Details
Bug
Description
For example, using the Chat portlet as an example, assume there are 2 instances on the portal page.
1) Interact with the first portlet by logging in.
2) Interact with the second portlet by logging in. At this point, the response contains an update to the first portlet at the <form> level (this should likely be more granular but that's a different issue). The updated form does NOT include the ViewState. There is another update for the ViewState in the response, but it is not correctly applied to the form.
3) Interact with the first portlet again by sending a message. Since this form does not contain the ViewState, the resulting request results in a new View and the UIViewRoot does not contain the old DOM as an attribute (which it would if the ViewState was provided). The resulting update comes back as a ViewRoot update, replacing the entire page.
So the issue seems to be center around why the ViewState is not properly restored to the form when it's updated.
Issue Links
- blocks
-
ICE-6018
Upgrade Chat portlet example
-
- Closed
-
Activity
- All
- Comments
- History
- Activity
- Remote Attachments
- Subversion
Attaching a patch to jsf.js that corrects the issue.
Recorded issue in Mojarra's bug tracking system: https://javaserverfaces.dev.java.net/issues/show_bug.cgi?id=1814
After wading through the repositories that are in a bit of flux, the change has been applied as of Nov 19 and is available as part of the JSF 2.1.0_b08 release:
------------------------------------------------------------------------
r8730 | sheetalv | 2010-11-23 18:00:08 -0800 (Tue, 23 Nov 2010) | 1 line
Tag 2.1.0 Build 08
------------------------------------------------------------------------
r8727 | rogerk | 2010-11-19 13:51:47 -0800 (Fri, 19 Nov 2010) | 1 line
Mojarra Issue: 1814: Ensure jsf.js is loaded once if referenced more than once.
The latest Glassfish that I could download as of today only seemed to included 2.1.0_b07.
Re-opening the case.
Due to a bug in their version checking, the Mojarra jsf.js script gets evaluated multiple times - once for each portlet. This causes the some of our handlers to be unregistered which, in turn, prevents our ViewState fix for forms from being applied.
The bug that was submitted to Mojarra (https://javaserverfaces.dev.java.net/issues/show_bug.cgi?id=1814) has been fixed and is available in a 2.1 tag (http://java.net/projects/mojarra/sources/svn/show/tags/2.1.0-b08) but is not available in the latest 2.0 tag (http://java.net/projects/mojarra/sources/svn/show/tags/2.0.4-b06). I tried the 2.1 release but there are a number of issues with the the PortletFaces Brdige that currently prevent it from working reliably. We've been using a 2.0.4-b05 release for portlets up to this point so it's been relatively well tested. I've checked out the 2.0.4-b06 version myself from their repository, applied the very simple patch that we originally supplied to them, and rebuilt our own custom version of 2.0.4_b06 that includes this change. The slightly modifed versions of the libraries are now checked in under icefaces/lib/portlets/jsf-*.jar.
Modified versions of 2.0.4-b06 libs are checked in with the patch applied. Resolving as fixed until such time as an official release which includes the patch can be used. Testing the the chat portlet shows that the problem of the missing ViewState is gone and the portlets no longer do a full ViewRoot update because of this.
It turns out that the check that jsf.js makes to prevent itself from being re-evaluated if there is more than one copy loaded is faulty. The current version of jsf.js does this:
if (!((jsf&& jsf.specversion&& jsf.specversion > 20000 ) && (jsf.implversion&& jsf.implversion > 2))){...
with the current specversion == 20000 and the current implversion being 3. The check should be changed to:
if (!((jsf&& jsf.specversion&& jsf.specversion >= 20000 ) && (jsf.implversion&& jsf.implversion >= 3))){...
so that it will only evaluate the file if jsf has not yet been defined by something equivalent or higher than the current values. The problem we have is that we are relying on registering callbacks to repair the ViewState in forms (a different bug that we are working around currently). We do this by checking every incoming update, retrieving the ViewState update if it's present, and applying it to any forms that don't have it applied. If a second jsf.js file gets re-evaluated, our callbacks are no longer registered and the ViewState fix doesn't get applied.