ry: the simplest ruby virtual env
If you're on a Mac with homebrew installed, you may simply
brew install ry and follow the instructions.
Otherwise, install the files from the repo to your favorite prefix (I use
~/.local, which is the default).
git clone git://github.com/jneen/ry cd ry PREFIX=$HOME/.local make install
and add the following lines to your bashrc (or zshrc):
# If you're using the default ~/.local prefix, # make sure ~/.local/bin is on your $PATH. export PATH="$HOME/.local/bin:$PATH" eval "$(ry setup)"
or, if you don't like
eval, you can do it manually:
For ZSH completion, see this.
If you want to specify a different directory for installing rubies:
# rubies are installed into $RY_PREFIX/lib/ry/rubies # set RY_RUBIES for an alternate location export RY_RUBIES="$HOME/.rubies"
Ry is a bit different from other version managers. The major design goal of ry is to be explicit, unobtrusive, and easy to query. In the vein of the n package manager for node, there are no subshells, and the only thing it needs to add to your environment is a single entry to your
$PATH (also tab completion if you like). For example, here's how you create a new installation:
ry install http://ftp.ruby-lang.org/pub/ruby/1.9/ruby-1.9.3-p125.tar.gz mri-1.9.3
This creates an installation of Ruby called
mri-1.9.3 using the tarball from
ruby-lang.org. To switch to this ruby, use
ry use mri-1.9.3 # or: ry mri-1.9.3
Want to use a custom commit?
ry install https://github.com/ruby/ruby/tarball/<hash> mri-custom
If you have
ruby-build installed, you can use it very easily with ry:
ry install 1.9.3-p125
shell-local ruby selection
ry use changes a symlink, that ruby is globally activated across all shells. To use a shell-local ruby, simply set your
$PATH to include that ruby's bin dir. The easiest way to do this is to use the
ry fullpath <ruby> helper, which outputs a copy of your
$PATH variable modified for use with ry.
export PATH="$(ry fullpath my-cool-ruby)"
You can activate this automatically with something like direnv.
For more information, see
All of the magic is in the bash script
bin/ry. Here are a couple of bash features I use that aren't common elsewhere:
Poor man's namespacing - the character
:is a perfectly valid character to use in a bash function's name. All of ry's subcommands are implemented as functions looking like
ry::foo. At the bottom of the file is the function
rywhich essentially delegates to
ry::$1- so to add a new subcommand, all you need to do is define the bash function and document it in
Piping from heredocs is awesome. The syntax
cmdwith stdin as the content of
$variable. I use this extensively, and you should too.