The HDF5 table-writing (called by hdf5_copy_group at the start of Hyperion) causes a segmentation fault when field names have different lengths. The following script reproduces the issue:
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
air 00000001001EB932 Unknown Unknown Unknown
air 00000001001732E2 Unknown Unknown Unknown
air 0000000100175887 Unknown Unknown Unknown
air 000000010015DCEB Unknown Unknown Unknown
air 000000010005A956 Unknown Unknown Unknown
air 000000010005B8AC Unknown Unknown Unknown
air 000000010000CF25 Unknown Unknown Unknown
air 0000000100001A30 Unknown Unknown Unknown
air 0000000100002959 Unknown Unknown Unknown
air 00000001000013CA Unknown Unknown Unknown
air 000000010000110C Unknown Unknown Unknown
air 00000001000010C4 Unknown Unknown Unknown
The release notes for HDF5 indicate that this was a known bug, and is fixed in 1.8.6+:
h5tbmake_table_f: Fixed error in passing an array of characters with different length field names.
Waiting for a reply from HDF5 developers for a workaround (if possible).
The HDF5 table-writing (called by hdf5_copy_group at the start of Hyperion) causes a segmentation fault when field names have different lengths. The following script reproduces the issue:
The error is just a normal segmentation fault:
The release notes for HDF5 indicate that this was a known bug, and is fixed in 1.8.6+:
Waiting for a reply from HDF5 developers for a workaround (if possible).