No transient fields. Everything is serializable. We're storing the complete objects in the model for a number of reasons. When we come back to the page we have some object but the state is gone. I think it's because we're redirecting back to the same page class/type so the old one gets dumped. Probably what I'll have to do,then, is to build special subclasses for the drilldown so that the page type changes on each step. Unless there's some way to increment the number of versions per pagemap...
On Nov 27, 2007 11:39 AM, Johan Compagner <[EMAIL PROTECTED]> wrote: > Yes in wicket 1.3 there is only 1 active page per pagemap > > But it shouldn't matter where the page comes from. If it is serialized > from > disk > or serialized in the session somehow by the container, You still should be > able > to fully construct all your data again in the objects. > > So what is suddenly null? All the transient fields? Why aren't those > reloaded > from a database or something? > > johan > > > > On Nov 27, 2007 5:01 PM, Evan Chooly <[EMAIL PROTECTED]> wrote: > > > We have a page that presents a report in which we can "drill down" into > > the > > data. When this happens, we setResponsePage() back to a new instance of > > the > > page with some different parameters and that all works dandy. But when > we > > hit the back button, it seems the old version is gone. All the model > data > > comes back null (though the model object itself is not). So when we > click > > back and then click another link to drilldown in another direction we > get > > NPEs. From stepping through DiskPageStore, et. al, it would appear that > > since the page class isn't changing, the new page bumps the old one from > > the > > cache so that only one version of the page. > > > > Am I correctly interpreting what I'm seeing and is there a way to set > the > > number of versions of a page? > > >
