The issue still exists. See ICE-6570 .
There is a patch attached.

Using some of the test code from this jira, I was able to reproduce the issue. And using code based from the patch from ICE-6570, the issue was resolved.

One lingering side note is that the dom differencing still detects a change for the inputTextarea rendering, since the old dom contains the regular string, and the new dom contains and escaped sequence. This might indicate that our dom save/restore deserialization code unescapes characters.

<textarea id="example-form:Comments">[0]<#text> :: Value changed from 'HELLO ????? THERE' to 'HELLO £¢?üå THERE'

At least the text is no longer being corrupted.

icefaces 3 trunk
Subversion 28105

icefaces-3.0.x-maintenance
Subversion 28106

To reproduce / debug the issue, I made the following code changes to the showcase ace:fileEntry example.

Added this to: icefaces3/samples/showcase/showcase/src/main/webapp/resources/examples/ace/fileentry/fileentry.xhtml
<h:panelGrid columns="2">
<ice:outputText value="Name :"/>
<ice:panelGroup>
<ice:inputText id="Name" value="#

"/>
<ice:message for="Name"/>
</ice:panelGroup>
<ice:outputText value="Comments :"/>
<ice:panelGroup>
<ice:inputTextarea id="Comments" value="#

" cols="40" rows="3" />
<ice:message for="Comments"/>
</ice:panelGroup>
</h:panelGrid>

And made it so it no longer had required="true" on the fileEntry in that form, so I wouldn't have to actually upload a file to test this.

And this to: icefaces3/samples/showcase/showcase/src/main/java/org/icefaces/samples/showcase/example/ace/file/FileEntryBean.java

private String testName = "HELLO £¢≠üå", testComments = "HELLO £¢≠üå";
public String getTestName()

public void setTestName(String tn)

public String getTestComments()

public void setTestComments(String tc)