Stephen White's proposal involving a `private' group
(I'll not repeat things that Matthew Birkholz has said - I agree with
him here and also regarding the question of revision control.)
Stephen White's scheme has a serious disadvantage compared to most
other arrangments: it becomes very hard for users to control access to
their files effectively.
This is because they can't chgrp any files with different ownerships
back to the state they were in when they were created in their home
directory (because they're not a member of the group). The only way
to do it is to make a new copy of the file and delete the old one.
This would be quite tedious in the case of directory trees and even
more so in the case of files with multiple hard links.
If and when someone writes a set of programs to allow user-maintained
groups this problem will become even more serious, as users will need
to use the chgrp command on a regular basis to grant and deny access.
It should also be noted that Stephen White's proposal won't work
unless the home directories filesystem and /tmp are both mounted with
BSD semantics.
This is because the setgid bit doesn't get propagated to
subdirectories you create unless you're a member of the group in
question.
Ian.
Reply to: