To be honest, I'd use something like DBD::CSV for this.

One reason is to let someone else handle it for me, and the other reason is that anything that's in a CSV file today makes a perfect candidate to moving to an RDBMS later. It may be a bit slow, but I'll take slow and correct over fast and 7 headers any day of the week :-)

• You want to create a single output csv file, based on a set of distinct csv input files.
• For each input file, you are going to run your script once, and this will append data to the common output file
• (Or, for each input file, you will do one iteration of a loop in your script to append to the output file? This seems like a reasonable approach.)
• You need to make sure that the line containing the "csv header" (listing of column names) is written only once, at the very beginning of the output file.

If that's a fair description of the problem, then you could use the following logic (expressed here in pseudocode):

open a given input file
open the output file
read first line of input (this will be the csv header)
look at the size of the output file
if output file size is zero, write the csv header to output
loop through remaining lines of input file
   write lines to output when appropriate
[download]

The easiest way to deal with your program is to make the ERROR-file by hand (just typing the header-line in a text-editor or so) and never, ever write the header to it, just the lines you extracted from the other files.

Sometimes automation just gets in the way.

#!/usr/bin/perl -w

# filename: catskiphead
# Merge files skipping the first line from the seocnd file onwards
my $line = 1;
while(<>) {
    next if ($line++ == 0);
    $line = 0 if eof;
    print ($_);
}
[download]

file1:
header
1
2
3
file2:
header
4
5
6
file3:
header
7
8
9
<p><c>Usage: catskiphead file1 file2 file3
[download]

Output

header
1
2
3
4
5
6
7
8
9
[download]

cheers

SK

PS: Note that you can change the print to take file handles and also do any checking on the input records before you write out.