Shell Programming and Scripting

I have been working on a script that executes on a number of different operating systems. As a result I was trying to set a variable or perhaps variable array depending on the OS. I tried the following using eval and such but so far have not had any luck. Is there a way to do something like the following.

DF="df -h | sed -e :a -e '$!N;s/\n / /;ta' -e 'P;D' | /usr/xpg4/bin/grep -ivE "filesystem|$FSIGNORE" | awk '{ print $6" "$5}'"

Later in the script do this:

$DF | while read LINE; do ...

I tried a host of different things, like putting eval into the variable and escaping the quotes but it just did not seem to work correctly. Any ideas, or am I better off writing the output of the command to a temp file and reading from the file?

Thanks in advance.

Why not put all of that in a function which you can then call?

Code:

DF()
{
    df -h | sed .....
}

DF | while read ...

also if you want portability don't expect all platforms to have /usr/xpg4/bin/...

I already have the df command broken out for every OS ( I realize I will have to make it different on each one. ) This was only two lines from the entire script.

I will try putting each version of the df into a function, that might work. Thanks for the reply. Any other suggestions are welcome.

I like porter's idea of using a function, but if you wanted to do it like you've specified, you'd put it in backticks.

Code:

DF=`df -h | sed -e :a -e '$!N;s/\n / /;ta' -e 'P;D' | /usr/xpg4/bin/grep -ivE "filesystem|$FSIGNORE" | awk '{ print $6" "$5}'`

I know about backticks, but that will not work in this case. The output of the command returns on one line. Since I process the data with read line and there could be any number of filesystems that method is not useful. I needed a way to execute the command into the pipe. The function should work nicely, I will try in the morning. The other alternative would be to hold the data in a file and read each line in.

Thanks for your help!