Skip to main content
Version: Next

Build System

The Backstage build system is a collection of build and development tools that help you lint, test, develop and finally release your Backstage projects. The purpose of the build system is to provide an out-of-the-box solution that works well with Backstage and lets you focus on building an app rather than having to spend time setting up your own tooling.

The build system setup is part of the @backstage/cli, and already included in any project that you create using @backstage/create-app. It is similar to for example react-scripts, which is the tooling you get with create-react-app. The Backstage build system relies heavily on existing open source tools from the JavaScript and TypeScript ecosystem, such as Webpack, Rollup, Jest, and ESLint.

Design Considerations

There are a couple of core beliefs and constraints that guided the design of the Backstage build system. The first and most important is that we put the development experience first. If we need to cut corners or add complexity we do so in other areas, but the experience of firing up an editor and iterating on some code should always be as smooth as possible.

In addition, there are a number of hard and soft requirements:

  • Monorepos - The build system should support multi-package workspaces
  • Publishing - It should be possible to build and publish individual packages
  • Scale - It should scale to hundreds of large packages without excessive wait times
  • Reloads - The development flow should support quick on-save hot reloads
  • Simple - Usage should be simple and configuration should be kept minimal
  • Universal - Development towards both web applications, isomorphic packages, and Node.js
  • Modern - The build system targets modern environments
  • Editors - Type checking and linting should be available within most editors

During the design of the build system this collection of requirements was not something that was supported by existing tools like for example react-scripts. The requirements of scaling in combination of monorepo, publishing, and editor support led us to adopting our own specialized setup.

Structure

We can divide the development flow within Backstage into a couple of different steps:

  • Formatting - Applies a consistent formatting to your source code
  • Linting - Analyzes your source code for potential problems
  • Type Checking - Verifies that TypeScript types are valid
  • Testing - Runs different levels of test suites towards your project
  • Building - Compiles the source code in an individual package
  • Bundling - Combines a package and all of its dependencies into a production-ready bundle

These steps are generally kept isolated from each other, with each step focusing on its specific task. For example, we do not do linting or type checking together with the building or bundling. This is so that we can provide more flexibility and avoid duplicate work, improving performance. It is strongly recommended that as a part of developing within Backstage you use a code editor or IDE that has support for formatting, linting, and type checking.

Let's dive into a detailed look at each of these steps and how they are implemented in a typical Backstage app.

Package Roles

Package roles were introduced in March 2022. To migrate existing projects, see the migration guide.

The Backstage build system uses the concept of package roles in order to help keep configuration lean, provide utility and tooling, and enable optimizations. A package role is a single string that identifies what the purpose of a package is, and it's defined in the package.json of each package like this:

{
"name": "my-package",
"backstage": {
"role": "<role>"
},
...
}

These are the available roles that are currently supported by the Backstage build system:

RoleDescriptionExample
frontendBundled frontend applicationpackages/app
backendBundled backend applicationpackages/backend
cliPackage used as a command-line interface@backstage/cli, @backstage/codemods
web-libraryWeb library for use by other packages@backstage/plugin-catalog-react
node-libraryNode.js library for use by other packages@backstage/plugin-techdocs-node
common-libraryIsomorphic library for use by other packages@backstage/plugin-permission-common
frontend-pluginBackstage frontend plugin@backstage/plugin-scaffolder
frontend-plugin-moduleBackstage frontend plugin module@backstage/plugin-analytics-module-ga
backend-pluginBackstage backend plugin@backstage/plugin-auth-backend
backend-plugin-moduleBackstage backend plugin module@backstage/plugin-search-backend-module-pg

Most of the steps that we cover below have an accompanying command that is intended to be used as a package script. The commands are all available under the backstage-cli package category, and many of the commands will behave differently depending on the role of the package. The commands are intended to be used like this:

{
"scripts": {
"start": "backstage-cli package start",
"build": "backstage-cli package build",
"lint": "backstage-cli package lint",
...
}
}

Formatting

The formatting setup lives completely within each Backstage application and is not part of the CLI. In an app created with @backstage/create-app the formatting is handled by prettier, but each application can choose their own formatting rules and switch to a different formatter if desired.

