FAQ

Here you’ll find answers to some frequently asked questions.

Breaking changes in tox 4

Version 4 of tox should be mostly backwards compatible with version 3. Exceptions and guidance for migrating are given in the upgrading doc.

New features in tox 4

Here is a non-exhaustive list of these.

  • You can now build wheel(s) instead of a source distribution during the packaging phase by using the wheel setting for the package setting. If your package is a universal wheel you’ll likely want to set the wheel_build_env to .pkg to avoid building a wheel for every Python version you target.

  • Editable wheel support was added as defined by PEP 660 via the package setting to editable.

  • We redesigned our CLI interface, we no longer try to squeeze everything under single command, instead now we have multiple sub-commands. For backwards compatibility if you do not specify a subcommand we’ll assume you want the tox 3 legacy interface (available under the legacy subcommand), for now the list of available commands are:

    subcommands:
      tox command to execute (by default legacy)
    
      {run,r,run-parallel,p,depends,de,list,l,devenv,d,config,c,quickstart,q,exec,e,legacy,le}
        run (r)                   run environments
        run-parallel (p)          run environments in parallel
        depends (de)              visualize tox environment dependencies
        list (l)                  list environments
        devenv (d)                sets up a development environment at ENVDIR based on the tox configuration specified
        config (c)                show tox configuration
        quickstart (q)            Command line script to quickly create a tox config file for a Python project
        exec (e)                  execute an arbitrary command within a tox environment
        legacy (le)               legacy entry-point command
    

    The exec and depends are brand new features. Other subcommands are a more powerful versions of previously existing single flags (e.g. -av is now succeeded by the list subcommand). All subcommands have a one or two character shortcuts for less typing on the CLI (e.g. tox run can be abbreviated to tox r). For more details see tox - CLI interface.

  • Startup times should be improved because now we no longer eagerly load all configurations for all environments, but instead these are performed lazily when needed. Side-effect of this is that if you have an invalid configuration will not be picked up until you try to use it.

  • We now discover your package dependency changes (either via PEP 621 or otherwise via PEP 517 prepare_metadata_for_build_wheel/build_wheel metadata). If new dependencies are added these will be installed on the next run. If a dependency is removed we’ll recreate the entire environment. This works for requirements files within the deps. This means that you should never need to use --recreate flag, tox should be smart enough to figure out when things change and automatically apply it.

  • All tox defaults can now be changed via the user level config-file (see help message output for its location, can be changed via TOX_CONFIG_FILE environment variable).

  • All tox defaults can now be changed via an environment variable: TOX_ prefix followed by the settings key, e.g. TOX_PACKAGE=wheel.

  • Any configuration can be overridden via the CLI -x or --override flag, e.g. tox run -e py311 -x testenv:py311.package=editable would force the packaging of environment py311 to be an editable install independent what’s in the configuration file.

  • basepython is now a list, the first successfully detected python will be used to generate python environment.

  • We now have support for inline tox plugins via the toxfile.py at the root of your project. At a later time this will allow using Python only configuration, as seen with nox.

  • You can now group tox environments via labels configuration, and you can invoke all tox environments within a label by using the -m label CLI flag (instead of the -e list_of_envs).

  • You can now invoke all tox environments within a given factor via the -f factor CLI flag.

Using a custom PyPI server

By default tox uses pip to install Python dependencies. Therefore to change the index server you should configure pip directly. pip accepts environment variables as configuration flags, therefore the easiest way to do this is to set the PIP_INDEX_URL environment variable:

set_env =
  PIP_INDEX_URL = https://tox.wiki/pypi/simple

It’s considered a best practice to allow the user to change the index server rather than hard code it, allowing them to use for example a local cache when they are offline. Therefore, a better form of this would be:

set_env =
  PIP_INDEX_URL = {env:PIP_INDEX_URL:https://tox.wiki/pypi/simple}

Here we use an environment substitution to set the index URL if not set by the user, but otherwise default to our target URI.

Using two PyPI servers

When you want to use two PyPI index servers because not all dependencies are found in either of them use the PIP_EXTRA_INDEX_URL environment variable:

set_env =
  PIP_INDEX_URL = {env:PIP_INDEX_URL:https://tox.wiki/pypi/simple-first}
  PIP_EXTRA_INDEX_URL = {env:PIP_EXTRA_INDEX_URL:https://tox.wiki/pypi/simple-second}

If the index server defined under PIP_INDEX_URL does not contain a package, pip will attempt to resolve it also from the URI from PIP_EXTRA_INDEX_URL.

