[lug] compressed cpio archive problems

carl.wagner at level3.com carl.wagner at level3.com
Tue Nov 20 14:31:33 MST 2001


3b1, wasn't that also called the 7300 or Unix-PC?  Around 86'-88'?
I would guess that 99% of them are in landfills by now.

Have you tried to decompress and de-cpio them on other boxes? 
Old SunOS, Ultrix, HPUX, AUX, AIX, SCO, etc. ?

Just in case the algorithms are so old that no one supports them
anymore.

It would be worth a shot, and a lot less work than trying to decode
them by hand!

Do you care if other people see what is in them?  If not you could 
post them somewhere and someone might be able to give it a try, if
you don't have access to those types of machines.  (I have a few
of the different types.)

Carl.


"S. Luke Jones" wrote:
> 
> Tom Tromey wrote:
> > Luke> Nobody has uncompress anymore (thanks a lot, Unisys) but gunzip
> > Luke> has the equivalent capability.
> > As others have pointed out, there is an uncompress for Linux.  On my
> > RHL box it is in the ncompress RPM.
> 
> I meant to write "uses" not "has". Sure enough, there's not only
> an uncompress (I understood the patent to allow that) but a compress
> as well. I'll remember that in case I want to lose some data beyond
> any hope of retrieval in the future.
> 
> > I don't think it is possible.  But don't take my word for it.  Find a
> > coding expert.  Or you can read the paper on which compress is based;
> > it is referenced in the man page.
> 
> Sigh. We had to do a project with one of those compression schemes
> in graduate school. Afterwards, I told myself I'd never use anything
> harder to wrap my brain around than RLE. If I were to go that route
> it would tempt me to estimate the value of the content of these cpio
> archives, which, since they contain old graduate school projects
> (buggy implementations of compression schemes and the like) is --
> sadly -- very close to zero.
> 
> > There's probably some small chance that there are bugs or
> > incompatibilities between the compress you used to create this file
> > and the current uncompress.  What media did you store these files on?
> 
> Let's see: they were created on a 3B1 and later transferred to a
> 386 PC running UNIX(tm). After that, I came up with .cpz as a three
> letter extension so I could move them to a Windows PC, where they
> spent a few years as unreadable collections of bits. I couldn't try
> to estimate the number of times they survived system rebuilds on
> consumer-grade tapes (CMS Jumbo 250 and Travan 3). After I got a
> Linux PC and a CD-R I started using CD's for archival purposes.
> 
> > I.e., is there a reasonable chance that they are really corrupt?
> 
> There is a small but measurable chance they are not corrupt, but
> the same could be said of any politician.  I wouldn't want to bet
> on it, though; at least, not with my own money.
> 
> > Anyway one thing that might be worth trying is to find a system
> > similar to the creating system and uncompress and unarchive there.
> 
> Other than Ebay, no. (See the provenance above.)
> 
> > Luke> cpio: warning: skipped 1247 bytes of junk
> >
> > One route open to you, which you may not like, is to read up on the
> > various cpio formats, and then examine the archive file using a hex
> > editor.  As I recall (I hacked on GNU cpio for a short time back in
> > 1994 or so), the cpio formats are pretty straightforward.  So this
> > wouldn't be as hard a job as it sounds.
> 
> Sigh again. I realize you're being helpful here. It's just I wanted
> some fairy godmother type to point out the -pdq switch to cpio that
> I had missed in its 21-line usage screen. If I can successfully get
> them uncompressed, I might actually try to un-cpio them this way.
> I'm always -- well, really almost never -- looking for new ways to
> learn how to use perl's unpack function. But the uncompression
> double-whammy is sapping my microscopic enthusiasm for new projects
> like this.
> 
> --
> Luke Jones  luke vortex frii fullstop com
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug



More information about the LUG mailing list