Hello.
I've found this bug while using firefox. on the _updateScripts methods of the atlas platform, you can find this:
if
(xmlScriptNode.childNodes.length != 0) {
for (var c = xmlScriptNode.childNodes.length - 1; c >= 0; c--) {
var nodeType = xmlScriptNode.childNodes[c].nodeType;
if ((nodeType == 3) || (nodeType == 4) || (nodeType == 8)) {
text = xmlScriptNode.childNodes[c].nodeValue;
break;
}
}
}
The problem is in the bold line (it's around line 11289 of the atlas.js file). if you receive a script line this from an atlas postback:
<script type="text/javascript">
<!--
var Page_Validators = new Array(document.getElementById("ButtonLabelContainer_ctl00_ButtonTextValidator"), document.getElementById("ButtonLabelContainer_ctl01_ButtonTextValidator"));
// -->
</script>
it won't be interpreted correctly by firefox since it consideres "\n" to be a correct node. to solve this, i've modified the previous excerpt to look like this:
if
(xmlScriptNode.childNodes.length != 0) {
for (var c = xmlScriptNode.childNodes.length - 1; c >= 0; c--) {
var nodeType = xmlScriptNode.childNodes[c].nodeType;
if ((nodeType == 3) || (nodeType == 4) || (nodeType == 8)) {
text = xmlScriptNode.childNodes[c].nodeValue;
if( text == "\n")
{
continue;
}
break;
}
}
}
this garantees that when firefox gets one of those "\n" it jumps to the next instruction (which in the previous case, will be the Page_Validators array.
btw, you can see a real example where this kind of behvaior happens in this post:
http://forums.asp.net/1247033/ShowThread.aspx#1247033
thanks.