Re: [dita] references to ditabases without an explicit topicid

[Robert D Anderson <robander@us.ibm.com>]

Hi Paul - maybe somebody else can give a more authoritative answer, but
from my experience...

The toolkit does not break files up when creating XHTML. So, if you
reference "file.dita", or "file.dita#first-topic", or
"file.dita#last-topic", the full file will generate XHTML. Once the spec
fully explains how to use the chunk attribute, I expect that the toolkit
will add support for that attribute as a way to break apart topics when
generating XHTML.

The main impact of that "file name equals first topic" assumption, at least
today, comes about when linking. If you reference "file.dita", then any
generated links are added after the body of the first topic. If there are
sub-topics, they appear below those links. Any generated links to this file
will pull the title from the first topic in the file. This is true whether
it's a <dita> mixed file, or of a single topic type. So, this demonstrates
the behavior that links to/from "file.dita" are treated as links to/from
the first topic.

If you give a topic ID in your reference, then links will be placed after
the body of that topic, regardless of what type of file it is or what else
is included. Generated links to this file will get the title from the
specific topic.

In terms of conref - my understanding is that conref="file.dita" is
supposed to go to the first topic in the file. In a <dita> mixed file, this
means the first child of <dita>. In a single-type file, this will bring in
that main topic and anything nested inside it. This clarification was made
after we wrote the conref code, and I haven't thought about it in a long
time, so I can't say for sure if the toolkit code was updated to work this
way. If not, we should open a bug report on it.

Does that help?

Robert D Anderson
IBM Authoring Tools Development
Chief Architect, DITA Open Toolkit
(507) 253-8787, T/L 553-8787

"Grosso, Paul" <pgrosso@ptc.com> wrote on 07/12/2006 02:29:41 PM:

> Our developers are trying to understand what it means
> to have a reference to a ditabase from a topicref in
> a map where the href value does not include a fragment
> identifier (e.g., "path" rather than "path#topicid").
> Specifically, is just the first topic picked up or
> all the topics?
>
> The DITA 1.0 Architectural Spec doesn't seem to
> address this.
>
> In the DITA 1.0 Language Specification, there are
> a few references that seem to address this issue:
>
> In the Language Spec. in the descriptions for the
> conref attribute on topic, concept, reference, and
> task elements it says in part:
>
>  During output processing, a lookup process will pull the
>  contents of the first topic into the calling topic markup
>  that has the conref attribute.
>
> And a bit later in the same description it says:
>
>     Target elsewhere in the same file:
>         conref="#topicid"
>
>     First topic in a different file:
>         conref="filename.xml"
>
>     Specific topic in different file:
>         conref="filename.xml#topicid"
>
> And in the href description for the link element in a
> topic, it says:
>
>    Target elsewhere in the same file:
>         conref="#topicid"
>
>     First topic in a different file:
>         conref="filename.dita"
>
>     Specific topic in different file:
>         conref="filename.dita#topicid"
>
> There is nothing specific about this one way or another in the
> other href descriptions for the elements topicref and xref.
>
> Using the DITA Open Toolkit, for html output, only one link
> appears in the index.html file, but all of the topics from
> the ditabase were included in the other .html file.
>
> Can someone confirm what is the expectation when one encounters
> an href to a ditabase without a fragment identifier on a topicref.
> Should just the first topic from the ditabase be included or
> should all of the topics from the ditabase be included?
>
> paul