title: vcpkg-- A C++ package manager for Windows, Linux and MacOS

description: "vcpkg is a command line package manager that greatly simplifies the acquisition and installation of open-source C++ libraries on Windows."

ms.date: "05/16/2019"

vcpkg is a command-line package manager that greatly simplifies the acquisition and installation of third-party libraries on Windows, Linux and MacOS. If your project uses third-party libraries, we recommend that you use vcpkg to install them. vcpkg supports both open-source and proprietary libraries. All libraries in the vcpkg Windows catalog have been tested for compatibility with Visual Studio 2015, Visual Studio 2017, and Visual Studio 2019. As of May 2018 there are over 900 libraries in the Windows catalog and over 350 in the Linux/MacOS catalog. The C++ community is adding more libraries to both catalogs on an ongoing basis.

With a single command, you can download sources and build a library. vcpkg is itself an open-source project, available on GitHub. You can customize your private clone(s) in any way you like. For example, you can specify different libraries, or different versions of libraries than what are found in the public catalog. You can have multiple clones of vcpkg on a single machine, each one producing custom sets of libraries and/or compilation switches, etc. Each clone is a self-contained environment with its own copy of vcpkg.exe that operates only on its own hierarchy. vcpkg is not added to any environment variables, and has no dependency on the Windows Registry or Visual Studio.

For libraries in the Windows catalog, vcpkg downloads sources instead of binaries[1]. It compiles those sources using the most recent version of Visual Studio that it can find. In C++, it is very important that any libraries you use are complied with the same compiler, and compiler version, as the application code that links to it. By using vcpkg, you eliminate or at least greatly reduce the potential for mismatched binaries and the problems they can cause. In teams that are standardized on a specific version of a compiler, one team member can use vcpkg to download sources and compile a set of binaries and then use the export command to zip up the binaries and headers for other team members. For more information, see [Export compiled binaries and headers](#export_binaries_per_project) below.

If you create a vcpkg clone with private libraries in the ports collection, you can add a port that downloads prebuilt binaries and headers and write a portfile.cmake file that simply copies those files to the desired location.

[1] *Note: for some proprietary libraries, sources are not available. Vcpkg will download compatible prebuilt binaries in these cases.*

This command enumerates the control files in the vcpkg/ports subfolders. You will see a listing like this:

If the portfile specifies dependencies, vcpkg downloads and installs those also. After downloading, vcpkg builds the library by using whatever build system the library uses. CMake and (on Windows) MSBuild projects are preferred, but MAKE is supported along with any other build system. If vcpkg cannot find the specified build system on the local machine, it downloads and installs it.

After you have installed some libraries, you can use **vcpkg list** to see what you have:

Run **vcpkg integrate install** to configure Visual Studio to locate all vcpkg header files and binaries on a per-user basis without the need for manual editing of VC++ Directories paths. If you have multiple clones, the clone from which you run this command becomes the new default location.

Now you can #include headers simply by typing the folder/header, and auto-complete assists you. No additional steps are required for linking to libs or adding project references. The following illustration shows how Visual Studio finds the azure-storage-cpp headers. vcpkg places its headers in the **/installed** subfolder, partitioned by target platform. The following diagram shows the list of include files in the **/was** subfolder for the library:


If you need to use a specific version of a library that is different from the version in your active vcpkg instance, follow these steps:

Run **vcpkg integrate install** to configure Visual Studio Code on Linux/MacOS with the location of the vcpkg enlistment and enable IntelliSense on source files.

You can produce Linux binaries from a Windows machine by using the Windows Subsystem for Linux (WSL). Follow the instructions to [Set up WSL on Windows 10](/windows/wsl/install-win10), and configure it with the [Visual Studio extension for Linux](https://blogs.msdn.microsoft.com/vcblog/2017/02/08/targeting-windows-subsystem-for-linux-from-visual-studio/). You can put all your built libraries for both Windows and Linux into the same folder and access it from both Windows and WSL.

Requiring everyone on a team to download and build libraries can be inefficient. A single team member can do that work, and then use **vcpkg export** to create a zip file of the binaries and headers, or a NuGet package (various format available), that can be easily shared with other team members.

The public catalog is kept up-to-date with the latest versions of the libraries. To determine which of your local libraries are out-of-date, use **vcpkg update**. When you're ready to update your ports collection to the latest version of the public catalog, run the **vcpkg upgrade** command to automatically download and rebuild any or all of your installed libraries that are out of date.

By default, the **upgrade** command only lists the libraries that are out of date; it doesn’t upgrade them. To perform the upgrade, use the **--no-dry-run** option.

The following example shows how to upgrade only specified libraries. Note that vcpgk automatically pulls in dependencies as necessary.

Type **vcpkg remove** to remove an installed library. If any other libraries depend on it, you are asked to rerun the command with **--recurse**, which causes all downstream libraries to be removed.

You can modify your clone of vcpkg in any way you like. You can create multiple vcpkg clones and modify the portfiles in each one to obtain specific versions of libraries or specify command-line parameters. For example, in an enterprise, one group of developers might be working on software that has one set of dependencies, and another group might have a different set. You can set up two clones of vcpkg, and modify each one to download the versions of the libraries and the compilation switches, and so on, according to your needs.

All vcpkg functionality and data is self-contained in a single directory hierarchy, called an "instance". There are no registry settings or environment variables. You can have any number of instances of vcpkg on a machine and they do not interfere with each other.

- downloads -- cached copies of any downloaded tools or sources. vcpkg searches here first when you run the install command .

- installed-- Contains the headers and binaries for each installed library. When you integrate with Visual Studio, you are essentially telling it add this folder to its search paths.

|**vcpkg search [pat]**|Search for packages available to install|

|**vcpkg hash \<file> [alg]**|Hash a file by specific algorithm, default SHA512|

|**vcpkg export \<pkg>... [opt]...**|Export a package|

|**vcpkg create \<pkg> \<url> [archivename]**|Create a new package|