> > Now, the instants of time where PostAddToViewEvent is published are: > > - With Partial State Saving enabled > > * When the view is build at first time. > * In a postback when the view is build but before the state of the > component is restored. > > - With Partial State Saving disabled > > * When the view is build at first time. > * In a postback when the view is "refreshed", because all component nodes > are detached and attached to the tree. In other words, on render response > phase, vdl.buildView is called and in this case facelets algorithm add all > transient components (usually all html markup not saved), and to do that, it > detach and attach all components to be in right order. This also has some > other implications, like c:if tag depends on this to work correctly. > This sounds ugly. You are sure that this is the behaviour in both MyFaces and Mojarra. Here my 2cents:

- Partial state saving should not make any difference at all (also in partial state saving, the template should be reapplied before rendering, see my other mail) - PostAddToViewEvent should be called always after a component has been added to the view fully, and all its state has been restored

Hi, > - Create a new event (let's call it PostBuildViewEvent for the moment), that > is > propagated every time the view is build (when it is created, restored or > refreshed before render response). h:outputScript, h:outputStylesheet, > cc:insertChildren and cc:insertFacet should add a listener for this > event and handle all relocation code at response of this event. > > - Fix PostAddToViewEvent/PreRemoveFromViewEvent publishing conditions. > In my opinion this ones should not be called when the view is "refreshed". > Martin suggestion about the publishing conditions are reasonable. yes, this sounds good. Leonardo, for the second event, how would you suggest the publishing to happen (actual sourcecode)? best regards, Martin -- http://www.irian.at Your JSF powerhouse - JSF Consulting, Development and Courses in English and German Professional Support for Apache MyFaces

Hi On this link: https://issues.apache.org/jira/browse/MYFACES-2638 attachment: MYFACES-2638-3.patch I attached a proposal to fix this issue, based on myfaces code. The idea is create a event called javax.faces.event.PostBuildRefreshViewEvent. This event is quite different from the one proposed in: [jsr-314-open] add component resources depending on the owner component state because there is one case this one is not published: when partial state saving is enabled and there is no call to vdl.buildView for refresh. I also notice that there is no "inheritance" between events, that means, if I have two events and one extends from the other and there is a listener attached to the parent, when the child event is raised the listener is not notified. h:outputScript, h:outputStylesheet, cc:insertChildren, cc:insertFacet needs to attach its listeners to that event too, to give the chance to relocate their related components correctly. The listener that track changes on the component tree and mark them to be saved/restored fully does not need to listen this event. PostAddToViewEvent/PreRemoveFromViewEvent should continue working as always, the important here is disable event processing while we are refreshing the view. The side effect is components added to the component tree when view is refreshed (c:if case) will not receive that event. If we want to avoid that side effect, the proposal is change PostBuildRefreshViewEvent to PostBuildViewEvent, this one will be an event specific for relocation and that one should be published after the view is build, in other words, in the same place as we are doing PostBuildRefreshViewEvent but for all cases, but that one requires remove the register of the listeners to PostAddToViewEvent on h:outputScript and related. Also, it is necessary to indicate in some way when publish PostAddToViewEvent/ PreRemoveFromViewEvent by facelets algorithm for build/refresh view, because it is there where we have the knowledge about when propagate it or not. I think at this point it is clear the problem and the possible direction to solve it, so I'll create an issue for this one on the spec issue tracker. regards, Leonardo Uribe 2010/3/30 Leonardo Uribe <lu4242(a)gmail.com&gt; Hi Martin > > In theory the secod event (PreRemoveFromViewEvent) should not be called > when the view is "refreshed". > > The important for me at this point is make clear the problem, but in this > point > I think the better is do a prototype with myfaces. It will take some time, > but as > soon as I have something I'll publish the results, so we can discuss > a more concrete proposal. I'll integrate also the solution for the problem > > "[jsr-314-open] add component resources depending on the owner component > state" > because it seems it is related. > > regards, > > Leonardo Uribe > > 2010/3/30 Martin Marinschek <mmarinschek(a)apache.org&gt; > > Hi, >> >> > - Create a new event (let's call it PostBuildViewEvent for the moment), >> that >> > is >> > propagated every time the view is build (when it is created, restored >> or >> > refreshed before render response). h:outputScript, h:outputStylesheet, >> > cc:insertChildren and cc:insertFacet should add a listener for this >> > event and handle all relocation code at response of this event. >> > >> > - Fix PostAddToViewEvent/PreRemoveFromViewEvent publishing conditions. >> > In my opinion this ones should not be called when the view is >> "refreshed". >> > Martin suggestion about the publishing conditions are reasonable. >> >> yes, this sounds good. Leonardo, for the second event, how would you >> suggest the publishing to happen (actual sourcecode)? >> >> best regards, >> >> Martin >> -- >> >> http://www.irian.at >> >> Your JSF powerhouse - >> JSF Consulting, Development and >> Courses in English and German >> >> Professional Support for Apache MyFaces >> > >