Echo Web Framework

I did download the echogo of today builds, and it seems that there is a problem in deregistration of event listeners on the client side.

When the application is closing the (modal) login dialog I receive a JS error in firefox with this message: element is undefined

The script code (according to the JS console of FF3.03)

Core.Web.DOM.stopEventPropagation(e);
},
remove: function(element, eventType, eventTarget, capture) {
Core.Web.Event._lastId = null;
if (!element.__eventProcessorId) {
return;
}
var listenerMap = capture ? Core.Web.Event._capturingListenerMap 
: Core.Web.Event._bubblingListenerMap;
var listenerList = listenerMap.map[element.__eventProcessorId];
if (listenerList) {
listenerList.removeListener(eventType, eventTarget);
if (listenerList.isEmpty()) {
listenerMap.remove(element.__eventProcessorId);
}
if (!listenerList.hasListeners(eventType)) {
Core.Web.DOM.removeEventListener(element, eventType, Core.Web.Event._processEvent, false);
}
}
},
removeAll: function(element) {
Core.Web.Event._lastId = null;
if (!element.__eventProcessorId) {
return;
}
Core.Web.Event._removeAllImpl(element, Core.Web.Event._capturingListenerMap);
Core.Web.Event._removeAllImpl(element, Core.Web.Event._bubblingListenerMap);
},
_removeAllImpl: function(element, listenerMap) {
var listenerList = listenerMap.map[element.__eventProcessorId];
if (!listenerList) {
return;
}
var types = listenerList.getListenerTypes();
for (var i = 0; i < types.length; ++i) {
Core.Web.DOM.removeEventListener(element, types[i], Core.Web.Event._processEvent, false); 
}
listenerMap.remove(element.__eventProcessorId);
},
toString: function() {
return "Capturing: " + Core.Web.Event._capturingListenerMap + "\n"
+ "Bubbling: " + Core.Web.Event._bubblingListenerMap;
}
};

The error occurs in the line

if (!element.__eventProcessorId) {

direct after the removeAll: function

Then pressing F5 shows the new screen correctly and the application works as expected.

Just a minor one: There's a var-declaration missing in Serial.js.

Cheers, Maik

Index: src/client/echo/Serial.js
===================================================================
--- src/client/echo/Serial.js (revision 1380)
+++ src/client/echo/Serial.js (working copy)
@@ -435,7 +435,7 @@
var element = Core.Web.DOM.getChildElementByTagName(pElement, "b");
var border = {};

- value = element.getAttribute("t");
+ var value = element.getAttribute("t");
if (value) {
border.top = value;
value = element.getAttribute("r");

Index: src/server-java/webcontainer/nextapp/echo/webcontainer/resource/RemoteClient.js
===================================================================
--- src/server-java/webcontainer/nextapp/echo/webcontainer/resource/RemoteClient.js (revision 1380)
+++ src/server-java/webcontainer/nextapp/echo/webcontainer/resource/RemoteClient.js (working copy)
@@ -1027,7 +1027,7 @@
if (!translator) {
throw new Error("Translator not available for property type: " + propertyType);
}
- propertyValue = translator.toProperty(this._client, propertyElement);
+ var propertyValue = translator.toProperty(this._client, propertyElement);
if (!this._referenceMap) {
this._referenceMap = {};
}

Hello,

after installing Opera I did some tests with the current svn release and noticed some problems.
Should these also be "critical" ? (We and our clients don't use Opera)

A few problems:

- Menubar (with toplevel icons?) is not rendered at all
- Tabpane don't work
- Under some conditions in modal windowpanes the text is selected (Of labels, buttons etc.), might be Tabpane problem related

André

I'm going to hold another day (until Wednesday) on Beta2.

I do have a new Echo3 CSJS demo up at http://demo.nextapp.com/echo3csjs. This one features dynamic module loading, so it's roughly half the size it used to be for startup. Would appreciate any testing/bug reporting in it as well. It's running on the latest Echo3/Extras/CoreJS as well. The demo source is available from the download page, and/or from SVN @ https://svn.nextapp.com/svn/echo3demo

I did take some time today to JSLint the entire platform. Had to tweak the JSLint script to avoid checking for == null, as we have a legitimate use for that and it appears quite a bit.

Please feel free to continue reporting any last-minute critical bugs as well, and thanks again for all the issues mentioned and solved so far.