is using display property for responsivness bad behaviour? [duplicate] - css

TL;DR Is it a bad practice to change default display property in my CSS?
Issue
Recently, in our project we had to position 2 header tags so they would look like one. They had the same font size and similar styling so the only issue was how to place one next to another. We had 2 different ideas on that and it le do a discussion on whether or not is a good practice to change default display property
So, our very basic code
<div class="container">
<h1>Header:</h1>
<h2>my header</h2>
</div>
The outcome we would like to have:
Header: my header
Note:
The code needs to consists of 2 different headings because on mobile version we want to display them in in separate lines (so leaving default display: block).
Approach #1: Use display: inline
This is pretty stright forward. Block elements became inline so they are positioned in the same line. The disadvantage of this approach is that default display properties of both h1 and h2 were changed.
Approach #2: Use float
H1 can be positioned on the left using float: left property. This approach leaves the default display property intact, but will requires some hacks if the .container is not long enough to fit both headers in single line.
The question
It all leads to a simple question: Is it a bad practice to change the default display property of HTML elements? Is it breaking the standard and should be avoided if possible? Or is it our bread and butter and it does not really matter, as long as code is semantically correct (so headers are placed in h1, articles are placed in article etc...)

Answering your main question:
tl;dr is it a bad practice to change default display property in my CSS?
NO
WHY?
A: Because it is all about semantics
Elements, attributes, and attribute values in HTML are defined (by
this specification) to have certain meanings (semantics). For example,
the ol element represents an ordered list, and the lang attribute
represents the language of the content.
These definitions allow HTML processors, such as Web browsers or
search engines, to present and use documents and applications in a
wide variety of contexts that the author might not have considered.
So, in your case if you really need to have 2 headings semantically then you can change their styles, including the display property.
However If you don't need to have 2 headings semantically, but only for purely cosmetics/design (responsive code), then you are doing it incorrectly.
Look at this example:
<h1>Welcome to my page</h1>
<p>I like cars and lorries and have a big Jeep!</p>
<h2>Where I live</h2>
<p>I live in a small hut on a mountain!</p>
Because HTML conveys meaning, rather than presentation, the same page
can also be used by a small browser on a mobile phone, without any
change to the page. Instead of headings being in large letters as on
the desktop, for example, the browser on the mobile phone might use
the same size text for the whole the page, but with the headings in
bold.
This example has focused on headings, but the same principle applies
to all of the semantics in HTML.
** Emphasis in the quote above is mine **
P.S - Remember that headings h1–h6 must not be used to markup subheadings (or subtitles), unless they are supposed to be the heading for a new section or subsection.
With all this above in mind, here is a few (good) approaches:
If you're doing the two headings purely for design then:
add a span inside of the h1, using a media query either using mobile first approach (min-width) or the non-mobile approach (max-width).
PROs - easily manageable through CSS, changing only properties.
CONs - adding extra HTML markup, using media queries as well.
h1 {
/* demo only */
background: red;
margin:0
}
#media (max-width: 640px) {
span {
display: block
}
}
<div class="container">
<h1>Header:<span> my header</span></h1>
</div>
If you need to use the two headings semantically then:
use flexbox layout.
PROs - no need to add extra HTML markup or the use of media queries, being the most flexible currently in CSS (basically the cons from option above mentioned).
CONs - IE10 and below has partial or none support, Can I use flexbox ? (fallback for IE10 and below would be CSS TABLES)
.container {
display: flex;
flex-wrap: wrap;
align-items: center;
/*demo only*/
background: red;
}
h1,
h2 {
/*demo only*/
margin: 0;
}
h2 {
/*640px will be flex-basis value - can be changed as prefered */
flex: 0 640px;
}
<div class="container">
<h1>Header:</h1>
<h2>my header</h2>
</div>
Sources:
W3C specs - 3.2.1 Semantics
W3C specs - 4.12.1 Subheadings, subtitles, alternative titles and taglines

