You're right, the maven dependency tree API ran into the same issue. Since using the plugin annotation for automatic resolution didn't make the problem any better (and made it worse in some cases), I switched the plugin back to resolving dependencies inside the plugin execution, and merged back the manual purge option. Hopefully this is ok with everyone.
This definitely seems like a bug in Maven 3.0.4 behaviour not allowing forced resolution with resolver.resolveAlways(). I set the invoker plugin to skip the failing ITs for the relevant versions of Maven. So here is the status of the ITs: Maven 2.0.x and 2.2.1 - 6 of 7 are passing, only the broken pom use case is failing, which is kind of an edge case anyway. Maven 3.0.x (not 3.0.4) - All 7 ITs are passing Maven 3.0.4 - 4 of 7 are passing. The three that are failing all involve resolving, purging, and then re-resolving, so they are affected by the same bug. IMO, this is ok for a release considering that this goal in the previous version of the dependency plugin didn't really work with Maven 3 at all because it couldn't find any remotely downloaded dependencies in the local repo (http://jira.codehaus.org/browse/MDEP-383). On 10/24/2012 12:25 PM, Hervé BOUTEMY wrote: > I had a look at the IT output and found this log message > [DEBUG] Skipped remote update check for org.apache.maven.its.dependency:purge- > local-repository:pom:1.0, already updated during this session. > > the code causaing this message is in Aether > > I doubt maven-dependency-tree will do anything about this problem: this seems > to be a cache feature of Aether we exceptionnaly don't want > > IMHO, if we used resolver.resolveAlways() API (like it was done in previous m- > dependency-p version), this cache should be avoided but I tried and it didn't > change anything > but this time, this would be a Maven core bug :) > > WDYT? > > Regards, > > Hervé > > Le mercredi 24 octobre 2012 08:56:32 Paul Gier a écrit : >> Hi Arnaud, >> >> I should have some time today to try out the shared dependency tree API, >> if it doesn't work in Maven 3.0.4 with that, then I'm not sure what else >> I can do since this seems to be a regression in 3.0.4. At that point I >> will just configure the failing ITs to be skipped for 3.0.4. >> >> Jenkins also confirmed what I'm seeing locally, that all the ITs pass >> with 3.0.3: >> https://builds.apache.org/job/maven-plugins-ITs-m3.0.3/ >> >> All the dependency plugin ITs pass, however, the job shows failure >> because some of the invoker plugin its are failing in 3.0.3. >> >> On 10/24/2012 08:28 AM, Arnaud Héritier wrote: >>> ok. thanks >>> I'll launch the release process tonight. >>> >>> On Wed, Oct 24, 2012 at 2:18 PM, Benson Margulies > <[email protected]>wrote: >>>> I reran the build with a clean local repo and it's good. So I withdraw >>>> my complaint, go right ahead. >>>> >>>> On Wed, Oct 24, 2012 at 7:56 AM, Arnaud Héritier <[email protected]> >>>> >>>> wrote: >>>>> Strange your issue with the report. >>>>> I don't have it on my side. >>>> >>>>> Example : >>>> https://projects.exoplatform.org/forum/forum-component/forum-component-co >>>> mmon/dependency-analysis.html>> >>>>> I need to investigate more but it won't be before tonight >>>>> >>>>> >>>>> On Wed, Oct 24, 2012 at 1:48 PM, Benson Margulies <[email protected] >>>>> >>>>> wrote: >>>>>> On Wed, Oct 24, 2012 at 7:30 AM, Benson Margulies < >>>> >>>> [email protected]> >>>> >>>>>> wrote: >>>>>>> On Wed, Oct 24, 2012 at 4:57 AM, Arnaud Héritier <[email protected] >>>>>> >>>>>> wrote: >>>>>>>> Ok if we are all agree I'll process the release this evening. >>>>>>>> Thanks for your help guys >>>>>>> >>>>>>> I am -1 to a release which does not work, at all, on Maven 3.0.4. I >>>>>>> tested the 'report' mojo last night with the current snapshot and it >>>>>>> died with the same wierd classworlds error that the ITs die with. I'm >>>>>>> willing to help fix this up, but I;m not following the logic of a >>>>>>> release. Feel free to explain why I'm confused. >>>>>> >>>>>> I'll do more diagnosis later today. >>>>>> >>>>>>>> Arnaud >>>>>>>> >>>>>>>> >>>>>>>> On Wed, Oct 24, 2012 at 1:02 AM, Hervé BOUTEMY < >>>> >>>> [email protected] >>>> >>>>>>> wrote: >>>>>>>>> Le lundi 22 octobre 2012 17:38:10 Paul Gier a écrit : >>>>>>>>>> I did some more testing, and the tests only seem to fail using >>>> >>>> Maven >>>> >>>>>>>>>> 3.0.4 for me. I was testing locally with Maven 3.0.3 previously. >>>>>> >>>>>> For >>>>>> >>>>>>>>>> Robert, the ITs passed with Maven 2.2.1, but failed with 3.0.x. >>>> >>>> If >>>> >>>>>> you >>>>>> >>>>>>>>>> want to release this week it's probably ok to release even with >>>> >>>> these >>>> >>>>>>>>>> failures because the previous release didn't work at all with >>>> >>>> Maven >>>> >>>>>> 3. >>>>>> >>>>>>>>>> We just didn't have ITs to catch it. We could set the invoker >>>> >>>> not to >>>> >>>>>>>>>> execute those for Maven 3, and maybe put a note in the docs, and >>>>>> >>>>>> find a >>>>>> >>>>>>>>>> better fix for a 2.6.1 release. >>>>>>>>> >>>>>>>>> +1 >>>>>>>>> >>>>>>>>> yes, this is not a regression but better ITs >>>>>>>>> >>>>>>>>> Regards, >>>>>>>>> >>>>>>>>> Hervé >>>> >>>> --------------------------------------------------------------------- >>>> >>>>>>>>> To unsubscribe, e-mail: [email protected] >>>>>>>>> For additional commands, e-mail: [email protected] >>>>>>>> >>>>>>>> -- >>>>>>>> ----- >>>>>>>> Arnaud Héritier >>>>>>>> 06-89-76-64-24 >>>>>>>> http://aheritier.net >>>>>>>> Mail/GTalk: [email protected] >>>>>>>> Twitter/Skype : aheritier >>>>>> >>>>>> --------------------------------------------------------------------- >>>>>> To unsubscribe, e-mail: [email protected] >>>>>> For additional commands, e-mail: [email protected] >>>>> >>>>> -- >>>>> ----- >>>>> Arnaud Héritier >>>>> 06-89-76-64-24 >>>>> http://aheritier.net >>>>> Mail/GTalk: [email protected] >>>>> Twitter/Skype : aheritier >>>> >>>> --------------------------------------------------------------------- >>>> To unsubscribe, e-mail: [email protected] >>>> For additional commands, e-mail: [email protected] >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: [email protected] >> For additional commands, e-mail: [email protected] > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
