Need to Relax Rule for Generating Chunk Names

[Eliot Kimber <ekimber@contrext.com>]

In DITA 1.2 and still in DITA 1.3 the rules for generating resource names
as a result of to-content chunk processing are (from the topic "Chunking"
in the Arch Spec):

[quote]
When creating new documents via chunk processing, the storage object name
or identifier (if relevant) is determined as follows:

1.If an entire map is used to generate a single chunk (by placing
to-content on the map element), the name is taken from the name of the map.

2. If the @copy-to attribute is specified, the name is taken from the
@copy-to attribute.

3. If @copy-to is not specified and the by-topic policy is in effect, the
name is taken from the @id attribute of the topic.

4. If @copy-to is not specified and the by-document policy is in effect,
the name is taken from the name of the referenced document.
[/quote]


In this set of rules, rule (3) is currently written as a MUST statement:
"the name *is taken* from the @id attribute"


This is too restrictive because it doesn't make any allowance for
processors to handle the case where two different chunks would have the
same resource identifier, e.g. "topicid.dita" because the topics happen to
have the same ID. (As it happens, the OT today handles this case by
generating a unique resource ID for the chunk, which is technically
non-conforming by the language quoted here.)

Robert and I are in agreement that this MUST should be a SHOULD and we
should add some language to clarify what processors are allowed to do,
something like:

"
If @copy-to is not specified and the by-topic policy is in effect, the
resource identifier SHOULD be taken from the @id attribute of the topic.
When two such resource identifiers would conflict processors MAY recover
by generating resource identifiers, for example, by combining the resource
identifier of the target topic's XML document and the topic's @id value.

"

The spec intent is to encourage consistency and predictability of the
generated resource identifiers, but the rule as stated in DITA 1.2 does
not satisfy that requirement without effectively imposing the requirement
to either use @copy-to in all cases where IDs are not reliably unique or
to modify your content to make topic IDs globally unique where they
otherwise would not need to be.

By contrast, a combination of the XML document filename and the topic ID
will always be unique within the same directory because you can't have two
XML documents with the same filename in the same directory and you can't
have two topics with the same ID in the same XML document.

Cheers,

E.

—————
Eliot Kimber, Owner
Contrext, LLC
http://contrext.com