I'm trying to write a shell script to monitor certain files and display a warning when they reach a certain size. The idea being to give a warning BEFORE the ulimit is hit! I've fallen at the first hurdle however, how can you RELIABLY get JUST the file size? I'm trying variations along the following lines: ls -l filename|cut -d" " -f5 However on my Unixware 2.1.3 (and on 7) system the cut seems to give different results depending on the file listed. Presumably there are different numbers of spaces in the resulting line? My other option is to use "cut -c35-41". However, being really pedantic, I'd like the script not to be platform specific so I can't rely on the character positions. Any ideas?

Try this .... ls -l "filename" | awk '{print $5}'

Correct on the varying number of spaces. The cut command is using each space as a field delimiter, so a bunch of spaces makes for a bunch of null fields. By default, awk delimites each field with any amount of whitespace. I check a directory for large files with: ls -l | awk '$5>999999' And the read command uses whitespace, by default, as its delimiter: ls -l fname | read x x x x fsize remainder echo $fsize