Re: Comments on the XHTML 2.0 WD

On 27-May-05, at 0:49, Johannes Koch wrote:
> Christian Johansen wrote:
>> I don't think this is a very good example either. Consider this:
>> <nl>
>>    <li href="" title="This link takes you to the Home page of this 
>> site.">First Link</li>
>>    <li href="" title="This link takes you to the Sitemap.">Second 
>> Link</li>
>> </nl>
>
> [...] But with using an attribute like title, the value becomes atomic 
> and cannot marked up any further, which might be necessary.

In the case where additional mark-up is necessary, a definition list 
can be nested inside of the list-item.  This is much preferrable to 
adding a description tag to the other list models.  It keeps the models 
simple, and allows for a greater variety of structures.


On  26-May-05, at 12:36, David Håsäther wrote:
> I'm not sure I agree using a definition list with only terms, and no 
> definitions makes sense, though. Can't think of any circumstances 
> where that would be appropriate, got any examples?

I agree that there should always be at least one definition in a 
definition list.  In the examples given so far, using <dd /> would be 
appropriate.  In both cases, it was conceded that there /should/ be a 
definition ("_incomplete_ glossary", "not all attribute values have 
been documented _yet_") - the value of the definition simply has not 
been set.  Providing an empty definition clearly represents this 
situation.  Also, it is important to note that omitting a dd makes it 
impossible to distinguish between several undefined terms which will be 
given the same definition, and several undefined terms which have 
different definitions (neglecting the use of di)


I mostly agree with David Håsäther's amended content models for dl and 
di.

The current models allow lists containing only a single dt or a single 
dd.  This is against the whole purpose of the definition list, which 
pairs terms with their definitions.  Also, this admission encourages 
the improper use of definition lists to indent blocks.

David's MCMs require at least one dd and one dt. Furthermore, they 
enforces proper ordering of the elements (dt before dd), which the 
current model does not.

However, why should the dl model not be "label?, ((dt+,dd+) | di)+"?  I 
don't see any reason to discourage mixing di's with unwrapped dt's and 
dd's.


--Maxwell Terpstra

Received on Friday, 27 May 2005 20:16:54 UTC