Hi all. ***This is an architectural question about STRUTS, not a syntactic one*** I'm new to the STRUTS-user group, so apologies if this is a FAQ-type question - would appreciate redirection if it is... I have been reading about the JSP Model 2 architecture - seems great for what we need. But a few of us are having a discussion that I'm hoping you may be able to help with. Here's one proposed J2EE stack for our work -JSP -DHTML using XML/XSLT data islands -using usebean calls to populate the XML -RequestController -creating/referencing a number of request handlers -using a central 'browse session' object for application state -forwarding the user onto the relevant JSP for a particular request -EJB (stateless session) -Data abstraction layer (not entity EJB) -Data sources (format unimportant) I'm going to leave the description there, because I'd like some of the more experienced designers out there to suggest questions about the above stack. If no questions arise, allow me to ask a question of you: =========== What kind of 'state' do you think could/should be stored in the central 'browse session' object? =========== Should this be UI-specific information (such as 'this user is currently on page 4 out of 10 of search results') or more generic UI-independent information (such as the 100 search results themselves, which may be displayed 10 at a time, or 100 all at once). In the above stack, the 'browse session' speaks to the stateless session EJB, gets the results, stores all 100 results and then forwards the user to a JSP that either shows 10 results at a time, or all 100 results. It also deals with the pagination between pages when there are 10 results/page. Is this sensible, or is there a point to be made in not bundling it all together? Cheers Dave Small