tl;dr is it a bad practice to change default display property in my CSS?
No. As expressed by W3C themselves; HTML conveys meaning, not presentation.
As an HTML author, it's your job to structure a page so that every section of the page carries the intended semantics as described by the documentation, so that software (browsers, screen readers, robots...) can correctly interpret your content.
As a CSS author, it's your job to alter the default styling of correct markup to present it the way you want to. This includes changing the default display properties just as much as changing the default color.
Any software can, however, decide that certain usage of CSS properties changes the way they interpret your page. For instance, a search engine could decide that text that has the same color as their parent's background should carry no weight for their ranking system.
In regards to subheadings, it's considered incorrect to markup a subheading with an <hX> element. What you should do is to decide on one <hX> element, wrap it in a <header> and wrap subheading-type text in <p>, <span> or similar.
The following is an example of proper subheadings, taken from the W3C documentation:
<header>
<h1>HTML 5.1 Nightly</h1>
<p>A vocabulary and associated APIs for HTML and XHTML</p>
<p>Editor's Draft 9 May 2013</p>
</header>
Note that there's a discrepancy between the W3C specification and the WHATWG specification where the latter uses the <hgroup> element for this specific purpose, while the former has deprecated it. I personally go with W3C's example, but most software will still understand hgroup, likely for many, many years to come, if you prefer the WHATWG approach. In fact, some argue that WHATWG should be followed over W3C when the specs differ.
In your particular example, however, I'm not sure why you chose to split the <h1> into two elements in the first place. If what you marked up as an <h1> is actually supposed to be a generic "label" for the heading, then it should probably be considered a subheading instead. If you need to split it for styling purposes, wrap the two parts of text in <span> as such:
<h1>
<span>Header:</span>
<span>my header</span>
</h1>

tl;dr is it a bad practice to change default display property in my CSS?
Its a good practice but choose carefully when to use it because it can cause some critical structure mistakes.
Why is it a good practice
The display property is open for changes. It makes HTML simple and generic. HTML elements come with a default display value that match the general behavior - what you would usually want. But they dont have to be kept and manipulated around to imitate another display property. Think about <div> for example. Obviously most of the times you want it to have display: block;, but display: flex; is much more suitable once in a while.
Lets look at a really common example of lists. <li> comes with the display property of list-item that breaks the lines for every new item.
<ul>
<li>Item 1</li>
<li>Item 2</li>
<li>Item 3</li>
</ul>
But horizontal lists are very common too. So why there is no special element for horizontal list items? Writing a special element for every common display behavior adds complexity. Instead, the convention, as also suggested by W3C is to set the <li> display property to inline.
ul li {
display:inline;
}
<ul>
<li>Item 1</li>
<li>Item 2</li>
<li>Item 3</li>
</ul>
display: inline-block; as an alternative to float
float has been used massively in page layout for many years. The problem is that it wasnt created for this task and was originally designed to wrap text around elements. A well-known float issue is that non floated elements dont recognize floated children because they are being removed from the normal flow of the document. You also cannot centrally float an element. you are limited to left or right floats only.
display is much more suitable for layout many times. display: inline-block; tells browsers to place that element inline, but to treat it as though it were a block level element. This means that we can use inline-block instead of floats to have a series of elements side by side. It is more intuitive and eliminates floats <div class="clearfix"></div> which is an additional non semantic element in your HTML.
Floats are useful when there is a need to float an element so that other page content flows around it. But there is no need to always press them into the service of a complicated layout.
Things to avoid when changing display
When you change the display property remember:
Setting the display property of an element only changes how the element is displayed, NOT what kind of element it is.
<span> test case:
In HTML early versions <span> is considered an inline-level element and <div> is block-level. Inline-level elements cannot have block-level elements inside them. Giving the <span> a display:block; doesn't change his category. It is still an inline-level element, and still cannot have <div> inside.
HTML5 introduced content models. Each HTML element has a content model: a description of the element's expected contents. An HTML element must have contents that match the requirements described in the element's content model. <span> can contain only phrasing content. It means that still you cannot nest a <div> (flow content) inside a <span>. Giving <span> a display:block; still doesn't change it.
Avoid:
span {
display:block;
}
<span>
<div>
Still Illegal!
</div>
<span>
In conclusion, changing the default display property is certainly our bread and butter. Remember that it only changes how the element is displayed, NOT what kind of element it is and use it correctly.
Now about the original two heading issue:
With respect to the comments:
Let's assume for the sake of the question, that we need to have two
headings. Or let's forget about the headings for the time being. - by the author
And also to the comment:
This question is not about resetting the display value globally. Using
selectors to target only the specific elements is implied. The
question is what we should do with these elements once selected. - by the person who set the bounty
Two headings side by side not only to handle mobile layout changes, can be done in many ways. The original example is simple and correct so its actually a good way.
h1, h2 {
display: inline;
}
<div class="container">
<h1>Header:</h1>
<h2>my header</h2>
</div>
It follows HTML rules and doesnt require any additional hacks.

