Check out

my $stdout_and_stderr1;
my $stdout_and_stderr2;
my $stdout_and_stderr3;

{
   print "running script1.pl\n";

   local *STDOUT;
   close(STDOUT);
   open(STDOUT, '>', \$stdout_and_stderr1) or die("...");

   local *STDERR;
   close(STDERR);
   open(STDERR, '>&STDOUT') or die("...");

   do 'script1.pl';
}

{
   print "running script2.pl\n";

   ...
[download]

The above requires 5.8(.0? .1?). IO::Scalar can be used with older versions of Perl, but it has limitations.

If you don't mind running the script in a seperate process, I think this works:

$stdout_and_stderr1 = `script1.pl 2>&1`;
$stdout_and_stderr2 = `script2.pl 2>&1`;
$stdout_and_stderr3 = `script3.pl 2>&1`;
[download]

Untested

There are couple of different ways, take a look at perlipc.

For example. Parent:

use strict;
use warnings;
    
my $pid = open(KID_TO_READ, "-|", "perl -w b.pl");
while (<KID_TO_READ>) {
    print "from parent: $_";
}
[download]

Child:

use strict;
use warnings;

print "Hello\n";
print "World\n";
[download]

This gives you:

from parent: Hello
from parent: World
[download]

However this does not handle STDERR, one way to resolve it is to do this in child:

use strict;
use warnings;

open STDERR, '>&STDOUT';

print "Hello\n";
print STDERR "World\n";
[download]

Yeah, it was intentional to run the scripts with the do command, are you saying that there may be advantages to running the scripts with a system command instead?
Also, I'm fairly new to perl, so I don't understand what you mean when you say I can dup STDOUT and STDERR

Thanks for your quick response though
Jonathan

Yes, do is a bit less useful than two different appoaches: modules, and completely separate scripts.

If your child scripts have any code in common with each other, and especially if they can share resources (files, data), it makes sense to write a module for each, and let everything run in equal standing in the same process. If, however, you want each task to really be a separate script, don't risk lefover values leaking over, and have the master invoke each script in a completely separate process.

(You can still share code by using modules if you run separate processes, of course: simply have them use the common code.)

As for duping, you can copy -- dup -- a filehandle over. perlopentut has details. It doesn't finish solving your original problem but it might help; alternatively pg's suggestion might be a better route.