[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Subject: RE: [regrep] V3 Proposal: Namespace Manager Function
Message text written by "CHIUSANO, Joseph" >The Namespace Manager proposal calls for the first 2 elements above to be associated within the registry with the http://www.acme.org/invoices namespacem and the 3rd element above to be associated with the http://www.acme.org/temp namespace. This would allow Acme to (for instance) determine at a review time all of the constructs associated with the http://www.acme.org/temp namespace (i.e. all constructs that should be reviewed). This is just one simple case where I believe this functionality could be useful. I am hearing clients express a need for this type of functionality more often every day. <<<<<<<<<<< Joe, The more I think about this - the more I think this is the tail wagging the dog. In the eBTWG work on core components I'm being gently admonished for not being the UML zealot and using the modelling tool to drive down into the content generation [ well that's what some people think anyway ; -) ]. Now I'm seeing here we're looking at a situation where the information "model" is hidden in these namespace linkages in the schemas, and not even the programmers can "see" the wood for the trees. This was the classic problem with XSD the way Microsoft allowed - as a proprietary extension - inline declarations of namespaces within the XML content and then using them like includes. Result = you can look at the XSD all you like - but you have no way of knowing what the actual content is allowed until someone sends you a sample message. DTD's may be dumb - but at least there are no hidden surprises - looking at the DTD allows you to exactly know the content model. Now - it seems to me that building around the Classification system in the Registry is the proper way to go here. It avoids hard wiring this to any one particular schema flavour de jour - and provides a very nice potential tie in to the UML and CCR/ CRI along the way. Notice - if someone wants to write a Perl or Python script to run thru Schema, or RELAX or DTD's or EDI messages, and trawl out the heirarchy and generate nice Classification tree constructs from that - that's a cool tool. But we should not have to specify all that in our spec's. We simply allow for a link between the logical classification node and the physical content path. I believe we have that already! And then the "Search on Path content" filtered query can also be easily vendor implemented using the underpinning - as a menu option. We should resist the urge to spawn another flavour of Lego brick everytime someone has a use case we've not encountered before. Making people exploit the Lego bricks we have already is their own value add. Thanks, DW.
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Powered by eList eXpress LLC