Håkon Responds to Questions About CSS and...

Håkon Responds to Questions About CSS and... 204

Posted by Roblimo on Friday June 23, 2006 @10:30AM from the partly-to-blame-for-web-2.0 dept.

You submitted questions for Håkon Wium Lie on June 20. Today we have his answers, not only to the (+5 moderated) questions we sent him, but to a bunch of others he thought would also be interesting to answer.

 > Where... by bcat24
 >
 > Do you think the W3C development process is too slow? I know that
 > you guys want everything to be perfect, but it seems to take far
 > longer than necessary. CSS 3 shows promise and I wouldn't want it
 > to die a slow death in standardization.

No, I don't think W3C is too slow. W3C isn't the bottleneck, browsers are. The dominant browser on the web hasn't been updated for years, and it doesn't make sense for specifications to get too far ahead. Rather, the CSS Working Group in W3C has focused on specification maintenance and achieving interoperability between implementations. This work is not so glamorous and some people — even within W3C — would prefer if they concentrate on new specifications. However, I think the focus on interoperability (which has resulted in CSS 2.1) has been crucial to the success of CSS.

CSS3 is a set of specifications that are developed more or less independently of each other. The best way to push a specification forward is to implement it. In the past year, we've seen some encouraging CSS3 implementations come along. For example, Mozilla supports multi-column layouts, Opera supports media queries, Prince supports cross-references and Safari supports borders and backgrounds. A few years from now, I think a select group of CSS3 modules will be interoperably supported in all browsers.

 > Why is CSS such a good idea but a pain to use?
 > by rar
 >
 > CSS is clearly very useful for separating style from content. But
 > apparently people tend to have problems when using it for layouts.
 > Would you say this is because people have not yet understood how to
 > properly do layout in CSS, or is it CSS that is lacking in this
 > area? What can be done to improve the situation? --- Would the web
 > benefit from HTML and CSS being complemented with some kind of
 > "layout language"?

I think layout and style should be tackled by the same language and the two are intertwined. Trying to split the two is like splitting the HTML specification in two, one specification describing inline elements and the other describing block elements. It's not worth the effort.

I think CSS is capable of describing beautiful and scalable layouts. The CSS Zen Garden has been a eye-opening showcase of what is possible today. If MS IE had supported CSS tables, another set of layouts would have been possible. So, there is still lots of potential in the existing CSS specifications which should be the next milestone. Beyond that, the CSS Working Group has started work on a new CSS3 module for advanced layout. Feedback is welcome.

 > CSS Evolution!
 > by eieken
 >
 > Is the wave of webpages designed completely in CSS what you
 > intially intended when you came up with CSS? Do you see that
 > changing? Is that good or bad?

I saw a clear need for a web style sheet language when proposing CSS in 1994. I also wanted CSS to fully describe the presentation of a web pages -- not just add some styling. All in all, I think it has turned out quite well. It has taken longer than I expected, but the scale -- due to the growth of the web -- is more than anyone could imagine.

I used "I" too many times in the previous paragraph. It's important to realize that CSS is a community effort rather that one man's work. Bert Bos joined me early and we worked out the initial designs on a whiteboard during the summer of 1995. The www-style mailing lists and the W3C CSS Working Group have also been crucial in ensuring the success of CSS.

If you're interested in the history of web style sheets, you'll find plenty of material in my PhD thesis on the subject.

 > Two questions (cut to 1.5 by editor Roblimo)
 > by Dolda2000

 > If you were allowed (perhaps by court order, which wouldn't be
 > unthinkable) to force Microsoft to do one (1) change in Internet
 > Explorer, what would that be?

I would force them to support one (1) single web page before shipping IE7, namely Acid2. By using a tiny amount of resources to get Acid2 right, Microsoft can save web designers and users endless amounts of frustration in the future. It would also be an honorable thing to do. This is what Microsoft's W3C representative wrote in 1998:

Microsoft has a deep commitment to working with the W3C on HTML and CSS. We have the first commercial implementation of HTML4, we were the first vendor anywhere to implement even portions of CSS, and we have put a tremendous amount of energy into seeing CSS mature to Level 2. We are still committed to complete implementations of the Recommendations of the W3C in this area (CSS and HTML and the DOM).

