I am trying the VDS feature with Hdf5 1.10. If the virtual dataset references elements that don't exist in the source datasets, tools to dump the dataset hang. If a source file is missing, then it works and I get the fill value.
For instance, If I create 3 src files with 1D datasets of 6 elements, say I have fileA/dsetA: 0,1,2,3,4,5 fileB/dsetB: 10,11,12,13,14,15 fileC/dsetC: 20,21,22,23,24,25 and then make a 1D vds in a new file which has a size of 4*6=24 elements mapped as follows fileA/dsetA -> start 0, stride 4, count 1, block 1 fileB/dsetB -> start 1, stride 4, count 1, block 1 fileC/dsetC -> start 2, stride 4, count 1, block 1 fileD/dsetD -> start 3, stride 4, count 1, block 1 where fileD doesn't exist, then h5ls -d shows what I'd expect, where the fill value for the vds is -1 0, 10, 20, -1, 1, 11, 21, -1, 2, 12, 22, -1, 3, 13, 23, -1, 4, 14, 24, -1, 5, 15, 25, -1 but if I change the size of the vds to be 4*10=40 elements, without altering the source files, than h5ls -d hangs. I was expecting to see 16 more -1's following the 24 elements printed above. If it sounds like a bug I can prepare an example to re-produce, but maybe this is how the VDS works? best, David Schneider SLAC/LCLS _______________________________________________ Hdf-forum is for HDF software users discussion. [email protected] http://lists.hdfgroup.org/mailman/listinfo/hdf-forum_lists.hdfgroup.org Twitter: https://twitter.com/hdf5
