What prevents mosaic"k"s from being created?

Question

I have created two base tables and two rdt tables. The base tables are named RASTER_CIP and MOSAIC. The rdt tables are RASTER_CIP_RDT and MOSAIC_RDT. The database is 11gR2.
The base tables have three attributes, NAME - VARCHAR2(64), SPATIAL_MBR SDO_GEOMETRY, IMAGE SDO_GEORASTER.
The RDT tables are named as standard.
Using FME I loaded three images into the RASTER_CIP/RDT tables with a SRID of 4326. Using the validategeoraster and schemavalidate functions, all three rows are valid. I can view the images with the FME viewer without problems. I can build (or delete) pyramid layers just fine. All three images were loaded, and then compressed with JPEG-B. They currently have no pyramids. The images do not overalap, and are not adjacent.
When I try to run the following mosaic operation (adopted from the georaster documentation), I get the following error. I've tried it with compressed (JPEG-B), not compressed, different block sizes, no options, etc. Same error.
What exactly does it not like?
-- Make MOSAIC
DECLARE
   gr   sdo_georaster;
BEGIN
   INSERT INTO   mosaic (name, image)
        VALUES   ('HIGHRES', SDO_GEOR.init ('MOSAIC_RDT'))
     RETURNING   image
          INTO   gr;
   SDO_GEOR.mosaic ('RASTER',
                    'IMAGE',
                    gr,
                    'blocksize=(512,512,3) compression=NONE interleaving=BSQ pyramid=FALSE',
                    SDO_NUMBER_ARRAY(0));
   UPDATE   mosaic
      SET   image   = gr
    WHERE   name = 'HIGHRES';
END;
/ORA-13419: cannot perform mosaick operation on the specified table column
ORA-06512: at "MDSYS.MD", line 1723
ORA-06512: at "MDSYS.MDERR", line 8
ORA-06512: at "MDSYS.SDO_GEOR_INT", line 963
ORA-06512: at "MDSYS.SDO_GEOR_INT", line 1268
ORA-06512: at "MDSYS.SDO_GEOR_INT", line 1539
ORA-06512: at "MDSYS.SDO_GEOR", line 1401
ORA-06512: at line 11

Answer 1

Bryan,
with raster applications, the flexibility of the solutions is extremely important, in other word, sometime, it's probably hard to find a single best solution for all application cases. GeoRaster provides such flexibility. Depending upon your application requirements, you can choose to use a set of individual georaster objects and treat them as a "virtual mosaic" or mosaick them together to create a physical mosaic, or a combination of both.
The sparse mosaic is very useful too. it creates a single mosaic without wasting storage space for no-data areas. it simplifies your database by creating a single georaster object out of a large sparse set of individual georaster objects thus ease the management task and application development. of course, the true scalability of GeoRaster is the prerequisite of such a solution. As real application examples, rarely there is an orthogonal political boundaries for a country, province or city. Mosaics of many thematic or image layers are generally required for such GIS systems for many reasons, but it is probably better to not fill all those void areas outside of the boundary with cell values thus save space and in many cases, improve application performance. In another application, users might be doing a mapping or photogrammetric survey of a large area such as for a big state, but the work progress could take years. You simply create a sparse mosaic for the whole region and then once new data coming in, you can call sdo_geor.updateRaster to add the new images into your sparse mosaic. that might help avoid losing the "big picture".
thanks,
Jeffrey