Exception after migrating from HDFView 2.13 to 2.14 or higher


#1

Hi,

I've written an application that uses the object interface in HDFView. After migrating from 2.13 to 2.14 or 3.0.0, an exception is being thrown when writing the second "slice" of data. Here's the exception and the relevant code. Any ideas what I've done wrong?

Thanks,
Scott

hdf.hdf5lib.exceptions.HDF5DataspaceInterfaceException: Out of range
hdf.hdf5lib.exceptions.HDF5DataspaceInterfaceException: Out of range
   at hdf.hdf5lib.H5.H5Dwrite_double(Native Method)
   at hdf.hdf5lib.H5.H5Dwrite(H5.java:1972)
   at hdf.hdf5lib.H5.H5Dwrite(H5.java:1905)
   at hdf.object.h5.H5ScalarDS.write(Unknown Source)
   at com.agi.stk.volumetric.VolumetricFile.VolumetricFileFromRasterInfo(VolumetricFile.java:642)
   at Main.originalMain(Main.java:517)
   at Main.main(Main.java:105)

FYI: numCoord1 = 180, numCoord2 = 360, numCoord3 = 1, numTimes = 12

      // Create the Values Dataset
      {
         long dims[] = {
               numCoord1,
               numCoord2,
               (numCoord3 > 1) ? numCoord3 : 2,
               numTimes };
         long maxDims[] = dims;

         Datatype dataType = null;
         try {
            dataType = h5File.createDatatype(Datatype.CLASS_FLOAT, 8,
                  Datatype.ORDER_LE, -1);
         } catch (Exception e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
         }

         hdf.object.Dataset dataset = null;
         try {
            dataset = h5File.createScalarDS("Values",
                  volumetricData, dataType, dims, maxDims, null, 0, null);
         } catch (Exception e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
         }

         try {
            // Define the dataset
            dataset.write();
         } catch (Exception e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
         }

         // Write each raster to dataset
         dataset.init();

         long[] start = dataset.getStartDims();
         double[] values = new double [numCoord1 * numCoord2];
         double[] rowValues = new double [numCoord2];
         int valuesOffset;

         // Each file contains data for one altitude at one forecast time
         int a = 0;
         int t = 0;
         for (int r = 0; r < rasterFilePaths.getLength(); r++) {
            System.out.printf("Reading raster: %s%n",
                  rasterFilePaths.item(r).getTextContent());
            Dataset hDataset2 = gdal.Open(rasterFilePaths.item(r).getTextContent(),
                  gdalconst.GA_ReadOnly);

            Band hBand = hDataset2.GetRasterBand(1);
            int bandDataType = hBand.GetRasterDataType();

            for (int row = 0; row < numCoord1; ++row) {
               // GDAL raster (0, 0) is top left. STK wants (0, 0) to be bottom left.
               hBand.ReadRaster(
                     0, row, // xOff, yOff
                     numCoord2, 1, // xSize, ySize
                     numCoord2, 1, // bufXSize, bufYSize
                     bandDataType, rowValues, // bufType, buffer
                     0, 0); // nPixelSpace, nLineSpace

               valuesOffset = ((numCoord1 - row - 1) * numCoord2);
               System.arraycopy(rowValues, 0, values, valuesOffset, numCoord2);
            }

            hDataset2.delete();

            try {
               start[2] = a++;
               start[3] = t;

               dataset.write(values);

               if (numCoord3 == 1) {
                  // Need to duplicate altitude values to create volume
                  // See if skipping this avoids exception. Did for Thunderstorm.
                  start[2] = a;
                  dataset.write(values); // ***** THIS IS LINE 642 *****
               }

               if (a >= numCoord3) {
                  a = 0;
                  t++;
               }
            } catch (HDF5DataspaceInterfaceException e) {
               // TODO Auto-generated catch block
               e.printStackTrace();
            } catch (Exception e) {
               // TODO Auto-generated catch block
               e.printStackTrace();
            }
        }
      }


#2

Scott,

First it is important that the main difference of the object library between version 2 and version 3 is the id size changed from int to long.
The second note is that the exception handling improved in version 2.14 and version 3 libaries.