May I have one (1) more change? Please? Then I would make IE7 support TrueType downloadable fonts. Microsoft's record in fonts isn't that bad. They made their core fonts available for anyone to use, and IE supports downloadable fonts. Unfortunately, only the proprietary EOT format is supported. A few lines of code would be sufficient to support zipped TrueType fonts as well, and this would unleash a new wave of typography on the web. (To protect yourself, make sure you use a browser where author style sheets can be turned off — Shift-G in Opera). > As a bonus question: What do you think of Slashdot's CSS? ;)

The new design looks great! The style sheets behind the scene are more complex than what the average web page needs. But, we wouldn't expect anything average from Slashdot, would we?

> 6)  Opera
 > by taskforce
 >
 > Opera 9.0 seems to offer a lot of decent additions to Opera's
 > standards pool. How satisfied are you personally with the work the
 > team has done on implementing standards, and is there anything in
 > there you feel is superflous and anything you would have preferred
 > to see which wasn't in there?

I'm very proud of the standards support in Opera 9. Acid2 is an obvious favorite of mine and seeing that smiley face makes me very happy.

Among the more experimental features is support for Audio in HTML5. Web applications can now make sounds in a sensible way! Combined with the canvas element, developers can create Flash-like content without resorting to a proprietary format.

Having support for Bittorrent is also great. From a technical point of view, it makes much sense. Also, it's a political statement of sorts.

During the development of Opera9. Geir Ivarsøy, who founded Opera with Jon von Tetzchner, died after fighting cancer for years. Geir did a spectacular initial CSS implementation in Opera, thus convincing me to join the company. In music, the 9th is legendary. Beethoven, Schubert, Bruckner and Mahler all did 9 symphonies. Opera 9 was Geir's last symphony.

> Included styles, aliases
 > by Spy der Mann
 >
 > I always wanted to have "included" substyles or "aliases" in my
 > CSS definition, to save redundancy.
 >
 > (For includes)
 >
 >     .class1 { color:#ff0000; }
 >     .class2 { background-color:#ffffff; }
 >     .class3 { include:class1,class2;font-weight:bold; }
 >
 > (For aliases)
 >
 >     @alias color1 #ff0000;
 >     @alias color2 #ffffff;
 >     @alias default_image url('/img/image1.jpg');
 >
 >     .class1 { color:color1; }
 >     .class2 { background-image:default_image;background-color:co lor2; }
 >
 > This way we could change colors or images for a whole webpage
 > by editing a reduced number of lines.
 >
 > Had you considered any of these ideas in the past? If so,
 > why were they rejected?

Yes, aliases and constants have been considered. As David Wheeler noted, "Any problem in computer science can be solved with another layer of indirection."

CSS is already an indirection. Instead of putting properties and values directly on elements, it associates properties and values with selectors. What you (and others) are proposing is to add another layer of indirection. By doing so, one could possible write shorter, more manageable style sheets. However, there are also some downsides. It requires a new syntactic construct (@alias) and implementations must be able to remember a list of aliases. What if aliases are defined in one style sheet and referenced in another -- should that work? If so, what if the first style sheet isn't available?

For CSS1, the downsides of aliases were considered more significant than the benefits.

 > Definition of pixel
 > by Sara Chan
 >
 > The word pixel meant "picture element", but CSS redefined it
 >to mean something quite different (a particular subtended angle
 >of view [w3.org]). This causes confusion: CSS pixels are not pixels.
 >(Indeed, I have seen misinformed comments on Slashdot due to
 >that confusion.)
 >
 > My question is this: why call the subtended angle a "pixel", instead of
 >something else (e.g. "subangle")? If CSS wanted to use the subtended
 >angle for something, that is fine, but calling it a pixel seems to follow
 >the approach of Humpty Dumpty "When I use a word, it means just
 >what I choose it to mean".

