Displaying #maven-dev/2017-02-15.log:

Wed Feb 15 00:19:14 2017  olamy:Joined the channel
Wed Feb 15 08:55:21 2017  Michael-O:Joined the channel
Wed Feb 15 09:12:51 2017  tibor_:Joined the channel
Wed Feb 15 10:11:46 2017  olamy:Joined the channel
Wed Feb 15 15:47:36 2017  tibor_:Joined the channel
Wed Feb 15 15:48:28 2017  tibor_:Michael-O: I am at home. Just ping me when you like.
Wed Feb 15 16:42:45 2017  Michael-O:Joined the channel
Wed Feb 15 16:42:55 2017  Michael-O:tibor_: got the output of the test run
Wed Feb 15 16:43:21 2017  Michael-O:uploading it
Wed Feb 15 16:43:53 2017  tibor_:did you apply those three java files from latest email?
Wed Feb 15 16:45:52 2017  Michael-O:Yes, I did. The second one.
Wed Feb 15 16:45:56 2017  Michael-O:Here is the tarball: http://home.apache.org/~michaelo/maven/surefire/maven-surefire-extended-patch.tar.gz
Wed Feb 15 16:52:13 2017  Michael-O:Do you have it?
Wed Feb 15 16:53:41 2017  tibor_:yes
Wed Feb 15 16:53:56 2017  tibor_:The problem is that Z event is not sent
Wed Feb 15 16:54:20 2017  tibor_:this means the JVM does not say "good bye"
Wed Feb 15 16:54:41 2017  tibor_:and there is no dump file
Wed Feb 15 16:55:23 2017  tibor_:did you check java processes after build?
Wed Feb 15 16:55:30 2017  tibor_:are they killed all?
Wed Feb 15 16:56:30 2017  tibor_:there can be only java process hanging or some exception which skipped sending Z
Wed Feb 15 16:56:30 2017  Michael-O:The build completed hours ago, but there are no zombie processes
Wed Feb 15 16:56:37 2017  tibor_:ok
Wed Feb 15 16:56:47 2017  Michael-O:The PID isn't logged either
Wed Feb 15 16:56:52 2017  Michael-O:which IT did you check?
Wed Feb 15 17:03:45 2017  tibor_:Surefire1080ParallelForkDoubleTestIT
Wed Feb 15 17:06:50 2017  Michael-O:Indeed, there is no good bye
Wed Feb 15 17:07:04 2017  tibor_:(ForkStarter.java:679) does not match any code
Wed Feb 15 17:07:42 2017  Michael-O:However, WorkingDirectoryMissingIT looks competely differnt
Wed Feb 15 17:07:59 2017  Michael-O:[ERROR] org.apache.maven.surefire.booter.SurefireBooterForkException: There was an error in the forked process
Wed Feb 15 17:08:01 2017  Michael-O:[ERROR] Error occurred in starting fork, check output in log
Wed Feb 15 17:08:02 2017  Michael-O:[ERROR] null
Wed Feb 15 17:08:05 2017  Michael-O:looks like a bug, doesn't it
Wed Feb 15 17:09:09 2017  tibor_:yes
Wed Feb 15 17:09:17 2017  tibor_:but we are not running HEAD
Wed Feb 15 17:09:32 2017  tibor_:what command did you run? mvn install?
Wed Feb 15 17:09:59 2017  tibor_:the stack does not match lines of code
Wed Feb 15 17:10:25 2017  Michael-O:I ran master with your patch, command was: mvn clean verify -Prun-its
Wed Feb 15 17:11:01 2017  Michael-O:179abbf
Wed Feb 15 17:11:29 2017  tibor_:org.apache.maven.plugin.surefire.booterclient.ForkStarter has revision e36fe19
Wed Feb 15 17:14:02 2017  tibor_:I have second project older
Wed Feb 15 17:14:19 2017  tibor_:are you able to run a single IT now?
Wed Feb 15 17:14:32 2017  Michael-O:I already have 57295 on that fie
Wed Feb 15 17:14:34 2017  Michael-O:yes, I can
Wed Feb 15 17:14:37 2017  Michael-O:which one?
Wed Feb 15 17:14:42 2017  tibor_:just a moment
Wed Feb 15 17:15:15 2017  tibor_:I will stash my changes I sent you, and I will revert to HEAD and then I will update those three files again
Wed Feb 15 17:15:39 2017  Michael-O:Ok, why don't you push a branch with your changes?
