RE: [dita] Specentry attribute

Sorry I wasn't clear. I'm not trying to design anything but to clarify
the existing design. The current design of simpletables does not allow
multiple header rows so I didn't think about that. Most likely it is
best that it doesn't support multiple header rows given that it is a
"simple table".

> -----Original Message-----
> From: JoAnn Hackos [mailto:] 
> Sent: Friday, December 30, 2005 10:48 AM
> To: Paul Prescod; 
> Subject: RE: [dita] Specentry attribute
> 
> Paul,
> How will you account for tables that have more than one 
> heading row? I am thinking of tables in which you may want to 
> vary the headings mid-way in the table to accommodate a 
> change in the content. This without having to introduce a 
> more complex table style. 
> Joann 
> 
> -----Original Message-----
> From: Paul Prescod [mailto:]
> Sent: Friday, December 30, 2005 11:08 AM
> To: 
> Subject: [dita] Specentry attribute
> 
> I don't think that the DITA spec is explicit about how to 
> derive a header from specentries. For example, the DITA 
> toolkit's FO implementation uses the specentry elements from 
> the first row, but I don't think that the DITA specification 
> says that it should. It also seems unclear on what should 
> happen if there is already a header row. In addition, there 
> are issues around translation that are not addressed by the 
> spec. I propose the following text:
> 
> 'The specialized entry attribute allows architects of 
> specialized types to define a fixed or default column title 
> for a specialization of simpletable. It is not intended to be 
> directly used by authors.
> 
> If a simple table does not have an sthead, and there is an 
> stentry element in the first strow that has a specentry 
> attribute, then DITA processing software should generate the 
> header. The software should generate a column header for each 
> stentry in the first strow, using the specentry attribute 
> value when available and the empty string otherwise.
> The software shold generate the header text by using the 
> specentry value as a lookup key for finding 
> language-appropriate text, falling back to the actual content 
> of the attribute if no language-appropriate text is defined. 
> 
> DITA processing software should ignore specentry attributes 
> in the following circumstances:
> 
>  * when there is an sthead in the simpletable
>  * when they appear on strows other than the first in the 
> simpletable '
> 
>  Paul Prescod
> 
> 
>