Comments on RDF last call working drafts

Dear Colleagues:

JSR-188, the Java Specification Request for CC/PP processing,
http://www.jcp.org/en/jsr/detail?id=188
has examined the RDF last call working drafts published on the 23rd of
January 2003. First, we commend the RDF WG for excellent work, and
congratulate you on bringing your REC to last call. However we would like to
raise an issue with the documents, concerning the adoption of local
datatyping. Here we use the term "local datatyping" and "global datatyping"
as proposed by Mike Dean
http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Jan/0173.html

We agree with Mike's comments. Specifically the JSR-188 Expert Group would
like to suggest that CC/PP definitely need "global" datatyping not "local"
datatyping. As CC/PP is based on RDF, this means that RDF should provide
some mechanism for global data typing. Adopting local datatyping will make
CC/PP profiles unnecessarily verbose using up valuable network bandwidth. As
one early adopter of CC/PP is UAProf which is aimed at wireless phone
networks, profile verbosity is of a particular concern. In addition as Mike
notes local datatyping also increases the potential for inconsistencies,
which we found to be a considerable problem in the deployment of CC/PP and
UAProf.

Furthermore, we anticipate that local datatyping, which is clearly the wrong
choice for CC/PP, will also be the wrong choice for other applications of
RDF which require either
1. require RDF/XML to be entered by hand, due to the increased risk of
inconsistencies or 
2. for RDF/XML documents to be exchanged within protocols, due to increased
document verbosity. 

Therefore we would like to raise the issue that we think the RDF core
working group should reconsider its position on datatyping.

Mark Butler, Hewlett Packard
Luu Tran, Sun Microsystems
Andreas Schade, IBM
Jason Williams
Reto Hermann, IBM
Rotan Hanrahan
Stan Wiechers
Steve Geach, Elata plc

Received on Tuesday, 18 February 2003 09:49:53 UTC