Hi guys, until now we were merging all beans.xml in a single one. This is more or less ok but with CDI 1.1 it can be really broken.
Issue is it works for almost everything (including scanning for most apps) but not for bean-discovery-mode which specify if a jar/folder should be scanned as before, ignored or scanned for CDI classes only (ie @Dependent mandatory if no other annotation are specifiied). Today we'll take the config of the WEB-INF/classes folder which is surely not what we want. In CdiScanner we have a map of beans.xml url -> classes but we don't link it yet with Beans (our representation of beans.xml). Main question I have is: do we create in our info tree a Beans by url (which should have been the case since EE 6) or do we hack something else? wdyt? Personally I'm to keep track of beans.xml url in org.apache.openejb.config.DeploymentLoader#addBeansXmls(org.apache.openejb.config.WebModule) and propagate it in info tree then use it to read it correctly. Romain Manni-Bucau Twitter: @rmannibucau Blog: http://rmannibucau.wordpress.com/ LinkedIn: http://fr.linkedin.com/in/rmannibucau Github: https://github.com/rmannibucau