The exception you note actually is from the H5Dwrite call in the HDF5 library. You should investigate the caller of "hdf.object.h5.H5ScalarDS.write" and double check the values of the dataspace and also verify that error handling is correct.

Allen

···

On Friday, November 17, 2017 7:34:49 AM CST Reynolds, Scott wrote:

Hi,

I've written an application that uses the object interface in HDFView. After migrating from 2.13 to 2.14 or 3.0.0, an exception is being thrown when writing the second "slice" of data. Here's the exception and the relevant code. Any ideas what I've done wrong?

Thanks,
Scott

hdf.hdf5lib.exceptions.HDF5DataspaceInterfaceException: Out of range
hdf.hdf5lib.exceptions.HDF5DataspaceInterfaceException: Out of range
   at hdf.hdf5lib.H5.H5Dwrite_double(Native Method)
   at hdf.hdf5lib.H5.H5Dwrite(H5.java:1972)
   at hdf.hdf5lib.H5.H5Dwrite(H5.java:1905)
   at hdf.object.h5.H5ScalarDS.write(Unknown Source)
   at com.agi.stk.volumetric.VolumetricFile.VolumetricFileFromRasterInfo(VolumetricFile.java:642)
   at Main.originalMain(Main.java:517)
   at Main.main(Main.java:105)

FYI: numCoord1 = 180, numCoord2 = 360, numCoord3 = 1, numTimes = 12

      // Create the Values Dataset
      {
         long dims[] = {
               numCoord1,
               numCoord2,
               (numCoord3 > 1) ? numCoord3 : 2,
               numTimes };
         long maxDims[] = dims;

         Datatype dataType = null;
         try {
            dataType = h5File.createDatatype(Datatype.CLASS_FLOAT, 8,
                  Datatype.ORDER_LE, -1);
         } catch (Exception e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
         }

         hdf.object.Dataset dataset = null;
         try {
            dataset = h5File.createScalarDS("Values",
                  volumetricData, dataType, dims, maxDims, null, 0, null);
         } catch (Exception e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
         }

         try {
            // Define the dataset
            dataset.write();
         } catch (Exception e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
         }

         // Write each raster to dataset
         dataset.init();

         long[] start = dataset.getStartDims();
         double[] values = new double [numCoord1 * numCoord2];
         double[] rowValues = new double [numCoord2];
         int valuesOffset;

         // Each file contains data for one altitude at one forecast time
         int a = 0;
         int t = 0;
         for (int r = 0; r < rasterFilePaths.getLength(); r++) {
            System.out.printf("Reading raster: %s%n",
                  rasterFilePaths.item(r).getTextContent());
            Dataset hDataset2 = gdal.Open(rasterFilePaths.item(r).getTextContent(),
                  gdalconst.GA_ReadOnly);

            Band hBand = hDataset2.GetRasterBand(1);
            int bandDataType = hBand.GetRasterDataType();

            for (int row = 0; row < numCoord1; ++row) {
               // GDAL raster (0, 0) is top left. STK wants (0, 0) to be bottom left.
               hBand.ReadRaster(
                     0, row, // xOff, yOff
                     numCoord2, 1, // xSize, ySize
                     numCoord2, 1, // bufXSize, bufYSize
                     bandDataType, rowValues, // bufType, buffer
                     0, 0); // nPixelSpace, nLineSpace

               valuesOffset = ((numCoord1 - row - 1) * numCoord2);
               System.arraycopy(rowValues, 0, values, valuesOffset, numCoord2);
            }

            hDataset2.delete();

            try {
               start[2] = a++;
               start[3] = t;

               dataset.write(values);

               if (numCoord3 == 1) {
                  // Need to duplicate altitude values to create volume
                  // See if skipping this avoids exception. Did for Thunderstorm.
                  start[2] = a;
                  dataset.write(values); // ***** THIS IS LINE 642 *****
               }

               if (a >= numCoord3) {
                  a = 0;
                  t++;
               }
            } catch (HDF5DataspaceInterfaceException e) {
               // TODO Auto-generated catch block
               e.printStackTrace();
            } catch (Exception e) {
               // TODO Auto-generated catch block
               e.printStackTrace();
            }
        }
      }