Sent by Eric A. Meyer on 18 August 2002 23:11
At 8:02 +0000 8/18/02, Kevin W wrote:
>Also, some of the patterns are just useless.
I would prefer that we not start arguing about which are not
sufficiently useful, as it's been my experience that for every piece
of CSS branded useless, someone will find themselves in a situation
where the same piece is a lifesaver. To illustrate my point, I
thought I'd try to provide some examples of cases where the selector
types in question might be useful, at least to some authors.
>E:root
This one makes life a lot easier for implementors, not necessarily
for authors. However, I can envision cases where I want to do things
to the root element without knowing ahead of time what it might be.
Regardless of whether it's 'body' or 'html' in HTML documents-- and
it's not clear which it should be-- XML documents will have their own
root elements. CSS is looking forward to styling XML as much as it
is looking back at styling HTML.
>E[foo~="bar"]
a[class~="warning urgent"] {color: red;}
This was introduced of CSS2, actually. Since classes, for example,
can contain values composed of multiple space-separated words, it can
be useful to select an element based on two or more of those values.
Also, the following two are exactly equivalent:
div.example {color: red;}
div[class~="example"] {color: red;}
If you remove the tilde from the second line, the two are no longer
the same thing.
>E[foo^="bar"]
a[href^="mailto"] {color: green;}
See also the use to which it was put in the article "Going to Print,"
found at <http://www.alistapart.com/stories/goingtoprint/>-- possibly
a rare use, but nonetheless very useful for those who want or need it.
>E[foo$="bar"]
a[href$=".pdf"] {font-weight: bold;}
>E[foo*="bar"],
*[src*="doubleclick.net"], *[href*="doubleclick.net"] {display: none;}
>E:nth-of-type(n)
tr:nth-of-type(2n) {background: yellow;}
That can also be accomplished with ':nth-child' when you're using
HTML documents, but it's possible that an XML language would permit
table-row elements to be interspersed with others. For such
circumstances, you need the ':nth-of-type' pseudo-class to get the
alternate rows selected. (Or every third, or whatever.)
>E:target
*:target {background: yellow;}
That's one I've wanted for a long, long time. The ability to style
the element targeted by a link is really nice. That way, the user
doesn't have to guess which piece of the end of a document they're
supposed to be reading when you point them to it.
>E:selection
That's another one that implementors will probably find much more
useful than authors-- but still, it would be nice to set the
text-selection background color, wouldn't it? That way, you prevent
someone's light purple selection scheme from disappearing into your
light-purple background color.
>But it'll be hard to read someone else's CSS without consulting the
>reference. SelectORacle[1] is evidence of the unneccesary
>complicatedness that CSS3 is going into.
When I proposed and Kevin wrote the SelectORacle, it was to
explain CSS2 selectors, which many authors already found confusing
and complicated. I was talking about how nice it would be to have an
automated tool people could use, and Kevin got that look in his eyes.
He finished it in less than a week, I slapped some styles onto the
interface, and there it was. I wouldn't say this is evidence of
unnecessary complication in CSS-- I'd say it's evidence that people
can, and often do, like to have tools around to teach and remind them
how things work.
--
Eric A. Meyer (http://www.meyerweb.com/eric/), List Chaperone
"CSS is much too interesting and elegant to be not taken seriously."
-- Martina Kosloff (http://www.mako4css.com/)