Wed Feb 15 17:15:44 2017  Michael-O:Makes is easier than pathes
Wed Feb 15 17:15:48 2017  Michael-O:patches
Wed Feb 15 17:16:01 2017  tibor_:Surefire1080ParallelForkDoubleTestIT has a single test, so this
Wed Feb 15 17:17:11 2017  Michael-O:tests started
Wed Feb 15 17:17:15 2017  Michael-O:will take a few minutes
Wed Feb 15 17:18:36 2017  Michael-O:CheckTestNgReportTestIT is really confusing
Wed Feb 15 17:18:55 2017  Michael-O:.assertThatLogLine( containsString( "[Parser] Running:" ), is( 1 ) );
Wed Feb 15 17:19:03 2017  Michael-O:this fails, though the log file contains this once
Wed Feb 15 17:20:15 2017  Michael-O:Looks like a false positive to me
Wed Feb 15 17:20:55 2017  tibor_:I will have a look, but now I will send you 3 files. You can avoid using run-its, because it's only for testing failsafe-plugin - will be faster: mvn install
Wed Feb 15 17:22:45 2017  Michael-O:ok
Wed Feb 15 17:23:03 2017  Michael-O:One important thing, my home is on NFS, could this influence the tests somehow?
Wed Feb 15 17:27:43 2017  Michael-O:Copying files...
Wed Feb 15 17:27:59 2017  tibor_:not sure if NFS breakes it
Wed Feb 15 17:28:11 2017  tibor_:null means returned value from errorInFork.getThrowable().getLocalizedMessage()
Wed Feb 15 17:28:20 2017  tibor_:no message in exception
Wed Feb 15 17:29:52 2017  Michael-O:Running now Surefire1080 with your patch now
Wed Feb 15 17:33:06 2017  tibor_:there must be exception in ForkedBooter we do not see , therefore null in strace
Wed Feb 15 17:40:29 2017  Michael-O:first run passed 1080
Wed Feb 15 17:40:32 2017  Michael-O:Running again
Wed Feb 15 17:43:45 2017  tibor_:wow
Wed Feb 15 17:44:21 2017  tibor_:and pls try CheckTestNgReportTestIT as well
Wed Feb 15 17:45:18 2017  tibor_:WorkingDirectoryMissingIT was strange as well
Wed Feb 15 17:47:29 2017  Michael-O:OK
Wed Feb 15 17:47:35 2017  Michael-O:give me 20 min
Wed Feb 15 17:48:18 2017  tibor_:are you running all or only some ITs?
Wed Feb 15 17:48:43 2017  Michael-O:Yes, 1080 passed again with your patch.
Wed Feb 15 17:48:49 2017  Michael-O:Will test the other two.
Wed Feb 15 17:49:17 2017  tibor_:and what patch, you mean the email with 3 files, or the last with one file in attachment?
Wed Feb 15 17:49:49 2017  Michael-O:the last one with the forkedbooter and the one from 18:23
Wed Feb 15 17:51:35 2017  rfscholte:Joined the channel
Wed Feb 15 17:51:44 2017  Michael-O:Running CheckTestNgReportTestIT and WorkingDirectoryMissingIT
Wed Feb 15 18:07:47 2017  Michael-O:Tests have completed, both failed again
Wed Feb 15 18:09:33 2017  tibor_:you think NFS can make it?
Wed Feb 15 18:09:41 2017  Michael-O:This is just an assumption
Wed Feb 15 18:09:57 2017  Michael-O:I will reupload log output for your and rerun stuff on a local drive
Wed Feb 15 18:10:40 2017  Michael-O:Please redownload the tarball
Wed Feb 15 18:10:46 2017  tibor_:ok
Wed Feb 15 18:11:17 2017  hboutemy:Joined the channel
Wed Feb 15 18:11:31 2017  hboutemy:rfscholte: Hi Robert
Wed Feb 15 18:20:58 2017  Michael-O:tibor_: running for a second time now
Wed Feb 15 18:26:14 2017  hboutemy:rfscholte: I need to go now, back at 11PM
Wed Feb 15 18:26:22 2017  Michael-O:tibor_: same stuff again. Fork error on a local disk, so it is not NFS. Confusingly with WorkingDirectoryMissingIT is the NullPointerException in the dumpstream file
Wed Feb 15 18:26:39 2017  tibor_:cool
Wed Feb 15 18:26:45 2017  Michael-O:ForkingRunListener:326
Wed Feb 15 18:26:48 2017  hboutemy:If you're here, we can discuss the result of beers with Remi Forax yesterday :)
Wed Feb 15 18:26:58 2017  Michael-O:encodeAndWriteToTarget
Wed Feb 15 18:27:05 2017  hboutemy:(and short night, since we stopped at 2h30 AM...)
