[netcdf-java] Problem with Fortran syntax and OpenDAP files

[Sachin Kumar Bhate]

John,

Using certain fortran type syntax to read data from OpenDAP files
is resulting in 'NullPointer Exception'.

For example, when I call the 'read' method on variable and supply the 
'range' to read 'all'
data for a particular dimension using colon ":" syntax, it gives null 
pointer, otherwise it works fine.
 >>>
String range = ":,50,100";
..........
varID=ncid.findVariable(var);
data = varID.read(range);
........
<<<<<
Test file: 
'http://stellwagen.er.usgs.gov/cgi-bin/nph-dods/models/vs/vs012/his/ocean_his_0001.nc'

With little debugging, I found out that the method  "public static List 
parseSpec(String sectionSpec)"
in "ucar.ma2.Range" assigns "null" for colon syntax ":" while parsing 
the string. Later, before the actual data is being read,
the 'null' assignment in the range list never gets changed to actual 
dimension size, thus resulting in exception.

For quick solution (till you find a better one) I have found that 
modifying the method  'public Array read(String sectionSpec)'
in 'ucar.nc2.Variable' with code below works fine.

  public Array read(String sectionSpec) throws IOException, 
InvalidRangeException  {
    List section = Range.parseSpec(sectionSpec);

   //Update ----start ----
  // If any 'nulls' in the range list, try and change with actual size 
of the dimension.
    Variable vs = new Variable( this);
    Range.setDefaults( section, vs.shape);
  //---end

    return read( section);
  }


Is this really a problem with API, or I am doing something wrong while 
reading data using Fortran type syntax ?

btw, if I use toolsUI GUI to read data (for test purpose) using the 
range ':,50,100', it works fine.
I think it's using 'Ncdump' class to read data which may be handling the 
':' syntax differently.
Just my guess.

thanks..

Sachin.


-- 
Sachin Kumar Bhate, Software Engineer
MSU-High Performance Computing Collaboratory, NGI
John C. Stennis Space Center, MS 39529
Ph: (601) 466-9692