Linting

The Backstage CLI includes a lint command, which is a thin wrapper around eslint. It adds a few options that can't be set through configuration, such as including the .ts and .tsx extensions in the set of linted files. The lint command simply provides a sane default and is not intended to be customizable. If you want to supply more advanced options you can invoke eslint directly instead.

In addition to the lint command, the Backstage CLI also includes a set of base ESLint configurations, one for frontend and one for backend packages. These lint configurations in turn build on top of the lint rules from @spotify/web-scripts.

In a standard Backstage setup, each individual package has its own lint configuration, along with a root configuration that applies to the entire project. The configuration in each package starts out as a standard configuration that is determined based on the package role, but it can be customized to fit the needs of each package.

A minimal .eslintrc.js configuration now looks like this:

module.exports = require('@backstage/cli/config/eslint-factory')(__dirname);

But you can provide custom overrides for each package using the optional second argument:

module.exports = require('@backstage/cli/config/eslint-factory')(__dirname, {
ignorePatterns: ['templates/'],
rules: {
'jest/expect-expect': 'off',
},
});

The configuration factory also provides utilities for extending the configuration in ways that are otherwise very cumbersome to do with plain ESLint, particularly for rules like no-restricted-syntax. These are the extra keys that are available:

KeyDescription
tsRulesAdditional rules to apply to TypeScript files
testRulesAdditional rules to apply to tests files
restrictedImportsAdditional paths to add to no-restricted-imports
restrictedImportPatternsAdditional patterns to add to no-restricted-imports
restrictedSrcImportsAdditional paths to add to no-restricted-imports in src files
restrictedTestImportsAdditional paths to add to no-restricted-imports in test files
restrictedSyntaxAdditional patterns to add to no-restricted-syntax
restrictedSrcSyntaxAdditional patterns to add to no-restricted-syntax in src files
restrictedTestSyntaxAdditional patterns to add to no-restricted-syntax in test files

Type Checking

Just like formatting, the Backstage CLI does not have its own command for type checking. It does however have a base configuration with both recommended defaults as well as some required settings for the build system to work.

Perhaps the most notable part about the TypeScript setup in Backstage projects is that the entire project is one big compilation unit. This is due to performance optimization as well as ease of use, since breaking projects down into smaller pieces has proven to both lead to a more complicated setup, as well as type checking of the entire project being an order of magnitude slower. In order to make this setup work, the entrypoint of each package needs to point to the TypeScript source files, which in turn causes some complications during publishing that we'll talk about in that section.

The type checking is generally configured to be incremental for local development, with the output stored in the dist-types folder in the repo root. This provides a significant speedup when running tsc multiple times locally, but it does make the initial run a little bit slower. Because of the slower initial run we disable incremental type checking in the tcs:full Yarn script that is included by default in any Backstage app and is intended for use in CI.

Another optimization that is used by default is to skip the checking of library types, this means that TypeScript will not verify that types within node_modules are sound. Disabling this check significantly speeds up type checking, but in the end it is still an important check that should not be completely omitted, it's simply unlikely to catch issues that are introduced during local development. What we opt for instead is to include the check in CI through the tsc:full script, which will run a full type check, including node_modules.

For the two reasons mentioned above, it is highly recommended to use the tsc:full script to run type checking in CI.

Testing

As mentioned above, the Backstage CLI uses Jest, which is a JavaScript test framework that covers both test execution and assertions. Jest executes all tests in Node.js, including frontend browser code. The trick it uses is to execute the tests in a Node.js VM using various predefined environments, such as one based on jsdom that helps mimic browser APIs and behavior.

The Backstage CLI has its own command that helps execute tests, backstage-cli test, as well as its own configuration at @backstage/cli/config/jest.js. The command is a relatively thin wrapper around running jest directly. Its main responsibility is to make sure the included configuration is used, as well setting the NODE_ENV and TZ environment variables, and provided some sane default flags like --watch if executed within a Git repository.

The by far biggest amount of work is done by the Jest configuration included with the Backstage CLI. It both takes care of providing a default Jest configuration, as well as allowing for configuration overrides to be defined in each package.json. How this can be done in practice is discussed in the Jest configuration section.

