Re: namespace viz validation [was RE: DSIGS]

It appears to me, that the discussion is actually somewhat abstract.

The practical problem to be solved with namespaces is IMHO the
following:

   1. There is a need to reuse fragments of DTDs
   2. There is a need to reuse fragments of instances - at least be able
to 
      simply cut & paste such kind of fragments.

   As long as we have WF documents, item #1 is obviously not a problem.
   Item #2 is no problem as long as entire subtrees are shared. 
   The namespace can be determined by the GI of one of the ancestors
(which delimits
   the namespace). This can be done using most of the programming resp.
style languages.

   The problem comes with valid documents! 
   Item #1 requires to intermix the contentmodels
   without ambiguity (and without ANY :-). With the current spec, we get
a sum
   over all content models (if we do not want to qualify the GIs of all
tags). 
   But what namespaces shoud switch content models and not mess them up
esp.
   if we want to use a structured authoring tool.

   Reusing instance fragments should be possible with out renaming all
elements,
   so a GI mangling approach isnīt helpful.



After all that, we should look for a way to define context dependant
content models.
But I donīt see how this can be specified without changing DTD-Syntax.
So its nothing
for July 1st.

Prefixing element names is a common practice in my daily work to 
distinguish varioius namespaces and to allow to share architectures.
We have a minimal set of prefixes. So I do not see an urgent need
for a namspace separation character (but i donīt oppose either).
>

Regards/Mit freundlichen Gruessen
===================================================================
Bernhard Weichel              Phone:  (49) 711 811 8322
Robert Bosch GmbH               Fax:  (49) 711 811 8262
Dept. K3/EES4                 eMail:  bernhard.weichel@pcm.bosch.de
P.O. Box 30 02 40                     
D-70442 Stuttgart
Germany

Received on Thursday, 19 June 1997 07:38:15 UTC