Wed Feb 15 18:27:22 2017  hboutemy:guys, ready for 3.5.0?
Wed Feb 15 18:27:44 2017  Michael-O:hboutemy: I'd say no
Wed Feb 15 18:27:49 2017  hboutemy:(without useless alphas or betas: eventually one RC would be sufficient)
Wed Feb 15 18:27:55 2017  hboutemy:what is missing?
Wed Feb 15 18:28:02 2017  Michael-O:either encodeBytes is null or OUT
Wed Feb 15 18:28:25 2017  Michael-O:hboutemy: we have more test failures in Surefire when run with Maven master than with 3.3.9
Wed Feb 15 18:28:41 2017  hboutemy:ah!
Wed Feb 15 18:28:45 2017  Michael-O:Stuff some is competely unclear at the moment
Wed Feb 15 18:28:59 2017  Michael-O:I am running at tests on FreeBSD for two days now in various configs
Wed Feb 15 18:29:07 2017  hboutemy:ok, can't help now, but let me know if I can be of any help
Wed Feb 15 18:29:19 2017  Michael-O:http://home.apache.org/~michaelo/maven/surefire/
Wed Feb 15 18:29:24 2017  Michael-O:here are some tarball
Wed Feb 15 18:29:25 2017  hboutemy:failuers are only on FreeBSD, or also on Linus/Windows?
Wed Feb 15 18:30:16 2017  tibor_:Michael, you are not in Europe?
Wed Feb 15 18:30:25 2017  Michael-O:Likely.
Wed Feb 15 18:30:29 2017  Michael-O:I am!
Wed Feb 15 18:30:56 2017  tibor_:you stopped at 2h30 AM?
Wed Feb 15 18:31:14 2017  hboutemy:yes, so much things to discuss with Remi about Java 9
Wed Feb 15 18:31:23 2017  hboutemy:and Jigsaw, and central...
Wed Feb 15 18:31:35 2017  Michael-O:no, I left at 23:30 yesterday
Wed Feb 15 18:33:46 2017  tibor_:can you send me the log?
Wed Feb 15 18:33:53 2017  Michael-O:Yes, just a minute
Wed Feb 15 18:34:00 2017  Michael-O:OUT seems to be null
Wed Feb 15 18:34:07 2017  Michael-O:there are two NPEs
Wed Feb 15 18:35:18 2017  tibor_:final FileOutputStream out = new FileOutputStream( "target/stream.out" );
Wed Feb 15 18:35:32 2017  tibor_:is the way we dump the stream
Wed Feb 15 18:35:45 2017  Michael-O:TAR updated
Wed Feb 15 18:35:50 2017  tibor_:but it throws exception because the stream.out file exists
Wed Feb 15 18:36:02 2017  tibor_:we must delete them recursively
Wed Feb 15 18:36:07 2017  Michael-O:$TS_138.dumpstream is the fie
Wed Feb 15 18:36:13 2017  Michael-O:not this out
Wed Feb 15 18:36:18 2017  Michael-O:but OUT, upper case
Wed Feb 15 18:36:56 2017  Michael-O:I think it could be a race condition
Wed Feb 15 18:38:07 2017  tibor_:new File( "target/stream.out" ).delete();
Wed Feb 15 18:38:29 2017  tibor_:try to put it at Line 72 in ForkingRunListener
Wed Feb 15 18:38:55 2017  Michael-O:Why do you think it needs to be deleted first? We can overwrite it, can't we?
