karaf-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Brian Topping (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (KARAF-884) karaf-maven-plugin should more closely map to POMs
Date Wed, 21 Sep 2011 17:48:08 GMT

    [ https://issues.apache.org/jira/browse/KARAF-884?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13109700#comment-13109700

Brian Topping commented on KARAF-884:

I stepped through the GenerateDescriptorMojo some and here's what I'm seeing:  

When project A is having it's features built (call it features A), the only time Project B's
feature file will generate a <features> element is if A's POM has a dependency on B
for an artifact of type=xml and classifier=features.  

Even so, the generated feature for A still has the <bundle> for B.  

My intuition as a user is that if Project A depends on Project B and Project B has a Feature
B, then Feature A should have a <feature> entry for B, not a reference to Bundle B.

https://github.com/topping/karaf/commit/4d6a10c0a10f183e4e1f34077f230676534222c3 has a rather
rough implementation of this.

> karaf-maven-plugin should more closely map to POMs
> --------------------------------------------------
>                 Key: KARAF-884
>                 URL: https://issues.apache.org/jira/browse/KARAF-884
>             Project: Karaf
>          Issue Type: Improvement
>          Components: karaf-tooling
>    Affects Versions: 3.0.0
>            Reporter: Brian Topping
> If karaf-maven-plugin is run against a set of POMs, the output of a feature generated
for a given project seems to be the transitive closure of the dependencies in the form of
a <bundle> entry for each dependency of the project.  This seems incorrect though. 
Consider if a parent project P builds A and B where A depends on X, B depends on A and Y.
 Currently, the feature for A will be a single bundle entry for X and the feature for B will
be three bundle entries for A, X & Y.  
> It seems to make more sense that the feature for B would actually be one <feature>
entry for A and a <bundle> entry for Y.  The <bundle> for X would be transitively
found through the feature definition for A.

This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira


View raw message