Monthly Downloads: 26
Programming language: Haskell
License: MIT License
Tags: Web    
Latest version: v1.8.0

duplo alternatives and similar packages

Based on the "Web" category.
Alternatively, view duplo alternatives based on common mentions on social networks and blogs.

Do you think we are missing an alternative of duplo or a related project?

Add another 'Web' Package



A opinionated, framework-less build tool for web applications


If "npm" sounds familiar:

  1. Get a Mac
  2. Install Homebrew
  3. Get npm: brew install npm
  4. Get GMP: brew install gmp
  5. Get pre-1.x.x Component: npm install -g [email protected]
  6. Get duplo: npm install -g duplo

If "cabal" sounds more familiar:

$ git clone [email protected]:pixbi/duplo.git
$ cd duplo
$ cabal install


  • duplo help displays all commands.
  • duplo info displays the version for this duplo installation.
  • duplo init <user> <repo> scaffolds a new duplo repo in the current directory.
  • duplo build builds the project. DUPLO_ENV defaults to developoment.
  • duplo dev: starts a webserver, watches for file changes, and builds in development environment.
  • duplo test builds test cases and run it in a browser.
  • duplo production: like duplo dev but builds in production environment
  • duplo patch bumps the patch version.
  • duplo minor bumps the minor version.
  • duplo major bumps the major version.

Guiding Principle

This is a build tool, not an application framework. It simply compiles and builds your codebase for you and does not inject a runtime into or impose a structure on your application.

However, it does have opinions, specifically:

The idea is to manage and deploy your code exclusively with git and have CircleCI deals with deployment for you. However, duplo is a build tool; it doesn't care about the exact structure of your application. This means that all scripts are dumped into one single file, and so are the stylesheets and the markup.

File Structure

app/            --> Application code
app/index.jade  --> Entry point for markups. Only this file is compiled.
                    Use Jade's include system to pull in other markups.
app/index.js    --> Application entry point. Only the top-level
                    application's `index.js` is included and run. Its
                    dependencies' are ignored.
app/assets/     --> Asset files are copied as-is to build's top-level
app/styl/       --> It contains "special" stylesheets that get loaded
                    before any other stylesheets.
app/modules/    --> All other application code not listed above must be
                    placed here. All files at the `app/` level are not
                    included by default.
components/     --> Other repos imported via Component.IO
component.json  --> The Component.IO manifest
dev/            --> Files here are included only when building in
                    development mode.
dev/assets/     --> Copied as-is just like `app/assets/`. Files here would
                    replace those with the same name under `app/assets/`.
dev/modules/    --> Works just like `app/modules/`.
public/         --> Built files when developing. Not committed to source
tests/          --> Test files go here


During development, everything in the dev/assets/ directory is copied over as-is at the end of the build process. This means that files in the directory would replace whatever that has been built (or copied over from app/assets/) at their respective locations.

Anything under dev/modules/ would be treated just like those under app/modules/, that they would be concatenated/compiled into the respective output files (i.e. index.html, index.css, or index.js).

This repo is checked with git-vogue. It's highly recommended that you use it for duplo development as well.


$ duplo test

The test suite contains:

Write a test suite for your duplo project

|-- app/modules/
    |-- a.js
    |-- b.js
    |-- c/
        |- d.js
|-- tests/
    |-- test-a.js
    |-- test-b.js
    |-- c/
        |-- d.js

When testing your codebase, structure your project like the above. Note that the path relative to tests/ should correspond to the path relative to app/modules/.

An example of a test suite:

define('name this to whatever but do not conflict with your module (e.g. `test-a`)',
['moduleA'], function (a) {

  describe('some text', function () {
    it('should ...', function () {
      // now you can use:
      //  expect()....
      //  assert()....


Duplo's test suite uses mocha and chai.js. It also supports another powerful testing tool, SinonJS, so you may fake/mock any functions, ajax requests, and timers yourself.

You may therefore use these functions:

  • mocha: describe, it and etc.
  • chai.js: expect and assert.
  • sinon.js: sinon.spy, sinon.stub, sinon.useFakeTimers and etc.


To make your repo BrowserStack-runnable, modify this template and save it to the root directory of your project (which is added to .gitignore to prevent information leakage into your git history):

  "username": "your-username-here",
  "key": "your-key-here",
  "test_path": "index.html",
  "test_framework": "mocha",
  "browsers": [
      "browser": "chrome",
      "browser_version": "latest",
      "os": "OS X",
      "os_version": "Mountain Lion"


duplo injects the DUPLO_ENV global variable with the value from the environment variable of the same name when building. There is no default value.

Entry Point

Every application has a main entry point. In a duplo application, it is app/index.js. Each repo may contain its own app/index.js but only the repo on which duplo is run does duplo execute app/index.js. Note that app/index.js is excluded when duplo commits via Component.IO so that the consuming application does not see library index files when building the project.

