- From: Matthew Raymond <mattraymond@earthlink.net>
- Date: Wed, 09 May 2007 07:31:47 -0400
- To: Dão Gottwald <dao@design-noir.de>
- CC: Rene Saarsoo <nene@triin.net>, "public-html@w3.org" <public-html@w3.org>
Dão Gottwald wrote: > Rene Saarsoo schrieb: >> Anyway, this list of applicable elements seems kind of >> limiting. For example it seems natural, that I would like >> to write an example of computer code, like this: >> >> <pre class="example">Some code...</pre> >> or >> <code class="example">Some code...</code> >> >> But predefined class "example" is only allowed for aside, >> figure, p, section, span :( > > For any other element, it would be equal to a freely chosen, unspec'd > class, no? Exactly. By specifying the particular situation where a class is most likely to have a given semantic, we can associate semantics with that class without causing problems in other circumstances where the class may have a different meaning. Note that |role| is not immune to the problem of naming conflicts. Because new "roles" may be added in future HTML specifications, earlier user agents will have to ignore unidentified roles so that they can preserve graceful degradation for future HTML specs. Thus, people may start using roles that are undefined for a given namespace (or use undefined roles without namespaces). This is especially true of clueless web authors who don't necessarily understand what |role| is: | <p role="to tell a story"> As a result, when HTML 6 comes along, we may run into the same problems with |role| that we currently have with trying to create predefined classes. Authors will have been using role names for purposes slightly or even significantly different from what HTML6 would define them for. Thus, |role| doesn't solve the problem of naming conflicts with legacy content. It just delays dealing with the problem until the next specification. At the same time, it prevents us from adding content to existing class names in situations where it does no harm to do so. (That's not to say that I'm convinced that classes or roles should have predefined semantics in the HTML specification to begin with, but that's a separate issue.)
Received on Wednesday, 9 May 2007 11:28:53 UTC