  Michael-O:so they are really swallowed
Tue Mar 21 22:18:58 2017  tibor_:no
Tue Mar 21 22:19:53 2017  tibor_:ForkedBooter is in forked jvm and not in the master process. So the ForkedBooter is able to write the stack trace to dump file. In other words it does not need stdout.
Tue Mar 21 22:21:57 2017  tibor_:Michael, maybe i did not get your point.

Comments