[rescue] Getting QIC tapes images

Fri May 15 18:16:03 UTC 2026

The original is comp.sources.unix - volume10 - copytape.Z [
https://github.com/Cutlery-Drawer/comp.sources.unix/blob/master/volume10/copytape
]

As I understand it, tapetool is a SunOS GUI to the core Unix tape tools via
its ToolTalk service protocol, trying to hide the specifics of the tape
system from you, which is the opposite of what you want to know; I would
avoid it.   As I said, I would highly recommend raw reads of all data,
including tape metadata [*i.e.*, "tape file" marks and actual tape EOT
marks].  This is what copytape does.  I will send you a slightly updated
version of the original that modern C89 or later compilers like clang will
understand (althought modern systems like macOS do not include sys/mtio.h,
so you have to install it locally).  If the local C compiler on your
system supports
C89, the updated version should compile and run without warnings.
 Otherwise, please grab the original; you may need to tweak the Makefile
slightly.

On Fri, May 15, 2026 at 9:34 AM Alan Perry <alanp at snowmoose.com> wrote:

> I have the green plastibands and the clear bands. I have found that the
> clear bands have too much tension and can lift material from the tape. The
> tension of the green bands vary and it seems like there is a window where
> they work best.
>
> The tapes came with a Sun 3/160 based Model 32, and some are clearly for
> it. My goal is to restore the CADDStation and I hope all of the software is
> there on the tapes.
>
> Where can one find copytape for SunOS?
>
> I was able to use tapetool to extract a file called tapedir from two
> different tapes
>
> alan
> On 5/14/26 11:35 AM, Clem Cole wrote:
>
> First, what are you using for bands?  (size and manufacturer).
>
> With regard to the tape itself, given that these are old tapes, assume you
> will get one shot in the transport.  So I >>highly<< recommend using the
> old 1985 copytape utility from the USENET (pdf of its man page of the
> command and the format it writes attached), and creating a tape image you
> can decode without needing to touch either the tape or the transport.   If
> you don't have it, send me an email off-list, and I'll be happy to provide
> it to you.
>
> copytape will attempt to read 262144 (256K) bytes at a time.  On a QIC
> drive, which has fixed-size (512-byte) records, that will equate to 512 blocks
> (on a 9-track, which used variable-sized records, each 256k read, will
> return as many bytes as were written on that tape record, which can be a
> small integer to no more than 65536 (64K bytes).   Copytape will read
> tape marks. Tape "files" are delineated by tape marks.  On drivers that
> were written properly, the last "tape file" will have a second tape mark [
> *i.e.*, two in a row] to delineate EOT. Note that many of the QIC drivers
> do not do that (and neither did Ken's original 9-track tape driver for
> Fifth Edition and IIRC Sixth, but by then a number of us had rewritten to
> 9-track drivers to be  much smarter, but I digress)
>
> Once you have a copy of the tape in "copytape" format, you can decode it
> much more easily.
>
> As for the format itself, it depends on which Computer Vision system
> wrote it.   Their early systems were based on DG Novas and ran RDOS
> (similar to DEC's RT11).  From the late 70s until the mid 1980s, they
> replaced RDOS with their own OS called CGOS (Computervision Graphics
> Operating System), and by the late 80's, early 90's, they had ported their
> CADDS system to run on UNIX and started shipping it on Suns. Finally, they
> got bought by Prime, and then moved it again to PRIMOS.
>
> Given that this is a QIC tape, the time frame says either late CGOS or
> SunOS, although Prime might have supported QIC; I never saw anything but
> 9-track on their systems.
>
> So ... if it was written on an early CV system, it is likely to be in what
> DG called MTIO format or possibly in the DG backup format.  Data
> General’s RDOS natively used raw, streaming sequential blocks without a
> complex, metadata-heavy file system structure, such as ANSI labeled tapes
> (DEC often used a superset - embrace and extend - ANSI tape format).   I
> don't know what CGOS used, but I suspect it was similar to the DG style.
>  If it was written on a Sun or other UNIX box, it is likely to be TAR
> (hopefully) or CPIO.  If the latter, there is a slew of different CPIO on
> tape formats - this is historical because it was created for PDP-11s and
> was a binary format.   It could also be one of the many UNIX backup
> formats, so you will need to do a little examination to determine which
> one.  Modern versions of the original V7 Unix file(1) >>might<< be able to
> identify the format. Finally, like DG, Prime opted not to use DEC-style
> (pseudo-ANSI) labeled file structures natively. Instead, they also created
> their own proprietary, streaming binary format deeply tied to PRIMOS’s
> unique disk architecture and character encodings.  I've never seen
> documentation on this format.
>
>
> On Thu, May 14, 2026 at 9:44 AM Alan Perry via rescue <rescue at sunhelp.org>
> wrote:
>
>> I seem to have found a working combination of drive and replacement's
>> bands to be able to get images of the dozens of Computervision install
>> tapes that I have.
>>
>> They are QIC-24 from the late 80s and aren’t in the “collection of tar
>> files” format that I usually use mt fsf and dd to read. Each tape starts
>> with a file with a “tapedir” record with permission and other stuff,
>> followed by a bunch of records that look like a number followed by a topic
>> or title. What am I looking at and what do I use to get the info off of the
>> tape?
>>
>> alan
>>
>>
>>
>>
>> _______________________________________________
>> rescue list - http://sunhelp.org/mailman/listinfo/rescue_sunhelp.org
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sunhelp.org/pipermail/rescue_sunhelp.org/attachments/20260515/e40fa317/attachment-0001.html>