Wed Feb 15 18:39:25 2017  tibor_:yes
Wed Feb 15 18:39:49 2017  tibor_:we can override it, but deleting in every jvm run is safer
Wed Feb 15 18:40:26 2017  tibor_:because
Wed Feb 15 18:40:30 2017  tibor_:the Javadoc says
Wed Feb 15 18:40:44 2017  tibor_:FileNotFoundException if the file exists
Wed Feb 15 18:41:06 2017  tibor_:so the jvm started but the file was there and new one could not be created
Wed Feb 15 18:41:25 2017  Michael-O:This could be the reason for the NPE and the failure
Wed Feb 15 18:41:36 2017  tibor_:definitely
Wed Feb 15 18:41:54 2017  tibor_:i did not realize this case
Wed Feb 15 18:42:16 2017  tibor_:but now we can run only those last two tests only
Wed Feb 15 18:42:57 2017  tibor_:why the tests are so slow ? In my file system it is several seconds. I have SSD.
Wed Feb 15 18:44:53 2017  Michael-O:Those are old SCSI disks, 4 core Xeon, old machine, still i386
Wed Feb 15 18:45:04 2017  Michael-O:but rock stable, runs for years
Wed Feb 15 18:47:18 2017  Michael-O:just added the delete before FileOutputStream
Wed Feb 15 18:47:34 2017  tibor_:very good
Wed Feb 15 18:57:47 2017  rfscholte:Joined the channel
Wed Feb 15 18:57:55 2017  Michael-O:No avail, the NPEs are still there
Wed Feb 15 18:59:16 2017  tibor_:so the "OUT" is null?
Wed Feb 15 18:59:32 2017  rfscholte:hboutemy: just finished dinner, let's try later tonight
Wed Feb 15 19:01:30 2017  Michael-O:Yep, it is.
Wed Feb 15 19:07:11 2017  tibor_:it looks like the file cannot be created
Wed Feb 15 19:07:28 2017  tibor_:we can run the test directly from target
Wed Feb 15 19:08:01 2017  tibor_:cd CheckTestNgReportTestIT_shouldBeVerbose
Wed Feb 15 19:08:18 2017  tibor_:mvn -o test -Dsurefire.version=2.19.2-SNAPSHOT
Wed Feb 15 19:10:00 2017  tibor_:add -DtestNgVersion=5.10 -DtestNgClassifier=jdk15 -Dsurefire.testng.verbose=10
Wed Feb 15 19:12:06 2017  tibor_:I think we found the fix
Wed Feb 15 19:12:26 2017  tibor_:Let's remove all around OUT because one tests passed
Wed Feb 15 19:12:42 2017  tibor_:I think the synchronized block anf flush() make it
Wed Feb 15 19:16:03 2017  tibor_:I am preparing new sources without OUT variable
Wed Feb 15 19:20:49 2017  tibor_:Michael-O: I have sent you the sources. Try with previous tests as well. Better to delete the target in surefire-integration-tests
Wed Feb 15 19:21:07 2017  Michael-O:Could be possible that target is not available that time in the static block?
Wed Feb 15 19:21:51 2017  tibor_:no, because the static block is running in forked jvm
Wed Feb 15 19:22:17 2017  tibor_:and surefire folder has to be in target
Wed Feb 15 19:22:51 2017  tibor_:it contains jar file which starts the tests
Wed Feb 15 19:23:50 2017  tibor_:but in one test we saw the dump which is OUT variable
Wed Feb 15 19:24:04 2017  tibor_:I mean the events like Z, etc
Wed Feb 15 19:24:11 2017  Michael-O:thats right
Wed Feb 15 19:25:00 2017  tibor_:this was not fix, only a hint to dump events. But the fix with flushing the stream is in there
Wed Feb 15 19:25:14 2017  tibor_:and you said that one test was ok
Wed Feb 15 19:25:21 2017  tibor_:it means with the fix
Wed Feb 15 19:26:12 2017  tibor_:and the question is why the last event was not sent, but all previous always. I guess because the flush was missing in OutputStream
Wed Feb 15 19:26:24 2017  Michael-O:Let me update those classes first
Wed Feb 15 19:26:40 2017  tibor_:so the plugin process did not receive and very fast the JVM finished
Wed Feb 15 19:26:55 2017  Michael-O:This is a good explanation
Wed Feb 15 19:27:03 2017  Michael-O:Likelihood of race conditions
Wed Feb 15 19:27:45 2017  tibor_:because the Z event is just before finishing the process, and without flushing no way to receive it in maven process
Wed Feb 15 19:28:19 2017  Michael-O:You think that the VM is existed without flushing buffers first?
