http://www.perlmonks.org?node_id=1016621


in reply to Variable as module name?

You cannot use use in this context because it is wrapped in an implicit BEGIN block (see BEGIN, UNITCHECK, CHECK, INIT and END), and thus gets evaluated at compile time and before you could initialize the value of your variable. The usual work around here is to use a string eval, a la: 

my $module_name = 'Foo';
eval "use $module_name;1;" or die "Couldn't use $module_name: $@";

[download]
Of course, this has security implications, since you are evaluating arbitrary code, depending on where your value comes from.

But it complains that I'm calling "new" on an unblessed reference when I do my $object = $module->new(); on the next line.

This should not be happening. The following code works for me (assuming an OO-module Foo in your path): 

#!/usr/bin/perl
use strict;
use warnings;
use Data::Dumper;

my $module = 'Foo';
my $module_path = "$module.pm";
$module_path =~ s|::|/|g;
require $module_path;

print Dumper $module->new;

[download]

Note the s|::|/|g to allow non-root modules. Also note that by requiring you are bypassing the import mechanism.

#11929 First ask yourself `How would I do this without a computer?' Then have the computer do it the same way.

Replies are listed 'Best First'.
Re^2: Variable as module name?
by Anonymous Monk on Feb 01, 2013 at 23:38 UTC
[reply]
Re^2: Variable as module name?
by Cody Fendant (Hermit) on Feb 03, 2013 at 01:03 UTC
    Thank you very much, that's all working fine now. My unblessed reference problem was an unrelated error in my module.
[reply]

In Section Seekers of Perl Wisdom