Displaying #maven-dev/2017-07-18.log:

Tue Jul 18 00:19:05 2017  olamy:Joined the channel
Tue Jul 18 02:07:06 2017  aheritier_:Joined the channel
Tue Jul 18 11:13:13 2017  tibor_:Joined the channel
Tue Jul 18 11:41:31 2017  Michael-O:Joined the channel
Tue Jul 18 12:34:06 2017  tibor_:Michael-O: If we do not read std/err should we leave it unread?
Tue Jul 18 12:34:27 2017  tibor_:I will close all three streams in finally block.
Tue Jul 18 12:34:52 2017  Michael-O:Good question.
Tue Jul 18 12:35:03 2017  Michael-O:I would send them to /dev/null, just to make sure
Tue Jul 18 12:35:06 2017  tibor_:These commands write short strings to shared memory.
Tue Jul 18 12:52:20 2017  tibor_:This works in cmd /bin/ps -o etime= -p 1 > /dev/null 1>&2 on FreeBSD. This would require reading from std/err, right?
Tue Jul 18 13:00:48 2017  Michael-O:correct
Tue Jul 18 13:01:05 2017  Michael-O:but you redirected all output to null
Tue Jul 18 13:04:00 2017  tibor_:howcome, i can see the output in commandline
Tue Jul 18 13:05:40 2017  Michael-O:ps writes to 2 and not 1
Tue Jul 18 13:06:10 2017  Michael-O:$ /bin/ps -o etime= -p 1 > /dev/null 1>&2
Tue Jul 18 13:06:10 2017  Michael-O:01:20:48
Tue Jul 18 13:06:10 2017  Michael-O:[mosipov@mika-ion ~]$ /bin/ps -o etime= -p 1 2> /dev/null 1>&2
Tue Jul 18 13:06:10 2017  Michael-O:[mosipov@mika-ion ~]$
Tue Jul 18 13:14:15 2017  tibor_:Michael-O: /bin/ps -o etime= -p 1 1>&2
Tue Jul 18 13:17:29 2017  Michael-O:[mosipov@mika-ion ~]$ /bin/ps -o etime= -p 1 1>&2
Tue Jul 18 13:17:29 2017  Michael-O:01:32:53
Tue Jul 18 13:19:07 2017  tibor_:So we do not need to empty std/error with this easy solution and no deadlock between two processes, right?
Tue Jul 18 13:19:46 2017  tibor_:I mean no reading of error stream in a separate Thread
Tue Jul 18 13:19:53 2017  tibor_:cool
Tue Jul 18 13:20:11 2017  Michael-O:Yep, if you redirect correctly to null
Tue Jul 18 13:26:01 2017  tibor_:Michael-O: ? you mean /bin/ps -o etime= -p 1 1>/dev/null 1>&2
Tue Jul 18 13:28:21 2017  Michael-O:What is actually the purpose of the command for you if you don't want to consume the output?
Tue Jul 18 13:29:25 2017  Michael-O:I'll be back later this day, we can continue talking then
Tue Jul 18 13:29:32 2017  tibor_:I consume output.
Tue Jul 18 13:29:53 2017  tibor_:The purpose is to read out and empty buffers.
Tue Jul 18 13:30:24 2017  tibor_:If read error stream still okay because I cannot find there time patter => error
Tue Jul 18 13:30:51 2017  tibor_:joining both together is okay
Tue Jul 18 16:44:44 2017  tibor_:Joined the channel
Tue Jul 18 19:03:30 2017  Michael-O:Joined the channel
Tue Jul 18 19:29:24 2017  Michael-O:tibor_, I am back
Tue Jul 18 19:30:02 2017  tibor_:I am playing with programming 1302
Tue Jul 18 19:39:56 2017  rfscholte:Joined the channel
Tue Jul 18 20:00:39 2017  Michael-O:(Y)
Tue Jul 18 20:00:41 2017  Michael-O:+1
Tue Jul 18 20:37:01 2017  tibor_:Michael-O: I will fix unit test. My plan is to push a new branch 1302_4 and then we will conclude it together and I will create git diff SUREFIRE-1302_3...SUREFIRE-1302_4, and then I will update SUREFIRE-1302_3, then git squash and push to master.
Tue Jul 18 20:37:45 2017  Michael-O:sounds reasonable to me
Tue Jul 18 20:39:01 2017  tibor_:It looks quite well. Then we can add JNA in a new Jira because this can be improved and the fixes in GitHub must be taken in parallel. The users are looking for a new release so therefore new Jira ticket and paralell dev.
Tue Jul 18 20:39:39 2017  tibor_:I want to ask the reporters in 1302 for verifying the fix.
Tue Jul 18 20:40:33 2017  tibor_:Then three fixes from users in GitHub, then fix JUnit Parameterized tests, Java 9, and one more, and then release 2.20.1
Tue Jul 18 20:41:10 2017  tibor_:Basically 2.20.1 is fixing 2.20
Tue Jul 18 20:42:23 2017  rfscholte:tibor_: FYI, I'm working on a library which can divide jars over classpath and modulepath, you'll probably need that as well
Tue Jul 18 20:42:58 2017  tibor_:what it means for Surefire?
Tue Jul 18 20:44:24 2017  rfscholte:required for SUREFIRE-1262
Tue Jul 18 20:44:34 2017  tibor_:In Surefire we wanted to add --add-modules ALL-SYSTEM
Tue Jul 18 20:44:54 2017  tibor_:and activate java.se.ee in IsolatedCL
Tue Jul 18 20:45:02 2017  rfscholte:are you sure, I think we can do better
Tue Jul 18 20:45:11 2017  tibor_:I think no need to divide CL
Tue Jul 18 20:45:29 2017  tibor_:but this may introduce more issues
Tue Jul 18 20:46:00 2017  rfscholte:for the first release it may be good enough, but in the end we must do it correct
Tue Jul 18 20:46:01 2017  tibor_:filtering dependencies, classes in artifacts, there is a lot
Tue Jul 18 20:46:46 2017  rfscholte:I'd say: focus on JNA stuff first ;)
Tue Jul 18 20:46:55 2017  tibor_:no no
Tue Jul 18 20:47:28 2017  tibor_:THis is important, JNA is not needed now, JNA is improvement if JMX is not accoring to Oracle TCK
Tue Jul 18 20:47:43 2017  tibor_:it's nice o have it of course
Tue Jul 18 20:48:31 2017  tibor_:@rfscholte: Try to open PR on github even if incomplete
Tue Jul 18 20:48:59 2017  rfscholte:I first need to finish the library
Tue Jul 18 20:50:57 2017  tibor_:What library you are talking about?
Tue Jul 18 20:52:10 2017  rfscholte:plexus-java, it is new
Tue Jul 18 21:05:47 2017  tibor_:Michael, Robert: Now I am runnign unit test. Failed. Debugging..
Tue Jul 18 22:39:04 2017  olamy:Joined the channel
Tue Jul 18 22:46:10 2017  olamy:Joined the channel

Comments