This document describes how to set up your development environment to build and test AngularJS, and
explains the basic mechanics of using git
, node
, yarn
and grunt
.
Before you can build AngularJS, you must install and configure the following dependencies on your machine:
-
Git: The Github Guide to Installing Git is a good source of information.
-
Node.js v8.x (LTS): We use Node to generate the documentation, run a development web server, run tests, and generate distributable files. Depending on your system, you can install Node either from source or as a pre-packaged bundle.
We recommend using nvm (or nvm-windows) to manage and install Node.js, which makes it easy to change the version of Node.js per project.
-
Yarn: We use Yarn to install our Node.js module dependencies (rather than using npm). See the detailed installation instructions.
-
Java: We minify JavaScript using Closure Tools, which require Java (version 7 or higher) to be installed and included in your PATH variable.
-
Grunt: We use Grunt as our build system. We're using it as a local dependency, but you can also add the grunt command-line tool globally (with
yarn global add grunt-cli
), which allows you to leave out theyarn
prefix for all our grunt commands.
To contribute code to AngularJS, you must have a GitHub account so you can push code to your own fork of AngularJS and open Pull Requests in the GitHub Repository.
To create a Github account, follow the instructions here. Afterwards, go ahead and fork the main AngularJS repository.
To build AngularJS, you clone the source code repository and use Grunt to generate the non-minified and minified AngularJS files:
# Clone your Github repository:
git clone https://github.com/<github username>/angular.js.git
# Go to the AngularJS directory:
cd angular.js
# Add the main AngularJS repository as an upstream remote to your repository:
git remote add upstream "https://github.com/angular/angular.js.git"
# Install JavaScript dependencies:
yarn install
# Build AngularJS:
yarn grunt package
Note: If you're using Windows, you must use an elevated command prompt (right click, run as
Administrator). This is because yarn grunt package
creates some symbolic links.
The build output is in the build
directory. It consists of the following files and
directories:
-
angular-<version>.zip
— The complete zip file, containing all of the release build artifacts. -
angular.js
/angular.min.js
— The regular and minified core AngularJS script file. -
angular-*.js
/angular-*.min.js
— All other AngularJS module script files. -
docs/
— A directory that contains a standalone version of the docs (same as served indocs.angularjs.org
).
To debug code, run end-to-end tests, and serve the docs, it is often useful to have a local HTTP server. For this purpose, we have made available a local web server based on Node.js.
-
To start the web server, run:
yarn grunt webserver
-
To access the local server, enter the following URL into your web browser:
http://localhost:8000/
By default, it serves the contents of the AngularJS project directory.
-
To access the locally served docs, visit this URL:
http://localhost:8000/build/docs/
We write unit and integration tests with Jasmine and execute them with Karma. To run all of the tests once on Chrome run:
yarn grunt test:unit
To run the tests on other browsers use the command line flag:
yarn grunt test:unit --browsers=Chrome,Firefox
Note: there should be no spaces between browsers. Chrome, Firefox
is INVALID.
If you have a Saucelabs or Browserstack account, you can also run the unit tests on these services via our pre-defined customLaunchers. See the karma config file for all pre-configured browsers.
For example, to run the whole unit test suite on selected browsers:
# Browserstack
yarn grunt test:unit --browsers=BS_Chrome,BS_Firefox,BS_Safari,BS_IE_9,BS_IE_10,BS_IE_11,BS_EDGE,BS_iOS_10
# Saucelabs
yarn grunt test:unit --browsers=SL_Chrome,SL_Firefox,SL_Safari,SL_IE_9,SL_IE_10,SL_IE_11,SL_EDGE,SL_iOS_10
Running these commands requires you to set up Karma Browserstack or Karma-Saucelabs, respectively.
During development, however, it's more productive to continuously run unit tests every time the source or test files change. To execute tests in this mode run:
-
To start the Karma server, capture Chrome browser and run unit tests, run:
yarn grunt autotest
-
To capture more browsers, open this URL in the desired browser (URL might be different if you have multiple instance of Karma running, read Karma's console output for the correct URL):
http://localhost:9876/
-
To re-run tests just change any source or test file.
To learn more about all of the preconfigured Grunt tasks run:
yarn grunt --help
AngularJS's end to end tests are run with Protractor. Simply run:
yarn grunt test:e2e
This will start the webserver and run the tests on Chrome.
