mjk@ttrdc.UUCP (Mike Kelly) (05/07/85)
There is also a limit on 3b20's. We've been getting tar format tapes written on a vax running 4.2 and the blocksize = 20. Our controller can't read the tapes. If you're sending tar tapes for general distribution, please use a smaller blocksize; whatever the default is seems to work. Mike Kelly AT&T Teletype
bill@ur-cvsvax.UUCP (Bill Vaughn) (05/10/85)
> There is also a limit on 3b20's. We've been getting tar format tapes > written on a vax running 4.2 and the blocksize = 20. Our controller > can't read the tapes. If you're sending tar tapes for general distribution, > please use a smaller blocksize; whatever the default is seems to work. > > Mike Kelly > AT&T Teletype But the default blocksize for TAR is 20. (Always has been, always will be :-) Bill Vaughn, Center for Visual Science, UNIV. OF ROCHESTER {allegra,seismo,decvax}!rochester!ur-cvsvax!bill
ed@mtxinu.UUCP (Ed Gould) (05/12/85)
> But the default blocksize for TAR is 20. (Always has been, always will be :-)
When tar was originally released by Bell, with the Seventh Edition,
its default blocksize was 512 bytes. That's still the default on
many systems that include it. The one major exception is for tar's
derived from Berkeley VAX distributions, where the default is 10240.
--
Ed Gould mt Xinu, 2910 Seventh St., Berkeley, CA 94710 USA
{ucbvax,decvax}!mtxinu!ed +1 415 644 0146adm@cbneb.UUCP (05/12/85)
>But the default blocksize for TAR is 20. (Always has been, always will be :-)
ok, I give up. 20 *what*? kilobytes? 512-byte blocks? bytes? or some
logical block size?
Whatever it is, its a pity it doesn't work on the 3B20, I've run into this
problem several times already. The single outstanding quality of TAR was
that it was so universal among UNIX* machines. Not any more!mark@cbosgd.UUCP (Mark Horton) (05/12/85)
In article <192@ur-cvsvax.UUCP> bill@ur-cvsvax.UUCP (Bill Vaughn) writes: >> There is also a limit on 3b20's. We've been getting tar format tapes >> written on a vax running 4.2 and the blocksize = 20. Our controller >> can't read the tapes. If you're sending tar tapes for general distribution, >> please use a smaller blocksize; whatever the default is seems to work. > >But the default blocksize for TAR is 20. (Always has been, always will be :-) > Sorry, but that's not quite true. V7 was the first distribution that had tar. The default blocksize was 1, but the maximum (ang generally recommended) blocksize was 20 (e.g. 10K, since in those days everyone "knew" a block was 512 bytes. Everyone inside AT&T still does.) Some time ago, Berkeley fixed a number of problems with the tar command. It made the default blocksize 20 (since experience had shown that this was the best choice, and 3B20 tape drives were just a gleam in the eye of the evil Dr. Foo.) It also made the defaults work properly on an arbitrarily blocked tape, and wrote directories onto the tape so they could be restored properly when the tape was read in. Unfortunately, the Berkeley tar now has lots of Berkeleyisms in it, so it won't run on System V. (Has anyone done a port?) Meanwhile, tar was also put into System III, essentially untouched and untouchable. The chown command was allowed for any user, but tar was never fixed to avoid giving away your files irretrevably (it's a 2 line fix, and it was reported to them.) The default blocksize was left at 1. You see, they all use cpio, and they all forgot tar was there. The 3B20 buffer can only hold 6144 bytes, as I recall. So if you're making tar tape and want it to be portable, block it at 10. (Don't use 1 unless you want to waste most of your tape.) There is no such thing as a portable cpio tape; most of the UNIX machines in the world with 9 track tape drives can't read it, even with the -c option, because V7, V8, 32V, and nBSD don't come with cpio. I only know of two UNIX distributions that don't come with tar: PC/IX and the 3B2's System V. People with these systems are just out in the cold.