I multi-boot several Linux distributions with an assortment of additional data partitions. I get frustrated whenever fsck is forced during boot. (It ONLY happens when I'm in a hurry don't you know...) So I wrote a script to automate forced fscking when I do have the time. (And/or while I'm doing something else in another workspace.
Because I multi-boot, I've learned that udev doesn't always assign the same device name to each drive for all distributions. I've had the same partition identified as hda5, sda5, & sdb5 by different distributions (without doing anything to affect the boot order) So my solution is to keep a list of partitions in a specific file on each distro with valid device names according to that distribution's udev process. Actually I'd use LABEL= instead but the labels don't show up in /etc/mtab, and I like to make sure a partition isn't mounted before I try to fsck it.

I can make this work in a for loop using cat. But I've seen so many things about NOT using cat that I wanted to rebuild my script. I can make this work with a redirect instead of cat via a while loop, But I "LIKE" old style for loops. But I can't seem to find a way to make a redirect work with one.

I thought this might make a good first «LinuxQuestions.org» question. I'm also open to any other suggestions on better/alternative methods...

Is it possible to redirect from a file into an actual for loop???

My script is as follows:
Thanks

You can redirect to a file with > which will overwrite the contents of a file.
You can redirect to a file with >> which will append the contents to a file.
You can redirect from a file with < which will take input from a file.

So, your for loop might work like this: for Fsckit in < /root/tmp/FsckEm.txt

Using a redirect is relatively easy except for one thing; for loops do word-breaking when reading the input. The way to work around that is to set your IFS to newline first.
I'd still recommend a while loop instead though as the standard way to handle input-by-line. In addition to not having to muck about with IFS, "$(..)" process substitution branches off a sub-shell.

I think I should clarify a bit more.

Unlike the while+read combination, a for loop can't directly accept input from stdin, which is why a file redirect alone doesn't work (although you can redirect stdout from it to a file). It simply processes a list of "words", either whatever follows "in" on the line, or the "$@" parameters if you don't specify anything.

What's considered a "word" separator is of course determined by your IFS shell variable. This is space+tab+newline by default. So in the above example changing it to newline means spaces and tabs will be ignored, and each "word" will actually be a whole line.

Anyway, to feed the loop you have to first build your word list somehow and insert it where for can read them. In the above I used process substitution. And now that I think about it, here's another way that uses an array instead, along with bash version 4's new mapfile built-in.
No need to worry about IFS or subshells here, since mapfile works on a per-line basis like read.

1 members found this post helpful.

Well That is the short answer. And explains why almost all the examples I've found use the while loop. <sigh> Thanks!

To be specific, even the while loop isn't reading directly from stdin. It's the read built-in that's doing that. It's just that while evaluates the command you give it during every iteration, whereas for only takes in the next "word" in the list.

http://www.tldp.org/LDP/abs/html/loops1.html