- From: David G. Durand <dgd@cs.bu.edu>
- Date: Fri, 13 Jun 1997 10:42:34 -0500
- To: w3c-sgml-wg@w3.org
When I started to write this I had the simple idea it would not be too harmful (though I don't see the utility as large) to add colon to namestart, for people to use today, and reserving double colon for future expansion, once we figure out what namespaces are supposed to be. But, after reading the very nice metadata proposal, I'm not convinced that we really solve much that way. The problem is that the same element and the same data in that element may conform to _several_ namespaces. In fact, to the extent that my quick reading of the proposal makes sense, it actually allows the creation of semantic structures like that in its own context. So attaching one identifier at the beginning doesn't solve the problem without additional machinery. The only problem that I can clearly understand from the namespace discussion so far is one of attaching external meaning(s) to markup in context. I still think that attributes already exist, and are the standard way to do this. The problem is that there isn't a universal namespace for attaching such attributes to elements. Maybe we need a PI to attach an FPI or URI (whether of a human readable spec, or a processor to be let vague in exactly the way that it is in NOTATION) to an attribute name. Then that attribute name's presence on an element means that the semantics associated with its value (as described in or implemented by the FPI or URI) applies to that element. In practice, we might be better served by just picking good attribute names, and letting the marked sort out conflicts, but the PI would at least serve a real need. Given that the URI identification would probably work pretty much the way NOTATIONs do, there is some case for using a notation declaration... So my proposal would lead to declarations like: <!NOTATION Netscape-wizardry PUBLIC "//NETSCAPE::Department of Meta-Standards//NOTATION Meta Description of Meta Data Format//EN" "http://defs.netscape.com/meta/meta/desciption/meta/DMS-MDMDF.class"> <?EXTERNAL-ARCH MDMDF Netscape-wizardry?> These would allow applications that care about architectures (and know how to deal with the MDMDF architecture, to interpret MDMDF attributes that occur in the document instance. I'm not sure that this is necessary -- we could just assign an attribute name like com-netscape-MDMDF and be done -- but it seems to meet the namespace requirements that I understand in a nicer way. Note that it also uses the theory of URN/FPI and the practice of Architecture attributes, so that we have _some tenuous_ theroretical/practical grip on it. -- David _________________________________________ David Durand dgd@cs.bu.edu \ david@dynamicDiagrams.com Boston University Computer Science \ Sr. Analyst http://www.cs.bu.edu/students/grads/dgd/ \ Dynamic Diagrams --------------------------------------------\ http://dynamicDiagrams.com/ MAPA: mapping for the WWW \__________________________
Received on Friday, 13 June 1997 11:02:52 UTC