Dear Monks,

I am trying to execute bp_genbank2gff3.pl (bioperl) from inside a perl script. I just need it to convert a file between formats and I don't read the resulting files later in my script, but pass them to a java program. This is why I think it's OK to execute this script as-is (generally, I guess it would be "prettier" to call some bioeprl subroutine that does the same job as the script does, but I couldn't find a simple one).

Anyway, I am using a general helper subroutine that I wrote (run_command) as described below.

use strict;
use warnings;
use Cwd;
use 5.010;

# run command and print output to stdout, file, both or none.
# $print_to_stdout is boolean, $output_filename is optional
sub run_command {
my ( $command_string, $print_to_stdout, $output_filename ) = @_;

say "### run_command (from ", getcwd(), " ): ", $command_string, " ###
+";

# create output file if needed
my $output_fh;
if ( defined $output_filename ) {
    open $output_fh, '>', $output_filename
      or die "can't open $output_filename: $!";
    print $output_fh "### run_command: ", $command_string, " ###\n";
}

# execute command
open( my $command_out, "-|", $command_string );

if ( $print_to_stdout || defined $output_fh ) {
    while (<$command_out>) {
        print if ($print_to_stdout);
        print $output_fh $_ if ( defined $output_fh );
    }
}
close $command_out;    # wait until command has finsihed
close $output_fh if ( defined $output_fh );
}
[download]

And define:

my $command = "bp_genbank2gff3.pl -y -o out_dir some_genbank_filename"
+;
[download]

(replace out_dir and some_genbank_filename with real names)

Now, any of these will work fine (the files will be created in the output dir):

run_command( $command, 1, undef );
run_command( $command, 0, "some_file_name");
run_command( $command, 1, "some_file_name");
[download]

but this will not work (no error but also no files created):

run_command( $command, 0, undef );
[download]

Why? It's as if the script knows whether I'm keeping track of it's stdout.

Thank you, Dave