How to add encapsulated event listeners in Mac/IE5?

[brothercake]

(With apologies for the deja-vous )

Mac/IE5 doesn't support the addition of event handlers using addEventListener or attachEvent, so is there any way of binding an encapsulated load handler, anything other than window.onload?

Here's what I've currently got. The problem with it is that the final group of browsers are only sensitive to exisiting onload functions if they come before this one:

//identify konqueror and opera 7
var kde = (navigator.vendor == 'KDE');
var op7 = /opera[\/ ]7/i.test(navigator.userAgent);

//.. gecko, safari and generic
//need to exclude konqueror specifically
//because it returns a function for window.addEventListener
//but window.addEventListener('load' doesn't work
if(!kde && typeof window.addEventListener != "undefined")
{
	//setup window listener to call initialising function
	window.addEventListener('load',onloadFunction,false);
}

//.. opera 7
else if(op7)
{
	//setup document listener to call initialising function
	document.addEventListener('load',onloadFunction,false);
}

//.. windows IE
else if(typeof window.attachEvent != "undefined")
{
	//setup window handler to call initialising function
	window.attachEvent('onload',onloadFunction);
}

//.. mac/ie, osx/msn and konqueror
//or anything else that gets this far
else
{
	//if there's an existing onload function
	if(typeof window.onload == "function")
	{
		//store it
		var currentOnload = onload;
		
		//setup handler
		window.onload = function()
		{
			//call it
			currentOnload();
			
			//call initialising function
			onloadFunction();
		};
	}
	else
	{
		//setup handler to call initialising function
		window.onload = onloadFunction;
	}
}



//onload function
function onloadFunction()
{
	alert("onload");
}

[liorean]

Of course, you already know about my registerEvent experiment, but it does something of the kind.

[brothercake]

Yeah I remember - but aren't you getting an existing load event in fundamentally the same way, such that if it came *after* your script it wouldn't be registered?

[liorean]

Welll, yes. I can't affect that. However, my registerEvent function uses addEventListener in cases it can be used in browsers you throw to the DOM0 way, and it adds both DOM0 binding of this and the most commonly used [DOM2Events] Event object properties.

It's still not entirely safe, however. I have not yet fixed bugs in it that affect at least saf and iem, though I think the op7.5 bugs are fixed at least.

I'm wondering about whether extending this to a more complete API, including among other things an event remover would be a good idea...