Hi,
I'm relatively new to JSF. The project I'm working on required use of the tree2 component. I've had to do some customizations to this and I've also incorporated AJAX into it using Ajax4JSF. I followed this, http://www.jroller.com/plainoldweblog/entry/use_tomahawk_tree2_and_ajax4 jsf, not exactly the same but I adapted his example. I've also rewritten the renderer based on the existing one. I got rid of all those tables. It would be really nice to see an implementation that doesn't rely on tables. Mine doesn't, but it's incomplete ... it works for what we need it for but it's far from a replacement to the one that comes with Tomahawk. If I have time (unlikely) maybe I'll write a complete one. So onto the problem: I've extended the TreeModelBase and TreeNodeBase to meet our needs here. He have some backend data that will be loaded into the tree. We don't want a round trip to the server every time, and loading everything up front would be too slow (sound familiar?), hence AJAX. I was able to get that part working quite well. Only problem is that it requires use of a session managed bean instead of a request bean. I can live with this. Now we have a drop down list of sites, or I guess you could call them topics, not important, but when a user changes topics the tree needs to be reloaded with a different set of data. Here's where that session bean became a problem. Since it was a session bean, it kept the same data from before. I came up with a fix for this, I was able to use an application context variable to detect when a user changes sites, and then just reinitialize the data model when this happens. That part works fine. Now for my actual problem... the data model initializes fine, but the tree state does not reset itself. So the end result is nodes already expanded in the new tree. They're not fully expanded since the child nodes aren't actually loaded, but they appear expanded, and a user would have to click twice to actually expand them. First click collapses the node, then second click re-expands it and actually loads the data. Stepping through the renderer code I found that the call: TreeState state = tree.getDataModel().getTreeState(); Is returning a previously stored TreeState. Here's my code: protected void buildTree(List list, String type) { TreeNode rootNode ; if (model == null) { rootNode = new MenuTreeNode(Constants.TREE_ROOT_NODE, null, "root", false); model = new MenuTreeModel(rootNode); model.setTreeState(new TreeStateBase()); // this isn't necessary, was just using for testing } else { rootNode = model.getRootNode(); } this.addChildNodes(rootNode, list, type); } If I use the debugger and stop in the above method, I can see it gets executed before the renderer, and comparing the object ids (visually via the debugger) I can see they're different. Basically to reset the tree I just set it null which forces a call to this method here. A bit of further browser of the tree2 source, I found that the above mentioned call... tree.getDataModel().getTreeState() is doing something like this... if (_cachedModel != null) { return _cachedModel; } What I don't understand is why the tree state is cached yet the model is not, since it loads the new model properly. So... how can get it to use a clean TreeState? Thanks for any help on this.
