I’m curious. When designing sites using css layout is there value in setting a bar for browsers you’ll bend over backwards for? Is there ever a point where it is okay to say, “I’ll design for standards complaint browsers, and I’ll even degrade gracefully down to a point…but past that point, I’ll strip the css and you will just get the unstyled page.”
Now, while I’ve got a really good idea what the philosophical answer would be, I’m much more interested in your real world answer.
If I said the following browsers get the CSS (and browsers not listed do not get the CSS), would you think I was evil?
- Firefox (all versions)
- Internet Explorer on the Mac (version 5.22 and higher)
- Internet Explorer on the PC (version 5.5 and higher)
- Konqueror (all versions)
- Mozilla (all versions)
- Netscape (version 7.1 and higher)
- Opera (version 8.0 and higher)
- Safari (all versions)
What if I added the fact that according to our Urchin browser stats, that only 1.3% of my viewers would see the page without styles.
Okay, now….hit me with your best shot. I asked ’cause I really want to know what you think!
I was thinking about this the other day, and came to the conclusion that it’s legit not to worry about aesthetics in browsers so old that they’re unsupported by the manufacturer (eg. IE5/ Mac, Netscape 4, beta versions of Firefox etc). I don’t know which versions of IE/ Win are officially end-of-lifed by Microsoft.
The great thing about designing with standards is you can be reasonably sure that the sites will be *usable* even with no styles.
Plus, from what I’ve heard on the grapevine, Mac are discontinuing use of IE 5.2 ….. So theres one less thing to worry about I guess?
I tend to work on per-site stats to make an exact decision, but I’m personally inclined towards being hard on old browsers: Especially those which have recieved a lot of recent standards-related upgrades. I’ve also got much harder on Opera since it went free. To be honest, your list is possibly more more accomodating than mine would be.
Generally, I aim for full rendering in:
* IE6/PC
* Safari
* Firefox 1.0+
* Opera 8+
* Current Konquorer
Then I’ll accept substandard rendering in IE5.*, Opera 7, Netscape 7 and anything older will get unstyled content.
In the case of personal sites, IE6/PC will slip into an inbetween catagory of rendering, where I’m prepared to accept some small degredation in exchange for a super-cool bit of CSS in O/Fx/Saf.
I so appeciate your thoughts on this, guys! I had read a recent article on Molly’s blog Javascript Accident at the AJAX Inn that had me sensitive to browser sniffin’ logic.
I should add, that we have an “explaination” page on our site that encourages you to email us if you think we need to include your browser in the CSS friendly list.
Why bother to browser sniff and strip the style? Older browsers will ignore style rules they don’t understand – causing quirks, but I’m not sure that’s any worse than no styles. Are you maintaining different CSS for different browsers?
The older browsers really do choke on the positioning to the point that the page becomes non-functional. So, no style is a significantly better experience than styles.
As for different css for different browsers, nope, we are pure. Well…we were pure until this afternoon. But we are 99.9% pure, save a special additional style (tucked into an opera only stylesheet). So…all browsers that get styles, are getting the exact same stylesheets (with the tiny exception of Opera who gets 1 more itty bitty style sheet to keep her happy).
Specific example of broken functionality: Netscape 7.02 chokes on the pulldown menu on the UT Home Page Search box. You cannot make a selection from this search box if we leave the CSS on. In addition, the “go” button for the search vanishes.
Rather than hack for Netscape 7.02, we made the decision to just strip the CSS. Leaving them with full functionality.
testing 1 2 3
There are some users who are not surprised by styling bugs. I would expect Mac IE users, for instance, to be used to a little weirdness. Every Mac IE version since 4.5 has been totally bizarre in some way, and so few sites tailor to that browser. I might have said the same thing about Netscape 4, until I got a call from a woman who was complaining that none of the links worked on our site. I detemined during the course of the conversation that she was using Netscape 4.x, and that the drop-down menus (lists whose styles are hidden to NS 4) at the top of the page were pushing the actual content below the fold. So that when you followed a link, you got what looked like the same exact page every time. (An aside – the skip to main content link proved the solution to this problem. But eventually I should probably push the global navigation code to the bottom and position it where it’s supposed to appear visually.)
Older browser users might be a tiny minority, but they’re the users who won’t understand what’s happening on the screen. Or the super-picky “I won’t bother to come back to a site that doesn’t display in Mozilla 1.1 for Linux” users. Or retro geeks like me who don’t want to pay for broadband and surf with Lynx when the S.O. is tying up the dial-up line with windoze updates. A little testing on older stuff and some creative solutions will make you a friend for life.
Of course having said that I’m realizing that I probably never fixed that Opera 7 bug. Bleh.
OK, that comment was confused.
What I meant was: Theoretically I would only exclude all the pre-7.1-Netscapes from my style sheets if I could
1. label the navigation effectively and non-jargony
2. keep a (XHTML strict, flexible-width) layout table
There really isn’t any way to distinguish navigation, other than its styling and positioning. Take away both, and you get bad user experience.
Practically, you do what you can. But I’m usually willing to trade some development time and some display oddities to keep the major visual cues for older browsers. Those users may be small in number, but so are some of our other constituencies – JAWS users, million-dollar donors – and we design for them, too.