comm with a variable

Shell Programming and Scripting
1

Hello all, I have two flat files that are colon delineated and I am trying to run a compare (Solaris v8 ksh) of $1 within a script to access a mysql database based on the results. Unix is telling me that it has to have physical file names. Is there a way to run a compare using variables? This is what I have tried. Thanks in advance for your responses.

$> comm -13 $( awk -F: '{ print $1 }' testfile.lst | sort | uniq ) $( awk -F: '{ print $1 }' testfile.lst.old | sort | uniq ) > results.lst
usage: comm [-123] file1 file2

then I tried:

$> file1=$( awk -F: '{ print $1 }' $testfile.lst | sort | uniq )
$> file2=$( awk -F: '{ print $1 }' $testfile.lst.old | sort | uniq )
$> comm -13 $file1 $file2 > results.lst
usage: comm [-123] file1 file2

2

try echoing the value of $file1 and $file2, I bet each produces more than one filename.

You can only comm two files, not two lists of files.

3

Well I echo the results of:
file1=$( awk -F: '{ print $1 }' $testfile.lst | sort | uniq )
and it's all on one line instead of in individual rows in a single column. I guess it would be possible to massage the results, but it will probably just be easier to output it to a physical file, instead. Crummy, I like to keep things clean, instead of having tmp files all over the place that need to be deleted etc. Oh well, as long as it works right?
As always thanks for the post, I appreciate everyone's interest in helping me solve problems.

4

it works with GNU bash(version 2.05b) on gentoo.
try:

comm -13 <(cut -d: -f1 $testfile.lst | sort | uniq) <(cut -d: -f1 $testfile.lst.old | sort | uniq) >results.lst