Install Vesta#

In development. There are no published binaries yet, so for now the way to try Vesta is to build it from source.

Once they exist, the packages for Windows, Linux and macOS will appear here alongside their checksum and signature.

No binaries does not mean unusable. On Windows the project even builds its own installer with a single command.

Build from source

1. Dependencies#

Keystone, Capstone and LibPEparse ship as submodules and clone themselves. Nothing else is required.

2. Clone and build#

git clone --recursive https://github.com/vesta-lang/vesta.git
cd vesta

cmake -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build -j

On Windows with TDM-GCC you need to ask for the MinGW generator:

cmake -G "MinGW Makefiles" -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build -j

With CMake 4 or newer you may need to add -DCMAKE_POLICY_VERSION_MINIMUM=3.5, because one of the submodules declares a minimum older than that version accepts.

The resulting executable is build/vm. Once installed it is called vesta.

3. Check that it works#

./build/vm --version

4. Your first program#

Save this as hello.vx:

i32 main() {
    println("Hello from Vesta ${1 + 1}!");
    return 0;
}

Then build it, from the repository root (see the note on the standard library below):

./build/vm --vx hello.vx -o hello
./build/vm --run hello.velb

The same source can be compiled into a standalone native executable, without the virtual machine:

./build/vm --vx hello.vx -m aot -o hello
./hello

There is no make install. Project packaging is Windows-only today, so on Linux and macOS installing means copying files by hand. It is four commands.

Copying the binary alone is not enough. The compiler resolves the standard library, the native plugins and the preprocessor headers relative to its own location, so they have to travel with it.

sudo mkdir -p /usr/local/lib/vesta
sudo cp build/vm            /usr/local/lib/vesta/vesta
sudo cp -r stdlib           /usr/local/lib/vesta/
sudo cp -r preprocessor/include_lib /usr/local/lib/vesta/

That leaves a layout which satisfies all three lookups at once:

/usr/local/lib/vesta/
    vesta              the compiler
    stdlib/vx/         standard library modules
    stdlib/native/     native plugins
    include_lib/       preprocessor headers

And to call it from anywhere, a symbolic link:

sudo ln -s /usr/local/lib/vesta/vesta /usr/local/bin/vesta
vesta --version

The link works because on Linux the compiler finds out where it lives by reading /proc/self/exe, which resolves the link and returns the real path.

On macOS, avoid the symbolic link. There the executable path is obtained differently and may come back as the link itself, which would send the compiler looking for its resources in /usr/local/bin. Add the directory to your PATH instead:

echo 'export PATH="/usr/local/lib/vesta:$PATH"' >> ~/.zshrc

This is the easiest route on Windows. The project builds its own installer through CMake, and downloads NSIS by itself if you do not have it.

cmake --build build --target installer

That leaves a VestaVM-<version>-win64.exe which takes care of everything: it copies the standard library and the native plugins next to the executable, adds Vesta to the PATH and creates the Start Menu shortcut.

From then on vesta works from any shell and finds its resources by itself. No environment variables, no working from one particular folder.

vesta --version

If you would rather not install anything, there is a portable ZIP with the same file layout:

cmake --build build --target installer-zip