Coda File System

Re: Coda and RPM/tar/lchown()

From: Ivan Popov <>
Date: Fri, 29 Apr 2005 19:21:17 +0200
On Fri, Apr 29, 2005 at 12:24:47PM -0400, Jan Harkes wrote:
> > It would be a totally another and good thing to have something like
> >  cfs ls-l
> > which could possibly show a lot of metainformation,
> I think AFS was actually using a modified ls which would look up the
> usernames assocated with the AFS id's in their equivalent of the pdb
> database (pts?) and would show only the parts of the modebits that
> mattered.

Huh, really, did it? I had such thoughts while writing for this thread
but declined the idea.

"ls" output format is by definition Posix-bound. As such, ls conforming to
"ls standard specifications" can not produce valid results on a global
file system.

We can never build a conforming replacement, hence the choice should be
possibly a different program, with a different name. I think "cfs <something>"
would be least confusing, it makes it clear where the command belongs.

The best we can do for the users is to teach them to ignore mode bits
and the "owner" concept - instead of imitating the semantics which
never can be right on a global fs...

Otherwise they would believe that Posix semantics is the absolute truth
and that it is Coda's fault that it does not "fully" implement it.

Global privilege management is nothing they have taken courses on, yet,
and Posix is the generally taught way of thinking. Too bad.
Coda will become fully useful when people learn the corresponding way
of thinking. Otherwise they never can make real use of its strengths.

My 0.2c
Received on 2005-04-29 13:22:13