Re: [dita] Issue: Map element IDs and references to them

[Michael Priestley <mpriestl@ca.ibm.com>]

OK, Robert talked some sense into me
off-line. 

I now get Eliot's point during the call
today that a link into the substructures of a conref'd section (for example)
is not reliable, if only because it requires that link resolution be done
as a second pass, after conref resolution, where many processes may be
resolving both conrefs and links at the same time as part of a resolve-references
pass.

I also get Jeff's point about the closest
target being preferable to the first target: for example, if I have an
xref to a list item, both in the same document, then I'd want it to continue
working even after I conref in something between them that introduces a
duplicate id.  So in this case, same document=closer.

That said, I still want our behaviors
to be predictable, ie the same across processors. But I don't want to make
a backwards-incompatible change either, if I can avoid it.

So how about:

- map documents, and individual topics,
SHOULD NOT contain duplicate ids on their elements (note should not, rather
than must not)
- conrefs that bring in an element with
an id that already exists in the conreffing context SHOULD change the id
of the element being brought in, to avoid creating a collision (again note
should not rather than must not)

That should give a rule similar to what
Jeff described in the call today, and makes it recommended but not required.

Michael Priestley, Senior Technical
Staff Member (STSM)
Lead IBM DITA Architect 
mpriestl@ca.ibm.com
http://dita.xml.org/blog/25

Eliot Kimber <ekimber@reallysi.com>

07/06/2009 09:27 AM

To	dita <dita@lists.oasis-open.org>
cc
Subject	[dita] Issue: Map element IDs and references to them

There appears to be serious inconsistency between
what at least I understand
our decisions about addressing elements within maps to be and what the
arch
spec says. In addition, the arch spec as currently drafted is inconsistent
on this matter.

In particular, we have established that the fragment identifier for elements
within maps is simply the @id attribute value, e.g. "#sometopicref".

However, the draft arch spec says this under "Map IDs and element
IDs within
a map":

"The id attributes for other elements in map are not of type ID and
are not
required to be unique."

If this statement is true then a fragment identifier consisting of just
the
element ID is not sufficient to enable reliable addressing of elements
within maps.

So something has to give. I see the following possible solutions:

A. Define a rule for resolving ambiguous references, e.g. "first occurrence
in document order". This probably reflects current behavior of most
implementations.

B. Require element IDs to be unique within map documents. Note that because
of shared elements between topics and maps, it's not possible to declare
the
ID attribute for most elements to be of type ID, so this requirement has
to
be validated by processors.

C. Make topicref IDs XML IDs and scope all other element IDs to the nearest
ancestor with a specified @id attribute (or the map element, whichever
is
nearer). Allow two-part fragment identifiers. Single-part fragment
identifiers address the first occurrence in document order.

Option (A) is the simplest to implement but the least complete. Option
C is
the most complete but changes current processing and address resolution
behavior.

As for use cases, references to topicrefs is the primary use case for
pointing to elements within maps, but certainly the current spec doesn't
disallow other references and there could be reasons to, e.g., data-about,
conref from "resource" maps, etc.
 
Cheers,

Eliot
----
Eliot Kimber | Senior Solutions Architect | Really Strategies, Inc.
email:  ekimber@reallysi.com <mailto:ekimber@reallysi.com>
office: 610.631.6770 | cell: 512.554.9368
2570 Boulevard of the Generals | Suite 213 | Audubon, PA 19403
www.reallysi.com <http://www.reallysi.com>  | http://blog.reallysi.com
<http://blog.reallysi.com> | www.rsuitecms.com <http://www.rsuitecms.com>


---------------------------------------------------------------------
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