OASIS Darwin Information Typing Architecture (DITA) TC

Michael Priestley wrote:
> 
>  >in the
>  >specialization modules I create I'm under no obligation to *allow*
>  >attributes from any particular domain.
> 
> Specialization modules don't control what attributes or elements get 
> added from a domain. The doctype shell controls that, when the domain 
> gets integrated with structural types. At that point, you cannot choose 
> which elements will get the new attributes - it's all or nothing: if you 
> integrate the attribute domain, then all the elements in your doctype 
> get the new attribute.

I think I see: every element type is required to use the attribute group 
(parameter entity) that lists all domain-defined attributes, so there's 
no way to limit a given element type to a specific set of attribute domains.

Hmmm.

It can't happen for 1.1 but I still thinking we will need a way to allow 
specialization-specific attributes on individual specialized element 
types. Otherwise I am forced to put anything that would otherwise be an 
attribute into a specialized subelement, and only then in a context 
where there is a repetition of a base type I can specialize from (e.g., 
elements that allow 

"W. Eliot Kimber" <ekimber@innodata-isogen.com>
wrote on 01/26/2007 04:36:22 PM:


> It can't happen for 1.1 but I still thinking we will need a way to
allow

> specialization-specific attributes on individual specialized element

> types. Otherwise I am forced to put anything that would otherwise
be an

> attribute into a specialized subelement, and only then in a context

> where there is a repetition of a base type I can specialize from (e.g.,

> elements that allow <ph> in their content.

Absolutely. I believe there's a proposal on the slate
for 1.2.

>

> For example, in a specialization I'm doing now, I have specialized

> paragraphs that have invariant paragraph numbers that needs to be

> captured in the markup. Normally this number would be an attribute
since

> it's a property of the paragraph and not part of it's core content
(that

> is, it's an identifier that happens to be displayed in most renderings).

>

> But since I can't (legally) add attributes, it seems like my only
choice

> is to create a specialized phrase that holds the paragraph number.

>

> Or have I missed something?


<data> might be a better choice than <ph>
but otherwise I think you are correct.

Michael Priestley

IBM DITA Architect and Classification Schema PDT Lead

mpriestl@ca.ibm.com

http://dita.xml.org/blog/25