Web tech

Microformats and what’s wrong with them

I just came from a morning-long presentation on microformats, by André Luís and I must confess I was not very impressed at all. Not by André, who’s a great guy who obviously knows his stuff and managed to make a clear presentation and still fend off some hard questions.

I was less-than-impressed by the microformats idea itself.

For a while, I couldn’t quite put my finger on what was really wrong with microformats and then it hit me.

Microformats reminds me of the mid-90s when people started using tables to build layouts in HTML. We looked at boring block-based webpages and felt frustrated that we could not get the simplest two column layout out of HTML. And then, we noticed tables and how you could hide the borders and place stuff inside the cells thus creating all sorts of different page layouts.

That’s what we did for years on the web until, of course, CSS-based layouts became the norm. As it turned out, using tables to build layouts was just plain wrong, but it was the only way webdesigners and developers found, at the time, to do it.

It was a hack.

Years later, we’ll still fighting for people to leave that behind and build their content and presentation separately and to stop using tables for anything other than displaying tabular data.

Microformats seem to be doing exactly the same thing. Using pre-existing HTML elements and trying to give them some sort of other meaning in order to create semantic value for the content. It all seems very slapdash and confusing.

Now that everyone’s used to use classes to declare CSS properties, you can have certain kinds of special class names that mean something in a microformat context. But the values aren’t reserved, so you’re never quite sure.

You’re also never quite sure of what might happen later on, when someone decides to change property values or names around. I think that maybe, some time from now, people will be going around telling us not to use classes and links to try to define relationships and meaning in content.

On the other hand, I think there’s a huge barrier in what comes to user generated content. People just do not have the patience to tag their content silly with meaningful semantic markers. You have to build them specific, unobtrusive, backend tools to markup their content for them.

Like WYSIWYG editors write your HTML for you, you’d need something that would write microformats for the users, invisibly. But that’s almost impossible.

If you have the intelligence to ‘read’ what the user’s writing and automagically tag stuff, then you have that intelligente to build a crawler that can interpret human-written text and have no need for microformats in the first place.

While I can clearly see the need for something like microformats to exist to face that lack of intelligence, I don’t think the current implementation is simple or practical for anything other than coders playing code-masturbation.

Users, common, everyday users, don’t care. And in the end, that’s who we’re always working for.


Horizontal CSS menu using a pointer and background images

OK, so the title is a bit too descriptive but I’m trying to make this post findable in case anyone out there needs to solve a similar problem. So this will be about… that’s right: creating a horizontal menu, using CSS and XHTML. Each menu item will have a small pointer underneath (an arrow if you will), and I want it all done using images, just to make it a little harder.

Here’s what I wanted the menu to look like:


As you can see, some menu items are longer, some shorter. The menu sits in a gray bar and is separated from the remainder of the page by a green “tube”. I want the selected item and the hover state to be represented by a dark box with a pointer that breaks through the green tube.

At first I made the item an li which had the dark box as a background image and was immediately followed by an empty em tag which then had the pointer as a background image and was offset using relative positioning at 50% left.

A menu item looked a bit like this:

<li><a href="whatever">Homepage</a><em></em></li>

The problem was the positioning of the whole thing as in some browsers, the pointer would become detached from the dark box, and I couldn’t fix that with negative top margins because then the pointer would be offset in yet another browser. A mess. Plus, with this method, I needed a minimum width for each item, so I could then calculate the mid point to place the pointer. No good.

But there is quite a practical solution that came to me while thinking about the sliding doors method of making tabbed navigation menus: marvel then, at the ability to stack background images.

I couldn’t have used a single background image, because then, all my items would have to be the same width, since stretching the background image would distort the triangular pointer, so enter the bkg stack!

I took the li elements of an ul and styled them so they had the correct height: dark box plus triangular pointer. Then, I added a span inside each li, so I could style two overlapping elements with two different images. Then, in the CSS, I applied a background image to the li and another (the pointer) to the span. Keeping the stretching and centering is dead easy like this: you make the dark box around the text repeat on the x-axis and you turn off stretching on the pointer. Plus, you can easily center the pointer by just using “center” on the background declaration.

Here’s the HTML:

<div id="menuarea">
  <ul class="menu">
    <li class="current"><a href="#"><span>Homepage</span></a></li>
    <li><a href="#"><span>Funcionalidades</span></a></li>
    <li><a href="#"><span>Contactos</span></a></li>
    <li><a href="#"><span>SMS</span></a></li>
    <li><a href="#"><span>Chamadas</span></a></li>
    <li><a href="#"><span>Webmessenger</span></a></li>
    <li><a href="#"><span>Comunidade</span></a></li>
    <li><a href="#"><span>Passatempos</span></a></li>

And the CSS:

#menuarea {
  background:url('images/menuarea_background.gif') repeat-x;

.menu {
  margin:0 auto;
  padding:6px 0 0 0;

.menu li {

.menu a {

.menu a span {
  padding:9px 8px 8px 8px;

* html .menu a span {
} /*IE Fix*/

.menu a em {

.menu a:hover {
  background:url('images/menuhover_background.gif') repeat-x;

.menu a:hover span {
  background:url('images/menu_pointer.gif') top center no-repeat;

.menu a:hover {
  background-position:0 0;
} /*IE fix*/

.menu .current a {
  background:url('images/menuhover_background.gif') repeat-x;

.menu .current span {
  background:url('images/menu_pointer.gif') top center no-repeat;

This is what the images look like:

menu area background The menu area background, which repeats along the x axis.

Menu item background The “dark box” that envelops each “current” or “hover” menu item

Menu pointer And the pointer.

As you can see, the dark box is as tall as a menu item plus the pointer and the pointer is the same height. They’re both transparent where they overlap and the pointer includes a bit of the green tube.

Overlapping background is a really powerful method of styling elements to make them look more complex than they actually are. It’s not *very* easy and you’ll notice the need for some IE 6 fixes and some weird height definitions that seem to make no sense. I know, I still can’t really explain some of it. But the truth is that the menu looks perfect in IE6, IE7 and Firefox for Windows as well as in Safari and Firefox for the Mac.

Toys and Software


Keeping up with my reading suggestions, rather than artcticles, here’s something interesting I found on “A list apart”, about the future of HTML.