• vf new [<options>] <envname> - Create a virtual environment.

  • vf ls [--details] - List the available virtual environments.

  • vf activate <envname> - Activate a virtual environment. (Note: Doesn’t use the script provided by Virtualenv.)

  • vf deactivate - Deactivate the current virtual environment.

  • vf upgrade [<options>] [<envname(s)>] - Upgrade virtual environment(s).

  • vf rm <envname> - Delete a virtual environment.

  • vf tmp [<options>] - Create a temporary virtual environment with a randomly generated name that will be removed when it is deactivated.

  • vf cd - Change directory to currently-activated virtual environment.

  • vf cdpackages - Change directory to currently-active virtual environment’s site-packages.

  • vf globalpackages - Toggle system site packages.

  • vf addpath - Add a directory to this virtual environment’s sys.path.

  • vf all <command> - Run a command in all virtual environments sequentially.

  • vf connect [<envname>] - Connect the current working directory with the currently active (or specified) virtual environment. This requires the auto-activation plugin to be enabled in order to have any effect besides creating a .venv file in the current directory.

If you are accustomed to virtualenvwrapper commands (workon, etc.), you may wish to enable the Virtualenvwrapper Compatibility Aliases (compat_aliases) plugin.

Using Different Pythons

By default, the environments you create with VirtualFish will use the same Python version that was originally used to Pip-install VirtualFish, which will usually be your system’s default Python interpreter.

If you want to create a new virtual environment with a different Python interpreter, add the --python PYTHON_EXE (-p for brevity) flag to vf new, where PYTHON_EXE is any Python executable. For example:

vf new -p /usr/bin/python3 my_python3_env

Specifying the full path to the Python executable avoids ambiguity and is thus the most reliable option, but if the target Python executable is on your PATH, you can save a few keystrokes and pass the bare executable instead:

vf new -p pypy my_pypy_env

Sometimes there may be Python interpreters on your system that are not on your PATH, with full filesystem paths that are long and thus hard to remember and type. VirtualFish makes dealing with these easier by automatically detecting and using Python interpreters in a few known situations, in the following order:

  1. asdf Python plugin is installed and has built the specified Python version.

  2. Pyenv is installed and has built the specified Python version.

  3. Pythonz is installed and has built the specified Python version.

  4. Homebrew keg-only versioned Python executable (e.g., 3.8) found at: /usr/local/opt/python@3.8/bin/python3.8

For asdf, Pyenv, and Pythonz , in addition to passing option flags such as -p python3.8 or -p python3.9.0a4, you can even get away with specifying just the version numbers, such as -p 3.8 or -p 3.9.0a4.

Upgrading Virtual Environments

Virtual environments contain links to Python interpreters that can become outdated over time. In addition, sometimes the underlying Python interpreter can be removed by Python upgrades, putting the virtual environment into an unusable state. Thankfully, VirtualFish includes a mechanism for upgrading outdated/broken environments.

To understand which environments might be outdated/broken, run:

vf ls --details

You can maintain a list of target Python versions via a line such as the following in a ~/.tool-versions file:

python 3.9.7 3.8.12 3.7.11 3.6.14

Environment Python versions that match one of those versions will be shown as up-to-date (green). If target Python versions are not specified in that file, VirtualFish compares environment Python versions to the current default Python version, as specified by the VIRTUALFISH_DEFAULT_PYTHON variable (see below), if defined. To perform a minor (point-release) upgrade to the currently-active virtual environment, run:

vf upgrade

Minor point-release upgrades will modify in-place the virtual environment’s Python version number and symlinks. (While this should work correctly in the majority of cases, there is the possibility that future changes to virtual environment structure will interfere with this in-place upgrade.)

For major version upgrades, say from Python 3.8.x to 3.9.x, you must instead re-build the environment via:

vf upgrade --rebuild

Re-building an environment will record its current package versions, remove the old environment, create a new environment with the same name, and re-install the list of recorded package versions.

If VirtualFish determines that a virtual environment is in a broken state, it will re-build that environment, even if --rebuild is omitted.

To upgrade to a specific Python interpreter or version, use the --python option:

vf upgrade --rebuild --python /usr/local/bin/python3.8

Virtual environments need not be active in order to upgrade them. To upgrade one or more virtual environments, specify their names:

vf upgrade project1 project2

Upgrades can also be applied to all environments. To re-build all existing environments:

vf upgrade --rebuild --all

Configuration Variables

The vf install […] installation step writes the VirtualFish loader to a file at $XDG_CONFIG_HOME/fish/conf.d/, which on most systems defaults to: ~/.config/fish/conf.d/

You can edit this file to, for example, change the plugin loading order. You can also add the following optional variables at the top, so that they are set before is sourced.

  • VIRTUALFISH_HOME (default: ~/.virtualenvs) - where all your virtual environments are kept.

  • VIRTUALFISH_DEFAULT_PYTHON - The default Python interpreter to use when creating a new virtual environment; the value should be a valid argument to the Virtualenv --python flag.

Regardless of the changes that you make, you must run exec fish afterward if you want those changes to take effect for the current shell session.