What does browser testing mean today?

The most ridiculous demand I've ever heard from a client was that "the letters look different on my home computer". Same font, just a different size. He had his browser settings different at home. This dear fellow was seriously going to pop his cork over the issue. It was rather sad.

Just goes to prove the point that you can't control the end users preferences unless you make each web page a PDF and trust me, I've had requests for an entire site made up of PDF files. They made them in house and they were chock full of clip art and unreadable fonts. It looked like a child made them with Broderbund Print Shop circa 1984.

Anyone that demands pixel perfect likeness cross browser is just irrationally controlling and likely to go insane when they realize they can't make a grand dad in Japan use Verdana.

Beyond user preferences, browser difference are there and the issue isn't with CSS or (X)HTML as stated in the article. Browser differences also affect tables-based layouts just in case anyone thinks that's a panacea.

I had to work on a site like this recently as a subcontractor. The owner (a graphic designer who had 'some' experience with making websites when using tables was still the norm) was obsessing over things like the spacing between certain elements being a few pixels difference between browsers, rather than fixing more obvious usability issues like the fact that there were no main navigation links for important pages, that image links on the home page all did different things (some thumbnails weren't links, others were links but triggered different functionality), and that the order form had masses of fields that was going to deter most people from bothering to fill it out.

I found Andy's use of Modernizr, to control his use of modern techniques with non-supportive browsers, an attractive tool that warrants further research.

I don't really see the point of Modernizr at the moment. Most of the progressive stuff that I'd want to use can be used without needing to detect browser capabilities; if a browser doesn't support something, it doesn't render it and it doesn't really affect much, e.g. border-radius, text-shadow.

For things like RGBa as long as you specify a normal background-color that IE can understand before your RGBa value, there should be no problem.

About the only instance I can think of where you would need different styles is if you were using multiple background images as browsers that don't support it will choke on the syntax but I haven't found a case for using multiple backgrounds where I wouldn't all browsers to have the same experience yet.

My last (ongoing...) client wanted (still wants) his site in Humanist 521. He says he understands that people don't have that on their computer, but somehow it seems he doesn't really understand/realise that. The closest I could get was Trebuchet MS, which really, isn't anything like the Humanist fonts except the small g.

Lawlz. Luckily on my Linux box, I can show my clients what their competitors' sites look like with Gnome screwing up Firefox' font settings, with JS off, so they can see how different the competition is per machine (as the clients usually haz IE7 on Xp or Vista).

Still, they don't think it's unreasonable to say, everyone with Windows XP and IE7 should see the same thing, and I've spent a lot of time trying to make IE7 the "prettiest" version, re spacing and fonts. I suppose that makes sense in a country where most people use IE and Windows anyway.