Building

The primary purpose of the build process is to prepare packages for publishing, but it's also used as part of the backend bundling process. Since it's only used in these two cases, any Backstage app that does not use the Backend parts of the project may not need to interact with the build process at all. It can nevertheless be useful to know how it works, since all of the published Backstage packages are built using this process.

The build is currently using Rollup and executes in isolation for each individual package. The build is invoked using the package build command, and applies to all packages roles except the bundled ones, frontend and backend.

There are three different possible outputs of the build process: JavaScript in CommonJS module format, JavaScript in ECMAScript module format, and type declarations. Each invocation of a build command will write one or more of these outputs to the dist folder in the package, and in addition copy any asset files like stylesheets or images. For more details on what syntax and file formats are supported by the build process, see the loaders section.

When building CommonJS or ESM output, the build commands will always use src/index.ts as the entrypoint. All non-relative modules imports are considered external, meaning the Rollup build will only compile the source code of the package itself. All import statements of external dependencies, even within the same monorepo, will stay intact.

The build of the type definitions works quite differently. The entrypoint of the type definition build is the relative location of the package within the dist-types folder in the project root. This means that it is important to run type checking before building any packages with type definitions, and that emitting type declarations must be enabled in the TypeScript configuration. The reason for the type definition build step is to strip out all types but the ones that are exported from the package, leaving a much cleaner type definition file and making sure that the type definitions are in sync with the generated JavaScript.

Bundling

The goal of the bundling process is to combine multiple packages together into a single runtime unit. The way this is done varies between frontend and backend, as well as local development versus production deployment. Because of that we cover each combination of these cases separately.

Frontend Development

The frontend development setup is used for all packages with a frontend role, and is invoked using the package start command. The only difference between the different roles is that packages with the 'frontend' role use src/index as the entrypoint, while other roles instead use dev/index. When running the start command, a development server will be set up that listens to the protocol, host and port set by app.baseUrl in the configuration. If needed it is also possible to override the listening options through the app.listen configuration.

The frontend development bundling is currently based on Webpack and Webpack Dev Server. The Webpack configuration itself varies very little between the frontend development and production bundling, so we'll dive more into the configuration in the production section below. The main differences are that process.env.NODE_ENV is set to 'development', minification is disabled, cheap source maps are used, and React Refresh is enabled.

If you prefer to run type checking and linting as part of the Webpack process, you can enable usage of the ForkTsCheckerWebpackPlugin by passing the --check flag. Although as mentioned above, the recommended way to handle these checks during development is to use an editor that has built-in support for them instead.

Frontend Production

The frontend production bundling creates your typical web content bundle, all contained within a single folder, ready for static serving. It is used when building packages with the 'frontend' role, and unlike the development bundling there is no way to build a production bundle of an individual plugin. The output of the bundling process is written to the dist folder in the package.

Just like the development bundling, the production bundling is based on Webpack. It uses the HtmlWebpackPlugin to generate the index.html entry point, and includes a default template that's included with the CLI. You can replace the bundled template by adding public/index.html to your app package. The template has access to two global constants, publicPath which is the public base path that the bundle is intended to be served at, as well as config which is your regular frontend scoped configuration from @backstage/config.

The Webpack configuration also includes a custom plugin for resolving packages correctly from linked in packages, the ModuleScopePlugin from react-dev-utils which makes sure that imports don't reach outside the package, a few fallbacks for some Node.js modules like 'buffer' and 'events', a plugin that writes the frontend configuration to the bundle as process.env.APP_CONFIG, and lastly minification handled by esbuild using the esbuild-loader. There are of course also a set of loaders configured, which you can read more about in the loaders and transpilation sections.

During the build, the following constants are also set:

process.env.NODE_ENV = 'production';
process.env.BUILD_INFO = {
cliVersion: '0.4.0', // The version of the CLI package
gitVersion: 'v0.4.0-86-ge54815618', // output of `git describe --always`
packageVersion: '1.0.5', // The version of the app package itself
timestamp: 1678900000000, // Date.now() when the build started
commit: 'e548156182a973ed4b459e18533afc22c85ffff8', // output of `git rev-parse HEAD`
};

The output of the bundling process is split into two categories of files with separate caching strategies. The first is a set of generic assets with plain names in the root of the dist/ folder. You will want to serve these with short-lived caching or no caching at all. The second is a set of hashed static assets in the dist/static/ folder, which you can configure to be cached for a much longer time.

