ComboBox Renderer Changes in 3.3.0

[kragoth]

Hi,

I'm trying to upgrade my application from Richfaces 3.2.2 to 3.3.0 and I'm having a bit of a problem.

Upgrading makes my jsfunit tests break, so I decided to investigate what was going on and I came across what I consider to be a rather nasty change that's been made in richfaces with regards to the ComboBox.

In 3.2.2 the html output for the combo box looked something like this

...
<div class="rich-combobox-list-position"
 id="MasterPageForm:reasonTypelistPosition">
 <div class="rich-combobox-list-decoration"
 id="MasterPageForm:reasonTypelistDecoration">
 <div class="rich-combobox-list-scroll"
 id="MasterPageForm:reasonTypelist">
 <span class="rich-combobox-item rich-combobox-item-normal">
 Client Interaction
 </span>
 <span class="rich-combobox-item rich-combobox-item-normal">
 Disbursement
 </span>
 </div>
 </div>
</div>

The output from 3.3.0 is:

...
<div class="rich-combobox-list-position"
 id="MasterPageForm:reasonTypelistPosition">
 <div class="rich-combobox-list-decoration"
 id="MasterPageForm:reasonTypelistDecoration">
 <div class="rich-combobox-list-scroll"
 id="MasterPageForm:reasonTypelist">
 <span class="rich-combobox-item rich-combobox-item-normal">
 gekko.domain.certificate.CertificateReason@fbc8ce
 [description=Client Interaction,targetType=/null/,deliveryDays=1,
 workgroup={proxy.id}=1,batchType=/null/,
 districtOfficeToRestrictTo=/null/,homeForm=CLIENT_INTERACTION,
 active=true,dateCreated=09/04/2009 11:27:15,
 userCreated={proxy.id}=1,dateUpdated=09/04/2009 11:27:15,
 userUpdated={proxy.id}=1,id=13,version=0]
 </span>
 <span class="rich-combobox-item rich-combobox-item-normal">
 gekko.domain.certificate.CertificateReason@3c72d7
 [description=Disbursement,targetType=PROPOSAL,deliveryDays=5,
 workgroup={proxy.id}=1,batchType=/null/,
 districtOfficeToRestrictTo=/null/,homeForm=DISBURSEMENT_FORM,
 active=true,dateCreated=09/04/2009 11:27:15,
 userCreated={proxy.id}=1,dateUpdated=09/04/2009 11:27:15,
 userUpdated={proxy.id}=1,id=18,version=0]
 </span>
 <span
 </div>
 </div>
</div>

As you can see, richfaces is now just using a toString() on my value object instead of getting the stringValue from the converter.

So where I used to be able to set the value in my jsfunit test to the value returned by my converter if I passed an object to it, now it doesn't work. Not only that the amount of bandwith that is wasted here is quite large. If I had a large object that I was using in my ComboBox you could be potentially sending "Megabytes" of data down the wire because of the call to toString().

I hope this is not the intended solution and is just an oversight cause this is really not going to work.


Maybe I've completely missed some change that was made that I need to go retrofit into my code, but after analysing the renderer I think that this is a change made by richfaces. Specifically in the public List encodeItems(FacesContext context, UIComponent component) method inside ComboBoxBaseRenderer.java.

[kragoth]

Infact the lines that cause the problem are here:

*Note I don't actually have the .java file this is just peeking in the class file :P *

 public String encodeSuggestion(FacesContext context, UIComponent component, Object value, String classes)
 throws IOException
 {
 ResponseWriter writer = context.getResponseWriter();
 String encodedValue = null;
 if(writer != null)
 {
 encodedValue = getConvertedStringValue(context, component, value);
 writer.startElement("span", component);
 writer.writeAttribute("class", classes, null);
 writer.writeText(value, null);
 writer.endElement("span");
 }
 return encodedValue;
 }

The encodedValue is assigned from the getConvertedStringValue method call, but when you call writer.writeText you pass in the object instead of the encoded value. I think this is wrong as previously it was writer.write(the encoded value)

Should I go raise a JIRA for this?

[ilya_shaikovsky]

https://jira.jboss.org/jira/browse/RF-6735

thanks for your digging!