16–? November 2002
In Issue 155 of A List Apart, for people who make websites: “Flexible Layouts with CSS Positioning.” Want to spend less time on CSS hacking and more time on design? Need your site to look as good on a 160 x 160 PDA screen as it does on a 1024 x 768 PC monitor? Dug Falby shares two techniques for practical grid-building based on CSS’s sometimes abused, often besmirched, and widely misunderstood absolute positioning. Held over from last week: Flash Satay. Have your Flash and web standards, too.
24 hours from this writing, we’ll be landing in Boston for four days of peace, love, and Web Design World, where we’ll deliver three talks, participate in a panel or two, and dig on some of the smartest people in the web industry.
The Truth about Standards :::
In “The Truth About RDF,” Dave Winer says something that resonates with us:
“I am a disbeliever of anything that requires ... much documentation, head-scratching, hand-waving, and eyes-glazing-over.”
Dave is talking about RDF, but his comment applies to all web technologies. To become a web standard, a technology should:
- Solve practical problems. (A standard whose primary purpose is to prove a theory will be of little real use to designers and developers.)
- Be easy to understand. (Steep learning curves are okay for applications like Director and Photoshop. But if a web technology is so complex that even experts aren’t sure they understand it, it’s unlikely to be widely adopted or correctly used. If it’s vague as well as complex, the door is opened to incompatible implementations—1997 all over again.)
- Be capable of practical implementation. (Don’t expect anyone to take a new technology on faith. Be sure it works before asking companies to base their businesses on it.)
- Lend itself to widespread adoption. (A standard that isn’t widely adopted is merely a thesis.)
These principles are as obvious as the notion that a hammer should be hard enough to drive a nail. Yet lately some emerging specifications seem to have wandered away from the idea that standards are supposed to help people solve real problems in practical ways. Some proposed standards (and a few existing ones) smell like science projects to us. Overly complicated and obscure specifications with no rationale and no obvious benefits are more likely to be ignored than adopted.
HTML was basic and inconsistent but it empowered millions. CSS made the grade because it offered design control without bloat while allowing users with special needs to change what they saw. The W3C DOM saved designers and web users from the hell of incompatibility and “Best Viewed With...” banners and gave developers access to the data embedded in web documents. XHTML 1.0 brought consistent rules to traditional web markup and helped it play well with XML applications. These standards made sense because they solved real problems.
To keep making sense, standards must continue to solve real problems. They must not be designed to prove theories or to fix what’s not broken. They must be tested in implementations that have some bearing on the real world of web development. Their documentation must be clear, comprehensible, and free of the vagueness that leads to incompatible implementations based on guesswork. They should discard existing methods only when replacements offer overwhelmingly compelling—and realistic—benefits. And they should strive for compatibility with existing specs. A proposed standard that ignores these principles is an academic exercise. We prefer to take our exercise in the gym.
Watch a live webcast of Håkon Wium Lie’s lecture on “Scalable, Beautiful Web Documents” at the MIT Media Lab on Tuesday, 19 November at 4:00 p.m. EST. Håkon is the father of CSS and CTO of Opera.
Worth reading: “Accountability of Accessibility and Usability” by Anitra Pavka. The title’s a mouthful but the content is easy to digest. Pavka examines web accessibility in the light of recent legal challenges and invites our industry to create guidelines instead of waiting for Big Brother to tell us what to do.
From the mailbag: Anonymous Coward asks, “Why do you refer to yourself as ‘we?’” This old page answers that question and others.
15 November 2002
The demure little heart character that used to mark permanent links here was causing the IE/Windows crashes we reported yesterday. ASCII includes 127 basic English characters. The world, of course, uses thousands of characters. To make non-ASCII or special ASCII characters appear on a web page, you use an “escaped character sequence.” The sequence for a typographically correct apostrophe is
’. The heart is encoded
♥. A conformant browser or Internet device translates this gibberish into the appropriate character or symbol. The visitor enjoys an “image” at the cost of a few bytes of text.
Opera never displayed the heart character correctly. Not even the new Opera 7 beta got it right. In the scheme of things, that seemed unimportant. Opera users didn’t see a heart, but they did see a clickable box, and that was good enough. (Similarly, Netscape 4 users don’t see CSS borders on Fox Searchlight, but the site looks okay in their browser and it is usable. In web design, if you create something that looks good in conformant browsers and is usable in less capable browsers, you’ve done your job.) Even older versions of Opera got most of the new layout here right. The absence of a small, decorative flourish wasn’t worth worrying about.
But crashes are another story. All versions of IE/Win displayed the little heart, but some crashed or froze when visitors clicked it. That was unacceptable. This site is now heartless.
Software development is complex. Browser makers often don’t know about latent problems in their products until a designer or developer stumbles upon them. In an earlier iteration of this site’s redesign, we used a spade. IE/Win couldn’t display it. We switched to a heart. You know the rest. Readers who identified the innocent heart character as the cause of their IE/Win woes include David Roberts, Ale Piana, Chris (no last name given), and Don Hinshaw.
For what it’s worth, in our tests, the following browsers had no problem with the little heart: Mozilla 1.1, Netscape 6, Netscape 7, IE5/Mac, and IE6/Win XP. The heart died that other browsers might live in light everlasting.
