CSS hacks & browser detection

More and more web developers are ditching tables and coming round to the idea of using CSS to control the layout of their site. With the many benefits of using CSS, such as quicker download time, improved accessibility and easier site management, why not?

The problem with CSS

Historically the main problem with using CSS has been lack of browser support. This is no longer the case as version 5 browsers, which all have good support for CSS, now account for over 99% of browsers in use.

Instead, the problem is that browsers can sometimes interpret CSS commands in different ways, causing developers to throw their arms up in the air and switch back to pixel-perfect table layouts. Fear not though, as you learn more about CSS you'll gradually start to understand the different browser interpretations and realise that there aren't really that many.

How browser detection using CSS hacks works

The way browser detection using CSS hacks works is to send one CSS rule to the browser(s) you're trying to trick, and then send a second CSS rule to the other browsers, overriding this first command. If you have two CSS rules with identical selectors then the second CSS rule will always take precedence.

Say for example you wanted the space between your header area and the content to have a gap of 25px in Internet Explorer, or IE as it's affectionately known. This gap looks good in IE but in Firefox, Opera and Safari the gap is huge and a 10px gap looks far better. To achieve this perfect look in all these browsers you would need the following two CSS rules:

#header {margin-bottom: 25px;}
#header {margin-bottom: 10px;}

The first command is intended for IE, the second for all other browsers. How does this work? Well, it won't at the moment because all browsers can understand both CSS rules so will use the second CSS rule because it comes after the first one.

By inserting a CSS hack we can perform our browser detection by hiding the second CSS rule from IE. This means that IE won't even know it exists and will therefore use the first CSS rule. How do we do this? Read on and find out!

Browser detection for Internet Explorer

To send different CSS rules to IE, we can use the child selector command which IE can't understand. The child selector command involves two elements, one of which is the child of the other. So, html>body refers to the child, body, contained within the parent, html.

Using the example of the header margin, our CSS command would be:

#header {margin-bottom: 3em;}
html>body #header {margin-bottom: 1em;}

IE can't understand the second CSS rule due to the html>body CSS command so will ignore it and use the first rule. All other browsers will use the second rule.

Browser detection for Internet Explorer 5

It may seem strange at first to send different CSS rules to different versions of a browser, but in the case of IE5 it's very necessary. This is due to IE5's misinterpretation of the box model. When specifying the width of an element in CSS, padding and borders aren't included in this value. IE5 however, incoporates these values into the width value causing element widths to become smaller in this browser.

The following CSS rule would result in a width of 10em for all browsers, except IE5 which would give it a width of just 5em (IE5 would incorporate two sets of padding and border, on both the left and right, when calculating the width):

#header {padding: 2em; border: 0.5em; width: 10em;}

The solution to this problem? Perform browser detection and send a different CSS rule to IE5:

#header {padding: 2em; border: 0.5em; width: 15em; width/**/:/**/ 10em;}

IE5 will use the first width value of 15em, 5em of which will be taken up by the two sets of padding and border (one each for the left and for the right). This would ultimately give the element a width of 10em in IE5.

The 15em value will then be overridden by the second width value of 10em by all browsers except IE5, which for some reason can't understand CSS commands with empty comment tags either side of the colons. It doesn't look pretty but it does work!

Quite simply, IE on the Mac does strange things with CSS. The browser's become somewhat obsolete as Microsoft aren't going to be bringing out an updated version. As such, many web developers code their CSS-driven sites so that the site works in IE/Mac, although it may not offer the same level of advanced functionality or design. Provided IE/Mac users can access all areas of the site this is seen as a suitable way of doing things.

CSS rules as you like:

/* Hide from IE-Mac \*/
#header {margin-bottom: 3em;}
#footer {margin-top: 1.5em;}
/* End hide */

IE/Mac will simply ignore all these commands. This CSS hack can actually be quite useful if there's a certain area of the site not working properly in IE/Mac. If that section isn't fundamental to being able to use the site, you can simply hide it from IE/Mac like so:

#noiemac {display: none}

/* Hide from IE-Mac \*/
#noiemac {display: block;}
/* End hide */

The first CSS rule hides the entire section assigned the noiemac id (i.e. <div id="noiemac">). The second CSS rule, which IE/Mac can't see, displays this section.

Browser detection for IE 4 and Netscape 4

Version 4 browsers have limited and somewhat erratic support for CSS. Making a CSS layout in these browsers, whose market share has now slipped well below 1%, can be extremely challenging. It's become common practice nowadays to completely hide the CSS file from version 4 and earlier browsers. This can be achieved using the @import directive to call up the CSS document:

<style type="text/css">@import "styles.css";</style>

Version 4 (and earlier) browsers will display a non-styled version of the site as they can't understand this @import directive.