Dmitry,
The 'event' variable in your test case is not being passed into the
listener method.
It is a global variable on the window object. I was suggesting that you
try registering event listeners
in a different way that would ensure that the event object is being
passed in as a parameter.
Another question - can you reproduce it in other browsers besides IE?
Does IE8 behave the same way as IE7?
Regards,
Max Starets
Dmitry Barsukov wrote:
Max,
It was a bit of a misleading bug description initially.
The bug should be described as: "Event object parameter of "onkeyup"
JavaScript handler is being mistakenly unset when <tr:table> contains more
than 1 range". This becomes apparent in the rectified example.
However "event" object parameter is correct when there is no ranges
navigation bar on the table, i.e. there is just one table range.
The JS handlers do get called, but the "event" parameter passed is wrong
when the table displays ranges navigation bar.
For our application this is a critical bug because we need to capture
pressed key code from the event object.
The above could be easily observed with the example jsf form I showed
earlier.
Regards,
Dmitry
On 18 June 2010 22:05, Max Starets <[email protected]> wrote:
Dmitry,
I guess you are saying that event.type produces incorrect results if
the table navigation bar just happens to be on the page. This is definitely
strange.
I am not really familiar wit the navigation bar code, but I took a quick
look
to see if we are doing any event capturing there. I did not find anything.
The 'event' javascript variable is global to the entire window. I would try
assigning a listener in javascript, so that you get the event passed in as
a parameter:
function handleKeyUp(evt)
{
alert (evt.type);
}
document.getElementById("myClientId").onkeyup = handleKeyUp;
Or (in IE):
document.getElementById("myClientId").attachEvent('onkeyup', handleKeyUp);
Regards,
Max
Dmitry Barsukov wrote:
Rectified form (surely a bug!) :
<?xml version="1.0" encoding="ISO-8859-1" ?>
<f:view xmlns:f="http://java.sun.com/jsf/core";
xmlns:tr="http://myfaces.apache.org/trinidad";
xmlns:trh="http://myfaces.apache.org/trinidad/html";>
<trh:html>
<trh:head><title>TABLE BUG TEST CASE</title></trh:head>
<trh:body>
<tr:form id="frm_main">
<tr:inputText id="it_search" onkeyup="alert('onkeyup:
event.type='+event.type);"/>
<tr:commandButton text="placebo"/>
<tr:table id="tbl_result" rows="50"
value="#{tblBugList}" var="row">
<tr:column > <tr:outputText value="#{row}"/>
</tr:column>
</tr:table>
</tr:form>
</trh:body>
</trh:html>
</f:view>
tblBugList bean:
<managed-bean>
<managed-bean-name>tblBugList</managed-bean-name>
<managed-bean-class>java.util.ArrayList</managed-bean-class>
<managed-bean-scope>session</managed-bean-scope>
<list-entries>
<value-class>java.lang.String</value-class>
<value>LINE ONE</value>
<value>LINE TWO</value>
<value>LINE THREE</value>
</list-entries>
</managed-bean>
If <tr:table> attribute "rows" is set to "2" then "event" object passed to
JavaScript onkeyup event is screwed, i.e. it must point to event.type
"onkeyup" but it shows "hidden" instead of "onkeyup".
If "rows" attribute is set to > 3, i.e. table does not display ranges bar,
then event object is correct and it points to event.type "onkeyup". There
is
a feeling that all other JS handlers/events are handled similarly.
This does sound like a bug to me. It is also quite critical to our
application.
Regards,
Dmitry