The configuration of static assets is optimized for frequent changes and serving over HTTP 2.0. The assets are aggressively split into small chunks, which means the browser has to make a lot of small requests to load them. The upside is that changes to individual plugins and packages will invalidate a smaller number of files, thereby allowing for rapid development without much impact on the page load performance.

Backend Development

The backend development setup does not use any bundling process. It runs a Node.js process directly, with on-the-fly transpilation from TypeScript to JavaScript. The transpilation is done with a custom transform based on SWC.

During development the backend Node.js process will restart whenever there is a change to the source code. This means that any in-memory data will be lost. In order to store data between restarts, the backend process has an IPC channel available to store and restore data from the parent CLI process. The primary purpose of this, which is already built-in, is to restore the contents of databases when using SQLite for development. You can also use it for your own purposes, with the DevDataStore utility exported from the @backstage/backend-dev-utils package.

If you want to inspect the running Node.js process, the --inspect and --inspect-brk flags can be used, as they will be passed through as options to node execution.

Backend Production

The backend production bundling uses a completely different setup than the other bundling options. Rather than using Webpack, the backend production bundling instead collects the backend packages and all of their local dependencies into a deployment archive. The archive is written to dist/bundle.tar.gz, and contains the packaged version of each of these packages. The layout of the packages in the archive is the same as the directory layout in the monorepo, and the bundle also contains the root package.json and yarn.lock files.

Note that before creating a production bundle you must first build all of the backend packages. This can be done automatically when executing the backend:bundle command by passing the --build-dependencies flag. It is an optional flag since it is quite common that the packages are already built earlier on in your build process, and building them again would result in duplicate work.

In order to use the bundle, you extract it into a directory, run yarn install --production, and then start the backend using your backend package as the Node.js entry point, for example node packages/backend.

The dist/bundle.tar.gz is accompanied by a dist/skeleton.tar.gz, which has the same layout, but only contains package.json files. This skeleton archive can be used to run a yarn install in environments that will benefit from the caching that this enables, such as Docker image builds. To use the skeleton archive you copy it over to the target directory along with the root package.json and yarn.lock, extract the archive, and then run yarn install --production. Your target directory will then have all dependencies installed, and as soon as you copy over and extract the contents of the bundle.tar.gz archive on top of it, the backend will be ready to run.

The following is an example of a Dockerfile that can be used to package the output of building a package with role 'backend' into an image:

FROM node:20-bookworm-slim
WORKDIR /app

COPY yarn.lock package.json packages/backend/dist/skeleton.tar.gz ./
RUN tar xzf skeleton.tar.gz && rm skeleton.tar.gz

# install sqlite3 dependencies
RUN apt-get update && \
apt-get install -y libsqlite3-dev python3 cmake g++ && \
rm -rf /var/lib/apt/lists/* && \
yarn config set python /usr/bin/python3

RUN yarn install --frozen-lockfile --production --network-timeout 300000 && rm -rf "$(yarn cache dir)"

COPY packages/backend/dist/bundle.tar.gz app-config.yaml ./
RUN tar xzf bundle.tar.gz && rm bundle.tar.gz

CMD ["node", "packages/backend"]

Transpilation

The transpilers used by the Backstage CLI have been chosen according to the same design considerations that were mentioned above. A few specific requirements are of course support for TypeScript and JSX, but also React hot reloads or refresh, and hoisting of Jest mocks. The Backstage CLI also only targets up to date and modern browsers, so we actually want to keep the transpilation process as lightweight as possible, and leave most syntax intact.

Apart from these requirements, the deciding factor for which transpiler to use is their speed. The build process keeps the integration with the transpilers lightweight, without additional plugins or such. This enables us to switch out transpilers as new options and optimizations become available, and keep on using the best options that are available.

Our current selection of transpilers are esbuild and Sucrase. The reason we choose to use two transpilers is that esbuild is faster than Sucrase and produces slightly nicer output, but it does not have the same set of features, for example it does not support React hot reloading.

