NOTE: The galeon mailing list is no longer active. The list archives are made available for historical reasons.
Jon, Just a quick response to one question in your note. The JPEG2000/JPIP community is very actively engaged in WCS. In fact they are probably the most advanced in terms of formally proposing standard extensions, so they seem to be "believers" in that they are expending considerable effort to ensure that JPEG2000 and JPIP are among the office WCS standard extensions. That's why I had thought of them as a model for how we might address CF-netCDF and CF-OPeNDAP as complementary standard extensions. As to the GIS applications side of things, arcGIS already can deal directly CF-netCDF files that are on local disk. In their first release of WCS, they did not incorporate CF-netCDF but my understanding is that they are headed that direction. However, we have to make sure CF-netCDF is a standard extension in order for them to accomplish that. -- Ben On Wed, Sep 24, 2008 at 8:14 AM, Jon Blower <jdb@xxxxxxxxxxxxxxxxxxxx>wrote:
Hi Ben, The analogy with JPEG2000 and JPIP is a good one (although I'm not an expert either). My experience of applications that use JPIP streaming (e.g. accessing high-res satellite data from an Antarctic research ship with very low bandwidth) is that these applications depend upon the advanced features of the protocol to function correctly (e.g. to manage bandwidth). A WCS approach might effectively remove some of these features: a client doesn't know exactly how much data to expect in response to a request so it can't manage bandwidth. In other words, I see WCS as providing a common baseline, not as a protocol that attempts to support all of these features. Does anyone know how the JPEG2000/JPIP community feel about WCS?But my impression is that others think that we should abandon the CF-netCDF encoding spec. and ONLY be proposing CF-OPeNDAP. Is that the heart of the suggestion that's on the table in terms of the OPeNDAP part of the discussion?My personal opinion is that if I want to use the internet to access FES data losslessly, and my application understands the NetCDF data model and the CF conventions, I should use OPeNDAP as it gives me (very nearly) all the information I need and tools are readily available.* I see no need for WCS to satisfy this use case. If, on the other hand, I have a GIS application that has a map-based data model and understands GeoTIFFs, I might well want to access FES data through a WCS - but it's no good returning me a NetCDF file because I won't be able to understand it. Instead it would be much more useful to have a WCS that reads CF-NetCDF data and produces GeoTIFFs. A mapping of CF-NetCDF to WCS concepts is still necessary to understand how this mapping actually takes place - what information is preserved, what is lost and so forth. Best wishes, Jon
galeon archives: