I have a situation where many of my projects need to produce an adjunct
artifact (in addition to the primary JAR) for later collection into my
distribution image.
I'd like to simplify the process that I use to generate the distribution
image, as right now the POM for the distribution project has lots of
dependency declarations that look like this:
<dependency>
<groupId>com.ibm.group</groupId>
<artifactId>foo</artifactId>
<version>${project.version}</version>
</dependency>
<dependency>
<groupId>com.ibm.group</groupId>
<artifactId>foo</artifactId>
<version>${project.version}</version>
<classifier>adjunct</classifier>
<type>zip</type>
</dependency>
The second declaration is essentially a duplicate of the first, just with
the classifier & type added. I'd like to get rid of the second
declaration. I'd like the assembly plugin to automatically discover the
adjunct artifacts if they exist for projects that are already declared as
dependencies for the primary artifact.
I know that I can use the classifier & type mapping capabilities of the
dependency plugin to do this, unpack the adjuncts & then use a fileset in
my assembly descriptor. But I can't help think that it would be cleaner if
the assembly plugin could just handle this internally.
Has there ever been such a use case put forward? Is there a better way to
handle this requirement? (please do not suggest that I restructure my
projects to eliminate the need for the adjunct artifacts)
--JEC
