xfs
[Top] [All Lists]

Re: rdiff-backup / EAs / ACLs

To: Nathan Scott <nathans@xxxxxxx>
Subject: Re: rdiff-backup / EAs / ACLs
From: Greg Freemyer <freemyer-ml@xxxxxxxxxxxxxxxxx>
Date: 19 Aug 2003 09:46:11 -0400
Cc: linux-xfs@xxxxxxxxxxx, acl-devel@xxxxxxxxxxx
In-reply-to: <20030819031858.GE940@frodo>
Organization:
References: <1061244256.29493.72.camel@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <20030819031858.GE940@frodo>
Reply-to: freemyer-ml@xxxxxxxxxxxxxxxxx
Sender: linux-xfs-bounce@xxxxxxxxxxx
On Mon, 2003-08-18 at 23:18, Nathan Scott wrote:
> On Mon, Aug 18, 2003 at 06:04:17PM -0400, Greg Freemyer wrote:
> > XFS/EA/ACL experts,
> > 
> > I'm trying to test rdiff-backup (unstable) for xfs acl support.
> > 
> > So far it is not working:
> > 
> > 
> > My understanding of xfs acls is that they are implemented via EAs. 
> > right?
> 
> Yes, although libacl hides the user/kernel interface and
> presents a (draft-POSIX-standard-compliant) ACL interface
> to users which isn't aware of extended attributes.

Does libattr allow access to ACLs, or from userland are EAs and ACLs
truly distinct?
 
> 
> > If so, should the python module pyxattr handle them?
> 
> I don't know what that module does.

It only talks about EAs, but I was hoping in XFS's case this would be
enough.

> 
> > Or does rdiff-backup need to explicitely support EAs and ACLs
> > seperately?
> 
> You probably want to have a look into the work Andreas did in
> getting the cp/ls/... tools (now called "coreutils" I think) to
> work with ACLs and extended attributes, and start from there.
> 

I'll take a look at cp in particular.

Greg
-- 
Greg Freemyer


<Prev in Thread] Current Thread [Next in Thread>