*BSD News Article 2103


Return to BSD News archive

Newsgroups: comp.unix.bsd
Path: sserve!manuel!munnari.oz.au!mips!mips!darwin.sura.net!mojo.eng.umd.edu!news!djm
From: djm@eng.umd.edu (David J. MacKenzie)
Subject: Re: Sorry, everyone, but I'm cpio clueless...
Message-ID: <DJM.92Jul20122506@frob.eng.umd.edu>
Date: Mon, 20 Jul 92 17:25:06 GMT
Organization: Project GLUE, University of Maryland
In-Reply-To: terry@npd.Novell.COM's message of Mon, 20 Jul 1992 15:26:50 GMT
References: <1992Jul17.160256.4518@news.iastate.edu> <1992Jul17.180345.1296@nrao.edu>
	<1992Jul18.044401.2343@uvm.edu>
	<1992Jul20.152650.4070@gateway.novell.com>

   In article <1992Jul18.044401.2343@uvm.edu> wollman@sadye (Garrett Wollman) writes:
   >  - The only cpio program that tries to make up for byte-sex problems
   >is GNU cpio.

   Being a tar fan, and hoping for the quick death of cpio, can't stop me from
   correcting you here.  Whether or not there is a dependancy is based on the
   options you use.  If you use "oacumvB", then there is not a byte order
   problem.

I think what Garrett was referring to is the default cpio format --
what you get when you don't give cpio the -c option (or -H for SVR4).
If you have a cpio archive created that way on a machine with the
opposite byte-sex from yours, the UNIX cpio can't read it.  GNU cpio
and afio can.

If you give cpio the -c option (or use the other portable formats the
SVR4 cpio supports, and GNU cpio will soon support), then the archive
should be the same no matter what your machine's byte-sex.  There's no
good reason to use the default (binary) cpio format(s) anymore.