- From: Ron Daniel <rdaniel@interwoven.com>
- Date: Wed, 6 Jun 2001 12:10:48 -0700
- To: "Art Barstow" <barstow@w3.org>, "Jan Grant" <Jan.Grant@bristol.ac.uk>
- Cc: "Brian McBride" <bwm@hplb.hpl.hp.com>, "RDFCore Working Group" <w3c-rdfcore-wg@w3.org>
Art Barstow said: [Brian McBride] > > > Someone, I think it might have been Graham, suggested recently that > > > attributes in the xml namespace should not generate properties. > > > > > > Any thoughts? > > [Jan Grant] > > This looks like it might be a good idea; it occurred to me too and I > > remember someone mentioning it, but couldn't find the message when I > > went looking for it. [Art Barstow] > Why would it be a good idea to not generate triples > for these attributes? It seems like the language should > have as few execeptions as possible. I worried about that too. But there are some current facts that we can't ignore: 0) If you attempt to simply apply RDF rules to the current xml:base and xml:lang, you get nonsense. 1) It is quite possible that future working groups will discover general needs, such as setting base URIs, and propose solutions to those problems. We can't anticipate how many of those will happen in the future. "Not many, but probably some" is all I would be willing to venture. 2) It is easy for those groups to propose those solutions in arbitrary XML, and to specify rules like "applies to all contained content" (which both xml:base and xml:lang do). 3) All of the above operates in the XML syntax domain. Defining a property in RDF whose meaning is "applies to all contained content in XML serialization" is ... non-trivial. Setting a requirement on all future groups that any new entries in the xml namespace must fit the current RDF syntax is a non-starter - they are doing XML, not RDF. So, It seems to me that things making it into the xml namespace are going to be considered special, and of enough importance that all XML processors going forward are expected to adopt them. RDF's syntax convention is there so parsers can do something reasonable with stuff they know nothing about. The special case rules need to take precedence over the general case. Change the 'attributes in the xml namespace shall not generate triples' to 'unknown attributes...' and I'm OK with it. Regards, Ron Daniel Jr. Standards Architect Tel: +1 415 778 3113 Fax: +1 415 778 3131 Email: rdaniel@interwoven.com Visit www.interwoven.com Moving Business to the Web In fact, this is my biggest concern about RDF, based on my experiences so far. It is VERY hard to hack in little extensions in a way that makes sense according to RDF's model. XML is pretty easy, the RDF syntax conventions are not obvious to the average (or even above-average) production programmer. They see a need, they hack a new element or attribute into the file format, and they are done. Too bad it breaks somebody else's stuff later down the line.
Received on Wednesday, 6 June 2001 15:13:23 UTC