This release adds further support for large datasets where the size of
an HDU may exceed 2GB.  In ArrayDataInput (and the BufferedFile
and BufferedDataInputStream that implement it)
  int skipBytes(int) 
method of java.io.DataInput is now overloaded with 
  long skipBytes(long).  
  
In ArrayFuncs 
  int computeSize(Object) 
method is augmented with 
  long computeLSize(Object)
It was not possible to use the same name here since the argument type
is the same.
Similarly 
  int nElements(Object)
is now matched with 
  long nLElements(Object)
  
These changes should not affect current usage of the existing methods.
  
References to skipBytes and computeSize in the FITS classes now take
advantage of these new methods.

While these changes increase the support of the library for large datasets,
there are still a number of restictions that arise from Java's limit that
array indices must be ints.  E.g., no single dimension can exceed 2 GB, 
and the total size of the heap for a given binary table HDU cannot 
exceed 2 GB.  ASCII tables may also be limited to 2 GB in some circumstances. 
Files which exceed these limits  may be readable using line by line 
approaches, but users will need to use the library at a much lower level.

-- 

The Header.read() method may now throw an IOException in circumstances
where it would previously throw an Error.  It probably should throw
a FitsException, but that was not in the signature and might have broken
existing programs.

A bug in the new PaddingException was fixed which allows use of Tilers
with truncated Image HDUs.

Some obsolete comments indicating that BITPIX=64 was an extension of
FITS were deleted.  FITS has officially supported longs for a fair number
of years now.

--

The regression tests have been augmented to test the new features, but users
should note that the new BigFileTester test takes a very long time to run. E.g.,
on the primary development machine this takes 240 seconds while all of the
other tests finish in just a few seconds.  The time is simply the time it
takes to write a file of known content that is more than 2 GB in size.