Subject: Re: while we're looking at old bugs
To: None <email@example.com, firstname.lastname@example.org>
From: The Grey Wolf <email@example.com>
Date: 10/03/1997 01:03:14
Jim Wise wrote:
#: At any rate, the ability to mv(1) to a directory on another filesystem
#: is itself _new_. 4.3 did not allow this. SVR2 did not allow this. 4.4
#: added this. If the scripts are new enough to expect this, they're
#: surely new enough to expect restricted chown. Or they are in error...
The ability to mv(1) directories across filesystems has always failed.
("Can't mv directories across filesystems" was the error, even.)
The ability to do:
% mv /fs1dir/file /fs2dir
has _always_ been doable, even under 4.2 BSD. 4.3 BSD did not change
this. If SVR2 didn't allow it, it was because they weren't smart enough
to do a cp(1) after a failed rename(2) with EXDEV. And now that I recall,
mv(1), cp(1), rm(1) and ln(1) were ALL linked to each other and argv
checked (I got to be a source integrator for a SVR2/3/4 project).
At any rate, this concept is NOT new to 4.4BSD.
If mv(1) cannot chown my files as a normal user, fine. Shut up
about it. POSIX_CHOWN_RESTRICTED certainly defines the semantics
for whether or not the chown(2) will work.
In addition to this, the last time I did a mv(1) as root across
filesystems, the file turned out to be owned by root instead of its
previous owner (tells you how much I actually use 'mv' across file-
systems -- usually, I use a dual tar pipe or an explicit copy and
I hold with the motion that mv should NOT complain about not being
able to chown(2) or chgrp files; if a user wants to see this, they
should have to ask for it.
#: Jim Wise