We recently discovered a smart site called ReadingEd.com. In addition to offering opinions and information on various geek topics, the site sports a fascinating “Outside Reading” feature and a clean, readable CSS layout. Very nice work.
One more (to us) recent discovery is Usability Views, offering visitor-sortable listings of “1,786 articles about usability, IA, HCI and web design.” The site tracks online magazines (including A List Apart) and sorts articles by date, title, or “popularity.” According to its ALA popularity index, Mark Bernstein’s “10 Tips on Writing the Living Web” is the most-linked feature A List Apart has ever run.
14 November 2002
[? | 8 pm | 6 pm]
We’re making three presentations in Boston next week, and just finished preparing for them all. Ahead of schedule, even. Whew. Lotta work.
We’re an all-Mac shop again (the Dell machine went to a needy neighbor) and do all our Windows testing in Virtual PC. Great product. Subpixel antialiasing in Windows XP is a beautiful thing (screenshot, 217K jpeg). Seeing it on an Apple Cinema Screen is sweet and somewhat surreal.
Last week we made a fourth attempt to install OS X 10.2 on our primary production machine, a dual processor G4 with all the trimmings. As with the three previous attempts, the effort failed. OS X installed fine, but after booting out of it to get some work done in OS 9, we were unable to boot back into it (grey screen of death). Fortunately we had also spent money on OS X utilities designed to fix problems in Apple’s latest operating system. Unfortunately these utilities simply froze. Conclusion: OS X cannot run on this machine. Maybe it’s the SCSI card. No SCSI devices are connected, but maybe OS X chokes on the card itself (the card was factory-installed by Apple when we bought the machine in 2000). Maybe it’s the RAM (factory-installed by Apple when we bought the machine).
OS X may be the New Coke of operating systems. It may be Steve Jobs’s revenge for the failure of NeXT (a Mac-like GUI that ran on Unix). Or it may be the beginnings of the greatest operating system ever created for a desktop computer. All we know for sure is that it won’t run on one of the best and most expensive computers Apple ever made. Yet with Virtual PC, this very same Macintosh computer runs Windows XP without a hitch. We will not comment on the attendant irony.
Two readers have told us the redesigned zeldman.com crashes for them in IE6/Win XP after reporting an “illegal instruction” error. We’re unable to reproduce the crash or the error in our own tests. It’s always something. [Update.]
13 November 2002
[2 pm | 11 am]
The discussion being generated by ALA’s Flash Satay is as intriguing as the article itself. Topics include alternate methods that work and validate, problems with IE/Win’s handling of the OBJECT element, and the implications of XHTML 2.
A reporter has asked us to show him an accessible corporate site—or an accessible mockup of an inaccessible corporate site. We could not think of a single example. If you can, or if you’ve created one, please share with the class.
A public beta of Opera 7 for Windows has left the building and is said to be DOM-compliant, though Opera’s News page doesn’t mention this long-desired improvement. To compete with MSIE and Netscape, Opera not only needs to be faster and lighter. It must also support the same technologies its competitors do—and DOM support is where Opera has traditionally stalled. Early field reports indicate Opera 7 has broken the DOM logjam and enhanced CSS performance. Is this the version of Opera its users (and web developers) have been waiting for? (Opera’s scripting specs are now available.)
We dreamed Kelly Goto, Hillman Curtis, and Eric Meyer were mobsters, planning a hit on a competitor who’d gotten too big. The guy to be whacked? Josh Davis. We need to get out more.
12 November 2002
[4 pm | 1 pm]
“Jeffrey Zeldman is Not Obsolete.” The SXSW Tech Report interview.
The AltaVista search engine has relaunched, with an ugly design and crummy, invalid markup straight out of the 90s. Developer Trip Kirkpatrick wasn’t about to sit still for the latter. Details at webstandards.org.
iStockphoto has re-launched. Version 8 sports new features and a fresh facelift. iStockPhoto is a collection of over 26,000 royalty-free photos, illustrations, and multimedia files created by a growing international community of artists. The site adds around 1,000 new royalty-free photos each week.
9–11 November 2002
[holiday weekend edition]
In Issue 154 of A List Apart, for people who make websites: “Flash Satay” by Drew McLellan. “This site uses Flash. This site validates as XHTML. They said it couldn’t be done. Now it can be. Have your Flash and standards, too.” Please note, the ALA server may be slower than normal due to heavy traffic.
The veil of secrecy has lifted. Macromedia Contribute is now available. Based on the Dreamweaver MX engine, this new $99 desktop application allows non-technical people to update web content without breaking the design, code, or functionality. Designers and developers specify which pages can be changed, which templates can be used to generate new pages, and which parts of a page are off-limits. The intuitive interface has been tested on schoolteachers, writers, and other non-web-pros, all of whom were easily able to complete tasks. Users can grab content from non-standards-compliant apps like Microsoft Office; Contribute will clean up the code and can even convert presentational tags to CSS and generate accessibility elements and attributes. Version control is also included. We saw this product twice before it was released and plan to buy copies for selected clients. It is perfect for those who can’t afford or don’t need a full-blown Content Management System. Contribute for Windows is available now; a Mac version will follow. The Mac version will include an embedded copy of Opera 7, with greatly improved DOM support.