- From: Etan Wexler <ewexler@stickdog.com>
- Date: Thu, 10 Apr 2003 02:54:35 -0700
- To: Ernest Cline <ernestcline@mindspring.com>, www-html@w3.org, www-html-editor@w3.org
Ernest Cline wrote to <mailto:www-html@w3.org> on 9 April 2003 in "[XHTML2] Poor little old <a>" (<mid:3E94831D.26111.544D0F9@localhost>): > There is very little purpose that <a> serves in the > current working draft. I agree. > About the only thing it still has going for it > is that links specified by <a> are still supposed to look like links. How do the links look when one accesses the Web by telephone? Device independence aside, there is no reason that links specified on elements not of type 'a' should not appear as links. > Now have the behavior of <a> be defined so as to [allow nesting] so > that if the user agent cannot find or use doc.pdf then it tries > doc.txt. This case is crying out for a link that has multiple targets and some inband metadata associating the target resources as alternatives to each other. There is significant difference between the 'a' and 'object' element types. An 'object' element is to be replaced in rendition by its target resource, if possible. If not possible, the element is to render with its document content. If the element's document content is another 'object' element, the nested element is tried in the normal manner because its retains the normal semantics. An 'a' element is not intended to embed its target. Therefore the content of a parent 'a' element and its nested 'a' child element should both be rendered and accessible in a given document rendition. The alternatives will not seem like alternatives to the user. If you propose to change the intended behavior of the 'a' element type, I say that it is not worth the effort. Let's just drop the 'a' element type and include a suitably expressive linking mechanism. -- Etan Wexler <mailto:ewexler@stickdog.com>
Received on Thursday, 10 April 2003 05:55:03 UTC