Wed Feb 15 19:28:32 2017  tibor_:I don't know
Wed Feb 15 19:28:41 2017  tibor_:but it has to do with platform
Wed Feb 15 19:28:50 2017  tibor_:because here I have another platform
Wed Feb 15 19:29:17 2017  tibor_:and the tests are ok, and Jenkins amd64 in ASF is ok as well
Wed Feb 15 19:29:36 2017  tibor_:you know std/out is somehow native
Wed Feb 15 19:29:55 2017  Michael-O:Running CheckTestNgReportTestIT_shouldNotBeVerbose manually succeeds
Wed Feb 15 19:30:03 2017  Michael-O:I'd retry the two tests
Wed Feb 15 19:30:05 2017  tibor_:not sure about Javadoc but jvm may have a bug as always
Wed Feb 15 19:30:42 2017  Michael-O:The implementation point of flush() could be different on BSD compared to the Linux Kernel
Wed Feb 15 19:33:32 2017  Michael-O:Can you push your diff to the three files to a branch on Surefire?
Wed Feb 15 19:33:43 2017  tibor_:of course
Wed Feb 15 19:33:47 2017  Michael-O:with a discriptive name
Wed Feb 15 19:35:13 2017  tibor_:and the java.lang.System.out is PrintStream. It contains autoFlush boolean and buffer inside. Maybe it is disablem but I do not see inside JVM because we use the original stream in our code.
Wed Feb 15 19:35:42 2017  tibor_:I will cleanup the code and then make a branch
Wed Feb 15 19:40:06 2017  Michael-O:tibor_: guess what, the WorkingDirectoryMissingIT just passed twice
Wed Feb 15 19:40:12 2017  Michael-O:no failures
Wed Feb 15 19:40:20 2017  tibor_:coool
Wed Feb 15 19:40:23 2017  Michael-O:The other one is still failure which has a different root cause
Wed Feb 15 19:40:35 2017  tibor_:and how many tests you run?
Wed Feb 15 19:40:55 2017  Michael-O:it was the flush apparently
Wed Feb 15 19:41:21 2017  Michael-O:Just the two: WorkingDirectoryMissingIT and CheckTestNgReportTestIT
Wed Feb 15 19:41:43 2017  Michael-O:As soon as you prepare the branch, I will rerun all tests and see wether any VM is failing
Wed Feb 15 19:41:50 2017  Michael-O:I think will be it
Wed Feb 15 19:43:27 2017  tibor_:just a moment
Wed Feb 15 19:43:30 2017  Michael-O:sure
Wed Feb 15 19:44:11 2017  tibor_:I commented out the body of catch(). Maybe not important but I want to make sure it is ok.
Wed Feb 15 19:44:21 2017  tibor_:I will send you the file
Wed Feb 15 19:46:47 2017  tibor_:ok, email is coming
Wed Feb 15 19:49:02 2017  tibor_:I think you will later run several combinations with Maven, right? So we will know complete results in the morning.
Wed Feb 15 19:49:19 2017  tibor_:Is Maven blocked by Surefire?
Wed Feb 15 19:49:26 2017  Michael-O:Yes, that is right. The first shot will be Maven master
Wed Feb 15 19:49:33 2017  Michael-O:This will depend on the output
Wed Feb 15 19:49:43 2017  Michael-O:We will know more tomorrow
Wed Feb 15 19:50:01 2017  Michael-O:Just to make sure, I have to replace ForkedBooter, but leave the other patched files, right?
Wed Feb 15 19:50:49 2017  tibor_:and how is it with the issue in maven-clean-plugin? Is somebody working on it or it is not a bug?
Wed Feb 15 19:51:34 2017  Michael-O:I have send some log files to Karl Heinz, but I'd like to rerun the case with your patch. It might be resolved.
Wed Feb 15 19:51:36 2017  tibor_:yes only ForkedBooter to replace and not the others
Wed Feb 15 19:52:42 2017  tibor_:So to make sure 3 files are edited in your git clone.
Wed Feb 15 19:52:57 2017  Michael-O:That's the case now
Wed Feb 15 19:53:02 2017  tibor_:ok, good
Wed Feb 15 19:53:40 2017  Michael-O:OK, running all tets now. Will take an hour or so. I will ping you again at 22:00.
