Late to the party, but since I was specifically mentioned it's only polite to reply ;)
The issue here is about the meaning of the word "semantic". Semantics refer to the *meaning* attached to something. What is the meaning of a div? It's *use* is structural, but it's *meaning* is ... well, it doesn't have a meaning. Only by attaching meaning via a class or id does a div or span acquire meaning.
DIVs are an interesting one in that they are "neutral", but they aren't really neutral. They just have a very "light" semantic meaning and it's mostly that their usage creates meaning for the page by adding structure. To me, the structure of a document is inextricably linked with the semantics of the contained elements. So together with headings they define the structure of the page. Although they themselves don't have specific significance, their usage is significant for the page. It gets a bit meta. They enclose semantically significant items. To approach this slightly differently, div+heading is the general non-form content equivalent of fieldset+legend.
It seems to me that many people here have different ideas about what semantic means. It would be helpful it we shared a common understanding in our conversations. I welcome, and invite, a *polite and professional* debate about the use of the term "semantic" as it relates to our work on the web. The "use" of something, and its "meaning" are not necessarily the same.
They're not necessarily different either :)
To come back to the original discussion about fieldsets, everyone has made it very clear what the "correct" way to use them is, and I don't disagree with them. I'm not interested in their correct (as defined by the specifications) use. As far as I'm concerned, the "use" of a fieldset is to group form controls and labels. But the "meaning" is, as the w3schools site says, to group related content.
The actual W3C spec says related form controls, it's only a confusingly-named third party that broadens the wording. Personally I discount W3Schools in favour of the actual W3C. [snip re testing in a screen reader]
I too would love to see the results of this experiment.
Interesting that the code you propose does render ok; but useful functionality is lost. So it's not directly harmful, but substantially detracts from other usage paradigms.
<fieldset><legend>staff details</legend> <dl> <dt>email</dt><dd>[EMAIL PROTECTED]</dd> <dt>phone</dt><dd>12345678</dd> </dl> </fieldset> Is perfectly valid, semantic markup which a screen reader would render just fine.
Changing the legend to a header (say an h2 just for discussion) and removing the fieldset also works just fine in a screen reader. In english-language documents the page order associates the H2 with the DL. If the content after the DL is totally unrelated, it would need another heading.
But can I point out, Ben, that at no time did anyone ever suggest placing form elements in the middle of general content. I'm not sure where you got that one from.
I consider fieldset and legend to be form elements, hence the comment. Admittedly that goes right back to the original point of discussion so I should have clarified a bit :) cheers, Ben -- --- <http://weblog.200ok.com.au/> --- The future has arrived; it's just not --- evenly distributed. - William Gibson ******************************************************************* List Guidelines: http://webstandardsgroup.org/mail/guidelines.cfm Unsubscribe: http://webstandardsgroup.org/join/unsubscribe.cfm Help: [EMAIL PROTECTED] *******************************************************************