Note that duplo only inspects the top-level define(). If you use require(), your program may not execute as duplo is not aware of anything other than define() declarations. The proper way to declare an entry point in app/index.js is:

[/* ... dependencies ... */],
function (/* ... dependencies ... */) {
  // Code here ...

Application Parameterization

If you need some build-time customization of the app, such as customizing each build with a JSON object of unique IDs and metadata, you can pass any string as the environment variable DUPLO_IN. The string is then turned into a JavaScript string and stored into a global variable.

To avoid special characters, DUPLO_IN must be base-64 encoded.

For example, say you need to pass in a random ID for each build, you would invoke:

// Content decoded as: `{"id":"someId"}`
$ env DUPLO_IN="eyJpZCI6InNvbWVJZCJ9" duplo

Then in app/index.js:

var someId = DUPLO_IN.id;

Note that all newline characters are removed before the string is wrapped into a JavaScript string.

JavaScript Concatenation Order

JavaScript files are not concatenated in any particular order. You must wrap code inside an AMD module and declaring its dependencies. For code that needs to be executed at initialization, utilize the environment's initialization event such as document.addEventListener("DOMContentLoaded") to bootstrap the rest of the script.

CSS/Stylus Concatenation Order

Unlike script files, where you place your CSS files within app/ is significant. Stylus files will be concatenated in this order:

app/styl/variables.styl --> An optional variable file that gets injected
                            into every Stylus file
app/styl/keyframes.styl --> Keyframes
app/styl/fonts.styl     --> Font declarations
app/styl/reset.styl     --> Resetting existing CSS in the target
app/styl/main.styl      --> Application CSS that goes before any module
app/**/*.styl           --> All other CSS files

There is no particular concatenation order between different dependencies.

HTML/Jade Concatenation Order

Jade files are concatenated in no particular order as the Jade include system is used for explicit ordering.

Automatic rewriting for Jade

duplo does not and cannot peek into Jade's include system. However, it does automatically expand paths in include statements to make the inclusion process easier. Take this example:

include index.jade
include menu/index.jade
include pixbi-helper/index.jade

In the absence of a Component repo string (i.e. <user>-<repo>), the path is assumed to be pointing to a file under the modules directory in the current repo. With a component repo string, it is assumed to also be pointing to a file under the modules directory, but in the corresponding component's repo.

The above is effectively rewritten into these paths, relative to the top-level repo's directory.

include app/modules/index.jade
include app/modules/menu/index.jade
include components/pixbi-helper/app/modules/index.jade

A note on the modules directory

By now, it should be obvious that there are really two "modes" for any duplo repo: an application mode and a library mode. In application mode, duplo acts as the top-level program, including other duplo repos via Component as libraries. In this scenario, app/index.js and app/index.jade are included into the build. Contrast this to the library mode, where only those in the "second" level (e.g. modules/, assets/, styl/) are included into the build.

Component Versions

Each component's version is recorded in the DUPLO_VERSIONS global variable, in the form similar to:

  "pixbi-main": "4.1.9",
  "pixbi-launcher": "0.1.4"

Dependency Selection

Some cases require the repo to be polymorphic in the sense that we could generate different forms of the same codebase. For example, you may need to build the repo in an embeddable form which would exclude certain dependencies that are required in its standalone form.

In this case you would include a modes attribute in the component.json manifest file. The attribute would contain an embeddable and a standalone attributes, each of which would then contain an array of dependencies as specified in the dependencies attribute to include.

Running duplo with the environment variable DUPLO_MODE set to embeddable would build with the dependencies specified under embeddable while setting MODE to standalone would do the same with those specified under the standalone attribute. Otherwise duplo would just build with all dependencies.

Note that dependency selection applies at the dependency level but not at the file level within the components.

Also note that duplo caches between builds. When you switch dependency selection, remember to duplo clean your repo first.

Putting it all together, an example of a component.json:

  "dependencies": {
    "pixbi/sdk": "1.1.1",
    "pixbi/embeddable": "2.2.2",
    "pixbi/standalone": "3.3.3"
  "modes": {
    "embeddable": [
    "standalone": [

Duplo Log

Note that tasks are run in parallel so the display log may look scrambled from line to line. This is normal.

Developing duplo

Right now duplo is published to both Hackage and NPM. Because of various compatibility issues, each time duplo is published all of the following must be done:

  1. Manually bump version in both duplo.cabal and package.json
  2. $ git tag <version>
  3. $ util/publish.sh
  4. $ npm publish
  5. Publish to Hackage

Note: ALWAYS always use Stackage while do not specify versions in the Cabal file. Just specify the dependencies by name and let Stackage manage the versions.

Copyright and License

Code and documentation copyright 2014 Pixbi. Code released under the MIT license. Docs released under Creative Commons.

*Note that all licence references and agreements mentioned in the duplo README section above are relevant to that project's source code only.