In most cases, a CSS pixel will be equal to a device pixel. But, as you point out, the definition of a CSS pixel will sometimes be different. For example, on a laser printer, one CSS pixel can be equal to 3x3 device pixels to avoid printing illegibly small text and images. I don't recall anyone ever proposing another name for it. Subangle? Personally, I think most people would prefer the pragmatic "px" to the non-intuitive "sa".

> Vertical CSS Support
 > by infestedsenses
 >
 > As a developer who works with CSS every day, I find one
 > complication that continues to bother me in my daily work.
 > Support for CSS has always been good on the horizontal scope,
 > but vertical positioning has always been quite complicated.
 > Alone the procedure to affix a footer to the bottom of a screen
 > in dependance of the amount of content is unnecessarily difficult,
 > spawning hackish solutions such as "footerStickAlt"
 > [themaninblue.com]. Centering an object in the dead center of a
 > page also requires strange procedures such as this one [wpdfd.com],
 > which still aren't ideal (try making the viewport really small). The old
 > table method provided much easier methods for this. What are your
 > thoughts on this and do you see improvement following in future
 > CSS revisions?

Indeed, the CSS formatting model allows more control horizontally than vertically. This is due to (typically) having a known width, but an unknown height. As such, the height is harder to deal with.

However, CSS2 fixed positioning allows you to place content relative to the viewport (which is CSS-speak for window) instead of the document. For example, by setting position: fixed; bottom: 0 on an element, it will stick to the bottom. This works in Opera, Safari and Mozilla-based browsers. IE6 doesn't support it, however. It remains to be seen if IE7 will support it.

> About Microsoft...
 > by Chabil Ha'
 >
 > With MS's next browser release (IE 7), you mentioned in other
 > interviews that their decision to not supprt CSS2 was more a political
 > decision than a mechanical one. Aside from their obvious desire to
 > dominate the world, what politics do you think are in play that make
 > them not want to conform to the standard, and what do you think would
 > change that landscape so that they would have some initiative to
 > fully support it?

Great question. It's quite clear that Microsoft has the resources and talent to support CSS2 fully in IE and that plenty of people have reminded them why this is important. So, why don't they do it? The fundamental reason, I believe, is that standards don't benefit monopolists. Accepted, well-functioning, standards lower the barrier of entry to a market, and is therefore a threat to a monopolist.

From that perspective, it makes sense to leave CSS2 half-implemented. You can claim support (and many journalists will believe you), and you also ensure that no-one can use the unimplemented (or worse: buggily implemented) features of the standard. The only way to change the equation is to remind Microsoft how embarrassing it is to offer a sub-standard browser. And to use better browsers.

Another reason for not making a IE too good is that it will compete with Windows. A modern browser is an application platform; the combination of HTML, JavaScript, CSS and DOM allows developers to target the web instead of Windows, Linux, or Mac.

> From linvir
 > How long since you last used Linux?

I'm using it right now. Ubuntu on a IBM Thinkpad X41 is the environment I live in. Ubuntu rocks -- especially with Opera on top! (And Emacs right underneath.)

 > From Rob T Firefly (844560)
 > Why the curly brackets?

The initial CSS proposal didn't use them, instead relying on newlines to separate statements. TimBL didn't really like that and I therefore borrowed the curly braces from the C programming language. The syntax for comments came along as well. I think it works quite well.

 > why not XML?
 > by slashdot.org
 > Simple question (hopefully simple answer ;-)): why
 > did you not use XML?

The simple answer is that the development of CSS preceded XML by a year or so. However, if XML had been available, would we have used it? Probably not. And I suspect Brendan Eich of JavaScript fame would answer the same way on behalf of his language.

XML is a great syntax for structured data, but not suitable for all languages. Still, I think the SGML-based syntax for the FOSI style sheet language.

 > Padding
 > by Anonymous Coward

 > Why was the decision made to make padding apply outside
 > of the width of a 'box', rather than inside, which would seem
 > to make more sense?

