In this example, we will walk through a simple multirepo setup that can be used with Earthly. The entire code of this exercise is available in the

In this example, let's assume that we have a web application where HTML files are held in one repository and JS file are held in another. The complete application is a combination of both.

As such, each repository might have its self-contained Earthfile, specific to the programming language and setup of the repository. A third, main, repository might want to tie everything together in a complete web application.

Here is an example of what the main repository's Earthfile might look like

# Earthfile
FROM node:13.10.1-alpine3.11
WORKDIR /example-multirepo
RUN npm install -g http-server
COPY* ./
COPY* ./
ENTRYPOINT ["http-server", "."]
SAVE IMAGE example-multirepo:latest

Notice how the build is able to reference other repositories and copy artifacts from specific build targets. For example, the line

COPY* ./

references the html target of the repository and copies all its artifacts in the current build environment. Earthly takes care of cloning that repository, executing its build for the html target and extracting the artifacts to be used here.

This command is also cache-aware. If the HEAD of the repository points to a different Git hash, Earthly knows to re-clone and build the repository again, using as much cache as relevant, depending on which files have changed.

You can also specify a specific tag or branch of the remote repository, to help keep builds consistent and avoid surprising changes. For that, you can use the syntax

COPY* ./

where v0.1.1 is a tag or branch specifier.

To review this example with its complete code, check out the examples/multirepo directory on GitHub.