Trash spec 0.2, technical questions
alexl at redhat.com
Thu Sep 2 10:32:31 EEST 2004
On Wed, 2004-09-01 at 16:59 +0200, David Faure wrote:
> On Wednesday 01 September 2004 16:17, Jerry Haltom wrote:
> > > Yes. Thus the discusion of a setuid app to set the right permissions on
> > > this directory.
> > That is not going to work on the important stuff: remote mounts.
> When the creation (or use) of a .Trash-$uid directory on a mountpoint fails,
> we can always use fallback: refusing trashing (seems to be Nautilus's preference)
> or copying the data to $HOME/.Trash (as I implemented in KDE).
> So border cases that fail don't matter much. But indeed a setuid app
> failing over NFS isn't a border case. However for such cases, a .Trash created
> by the admin does the job.
> But that means we need a rather complex set of things to try:
> 1) create $topdir/.Trash-$uid as user
> 2) use $topdir/.Trash/$uid after root created .Trash
> 3) use a setuid app to create .Trash
> Since copying data to $HOME is a good enough fallback IMHO, I am fine
> with 3) being left out.
Yes. I agree. Especially with the always unknown danger of setuid apps.
Alexander Larsson Red Hat, Inc
alexl at redhat.com alla at lysator.liu.se
He's a lounge-singing crooked vagrant on the edge. She's a brilliant
out-of-work archaeologist from beyond the grave. They fight crime!
More information about the xdg