Archiving and Transporting Miriad Datasets

As Miriad datasets are written in a machine-independent format (the bits stored on disk to represent a particular value are independent of the host system, whether it be a VAX, Sun or Cray, for example), no translation is needed when transporting datasets between unlike architectures.

Instead of relying on any special tasks to write tapes for archiving or transport, Miriad relies on the facilities of the host system. Probably the most convenient archiving format is tar. This is universally supported on UNIX machines, and is quite common on VAX/VMS machines.

The two most common commands used with tar are to copy to and from tape. To copy to tape, use

% tar -cvf device datasets

where device is a tape device name such as /dev/nrst0 (the non-rewinding, raw interface, i.e. device names starting with nr), and datasets is a list of datasets and files to copy to tape (If a dataset, or directory, name is given, tar works recursively, copying all the contents of this directory). These can include wildcards. For example, to copy all datasets and files from the current directory to /dev/nrst0, use

      tar -cvf /dev/nrst0 *

To retrieve data from tape, use the tar command

% tar -xvf device

Multiple tar files can be stored on the one tape so that you do not overwrite old data. Again the UNIX mt command can be used to manipulate the tape position.

See the man pages on the tar and mt commands for more information.