Matthew Coertze

Hello Tarquin, first up, thanx for your importXML script, you've saved me
ages in tracking the various issues down. (Your other scripts look great
too, but I haven't had a chance to download/study them yet).

Anyhow, on to the reason for my e-mail - importXML.js, problems and possible
solutions:

1) IE6 on Xp sp2 returns gibberish when it downgrades to the iframe
solution - probably in much the same way Konquerer does - I refer to your
code at the start of createTable below, which traps the IE6 error as well:
[code]
/* Konqueror insists that the timer can keep running until there are 20
alerts (the file is still in the list of files to be checked until at least
one alert is dismissed), so I use a timer to delay the alert and allow my
script to stop checking the file */
if( xmlDoc.documentElement && xmlDoc.documentElement.tagName &&
xmlDoc.documentElement.tagName.toUpperCase() == 'HTML' ) {
setTimeout('alert(\'For no apparent reason, your browser has turned the
clean XML into HTML based garbage.\\nScript aborted.\');',50); return; }
[/code]

The fix: Upon investigating, I discovered that the iframes section of
importXML should return
window.frames['iframename'].window.document.XMLDocument for IE6, instead of
just window.frames['iframename'].window.document
Now I know this makes no sense. MSDN itself says the XMLDocument part is
optional, but there you have it. It works for IE6. Unfortunately, however,
it breaks the iframe solution for Firefox 1.5 and Netscape 8 (Opera 8.5 is
still happy). Since I don't have Konqueror, I don't know if the same fix
will work for that too.

Implementing the fix: Probably the fix should go into MWJ_checkXMLLoad,
something like

if(window.frames['MWJ_XML_loader_'+x].window.document.documentElement) {
setTimeout( MWJ_ldD[x].split('MWJ_SPLIT')[0] +
'(window.frames.MWJ_XML_loader_'+x+'.window.document.XMLDocument);',
parseInt(MWJ_ldD[x].split('MWJ_SPLIT')[1]) );
} else {
setTimeout( MWJ_ldD[x].split('MWJ_SPLIT')[0] +
'(window.frames.MWJ_XML_loader_'+x+'.window.document);',
parseInt(MWJ_ldD[x].split('MWJ_SPLIT')[1]) );
}

Your thoughts on this would be appreciated.

2) In checking the e-mails for a solution to 1) above, I saw a post about
importXML not catering properly for IE with activeX disabled. To cater for
this this I changed the code from:

[oldcode]
if( !navigator.__ice_version && window.ActiveXObject ) {

the Microsoft way - IE 5+/Win (ICE produces errors and fails to use
try-catch correctly)

try { //IE Mac has the property window.ActiveXObject but produces errors if
you try and use it

try { var tho = new ActiveXObject( 'Microsoft.XMLDOM' ); //newer

} catch(e) { var tho = new ActiveXObject( 'Msxml2.XMLHTTP' ); } //older

MWJ_ldD[MWJ_ldD.length] = tho;

MWJ_ldD[MWJ_ldD.length-1].onreadystatechange = new Function( 'if(
MWJ_ldD['+(MWJ_ldD.length-1)+'].readyState == 4 ) {
'+oFunct+'(MWJ_ldD['+(MWJ_ldD.length-1)+']); }' );

MWJ_ldD[MWJ_ldD.length-1].load(oURL);

return true;

} catch(e) {}

}

[/oldcode]

to...

[code]

if( !navigator.__ice_version && window.ActiveXObject ) {

//activeX XMLDOM/HTTP request - IE Win, Ice etc

var tho=MWJ_ActiveXDOM( 'Microsoft.XMLDOM' ); //newer

if(!tho) tho=MWJ_ActiveXDOM( 'Msxml2.XMLHTTP' ); // older

if(tho) return true;

}

[/code]

and created a new function as below:

[code]

function MWJ_ActiveXDOM(dom) {

var tho;

try {

tho = new ActiveXObject( dom );

MWJ_ldD[index] = tho;

MWJ_ldD[index].onreadystatechange = new Function( 'if(
MWJ_ldD['+index+'].readyState == 4 ) {
'+oFunct+'(MWJ_ldD['+index+'],'+index+'); }' );

MWJ_ldD[index].load(oURL);

} catch(e) { }

return tho;

}

[/code]

Which fails gracefully in IE6 with activeX disabled, and should catch the IE
mac problem too. (Although I can't test that).

Again, your thoughts would be appreciated.

3) The last point about importXML.js is that, as I'm sure you know, it is
not reentrant and can cause a memory leak. For example, say I have a xml
file called news.xml which I refresh on the server periodically, and hence
want to reload from my page every few minutes. Every call to importXML will
allocate a new array entry and hence use more memory. To get around this I
changed importXML to receive an index as input, and also now pass the
allocated index as part of the function callback. If you would like me to
post the changes I will gladly do so.

Your thoughts on this too appreciated.

Thanking you for a wonderful script

Jacob Matthew Coertze

Tarquin, thanx for the quick response.

> If ActiveX is enabled, and it still fails, then we have a problem,
> because it should not be going down the iframe branch. Is this the case?

No. I wanted to cater for the case where the user has consciously disabled activeX,
and thus I would need to use the iframe solution. The fix I posted works fine for
catching the activeX disable, and the 'if' for using XMLDocument or not (as posted) also works.
So basically, I'm all set :) I just wanted to give you the feedback and in case others also
needed the workarounds.

Thanx again

Matthew