Question

When I use convertpom task in Ivy to convert a pom.xml to ivy.xml, I get default publications

<publications>
    <artifact name="XYZ" type="jar" ext="jar" conf="master"/>
</publications>

How do I modify pom.xml, so that more artifacts are added to this in conversion. Where does converttopom pick up the artifacts from. Also, how do I change the type?

Is it possible to override these in ivy:publish call?

found this piece of code in convertpom ant task, not sure how it is used.

   private void addSourcesAndJavadocArtifactsIfPresent(
         PomModuleDescriptorBuilder mdBuilder, ParserSettings ivySettings) {
     if (mdBuilder.getMainArtifact() == null) {
         // no main artifact in pom, we don't need to search for meta artifacts
        return;
     }
     ModuleDescriptor md = mdBuilder.getModuleDescriptor();
     ModuleRevisionId mrid = md.getModuleRevisionId();
     DependencyResolver resolver = ivySettings.getResolver(
         mrid);

    if (resolver == null) {
        Message.debug("no resolver found for " + mrid 
                         + ": no source or javadoc artifact lookup");
    } else {
        ArtifactOrigin mainArtifact = resolver.locate(mdBuilder.getMainArtifact());

        if (!ArtifactOrigin.isUnknown(mainArtifact)) {
            String mainArtifactLocation = mainArtifact.getLocation();

            ArtifactOrigin sourceArtifact = resolver.locate(mdBuilder.getSourceArtifact());
            if (!ArtifactOrigin.isUnknown(sourceArtifact)
                    && !sourceArtifact.getLocation().equals(mainArtifactLocation)) {
                Message.debug("source artifact found for " + mrid);
                mdBuilder.addSourceArtifact();
            } else {
                // it seems that sometimes the 'src' classifier is used instead of 'sources'
                // Cfr. IVY-1138
                ArtifactOrigin srcArtifact = resolver.locate(mdBuilder.getSrcArtifact());
                if (!ArtifactOrigin.isUnknown(srcArtifact)
                        && !srcArtifact.getLocation().equals(mainArtifactLocation)) {
                    Message.debug("source artifact found for " + mrid);
                    mdBuilder.addSrcArtifact();
                } else {
                    Message.debug("no source artifact found for " + mrid);
                }
            }
            ArtifactOrigin javadocArtifact = resolver.locate(mdBuilder.getJavadocArtifact());
            if (!ArtifactOrigin.isUnknown(javadocArtifact) 
                    && !javadocArtifact.getLocation().equals(mainArtifactLocation)) {
                Message.debug("javadoc artifact found for " + mrid);
                mdBuilder.addJavadocArtifact();
            } else {
                Message.debug("no javadoc artifact found for " + mrid);
            }
        }
    }
}

Answer 1

This really demonstrates one of the key differences between Maven and ivy…

An ivy file explicitly lists all the files contained in the module. A Maven POM on the other hand does not. Instead zero or more additional files can be stored in Maven each with a different “classifier” to make the filename unique.

I can’t see any way to build the complete “publications” section in ivy without having access to the Maven module’s filesystem. Are you using a Maven repository manager? Nexus has a REST API that perhaps you could invoke to obtain all files in a module (Just a thought)

Another idea is to submit a request to extend the convertpom task. Create some optional child tags that enable you to list the available classifiers:

<ivy:convertpom pomFile="pom.xml" ivyFile="ivy.xml">
    <classifier name="sources"/>
    <classifier name="javadoc"/>
    <classifier name="archive" type="tar.gz/>
</ivy:convertpom>

I don’t see this change being very popular (or useful). Most people are converting in the other direction using the makepom task.