Skip to content

Latest commit

 

History

History
235 lines (189 loc) · 13.2 KB

README.md

File metadata and controls

235 lines (189 loc) · 13.2 KB

Action Status

Coverage Status

Quickstart with a C++ project template

Take a look at this C++ project template that applies all the following instructions, but also shows how to create a pure workflow without using special GitHub action that you cannot run locally on your development machine, but directly using the tools (CMake, Ninja, vcpkg, C++ compilers) you already use daily.

The run-vcpkg action restores from cache vcpkg along with the previously installed packages, and then setup vcpkg to be run in a subsequent step. Or it runs it for you.

Special features which provide added value over a pure workflow are:

  • automatic cache management of vcpkg's artifacts by computing a cache key based (among others) on content of vcpkg.json.
  • automatic computation of the set of keys (primary key and secondary keys) used maximize the reuse of existing cached vcpkg's artifacts.
  • automatic dump of log files created by CMake (e.g., CMakeOutput.log) and vcpkg. The content of those files flow into the workflow output log. Customizable by the user.
  • automatic parsing of CMake, vcpkg and gcc, clang, msvc errors, reporting them contextually in the workflow summary by means of annotations.

The provided samples use GitHub hosted runners.

Good companions are the run-cmake action and the get-cmake actions.


Quickstart with instructions

It is highly recommended to use:

jobs:
  build:
    steps:
      #-uses: actions/cache@v1   <===== YOU DO NOT NEED THIS!

      # Install latest CMake.
      - uses: lukka/get-cmake@latest
      # Or pin to a specific CMake version:
      # lukka/get-cmake@v3.21.2

      # Restore from cache the previously built ports. If a "cache miss" occurs,
      # then vcpkg is bootstrapped. Since a the vcpkg.json is being used later on
      # to install the packages when `run-cmake` runs, no packages are installed at
      # this time.
      - name: Restore artifacts, or setup vcpkg (do not install any package)
        uses: lukka/run-vcpkg@v10 # Always specify the specific _version_ of the
                                  # action you need, `v10` in this case to stay up
                                  # to date with fixes on v10 branch.
        #with:
          # This is the default location of the directory containing vcpkg sources.
          # Change it to the right location if needed.
          # vcpkgDirectory: '${{ github.workspace }}/vcpkg'

          # If not using a submodule for vcpkg sources, this specifies which commit
          # id must be checkout from a Git repo. It must not set if using a submodule
          # for vcpkg.
          # vcpkgGitCommitId: '${{ matrix.vcpkgCommitId }}'

          # This is the glob expression used to locate the vcpkg.json and add its
          # hash to the cache key. Change it to match a single manifest file you want
          # to use.
          # vcpkgJsonGlob: '**/vcpkg.json'

          # This is needed to run `vcpkg install` command (after vcpkg is built) in
          # the directory where vcpkg.json has been located. Default is false,
          # It is highly suggested to let `run-cmake` to run vcpkg (i.e. `false`)
          # (i.e. let CMake run `vcpkg install`) using the vcpkg.cmake toolchain.
          # runVcpkgInstall: true

      - name: Run CMake consuming CMakePreset.json and vcpkg.json by mean of vcpkg.
        uses: lukka/run-cmake@v10
        with:
          # This is the default path to the CMakeLists.txt along side the
          # CMakePresets.json. Change if you need have CMakeLists.txt and CMakePresets.json
          # located elsewhere.
          # cmakeListsTxtPath: '${{ github.workspace }}/CMakeLists.txt'

          # This is the name of the CMakePresets.json's configuration to use to generate
          # the project files. This configuration leverages the vcpkg.cmake toolchain file to
          # run vcpkg and install all dependencies specified in vcpkg.json.
          configurePreset: 'ninja-multi-vcpkg'

          # This is the name of the CMakePresets.json's configuration to build the project.
          buildPreset: 'ninja-multi-vcpkg'

    #env:
    #  By default the action disables vcpkg's telemetry by defining VCPKG_DISABLE_METRICS. 
    #  This behavior can be disabled by defining `VCPKG_ENABLE_METRICS` as follows.
    #  VCPKG_ENABLE_METRICS: 1 
    #
    #  [OPTIONAL] Define the vcpkg's triplet you want to enforce, otherwise the default one
    #  for the hosting system will be automatically choosen (x64 is the default on all platforms,
    #  e.g. `x64-osx`).
    #  VCPKG_DEFAULT_TRIPLET: ${{ matrix.triplet }} 

Action reference: all input/output parameters

Description of all input parameters: action.yml

Flowchart

Flowchart with related input in action.yml which let customize the flow.

