Flutterby™! : CSS and tables 2003-01-28 18:11:34+00

CSS and tables

2003-01-28 18:11:34+00 by Dan Lyke 7 comments

I am under the distinct impression that the rise of CSS has reduced the amount of semantic information in web pages. We see <span> with special styles used when <cite> would be better. We see the anti-table atmosphere blow away tabular data in favor of wacky

schemes. The latest is Dave Winer's request to redesign Weblogs.com, with an explanation of why and some solutions. Now I do some mucking about on the Flutterby pages to accomodate slow downloads, I understand why Dave wanted to do away with a table that takes a long time to render, but it bothers me that we're seeing semantic information stripped. Especially when the RSS

variants, with their entity encoded HTML, is what's being pushed as "the semantic web" these days.

But there's nothing particular to CSS that says not to use semantic tags. You can style cite more easily than "span class='whatever'"--people just don't know they mean whichever tag it is. I'm curious how many folks who use a span when they mean cite would, without CSS, just use i.

No, my whine is more aboutthe realities of what happens than whether or not something is possible within a given framework.

Anything can be implemented badly. Just because people are incompetent in their use of a technology doesn't provide a legitimate argument not to use that technology.

It's more about encouraging the use of technology. I'm a strong supporter of 2^nd Amendment rights, but from a social standpoint there are people in my neighborhood I do not want carrying. Licensing may be a little extreme, but we need to encourage education on the use of CSS before we encourage people to blindly start using it. And in Dave's case, we should be encouraging browser makers to optimize the use of the right tool rather than pushing workarounds.

Alright, I looked a little closer at the solution that Douglas Bowman suggested and the apparent solution that Dave used. You've still got a semantically correct document. Weblogs.com is a list. Isn't it? Douglas Bowman's example presents it as a list. The only thing that bothers me is that he's using an <em> tag for the "last updated time" but I'm not sure what I'd recommend to replace it with.

This brings me to another point. Yes, I understand that the <span> tag has no inherent semantic meaning. I agree that this should be the case. However, why can't the class attribute be used with it to give it a custom semantic meaning?

Actually, I see the information as clearly a table. It's a list, but the numbering is less important than that it's a set of paired values and attributes. Seems to me like a table is the best way to describe that data.

Actually, I'm a little surprised that it wasn't just done in XML with a linked stylesheet. I can see it now:

<item>
  <weblog order="n">Weblog Name</weblog>
  <updated>00:00:00GMT 00/00/0000</updated>
</item>
<item>
  <weblog order="n+1">Next Weblog Name</weblog>
  <updated>00:00:00GMT 00/00/0000</updated>
</item>

Then a standards group could organize and there could be a fork in the standard over RDF being included.