Warning

Using an extra PyPI index for installing private packages may cause security issues. For example, if package1 is registered with the default PyPI index, pip will install package1 from the default PyPI index, not from the extra one.

Using constraint files

Constraint files are a type of artifact, supported by pip, that define not what requirements to install but instead what version constraints should be applied for the otherwise specified requirements. The constraint file must always be specified together with the requirement(s) to install. While creating a test environment tox will invoke pip multiple times, in separate phases:

  1. If deps is specified, it will install a set of dependencies before installing the package.

  2. If the target environment contains a package (the project does not have package skip or skip_install is true), it will:

    1. install the dependencies of the package.

    2. install the package itself.

Starting in tox 4.4.0, when constrain_package_deps = true is set in the test environment, {env_dir}{/}constraints.txt will be generated during install_deps based on the package specifications listed under deps. These constraints are subsequently passed to pip during the install_package_deps stage, causing an error to be raised when the package dependencies conflict with the test environment dependencies.

For stronger guarantees, set use_frozen_constraints = true in the test environment to generate the constraints file based on the exact versions enumerated by the list_dependencies_command (pip freeze). When using frozen constraints, if the package deps are incompatible with any previously installed dependency, an error will be raised. To use constraints with url, path, or “extras” (.[tests]) deps, then you must use_frozen_constraints, as these types of deps are not valid constraints as specified (see pypa/pip#8210).

Note constraint files are a subset of requirement files. Therefore, it’s valid to pass a constraint file wherever you can specify a requirement file.

Platform specification

Assuming the following layout:

tox.ini      # see below for content
setup.py     # a classic distutils/setuptools setup.py file

and the following tox.ini content:

[tox]
min_version = 2.0  # platform specification support is available since version 2.0
envlist = py{310,39}-{lin,mac,win}

[testenv]
# environment will be skipped if regular expression does not match against the sys.platform string
platform = lin: linux
           mac: darwin
           win: win32

# you can specify dependencies and their versions based on platform filtered environments
deps = lin,mac: platformdirs==3
       win: platformdirs==2

# upon tox invocation you will be greeted according to your platform
commands=
   lin: python -c 'print("Hello, Linus!")'
   mac: python -c 'print("Hello, Tim!")'
   win: python -c 'print("Hello, Satya!")'

You can invoke tox in the directory where your tox.ini resides. tox creates two virtualenv environments with the python3.10 and python3.9 interpreters, respectively, and will then run the specified command according to platform you invoke tox at.

Ignoring the exit code of a given command

When multiple commands are defined within the commands configuration field tox will run them sequentially until one of them fails (by exiting with non zero exit code) or all of them are run. If you want to ignore the status code of a given command add a - prefix to that line (similar syntax to how the GNU make handles this):

[testenv]
commands =
  - python -c 'import sys; sys.exit(1)'
  python --version

You can also choose to provide a ! prefix instead to purposely invert the exit code, making the line fail if the command returned exit code 0. Any other exit code is considered a success.

[testenv]
commands =
  ! python -c 'import sys; sys.exit(1)'
  python --version

Customizing virtual environment creation

By default tox uses the virtualenv to create Python virtual environments to run your tools in. To change how tox creates virtual environments you can set environment variables to customize virtualenv. For example, to provision a given pip version in the virtual environment you can set VIRTUALENV_PIP or to enable system site packages use the VIRTUALENV_SYSTEM_SITE_PACKAGES:

[testenv]
setenv =
  VIRTUALENV_PIP==22.1
  VIRTUALENV_SYSTEM_SITE_PACKAGES=true

Consult the virtualenv project for supported values (any CLI flag for virtualenv, in all upper case, prefixed by the VIRTUALENV_ key).

Building documentation with Sphinx

It’s possible to orchestrate the projects documentation with tox. The advantage of this is that now generating the documentation can be part of the CI, and whenever any validations/checks/operations fail while generating the documentation you’ll catch it within tox.

We don’t recommend using the Make and Batch file generated by Sphinx, as this makes your documentation generation platform specific. A better solution is to use tox to setup a documentation build environment and invoke sphinx inside it. This solution is cross platform.

For example if the sphinx file structure is under the docs folder the following configuration will generate the documentation under .tox/docs_out/index.html and print out a link to the generated documentation:

[testenv:docs]
description = build documentation
basepython = python3.10
deps =
  sphinx>=4
commands =
  sphinx-build -d "{envtmpdir}{/}doctree" docs "{toxworkdir}{/}docs_out" --color -b html
  python -c 'print(r"documentation available under file://{toxworkdir}{/}docs_out{/}index.html")'

Note here we also require Python 3.10, allowing us to use f-strings within the sphinx conf.py.

