- From: Fisher Mark <FisherM@is3.indy.tce.com>
- Date: Thu, 15 Aug 96 12:35:00 PDT
- To: www-html <www-html@w3.org>
Marc Salomon writes in <9608121129.ZM8807@gaia.ckm.ucsf.edu>: >Wait a sec. You said earlier that there was, for all intents and purposes, >slim difference in the task of processing structure expressed through GI's or >through attributes. If its the same difference in processing, then how can it >adversely affect the deployment of sounder solutions? I could see how it >might >be problematic if content models (that get lost in CLASS) are important, but >retaining and enforcing content models with arbitrary GI's seems even a >tougher >nut to crack than implementing CLASS. The problem arises from having to create _another_ structure-handling layer (a layer to handle CLASS) over SGML's structure-handling layer. Whereas if SGML handles the structure directly, we can make use of the experience and tools derived from SGML usage. The problem of arbitrary content models should be easier with arbitrary GI's, as at least we have SGML for that -- implementing CLASS as the main structure expression avenue would require creating yet another structure handler for an as-yet undefined language. ====================================================================== Mark Leighton Fisher Thomson Consumer Electronics fisherm@indy.tce.com Indianapolis, IN
Received on Thursday, 15 August 1996 20:50:37 UTC