[thelist] Explorer like tree?

Marcus Andersson marcus at bristav.se
Wed May 26 17:35:05 CDT 2004


Simon Perry wrote:

> Surely you can recycle those images, you can even take things one step 
> further and use one image with all you icons on it. This approach was 
> started of by pixy's Fast rollovers[0] 

How does that help to reuse images in IE (which is the primary and perhaps only target browser)? I would 
really like to know.

I've done some testing on image reuse with IE6 and I can't preload (well, I can preload but when I try to use 
it it goes to the server and reloads anyway) and I can't clone images with DOM because it goes down to the 
server to get the image again (checked with web server logs). It's pretty funky actually... not! If someone 
can show a way to get around this I would be more than happy. The same goes for when you have 
background-images and change class (which might even have the same background-image), down to the server again 
and now you probably will have some cool flickering as well (the fast rollovers might help with this but not 
in all cases I think). MS seems to think that caching is a *really bad* idea!!!


> and expanded on in an article on 
> alistapart CSS Sprites[1]. The javascript to control the menu should 
> also be quite straight forward if you use the DOM[2] to transverse your 
> nested lists adding the events and behaviors you require. I would create 
> a static CSS styled tree first 

This won't be used on "normal" web sites but in a "rich client" where there is extensive scripting going on so 
it's not really an issue here. The tree, the html of it, will probably be rendered with XSLT on the client 
(ideally that is, the current tree is rendered this way, it's not impossible to change my mind though) since I 
want to do filtering on the XML data that is visualized, and then I just listen for every mouse event in the 
tree from a containing element and decides what to do with the help of the srcElement in the event object (or 
target if Moz).

Btw, since you talked about building the tree with DOM, did you know that it's much faster to create a large 
tree as a string of HTML and insert it with innerHTML in IE than to use the DOM? I did some tests and the 
differences are quite remarkable and I tested to both build and insert small chunks of nodes as you walked 
along and to build larger chunks and insert (some "guru" on MSDN claims that it's faster to insert small 
chunks of nodes often than to insert large chunks few times). It was a while ago but I think it was a 
difference of factor 5 or something like that. Using DOM in IE is slower than in Moz but using innerHTML in IE 
is a lot faster. Strange. But everything else is better in Moz (almost anyway, I wish SVG were standard (I use 
VML in IE and quite like it) and that the XSLT engine was a little better).

/Marcus


More information about the thelist mailing list