maven-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stephen Connolly <stephen.alan.conno...@gmail.com>
Subject Re: Jenkins job stability: was Re: [MNG-5551] Java 8 + Maven status
Date Fri, 28 Mar 2014 12:02:09 GMT
I would rather get some good pipelining going so that we have a better
quality of jobs in the first place, e.g. see the job pipeline I have set up
https://builds.apache.org/job/maven-3.2-release-status/

I want to add other tests into this pipeline and once we have a strong
template for a good pipeline then we can replicate for other needs (note
that the reality is I will transfer the pipeline to the literate job type
once I get that out the door as it makes the whole shebang a whole lot
easier to setup and visualize... plus it makes branch development nicer too)


On 28 March 2014 11:32, Mirko Friedenhagen <mfriedenhagen@gmail.com> wrote:

> Hello,
>
> as a fresh subscriber to notifications I really wonder how often the jobs
> have failed in the last two weeks.
>
> - One reason seems to be, that during staging of plugins and especially
> shared components the jobs are failing because of staged dependencies are
> not available in central. I proposed to include all of maven-staging to the
> settings of Jenkins for IT and shared, but there were some concerns by
> Karl-Heinz(?) Same goes for the pom staging.
> - Yesterday maven-dependency-analyzer was not able to compile it's tests,
> because the TestCase symbol from junit was missing, my guess would be a
> defect local repository because of concurrent downloads.
> - All in all right now the jobs do irritate me more than help me.
> - IMO once a Jenkins job fails oftenly out of the blue it rapidly starts to
> become useless :-)
>
> So I have two proposals:
> - Include maven-staging in a special settings file included in maven-shared
> and maven-plugin only to be used in Jenkins.
> - Configure jobs to use a private repository in the Jenkins workspace and
> purge it beforehand.
>
> What do you think?
>
> Regards
> Mirko
> --
> Sent from my mobile
> On Mar 28, 2014 8:12 AM, "Hervé BOUTEMY" <herve.boutemy@free.fr> wrote:
>
> why didn't we detect the failure when building the plugin and running ITs
> with
> JDK 8 = somthing we did a long time ago and that I was conviced would give
> us
> more accurate results than what we finally have?
>
> I had a quick look at ITs, and it seems that the compiler plugin is
> configured
> to generate 1.5 bytecode
> IMHO, we need to add an IT generating 1.8 bytecode to make tests and show
> the
> failure before fixing and being sure the fix is complete.
> I didn't have time to really test, but I hope such discussion can help us
> improve JDK8 support more rapidly
> And find every other place where JDK8 compatibility won't be automatic:
> looking
> for asm is one way, but I suppose there may be problems for tools not using
> asm
>
> Regards,
>
> Hervé
>
> Le vendredi 28 mars 2014 07:40:53 Kristian Rosenvold a écrit :
> > I have updated dependency (maven shade) to use asm 5.0.1, and I hope to
> get
> > version 0.8 of dependency released RealSoon (tm).
> >
> > Kristian
> >
> > 2014-03-27 21:16 GMT+01:00 Mirko Friedenhagen <mfriedenhagen@gmail.com>:
> > > Steven,
> > >
> > > thanks, I now could reproduce this. Installing a local SNAPSHOT of the
> > > shared library and plugin did resolve this.
> > > So I guess we have to release both pretty soon :-).
> > > Regards Mirko
> > > --
> > > http://illegalstateexception.blogspot.com/
> > > https://github.com/mfriedenhagen/ (http://osrc.dfm.io/mfriedenhagen)
> > > https://bitbucket.org/mfriedenhagen/
> > >
> > >
> > > On Thu, Mar 27, 2014 at 8:37 PM, Steven Schlansker
> > >
> > > <stevenschlansker@gmail.com> wrote:
> > > > Here is a reproduction case:
> > > >
> > > > https://github.com/stevenschlansker/mdep-439-analyze-java8
> > > >
> > > > On Mar 27, 2014, at 10:26 AM, Mirko Friedenhagen <
> > >
> > > mfriedenhagen@gmail.com> wrote:
> > > >> Steven, I can not reproduce that maven-dependency-plugin:analyze:2.8
> > > >> fails with JDK8. I have created a small library with a Lambda (call
> it
> > > >> L) and ran dependency:analyze without probems. I installed this
> > > >> library and made a new component depend on L and ran
> > > >> dependency:analyze successfully again. As stated in MDEP-439[1], can
