- From: Wendy A Chisholm <wendy@w3.org>
- Date: Wed, 30 May 2001 15:04:35 -0400
- To: "Sean B. Palmer" <sean@mysterylights.com>, <w3c-wai-er-ig@w3.org>
- Cc: dimitris.dimitriadis@improve.se
hello all, (Dimitris, I'm including you 1/2 way through the thread. It begins at [1]. We're trying to understand the DOM perspective re: EARL/metadata/XML. here's my response to Sean Palmer's latest post in the thread) >RDF itself isn't difficult at all to "learn", it's just that there is >a startling lack of tutorials, and a huge amount of assumed best >practices which are only encoded in streams of list archives. which makes it time consuming to learn. many people don't have time or patience. >Ooh, time for a bit of RDF cheerleading. rah! rah! siss boom bah! <grin/> >Why RDF and not XML? Why reinvent the wheel? With RDF, there is a >single standard data model (triples), fully decentralized, and with >numerous implementations, APIs and parsers. If you invent your own >language in pure XML, you are in essence also developing a data model. >There are numerous problems with this:- > >1) You are slowed down because you have to work out a data model - >just use RDFs (although you do have to learn it, but surely learning >is easier than creating). Depends on the person, no? If someone has invented several XML languages, then they could more easily do that than learn something new. Also, many people will continue using familiar tools rather than spending time learning the benefits of new tools. There are trade-offs and not all people will choose the same path. Some people just want to get things done and will use whatever tools are closest and that they know will work. Our job is to help with the transition since we believe we have a better tool. Basic marketing I suppose. but how many people do you know who still use an older version of a browser because they know it works or don't see the need to take the time to download the latest and greatest or perhaps use it for testing purposes or some other reason - perhaps it works better with other tools that they have? and they will continue to use it even though they know there are new features in the latest version but they might not see the need for those new features when they are comfortable with what they have. >2) Other people then have to learn your data model, and build tools to >process it. Again: why reinvent the wheel? >3) Interoperability with other languages is then next to impossible. >With RDF based languages, it is still difficult, but at least >possible. Isn't that the point of XML and namespaces, tho? e.g. embedding SVG into XHTML? >There are many more hints in TimBL's missive:- > > http://www.w3.org/DesignIssues/RDF-XML > >the best bit is:- > >[[[ >The expression you need for querying something in terms of the XML >tree is necessarily more complicated than the expression you need for >querying something in terms of the RDF tree. >]]] I don't know RDF well enough to fully appreciate the beauty of this statement, but I do know XSLT and SQL so I am trying to think of this in those terms. Looking at Tim's paper again, he is primarily discussing machine-readable info, right? At this point, I'm assuming people are still the primary readers of the content in some of the applications we're talking about. For example, in the DOM group, they know the schema, so that's not a "big if" for them. It is difficult to see past that. also, they have tools that help them write using their DTD (XED, for example). Are there the same types of tools for RDF? Where you can easily plug in a schema and author? I was amazed how easy it is to use a new DTD with XED. Therefore, perhaps this gets back to tools, and I again apologize for still getting up to speed on this stuff myself. The point I'm trying to make is that when someone is working closely with the semantics and they know how to express them comfortably in XML, then convincing them to learn a new language and think about things in a different way can be a difficult task and it seems that that might be the task at hand. I am only trying to look at this from another perspective. I do see the need for EARL to be machine-readable (and thus in RDF) and think that it would benefit the DOM test suites as well, but to give them a strong argument I am trying to see things from their perspective...some of this is conjecture, thus I CC'ed Dimitris on my reply. Daniel raised an interesting question in his post: What about integrating EARL into an existing framework. This seems to be what Curt and Dimitris had discussed [2], [3]. Curt describes three phases in which it might be possible to use EARL. He proposes keeping the current schema but adding a metadata element that can contain RDF info. His example uses Dublin Core info about an author - who created a test. It seems easy to then say, "who says that I pass the test" ala EARL. However, Dimitris proposes leaving the metadata element optional since developers will already have had to learn the NIST DTD and they don't want to further burden them. <wendy goes off to think some more..../> --w [1] http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2001May/0101.html [2] http://lists.w3.org/Archives/Public/www-dom-ts/2001May/0067.html [3] http://lists.w3.org/Archives/Public/www-dom-ts/2001May/0103.html -- wendy a chisholm world wide web consortium web accessibility initiative seattle, wa usa tel: +1 206.706.5263 /--
Received on Wednesday, 30 May 2001 14:59:12 UTC