- From: Dirk Herr-Hoyman <hoymand@gate.net>
- Date: Thu, 29 Dec 1994 20:11:56 -0500
- To: uri@bunyip.com
- Cc: "C. M. Sperberg-McQueen" <U35395@UICVM.UIC.EDU>
At 10:21 PM 12/29/94, Terry Allen wrote: >Larry writes: >>Just as HTML doesn't require all browsers to implement all of SGML, >but allows SGML-based tools to be used on HTML documents, we could use >SGML syntax, but require that a 'URC' be in some canonical form, to >facilitate easier handling by simpler tools. > >I say again, this won't work if the "canonical form" cannot be >expressed in a DTD and SGML declaration. "Using SGML syntax" means just >that, or it doesn't mean very much at all. Michael >Sperberg-McQueen has outlined the desirable SGML toggles for >achieving simplicity: > >http://www.ncsa.uiuc.edu/SDG/IT94/Proceedings/Autools/sperberg-mcqueen/sper >berg.html > >I strongly recommend it. > I would agree that if we are to use an SGML approach, that does mean we have a DTD. I think Larry's point is well taken, and does fit with the proposal of Michael Sperberg-McQueen, which calls for server-side validation of SGML. In the same manner, we could require that the URN resolution server only serve up valid URCs, which have been checked with a full-blown SGML parser at load time. I want to also add a side comment to the Sperberg-McQueen paper. A URC could be used to associate a DTD and style sheet with an SGML document, as opposed to doing this with HTTP. This would make his proposal protocol independent, or perhaps URN/URC resolution dependent. -- Dirk Herr-Hoyman <hoymand@gate.net> | I tried to contain myself CyberBeach Publishing | but * Internet publishing services | I got out Lake Worth, Florida, USA | Web: http://www.gate.net/cyberbeach.html Phone: +1.407.540.8309
Received on Thursday, 29 December 1994 20:11:16 UTC