The benchmarking of the various options was done in ts-build-bench. This benchmarking project allows for setups of different shapes and sizes of monorepos, but the setup we consider the most important in our case is a large monorepo with lots of medium to large packages that are being bundled with Webpack. Some rough findings have been that esbuild is the fastest option right now, with Sucrase following closely after and then SWC closely after that. After those there's a pretty big gap up to the TypeScript compiler run in transpilation only mode, and lastly another jump up to Babel, being by far the slowest out of the transpilers we tested.

Something to note about these benchmarks is that they take the full Webpack bundling time into account. This means that even though some transpilation options may be orders of magnitude faster than others, the total time is not impacted in the same way as there are lots of other things that go into the bundling process. Still, switching from for example Babel to Sucrase is able to make the bundling anywhere from two to five times faster.

Loaders

The Backstage CLI is configured to support a set of loaders throughout all parts of the build system, including the bundling, tests, builds, and type checking. Loaders are always selected based on the file extension. The following is a list of all supported file extensions:

ExtensionExportsPurpose
.tsScript ModuleTypeScript
.tsxScript ModuleTypeScript and XML
.jsScript ModuleJavaScript
.jsxScript ModuleJavaScript and XML
.mjsScript ModuleECMAScript Module
.cjsScript ModuleCommonJS Module
.jsonJSON DataJSON Data
.ymlJSON DataYAML Data
.yamlJSON DataYAML Data
.cssclassesStyle sheet
.eotURL PathFont
.ttfURL PathFont
.woff2URL PathFont
.woffURL PathFont
.bmpURL PathImage
.gifURL PathImage
.jpegURL PathImage
.jpgURL PathImage
.pngURL PathImage
.svgURL PathImage
.mdURL PathMarkdown File

Jest Configuration

The Backstage CLI bundles its own Jest configuration file, which is used automatically when running backstage-cli test. It's available at @backstage/cli/config/jest.js and can be inspected here. Usage of this configuration can be overridden either by passing a --config <path> flag to backstage-cli test, or placing a jest.config.js or jest.config.ts file in your package.

The built-in configuration brings a couple of benefits and features. The most important one being a baseline transformer and module configuration that enables support for the listed loaders within tests. It will also automatically detect and use src/setupTests.ts if it exists, and provides a coverage configuration that works well with our selected transpilers. The configuration will also detect the appropriate Jest environment for each package role, running web-libraries with the "jsdom" environment, node-libraries with "node", and so on.

The configuration also takes a project-wide approach, with the expectation most if not all packages within a monorepo will use the same base configuration. This allows for optimizations such as sharing the Jest transform cache across all packages in a monorepo, avoiding unnecessary transpilation. It also makes it possible to load in all Jest configurations at once, and with that run yarn test <pattern> from the root of a monorepo without having to set the working directory to the package that the test is in.

Where small customizations are needed, such as setting coverage thresholds or support for specific transforms, it is possible to override the Jest configuration through the "jest" field in package.json. For a full list of options, see the Jest documentation. These overrides will be loaded in from all package.json files in the directory ancestry, meaning that you can place common configuration in the package.json at the root of a monorepo. If multiple overrides are found, they will be merged together with configuration further down in the directory tree taking precedence.

The overrides in a single package.json may for example look like this:

  "jest": {
"coverageThreshold": {
"global": {
"functions": 100,
"lines": 100,
"statements": 100
}
}
},

Caching

Caching is used sparingly throughout the Backstage build system. It is always used as a way to squeeze out a little bit of extra performance, rather than requirement to keep things fast. The following is a list of places where optional caching is available:

  • TypeScript - The default tsconfig.json used by Backstage projects has incremental set to true, which enables local caching of type checking results. It is however generally not recommended in CI, where yarn tsc:full is preferred, which sets --incremental false.
  • Testing - The backstage-cli repo test command has a --successCache flag that enables caching of successful test results. This is done at the package level, meaning that if a package has not been changed since the last test run and it was successful, the testing will be skipped. This is recommended to be used in CI, but not during local development.
  • Linting - The backstage-cli repo lint command has a --successCache flag that enables caching of successful linting results. This is done at the package level, meaning that if a package has not been changed since the last lint run and it was successful, the linting will be skipped. This is recommended to be used in CI, but not during local development.
  • Webpack - It is possible to enable experimental caching of frontend package builds using the BACKSTAGE_CLI_EXPERIMENTAL_BUILD_CACHE environment variable. This will enable the Webpack filesystem cache.

