Re: Bug in cron postinst

["John H. Robinson, IV" <jaqque@debian.org>]

Matt Zimmerman wrote:
> On Wed, Sep 03, 2003 at 01:40:51PM -0700, John H. Robinson, IV wrote:
> 
> > Matt Zimmerman wrote:
> > > there:[~] /bin/bash
> > > mdz@there:~$ foo="a b"
> > > mdz@there:~$ for x in $foo; do echo "$x"; done
> >                         ^^^^
> > > a
> > > b
> > 
> > $ foo="a  b"; for x in "$foo"; do echo "$x" : $x; done
> > a  b : a b
> 
> That misses the whole point, which was to iterate over a list of items, some
> of which could contain whitespace.

i understand that. the whole point i was trying to make is that using
bash and sh variables is inadequate to that task.

% a=("apple  tony  the  one  and"\ \ only fred betty "wilma")
% for i in $a ; do echo $i ; done                              
apple  tony  the  one  and  only
fred
betty
wilma

as i said, zsh users are excused from this excercise.

for the bash purists:

$ a=("one  two" "a  b" "i  ii")
$ for i in "${a[@]}"; do echo "$i"; done
one  two
a  b
i  ii

good luck setting that array from the output of find(1), though.
remember, newline is a valid filename character.

quick testing indicates that ``a=(*)'' works, even for files with
embedded newlines. but wasn't the whole point to get away from shell
expansions?

in the following example, file?3 has an embedded newline, and "file 1"
has an embedded space

$ touch file\ 1 file2 file'
> '3
$ a=(*)
$ for i in "${a[@]}"; do echo :"$i":; done
:file
3:
:file 1:
:file2:


-john