Bash I/O redirection problem

R00ts · 10-26-2006, 02:48 PM

I'm having an issue with a bash script I wrote yesterday that I can't figure out. I need to run a series of experiments, so I wrote this script which calls the Perl script that performs the experiment. The Perl script writes the experimental results to stdout, so I redirect that output to a file that is named individually for each experiment.

The problem is that the redirection on the `eval` line in the else block seems to be directing all output to the same file. I've tried removing the eval, and then what I believe happened is that the files captured the output from this bash script, but the output from the Perl script was nowhere to be found.

Here is the code below:

Code:

### Subroutine which runs an experiment for each 
# Parameter #1: The location of the performance data to use
# Parameter #2: The location of the results set to use
# Parameter #3: The experiment name, used also in the name of the output file
perform_experiment() {

	for i in `seq 1 9`
	do
		out_file="$out_dir/$3.${name_opts[i]}.cdf"	
		if [ -f $out_file ];
		then
			echo "WARNING: File $out_file exists! I will not overwrite it"
		else
			echo "$run_exe -d $1 -r $2 ${score_opts[i]} > $out_file"
			eval "$run_exe -d $data_dir/ -r $results_dir/ ${score_opts[i]} > $out_file"
		fi
	done
}

I call this function multiple times to perform each experiment. My question is: what do I need to do to get the `eval` statement to properly re-direct the output from that invocation of the Perl script to stdout? I'm stumped.

Thanks

makyo · 10-26-2006, 07:50 PM

Hi.

I think the problem is elsewhere, and that you do not need eval.

Here's a simplified version of your function and a driver:

Code:

#!/bin/sh

# @(#) s1       Demonstrate output re-direction.

### Subroutine which runs an experiment for each
# Parameter #1: The location of the performance data to use
# Parameter #2: The location of the results set to use
# Parameter #3: The experiment name, used also in the name of the output file
perform_experiment() {

  for i in `/usr/bin/seq 1 9`
  do
    out_file="t${i}"
    echo " Outfile for iteration $i is :$out_file:"
    if [ -f $out_file ];
    then
      echo "WARNING: File $out_file exists!"
    else
      echo echo hi >$out_file
    fi
  done
}

# Remove debris, create t9 as a test.

rm -f t[1-9]
touch t9
echo
echo " Files before function call:"
ls

echo
echo " Run function."
perform_experiment

echo
echo " Files after  function call:"
ls

Which produces:

Code:

% ./s1

 Files before function call:
RCS  Readme.txt  s1  t9

 Run function.
 Outfile for iteration 1 is :t1:
 Outfile for iteration 2 is :t2:
 Outfile for iteration 3 is :t3:
 Outfile for iteration 4 is :t4:
 Outfile for iteration 5 is :t5:
 Outfile for iteration 6 is :t6:
 Outfile for iteration 7 is :t7:
 Outfile for iteration 8 is :t8:
 Outfile for iteration 9 is :t9:
WARNING: File t9 exists!

 Files after  function call:
RCS  Readme.txt  s1  t1  t2  t3  t4  t5  t6  t7  t8  t9

I recommend that you put some intermediate debugging echo statements in the function to see what is going on ... cheers, makyo

( edit 1: typo )