I need a better way to loop in the shell...

[Norberto Meijome]

user wrote:
> I always do loops in /bin/sh like this:
> 
> for f in `cat file` ; do rm -rf $f ; done
> 
> Easy.  I like doing it like this.
> 
> The problem is, when I am dealing with an input list that has multiple
> words per line, this chops it up and treats every word as a new line...
> 
> For instance, lets say I have a file full of filenames, like this:
> 
> # cat file
> 
> 10,000 Maniacs MTV Unplugged - 01 - These Are Days.mp3
> 10,000 Maniacs MTV Unplugged - 02 - Eat For Two.mp3
> 10,000 Maniacs MTV Unplugged - 03 - Candy Everybody Wants.mp3
> 
> and I try to use the above loop on it, it thinks that every word is a line
> ... the above loop will attempt to delete the following files:
> 
> 10,000
> Maniacs
> MTV
> Unplugged
> -
> 01
> -
> These
> 
> (and so on)
> 
> Even if I quote the variable $f, like:
> 
> for f in `cat file` ; do rm -rf "$f" ; done
> 
> it still does the same thing.
> 
> -----
> 
> So my question is, what is a nice simple way to loop in the shell, as
> close to what I am doing above as possible, that does not have this
> problem ?  Should I just be using something other than `cat` ?  Or what ?

you can redefine the character(s) used by your shell for delimiting 
words after expansion. In Bash , this is the env var IFS. so , if you 
know there are no ;, you can say

export IFS=";"
[your loop here]
export IFS=

and you should be set.

man bash for more info.

Beto