gnodet commented on a change in pull request #616:
URL: https://github.com/apache/maven/pull/616#discussion_r785711956
##########
File path:
maven-embedder/src/main/java/org/apache/maven/cli/internal/BootstrapCoreExtensionManager.java
##########
@@ -114,14 +123,42 @@ private CoreExtensionEntry createExtension( CoreExtension
extension, List<Artifa
{
String realmId =
"coreExtension>" + extension.getGroupId() + ":" +
extension.getArtifactId() + ":" + extension.getVersion();
- ClassRealm realm = classWorld.newRealm( realmId, null );
- log.debug( "Populating class realm " + realm.getId() );
- realm.setParentRealm( parentRealm );
+ final ClassRealm realm = classWorld.newRealm( realmId, null );
+ Set<String> providedArtifacts = Collections.emptySet();
+ String classLoadingStrategy = extension.getClassLoadingStrategy();
+ if ( STRATEGY_PARENT_FIRST.equals( classLoadingStrategy ) )
+ {
+ realm.importFrom( parentRealm, "" );
+ }
+ else if ( STRATEGY_PLUGIN.equals( classLoadingStrategy ) )
+ {
+ coreExports.getExportedPackages().forEach( ( p, cl ) ->
realm.importFrom( cl, p ) );
+ providedArtifacts = coreExports.getExportedArtifacts();
+ }
+ else if ( STRATEGY_SELF_FIRST.equals( classLoadingStrategy ) )
+ {
+ realm.setParentRealm( parentRealm );
+ }
+ else
+ {
+ throw new IllegalArgumentException( "Unsupported class-loading
strategy '"
+ + classLoadingStrategy + "'. Supported values are: " +
STRATEGY_PARENT_FIRST
+ + ", " + STRATEGY_PLUGIN + " and " + STRATEGY_SELF_FIRST );
+ }
+ log.debug( "Populating class realm {}", realm.getId() );
for ( Artifact artifact : artifacts )
{
- File file = artifact.getFile();
- log.debug( " Included " + file );
- realm.addURL( file.toURI().toURL() );
+ String id = artifact.getGroupId() + ":" + artifact.getArtifactId();
+ if ( providedArtifacts.contains( id ) )
+ {
+ log.debug( " Excluded {}", id );
+ }
+ else
+ {
+ File file = artifact.getFile();
+ log.debug( " Included {}", file );
Review comment:
> > If we can't agree on that solution, I'd go for the alternative using
two statements, one at debug and one at trace level.
> > ```
> > log.debug( " Included {}", id );
> > log.trace( " -> {}", file );
> > ```
>
> That's not really better because it is split now in two lines which makes
it even harder to grep.
Yes, I do agree and that's exactly why I did not want to use that.
> Counter questions:
>
> * Why not display the path always with debug?
Yes, let's do that, it may be the simplest way forward.
> * Why display the path with excludes, but not with inlcudes?
I think it only makes sense when included for inspection when dealing with
classloading problems. When not included, the exact content does not matter so
much imho.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
