We have an applet running inside the JRE plugin for Internet Explorer that updates a web page via LiveConnect. The updates occur in response to messages from JMS. It works correctly some of the time, but IE will frequently crash, usually when the user is interacting with IE in some way while the page is updating, such as resizing the frame. We've tried versions 1.2.2, 1.3.0, and 1.3.1 in IE5.5, all have the same problem.
DESCRIPTIONEND
TESTCASEBEGIN
APPLET:
private JSObject window;

try {
window = JSObject.getWindow(this);
        } catch (Exception e) {
            e.printStackTrace();
        }
        
QuoteField f = quoteItem.get(field);
Object[] args = {tag, quoteItem.toString(field, displayFormat), new Integer(f == null ? QuoteField.TIC_DIRECTION_NORMAL : f.getDisplayTicDirection())};
synchronized(window) {
try {
window.call("onData", args);
         } catch (Exception e) {
         e.printStackTrace();
         }
         }

JavaScript:

function onData(tag, value, ticDirection)
{
switch (ticDirection) {
case 0:
document.all[tag].style.backgroundColor = 0xFFCC00;
break;
case 1:
document.all[tag].style.backgroundColor = 0x33cc00;
break;
case 2:
document.all[tag].style.backgroundColor = 0xcc0033;
break;
default:
document.all[tag].style.backgroundColor = "";
break;
}

document.all[tag].innerHTML = makeFraction(value);
}
Basically, we run JMS messages from our server into the attached
'UpdaterApplet'. It turns these messages into updates to the DHTML content
by calling the LiveConnect classes. As you will see, these will frequently
fail. It does not happen using the standard MS VM.

I have tried running our application with jre 1.3.0.01, but it doesn't work.
It appears that the LiveConnect API no longer passes data to the web page in
the same way. For example, when we pass an Integer as an argmument to
JSObject.call(), it shows up on the JavaScript side as type Object under
1.2.2. With 1.3.0.01, it shows up as type Number. According to the
LiveConnect documentation from Netscape, the way it is implemented in 1.2.2
is correct. Was this change intentional, or is it a bug?