The standard ActionForward and ActionInclude Actions create a RequestDispather in the Action, and do call another servlet directly, without going back to the container.
http://jakarta.apache.org/struts/api-1.0/org/apache/struts/actions/ForwardAction.html I recently suggested some additional methods for ActionServlet which would allow one Action to call another, in case there was a centralized Action that was rendering a view. I'm just waiting for someone to offer to test it, since this is not something I need to do myself. To support JavaServer pages, we do eventually have to forward back to the container, so the JSP can be rendered. AFAIK, there is not a portable way that the ActionServlet could "call" a JSP directly. So forwarding is the default because it should be able to get you anyplace you want to go. But nearly anything that is possible in a servlet, is possible in an Action, if you want to do it. Incze Lajos wrote: > No, the question is entirely independent from velocity. There are > many possibilities to render a view. E.g. Maverick "domifies" a data > structure and gives it to an XSL transformer, etc. What I was wonder > was simply the WRAPPING of the view helper. To me it seems to be > natural to hide (and make it configurable) the calling mechanism > of the view renderer. You can dispatch a servlet from a wrapper object > if yo want. What bothers me is that in ActionServlet a new servlet > dispath is wired. (I do understand that it can be a velocity servlet, > an xsl servlet etc.) The question is: why is the default to dispatch > a new servlet? What is my benefit over a simple call (that could dispatch > a new servlet if needed)? > -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
