nave

Virtual Environments for Node.

Switch out your node version and global npm install space in one command. Supports named environments. Uses subshells by default so that you can ^D or exit out of an environment quickly.

No need to edit .bashrc or .profile, just install and go.

Installation

If you want a global nave command, you could install this thing with npm. But that's not really necessary. You can run the nave.sh shell script from here, or symlink it wherever you want, or even just download the script and just put it where you want it.

curl https://raw.githubusercontent.com/isaacs/nave/main/nave.sh > /usr/local/bin/nave
chmod 0755 /usr/local/bin/nave

with npm

If you have npm, presumably you already have Node, so it's a tiny bit silly, but maybe you like installing the top-level Node some other way, and install your subshell version switcher with npm. Why is a bash program in npm anyway? It's fine. Bits don't judge.

npm install -g nave

Usage

To use a version of node, you do this:

nave use <some version>

If you want to name a virtual env, you can do this:

nave use <some name>

If that virtual env doesn't already exist, it'll prompt you to choose a version.

Both of these commands drop you into a subshell. Exit the shell with exit or ^D to go back from whence you came.

Here's the full usage statement:

Usage: nave <cmd>

COMMANDS

  install <version>     Install the version specified (ex: 12.8.0)
  install <name> <ver>  Install the version as a named env
  use <version>         Enter a subshell where <version> is being used
  use <ver> <program>   Enter a subshell, and run "<program>", then exit
  use <name> <ver>      Create a named env, using the specified version.
                        If the name already exists, but the version differs,
                        then it will update the link.
  usemain <version>     Install in /usr/local/bin (ie, use as your main nodejs)
  clean <version>       Delete the source code for <version>
  uninstall <version>   Delete the install for <version>
  ls                    List versions currently installed
  ls-remote             List remote node versions
  ls-all                List remote and local node versions
  latest                Show the most recent dist version
  cache                 Clear or view the cache
  help                  Output help information
  auto                  Find a .naverc and then be in that env
                        If no .naverc is found, then alias for 'nave exit'
  auto <dir>            cd into <dir>, then find a .naverc, and be in that env
                        If no .naverc is found, then alias for 'nave exit' in
                        the specified directory.
  auto <dir> <cmd>      cd into <dir>, then find a .naverc, and run a command
                        in that env
                        If no .naverc is found, then alias for 'nave exit <cmd>'
                        in the specified directory.
  should-auto           Exits with 1 if the nave auto env already
                        matches the config, or 0 if a change should
                        be made (ie, by calling 'nave auto')
                        An explicit call to 'nave use' or 'nave exit' will
                        tell nave that it should NOT auto.
  get <variable>        Print out various nave config values.
  exit                  Unset all the NAVE environs (use with 'exec')
  exit <cmd>            Run the specified command in a nave-free environment
                        (Note that nave will still set NAVE_EXIT=1 in order to
                        prevent 'nave should-auto' from evaluating true.)

VERSION STRINGS

  Any command that calls for a version can be provided any of the
  following "version-ish" identifies:

  - x.y.z       A specific SemVer tuple
  - x.y         Major and minor version number
  - x           Just a major version number
  - lts         The most recent LTS (long-term support) node version
  - lts/<name>  The latest in a named LTS set. (argon, boron, etc.)
  - lts/*       Same as just "lts"
  - latest      The most recent (non-LTS) version
  - stable      Backwards-compatible alias for "lts".

  To exit a nave subshell, type 'exit' or press ^D.
  To run nave *without* a subshell, do 'exec nave use <version>'.
  To clear the settings from a nave env, use 'exec nave exit'

ENVIRONMENT VARIABLES

  The following environment variables can be set to change nave's behavior.

  NAVE_DIR        Root directory for nave to operate in.  Defaults to
                  $XDG_CONFIG_HOME/nave if set (eg, ~/.config/nave), or
                  ~/.nave otherwise.
  NAVE_NPX        Set this to '1' to add node_modules/.bin to the PATH
  NAVE_DEBUG      Set this to '1' to run in debug mode.
  NAVE_CACHE_DUR  Duration in seconds to cache version information (86400)
  NAVEUA          User-agent header to send when fetching version information
  NAVE_SRC_ONLY   Set to '1' to *only* build node from source, and never use
                  binary distributions.  (This is much slower!)
  NAVE_JOBS       Set to the number of JOBS to use when building node.
                  Defaults to the number of CPUs on the system.
  NODEDIST        The distribution server to fetch node from.  Defaults to
                  https://nodejs.org/dist
  NAVE_CONFIG     Arguments to pass to ./configure when building from source.

