On 7 September 2015 at 20:44, P. Ottlinger <pottlin...@apache.org> wrote: > Hi all, > > Am 30.06.2015 um 07:48 schrieb Paul Merlin: >> Apache Jenkins Server a écrit : >>> * What went wrong: >>> Execution failed for task ':compileTestGroovy'. >>>>> nebula/test/PluginProjectSpec : Unsupported major.minor version 51.0 >> Looks like nebula-test, the test harness for Gradle plugins, has been >> built using Java 7 but buildbot is running the build using Java 6. > > As it seems there is no nebula artifact available that builds with an > ancient JDK5 ..... > > The used nebula 2.2.1 is build with JDK7 as its pom indicates: > https://jcenter.bintray.com/com/netflix/nebula/nebula-test/2.2.1/nebula-test-2.2.1.pom > <snip> > <nebula_X_Compile_Target_JDK>1.7</nebula_X_Compile_Target_JDK> > <nebula_X_Compile_Source_JDK>1.7</nebula_X_Compile_Source_JDK> > </snip> > > Is it time to update at least to JDK6 or JDK7?!
I don't think we should force users of RAT to upgrade their Java just because our build engine is causing us problems... > How else can we get green builds again? Just use a more recent version of Java for the build. For example Commons Logging still targets Java 1.2+ but builds happily on later versions > I like the idea of being able to build with gradle ... my fear is that > our project is going to loose momentum if we stay in the dark ages for > all times. > > What do you think? Is there a way to determine the minimal JDK used by > our clients? Only by asking, but where? Or releasing a new version that requires JavaN and wait for the complaints that it no longer runs on previous versions... It's not easy finding out, and some users (particularly larger companies) tend to be very slow to upgrade. That is why I am generally against requiring more recent versions of Java just because it's there. However, RAT is not a low-level library and is generally used on development systems which tend to be less restricted when it comes to upgrading. So requiring a more recent version of Java for RAT is less likely to cause problems, than say updating Commons Lang. But if there is no pressing need to do so, why take the risk? > Thanks, > Phil >
