nave

Virtual Environments for Node

Node.js version management
isaacs/nave

README

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.

  1. ```
  2. curl https://raw.githubusercontent.com/isaacs/nave/main/nave.sh > /usr/local/bin/nave
  3. chmod 0755 /usr/local/bin/nave
  4. ```

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.

  1. ```
  2. npm install -g nave
  3. ```

Usage


To use a version of node, you do this:

  1. ```
  2. nave use <some version>
  3. ```

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

  1. ```
  2. nave use <some name>
  3. ```

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:

  1. ```
  2. Usage: nave <cmd>

  4. COMMANDS

  6.   install <version>     Install the version specified (ex: 12.8.0)
  7.   install <name> <ver>  Install the version as a named env
  8.   use <version>         Enter a subshell where <version> is being used
  9.   use <ver> <program>   Enter a subshell, and run "<program>", then exit
  10.   use <name> <ver>      Create a named env, using the specified version.
  11.                         If the name already exists, but the version differs,
  12.                         then it will update the link.
  13.   usemain <version>     Install in /usr/local/bin (ie, use as your main nodejs)
  14.   clean <version>       Delete the source code for <version>
  15.   uninstall <version>   Delete the install for <version>
  16.   ls                    List versions currently installed
  17.   ls-remote             List remote node versions
  18.   ls-all                List remote and local node versions
  19.   latest                Show the most recent dist version
  20.   cache                 Clear or view the cache
  21.   help                  Output help information
  22.   auto                  Find a .naverc and then be in that env
  23.                         If no .naverc is found, then alias for 'nave exit'
  24.   auto <dir>            cd into <dir>, then find a .naverc, and be in that env
  25.                         If no .naverc is found, then alias for 'nave exit' in
  26.                         the specified directory.
  27.   auto <dir> <cmd>      cd into <dir>, then find a .naverc, and run a command
  28.                         in that env
  29.                         If no .naverc is found, then alias for 'nave exit <cmd>'
  30.                         in the specified directory.
  31.   should-auto           Exits with 1 if the nave auto env already
  32.                         matches the config, or 0 if a change should
  33.                         be made (ie, by calling 'nave auto')
  34.                         An explicit call to 'nave use' or 'nave exit' will
  35.                         tell nave that it should NOT auto.
  36.   get <variable>        Print out various nave config values.
  37.   exit                  Unset all the NAVE environs (use with 'exec')
  38.   exit <cmd>            Run the specified command in a nave-free environment
  39.                         (Note that nave will still set NAVE_EXIT=1 in order to
  40.                         prevent 'nave should-auto' from evaluating true.)

  42. VERSION STRINGS

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

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

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

  60. ENVIRONMENT VARIABLES

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

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

  79.   Nave sets the following environment variables when in use:

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

  91. CONFIGURATION FILES

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

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

  103.     .nave_profile_${NAVE}
  104.     .nave_profile_${NAVENAME}, if a named environment
  105.     .nave_profile_${NAVEVERSION}, eg .nave_profile_16.19.0
  106.     .nave_profile_${NAVEVERSION major.minor}, eg .nave_profile_16.19
  107.     .nave_profile_${NAVEVERSION major}, eg .nave_profile_16
  108.     .nave_profile

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

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

  117.   The 'nave auto' command will walk up the directory tree looking for a
  118.   '.naverc' or '.nvmrc' file, and use the contents as arguments to 'nave use'.
  119. ```

Subshell-free operation


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

  1. ```bash
  2. exec nave use lts/argon
  3. ```

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

  1. ```bash
  2. n () {
  3.   exec nave "$@"
  4. }
  5. ```

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:
  1. ```bash
  2. [ "${IN_SUBSHELL}" != "$0" ] && exec env IN_SUBSHELL="$0" nave use 5.0.0 bash "$0" "$@" || :
  3. ```

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

  1. ```
  2. echo lts/boron > ~/projects/my-project/.naverc
  3. ```

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)

  1. ```
  2. alias cd='exec nave auto'
  3. ```

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).

  1. ```bash
  2. export PROMPT_COMMAND='nave should-auto && exec nave auto'
  3. ```

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:

  1. ```bash
  2. __prompt () {
  3.   if nave should-auto; then
  4.     exec nave auto
  5.   if
  6.   # Show the nave version in white-on-blue, but the "normal" node
  7.   # version in green
  8.   if [ "$NAVE" != "" ]; then
  9.     echo -ne " \033[44;37mnode@$NAVE\033[0m"
  10.   else
  11.     echo -ne " \033[32mnode@$(node -p 'process.version.slice(1)' 2>/dev/null)\033[0m"
  12.   fi
  13. }
  14. PS1="\\$ "
  15. export PROMPT_COMMAND='__prompt'
  16. ```

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:

  1. ```
  2. npm test
  3. # or...
  4. bash test/run.sh
  5. ```

And you can check coverage with:

  1. ```
  2. npm run cov
  3. # or...
  4. COV=1 bash test/run.sh && open coverage-all/kcov-merged/nave.sh.*.html
  5. ```

The latest coverage report can be found at

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.