Naming UI components in OOCSS

Written by Harry Roberts on CSS Wizardry.

One of the biggest—if not most common—complaints about OOCSS is its use of ‘insemantic classes’. Unfortunately, the idea that classes are semantic (in the HTML sense of the term) is something of a fallacy; classes aren’t understood by machines, they’re simply read and/or matched—machines cannot glean any meaning from something whose content is entirely subjective.

If you are still on the fence about semantic classes, I would recommend reading Nicolas Gallagher’s excellent article, About HTML semantics and front-end architecture. In it, he discusses what we mean when we talk about semantics, and how HTML semantics differs from developer semantics. The short version is that we should write classes that are useful for developers; classes that are highly reusable, that don’t couple themselves to specific types of content, and classes that describe the styling’s function rather than the content’s function. Traditionally we would refer to these as insemantic classes, but Nicolas does a great job of debunking that. Having a solid grasp of (and, ideally, being in agreement with) his article will really make this one make more sense.

tl;dr Keep using agnostic, abstract, OOCSS classes in your markup, but add any desired meaning to your HTML via a data-ui-component attribute, e.g.: <ul class="ui-list" data-ui-component="users-list">.

One of the best things about OOCSS, and ‘insemantic’ classes, is that we have many design patterns tied to highly reusable names—we have very recyclable CSS that we can apply over and over again, keeping our codebase small, neat, and consistent.

One of the cited downsides of OOCSS is that these classes don’t tell you anything about the content. Nicolas explains why this isn’t that important (basically classes shouldn’t describe content when content describes itself), but when you have a series of objects and abstractions that come together to form one complex UI component, it is often advantageous to be able to refer to that component by a unique name.

I firmly believe that classes should not describe content, because it inhibits their reusability, and that there is no such thing as an insemantic class, but I don’t see any harm in having the best of both worlds. To this end, I came up with something of a solution: highly abstracted, reusable classes, along with a method of giving distinct names to particular UI components. Let’s take an example…

Imagine we have some abstractions:

• The media object which places some text next to an image.
• The bare-list object which removes the indents and bullets from a list.
• The UI-list object which takes the indents and bullets off of a list, gives the list items some padding, and places a small border between each list item.

We combine these three abstractions to have a UI-list; each list item in the UI-list contains a media object; each media object contains a picture of a user and their bio; each bio contains a bare-list of their Twitter and website URLs. Three individual UI objects which combine to create a list of users and their bio information:

<ul class="ui-list">

    <li class="ui-list__item">
        <div class="media">
            <img src="" alt="" class="media__img" />
            <div class="media__body">
                <p>...</p>
                <ul class="bare-list">
                    <li>...</li>
                </ul>
            </div>
        </div>
    </li>

    ...

</ul>

Take a look on jsFiddle

Of course, the beauty of these classes is that they could be rearranged in any order or combination to make another complex UI component that is entirely different. But, when a client wants to duplicate a piece of content, they won’t ask you to copy the UI-list and media object and bare-list component, they’ll probably ask you to duplicate the user-list. When you ask a software engineer to write a loop to populate that list, you’ll probably also want to refer to it as a user-list to them. What we need is a way of assigning useful names to these composites for when we have discussions about them; there is no harm in having meaningful names as long as they don’t impact our ability to reuse things.

Using Sass’ @extend

One method, obviously, would be to wrap all these classes up into one unique one using Sass’ @extend directive, but there are problems with @extend that both Oliver J Ash and I have already covered: chiefly, that @extend is very greedy, and can cause serious bloat if you’re not careful.

The other downside to wrapping these objects up into a more meaningful class is that you have to pop open your Sass file(s), think up a brand new name, potentially create a new partial, and add some more code to replicate functionality that already existed free of charge. You have to do this every time you want to reuse that object/abstraction anywhere new. This is just increasing the amount of CSS you output for no real, tangible gains.

data-* attributes

This desire to give UI components meaningful names is one that has been around for a long time, and one that I frequently get asked about in workshops, etc. I recently began to give it a little more thought, and it hit me: data attributes are the perfect candidate. This, from MDN:

HTML5 is designed with extensibility for data that should be in the HTML, but not visible. data-* attributes allow us to store extra information on standard, semantic HTML elements without polluting the class name.

We can now attach our meaningful names via a data-ui-component attribute, for example:

<ul class="ui-list" data-ui-component="users-list">

Take a look on jsFiddle

When this list gets repurposed as something else, we can keep using the same, agnostic, decoupled classes but we can also give it a name that describes how and where that component might get used:

<ul class="ui-list" data-ui-component="articles-list">

Reusable classes and meaningful names!

Take a look on jsFiddle

Now we can use as many nicely abstracted, agnostic classes as we like, but still neatly give useful names to complex UI composites:

<div class="media" data-ui-component="mini-bio">
    ...
</div>

<div class="media  media--large" data-ui-component="album-info">
    ...
</div>

<ul class="inline-list" data-ui-component="site-nav">
    ...
</ul>

<div class="box  box--promo" data-ui-component="promo-area">
    ...
</div>

It’s the best of both worlds: you can still keep your CSS super-lean, abstracted, recyclable, and reusable, but you can also give specific chunks of markup meaningful names which can be used in discussions (e.g. referring to the articles-list rather than any given UI-list).

Another really handy thing that this allows us to do—when in a debugging mode—is to quickly view any parts of a site that we deem components (the kinds of thing that you’d have in your pattern library or UI toolkit):

[data-ui-component] {
    outline: 5px solid yellow;
}

This will quickly put an outline around any explicitly named parts of the UI.

It’s important to note that although we can style HTML via its data-* attributes, we probably shouldn’t. data-* attributes are meant for holding data in markup, not for selecting on. This, from the HTML Living Standard (emphasis mine):

Custom data attributes are intended to store custom data private to the page or application, for which there are no more appropriate attributes or elements.

I’ve been using this on a client project recently, and it’s proved very useful in encapsulating and discussing UI components, whilst also allowing me to keep my CSS as abstracted and OO as usual. My client doesn’t need to know—or even care—that the share with friends component is actually a combination of a .box, .media, and .btn object, they just need to refer to is at the same thing everyone else does—the share with friends component!

I would strongly recommend trying it out.

By Harry Roberts

Harry Roberts is an independent consultant web performance engineer. He helps companies of all shapes and sizes find and fix site speed issues.

Did this help? We can do way more!