- From: William D. Lindsey <blindsey@bdmtech.com>
- Date: Fri, 20 Sep 1996 16:22:56 -0600
- To: Paul Prescod <papresco@calum.csclub.uwaterloo.ca>
- Cc: w3c-sgml-wg@w3.org
Paul Prescod writes: > As I understand 15.3, A > Conforming SGML System does not have to support short references. Is RS/RE > worth breaking compatibility with a majority of SGML systems? > I don't know, maybe. I don't know of a sufficiently comprehensive survey of SGML systems to even know for sure when we are or are not being compatible with a majority of them. So, my problem wasn't so much with your argument as it was in trying to figure out how to interpret point #1 of design principal three. "Existing SGML tools shall be able to read and write XML data". In the absence of that comprehensive survey of tools, I jumped to the conclusion that the best assurance of meeting that goal is to make sure that all XML documents conform to ISO 8879 as it stands today. That led me to the idea that is was OK to use any part of SGML that might offer solutions in meeting the other design principles. Now, if we want to define "existing SGML tools" as "any system meeting the requirements of section 15.3", I know what our boundaries are, and know that stupid net tricks and fixed shortrefs are out the door. Is this how we want to interpret design principle three? -Bill -- William D. Lindsey blindsey@bdmtech.com +1 (303) 672-8954
Received on Friday, 20 September 1996 18:24:07 UTC