General collation-override element

[Deborah_Pickett@moldflow.com]

Please don't think about this until after DITA 1.1. :)

I notice in the DITA 1.1 draft that the indexing domain gives us 
<index-sort-as> to allow writers to specify a non-standard collation value 
for an <indexterm> entry.  Ostensibly this will have a lot of use to 
Japanese writers but Westerners might also get some use out of it in edge 
cases.  Good, I like that.

My contention is that this feature is useful for more than just index 
generation.  It will be useful in any situation where sorted, generated 
lists happen.

One example is auto-generated lists of trademarks (for which an element 
exists in bookmap).  Some books I've looked in alphabetize their trademark 
list in the front matter (Adobe CS is one I found.)  If one of my 
trademarks is the Japanese <tm>会社の日本</tm>, where does it go in the 
list?

I suggest a specialized element (perhaps from topic/data, but the specific 
implementation isn't important) that permits an element to override its 
default collation value (i.e., its text content).  Normally the element 
would be ignored, but list-generation processing could detect it and use 
its value rather than the containing element's.

In my example, I could then say (with Hiragana) <tm>会社の日本<collate-as>
かいしゃのにほん</collate-as></tm> or (with romaji) <tm>会社の日本
<collate-as>kaishanonihon</collate-as></tm>, and let the collation value 
be picked up by processing.

Other cases where collation values might be useful:
- Lists of table or figure titles for collation into a list of same.
- Lists of first lines of a collection of poetry.
- Lists of keywords or apinames (say, in a programming reference).
- Bibliographies (to get correct author family name in a multi-word name).

Looked at from this angle, an index is just another generated list, albeit 
the most common kind.

-- 
Deborah Pickett
Deborah_Pickett@moldflow.com