A unique ID question ?

[Len Bullard]

W. Eliot Kimber wrote:
> 
> This is why external parsed entities ("external text entities" in SGML
> parlance) are evil and should never have been included in XML.

Evil or inconvenient, we are stuck with them.  The same issues 
has come up in the DTD for X3D/VRML.  Well-formedness 
and namespaces don't exactly cohere as concepts in practice.

For that reason, there was a short consideration to declare them as
NMTOKEN.  
At least, they would weakly validate.  If we declare DEF/USE
combinations 
as ID/IDREF:

1. Cannot duplicate DEFs:  VRML97 allows this with undefined results.  
Essentially, in implementation, the last guy declared is used 
by the next guy.  Some authoring tools also flag an error for the author 
to clean up.

2.  Name collisions:  Inlines have to be scoped.  PROTOS and 
extenal PROTOS have to be scoped.  If one aggregates, one has 
to declare all of the scopes and validate them individually, 
or not use ID/IDREF.  It is probably better to support the 
semantics upstream of the broadcast products and keep the 
client thin as a fashion model.

> The only complete and managable solution to this problem is to manage
> each component as a complete and independent document and combine them
> together *semantically* using something like XLink or HyTime's value
> reference facility (explicit use-by-reference as distinct from
> hyperlinking).

Ensure that any translation or conversion tool uniquefies names 
where used as IDs.  

Whether done with Xlink or valuerefs, some means to explicitly 
spec name resolution is needed.  Given the time required to get 
changes through multiple language ecosystems, my intuition is 
this will be a problem for a long time. 

The authoring tools do a lot of work, and while that is not all bad, 
it should be something developers and customers are more aware of.  The
features of 
the tools to manage XML applications are more important to understand
than 
the specification because of the need to validate.  That brings up 
the subject of the number of unicorns required to approve a purchase. 
For the fresh faces, conformance tests for systems must be 
spec'd at several levels of the system and normatively declared 
in the contracts.  Remember, contract writers aren't usually 
deeply technical, so, take time to cover this subject.  
We had many similar issues with SGML procurements.

len



xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/ and on CD-ROM/ISBN 981-02-3594-1
To unsubscribe, mailto:majordomo at ic.ac.uk the following message;
unsubscribe xml-dev
To subscribe to the digests, mailto:majordomo at ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at ic.ac.uk)