Sure changing the default behaviour is redundant and even can hit performance. As a subjective solution, would recommend to use flex (but i'm not sure about performance of it, altho you can google it), it's broadly supported, and doesn't change any element css properties, it's just a layout thing, check this out
.container {
display: flex;
justify-content: flex-start;
flex-direction: column;
align-items: baseline;
}
.container.mobile {
flex-direction: row;
}
web
<div class="container">
<h1>Header:</h1>
<h2>my header</h2>
</div>
<hr />
mobile
<div class="container mobile">
<h1>Header:</h1>
<h2>my header</h2>
</div>
Notice that h1 styles stay the same

Changing default css properties is not a good idea, and should be avoided to prevent unwanted shortcomings in your markup. Instead, you should give "id" or better "class" to all html elements you want to customize and do the styling for those.
Besides, using css like "h1", "div" etc. is the slowest way as the engine try to find all those elements in the page.
In your example, it doesnt matter to use display or float as long as you give your h1 elements a css class.
Also, using correct html elements for better semantics can be useful for things such as SEO etc.

best Practice is to group the two heading in hgroup and change the display property for mobile and other views using #media query.
<hgroup class="headingContainer">
<h1>Main title</h1>
<h2>Secondary title</h2>
</hgroup>
The HTML Element (HTML Headings Group Element) represents the
heading of a section. It defines a single title that participates in
the outline of the document as the heading of the implicit or explicit
section that it belongs to.
As hgroup defines a single title for a section ,therefore changing display property within hgroup is not bed practice.