Wed Feb 15 19:53:51 2017  tibor_:thx
Wed Feb 15 20:19:42 2017  tibor_:the branch is named SUREFIRE_STDOUT_FLUSH
Wed Feb 15 20:30:10 2017  tibor_:I go shopping and will be available after 10.
Wed Feb 15 21:10:55 2017  Michael-O:tibor_: Build has finished
Wed Feb 15 21:11:10 2017  Michael-O:There are less errors now, but now a few.
Wed Feb 15 21:11:22 2017  Michael-O:I will tar the output and upload again.
Wed Feb 15 21:11:53 2017  Michael-O:ForkMode*IT is problematic
Wed Feb 15 21:12:08 2017  tibor_:let's see
Wed Feb 15 21:16:07 2017  tibor_:i see you are uploading now
Wed Feb 15 21:17:15 2017  Michael-O:OK, please readownload
Wed Feb 15 21:23:44 2017  Michael-O:ForkMultiModuleIT is weird, in most cases exit code is 0, but the test faisl
Wed Feb 15 21:29:25 2017  tibor_:The forked VM terminated without properly saying goodbye.
Wed Feb 15 21:33:47 2017  Michael-O:yes but exit code is 0
Wed Feb 15 21:33:53 2017  Michael-O:which doesn't make any sense to me
Wed Feb 15 21:34:07 2017  tibor_:ForkMultiModuleIT is failed?
Wed Feb 15 21:34:49 2017  tibor_:I am not sureif I have right log open.
Wed Feb 15 21:35:04 2017  Michael-O:testForkCountOneNoReuse(org.apache.maven.surefire.its.ForkModeMultiModuleIT): Exit code was non-zero: 1; command line and log = (..)
Wed Feb 15 21:35:06 2017  Michael-O: testForkCountTwoNoReuse(org.apache.maven.surefire.its.ForkModeMultiModuleIT): Exit code was non-zero: 1; command line and log = (..)
Wed Feb 15 21:35:07 2017  Michael-O: testForkCountTwoReuse(org.apache.maven.surefire.its.ForkModeMultiModuleIT): Exit code was non-zero: 1; command line and log = (..)
Wed Feb 15 21:35:09 2017  Michael-O: testForkCountOneReuse(org.apache.maven.surefire.its.ForkModeMultiModuleIT): Exit code was non-zero: 1; command line and log = (..)
Wed Feb 15 21:35:10 2017  Michael-O: testForkCountTwoNoReuse(org.apache.maven.surefire.its.ForkModeTestNGIT): Exit code was non-zero: 1; command line and log = (..)
Wed Feb 15 21:35:12 2017  Michael-O: testForkCountTwoReuse(org.apache.maven.surefire.its.ForkModeTestNGIT): Exit code was non-zero: 1; command line and log = (..)
Wed Feb 15 21:35:14 2017  tibor_:maven-surefire-extended-patch.tar.gz ?
Wed Feb 15 21:35:22 2017  Michael-O:yes
Wed Feb 15 21:35:27 2017  Michael-O:you should have this: Finished at: 2017-02-15T21:57:26+01:00
Wed Feb 15 21:47:02 2017  tibor_:Something is strange
Wed Feb 15 21:47:08 2017  tibor_:the lines
Wed Feb 15 21:47:08 2017  tibor_:awaitResultsDone(ForkStarter.java:494)
Wed Feb 15 21:48:56 2017  Michael-O:The weirdest one is this: CheckTestNgReportTestIT
Wed Feb 15 21:50:34 2017  Michael-O:All of the VM failues, I see no dump and not stream.out
Wed Feb 15 21:50:37 2017  Michael-O:is that correct?
