I certainly agree that a loose structure has a practical
value for some activities, but it is being seriously corrupted by people who
have convinced others that DITA "does not work." They use ditabase to turn DITA
into a sequential book structure. They eliminate DITA map because it is
inconvenient and means less money for them to maintain the mess they've created.
As Yas suggests, it might be quickly deprecated to allow
those who want intertwined bookish structures to move to
Docbook.
JoAnn
Implementors
can already constrain users to
single typed topics by changing the doctype to use the appropriate concept/task/concept/reference DTD.
In terms of conversion, there could be cases where legacy
conversion to DITA would result in a single topic with mixed
information type sub-topics.
In terms of specialization, ditabase is
intentionally lax to serve as an appropriate base for specialization.
I agree with JoAnn that authoring with ditabase might not
follow the best practice but I don't see the justification for eliminating it
for DITA 1.1.
Lets consider leaving ditabase loose and encourage
implementors to use the single-typed DTDs for the appropriate information
type. Providing a ditabase DTD for composite documents might be an option
that some users, conversion vendors and specializers might want to
keep.
Finally, this feels like a large change given the
timeline for DITA 1.1's release. If the rest of the TC members feel
strongly that it should be eliminated, I'd hope that we'd start with
deprecation rather than elimination.
Yas Etessam
This is my position exactly. Let's eliminate ditabase.
I'm attending a meeting this week battling with a "consultant" who has used
ditabase to recreate docbook inside a dita topic.The entire book is in one
topic. It's infuriating that people want to corrupt the entire
concept.
JoAnn
Then the most logical choice would be to
eliminate ditabase from the standard - and let implementors do their own
ditabases as a practical measure, if they want to give authors a way of
writing non-conformant topic collections prior to splitting them up into
conformant topics.
--Dana
W. Eliot Kimber wrote:
Dana
Spradley wrote:
This would be a rather extreme change of policy,
wouldn't it?
As I understand it, ditabase is expliticly
*non*-normative, and as the spec currently says any nesting or other
arrangement of topics in it "has no particular output implications; it
simply allows you to create multiple topics of different types at the same
level in a single document."
Unless I've completely
misunderstood the implications of how things are delivered, all the
declarations are normative. That is, the DITA standard consists of the
architecture specification, the language reference, and the accompanying DTD
and XSD declarations, all of which are normative.
That is, the very
fact that we need to have language in the language reference about when
different containment rules apply indicates that we have two different but
normative rules.
If it wasn't normative then we wouldn't have the
language in the spec.
Cheers,
E.