Building documentation with mkdocs

It’s possible to orchestrate the projects documentation with tox. The advantage of this is that now generating the documentation can be part of the CI, and whenever any validations/checks/operations fail while generating the documentation you’ll catch it within tox.

It’s best to define one environment to write/generate the documentation, and another to deploy it. Use the config substitution logic to avoid duplication:

[testenv:docs]
description = Run a development server for working on documentation
deps =
  mkdocs>=1.3
  mkdocs-material
commands =
  mkdocs build --clean
  python -c 'print("###### Starting local server. Press Control+C to stop server ######")'
  mkdocs serve -a localhost:8080

[testenv:docs-deploy]
description = built fresh docs and deploy them
deps = {[testenv:docs]deps}
commands = mkdocs gh-deploy --clean

Understanding InvocationError exit codes

When a command executed by tox fails, it always has a non-zero exit code and an InvocationError exception is raised:

ERROR: InvocationError for command
       '<command defined in tox.ini>' (exited with code 1)

Generally always check the documentation for the command executed to understand what the code means. For example for pytest you’d read here. On unix systems, there are some rather common exit codes. This is why for exit codes larger than 128, if a signal with number equal to <exit code> - 128 is found in the signal module, an additional hint is given:

ERROR: InvocationError for command
       '<command>' (exited with code 139)
Note: this might indicate a fatal error signal (139 - 128 = 11: SIGSEGV)

The signal numbers (e.g. 11 for a segmentation fault) can be found in the “Standard signals” section of the signal man page. Their meaning is described in POSIX signals. Beware that programs may issue custom exit codes with any value, so their documentation should be consulted.

Sometimes, no exit code is given at all. An example may be found in pytest-qt issue #170, where Qt was calling abort() instead of exit().

Access full logs

If you want to access the full logs you need to write -q and -v as individual tox arguments and avoid combining them into a single one.

Running within a Docker container

If you want to run tox within a Docker container you can use 31z4/tox. This Docker image neatly packages tox along with common build dependencies (e.g., make, gcc, etc) and currently active CPython versions. See more details in its GitHub repository.

The recommended way of using the image is to mount the directory that contains your tox configuration files and your code as a volume. Assuming your project is within the current directory of the host, use the following command to run tox without any flags:

docker run -v `pwd`:/tests -it --rm 31z4/tox

Because an entry point of the image is tox, you can easily pass subcommands and flags:

docker run -v `pwd`:/tests -it --rm 31z4/tox run-parallel -e black,py311

Note, that the image is configured with a working directory at /tests.

If you want to install additional Python versions/implementations or Ubuntu packages you can create a derivative image. Just make sure you switch the user to root when needed and switch back to tox afterwards:

FROM 31z4/tox

USER root

RUN set -eux; \
    apt-get update; \
    DEBIAN_FRONTEND=noninteractive \
    apt-get install -y --no-install-recommends \
        python3.12; \
    rm -rf /var/lib/apt/lists/*

USER tox

Testing end-of-life Python versions

tox uses virtualenv under its hood for managing virtual environments. Virtualenv 20.22.0 dropped support for all Python versions smaller or equal to Python 3.6.

If you need to test against e.g. Python 2.7, 3.5 or 3.6, you need to add the following requires statement to your tox.ini configuration files.

[tox]
requires = virtualenv<20.22.0

In case you need to do this for many repositories, we recommend to use all-repos.

Support for Python 3.7 was dropped in virtualenv 20.27.0. In order to test against Python 3.7, you can limit the version with requires = virtualenv<20.27.0 instead.

Testing with Pytest

Running pytest from tox can be configured like this:

[tox]
envlist = py311, py312

[testenv]
commands = pytest

If required, tox positional arguments can be passed through to pytest:

[testenv]
commands = pytest {posargs}

When running tox in parallel mode (tox run-parallel (p)), care should be taken to ensure concurrent pytest invocations are fully isolated.

This can be achieved by setting pytest’s base temporary directory to a unique temporary directory for each virtual environment as provided by tox:

[testenv]
commands = pytest --basetemp="{env_tmp_dir}"

Setting the pytest --basetemp argument also causes all temporary pytest files to be deleted immediately after the tests are completed. To restore the default pytest behavior to retain temporary files for the most recent pytest invocations, the system’s temporary directory location could be configured like this instead:

[tox]
set_env =
  TEMP = {env_tmp_dir}

[testenv]
commands = pytest