- From: Michael Sperberg-McQueen <U35395@UICVM.UIC.EDU>
- Date: Fri, 23 May 97 16:49:34 CDT
- To: W3C SGML Working Group <w3c-sgml-wg@w3.org>
On Thu, 22 May 1997 16:18:59 -0400 (EDT) Eliot Kimber said:
>In the discussions I've seen (both here and elsewhere), there seem to be
>several distinct and largely independent problems:
>
>1. Connecting instance element types to known semantics
> ("architectures")
>2. Managing "DTD modules" in some semi-automatic way ("schema import")
>3. Enabling the literal inclusion of document fragments with different
> schemas in the same document ("inline subdocs")
In various discussions of what I mentally call "the name-space
problem" I've encountered other related problems, or variations
on these:
2a. schema import en bloc (inline SUBDOC): allow CALS tables to
appear in this document; once you're in a CALS table, only CALS
elements are legal.
2b. schema import with merger: allow CALS tables in this document, but
once you get down to the table cell, allow the same phrase-level
elements defined by the outer DTD. This allows identifier shadowing
within nested blocks, and is one of the main differences between the
Japanese proposal I read long ago (does anyone remember the WG8 document
number for it so people can go read it on the WG8 web site?) and Martin
Bryan's proposal, at least as far as I remember the former and
understand the latter.
4. Delimit the scope of parameter entities, IDs, (general entities?),
notations, etc. -- it's not quite clear how to fit this nicely into the
current structure of 8879; moving to something like the Algol or C block
structure might make it easier to solve.
It seems to me that something like Chris Maden's name-structure sorcery
may be our best bet in the short term. Something like the Japanese
proposal (perhaps extended to cover scoping of things other than GIs) is
the best bet for 8879 in the longer term.
-CMSMcQ
Received on Friday, 23 May 1997 18:16:56 UTC