Richfaces 4.5 - wrong reference to spacer.gif

[johannes2014]

Richfaces 4.5.17.Final (after upgrading from 4.3.6.Final)

This error also shows up with 4.5.9.Final

After upgrading from 4.3.6.Final to Richfaces 4.5.17.Final I see the following Exception when I open I page including an rich:extendedDataTable:

13:50:08,594 ERROR [org.apache.catalina.core.ContainerBase.[jboss.web].[default-host].[/myapp].[Faces Servlet]] (http-127.0.0.1:8080-1) JBWEB000236: Servlet.service() for servlet Faces Servlet threw exception: java.lang.UnsupportedOperationException

    at org.richfaces.resource.URLResource.userAgentNeedsUpdate(URLResource.java:79) [richfaces-core-4.5.17.Final.jar:4.5.17.Final]

    at com.sun.faces.application.resource.ResourceHandlerImpl.handleResourceRequest(ResourceHandlerImpl.java:255) [jsf-impl-2.1.28.SP10-redhat-1.jar:2.1.28.SP10-redhat-1]

    at javax.faces.application.ResourceHandlerWrapper.handleResourceRequest(ResourceHandlerWrapper.java:125) [jboss-jsf-api_2.1_spec-2.1.28.SP2-redhat-1.jar:2.1.28.SP2-redhat-1]

    at javax.faces.application.ResourceHandlerWrapper.handleResourceRequest(ResourceHandlerWrapper.java:125) [jboss-jsf-api_2.1_spec-2.1.28.SP2-redhat-1.jar:2.1.28.SP2-redhat-1]

    at org.richfaces.resource.ResourceHandlerImpl.handleResourceRequest(ResourceHandlerImpl.java:219) [richfaces-core-4.5.17.Final.jar:4.5.17.Final]

    at javax.faces.webapp.FacesServlet.service(FacesServlet.java:591) [jboss-jsf-api_2.1_spec-2.1.28.SP2-redhat-1.jar:2.1.28.SP2-redhat-1]

The cause for this is that there is a missing reference to spacer.gif:

http://localhost:8080/myapp/javax.faces.resource/spacer.gif.myportal-resource?ln=org.richfaces&v=sLjXbVU

The spacer.gif is loaded in

http://localhost:8080/myapp/org.richfaces.resources/javax.faces.resource/org.richfaces/extendedDataTable.ecss?db=eAG7dPvZfwAIqAOT

This is how extendedDataTable.ecss looks rendered in the browser:

.rf-edt-rsz {

background-image: url(/ddsadmin_2/javax.faces.resource/spacer.gif.myportal-resource?ln=org.richfaces&v=sLjXbVU);

cursor: e-resize;

height: 20px;

position: absolute;

left: 100%;

margin-left: -3px;

width: 6px;

}

and how extendedDataTable.ecss looks like in the WAR:

.rf-edt-rsz {

  background-image: "#{a4jSkin.imageUrl('spacer.gif')}";

  cursor: e-resize;

  height: 20px; /*TODO*/

  position: absolute;

  left: 100%;

  margin-left: -3px;

  width: 6px;

}

This is how to fix it yourself easily by overriding the css class (see also Chapter 3. RichFaces overview ):

Override the CSS class and remove the background-image reference:

<h:outputStylesheet target="head">

.rf-edt-rsz {

    background-image: none;

}

</h:outputStylesheet>

The error itself has nothing to do with the skin used (I turned it off using org.richfaces.enableControlSkinning=false in web.xml), I verified it also occurs with blueSky or DEFAULT.

The spacer.gif is present in richfaces-4.5.17.Final\META-INF\resources\org.richfaces.staticResource\4.5.17.Final\PackedCompressed\blueSky\org.richfaces.images\spacer.gif.

The error is caused by a wrong reference to the existing spacer.gif:

The correct rendered reference would be

This is the wrong reference rendered in the ecss file:

http://localhost:8080/myapp/javax.faces.resource/spacer.gif

This it how it should look like (this reference works):

http://localhost:8080/myapp/org.richfaces.resources/javax.faces.resource/org.richfaces/spacer.gif

So the "org.richfaces.resources" is missing.

[michpetrov]

It's possible a new JSF version broke the resource handling, what server are you using?

[johannes2014]

I'm using Jboss EAP 6.4.9.GA / JSF 2.1.28. For me it seems there is a problem with the reference "#{a4jSkin.imageUrl('spacer.gif')}" when being translated to the real url in the ecss file.

[michpetrov]

Well, everything works fine for me. Are you using any filters?

[johannes2014]

Only one filter to add no-cache headers to every request, but nothing more.

[michpetrov]

It's still probably due to some settings, what does your web.xml look like?

[johannes2014]

I cleaned out any richfaces settings in web.xml, but the error still occurs (only javax.faces.FACELETS_LIBRARIES/javax.faces.webapp.FacesServlet/security-constraints left).
However, I'm running at a customized Jboss EAP container, so that could be causing it.

Where can I find the sources for the Richfaces samples for 4.5.17.Final so I can try it myself?

Please make sure you open a page with an rich:extendedDataTable in it, otherwise the error will not show up.

[michpetrov]

Our sources are on github, the jpa-demo is probably the simplest, just change the dataTable to EDT.

[johannes2014]

You're right, with the jpa-demo everything works fine.

I tracked it down that the external container is somehow influencing the way the urls are resolved in the ecss file.

However, I cannot influence the external framework directly.

Is there any way to overrule the way the urls are resolved in my WAR?

I nearly solved it by overriding the css class, however, for some browsers like Internet Explorer this is not working.

Do you have another idea maybe?

[johannes2014]

I tracked it down some more, the problem is caused by the external container using Richfaces 4.2.1.Final. This is the cause why the a4jSkin.imageUrl() is resolved differently.

However, I think it would be fine if there would be no stacktrace if an image is not found (404), probably a warning would be sufficient. Otherwise stacktraces are produced with every click onto a view with a rich:extendedDataTable.

I now have to revert to 4.3.6.Final and wait until the container itself is updated to a newer version or there are at least no more stacktraces produced by org.richfaces.resource.URLResource.userAgentNeedsUpdate().

[michpetrov]

The issue is that org.richfaces.resource.URLResource.userAgentNeedsUpdate() is not supposed to be called, that's why UnsupportedOperationException is thrown. But I think that's a separate issue from the a4jSkin.imageUrl() producing a wrong path. Does it occur if you use packed resources?

[johannes2014]

Yes, the org.richfaces.resource.URLResource.userAgentNeedsUpdate() stacktrace is a separate issue, probably every missing resource will cause this exception.

I'm not using any packed resources.

[michpetrov]

Well, have you tried?