Processing Model and ISO DSDL

I see in this week's TAG notes that there is discussion on processing models.

I hope there is scope for co-ordination with ISO DSDL on this.
Part 1 of that is to be a processing model specification language
of some kind; currently it says "based on RELAX Namespaces"
but I am also suggesting we look at XML Pipeline (which is
almost a superset of a little language of mine called Connect, so I think it is a useful approach ).   

I think this means that we define the "pipelines" for processing,
but the "pipelines" are selected according to namespaces (i.e.
as functions, not like UNIX Pipes processing a whole document
necessarily).  In other words, what XML Pipelines lacks is
a way to attach functions to namespaces or elements, and
RELAX Namespaces provides good ideas for that.

I don't think TAG needs to endorse DSDL, because DSDL really
is just attempting to put the ISO rubber stamp on existing initiatives
that have industry acceptence (e.g. more than one implementation): 
W3C XML Schemas, RELAX NG, Schematron, etc. in the super-DTD 
area (namespaces, validation,  augmentation, type allocation, 
inclusions, simple transformations)

The big gap in industry initiatives is with the processing model.
If there is nothing out there, then ISO will have to go ahead and
fill in the gaps itself (i.e. based on RELAX Namespaces and,
I hope, XML Pipeline.)

This raises the prospect of W3C then going ahead with its own
processing model, and the two efforts disapating or wasting our
time.  So it would be great if TAG or W3C could get some 
definite POV about processing models (say, before the next
ISO meeting in May) so that we ISO can react and enhance 
the W3C lead (if W3C wishes to take a lead in this). 

Rick Jelliffe  (writing unofficially)

Received on Thursday, 14 March 2002 08:38:06 UTC