I have been using h5dump to help provide diagnostics on our data, and have noticed behavior that I don't understand nor expect.
If I use a command like "h5dump -o somefile --dataset=somedataset -s 5630,0,0,0 -c 10,2,32,1 somefile.h5", the output is as I expect: a dump of all elements from 5630,0,0,0 through 5639,1,31,0
However, if I change that command from -c to -k (changing the count to a block size instead), I end up with a very strange set of blocks in no apparent pattern. First it dumps 5630,0,0-19,0 then 5630,0,1-20,0 and so on. I guess there is a pattern here, but it's a really odd one, and h5dump never goes beyond 5630 in the first dimension.
I would have expected to get the exact same set of elements. Why am I seeing what I'm seeing?
John,
Thank you for reporting the problem.
I believe there is a bug in h5dump for the -k option for datasets of 3 or higher
dimensions. The "-c" should give you the correct data.
Thanks
--pc
ยทยทยท
On 2/28/2011 1:29 PM, John Knutson wrote:
I have been using h5dump to help provide diagnostics on our data, and have noticed behavior that I don't understand nor expect.
If I use a command like "h5dump -o somefile --dataset=somedataset -s 5630,0,0,0 -c 10,2,32,1 somefile.h5", the output is as I expect: a dump of all elements from 5630,0,0,0 through 5639,1,31,0
However, if I change that command from -c to -k (changing the count to a block size instead), I end up with a very strange set of blocks in no apparent pattern. First it dumps 5630,0,0-19,0 then 5630,0,1-20,0 and so on. I guess there is a pattern here, but it's a really odd one, and h5dump never goes beyond 5630 in the first dimension.
I would have expected to get the exact same set of elements. Why am I seeing what I'm seeing?
_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@hdfgroup.org
http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org
