Archived

This repository has been archived on 2024-04-08. You can view files and clone it, but cannot push or open issues or pull requests.

Sam Alba 29ace62b72 docs: 102: use source link instead

Signed-off-by: Sam Alba <sam.alba@gmail.com>

2021-06-23 10:53:37 +02:00

9.6 KiB

Raw Blame History

slug
/learn/102-dev

Dagger 102: create your first environment

Overview

In this guide you will create your first Dagger environment from scratch, and use it to deploy a React application to 2 locations in parallel: a dedicated Amazon S3 bucket, and a Netlify site.

Anatomy of a Dagger environment

A Dagger environment contains all the code and data necessary to deliver a particular application in a particular way. For example the same application might be delivered to a production and staging environment, each with their own configuration.

An environment is made of 3 parts:

A plan, authored by the environment's developer, using the Cue language.
Inputs, supplied by the environment's user via the dagger input command, and written to a special file. Inputs may be configuration values, artifacts, or encrypted secrets.
Outputs, computed by the Dagger engine via the dagger up command, and recorded to a special directory.

We will first develop our environment's plan, then configure its initial inputs, then finally run it to verify that it works.

Anatomy of a plan

A plan specifies, in code, how to deliver a particular application in a particular way. It is your environment's source code.

Unlike regular imperative programs which specify a sequence of instructions to execute, a Dagger plan is declarative: it lays out your application's supply chain as a graph of interconnected nodes.

Each node in the graph represents a component of the supply chain, for example:

Development tools: source control, CI, build systems, testing systems
Hosting infrastructure: compute, storage, networking, databases, CDNs
Software dependencies: operating systems, languages, libraries, frameworks, etc.

Each link in the graph represents a flow of data between nodes. For example:

source code flows from a git repository to a build system
system dependencies are combined in a docker image, then uploaded to a registry
configuration files are generated then sent to a compute cluster or load balancer

Introduction to Cue development

Dagger delivery plans are developed in Cue. Cue is a powerful declarative language by Marcel van Lohuizen. Marcel co-created the Borg Configuration Language (BCL), the language used to deploy all applications at Google. It is a superset of JSON, with additional features to make declarative, data-driven programming as pleasant and productive as regular imperative programming.

If you are new to Cue development, don't worry: this tutorial will walk you through the basic steps to get started, and give you resources to learn more.

In technical terms, our plan is a Cue Package. In this tutorial we will develop a new Cue package from scratch for our plan; but you can use any Cue package as a plan.

Initial setup

Install Cue

Although not strictly necessary, for an optimal development experience we recommend installing a recent version of Cue.

Prepare Cue learning resources

If you are new to Cue, we recommend keeping the following resources in browser tabs:

The unofficial but excellent Cuetorials in a browser tab, to look up Cue concepts as they appear.
The official Cue interactive sandbox for easy experimentation.

Setup example app

You will need a local copy of the Dagger examples repository. NOTE: you may use the same local copy across all tutorials.

git clone https://github.com/dagger/examples

Make sure that all commands are run from the todoapp directory:

cd examples/todoapp

Develop the plan

Initialize a Cue module

Developing for Dagger takes place in a Cue module. If you are familiar with Go, Cue modules are directly inspired by Go modules. Otherwise, don't worry: a Cue module is simply a directory with one or more Cue packages in it. A Cue module has a cue.mod directory at its root.

In this guide we will use the same directory as the root of the Dagger workspace and the root of the Cue module; but you can create your Cue module anywhere inside the Dagger workspace.

cue mod init

Create a Cue package

Now we start developing our Cue package at the root of our Cue module.

In this guide we will split our package in multiple files, one per component. But you can organize your package any way you want: the Cue evaluator simply merges together all files from the same package, as long as they are in the same directory and start with the same package clause. It is common for a Cue package to have only one file. See the Cue documentation for more details.

We will call our package multibucket because it sounds badass and vaguely explains what it does. But you can call your packages anything you want.

Let's create a new directory for our Cue package:

mkdir multibucket

Component 1: app source code

The first component of our plan is the source code of our React application.

In Dagger terms, this component has 2 important properties:

It is an artifact: something that can be represented as a directory.
It is an input: something that is provided by the end user.

Let's write the corresponding Cue code to a new file in our package:

package multibucket

import (
    "dagger.io/dagger"
)

// Source code of the sample application
src: dagger.#Artifact & dagger.#Input

This defines a component at the key src, and specifies that it is both an artifact and an input.

Component 2: yarn package

The second component of our plan is the Yarn package built from the app source code:

package multibucket

import (
    "dagger.io/js/yarn"
)

// Build the source code using Yarn
app: yarn.#Package & {
    source: src
}

Let's break it down:

package multibucket: this file is part of the multibucket package
import ( "dagger.io/js/yarn" ): import a package from the Dagger Universe.
app: yarn.#Package: apply the #Package definition at the key app
&: also merge the following values at the same key...
{ source: src }: set the key app.source to the value of src. This connects our 2 components, forming the first link in our DAG.

Component 3: dedicated S3 bucket

FIXME: this section is not yet available, because the Amazon S3 package does not yet support bucket creation. We welcome external contributions :)

Component 4: deploy to Netlify

The third component of our plan is the Netlify site to which the app will be deployed:

package multibucket

import (
    "dagger.io/netlify"
)

// Netlify site
site: "netlify": netlify.#Site & {
    contents: app.build
}

This is very similar to the previous component:

We use the same package name as the other files
We import another package from the Dagger Universe.
site: "netlify": site.#Netlify: apply the #Site definition at the key site.netlify. Note the use of quotes to protect the key from name conflict.
&: also merge the following values at the same key...
{ contents: app.build }: set the key site.netlify.contents to the value of app.build. This connects our components 2 and 3, forming the second link in our DAG.

Exploring a package documentation

But wait: how did we know what fields were available in yarn.#Package and netlify.#Site? Answer: thanks to the dagger doc command, which prints the documentation of any package from Dagger Universe.

dagger doc dagger.io/netlify
dagger doc dagger.io/js/yarn

You can also browse the Dagger Universe reference in the documentation.

Setup the environment

Create a new environment

Now that your Cue package is ready, let's create an environment to run it:

dagger new 'multibucket'

Load the plan into the environment

Now let's configure the new environment to use our package as its plan:

cp multibucket/*.cue .dagger/env/multibucket/plan/

Note: you need to copy the files from your package into the environment, as shown above. If you make more changes to your package, you will need to copy the new version, or it will not be used. In the future, we will add the ability to reference your Cue package directory, making this manual copy unnecessary.

9.6 KiB Raw Blame History