UPDATE
It seems that I might've obscured the Plunker, since Anthony Rutledge obviously failed to see (or neglected to review) it. I have provided a screen shot with a few tips on how to use the Plunker.
PLUNKER - Embed
PLUNKER - iNFO
PLUNKER - Preview
Q & A
It all leads to a simple question: Is it a bad practice to change the default display property of HTML elements?
No, not at all. Matter of fact it's a very common practice of web developers (myself included), to alter not only properties of an element, but also attributes, and it's contents to name a few.
Is it breaking the standard and should be avoided if possible?
No, but perhaps the way one goes about doing it may break the code itself which IMO is a greater concern than standards. Standards of course plays an important role but not an essential one. If that were the case, then web browsers should comply under one common set of standards (I'm talking to you IE :P). Off the top of my head, here's things that should be avoided:
Using the table element for a layout
<table>
<tbody>
<tr>
<td><img></td>
<td><input type="button"/></td>
</tr>
...
Using inline styles
<div style="display: inline-block"></div>
Using inline event handlers
<div onclick='makeASandwich();'></div>
Or is it our bread and butter and it does not really matter, as long as code is semantically correct (so headers are placed in h1, articles are placed in article etc...)
Changing an element's display property is a very small yet fundamentally essential aspect of web developing. So yes I suppose it can be considered bread and butter, which would make semantics the parsley that's used as garnish and never eaten. Semantics is subjective, a way of thinking, it is not a standard. I believe a novice should be aware of it's importance (or at least how it's important to others), but should not be pontificating between an <article> and a <section> being semantically better than using a <main> and an <aside>. In due time, semantics will just feel right.
Approach #1: Use display: inline
I have never found a good reason to use display: inline because display: inline-block is a far better choice.
Approach #2: Use float
Floats are fragile antiques. Just like handling Grandma's bone china dinner plates, you must take certain precautions if you plan on using them. Be mindful of how to clear floats and don't throw them in the dishwasher.
Basically, if given only these 2 options, Approach #1 is a better choice, especially if using inline-block. I'd stay away from floats, they are counter-intuitive and break easily. I recall only using them once because a client wanted text wrapping around an image.
CSS & CSS/JS
Provided is a Snippet comprising of 3 demos:
Pure CSS solution utilizing display: flex.
Pure CSS solution utilizing display: table-row/table-cell.
CSS and minimal JavaScript solution utilizing display: inline-block and the classList API
Each of these demos are identical on the surface:
HTML
<section id="demo1" class="area">
<!--==Pure CSS Demo #1==-->
<!--======Flexbox=======-->
<header class="titles">
<h1>Demo 1 - </h1>
<h2>display: flex</h2>
</header>
</section>
This is the original markup with the following changes:
div.container is now header.titles
h1 text is: "Demo #n"
h2 text is: "prop:value"
section#demo#n.area is wrapped around everything.
This is a good example of semantics: Everything has meaning
You'll notice at the bottom of the viewport, are buttons. Each button corresponds to a demo.
Details on how each demo works as well as pros and cons are in the following files located in the leftside menu of the Plunker (see screenshot):
demo1.md flexbox
demo2.md disply: table
demo3.md classList
PLUNKER
These notes are not for the purpose of informing the OP of anything relevant to the question. Rather they are observations that I would like to address later on.
Further Notes
Demo 1 and demo 2 are powered by the pseudo-class :target. Clicking either one of them will trigger the click event It resembles an event because it's invoked by a click, but there's no way of controlling, or knowing the capture or bubbling phase if it actually exists. Upon further clicking of the first and second button, it will exhibit odd behavior such as: toggling of the other button then eventually becoming non-functional. I suspect the shortcomings of :target is that CSS handles events in a completely different way with little or no interaction with the user.

You should use:
$('element').css('display','');
That will set display to whatever is the default for element according to the current CSS cascade.
For example:
<span></span>
$('span').css('display','none');
$('span').css('display','');
will result in a span with display: inline.
But:
span { display: block }
<span></span>
$('span').css('display','none');
$('span').css('display','');

You can use flex box to arrange elements also, like this
<div class="container" style="display: flex;">
<h1>Header:</h1>
<h2>my header</h2>
</div>
Try to read this tutorial about flex, it is really great and easy to use
https://css-tricks.com/snippets/css/a-guide-to-flexbox/

Related

Bootstrap "class=text-center" vs. CSS "text-align: center"

We may possibly stop using the Bootstrap framework, so I want to know if replacing class="text-center" with vanilla CSS will change anything. Is there any difference in behavior between
<!--Bootstrap files already included-->
<header>
<p class="text-center">Some text</p>
</header>
and
<header>
<p>Some text</p>
</header>
...
/*Separate style.css file*/
header p {
text-align: center;
}
?
The OP's question seems to have been edited after it was first posted. It essentially asks: whether Bootstrap does anything extra to elements with the class .text-center, in addition to text-align: center;. I also had this doubt and the answer is no. I think this question can be better answered by just looking at Bootstrap's source code:
.text-center {
text-align: center !important; }
Apparently all that Bootstrap does is to apply one CSS rule: text-align: center !important; to such an element. So yeah they can be considered equivalent.
Sure, just keep in mind that the p selector will apply that style to every <p> tag in your page. If this is a style you definitely want every <p> to have, then you can use it. If you only want some of them to have this style, then stick with the class. Sidenote, you can use the following syntax to target only elements that are <p class="text-center"> without affecting other tags with the same class.
p.text-center {
text-align:center
}
NOTE: To clarify something, this is basic CCS functionality, completely independent of bootstrap. You could do this whether you are using it or not.
As mentioned in other answers, there is no added value from a behavioral aspect of your code or functionality so I will not tackle that aspect of the question, however, I believe that this is a matter of making your code easily understood by other developers.
Simply finding a 'text-center' class justifies behavior that's happening to the tag element without the need to dig into the CSS class to understand it, and making the access to modify that class more convenient for other developers on task (developer already having the HTML page open on his text editor VS using devTools to find out the relevant class and/or CSS file causing the undesired effect, navigating onto that file and applying the necessary functions), and I believe it makes for a better semantic code.

