Yes, that's a good way of looking at it, Erik.
Now that you mention it, DITA abounds in this kind of "contradiction."
I see no particular reason to eliminate it in the case of ditabase vs
concept et al.
--Dana
Erik Hennum wrote:
Hi, Dana, Eliot, and Michael:
I'm wondering if the discussion underscores the distinction between a
vocabulary module and a document type shell.
The vocabulary module gives the normative definition of a content model
including the base element at each context. For instance, the
definition of <ph> in the topic vocabulary module allows
<keyword> as part of a choice.
The document type shell gives the normative definition for each content
model within that document type after domain substitions and
restrictions have been applied to those contexts.
Less formally, the vocabulary module provides the ingredients. The
document type shell assembles the ingredients as a meal.
For an example of domain substitions, the topic document type draws on
the programming, software, and UI domains to add the following
substitions in all <keyword> contexts: <option>,
<parmname>, <apiname>, <cmdname>, <msgnum>,
<varname>, and <wintitle>. So, the normative definition of
the <ph> content model for the topic document type provided with
the standard differs from the normative definition of the <ph>
content model for the topic vocabulary module. The two aren't in
conflict because of the nature of domain substitution groups. Other
conformant document type shells are possible that use the topic
vocabulary module but provide a different set of domain substitutions
for the <keyword> element and thus change the definition of the
<ph> content model.
For an example of restriction, the concept vocabulary module allows a
nested <topic> in the last context within the content model for
the <concept> element. The concept document type shell restricts
that context to the <concept> specialization of <topic>.
Other conformant document type shells are possible that allow the base
<topic> element in that context or that restrict the same context
to a choice of <concept> or <task> elements.
In DITA 1.0, a document type shell can only restrict <topic>
contexts. There is a DITA 1.2 requirement to extend that restriction
capability to any context (which would allow you, for instance, to
restrict the <keyword> context within <ph> to
<apiname> or <cmdname>).
Hoping that's useful,
Erik Hennum
ehennum@us.ibm.com
Michael
Priestley <mpriestl@ca.ibm.com>
>I think you're misunderstanding the implications of the content
models
>as shipped: they don't say you *must* do (as in "must allow all
topic
>types to nest"), they say what you *can* do, from the point of what
is
>allowed *for specializations*.
The nesting of different topic types is not governed through
specialization: you can change nesting rules without changing
specializations.
Michael Priestley
IBM DITA Architect and Classification Schema PDT Lead
mpriestl@ca.ibm.com
http://dita.xml.org/blog/25
Michael Priestley wrote:
> If we want to ditch normative nesting for one, we should ditch it
for
> both. Otherwise I don't see how we can say concept must allow
nesting of
> task in ditabase.dtd, and that's normative, but it can disallow it
in
> concept.dtd, and that's not.
I think you're misunderstanding the implications of the content models
as shipped: they don't say you *must* do (as in "must allow all topic
types to nest"), they say what you *can* do, from the point of what is
allowed *for specializations*.
That is, the existence of the value for info-types as declared in the
ditabase.dtd says "it is, as far as the standard is concerned, OK to
nest any type within any other type". But it does not say "you must
allow any type to nest within any other type".
Note that using ditabase.dtd as my specialization or configuration
base,
I can implement exactly the same constraints that the task-specific
shell DTDs impose.
Or said another way, the purpose of the normative part of the
specification is to define the *minimal set of required constraints*
needed to make all possible DITA elements sensible and interchangable.
Anything else we might choose to provide that is more constraining can
only be a non-normative example or opinion about best practice.
Cheers,
E.
--
W. Eliot Kimber
Professional Services
Innodata Isogen
8500 N. Mopac, Suite 402
Austin, TX 78759
(214) 954-5198
ekimber@innodata-isogen.com
www.innodata-isogen.com