Debugging Jest Tests

For your productivity working with unit tests it's quite essential to have your debugging configured in IDE. It will help you to identify the root cause of the issue faster.

IntelliJ IDEA

  1. Update Jest configuration template by:
  • Click on "Edit Configurations" on top panel
  • In the modal dialog click on link "Edit configuration templates..." located in the bottom left corner.
  • In "Jest package" you have to point to relative path of jest module (it will be suggested by IntelliJ), i.e. ~/proj/backstage/node_modules/jest
  • In "Jest config" point to your jest configuration file, use absolute path for that, i.e. --config /Users/user/proj/backstage/packages/cli/config/jest.js --runInBand
  1. Now you can run any tests by clicking on green arrow located on describe or it.

VS Code

{
"jest.jestCommandLine": "node_modules/.bin/jest --config node_modules/@backstage/cli/config/jest.js",
// In a large repo like the Backstage main repo you likely want to disable
// watch mode and the initial test run too, leaving just manual and perhaps
// on-save test runs in place.
"jest.autoRun": {
"watch": false,
"onSave": "test-src-file"
}
}

A complete launch configuration for VS Code debugging may look like this:

{
"type": "node",
"name": "vscode-jest-tests.v2",
"request": "launch",
"args": [
"repo",
"test",
"--runInBand",
"--watchAll=false",
"--testNamePattern",
"${jest.testNamePattern}",
"--runTestsByPath",
"${jest.testFile}"
],
"console": "integratedTerminal",
"internalConsoleOptions": "neverOpen",
"disableOptimisticBPs": true,
"program": "${workspaceFolder}/node_modules/.bin/backstage-cli"
}

Publishing

Package publishing is an optional part of the Backstage build system and not something you will need to worry about unless you are publishing packages to a registry. In addition to the documentation in the section, be sure to also read the section on metadata for published packages.

In order to publish a package, you first need to build it, which will populate the dist folder. Because the Backstage build system is optimized for local development along with our particular TypeScript and bundling setup, it is not possible to publish the package immediately at this point. This is because the entry points of the package will still be pointing to src/index.ts, but we want them to point to dist/ in the published package.

In order to work around this, the Backstage CLI provides prepack and postpack commands that help prepare the package for publishing. These scripts are automatically run by Yarn before publishing a package.

The prepack command will take entry point fields in "publishConfig", such as "main" and "module", and move them to the top level of the package.json. This lets you point at the desired files in the dist folder during publishing. The postpack command will simply revert this change in order to leave your project clean.

The following is an excerpt of a typical setup of an isomorphic library package:

  "main": "src/index.ts",
"types": "src/index.ts",
"publishConfig": {
"access": "public",
"main": "dist/index.cjs.js",
"module": "dist/index.esm.js",
"types": "dist/index.d.ts"
},
"scripts": {
"build": "backstage-cli package build",
"lint": "backstage-cli package lint",
"test": "backstage-cli package test",
"clean": "backstage-cli package clean",
"prepack": "backstage-cli package prepack",
"postpack": "backstage-cli package postpack"
},
"files": ["dist"],

Subpath Exports

The Backstage CLI supports implementation of subpath exports through the "exports" field in package.json. It might for example look like this:

  "name": "@backstage/plugin-foo",
"exports": {
".": "./src/index.ts",
"./components": "./src/components.ts",
},

This in turn would allow you to import anything exported in src/index.ts via @backstage/plugins-foo, and src/components.ts via @backstage/plugins-foo/components. Note that patterns are not supported, meaning the exports may not contain * wildcards.

As with the rest of the Backstage CLI build system, the setup is optimized for local development, which is why the "exports" targets point directly to source files. The package build command will detect the "exports" field and automatically generate the corresponding dist files, and the prepublish command will rewrite the "exports" field to point to the dist files, as well as generating folder-based entry points for backwards compatibility.

TypeScript support is currently handled though the typesVersions field, as there is not yet a module resolution mode that works well with "exports". You can craft the typesVersions yourself, but it will also be automatically generated by the migrate package-exports command.

To add subpath exports to an existing package, simply add the desired "exports" fields and then run the following command:

yarn backstage-cli migrate package-exports