RE: [docbook-tc] align common elements?

["Dick Hamilton" <rlhamilton@frii.com>]

Scott, 

Interesting thread.
>
> At what level do you think we should focus on DocBook
> interoperating with DITA and/or ODF?
>
I think the interoperability group demo puts a spotlight
on the most important areas of focus.  Namely, making it
possible for someone working in one standard to include
fragments authored using one of the other standards,
generate a combined document, and produce useful output
(web pages, help systems, printed docs, etc.).

I would argue that making it easier for an individual to
author content in more than one standard (which would be
the main value of aligning element names) is of secondary
importance, since pretty much any organization is going
to choose one standard for mainstream authoring and
production and treat other standards as "things to include"
rather than "things to author."  So, as long as they can
include content from the other standards and generate
output that uses those inclusions, I suspect they're
going to be happy.

That being the case (and I'll concede that's debatable),
I'd say that alignment of tag sets is not that big a
deal, as long as there are good conversions between the
standards.
>
> I was thinking that some of the more common elements/structures
> would have some benefit to being interchangeable. Perhaps at
> this point, the differences in content models prevent that and
> we just have to look to mapping the transformations...
>
I've been thinking about the same thing and I think the question
is how to create the cleanest transformations.  At the interop
group meeting last week, I suggested that instead of doing a
brute force conversion from DocBook to standard DITA, it might
be useful to create a DITA specialization that included a set
of DocBook elements and structures.  You'd still need a transform,
but that transform would focus on just those things that can't
be expressed in a specialization.

Norm's "DITA for DocBook" (http://norman.walsh.name/2005/10/21/dita)
discussion can be looked at as a similar exercise in the opposite
direction, with the additional advantage that it implements some
DITA capabilities that go beyond mapping tags/structures.

I'm not sure that either of these strategies is better than an
end to end XSL transform, but I do think that making good, 2-way,
lossless conversions by whatever means work best is more important
than trying to normalize tag names (unless you're starting from
scratch, in which case I'd suggest you just normalize to DocBook:-).

Dick Hamilton