Eric A. Meyer - CSS and Documents

CSS was first proposed in 1994, just as the Web was beginning to really catch on. In fact, the first draft of what would eventually become CSS (titled Cascading HTML Style Sheets ) was published mere days before the first release of Mozilla (soon to be Netscape Navigator) was announced.

At the time, browsers gave all sorts of styling power to the userthe presentation preferences in Mosaic, for example, permitted all manner of font family, size, and color to be defined by the user on a per-element basis. None of this was available to document authors; all they could do was mark a piece of content as a paragraph, as a heading of some level, as preformatted text, or one of a handful of other element types. If a user configured his browser to make all level-one headings tiny and pink and all level-six headings huge and red, well, that was his lookout.

It was into this milieu that CSS was introduced. Its goal was simple: provide a simple, declarative styling language that was flexible for authors and, most importantly, provided styling power to authors and users alike. By means of the cascade, these styles could be combined and prioritized so that both authors and readers had a saythough readers always had the last say.

Work quickly advanced and by late 1996, CSS1 was finished. While the newly established CSS Working Group moved forward with CSS2, browsers struggled to implement CSS1 in an interoperable way. Although each individual piece of CSS was fairly simple on its own, the combination of those pieces created some surprisingly complex behaviors. There were also some unfortunate missteps in early implementations, such as the now-infamous discrepancy in box model implementations. These problems threatened to derail CSS altogether, but fortunately some clever proposals were implemented, and browsers began to harmonize. Within a few years, thanks to increasing interoperability and high-profile developments such as the CSS-based redesign of Wired magazine and the CSS Zen Garden, CSS began to really catch on.

Before all that happened, though, the CSS Working Group had finalized the fairly weighty CSS2 specification in early 1998. Once CSS2 was finished, work immediately began on CSS3 (as well as a clarified version of CSS2 called CSS2.1). In keeping with the spirit of the times, CSS3 was constructed as a series of (theoretically) standalone modules instead of a single monolithic specification. This approach reflected the then-active XHTML specification, which was split into modules for similar reasons.

The rationale for modularizing CSS3 was that each module could be worked on at its own pace, and particularly critical (or popular) modules could be advanced along the W3Cs progress track without being held up by others. Indeed, this has turned out to be the case. By early 2012, three CSS3 modules (along with CSS1 and CSS 2.1) had reached full Recommendation statusCSS Color Level 3, CSS Namespaces, and Selectors Level 3. At that same time, seven modules were at Candidate Recommendation status, and several dozen others were in various stages of Working Draft-ness. Under the old approach, colors, selectors, and namespaces would have had to wait for every other part of the specification to be done or cut before they could be part of a completed specification. Thanks to modularization, they didnt have to wait.