With the -t option, tarx lists the file names it sees in the archive.
Without the -t option, tarx takes file or directory names as arguments and attempts to extract them from the archive. (If no names are given, tarx extracts everything it can find.) Tarx is not willing to create directories, however, so these must be made manually beforehand if they do not already exist. Files are owned by the user, and have his default permissions.
Tarx allocates a read buffer of blockfactor*512 bytes, where the default blockfactor is 400 (to allow for the very large blocks sometimes found on cartridge tapes), and attempts a full-sized read each time. It copes properly with getting less than it expected, although it does insist that the result be a multiple of 512 bytes. When dealing with cranky tape drives or other special situations, or severely short of memory, or working on a 16-bit machine, it may be useful to specify a smaller blockfactor. (Notably, on 16-bit machines 126 is often the maximum that will work.) Specifying a larger blockfactor may be necessary if tape blocks are truly enormous.
``tarx joe/precious </dev/rmt0'' restores the file ``joe/precious'' from the tape mounted on /dev/rmt0. The directory ``joe'' must already exist.
Arguably should use the tar header-block checksum, instead of the slightly-arcane pattern matcher, to identify header blocks.
A tar archive containing a file which is itself a tar archive will cause considerable confusion.