Dylan Schiemann

Being the elitest that I am, I like to poke fun at people who make the statement: “I designed this site entirely in css”. Recently, the designer of the newly launched mozilla store proclaimed that it was the first site he had designed entirely in css. Sure, CSS is a language used to describe presentational style, but you still need an html/xhtml/xml syntax of some kind to describe the content. Then the perverse thought occurred to me… what is the minimal amount of markup that could be used to develop a site? In Internet Explorer, because of its very loose rules, it turns out that you can define an entire web site with the following block of markup:

<link href="style.css" rel="stylesheet" type="text/css" />


Then, you simply set any css rule in the stylesheet that uses an expression to modify the dom of the document with the markup included in the css document. So really, you’re still using markup, and in this case JavaScript. However, because everything is in a stylesheet, you’d truly be designing a site entirely in css. If you have an idea for an equally useless trick in mozilla, please contact me.

P.S. No offense intended to Dan (you were just low hanging fruit) or to the mozilla store … in fact, I was wearing my mozilla shirt while writing this post. That’s almost as bad as wearing an In-N-Out Burger shirt to In-N-Out.

I was recently invited by David to setup a G-mail account. The system is currently by far the best e-mail interface of the major web mail providers, though that lead may diminish with Yahoo’s recent acquisition of OddPost. The main reason is that Google uses JavaScript and XMLHttpRequests to update data, rather than loading a new page for each message or mail action. What this means to the end user is that the application is fast and responsive. Google also has some interesting ideas in arranging messages by conversation and filter, and by showing/hiding message components in a convenient, intuitive manner. The service does have some negatives: there is currently no direct pop3 access (though there are ways around this limitaiton…), and the color scheme of the interface is a bit too bright (it matches Google’s color palette, which is ok in moderation, but a bit obnoxious in the quantities used within gmail.

Because of mozilla’s superior user experience and security, and Internet Explorer’s ongoing slew of security holes, numerous organizations have recently recommend switching away from IE, including the Department of Homeland Security, CERT, Security Focus, Open for Business, and even Microsoft’s own Slate Magazine.

I was recently thinking out loud that Google should punish web sites with markup that are not well-formed, invalid, and/or excessive in their use of presentational markup. Then it occurred to me that, in fact, they already do (excluding really large, popular web sites). They just don’t explicitly say that they do. Instead, what they do is reward sites with great markup. If they came out an explicitly said that they would reduce the rank of badly developed web sites, they could effect a lot of change amongst web developers.

The W3C recently held a Workshop on Web Applications. Opera and mozilla have formed the Web Hypertext Application Technology (WHAT) Working Group. Microsoft is trying to force xaml upon the world. Things are getting interesting again.

Daniel Glazman has added an entry about three ways to emulate getElementsByClass. Of course, there’s another method if your document is xhtml: use xpath like this…

function getElementByClassName(needle) {
var xpathResult = document.evaluate(‘//*[@class = “‘needle'”]’, document, null, 0, null);
var outArray = new Array();
var item;
while (item = xpathResult.iterateNext()) outArray[outArray.length] = item;
return outArray;
}

Code updated to address two minor issues brought to my attention by Biju.

Ian Hickson was written an interesting proposal for server-sent dom events, basically a way to avoid some of the iframe hacks to keep an open stream to a server which would deliver events to client-side code.

Daniel Glazman writes in his blog about 42 problems with xhtml 2.0. Daniel is a Netscape France employee, and someone I highly respect from my interaction with him on the W3C public mailing list.

Is it just me, or were the real April 1st news stories on slashdot scarier or more unbelievable than the fake stories this year?

With the common strategy on recent episodes of the Apprentice being to sell future services, inferior domain name provider network solutions has stooped to new lows, trying to convince people with a straight face that they should renew their domain name for 100 years. Network Solutions tactics bother me for several reasons:

Regarding the last point, Network Solutions appears to be taking lessons from Microsoft in promoting this service as a way to reduce confusion and worry. If they had built a simple system from the beginning that did not result in constant whois spamming, there would not be as many issues with renewal and domain theft. Why should a customer be convinced by a company to pay for a solution to a problem that was caused by the company to begin with? I make this reference to Microsoft because I believe that average people are afraid to buy and install software because Microsoft has intentionally made it confusing, difficult, and insecure, thus making it easier for them to bundle software with future versions of their operating system. In fact, I will go on to say that Microsoft’s legal team intentionally tanked the Eolas case, as this patent is a plug-in killer (Flash, Java, QuickTime), something Microsoft could not kill on its own without getting into even more legal trouble.