Meeting Minutes 13 April 2006

Meeting Minutes 13th April 2006
Present: David (Andre) Nickerson, Sarala Dissanayake, Peter Villiger, Carey Stevens, Shane Blackett, Matt Halstead, Poul Nielsen, Andrew Miller
Apologies: Dong Zhang

Poul: Noted that there had been extensive discussion about the CellML media type registration on the cellml-discussion list.
Andrew: Elaborated on the issues raised on the ietf-types@ list (forwarded to cellml-discussion@) related to the proposal to register new media types for every version.
Matt: Is in favour of a single media type, because he believes that we will never store all information needed to determine whether a model can be processed in the media type(as there is more to it than just the version).
Shane: Agrees with the issue raised by Matt on the list that if we have a different media type for different versions, we should have a different extension, and agrees that we should have a single type.
Matt: Believes that a WSDL-like service for content-negotiation and dynamic translation would be better than a single service.
Shane: Believes that breaking backwards compatibility is a good thing, as it encourages tool vendors to upgrade.

After extensive discussion, the group reached the consensus that we should register a single media type for an umbrella format, and from that format, it should be possible to determine the version, but nothing else. Each CellML version released will comply with the umbrella format.

Andrew: Asked whether we will develop a new binding("XCI") which allows XPCOM to call directly to the CellML API, without going through CORBA.
Matt: Believes it would be a good idea.
Andrew: Noted that it is purely for performance reasons, and there is very little change required to switch from the CORBA approach to the non-CORBA approach. As such it can be deferred.
Matt: Suggested that we try the CORBA approach, to allow it to be profiled, and then write the XCI if the CORBA approach proves too expensive.
Overall consensus was that the XCI will not be implemented straight away, but deferred for possible implementation in the future.