[
Lists Home |
Date Index |
Thread Index
]
- To: <xml-dev@lists.xml.org>
- Subject: Design of streamform declaration.
- From: "Chris Wilper" <cwilper@cs.cornell.edu>
- Date: Fri, 25 Oct 2002 11:26:55 -0400
- Thread-index: AcJ8OvORiE9ekcbjSBCN+z0L4nrxzA==
- Thread-topic: Design of streamform declaration.
Hi all,
Thanks to those that responded to my previous thread
http://lists.xml.org/archives/xml-dev/200210/msg00717.html
concerning identifying document types.
I am working on a prototype[1] that demonstrates the
utility of using non-schema-specific identifiers for
for streamforms[2] (what is commonly imagined as
a "document type" or "file format")
The closest I have seen to this being done is
in sgml arch. forms (architecture use declaration PI),
but that carries its own (rather large) baggage.
So my first thought was, if I want to declare a
"streamform", it'd be nice if it was an attribute
on the root node. Then I thought... Hmm, it's not
too realistic to change the schema of a document
just to be able to declare its streamform.
So I switched to a processing instruction.
My question: Is a PI the most natural (elegant) place
for this (internal) declaration of streamform? Are
there even any other realistic options?
[1]
I'll post the URL when I have a decent demo online.
[2]
To avoid confusion and semantic disputes, I'm using the
term "streamform" to mean: An identifier for a set
that an arbitrary piece of data belongs to or doesn't
depending on its physical data structure. Structure
may be defined any number of ways (formal grammar
(adaptive or not) or some set of rules that constitute
a structural "filter"), but the structure need not be
defined 1) formally, nor 2) via one grammar/rules
language in order to exist.
___________________________________________
Chris Wilper
Cornell Digital Library Research Group
http://www.cs.cornell.edu/~cwilper/
|