Wed Feb 15 21:54:22 2017  tibor_:I deleted stream.out
Wed Feb 15 21:54:30 2017  tibor_:it was that OUT
Wed Feb 15 21:54:46 2017  tibor_:we can add it if we want to
Wed Feb 15 21:54:54 2017  tibor_:but in this test
Wed Feb 15 21:55:07 2017  tibor_:the behaviour of TestNG is wrong
Wed Feb 15 21:55:21 2017  tibor_:it looks to me that
Wed Feb 15 21:55:41 2017  tibor_:the system properties are not set after the JVM started
Wed Feb 15 21:55:48 2017  tibor_:you know
Wed Feb 15 21:55:50 2017  tibor_:this
Wed Feb 15 21:56:08 2017  tibor_:<properties> <property> <name>surefire.testng.verbose</name> <value>${surefire.testng.verbose}</value> </property> </properties> </configuration>
Wed Feb 15 21:56:14 2017  tibor_:in the pom
Wed Feb 15 21:56:44 2017  tibor_:surefire-integration-tests\target\CheckTestNgReportTestIT_shouldNotBeVerbose
Wed Feb 15 21:57:51 2017  tibor_:I think FreeBSD is special in this
Wed Feb 15 21:58:33 2017  tibor_:Can you run the test from commandline?
Wed Feb 15 22:01:40 2017  tibor_:mvn -o test -Dsurefire.version=2.19.2-SNAPSHOT -DtestNgClassifier=jdk15 -DtestNgVersion=5.10
Wed Feb 15 22:02:21 2017  tibor_:You should not see this text in log "[Parser] Running:", and that's the assertion statement in IT: assertThatLogLine( containsString( "[Parser] Running:" ), is( 0 ) );
Wed Feb 15 22:05:26 2017  Michael-O:hold on a second, I'll try
Wed Feb 15 22:07:09 2017  tibor_:I will try on windows
Wed Feb 15 22:07:52 2017  Michael-O:Just did, it succeed
Wed Feb 15 22:09:04 2017  Michael-O:There is one problematic issue with the call:
Wed Feb 15 22:09:11 2017  tibor_:yes it succeeds but the IT fails because the text appears
Wed Feb 15 22:09:33 2017  Michael-O:TEST-TestSuite.xml says: <property name="sun.java.command" value="org.codehaus.plexus.classworlds.launcher.Launcher -e --batch-mode -Dmaven.repo.local=/usr/home/osipovmi/Projekte/maven-surefire/surefire-integration-tests/../surefire-setup-integration-tests/target/it-repo org.apache.maven.plugins:maven-clean-plugin:clean -Dsurefire.version=2.19.2-SNAPSHOT -DtestNgVersion=5.7 -DtestNgClassifier=jdk1
Wed Feb 15 22:09:34 2017  Michael-O:5 -DtestNgVersion=5.10 -DtestNgClassifier=jdk15 test"/>
Wed Feb 15 22:09:35 2017  tibor_:"[Parser] Running:"
Wed Feb 15 22:09:41 2017  Michael-O:note that -DtestNgVersion appears twice
Wed Feb 15 22:09:54 2017  Michael-O:Exactly, why is this verbose.
Wed Feb 15 22:10:07 2017  tibor_:jdk15? not jdk1
Wed Feb 15 22:10:11 2017  tibor_:or?
Wed Feb 15 22:10:17 2017  Michael-O:DtestNgVersion
Wed Feb 15 22:10:24 2017  Michael-O:5.7 and 5.10 is passed
Wed Feb 15 22:10:26 2017  tibor_:it's once
Wed Feb 15 22:10:50 2017  Michael-O:not here: sun.java.command
Wed Feb 15 22:10:51 2017  tibor_:i see only -DtestNgVersion=5.7
Wed Feb 15 22:10:58 2017  tibor_:I will check the logs
Wed Feb 15 22:11:12 2017  Michael-O:TEST-TestSuite.xml
Wed Feb 15 22:11:17 2017  Michael-O:sun.java.commmand
Wed Feb 15 22:11:37 2017  tibor_:ok
Wed Feb 15 22:11:59 2017  Michael-O:Do you see it now?
Wed Feb 15 22:12:31 2017  tibor_:I see
Wed Feb 15 22:12:43 2017  tibor_:5.10 is default hardcoded in test framework
Wed Feb 15 22:13:24 2017  tibor_:if you run build with only this particular IT with maven 3.3.9
Wed Feb 15 22:13:30 2017  tibor_:would pass?