> > > >> you or someone else provide a sample? Otherwise I will close this
> bug.
> > > >>
> > > >> [1] http://jira.codehaus.org/browse/MDEP-439
> > > >> Regards Mirko
> > > >> --
> > > >> http://illegalstateexception.blogspot.com/
> > > >> https://github.com/mfriedenhagen/ (http://osrc.dfm.io/mfriedenhagen
> )
> > > >> https://bitbucket.org/mfriedenhagen/
> > > >>
> > > >>
> > > >> On Thu, Mar 27, 2014 at 1:28 PM, Matt Benson <gudnabrsam@gmail.com>
> > >
> > > wrote:
> > > >>> Oh, good news on the dependency plugin bit--I almost forgot that
> you
> > >
> > > had
> > >
> > > >>> mentioned its underlying library having already upgraded its trunk
> to
> > > >>> version 4. I was thinking more about jdependency, which supports
> the
> > >
> > > shade
> > >
> > > >>> plugin.
> > > >>>
> > > >>> Matt
> > > >>>
> > > >>> On Mar 27, 2014 7:22 AM, "Matt Benson" <gudnabrsam@gmail.com>
> wrote:
> > > >>>> Oh, well... It's no secret that ASM 3, being interface-based,
is
> > >
> > > wholly
> > >
> > > >>>> incompatible with ASM 4, which took the approach of using
abstract
> > >
> > > classes
> > >
> > > >>>> to significantly reduce the amount of code needed to accomplish
a
> > >
> > > given
> > >
> > > >>>> task. ASM 5 claims to be compatible with 4, which is why I,
not
> > >
> > > realizing
> > >
> > > >>>> that the plugins in question were based on ASM 3, suggested
that
> > >
> > > simply
> > >
> > > >>>> "dropping in" the new jar should suffice. The good news is
that
> the
> > >
> > > upgrade
> > >
> > > >>>> process is not terribly onerous, if only someone steps to
do it.
> > > >>>>
> > > >>>> Matt
> > > >>>> On Mar 27, 2014 5:25 AM, "Mirko Friedenhagen" <
> > >
> > > mfriedenhagen@gmail.com>
> > >
> > > >>>> wrote:
> > > >>>>> Mark,
> > > >>>>>
> > > >>>>> the analyze goal depends on the
> > > >>>>> org.apache.maven.shared:maven-dependency-analyzer:1.4
which
> depends
> > >
> > > on
> > >
> > > >>>>> asm 3.3.1. The trunk already moved to 4.2. I will see
what
> happens
> > > >>>>> when switching to asm 5 :-)
> > > >>>>> Regards Mirko
> > > >>>>> --
> > > >>>>> http://illegalstateexception.blogspot.com/
> > > >>>>> https://github.com/mfriedenhagen/ (
> http://osrc.dfm.io/mfriedenhagen)
> > > >>>>> https://bitbucket.org/mfriedenhagen/
> > > >>>>>
> > > >>>>>
> > > >>>>> On Thu, Mar 27, 2014 at 5:45 AM, Mark Derricutt <mark@talios.com
> >
> > >
> > > wrote:
> > > >>>>>> What version of the maven-dependency-plugin?  I'm
using 2.8 fine
> > >
> > > under
> > >
> > > >>>>> JDK8
> > > >>>>>
> > > >>>>>> and have been for some time - this is using the
> `copy-dependencies`
> > > >>>>>
> > > >>>>> goal and
> > > >>>>>
> > > >>>>>> nothing else tho...
> > > >>>>>>
> > > >>>>>> On 27 Mar 2014, at 6:15, Steven Schlansker wrote:
> > > >>>>>>> Java 8 has now been out for a week and Maven is
still not
> really
> > > >>>>>>> compatible.
> > > >>>>>>> In particular, the maven-shade-plugin and
> maven-dependency-plugin
> > >
> > > do
> > >
> > > >>>>> not
> > > >>>>>
> > > >>>>>>> work
> > > >>>>>>> due to an old version of ASM that throws
> > > >>>>>
> > > >>>>> ArrayIndexOutOfBoundsException on
> > > >>>>>
> > > >>>>>>> Java 8 class files.
> > > >>>>>
> > > >>>>>
> --------------------------------------------------------------------
> > > >>>>> -
> > > >>>>> To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
> > > >>>>> For additional commands, e-mail: users-help@maven.apache.org
> > > >>
> > > >>
> ---------------------------------------------------------------------
> > > >> To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
> > > >> For additional commands, e-mail: users-help@maven.apache.org
> > > >
> > > > ---------------------------------------------------------------------
> > > > To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
> > > > For additional commands, e-mail: users-help@maven.apache.org
> > >
> > > ---------------------------------------------------------------------
> > > To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
> > > For additional commands, e-mail: dev-help@maven.apache.org
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
> For additional commands, e-mail: dev-help@maven.apache.org
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message