We could do this also in (on)beforeRender, so there wouldn't be another traversal.
-Matej On 9/27/07, Igor Vaynberg <[EMAIL PROTECTED]> wrote: > well, what if instead of detecting this on urlfor call you simply traverse > the object graph and see if any components implement the request listener > interface, and if they do then the page is stateful. so you basically > measure a potential for the urlfor call... > > -igor > > > On 9/27/07, Matej Knopp <[EMAIL PROTECTED]> wrote: > > > > Hi all, > > > > we have a problem currently. We have a situation, when session is > > bound in the middle of page rendering. That causes problems with > > appending session id to URL. > > > > Let's have a situation like this: > > Page with 1 bookmarkable link and one listener interface link. This is > > the first page rendered, session is not bound yet. > > > > When the bookmarkable link is being rendered, the container doesn't > > append session id to it, because the session is not bound yet. It's > > bound only when the listener interface link is being rendered, as at > > that point wicket knows that the page is stateful. So on rendered > > page, the first bookmarkable link is without session id. > > > > I think the proper solution would be doing a bit of less magic - each > > component should report it's statefullness even without rendering the > > URL, so we should be a able to determine if page is statefull after > > beforeRendere, before the actual rendering. So the session could be > > bound properly and all links on page would contain session Id. > > > > -Matej > > >