dagger/docs/faq/1232-chain-actions.md

slug	displayed_sidebar
/1232/chain-actions	europa

Chaining actions

Dependencies are materialized at runtime, when your Cue files are parsed and the corresponding DAG gets generated:

// Prerequisite action that runs when `test` is being called
_dockerCLI: alpine.#Build & {
    packages: {
        bash: {}
    }
}

// Main action
foo: bash.#Run & {
    input: _dockerCLI.output // <== CHAINING of action happening here
    always: true
    script: contents: #"""
        echo "bonjour"
        """#
}

On above example, _dockerCLI gets executed first, as the corresponding DAG shows that it is required for the action foo to be processed.

This is the input-output model: foo's input is _dockerCLI's output.

We currently don't support explicit dependencies at the moment (one of our top priority). But, if you need one, here is how your can hack your way around it:

foo: bash.#Run & {
    input: _dockerCLI.output
    always: true
    script: contents: #"""
        echo "bonjour"
        """#
}

// Main action
bar: bash.#Run & {
    input: _dockerCLI.output
    always: true
    script: contents: #"""
        echo "bonjour"
        """#
    env: HACK: "\(test.success)" // <== HACK: CHAINING of action happening here
}

foo and bar are similar actions. I don't want to rely on the input-output model but still want to force a dependency between them. The easiest way is to create an environment variable that relies on the other action's success