From: Arjun Ray Newsgroups: comp.text.sgml Subject: Data attributes: Extending the basic syntax Date: Mon, 27 Nov 2000 02:42:02 -0500 Organization: FUDGE Dispersal Systems Message-ID: Two use cases to think about: 1. Before Annex K, only in an external entity declaration could one associate both a notation and a set of data attributes: Now, we can also use this mechanism (notation name + data attribute list) to qualify the value of an attribute (K.4.4.2-3): But we still don't have a way to do this for the data content of an element. The basic problem appears to be syntactic: how to tell the data attributes from the regular attributes, and perhaps, how to fit such information into an ATTLIST declaration. 2. There is no clean way to invoke architectural processing. Never mind the ArcBase or ArchUse PIs to get everything going - we probably need something to that declarative effect in any case - there's also the problem of architectural support attributes looking more than a little factitious when I do something like this: After all, 'ArcFormA' and 'ArcNamrA' and their kin come from the AFDR, but somehow everyone (me, my software) has to "know" this. What I'd like to do is to associate those attributes with the AFDR *notation*, and then use data attribute mechanism to associate an instantiation AFDR [ArcFormA="html" ArcNamrA="htmln"] with something that identifies the "html notation". The general point is that the syntactic form of a name plus [attspecs] could also be profitably applied in other places. A. In an element declaration: which would be equivalent to a (#PCDATA) content model, except that we also tie an attributed data content notation at the same time. This is exactly the same feature K.4.4.2-3 is now giving us for attribute values only. B. In a notation declaration: C. In a start-tag: This partially solves the current problem that we can only specify a notation name through an attribute with NOTATION declared value, but the syntax also allows a wider interpretation. Since the form is clearly distinguishable from regular attribute specifications (even when minimized), it could refer to any declared notation for purposes defined by the application. Thoughts? -- :ar