HTML5 Semantic tag alternatives

HTML5 has some useful semantic tags for use. They're not actually necessary, but the point of 'Semantic Web' can be helpful for organizing a lot of content.
I just came across a simple <small> tag which I never saw before. It works as it sounds, the text gets smaller.
What if the !DOCTYPE html was not for HTML5 but other legacy doctypes, what are good alternatives besides simply doing following style/CSS adjustments?
CSS: p {font-size: smaller}
HTML attribute: <p style="font-size: ##%">This text.</p>
What if the !DOCTYPE html was not for HTML5 but other legacy doctypes
<small> was introduced in HTML 3.2, not in HTML 5, so you should still use the <small> element.
The semantic meaning that <small> holds in HTML 5 wasn't there in earlier versions of HTML, but there was nothing with equivalent semantics that would be better.
An alternative would be to use a class name such as .small-text to represent small text.
.small-text {
font-size: smaller;
}
However, you should also ask yourself whether defining a class for small text is necessary. For example, you might actually want this instead:
.subtitle {
font-size: smaller;
/* other styles */
}
and avoid creating too many useless class names.
Thanks for the answers people...
I believe the <span> tag is an excellent alternative for the similar results. The difference would be that the tag comes with a preset size. <span> brings a different value being unchanged until inheriting via css/attr or js.
You won't necessarily have to worry about adding a new class, you can have it inherit a style by relative placement.
ie.
<article><p>Text here to continue for who knows how long.<small>This is small text.</small></p></article>
<article><p>Text here to continue for who knows how long.<span>This is small text.</span></p></article>
The advantage is <small> </small> will already be smaller, but who knows at what default size?...
The <span></span> will just be there awaiting definition to it...
CSS:
article > span { font-size: ###;} //no class necessary unless you really want varying span sizes in the content paragraph.

Is it right to add style in css to semantic elements of HTML 5?

