将（>）h5dump输出重定向到文本文件会导致数据丢失

Question

I have a bash script that runs the following command:

h5dump --noindex particle_pos.6400_acetone > tempPythonFile.txt

The contents of the hdf5 file, particle_pos.6400_acetone, is a collection of x,y,z coordinates like a sample shown below.

HDF5 "particle_pos.6400_acetone" { GROUP "/" {    DATASET "particle position" {
      DATATYPE  H5T_COMPOUND {
         H5T_IEEE_F64LE "x";
         H5T_IEEE_F64LE "y";
         H5T_IEEE_F64LE "z";
      }
      DATASPACE  SIMPLE { ( 7372815 ) / ( 7372815 ) }
      DATA {
         {
            0.0985972,
            0.0051215,
            0.000543894
         },
         {
            0.0986334,
            0.00520195,
            0.000408405
         },

The file should look like the text above all the way through, but when I go to the end of the text file I see the following.

},
         {
            ,
            -0.162017,

         },
         {
            ,
            ,

         },
         {
            ,
            ,

         },
         {
            ,
            ,

To me it looks like something is happening that is dropping data because there is occasionally a floating point number in the sea of blank entries.

I haven't had an issue with this command on smaller data sets, but I started seeing data missing from the tempPythonFile.txt file when my data set contained 7372815 elements.

Is there a more accepted method for getting HDF5 data out to a text file? Is the redirect(>) command not suited for large amounts of data?

Thank you

Answer 1

我确实收到了HDF5小组的回音，他们确实发现这与我使用的HDF5版本有关。