To ensure consistency throughout the source code, keep these rules in mind as you are working:
- All features or bug fixes must be tested by one or more specs.
- All public API methods must be documented with ngdoc, an extended version of jsdoc (we added support for markdown and templating via @ngdoc tag). To see how we document our APIs, please check out the existing source code and see the section about writing documentation
- With the exceptions listed below, we follow the rules contained in
Google's JavaScript Style Guide:
- Do not use namespaces: Instead, wrap the entire AngularJS code base in an anonymous closure and export our API explicitly rather than implicitly.
- Wrap all code at 100 characters.
- Instead of complex inheritance hierarchies, we prefer simple objects. We use prototypal inheritance only when absolutely necessary.
- We love functions and closures and, whenever possible, prefer them over objects.
- To write concise code that can be better minified, we use aliases internally that map to the external API. See our existing code to see what we mean.
- We don't go crazy with type annotations for private internal APIs unless it's an internal API that is used throughout AngularJS. The best guidance is to do what makes the most sense.
When adding configuration (options) to providers, we follow a special pattern.
- for each option, add a
method
that ...- works as a getter and returns the current value when called without argument
- works as a setter and returns itself for chaining when called with argument
- for boolean options, uses the naming scheme
<option>Enabled([enabled])
- non-primitive options (e.g. objects) should be copied or the properties assigned explicitly to a new object so that the configuration cannot be changed during runtime.
For a boolean config example, see $compileProvider#debugInfoEnabled
For an object config example, see $location.html5Mode
User-facing errors should be thrown with minErr
, a special error function that provides
errors ids, templated error messages, and adds a link to a detailed error description.
The $compile:badrestrict
error is a good example for a well-defined minErr
:
code and description.
We have very precise rules over how our git commit messages can be formatted. This leads to more readable messages that are easy to follow when looking through the project history. But also, we use the git commit messages to generate the AngularJS change log.
The commit message formatting can be added using a typical git workflow or through the use of a CLI
wizard (Commitizen). To use the wizard, run yarn run commit
in your terminal after staging your changes in git.
Each commit message consists of a header, a body and a footer. The header has a special format that includes a type, a scope and a subject:
<type>(<scope>): <subject>
<BLANK LINE>
<body>
<BLANK LINE>
<footer>
The header is mandatory and the scope of the header is optional.
Any line of the commit message cannot be longer 100 characters! This allows the message to be easier to read on GitHub as well as in various git tools.
If the commit reverts a previous commit, it should begin with revert:
, followed by the header
of the reverted commit.
In the body it should say: This reverts commit <hash>.
, where the hash is the SHA of the commit
being reverted.
A commit with this format is automatically created by the git revert
command.
Must be one of the following:
- feat: A new feature
- fix: A bug fix
- docs: Documentation only changes
- style: Changes that do not affect the meaning of the code (white-space, formatting, missing semi-colons, etc)
- refactor: A code change that neither fixes a bug nor adds a feature
- perf: A code change that improves performance
- test: Adding missing or correcting existing tests
- chore: Changes to the build process or auxiliary tools and libraries such as documentation generation
The scope could be anything specifying place of the commit change. For example $location
,
$browser
, $compile
, $rootScope
, ngHref
, ngClick
, ngView
, etc...
You can use *
when the change affects more than a single scope.
The subject contains succinct description of the change:
- use the imperative, present tense: "change" not "changed" nor "changes"
- don't capitalize first letter
- no dot (.) at the end
Just as in the subject, use the imperative, present tense: "change" not "changed" nor "changes". The body should include the motivation for the change and contrast this with previous behavior.
The footer should contain any information about Breaking Changes and is also the place to reference GitHub issues that this commit closes.
Breaking Changes should start with the word BREAKING CHANGE:
with a space or two newlines.
The rest of the commit message is then used for this.
A detailed explanation can be found in this document.
The AngularJS project uses a form of jsdoc called ngdoc for all of its code documentation.
This means that all the docs are stored inline in the source code and so are kept in sync as it changes.
There is also extra content (the developer guide, error pages, the tutorial, and misceallenous pages) that live inside the AngularJS repository as markdown files.
This means that since we generate the documentation from the source code, we can easily provide version-specific documentation by simply checking out a version of AngularJS and running the build.
