How do I "use" a Perl module in a directory not in @INC?

I have a module in the parent directory of my script and I would like to 'use' it.

If I do

use '../Foo.pm';

I get syntax errors.

I tried to do:

push @INC, '..';
use EPMS;

and .. apparently doesn't show up in @INC

I'm going crazy! What's wrong here?

use takes place at compile-time, so this would work:

BEGIN {push @INC, '..'}
use EPMS;

But the better solution is to use lib, which is a nicer way of writing the above:

use lib '..';
use EPMS;

In case you are running from a different directory, though, the use of FindBin is recommended:

use FindBin;                     # locate this script
use lib "$FindBin::RealBin/..";  # use the parent directory
use EPMS;

There are several ways you can modify @INC.

• set PERL5LIB, as documented in perlrun

• use the -I switch on the command line, also documented in perlrun. You can also apply this automatically with PERL5OPT, but just use PERL5LIB if you are going to do that.

• use lib inside your program, although this is fragile since another person on a different machine might have it in a different directory.

• Manually modify @INC, making sure you do that at compile time if you want to pull in a module with use. That's too much work though.

• require the filename directly. While this is possible, it doesn't allow that filename to load files in the same directory. This would definitely raise eyebrows in a code review.

Personally I prefer to keep my modules (those that I write for myself or for systems I can control) in a certain directory, and also to place them in a subdirectory. As in:

/www/modules/MyMods/Foo.pm
/www/modules/MyMods/Bar.pm

And then where I use them:

use lib qw(/www/modules);
use MyMods::Foo;
use MyMods::Bar;

As an aside.. when it comes to pushing, I prefer the fat-arrow comma:

push @array => $pushee;

But that's just a matter of preference.

'use lib' is the answer, as @ephemient mentioned earlier. One other option is to use require/import instead of use. It means the module wouldn't be loaded at compile time, but instead in runtime.

That will allow you to modify @INC as you tried there, or you could pass require a path to the file instead of the module name. From 'perldoc -f require':

> If EXPR is a bareword, the require assumes a ".pm" extension and > replaces "::" with "/" in the filename for you, to make it easy to > load standard modules. This form of loading of modules does not risk > altering your namespace.

You have to have the push processed before the use is -- and use is processed early. So, you'll need a BEGIN { push @INC, ".."; } to have a chance, I believe.

As reported by "perldoc -f use":

> It is exactly equivalent to
> BEGIN { require Module; import Module LIST; }
> except that Module must be a bareword.

Putting that another way, "use" is equivalent to:

• running at compile time,
• converting the package name to a file name,
• require-ing that file name, and
• import-ing that package.

So, instead of calling use, you can call require and import inside a BEGIN block:

BEGIN {
  require '../EPMS.pm';
  EPMS->import();
}

And of course, if your module don't actually do any symbol exporting or other initialization when you call import, you can leave that line out:

BEGIN {
  require '../EPMS.pm';
}

Some IDEs don't work correctly with 'use lib', the favored answer. I found 'use lib::relative' works with my IDE, JetBrains' WebStorm.

see POD for lib::relative

The reason it's not working is because what you're adding to @INC is relative to the current working directory in the command line rather than the script's directory.

For example, if you're currently in:

a/b/

And the script you're running has this URL:

a/b/modules/tests/test1.pl

---

BEGIN {
	unshift(@INC, "..");	
}

The above will mean that .. results in directory a/ rather than a/b/modules.

Either you must change .. to ./modules in your code or do a cd modules/tests in the command line before running the script again.