Haven't checked the most recent weekly, but be glad you don't have to witness the mess that is KLM.com in the newest internal build. The reason? Regressions in recent builds that mean we don't parse things like this as expected:
<div> <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd"> <html><head> <meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /> <title>KLM Royal Dutch Airlines</title> </head> <body> <div id="langdata"><ul><li id="aw">Aruba<ul><li class="en">English</li><li class="nl" id="selected">Nederlands</li></ul></li></ul></div></body></html> </div>
Does that look ugly? This, ladies and gentlemen, is what you get when you do document.createElement('div') and follow up by setting the innerHTML of said div to the responseText of an XMLHttpRequest that just fetched an address that turned out to be a complete HTML document!
KLM is a weird site indeed.. Look closely at the markup. Note how they carefully avoid custom attributes, instead using HTML4's "id" and "class" to carry the information about two-letter country codes and language codes. You would think they actually cared about writing valid HTML…
And here is the other fun part: when this breaks, the whole site becomes unusable. Now, of course this is a bug in Opera and we won't ever release it so it isn't much of an issue but the question I'd like to ask them is: since you actually appear to care at least some about correct HTML coding, what part of the progressive enhancement idea did you not understand?