This GitHub Action can be used in a workflow to install Cygwin.

e.g.

- run: git config --global core.autocrlf input
- uses: actions/checkout@v6

- uses: cygwin/cygwin-install-action@v6
  with:
    packages: |
      git
      python3
      python3-pip

- run: bash tests/script.sh  # see note below

• Inputs

  • packages
  • allow-test-packages
  • work-vol
  • install-dir
  • platform
  • add-to-path
  • site
  • pubkeys
  • check-hash
  • check-installer-sig
  • check-sig

• Outputs

  • root
  • setup
  • package-cache

• Line endings

• PATH

• Symlinks

• Mirrors and signatures

A list of additional packages to install.

Example usage:

- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    packages: |
      git
      python3
      python3-pip

By default, packages marked test are not considered for installation.

Setting this input to 'true' will allow test packages to be found and installed.

- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    allow-test-packages: 'true'
    packages: |
      my-cool-package

The Windows volume to which the installer and packages are downloaded.

By default, work-vol is also the volume where Cygwin will be installed, but the install location can be customized using the install-dir input.

The default value is 'D:' for performance reasons.

Example:

- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    work-vol: 'C:' # This affects the Cygwin install directory, too.

The absolute path to the directory where Cygwin will be installed.

The default install directory is 'D:\cygwin', which is calculated using the work-vol input.

Example:

- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    install-dir: 'D:\cygwin64'

Select the architecture to install.

'x86_64' is the default value, but 'x86' can also be specified.

Example usage:

- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    platform: 'x86_64'

By default, Cygwin's /usr/bin directory is added to the system $PATH.

This behavior can be disabled by setting this input to 'false'.

- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    add-to-path: 'false'

Mirror sites to install from, separated by whitespace.

The default site is selected based on the value of the platform:

Example:

- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    site: |
      https://mirrors.kernel.org/sourceware/cygwin/
      https://domain.example/path/

Absolute paths to extra public key files (RFC4880 format).

Example:

# Prior to reaching this step in the workflow
# you would need to download the keys referenced below.
- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    pubkeys: |
      D:\key1.pub
      D:\key2.pub

By default, the hash of the Cygwin installer will be verified.

This behavior can be disabled by setting this input to 'false'.

Example:

- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    check-hash: 'true'  # Change to 'false' only if required.

By default, the Authenticode signature of the installer file will be verified.

This behavior can be disabled by setting this input to 'false'.

Example:

- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    check-installer-sig: 'true'  # Change to 'false' only if required.

By default, the installer will verify the signature of the package manifest.

This behavior can be disabled by setting this input to 'false'.

Example:

- uses: 'cygwin/cygwin-install-action@<version>'
  with:
    check-sig: 'true'  # Change to 'false' only if required.

The root directory of the Cygwin installation.

By default, this value will be 'D:\cygwin', but it is affected by the work-vol and install-dir options.

The absolute path to the Cygwin installer.

The absolute path to the package cache directory.

If you're going to use actions/checkout in your workflow, you should precede that with

- run: git config --global core.autocrlf input

to ensure that any shell scripts etc. in your repository don't get checked out with \r\n line endings (leading to '\r': command not found errors).

Likewise, if you have multiple lines of shell script in a YAML block for run: in your workflow file, the file this is written into on the runner ends up with \r\n line endings.

You can use >- (rather than |) to ensure that it doesn't contain any newlines.

Alternatively, you can invoke bash with -o igncr.

Warning: Putting igncr in the SHELLOPTS environment variable seems like it should have the same effect, but this can have unintended side-effects (by default, SHELLOPTS is a shell variable and moving it to the environment causes all shell options to propagate to child shells).

By default, this action prepends Cygwin's /usr/bin directory to the PATH.

If you do not want the system PATH changed, set add-to-path to 'false'.

However, if you want to ensure that PATH only contains Cygwin executables, and other stuff installed in the VM image isn't going to get picked up:

• Set PATH to something like /usr/bin:$(cygpath ${SYSTEMROOT})/system32 in your shell script

or,

• Put CYGWIN_NOWINPATH=1 into the environment
• start a login shell with bash --login
• because the profile script does cd ${HOME}, either:
  • cd ${GITHUB_WORKSPACE} in your shell script, or
  • prevent the profile script from changing directory by putting CHERE_INVOKING into the environment

Cygwin's installer creates Cygwin-style symlinks by default, and some executables (e.g. python) are symlinks.

Since CMD and PowerShell don't understand those symlinks, you cannot run those executables directly in a run: in your workflow. Execute them via bash or env instead.

Alternatively, putting e.g. CYGWIN=winsymlinks:native into the workflow's environment works, since the installer now honours that.

You probably don't need to change the setting for site, and you shouldn't change check-hash, check-installer-sig, or check-sig unless you're very confident it's appropriate and necessary.

These options are very unlikely to be useful except in some very isolated circumstances, such as using the Cygwin Time Machine.