On 10 Dec 2015, at 20:50, Denis Golovin wrote:
On Wed, Dec 2, 2015 at 1:21 AM Max Rydahl Andersen
<manderse(a)redhat.com>
wrote:
> Nice writeup.
>
> I think at this time obviously we have to change
> maven-dependency-plugin declaration
> in parent/pom.xml#build/pluginManagement to never skip downloading
> dependencies, because it is most common use case. After it is done we
> can
> remove unnecessary <skip>false</skip> throughout the projects.
>
> Yes - your conclusion matches mine on the jira and what mistria
> suggestion
> for fixes
> and with my updates handles.
>
> Can you make it happen ? :)
>
I sure can :)
Great!
> If there are libs for (2) and (3) downloaded from external URL
using
> download-maven-plugin/maven-download-plugin we should probably move
> them to
> locus.
>
> Main problem I know is forge, hibernate and openshift afaik.
>
> forge and openshift would not be able to move very fast if we moved
> their
> libs for Locus.
>
> Hibernate - we don't actually want/intend to expose the internal apis
> of
> hibernate so not
> sure if pushing all variations of hibernate into Locus is such a good
> idea....but it could
> be considered.
>
> One thing that might be worth looking for are those bundling mockito
> or
> other testing libs
> that we might already have put into locus or use from orbit so these
> are
> not even relevant anymore.
>
I mean if we have libs downloaded from external URL, that is stable
enough
to be included into locus, but it seems we don't have cases like this.
so fix would be removing skipping for maven-dependency-plugin if
tests
skipped and then clean up not needed <skip> tags throughout the
projects
and manually configure skipping for (1).
Yes, I think the last few comments on
https://issues.jboss.org/browse/JBIDE-15332
outline that specific solution too.
/max
http://about.me/maxandersen