Le mardi 20 novembre 2012 19:50:22 Benson Margulies a écrit : > On Tue, Nov 20, 2012 at 7:45 PM, Brett Porter <[email protected]> wrote: > > On 21/11/2012, at 7:37 AM, Kristian Rosenvold <[email protected]> wrote: > >> I think I found at least one cause of too many forked lifecycles within > >> site. > >> > >> Basically the aggregators are being run for every module in the > >> reactor, if I understand things correctly they should only be run for > >> "pom" > >> modules. > >> > >> I have a very simple patch at > >> https://github.com/krosenvold/maven-shared/commit/9536ef08946d4e0aa6d3d05 > >> 33642bb5640da4a11 I was hoping someone who actually understands this > >> stuff would comment on this patch ;) > >> > >> I will make some tests if the patch looks good ;) > > > > I'm not sure this covers every use case. For example, what about a project > > that has submodules that are POMs for hierarchy? While uncommon, I think > > it is also possible to build an artifact at the root that isn't a POM. > > IIRC there has been a method introduced to determine if the current > > project is the execution root - perhaps that's what's needed here. > I'm not sure I follow. This will go ahead and for any pom with > packaging 'pom'. > Is your concern that it will still run too many, as > it will aggregate at each intermediate level? for the moment, it is run at each level, even when not pom packaging, ie AFAIK the only case where there can be modules, then aggregator means something (or I may miss something)
> My inclination is to > think that aggregation should only happen at the root, not at > intermediates, but I could see room for disagreement. root is a relative notion: if you build starting from this intermediate, this is the root from the build. IMHO it would be better if partial builds gave the same result when built as part of a super-build to avoid that an aggregator report is run in modules, the user must do the effort of configuring <inherit>false</inherit>: yes, this requires user work, but at least there is no magic maybe the right algorithm to choose if an aggregator goal needs to be run is: packaging==pom and has modules Regards, Hervé > > > Cheers, > > Brett > > > > -- > > Brett Porter > > [email protected] > > http://brettporter.wordpress.com/ > > http://au.linkedin.com/in/brettporter > > http://twitter.com/brettporter > > > > > > > > > > > > > > --------------------------------------------------------------------- > > 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]
