[Jboss-dev-forums] [Design of JBoss Build System] - Re: More on component-info.xml usage

[<ad...@jb...>]

"sco...@jb..." wrote : So that is a good question. Actually the naming-tests should include the jnp-client.jar rather than jnp-server.jar, with the assumption that the consumer of the naming-tests export artifcact will configure the server given the naming-server artifcat. This illustrates a problem with semantic artifcat definition like a tests input artifcat. It depends on how the test is going to be run. If the test driver is a component like the jbossas testsuite, the jnpserver.jar should not be in the naming-tests. Even if the driver is a standalone unit test target that needs to configure a standalone naming server the, the unit test setup target should express that it needs both the naming-server and naming-tests component artifacts and dependencies. 
  | 

That was my comment above about junit-config.
However, I don't see any reason why jnpserver cannot be unit tested from standalone tests. Other complications like transport, security, ha enhancements would be
tested elsewhere as part of integration tests.

anonymous wrote : 
  | For the common component dependency, I was assuming this was being used by the tests as in the case of the org.jboss.logging.Logger. There should also be a dependency on some base test harness which likely includes common.
  | 

This already exists in the "test" project.
http://jira.jboss.com/jira/browse/JBAS-1768

View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3878095#3878095

Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3878095