  Nave sets the following environment variables when in use:

  NAVE            A descriptive string of the nave setting in use.
  NAVENAME        The name, in named subshells, otherwise $NAVEVERSION
  NAVEVERSION     The version of node in use.
  NAVELVL         The number of subshells currently in use (like bash $SHLVL)
  NAVE_LOGIN      '1' in interactive nave subshells, '0' otherwise.
  NAVE_ROOT       Location of nave installed environments
  NAVE_SRC        Location of downloaded Node.js source
  NAVE_AUTO_RC    The .naverc file used by 'nave auto'
  NAVE_AUTO_CFG   The contents of the .naverc file used by 'nave auto'

CONFIGURATION FILES

  Nave subshells will source the same .bashrc, .bash_profile, .zprofile, etc.
  configuration files as normal shells, based on whether it is being run as a
  login shell, or to run a specific command.

  In addition, the following files are sourced in all nave subshells if found,
  after the normal shell profile files, in the following order, based on the
  resulting environment variables described above. When run in a subdirectory,
  nave will walk up the directory tree looking for any of these that it finds,
  but will not walk up further than any folder containing a '.git' entry.

    .nave_profile_${NAVE}
    .nave_profile_${NAVENAME}, if a named environment
    .nave_profile_${NAVEVERSION}, eg .nave_profile_16.19.0
    .nave_profile_${NAVEVERSION major.minor}, eg .nave_profile_16.19
    .nave_profile_${NAVEVERSION major}, eg .nave_profile_16
    .nave_profile

  Finally, it will always source ${NAVEDIR}/../.naverc if present.
  (eg, ~/.config/.naverc)

  These may be used to set project-specific confirations, env variables, or
  other behavior based on the Nave environment in use, without the use of
  configuration files in the home directory.

  The 'nave auto' command will walk up the directory tree looking for a
  '.naverc' or '.nvmrc' file, and use the contents as arguments to 'nave use'.

Subshell-free operation

If you prefer to not enter a subshell, just run nave with exec

exec nave use lts/argon

You could even add something like this to your .bashrc file to save on typing:

n () {
  exec nave "$@"
}

Running shell script with specific version of Node.js

If there is need to run a shell script with version of node.js provided by nave following snippet can be inserted into script:

[ "${IN_SUBSHELL}" != "$0" ] && exec env IN_SUBSHELL="$0" nave use 5.0.0 bash "$0" "$@" || :

AUTOMAGICAL!

You can put a .naverc file in the root of your project (or anywhere). This file should contain the version that you want to use. It can be something like lts/boron or 16 or latest

echo lts/boron > ~/projects/my-project/.naverc

Then you can run nave auto to load the appropriate environment.

BUT THAT'S NOT NEARLY MAGICAL OR AUTO ENOUGH FOR ME THOUGH

If you want to get even more absurd/automated, put this in your bash settings (like ~/.bashrc or whatever)

alias cd='exec nave auto'

and then every time you cd into a different folder, it'll automatically load the correct nave settings, or exit nave-land if no automatic stuff could be found.

Note that doing this will also cause it to exit the nave environment when you cd to a directory that doesn't have a nave setting, so it can interfere with "normal" nave operation.

Also, aliasing cd is a very all-consuming type of change to make to one's system. You might wish to give it some other name, so that you can switch directories without affecting environment variables as a potentially surprising side effect, or even just run exec nave auto as an explicit action whenever you want this behavior to happen.

Bottom line, it's your shell, and I hope that this helps you enjoy it more :)

NO THAT'S TOO MAGICAL, BE JUST SLIGHTLY LESS MAGICAL THAN THAT

Ok, put this snippet in a PROMPT_COMMAND export in your bash profile (.bashrc or .bash_profile or whatever you use for that).

export PROMPT_COMMAND='nave should-auto && exec nave auto'

Now you'll always be in the configured nave environment in any project with a .naverc (or .nvmrc), and always not in a nave environment in your main shell in any folder that isn't set up for nave auto.

