NOTE: The galeon mailing list is no longer active. The list archives are made available for historical reasons.
Hi all, The Coverages DWG, the WCS SWG, and the CF-netCDF SWG had several "lively" and hopefull productive discussions regarding how to incorporate coverage encodings (e.g., geoTIFF, JPEG2000, netCDF) into OGC protocol specs (not just WCS but also possibly WFS, SOS, WPS, ...). There appears to be general agreement that, as much as possible, these coverage encoding specifications (e.g., the encoding data model mappings to GMLCOV and the special parameters for each binary encoding) should be decoupled from the data access protocols. If you recall, we had been proposing the CF-netCDF specifically as an encoding for WCS 2.0. So far several ideas have been proposed for how to go about doing this but, as yet, there is no clear agreed-upon path. During these discussions, it was noted that the OGC Architecture Board is also considering ways to streamline and simplify some of the rather rigid requirements for how the specification documents are written. Some possible mechanisms are aiming for breaking specs into fewer modules, providing a cleaner and less distracting means for dealing with HTTP URI requirements of the OGC Naming Authority, and perhaps less emphasis and dependence on UML diagrams. Having spent a large fraction of my time writing and rewriting those portions of the existing CF-netCDF documents, I think these are moves in the right direction. On the other hand, we needed to get the remaining specs relating to CF-netCDF adopted by OGC. After a discussion with Stefano (who was not at the OGC meetings) last Friday, we came up with the following: An avenue that seems promising is to take the document that has been created as a WCS extension for CF-netCDF encoding and rewrite it as an encoding specification extension to OWS Common. The chapters that describe the mapping between CF-netCDF and GMLCOV would remain pretty much the same. However the requirements that pertain to the WCS (2 from chapter 6 and the remainder in chapter 7) would be rewritten to refer to OWS Common instead of WCS. The specification would thus have two conformance classes: one for the GMLCOV mapping and one for the OWS Common connection. Using OWS Common would ensure the encoding would be decoupled from any one OGC protocol so it could be used by any of them. I believe housing such specs that apply to all the protocols is one of the primary reasons for the existence of OWS Common. My understanding is that this specification could be proposed by any SWG so we might consider proposing it in the CF-netCDF SWG instead of WCS. On the other hand, for a case like geoTIFF where there is no OGC SWG, it could be proposed by the WCS SWG. The point is that A SWG proposes the specification the TC. From our discussion at last week's TC, I understood the conclusion to be that, if the TC votes it in as an OGC standard, it does not matter which SWG initiated it. Our struggle in the TC discussions was to determine a logical place for such specification documents to reside. OWS Common seems to be such a logical place. As usual, these OGC specification requirement issues are difficult to summarize concisely and clearly, but I hope this note provides an understanding of the fundamental questions and a possible way forward for CF-netCDF. -- Ben
galeon archives: