Bill Silvert wrote:
>
> Jerry Peek writes:
>
> >>>find . -exec chown $USER {} \;
> >>>find . -exec chgrp $GROUP {} \;
> >
> >This is a good substitute for chown -R, I think. But if the chown
> >fails on some pathname, the chgrp will run, anyway, and might fail
> >too. If you write the find another way, a failed chown (with a nonzero
> >status) will prevent the chgrp from running on that particular
> >pathname... it also avoids descending the tree twice:
> >
> > find . -exec chown $USER {} \; -exec $GROUP {} \;
> >
> >There's an implied "-and" between the first and second -exec's.
> >"find" does left-to-right short-circuit evaluation on the expression;
> >and it will quit if it can tell that the expression will be false.
>
> An alternative would be to create a file called mjchown as part of the
> installation process, so that we could have
>
> find $MJ_HOME -exec $MJ_HOME/mjchown $USER $GROUP {} \;
>
> and write mjchown to be as general and bulletproof as desired, using
> exit codes and so on. I'm not sure about the exit codes, but maybe
>
> #/bin/sh
> if ( chown $1 $3 )
> then chgrp $2 $3
> fi
>
I had suggested:
#!/bin/sh
#
# Emulate the user:group and user.group 'chown'
# for max portability, assume V7 shell, sed and awk
# jmj
#
Param=`echo $1 | sed 's/[\.:]/ /'`
shift
User=`echo $Param | awk '{ print $1 }'`
Group=`echo $Param | awk '{ print $2 }'`
if [ "x$Group" = "x" ]; then
:
else
chgrp $Group "$@"
fi
chown $User "$@"
B
--
#include <std/disclaimer>
=========================
Jim Jagielski << jim@jaguNET.com >> | "Not the craw... the CRAW!"
References:
|
|
