Coda File System

Re: Inconsistent adherence to Unix permission conventions

From: Randy Harmon <rjh_at_fortheweb.com>
Date: Thu, 11 Oct 2001 14:49:11 -0700
----- Original Message -----
From: "Jan Harkes" <jaharkes_at_cs.cmu.edu>

> On Thu, Oct 11, 2001 at 12:35:38PM -0700, Randy Harmon wrote:
> > -r-------- root     root   filename   Permission denied (OK)
> > -r-----rw- root     root   filename   Permission denied (BUG)
> > -rw------- root     root   filename   file is written (BUG)
>
> ACL (rlidw) . ownerbits (r) = rlid, no write permission.
> ACL (rlidw) . ownerbits (r) = rlid, no write permission, not a bug.
> ACL (rlidw) . ownerbits (rw) = rlidw, write permission, not a bug.

OK, I've done development work myself, so I recognize that you're
perspective is, "this is the way we designed it to operate, so I'm not
surprised by this behavior.  No bug.".  From a user's perspective,
though, I have a unix filesystem that lets me do Unix things like
setting ownership and permission bits.  And then it's sometimes
honoring those conventions, but in a way that's not consistent with
Unix conventions.  So the *impression* is one of bugginess, even if
the behavior is consistent and predictable.  It seems clear to me that
the design could be... further optimized.

So, maybe it's a bug (from my fresh perspective on the software), or
maybe it's just an enhancement request.  I'll call it a design
enhancement: Venus should be configurable to honor the
owner/group/other id's and mode-bits for chmod/chown and read/write
requests.  As well as honoring the ACL settings.

For now, of course, I'm proceeding with the knowledge of *current*
behavior.  Thanks for the clarifications, and for your consideration
to this idea.  IMHO, it would improve Coda's usefulness and appeal to
Unix admins.

Randy
Received on 2001-10-11 17:49:17