Quickstart

This guide will teach you how to setup hlb and run a build to output hello world.

Installation

If you're on a MacOS or Linux (linux-amd64), head on over to Releases to grab a static binary.

Otherwise, you can compile HLB yourself using go:

go get -u github.com/openllb/hlb/cmd/hlb

You'll also need to run buildkitd somewhere you can connect to. The easiest way if you have Docker, is to run a local buildkit container:

# We're still waiting on some upstream PRs to be merged, but soon you'll be able to use standard moby/buildkit
docker run -d --name buildkitd --privileged openllb/buildkit:experimental

Run your first build

Now that you have installed hlb, we can run our first build. Typically, we will write our program in a file with a .hlb extension, but for our first build we can just pipe the program in from stdin. Try it yourself!

export BUILDKIT_HOST=docker-container://buildkitd
echo 'fs default() { scratch; mkfile "/output" 0o644 "hello world"; }' | hlb run --target default,download=.

Once the build has finished, you should end up with a file output in your working directory.

$ cat output
hello world

Congratulations! You've now ran your first hlb build and downloaded the output back to your system.

Tip

By default, once the build has finished, nothing is exported anywhere. You'll need to specify where the results go, e.g. to your host as a tarball, or pushed to a Docker registry.

Now that we've verified hlb is functioning, it's time to start the tutorial.