I am currently running the CI job #129 of Surefire with JDK 7, 8, 9, 10, Maven 3.2, 3.3, 3.5, Linux and Windows: This is the link to pipeline https://builds.apache.org/job/maven-wip/job/maven-surefire/job/SUREFIRE-1463/
The old job #128 passed the most of the Linux builds. My last push -f crashed git checkout in the Pipeline. Bad luck, I just forced changes on the git HEAD of the branch meanwhile the checkout was running and the command 'git checkout' just crashed in Jenkins. I have Windows 7 and the native library used in test Surefire1295AttributeJvmCrashesToTestsIT is working as expected but our Jenkins in ASF is using Windows NT and there the library used in the IT does not work properly. I already reported an issue in the open source: https://github.com/michaeltandy/crashjvm/issues/1 Cheers Tibor On Fri, Feb 16, 2018 at 3:52 PM, Tibor Digana <[email protected]> wrote: > The reason why I was idle is the fact I made everyting to to support > Surefire on Jdk 7-10 & Maven 3.1 - 3.5 and Linux/Windows and I am now happy > with that. I did not commit it in my branch yet. > I don't want to loose all my work, therefore the branch *SUREFIRE-1463* > will be finalized by today evening and then I will wait for Jenkins result > but my local build worked with many configurations so far. > I have made a quite a lot of improvements connected to broken tests > regarding the combination of Maven & Jdk & Platform version and I really do > not want to loose all of that. > And the best is that I did not use latest *commons-lang3* in > *src/main/java*, but I used that in *src/test/java*. This means I do not > have to force the Users to adapt to *JDK7*. This should be the plan in > version *3.0.0*. > So today I will finish *Jenkinsfile*, the jobs will run in a sequence as > a preference. > If you saw my e-mails you would notice that I also had a problem with > Jenkins because Jenkins started a lot of parallel runs in *ubuntu-1* > however it has only 2 executors - I expected only two runs but there was > much more and overloaded. > Why *ubuntu-1*? The answer is the large number of ITs we have. We have > also sophisticated tests with native libraries and these did not work in > every ASF machine. > I have very good results with *java.sql.Date*. I saw it but it's gone. > > > On Fri, Feb 16, 2018 at 1:42 PM, Robert Scholte <[email protected]> > wrote: > >> And this is not an issue with Java 9? >> >> It might be related to MNG-6275[1]. Classworlds seems to use the >> bootclassloader, which doesn't expose all the jdk/jre classes like >> java.sql.Date. >> We weren't able to fix this in 3.5.x >> >> Robert >> >> [1] https://issues.apache.org/jira/browse/MNG-6275 >> >> On Fri, 16 Feb 2018 12:16:59 +0100, Olivier Lamy <[email protected]> >> wrote: >> >> Hi, >>> I'm currently testing maven builds with jdk10 (especially surefire) >>> I noticed velocity tools need java.sql.Date (yes it is :-) ) >>> See stack trace: >>> https://gist.github.com/olamy/243bf93f77becdd698148b07065b84f7 >>> >>> Well ATM I'm not sure how to fix that... >>> >>> Cheers >>> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: [email protected] >> For additional commands, e-mail: [email protected] >> >> >
