Re: [dita] link to ditamap from a topic

[Jim Tivy <jimt@bluestream.com>]

Hi Eliot

Thanks for describing this - apologies for the slow reply...

I should have been more precise in my question.  But I went to the
DITA spec and there was no statement on map links that I could find.
<xref href="some-map.ditamap" scope="{any}" format="ditamap"/> - so I
asked the more general question.

Our customer wants to link to other peer root maps within a book.  So
perhaps in the operators manual they want to link to the configuration
guide.
<xref href="ConfigGuide.ditamap" scope="peer" format="ditamap">Please
see the configuration guide.</xref>

I think the implied meaning they want is that this is a reference to
the entire set of topics in sequence - the book.

Does that make sense and should it be allowed?

Note: They also want cross book links to specific topics but I think
that is defined in the cross book links work in 1.3 using keyscopes so
is not my question here.

cheers
Jim


On Wed, Feb 19, 2020 at 6:37 AM Eliot Kimber <ekimber@contrext.com> wrote:
>
> When you say "link to a ditamap" what exactly are you describing?
>
> An xref that refers directly to a map document with a scope of "local" seems non-sensical, i.e.:
>
> <xref href="some-map.ditamap" scope="local" format="ditamap"/>
>
> What could this possibly mean?
>
> First, the map is reference with no use context, so it's impossible to know which use of the map you might mean.
>
> Second, a map that is not a root map does not have a well-defined identity in the context of the map tree it participates in because after resolution of all map references, the existence of the root map is effectively gone, meaning that the map reference ends up being (effectively) replaced by the topicrefs the map contains [this is, I think, a flaw in the general DITA design in that the standard should provide some defined post-resolution representation for submaps but we haven't addressed this so far, but it is a practical requirement for tools like the OT, which uses a convention for topic groups as a way to remember the details of submaps during preprocessing.]
>
> The alternative would be to use a key reference to the map reference, i.e.:
>
> <xref keyref="some-map" scope="local"/>
>
> Where the map reference would be:
>
> <map>
>   ...
>   <mapref keys="some-map" href="some-map.ditamap"/>
>   ...
> </map>
>
> This makes the use context explicit but still has the problem that the meaning of a cross reference to a map is undefined.
>
> If maps were required to have at most one direct-child topicref you could say that the reference to the map was effectively a reference to that map's root topicref, but that's not a requirement--a map may have any number of direct-child topicrefs. In that case, what would a reference to a map imply? A reference to *all* the topicrefs? A reference to the first child topicref?
>
> I therefore have to conclude that a local-scope link from a topic to a map cannot be given any useful or reliable meaning.
>
> Cheers,
>
> E.
> --
> Eliot Kimber
> http://contrext.com
>
>
> ïOn 2/18/20, 5:43 PM, "Jim Tivy" <dita@lists.oasis-open.org on behalf of jimt@bluestream.com> wrote:
>
>     Hi Folks
>
>     We are discussing the matter of having topics link to ditamaps through
>     xrefs or link elements.
>
>     It seems the spec does not disallow a link to a ditamap.
>     (https://dita.xml.org/content/00340729-SD_3/SD00340023/)
>
>     It seems that cross deliverable links have some discussion in the 1.3
>     spec but that focus in linking to topics in a peer map.   What about
>     same deliverable links to a map?
>
>     1. Is it legal to link from a topic to a map for the same deliverable?
>
>     2. What would the possible semantics for a processor be for the same
>     deliverable - navigate to a table of contents?
>
>     3. Is this really just a bad idea that nobody should do - for example
>     would the open toolkit handle it.
>
>     Also, if something has been written on this let me know.
>
>     cheers
>
>     Jim
>
>
>
>     ---------------------------------------------------------------------
>     To unsubscribe from this mail list, you must leave the OASIS TC that
>     generates this mail.  Follow this link to all your TCs in OASIS at:
>     https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php
>
>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe from this mail list, you must leave the OASIS TC that
> generates this mail.  Follow this link to all your TCs in OASIS at:
> https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php
>