It makes sense in some situations, but not in others. For example, when a child element is set to width: 100%, I don't think it should cover the padding of its parent. The box-sizing property in CSS3 addresses this issue. Ideally, the issue should have been addressed earlier, though.

> by nuzak

 > why not float:

 > DSSSL had this sort of thing solved before HTML even existed,
 > let alone CSS. But scheme is too scary and icky, and the W3C
 > believes in a principle of least power, so CSS has to be fully declarative,
 > static, and crippled until patched later.

You're wrong about DSSSL -- it didn't support floating text (as in having text wrap around images) at all. And the DSSSL specification only became publically available around 1996, years after HTML.

> by MagicM

 > How frustrating is it to write a specification knowing
 > that you're at the browser vendors' mercy?

That's part of the game. I don't think any specification has a birthright to be fully supported by all browsers. There should be healthy competition between different specifications. I believe simple, author-friendly specifications will prevail in this environment.

Microformats are another way of developing new formats. Instead of having to convince browser vendors to support your favorite specification, microformats add semantics to HTML through the CLASS attribute. And style it with CSS.

> New standards
 > by iamsure
 >
 > In your work at Opera, you have clearly paved a path that includes
 > going beyond the W3C standards. Whether it is WhatWG
 > implementations, or new functionality specific to Opera (2dgame),
 > you are pushing into new territory. Can you explain why W3C isn't
 > sufficient, and why efforts at Opera to expand beyond the standards
 > differ from Microsoft's embrace/extend model?

It's a fair question. The WhatWG was set up when it seemed as if W3C didn't care much about browsers anymore. That has definitely changed and work items from WhatWG are now channeled into W3C (e.g., XMLHttpRequest).

At Opera, we sometimes include experimental features before they have been standardized. When this happens they are labeled as such, but we still try to document them. For example, we support some Opera-only CSS properties for XML. If these features gain traction, we are happy to work with other organizations to standardize them. If they don't become popular, the features will most likely disappear.

 > Beyond HTML
 > by pr1000

 > How far can CSS be taken beyond the web page--that is,
 > have generalized or non-web specific features for such things
 > as page formatting or type setting? Do you plan/wish/hope to
 > take it farther than it currently is?

Yes, I think it's possible to take CSS further in several directions. I'm eager to see CSS being used in paper-based publishing and I joined the board of YesLogic — which makes the Prince XML to PDF converter — to make sure they added my favorite features. Bert and I used Prince to generate PDF from HTML and CSS sources for the third edition of our book. W3C just published a new Working Draft which describes features for printing, e.g., footnotes, cross-references, and even generated indexes.

For mobile units, I think Media Queries will be important. For example, they can express that large images should not be sent to mobile devices.

Another great opportunity for CSS is Web Applications. Just like documents, applications need to be styled and CSS is an intrinsic component of AJAX. The "AJAX" name sounds great, but allow me to propose a few alternate spellings that I find more accurate:

AJACX: Asynchronous JavaScript, CSS and XMLHttpRequest
ADJACS: Asynchronous DOM, JavaScript and CSS
ADHJACS: Asynchronous DOM, HTML, JavaScript and CSS
AJAHCS: Asynchronous JavaScript, HTML and CSS
AJACS: Asynchronous JavaScript, HTML and CSS

Opera, Mozilla and Safari developers are collaborating in the WHAT WG and in W3C to make sure we have interoperable specifications for AJAX. I mean, ADHJACS.

 > by crush

 > Is the difficulty of producing a layout that consists of
 > three or more columns of equal height justification for
 > adding some new feature to the specification to make this easier?

I don't think so. CSS2 defines a table layout that can be used for this purpose. The problem is, and I'm repeating myself here, that the dominant browser doesn't support it. Adding yet more features to the specification wouldn't help.

 > How come, we um, lie?

Right. My name is, um, a bit troublesome to pronounce in English. But I'm a nice person who generally tells the truth. I often tell non-Norwegians that my first name is pronounced "howcome". That's close, but not quite. It does make for a great email address like howcome@opera.com, though.

So I may be a Lie, but I'm just a little white one :-)

This discussion has been archived. No new comments can be posted.