Is it right, to add styles to html 5 semantics (nav, header, footer, etc...) like we add them to divs?
To use them instead of regular divs?
One time I heard frome someone I respect, that we should not add any style to html 5 semantic elements - just only if it is really necessary add a bit, but no many styles to this elements. Is he right?
for example
<nav>
<ul>
................
<ul>
</nav>
nav {
background-color: .....
width: ....
height: ......
margin: .....
color: ......
padding: ......
}
instead of
<nav>
<div id="nav">
<ul>
................
<ul>
</div>
</nav>
#nav {
background-color: .....
width: ....
height: ......
margin: .....
color: ......
padding: ......
}
How we should do it right?
What is the proper way of coding this?
What is perfect way of handling it?
The idea of separating the structure/content (you html code) from your style gives you exactly this ability, and actually drives you in this direction.
The structure gives multiple devices the ability to better understand your content in order to give your users a better experience. For example - if you will just use a ul > li structure for your menu, some devices that are not regular browsers will not be able to fully understand that this is the menu for your website, while using nav > ul > li gives them exactly that.
The style only tells the device how it should display them - and you should use it in order to give your users better experience.
The semantic elements helps us separate the structure of the page, for example:
Well, HTML5 semantic (nav, header, footer, etc...) were created to help us give meaningful and self-descriptive names to sections of our web pages. They are expected to be unique.
So, before the introduction of these semantics, sections were done this way:
<div id="main-section">
your content
</div>
<div id="sidebar">
your sidebar content
</div>
HTML5 semantics were supposed to save us from situations like the above while also being descriptive.
Styles were added to sections like the above then, and I don't see why we can't do the same now. There isn't really any rule against adding styles to HTML5 semantics but ensure that HTML5 semantics are used for unique elements in the first place.
I agree. Avoid adding styles to semantic elements. This is because these elements add nothing new when we are speaking about how things 'look'.
Remember that you want to avoid redundancy. If you need a block element, you are thinking about the 'look', and so use a div and style that. Then immediately inside that div, add your semantic element if you wish. Leave the semantic elements as hidden as possible from your css, and even from your selectors, whether in the css, or the javascript. However, adding a custom class that marks these semantic elements is good for clarity. And so, instead of:
div > * h1,
you would write like this:
div > .semanticElement h1.
This should be better for clarity, while avoiding referencing the semantic element.
If the reason for some of the above is not clear, think of it this way. The semantic changes for one element, and now suddenly I also have to make a change in the css and javascript, which is absurd.

How to do CSS text replacement using generated :before or :after without absolute positioning?

I'm attempting to allow our CMS editors the ability to swap out the text used for a page title using only a css override.
<header data-alternate="An Alternate Title">
This Page's Default Title
</header>
Using the :before or :after tag, one could use one of many available alternate titles.
header:before {
content: attr(data-alternate);
display: inline-block;
}
If only we could also say,
header:text {
display: none;
}
Unfortunately, as far as I can tell, there is no good way to hide "This Page's Default Title" in order to replace it with "An Alternate Title". If this were a Sprite, we could use one of the well-worn image replacement techniques like Phark or otherwise. Not so much with text replacement generated by :before, because the :before is also affected by the CSS devices used to hide the default text so that, with Phark, for example, the :before content is also at -9999px.
There are solutions I'm trying to avoid.
Using the Phark method or somesuch to hide the default text and then using absolute positioning on the :before content to put it back at left: 0, top: 0. I want/need to preserve flow if possible.
Wrapping the "Page's Default Title" in a span and just setting it to display: none in the CSS when an alternate title is being used.
i.e.
<header data-alternate="An Alternate Title">
<span class="default">This Page's Default Title</span>
</header>
This works, but a span nested in a header is displeasing.
Is there a way to target a tag's text without also targeting its generated :before/:after content? Is there another way to do this?
I'm not sure if this is exactly what you want, but you could try something like this:
p {
visibility: hidden;
}
p:before {
content: attr(data-alternate);
display: inline-block;
visibility: visible;
}
http://jsfiddle.net/yJKEZ/
You can set the visibility of the p element to be hidden, and then set the visibility of the :before pseudo-element to be visible within it's parent (the p) despite it's setting.
If that doesn't quite work as expected, there isn't really anything tremendously wrong with adding an extra span in, to help the process. It might not be as clean, but it could work better.
I do, however, want to raise the question of why you might need to do this, and point out some concerns with an approach like this...
For starters, pseudo elements are not part of the DOM, so that alternate text can't be selected, and isn't as accessible to the browser (or the user). Screen readers or search engines will see the default text, and not pay any attention to the alternate text, but that's what your user will see... This could lead to some confusion.
While your question specifies that you want to be able to do this with CSS, and while it may be possible, it really isn't the best solution for doing something like this. Especially if your website is being viewed in an older browser which does not support pseudo elements (Now the user sees nothing at all!).
I would more recommend something like this for swapping an image out for alt text in a print stylesheet, or swapping a hyperlink's text for the full address that it links too (again, mainly for a print stylesheet). Changing important content like a heading in this fashion can cause a lot of other issues, especially in terms of accessibility.
Just something for you to consider along with my answer... I hope I've helped you with your problem!

