- From: Andreas Schade <san@zurich.ibm.com>
- Date: Wed, 19 Jun 2002 16:51:12 +0200
- To: "Butler, Mark" <Mark_Butler@hplb.hpl.hp.com>
- Cc: "'www-mobile@w3.org'" <www-mobile@w3.org>, www-mobile-request@w3.org
Mark, Here's is our list of proposed CC/PP issues and work items: External profile representation We need to clearly define the syntax for an externalized CC/PP profile to simplify parsing and to avoid proliferation of different CC/PP "flavors". Principally, we see the following possible approaches: - Replace RDF with simple XML. We envisage two DTDs, one for the schema definition and a another one for actual profile instances. This is our favorite approach as it would make things much easier and solve machine-readability issue of schema definitions (see 2). - Use RDF but define *one* valid RDF externalization which can be parsed with reasonable efficiency. - Use RDF but permit several syntactic forms. This must be done with care. Machine-readability of CC/PP schema definitions Schema definitions must be machine-readable thus enabling the development of schema-transparent CC/PP processing software that can cope with schema evolution. Making the schema definition machine-readable means that a particular profile schema can be loaded at runtime and does not have to be hardcoded. Property type system Typed properties simplify the programmatic use of CC/PP profiles. Some types imply default relations between property values that can be used when comparing profiles (for instance, the relational operators "<", "<=", "==", ">=", ">" for values of integer type). We need to define an extensible type system for properties and mandate the use of typed properties. As it may not be sufficient to define a set of base property types (boolean, literal, number etc) and a set of composition specifiers (bag, sequence, etc), we may need some sort of extension mechanism. XML schema would be a candidate for this. The machine-readability of the schema definition must be preserved whatever typing system and extension mechanism is adopted. The question what values are valid for a given property is related to the property type system. A schema definition must also enable the definition of valid values for any property it contains. This can be done implicitly via the property type or more specifically by explicitly defining the set of valid values, e.g., for literal enumeration types. Profile validation We need to specify when a CC/PP profile is valid vis-a-vis a CC/PP schema. The schema represents a sort of contract between the sender and the receiver of the profile information as it indicates - for the sender what information can be sent in what form - for the receiver what questions it may ask, i.e. what properties it may query and what value(s) it may expect in response. If profiles do not adhere to the schema they use, a receiving entity can only guess what properties are contained in the profile, what types are used to represent the property values, and what these values actually mean. In this case the profile does not provide much value for the receiver. Validation is in our opinion the only way to guarantee that the contract between the sender and the recipient is upheld. On the other hand, strictly requiring that a profile can only contain the properties defined in the schema it uses actually breaks profile extensibility. A way to deal with this is to define different levels of strictness for profile validation. (see my earlier note on validation on the mailing list). Transfer syntax Currently there is only a proposed syntax based on HTTPExt. Given the limited support of HTTPExt in current web servers and proxies we have to standardize an HTTP-compliant transfer syntax for transporting CC/PP profiles. Core vocabulary In order to encourage a wider deployment of CC/PP we have to define a core vocabulary (core profile schema). This would enable separate development efforts for both sending and receiving entities of CC/PP information as both parties would know what data they can sent in what format/what data they have to expect. Profile-diffs Adopt the useful concept of the UAProf profile-diffs that can be applied as deltas to base profiles. Explicitly specify how profile-diffs are to be applied to base profiles (resolution policies) and when this can be done (involves work of schema compatibility). Kind regards, Andreas Schade IBM Zurich Research Lab |---------+-----------------------------> | | "Butler, Mark" | | | <Mark_Butler@hplb.| | | hpl.hp.com> | | | Sent by: | | | www-mobile-request| | | @w3.org | | | | | | | | | 13.06.2002 18:02 | | | Please respond to | | | "Butler, Mark" | | | | |---------+-----------------------------> >---------------------------------------------------------------------------------------------------------------------------------------------| | | | To: "'www-mobile@w3.org'" <www-mobile@w3.org> | | cc: | | Subject: Issues and future work for CC/PP | >---------------------------------------------------------------------------------------------------------------------------------------------| Hi all A while back I said that it is not possible to submit issues to the CC/PP Working Group until we get to the next stage in the process (Candidate Recommendation). After some discussion with the W3C this week, my position has changed on this. If you have any issues with CC/PP, or you have any suggestions for future work, then please submit them to me via this email list. It may not be possible to reply to these issues until the next stage in the process, but there is no reason why we can't collect them now. I'll take responsibility for collecting and coordinating these issues and requests. Note this is a formal process so you need to make it clear that you want me to respond to the posting in that way. Furthermore it would be helpful if you could structure your issue / suggestion in the following way: - A summary sentence that describes the issue / suggestion. - A paragraph that gives more details Also note I'm not planning to go back through www-mobile looking for issues, so if you feel something needs to be addressed you need to post it again or refer me to the post you made. Hope that is okay! PS just in case people haven't seen it already, you may be interested in JSR-188 on CC/PP - see http://jcp.org/jsr/detail/188.jsp best regards Mark H. Butler, PhD W3C CC/PP Working Group Chair Research Scientist HP Labs Bristol mark-h_butler@hp.com Internet: http://www-uk.hpl.hp.com/people/marbut/
Received on Wednesday, 19 June 2002 10:53:19 UTC