Re: [dita] What Is A Topic

Defining a topic is, I suspect, going
to be as slippery as any other "you know it when you see it"
thing.  It's clear from the responses so far that some have a functional
definition (it answers one question or described one thing), others have
a content-model definition (what the DTD permits), others have a unit-based
one (to be authored, or read, in one piece).

Being a fan of specialization, and being
a programmer first, my perspective on what a Topic is is based on heredity.
 That is, my working definition of Topic is as the ancestor of *all*
DITA document types, past, present and future.  The existing core
specializations (task, concept, reference) are fairly light reworkings
of Topic, but so much more is possible with specialization.  I don't
find it difficult to imagine a specialization which benefits from, or perhaps
relies on, the minimal content model of a topic with neither shortdesc,
abstract nor body.

Case in point: look at how Topic allows
non-sections to follow sections.  It's hard to justify this content
model in Topic by itself, but in the context of a Task, it's perfectly
natural to follow a prereq with steps.  Another example has already
been referred to, with the body-less glossentry specialization.

I'd hate to render some future specialization
impossible because of a failure to sufficiently imagine the diversity that
specialization can bring.

[Oh, BTW, Hi everybody.  I'm new
to OASIS but I've been kicking about on the dita-users and DITA-OT lists
for quite a while.  I'm a programmer-turned-writer at a software-for-the-plastics-industry
company, and I'm inflicting DITA onto my co-workers mercilessly.]

-- 

Deborah Pickett