Wed Feb 15 22:13:36 2017  Michael-O:Just a second
Wed Feb 15 22:13:50 2017  tibor_:if yes then this testframework must be fixed
Wed Feb 15 22:14:19 2017  tibor_:You know TestNG has a bug from version to version
Wed Feb 15 22:14:34 2017  tibor_:so 5.10 may behave differently from 5.7
Wed Feb 15 22:14:41 2017  tibor_:This happens with 6.x
Wed Feb 15 22:14:46 2017  tibor_:as well
Wed Feb 15 22:14:57 2017  Michael-O:this is my assumption
Wed Feb 15 22:15:04 2017  Michael-O:Running this single test with 3.3.9 now
Wed Feb 15 22:15:11 2017  tibor_:yes
Wed Feb 15 22:15:31 2017  tibor_:I will try to find the 5.10
Wed Feb 15 22:17:18 2017  tibor_:The same problem is with -DtestNgClassifier=jdk15
Wed Feb 15 22:17:21 2017  tibor_:duplicatd
Wed Feb 15 22:17:57 2017  tibor_:goals1.add( "-DtestNgClassifier=jdk15" ); SurefireLauncher
Wed Feb 15 22:18:26 2017  tibor_:This is bad because we will fix it and other tests which rely on it will fail :(
Wed Feb 15 22:19:34 2017  tibor_:surefire-integration-tests POM has <testng.version>5.7</testng.version>
Wed Feb 15 22:19:44 2017  tibor_:private final String testNgVersion = System.getProperty( "testng.version" );
Wed Feb 15 22:20:15 2017  Michael-O:Looks like a conflict in properties
Wed Feb 15 22:20:21 2017  tibor_:I think this is not user's problem.
Wed Feb 15 22:20:35 2017  tibor_:our in Surefire project
Wed Feb 15 22:20:45 2017  Michael-O:What should be done in your opinion?
Wed Feb 15 22:20:56 2017  Michael-O:Use 5.10 everywhere?
Wed Feb 15 22:21:28 2017  tibor_:will be nice to have more hands
Wed Feb 15 22:21:46 2017  tibor_:because we have almost 900 ITs
Wed Feb 15 22:22:08 2017  tibor_:yes everywhere
Wed Feb 15 22:22:15 2017  Michael-O:The test just passed with 3.3.9
Wed Feb 15 22:22:48 2017  Michael-O:Will run a second time
Wed Feb 15 22:22:56 2017  tibor_:I do not know why ASF or Codehaus did this trick.
Wed Feb 15 22:23:23 2017  Michael-O:with duplicate versions? Safety belt?
Wed Feb 15 22:23:49 2017  tibor_:Now I got it
Wed Feb 15 22:23:56 2017  tibor_:let's do it other way
Wed Feb 15 22:25:27 2017  tibor_:in SurefireLauncher we should discover the goal which is List<String> parse it and override like in Set if added programatically in IT
Wed Feb 15 22:25:39 2017  tibor_:I will try on Windows
Wed Feb 15 22:26:00 2017  tibor_:if this would work, then it means we did not break the tests
Wed Feb 15 22:26:09 2017  tibor_:and we do not need to have more hands
Wed Feb 15 22:26:15 2017  tibor_:possible
Wed Feb 15 22:26:34 2017  tibor_:I will try
Wed Feb 15 22:29:11 2017  Michael-O:Sounds reasonable, what if we remove the testNgVersion property from the IT poms and use the -DtestNgVersion only?
Wed Feb 15 22:29:57 2017  Michael-O:We have too many versions at play
Wed Feb 15 22:30:02 2017  Michael-O:the range and fixed one
Wed Feb 15 22:30:10 2017  Michael-O:unpredicatable, I think
Wed Feb 15 22:31:41 2017  Michael-O:I will try to setup a test bench on RHEL 6 too tomorrow or on friday, I have a brand new 12-core machine at hand at work with plenty of memory
Wed Feb 15 22:34:47 2017  tibor_:It's good to remove testNgVersion property, but I would do it in second step; otherwise we would have too much work.
Wed Feb 15 22:35:35 2017  tibor_:In first step the MavenLauncher should have unique system properties in variable private final List<String> goals = new ArrayList<String>();
Wed Feb 15 22:36:05 2017  Michael-O:I see, I trust you here, you have more experience with the Surefire code.
Wed Feb 15 22:36:47 2017  tibor_:If the first step does not break tests on my machine then we can go on with second step and add default sys prop in ITs which fail or select all TestNG ITs one by one
Wed Feb 15 22:37:32 2017  Michael-O:I will go offline now. If you have some changes, just send me them and tell what you want to test (combinations) tomorrow and I will prepare it.

Comments