I think it largely depends on if you want service injection or not. If your "component" requires other services it might make sense to make it plexus-managed.
As an example, I have been irritated the the plexus-io collections are themselves plexus-components. This is because they get the plexus logger injected by the container. The alternative to container injection is to move the required service to the interface that the client code is exposed to. In some cases this means the client needs to "know" things it shouldn't; by moving the dependency to the "business-level" interface of your component you require your user to have a relationship to that component as well. (When it comes to logging, most log frameworks use static-method based injection - which probably might have been a better choice for plexus logging too, which makes it even more irritating that this was the sole reason they were made components in the first place). So using injection can make the client "see" less of your component, which may be a good thing. Kristian 2014-12-02 18:30 GMT+01:00 Benson Margulies <bimargul...@gmail.com>: > I'm working on a plugin where there's a use-case in which the user > will want to add something to the plugin's classpath and then > configure a class. > > Is there any advantage to using plexus injection for this as opposed > to just ordinary classpath fishing for a named class? > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org > For additional commands, e-mail: dev-h...@maven.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org For additional commands, e-mail: dev-h...@maven.apache.org