┌──────────────────────────┐
│  Compute cache key from: │   Inputs:
│  - vcpkg Git commit      │   - `appendedCacheKey`
│  - platform and OS       │   - `vcpkgGitCommitId`
│  - user provided key     │
└─────────────┬────────────┘
              │
              ▼
 ┌─────────────────────────┐
 │ Locate vcpkg.json.      │   Inputs:
 │ If found, add its hash  │   - `vcpkgJsonGlob`
 │ to cache key            │   _ `vcpkgJsonIgnores`
 └────────────┬────────────┘
              │
              ▼
 ┌─────────────────────────┐   Inputs:
 │ Restore vcpkg and       │   - `doNotCache`
 │ packages from cache     │   - `vcpkgDirectory`
 │ to cache key            │   - `binaryCachePath`
 └────────────┬────────────┘   Environment variable:
              │                - `VCPKG_DEFAULT_BINARY_CACHE`: where previous built packages
              ▼                  are going to be restored
 ┌─────────────────────────┐
 │ If vcpkg is not a       │   Inputs:
 │ submodule, fetch it     │   - `vcpkgGitCommitId`
 │                         │   - `vcpkgGitURL`
 └────────────┬────────────┘   - `doNotUpdateVcpkg`
              │                - `vcpkgDirectory`
              ▼
 ┌─────────────────────────┐
 │ Rebuild vcpkg executable│   Inputs:
 │ if not in sync with     │   - `vcpkgGitCommitId`
 │ sources                 │   - `vcpkgGitURL`
 └────────────┬────────────┘
              │
              ▼
  <Is `runVcpkgInstall:true`>┐    Inputs:
          ────┬────        No│   - `runVcpkgInstall`
              │ Yes          │
              ▼              │
 ┌─────────────────────────┐ │
 │ Launch `vcpkg install`  │ │   Inputs:
 │ where vcpkg.json has    │ │   - `runVcpkgFormatString`
 │ been located            │ │   Environment variables:
 └────────────┬────────────┘ │   - `VCPKG_DEFAULT_TRIPLET` is used. If not yet
              │              │     set, it is set to the current platform.
              ▼              │
 ┌─────────────────────────┐ │
 │ Set `VCPKG_ROOT` and    │ │
 │ `VCPKG_DEFAULT_TRIPLET` │ │
 │ workflow-wide env vars  │ │
 └────────────┬────────────┘ │
              ├───────────── ┘
              ▼
 ┌───────────────────────────┐
 │ Schedule this step at     │
 │ end of the workflow:      │
 │┌─────────────────────────┐│
 │| If no cache-hit,        ││  Inputs:
 ││ store `VCPKG_ROOT` and  ││  - `binaryCachePath`
 ││ binary packages in cache││  - `doNotCache`
 ││                         ││  Environment variables:
 │└────────────┬────────────┘│  - RUNVCPKG_NO_CACHE: Any step before the post action
 └────────────┬──────────────┘    may set this environment variable to disable saving
              ▼                   the cache.
              ⬬

Best practices

Use vcpkg as a submodule of your repository

It is highly suggested to pin the specific version of vcpkg you want to use to keep a consistent development experience between local and remote build environments. This is accomplished by using vcpkg as submodule of your Git repository; this way the version of vcpkg used is implied by the commit id specified by the submodule for vcpkg.

Use vcpkg's vcpkg.json file to specify the dependencies

The vcpkg.json is a manifest file that declaratively specifies the dependencies to be installed. The file is being used automatically by running CMake (e.g. by using run-cmake) when:

  • starting CMake with the vcpkg.cmake toolchain file.
  • the root CMake source directory contains a vcpkg.json file.

Or it can also be used by invoking vcpkg install in a directory where vcpkg.json is located (e.g., input runVcpkgInstall : true).

When conditions are satisfied, the toolchain execution starts vcpkg to install the packages declared in the manifest file.

Putting this manifest-like file under source control is highly recommended as this helps to run vcpkg the same exact way locally and remotely on the build servers.* The dependencies specified in the vcpkg.json file are installed when CMake runs (i.e. at run-cmake execution time).

Samples

Checkmarks indicates whether the samples "uses" or specifies the thing in the header or whether it is true.

workflow link vcpkg as submodule explicit triplet vcpkg toolchain CMake's Presets Ninja run-vcpkg runs vcpkg CMake runs vcpkg
link
link
link
link

License

All the content in this repository is licensed under the MIT License.

Copyright © 2019-2020-2021 Luca Cappa

Disclaimer

The software is provided as is, there is no warranty of any kind. All users are encouraged to improve the source code with fixes and new features.

Contributing

Read CONTRIBUTING.md