- From: Michael Sperberg-McQueen <U35395@UICVM.UIC.EDU>
- Date: Tue, 01 Apr 97 06:42:47 CST
- To: W3C SGML Working Group <w3c-sgml-wg@w3.org>
On Tue, 1 Apr 1997 04:17:26 -0500 Martin Bryan said: >The problem I have is the one you seem to recognize yourself. If >someone adds a keyword to a catalog because he has facilities for >interpreting it, how can he be sure that someone without those >facilities can use it? ... By 'use it' do you mean 'use the new keyword', or 'use the catalog'? The draft wording says keywords other than public may but need not be supported by XML processors. I don't know what part of this is unclear. If you add a FREEMJET keyword to the SGML Open Catalog, the XML-lang spec does not promise you that my XML processor will be able to 'use' it in the sense of correctly doing whatever it is you want processors to do with a FREEMJET entry. If you want that promise, you need to get it on the basis of other agreements which are, perforce, outside the scope of the XML-lang spec. > ... Is it a requirement that the addition of >keywords to a catalog means that the catalog can only be used on >sites that support that keyword? No, it is not. The draft wording defines a catalog file (which ought probably to be called a catalog entry file, in line with Paul Grosso's more careful terminology) as containing keywords other than PUBLIC. That means an XML processor must be able to read past them, whether it chooses to support them or not. So if you add a FREEMJET keyword to the SGML Open Catalog, the XML-lang spec does promise you that my XML processor will be able to read past the FREEMJET entries, and therefore that even processors which don't support FREEMJET (or, for that matter, CATALOG, ENTITY, DELEGATE, or any of the other keywords defined by the full SGML Open 9401 spec) will be able to use the PUBLIC entries in your catalog file (as well as any other entry types the processor supports). If there is another way to interpret the draft wording (e.g. as meaning that an XML processor is required to handle the FREEMJET keyword, or that an XML processor is allowed to signal an error on the CATALOG keyword), I would be grateful for an explanation of that other interpretation. At the moment, the draft still seems unambiguous to me. -C. M. Sperberg-McQueen
Received on Tuesday, 1 April 1997 08:02:27 UTC