On 8/10/2016 7:04 AM, Tor Arntsen wrote:
On 10 August 2016 at 15:22, <aswood at
t-online.de> wrote:
I successfully took a (factory new) DEC TSZ07
SCSI tape drive into operation using a Sun SS20 and a Linux box.
Now I do have a big pile of CDC, DEC, HP, Convex and IBM tapes and I'd like to create
tape images to file to save the tapes content.
What is the to be preferred procedure to image the tapes, which software to use and which
kind of format to store the images?
fwiw, I made my own format when I archived all
my old tapes. A very simple one:
- I read the tape record by record
- The output disk file contains a 4-byte integer in big endian format,
followed by one record.
- The 4-byte integer contains the physical record size as read from the tape.
- Then another 4-byte value plus a new record, and so on.
- End-of-file markers are encoded by using 0 as the value for the
4-byte integer, with no record following (logically enough).
I'd find a way to
get it to some format acceptable to simh as a lot of
the tapes you have will probably work on emulators. There are already
utilities for going to and from that, and perhaps even some that will
compile and run on your hardware to go from SCSI devices or serial block
devices to that format.
Also WRT to the above format, I believe that most tape formats have a
block size for the previous record so that the tape can be read in
reverse. Not sure I saw that from the description.
I have a simple program that creates a single file per tape block on
reading. A zero length file is a tape mark. File number on the tape is
the first part of the file name, and record number of the block is the
second part, with the file name being 4 hex ascii, a "-" character and a
4 hex ascii character.
I then read that to convert to the archival formats, or to examine the
file prior to encoding.
Al K's reader and format has meta data related to tape errors
encountered so that tape images can be padded out with some filler for
possible recovery or use of tapes that are unreadable. I have not
written any of that code, since it usually isn't available from raw
drivers on Linux. his reading method uses specialized hardware to
digitize the files, and is somewhat different when the metadata is
possible to create.
thanks
Jim