Have you ever set a class for your css that uses it multiple times?

I always was told to take out multiple properties in your css that you use more then once, and add them all in one rule. Like below. (please excuse the poor example)
I always seen this:
.button, .list, .items { color: #444; }
With multiple rules, can't that leave a lot of clutter?
Only in css tutorials and examples Ive seen this:
.someColor { color: #444; }
And in the css, just add another class of '.sameColor'. (div class="button someColor")
I've never seen this and feels like it would leave less clutter in your CSS. Would this be okay? Or do you think it could leave with more clutter in your HTML ?
Try to name your classes independently of their visual effect. It is a nature of CSS to play with the design and layout without having to change the HTML. Class names such as .someColor or .left-sidebar are a bad practice. Colors and position can change.
And also apply rules to semantic HTML elements rather than adding classes on all different divs and spans. It should be obvious, although many people get this wrong.
CSS is a limited set of rules and that makes it a perfect creativity stimulator.
It's all based on personal preference. I've tried both methods and prefer the second method you listed, except with more generic class names such as middleParagraph or headerGraphic so it applies to an area rather than a specific color because colors can change.
Good classnames and IDs are the first place you should optimize. THEN move onto multiple class names.
Multiple classnames can help out quite a bit though, consider:
<div class="leftColumn">Left</div>
<div class="rightColumn">Right</div>
<div class="middleColumn hasLeft hasRight">I have padding-left of 210px and padding-right of 210px</div>
<!-- alternatively, you could have -->
<div class="rightColumn">Right</div>
<div class="middleColumn hasRignt">I have padding right of 210px</div>
<!-- or -->
<div class="leftColumn">Left</div>
<div class="middleColumn hasLeft">I have padding left of 210px</div>
<!-- or -->
<div class="middleColumn">I have no padding</div>
and your css
.leftColumn { width:200px; float:left; }
.rightColumn { width:200px; float:right; }
.middleColumn.hasLeft { padding-left:210px; }
.middleColumn.hasRight { padding-right:210px; }
The result is floated right/left columns and the center area compensates for them with padding. This means you can style your middleColumn how you want to (e.g. .middleColumn .otherCoolSelector ).
It's perfectly acceptable to apply multiple classes to HTML elements. The trick is to be judicious; I usually find that when I do this, the additional classes are additions or exceptions to the basic styling being applied. For example, here are some classes I occasionally add to an element that already has a class:
error -- to style the current element if the user entered invalid data
first -- to style the first element in a list or in a table row, e.g. to suppress padding-left
last -- to style the final element in a list or in a table row, e.g. to suppress margin-right
even -- to apply zebra-striping to alternate elements
hidden -- to hide an element if it's not currently relevant
These extra classes are typically generated dynamically with a server-side language like ASP.NET or PHP. They can also be added or removed on the client side with JavaScript, esp. with a library like jQuery. This is especially useful to show or hide elements in response to an event.
There are a lot of good answers here. The trick is finding out which one fits your situation best.
One thing to consider is your markup size. In a high-traffic situation, your markup size is critical to the speed of your page loads...every byte counts. If this is the case for you, then you may want to create more CSS classes and put less in your markup. That way, the client is caching more and your website is serving up less.
What you're suggesting is a bit like an in-line style, e.g. style="color:#444". So if you want to change the color of your element you'd have to make a change to the html, which means you've defined style as part of your content. Which is exactly what css is supposed to avoid.
Imagine if you'd included 'someColor,' multiple times across multiple html files and you decide some of these elements shouldn't have 'someColor,' after all, you've got a lot of files to go through.
I'd probably avoid the list option too, if I'm making a component, say a button, I want to find .mybutton class in my css file and see all the rules for that component, without having to go through all sorts of unhelpful global classes. Also if someone comes along and changes the color in our global class he may break my button, where as if the button controlled it's own styles it can't be broken in this way.

Resources