Simone Giannecchini a écrit : > If you revise the draft document on metadata that I sent you a while > ago, you would see that we left space for nodata, offset and scale > values for each band.
Yes I remember and we agree on that. But based on previous emails and on a look to TransformCategory interface, I have the feeling that it was not clear that "sampleToGeophysics" (lets rename it "sampleToUnit" - I admit that the former was a bad name) is nothing else than scale, offset and nodata packaged in a function. So if we dissociate "sampleToUnit" from Category (so indirectly from SampleDimension), we have to dissociate scale, offset, nodata and units as well to remain consistent. > I am sugesting NOT to apply any automagic code to "guess" no data > values but to allow users to EXPLICITLY specify them. I don't remember automagic code to guess no data value. They are explicitly specified since every Catagory without "sampleToUnit" (lets said "every range of sample values without scale and offset") is "no data" from the "geophysics view" perspective. > Once no data values, scale and offset have been specified we > should try hard to PRESERVE them This is what current implementation does as far as I can see... > I think we should separate the management of the so-called > sampleToGeoPhysics and of the no data values from the portrayal of the > data (you mentioned more than once separation of concerns if I recall) Yes, but keep in mind the relationship between concepts. The proposed TransformCategory wipe out the relationship between "sampleToUnit" and the SampleDimension attributes (unit, scale...). If your proposal is dedicaced to portrayal only, and if portrayal doesn't need "sampleToUnit", then it should not appears at all. All those TransformCategory interfaces should disaspear. Thats my understanding of separation of concerns. So what I means by "remove Category interfaces" is: keep only what is needed for portrayal, nothing more (otherwise it looks like a proposed replacement for org.geotools.coverage.Category). Martin ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Geotools-devel mailing list Geotools-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/geotools-devel