Meeting Summary - 30 July 2007
[Meeting Summary - 30 July 2007] F2F Review. Confirmed publications. Outstanding issues re normative names. Renaming Requirements. Target languages. Opaque context key. Problem with ANY. XPath/JSON?
[Review] This was the first conference call of the DDWG since the London face-to-face meeting. The group considered the summary that was intended for publication this week. It was time to review the decisions made, and issues uncovered, during that meeting.
[Publications] Four new publications were noted. The first two will be the final publications of Landscape and Ecosystem. It is expected that this will occur sometime in September after the formatting and proofreading is complete. The first draft of the API specification will occur around the end of September. A new WG Note has been agreed, to cover the result of the research into the use of IDL.
[Names] It has been agreed that the normative names for properties will be located within the UWA's ontology. However, there still remains the issue of named concepts in the DDWG vocabulary that might not be identified with a single property in the ontology. For example, the vocabulary might contain the concept of "supported animation" as a named item. This would identify the list of animated image formats supported by a device (GIF 89a, MP4, etc.). However, this is an aggregate derived from an analysis of the complete list of supported image formats. The vocabulary item does not therefore have a one-to-one mapping to the ontology, although it references the ontology "under the hood". The ontology is intended to be a normative specification, while the DDWG Core Vocabulary is being presented via WG Note. It is unclear what status would be given to names that are defined in the vocabulary but not in the ontology.
[Rename] Given the change in focus of the Requirements document, it has been proposed that the group consider a new title that reflects the emphasis in this document: High-level Requirements for DDR implementations. In the meantime, the group will continue to use the generic "Requirements" title.
[Targets] The group has agreed that the IDL will be designed with several target languages in mind, although in theory any programming language should be supported. The target languages are: WSDL (normative), Java (possibly normative),
[Key] The context key is an opaque object that identifies sufficient information from the delivery context to enable a DDR instance to retrieve information about that context. A particular context key is likely to be a structured object, but the details of that structure are only known to a Discovery process and a corresponding DDR implementation. Context Keys should be unique, so that different delivery contexts will imply different Context Keys. The comparison of Context Keys is an operation that can only be performed by Discovery and/or DDR instances. It is necessary to represent the Context Key in the API. This will permit the key to pass from the Discovery process to the Repository as part of the request, in situations where these two are visible separately.
[ANY] The group considered how a Context Key should be described. A generic object type should suffice, although it would be given a specific name. In IDL, the use of "ANY" is suggested. However, the group noted that there could be problems mapping ANY in a WSDL binding and therefore agreed to do some more investigation.
Cosmin Basca (DERI)
José Manuel Cantera Fonseca (Telefónica)
Rotan Hanrahan (MobileAware)
Martin Jones (Volantis)
Rhys Lewis (Volantis)
Kevin Smith (Vodafone)
Mike Smith (W3C)
Andrea Trasatti (dotMobi)
Matt Womer (W3C)