On Thu, May 22, 2008 at 5:21 PM, tomas darbois (JIRA) < tuscany-dev@ws.apache.org> wrote:
> > [ > https://issues.apache.org/jira/browse/TUSCANY-2296?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12599068#action_12599068] > > tomas darbois commented on TUSCANY-2296: > ---------------------------------------- > > I'm facing the same problem, and I was hoping to see a solution or > workaround , instead of reply with Work as Designed :p > > I ve several composite files that are used depending on the deployement > (used by several applications, and in various servers that aren't identical) > and I was starting to see a good way to design it in my application, > grouping the composites in the same place so that I was having a quicker > control on them. > > I will use some tricks to workaround that problem, but it's considered for > me as a really bad point for future enhancements of my application. > > > SCADomain.newInstance() processes all composites in the contribution > irrespective of the parameter passed > > > --------------------------------------------------------------------------------------------------------- > > > > Key: TUSCANY-2296 > > URL: https://issues.apache.org/jira/browse/TUSCANY-2296 > > Project: Tuscany > > Issue Type: Bug > > Components: Java SCA Embedded Runtime > > Affects Versions: Java-SCA-1.2 > > Reporter: Vamsavardhana Reddy > > Fix For: Java-SCA-Next > > > > Attachments: TUSCANY-2296-recreate.patch > > > > > > I have two composites, namely "proper.composite" and "error.composite", > in my test project. "proper.composite" is used to test proper contribution > without any errors whereas "error.composite" is used to test a problematic > contribution. But when I initialize a domain using SCADomain. > newInstance("proper.composite"), I am getting an exception due to > "error.composite". Though it looks as if the call SCADomain. > newInstance("proper.composite") processes only "proper.composite", in > reality it is processing all the .composite files present in the project > irrespective of the parameter passed to SCADomain.newInstance(). > > -- > This message is automatically generated by JIRA. > - > You can reply to this email to add a comment to the issue online. > > I don't like the behaviour either. We have a new Node implementation [1] that does a very similar thing to the SCADomain class but allows us to separate the Domain configuration from the actual runtime (the Node). The node interface is a little complicated in this particular scenario at the moment but you can direct it toward a specific directory to read as a contribution and read a named composite. This is useful as it removes this behaviour where SCADomain just uses the composite name to locate a contribution and then loads all composites there. I used this in some of our validation tests [2] node = nodeFactory.createSCANode(new File("src/main/resources/ComponentReferenceTargetNotFound/Calculator.composite").toURL().toString(), new SCAContribution("TestContribution", new File("src/main/resources/ComponentReferenceTargetNotFound").toURL().toString())); We have talked about adding a helper method (from an old Node implementation we have) that allows a simpler API node = SCANode2Factory.createSCANodeWithComposite("DuplicateComponentName/Calculator.composite"); We haven't got round to this yet but I think we would make this use the directory where Calculator.composite is as the contribution and have it load just Calculator.composite. Thoughts? Simon [1] http://svn.apache.org/repos/asf/incubator/tuscany/java/sca/modules/node2-api/ [2] http://svn.apache.org/repos/asf/incubator/tuscany/java/sca/itest/validation/src/test/java/calculator/warning/DuplicateComponentNameTestCase.java
