For simple use cases, writing a cpanfile
is a good choice. A sample file might look like
requires 'Marpa::R2', '2.078';
requires 'String::Escape', '2010.002';
requires 'Moo', '1.003001';
requires 'Eval::Closure', '0.11';
on test => sub {
requires 'Test::More', '0.98';
};
That is, it's actually a Perl script, not a data format. The dependencies can then be installed like
$ cd /path/to/your/module
$ cpanm --installdeps .
This does not install your module! But it makes sure that all dependencies are satisfied, so we can do:
use lib '/path/to/your-module/lib'; # add the location as a module search root
use Your::Module; # works! yay
This is usually sufficient e.g. for a git repository which you want others to tinker with.
If you want to create a tarball that can be distributed and installed easily, I'd recommend Dist::Zilla
(although it's geared towards CPAN releases). Instead of a cpanfile
we use a dist.ini
:
name = Your-Module
version = 1.2.3
author = Your Self <you@example.com>
license = GPL_3
copyright_holder = Your Self
[@Basic]
[Prereqs]
Marpa::R2 = 2.078
String::Escape = 2010.002
Moo = 1.003001
Eval::Closure = 0.11
[Prereqs / TestRequires]
Test::More = 0.98
Then:
$ dzil test # sanity checks, and runs your tests
$ dzil build # creates a tarball
Dist::Zilla takes care of creating a Makefile.PL
and other infrastructure that is needed to install the module.
You can then distribute that tarball, and install it like cpanm Your-Module-1.2.3.tar.gz
. Dependencies are resolved, your packages are copied to a permanent location, and you can now use Your::Module
in any script without having to specify the location.
Note that you should adhere to the standard directory layout for Perl modules:
./
lib/
Your/
Module.pm # package Your::Module
Module/
Helper.pm # package Your::Module::Helper
t/ # tests to verify the module works on the target syste,
foo.t
bar.t
xt/ # optional: Author tests that are not run on installation
baz.t
bin/ # optional: scripts that will later end up in the target system's $PATH
command-line-tool