App::cpanminus - get, unpack, build and install modules from CPAN
Run cpanm -h for more options.
cpanminus is a script to get, unpack, build and install modules from CPAN.
Why? It's dependency free, requires zero configuration, and stands alone. When running, it requires only 10MB of RAM.
There are Debian packages, RPMs, FreeBSD ports, and packages for other operation systems available. If you want to use the package management system, search for cpanminus and use the appropriate command to install. This makes it easy to install cpanm to your system without thinking about where to install, and later upgrade.
If you want to build the latest from source,
git clone git://github.com/miyagawa/cpanminus.git
make install # or sudo make install if you're non root
This will install cpanm to your bin directory like /usr/local/bin (unless you configured INSTALL_BASE with local::lib), so you might need to sudo. Later you can say cpanm --self-upgrade --sudo to upgrade to the latest version.
cpanm --self-upgrade --sudo
chmod +x cpanm
# edit shebang if you don't have /usr/bin/env
just works, but be sure to grab the new version manually when you upgrade (--self-upgrade might not work).
perl 5.8 or later.
'tar' executable (bsdtar or GNU tar version 1.22 are rcommended) or Archive::Tar to unpack files.
C compiler, if you want to build XS modules.
make, if you want to reliably install MakeMaker based modules
Module::Build (core in 5.10) to install Build.PL based modules
OK, the first motivation was this: the CPAN shell runs out of memory (or swaps heavily and gets really slow) on Slicehost/linode's most affordable plan with only 256MB RAM. Should I pay more to install perl modules from CPAN? I don't think so.
First of all, I have no intention to dis CPAN or CPANPLUS developers. Don't get me wrong. They're great tools I've used for literally years (you know how many modules I have on CPAN, right?). I really respect their efforts of maintaining the most important tools in the CPAN toolchain ecosystem.
However, for less experienced users (mostly from outside the Perl community), or even really experienced Perl developers who know how to shoot themselves in their feet, setting up the CPAN toolchain often feels like yak shaving, especially when all they want to do is just install some modules and start writing code.
In particular, here are the few issues I've observed:
They ask too many questions and do not provide enough sane defaults. A normal user doesn't (and shouldn't have to) know what's the right answer for the question Parameters for the 'perl Build.PL' command? 
Parameters for the 'perl Build.PL' command? 
They provide very noisy output by default.
They fetch and rebuild their indexes almost every day, and this takes time.
... and they hog 200MB of memory and thrashes/OOMs on my 256MB VPS
cpanminus is designed to be very quiet (but logs all output to ~/.cpanm/build.log) and to pick the sanest possible defaults without asking any questions -- to just work.
Note that most of these problems with existing tools are rare, or are just overstated. They might already be fixed, or can be configured to work nicer. For instance, the latest CPAN.pm dev release has a much better first time configuration experience than ever before.
I know there's a reason for them to have many options and questions, because they're meant to work everywhere for everybody.
Of course I should have contributed back to CPAN/CPANPLUS instead of writing a new client, but CPAN.pm is nearly impossible (for anyone other than andk or xdg) to maintain (that's why CPANPLUS was born, right?) and CPANPLUS is a huge beast for me to start working on.
Yeah, I think my brain has been damaged since I looked at PyPI, gemcutter, pip and rip. They're quite nice and I really wanted something as nice for CPAN which I love.
Imagine you don't have CPAN or CPANPLUS. You search for a module on the CPAN search site, download a tarball, unpack it and then run perl Makefile.PL (or perl Build.PL). If the module has dependencies you probably have to resolve those dependencies by hand before doing so. Then you run the unit tests and make install (or ./Build install).
cpanminus automates that.
It queries the CPAN Meta DB site running on Google AppEngine at http://cpanmetadb.appspot.com/. The site is updated every hour to reflect the latest changes from fast syncing mirrors. The script then also falls back to the site http://search.cpan.org/. I've been talking to and working with with the QA/toolchain people for building a more reliable CPAN DB website.
Fetched files are unpacked in ~/.cpanm. You can configure this with the PERL_CPANM_HOME environment variable.
It installs to wherever ExtUtils::MakeMaker and Module::Build are configured to (via PERL_MM_OPT and MODULEBUILDRC). So if you're using local::lib, then it installs to your local perl5 directory. Otherwise it installs to the siteperl directory.
cpanminus at a boot time checks whether you have configured local::lib, or have the permission to install modules to the sitelib directory. If neither, it automatically sets up local::lib compatible installation path in a perl5 directory under your home directory. To avoid this, run the script as the root user, with --sudo option or with --local-lib option.
This local::lib automatic integration is still considered alpha and in the work -- more bootstrapping is under development. Stay tuned.
I tested installing MojoMojo, Task::Kensho, KiokuDB, Catalyst, Jifty and Plack using cpanminus and the installations including dependencies were mostly successful. More than half of CPAN behaves really nicely and appears to work.
However, there might be some distributions that will miserably fail, because of nasty edge cases. Here are some examples:
Packages uploaded to PAUSE in 90s which don't live under the standard authors/id/A/AA directory hierarchy.
Distributions with a Makefile.PL or Build.PL that asks you questions without using prompt function. However cpanminus has a mechanism to kill those questions with a timeout, and you can always say --interactive to make the configuration interactive.
Distributions that do not shipped with META.yml file but do require some specific version of toolchain for configuration.
Distributions that have a META.yml file that is encoded in YAML 1.1 format using YAML::XS. This will be eventually solved once we move to META.json.
cpanminus intends to work for 99.9% of modules on CPAN for 99.9% of people. It may not be perfect, but it should just work in most cases.
If this tool doesn't work for your very rare environment, then I'm sorry, but you should use CPAN or CPANPLUS, or build and install modules manually.
If you have CPAN or CPANPLUS working then you may want to keep using CPAN or CPANPLUS in the longer term, but I hope this can be a quite handy alternative to them for people in other situations. And apparently, many people love (at least the idea of) this software :)
Copyright 2010- Tatsuhiko Miyagawa
The standalone executable contains the following modules embedded.
Same as Perl.
Patches and code improvements were contributed by:
Goro Fuji, Kazuhiro Osawa, Tokuhiro Matsuno, Kenichi Ishigaki, Ian Wells, Pedro Melo, Masayoshi Sekimura, Matt S Trout, squeeky, horus and Ingy dot Net.
Bug reports, suggestions and feedbacks were sent by, or general acknowledgement goes to:
Jesse Vincent, David Golden, Andreas Koenig, Jos Boumans, Chris Williams, Adam Kennedy, Audrey Tang, J. Shirley, Chris Prather, Jesse Luehrs, Marcus Ramberg, Shawn M Moore, chocolateboy, Chirs Nehren, Jonathan Rockway, Leon Brocard, Simon Elliott, Ricardo Signes, AEvar Arnfjord Bjarmason, Eric Wilhelm, Florian Ragwitz and xaicron.
This software is provided "as-is," without any express or implied warranty. In no event shall the author be held liable for any damages arising from the use of the software.
CPAN CPANPLUS pip
To install App::cpanminus, copy and paste the appropriate command in to your terminal.
perl -MCPAN -e shell
For more information on module installation, please visit the detailed CPAN module installation guide.