One of the DHTML (Dynamic HTML) effects not built into my own libraries is a fish-eye effect. Those of you who have a Mac will know the effect I’m talking about: when you move your mouse over a menu bar, the items expand but in a way that emulates a ‘fish-eye’ magnifier.
This isn’t a trivial effect to create. You not only have to capture the web page reader’s mouse movements, you also have to size objects in relation to each other and the mouse. Luckily, the popular Ajax library Dojo Toolkit, created by a consortium of interested developers has just such an effect.
Though Dojo has documentation, it’s coverage is like Wikipedia’s coverage: it’s added by interested parties, and as such, there are gaps in how to use some of the library objects–including the Fisheye Widget. However, there are plenty of demos, includin one for this object.
To use this functionality, according to the demo, you create a series of div elements: one outer, one that acts as control, and several inner (for each menu item). You then add Dojo class names, as well as element attributes providing information for the menu caption, source file for the icon, minimum and maximum image sizes and so on. The markup in the demo is as follows:
<div class="outerbar">
<div class="dojo-FisheyeList"
dojo:itemWidth="50" dojo:itemHeight="50"
dojo:itemMaxWidth="200" dojo:itemMaxHeight="200"
dojo:orientation="horizontal"
dojo:effectUnits="2"
dojo:itemPadding="10"
dojo:attachEdge="top"
dojo:labelEdge="bottom"
dojo:enableCrappySvgSupport="false"
>
<div class="dojo-FisheyeListItem" onClick="load_app(1);"
dojo:iconsrc="images/icon_browser.png" caption="Web Browser">
</div>
<div class="dojo-FisheyeListItem" onClick="load_app(2);"
dojo:iconsrc="images/icon_calendar.png" caption="Calendar">
</div>
<div class="dojo-FisheyeListItem" onClick="load_app(3);"
dojo:iconsrc="images/icon_email.png" caption="Email">
</div>
<div class="dojo-FisheyeListItem" onClick="load_app(4);"
dojo:iconsrc="images/icon_texteditor.png" caption="Text Editor">
</div>
<div class="dojo-FisheyeListItem" onClick="load_app(5);"
dojo:iconsrc="images/icon_update.png" caption="Software Update">
</div>
<div class="dojo-FisheyeListItem" onClick="load_app(6);"
dojo:iconsrc="images/icon_users.png" dojo:caption="Users" >
</div>
</div>
</div>
It’s an interesting approach to take: embed the necessary information as tag attributes, so that the person doesn’t have to touch code. However it has two major drawbacks: it doesn’t validate, and it’s not accessible.
Dojo, like many other of the new Ajax libraries, make use of custom attributes on standard HTML and XHTML objects, which don’t validate as either HTML or XHTML. In addition, the menu is JS driven, so a person who doesn’t have JavaScript enabled won’t have access to the menu. Not only JavaScript driven, it’s also mouse driven menu, which makes it unusuable within text-to-speech browsers.
Modifying the code so that it validates is a bit tricky, but doable. What is required is removing the tag attributes for the elements, and adding these using the DOM, or Document Object Model, API.
I have six menu items, which means removing the custom attributes for one controller and the six menu item div elements:
<div class="dojo-FisheyeList" id="controller">
<div id="menu1" class="dojo-FisheyeListItem">
</div>
...
<div id="menu6" class="dojo-FisheyeListItem">
</div>
</div>
In JavaScript, I use the DOM setAttribute to re-set these custom attributes. The following code re-sets the attributes for the controller object:
var cont = document.getElementById("controller");
cont.setAttribute("itemWidth","60");
cont.setAttribute("itemHeight","100");
cont.setAttribute("itemMaxWidth", "200");
cont.setAttribute("itemMaxHeight", "300");
cont.setAttribute("orientation","horizontal");
cont.setAttribute("effectUnits","2");
cont.setAttribute("itemPadding","10");
cont.setAttribute("attachEdige","top");
cont.setAttribute("labelEdge","bottom");
cont.setAttribute("enableCrappySvgSupport","false");
These attribute settings are exactly as they were found in the tag attributes, other than altering the image sizes to fit my own GIFs.
For each of the menu options, again the element is accessed by identifier, and attributes added with setAttribute. The following sets the attributes for the first menu item, but all other menu objects are modified using the exact same code (but different images and captions):
var menu1 = document.getElementById("menu1");
menu1.setAttribute("onClick","load_page('http://learningjavascript.info')");
menu1.setAttribute("iconsrc","/dotty/dotty.gif");
menu1.setAttribute("caption","Learning JavaScript");
Since Dojo requires these attribute settings before its functionality, and it processes the data on page load, the function that contains the attribute setting needs to be called right after the div elements are created in the page. One way is to embed a script block calling the function in the web page right after the div elements are created:
<script type="text/javascript">
//<![CDATA[
setMenuProps();
//]]>
</script>
(Notice the use of the CDATA section surrounding the script? This, also, is required in order for the page to validate as XHTML.)
Once the attributes are set, the Dojo fisheye menu loads cleanly, without having to use custom attributes. But something’s still missing: attributes that are required. Each img tag requires an alt attribute, which is a legitimate X(HTML) attribute, but one that’s not provided.
I explored the Dojo code and tried a couple of alternatives to add attributes for the images, but nothing worked. There’s also nothing in documentation. So, back again to my own custom code.
Unlike setting the initial attributes, the alt attribute needs to be added afterDojo has done its work and created the menu. Dojo captures the window.onload event, which I also needed to capture. However, I had to do so in such a way as to not ‘break’, or override Dojo’s event handler.
I needed to use the DOM again, but this time to attach an event handler on the window onload event, chaining it with the Dojo event handler. The following code does the trick:
function addWindowOnLoad(func) {
// test for object model
if (window.addEventListener) {
window.addEventListener("load",finish,false);
} else if (window.attachEvent) {
window.attachEvent("onload", finish);
}
}
addWindowOnLoad(finish);
The finish method then accesses each image in the page, checks for class name, and when it matches the Dojo class name, checks the source attribute on the image. Based on the text, the related alt tag value is set:
function finish() {
for(var i = 0; i The page now validates as XHTML transitional. Thought it takes more code, it’s preferable than just blowing off XHTML as ‘not useful’, or unimportant. We haven’t walked three steps forward in our use of web standards the last decade, only to take two steps back now for the sake of a little sizzle.
Even if we decide to blow off valid markup, we can’t justify blowing off accessibilty (of which valid markup is one component). A pretty or cool effect is not worth putting barriers around our pages. Unfortunately, though, accessibility is both easier and more difficult to implement.
The simplest accessibility option is to provide a NOSCRIPT block with content to display if JavaScript is disabled. In this case, a straight menu with hypertext links around images is all that’s needed:
<noscript>
<a href="http://scriptteaser.com/learningjavascript/"><img src="/dotty/dotty.gif"
alt="ScriptTeasing News" /></a>
...
</noscript>
Unfortunately, other accessibility issues aren’t as easy to resolve. A truly accessible menu needs to be keyboard sensitive for many text-to-speech browsers. Adding this into the Dojo menu is going to take additional thought. In addition, the behavior of the menu is off with Internet Explorer 6.x (the entire menu doesn’t display until you move your mouse over the bar).
Then there’s the issue of the size of the Dojo libraries. There’s a noticeable delay loading this page with Dojo’s large code base in addition to mine, just to create a visual effect.
Sometimes, though, you want both the sizzle and the steak. I’ll return later with more on whether I’ll keep the Dojo fisheye menu, and how I’ll resolve the last issue of accessibility if I do.