i guess thats my point, why do we need a message-passing framework as part of core? its outside the scope of a web app framework.
i guess what i was hoping for, although i dont know if its possible, is installing a httpsessionattributelistener or something similar that can monitor the attributes in the session and then on the attribute changed we would check if its a page from another node and save it to disk. that way we piggy-back on whatever clustering mechanism the servlet container uses - we dont need to have our own parallel messaging framework to replicate the pages. like i said, dont know how feasible that is, matej should know more. -igor On Dec 14, 2007 11:09 PM, Eelco Hillenius <[EMAIL PROTECTED]> wrote: > > On Dec 14, 2007 11:07 PM, Eelco Hillenius <[EMAIL PROTECTED]> wrote: > > On Dec 14, 2007 10:57 PM, <[EMAIL PROTECTED]> wrote: > > > I don't know if these should go into core just yet. The cluster > > > support I put on the wiki was going to work differently without any > > > external libs. Basically each node would write the replicated current > > > page of the other node to disk so each node has all page of all other > > > nodes on disk. That's what I meant by transparent. Marek? > > > > That's not a configuration I would prefer, but can't we fold that into > > the same project? Did you start from scratch? It would be unfortunate > > if we only have wicket-stuff projects for this. > > Btw, page map replication is one optional part (project); the core > mainly exists for generic message passing that is used for normal > session clustering. That only imho is worth having a project for. > > Eelco >