This has no effect on the normal nave subshells you get from nave use.

The output of your PROMPT_COMMAND is used for the main bash prompt, so you can also do some fancy stuff like this:

__prompt () {
  if nave should-auto; then
    exec nave auto
  if
  # Show the nave version in white-on-blue, but the "normal" node
  # version in green
  if [ "$NAVE" != "" ]; then
    echo -ne " \033[44;37mnode@$NAVE\033[0m"
  else
    echo -ne " \033[32mnode@$(node -p 'process.version.slice(1)' 2>/dev/null)\033[0m"
  fi
}
PS1="\\$ "
export PROMPT_COMMAND='__prompt'

env vars

• $NAVE The current shell. Either a version, or a name and version.
• $NAVE_NPX Set to "1" to add node_modules/.bin to the $PATH in all nave shells (including the main shell when exec nave auto is used).
• $NAVE_AUTO_RC The .naverc file found by nave auto.
• $NAVE_AUTO_CFG The contents of the .naverc file that was read when entering the nave auto environment.
• $NAVENAME The name of the current shell. Equal to $NAVEVERSION in unnammed environments.
• $NAVEVERSION The version of node that the current shell is pointing to. (This should comply with node -v.)
• $NAVELVL The level of nesting in the subshell.
• $NAVE_DEBUG Set to 1 to run nave in bash -x style.
• $NAVE_DIR Set to the location where you'd like nave to do its business. Defaults to ~/.nave.
• $NAVE_CONFIG Set this to an array of arguments to pass to ./configure. Defaults to ("--debug"). (Note that parens are required to supply multiple arguments. I use ("--debug" "--without-npm") on my own system, since I'm usually using nave to test npm, so installing it in the subshell doesn't help much.) This can be set in the ~/.naverc file, or in your normal ~/.bash{rc,_profile} files.
• $NAVE_JOBS If set, this will be the number of jobs to run when building node. If this isn't set, it'll use the $JOBS env, then try to guess a reasonable value based on the number of CPUs, then fall back on 2 if sysctl -n hw.ncpu fails.
• $NAVE_SRC_ONLY Set to "1" to only build from source, rather than fetching binaries.

Contributing

Patches welcome! Before spending too much time on a patch or feature request, please post an issue to see if it's something that's going to be accepted or have unforeseen consequences.

Patches will usually not be accepted if they break tests or cause coverage to drop below 100%. You can run tests with:

npm test
# or...
bash test/run.sh

And you can check coverage with:

npm run cov
# or...
COV=1 bash test/run.sh && open coverage-all/kcov-merged/nave.sh.*.html

The latest coverage report can be found at https://isaacs.github.io/nave/kcov-merged/index.html

Compatibility

Nave is a bash program. It can still do most of its functionality if you use zsh or fish as your shell, as long as you have bash somewhere, but some of the magical stuff won't work (since obviously that has to run inline in your shell with exec).

Nave requires bash. It will probably never work on Windows, or other systems lack a native Bourne Again Shell. Sorry. (Patches welcome if you can get it to work properly on Windows that do have bash, like WSL and Cygwin.)

Nave logins work with any shell, but executing a command in the nave environment (ie, nave use 12 node program.js) requires that your shell support the -c argument. (Bash, sh, zsh, and fish all work fine.)

Configuration

Nave will source ~/.naverc on initialization of a new subshell, if it exists and is readable.

You may control the place where nave puts things by setting the NAVE_DIR environment variable. However, note that this must be set somewhere other than ~/.naverc, since it needs to be set in the parent shell where the nave command is invoked.

By default, nave puts its stuff in ~/.nave/. If this directory does not exist and cannot be created, then it will attempt to use the location of the nave.sh bash script itself. If it cannot write to this location, then it will exit with an error.

Credits

nave borrows concepts, inspiration, and code from Tim Caswell's "nvm" and Kris Kowal's "sea" programs.

Sea is really nice, but is very tied to Narwhal. Also, it's a require.paths manager, which nave is not.

Nvm is also really nice, but has to be sourced rather than being run, and thus is a little bit wonky for some use cases. But it doesn't involve subshells, which makes it better for some others.