Extracting the source code documentation, processing and building the docs is handled by the documentation generation tool Dgeni.
The docs can be built from scratch using grunt:
yarn grunt docs
This defers the doc-building task to gulp
.
Note that the docs app is using the local build files to run. This means you might first have to run the build:
yarn grunt build
(This is also necessary if you are making changes to minErrors).
To view the docs, see Running a Local Development Web Server.
The ngdoc utility has basic support for many of the standard jsdoc directives. But in particular it is interested in the following block tags:
-
@name name
- the name of the ngdoc document -
@param {type} name description
- describes a parameter of a function -
@returns {type} description
- describes what a function returns -
@requires
- normally indicates that a JavaScript module is required; in an Angular service it is used to describe what other services this service relies on -
@property
- describes a property of an object -
@description
- used to provide a description of a component in markdown -
@link
- specifies a link to a URL or a type in the API reference. Links to the API have the following structure:- the module namespace, followed by
.
(optional, defaultng
) - the
@ngdoc
type (see below), followed by:
(optional, automatically inferred) - the name
- the method, property, or anchor (optional)
- the display name
For example:
{@link ng.type:$rootScope.Scope#$new Scope.$new()}
. - the module namespace, followed by
-
@example
- specifies an example. This can be a simple code block, or a runnable example. -
@deprecated
- specifies that the following code is deprecated and should not be used. In The AngularJS docs, there are two specific patterns which can be used to further describe the deprecation:sinceVersion="<version>"
andremoveVersion="<version>"
The type
in @param
and @returns
must be wrapped in {}
curly braces, e.g. {Object|Array}
.
Parameters can be made optional by either appending a =
to the type, e.g. {Object=}
, or by
putting the [name]
in square brackets.
Default values are only possible with the second syntax by appending =<value>
to the parameter
name, e.g. @param {boolean} [ownPropsOnly=false]
.
Descriptions can contain markdown formatting.
In addition to the standard jsdoc tags, there are a number that are specific to the Angular code-base:
@ngdoc
- specifies the type of thing being documented. See below for more detail.@eventType emit|broadcast
- specifies whether the event is emitted or broadcast@usage
- shows how to use afunction
ordirective
. Is usually automatically generated.@knownIssue
- adds info about known quirks, problems, or limitations with the API, and possibly, workarounds. This section is not for bugs.
The following are specific to directives:
@animations
- specifies the animations a directive supports@multiElement
- specifies if a directive can span over multiple elements@priority
- specifies a directive's priority@restrict
- is extracted to show the usage of a directive. For example, for [E]lement, [A]ttribute, and [C]lass, use@restrict ECA
@scope
- specifies that a directive will create a new scope
This directive helps to specify the template used to render the item being documented. For instance, a directive would have different properties to a filter and so would be documented differently. The commonly used types are:
overview
- a general page (guide, api index)provider
- AngularJS provider, such as$compileProvider
or$httpProvider
.service
- injectable AngularJS service, such as$compile
or$http
.object
- well defined object (often exposed as a service)function
- function that will be available to other methods (such as a helper function within the ng module)method
- method on an object/service/controllerproperty
- property on an object/service/controllerevent
- AngularJS event that will propagate through the$scope
tree.directive
- AngularJS directivefilter
- AngularJS filtererror
- minErr error description
Any text in tags can contain markdown syntax for formatting. Generally, you can use any markdown feature.
Only use h2 headings and lower, as the page title is set in h1. Also make sure you follow the heading hierarchy. This ensures correct table of contents are created.
In line code can be specified by enclosing the code in back-ticks (`). A block of multi-line code can be enclosed in triple back-ticks (```) but it is formatted better if it is enclosed in <pre>...</pre> tags and the code lines themselves are indented.
It is possible to embed examples in the documentation along with appropriate e2e tests. These examples and scenarios will be converted to runnable code within the documentation. So it is important that they work correctly. To ensure this, all these e2e scenarios are run as part of the automated Travis tests.
If you are adding an example with an e2e test, you should run the test locally first
to ensure it passes. You can change it(...)
to fit(...)
to run only your test,
but make sure you change it back to it(...)
before committing.
This tag identifies a block of HTML that will define a runnable example. It can take the following attributes:
animations
- if set totrue
then this example uses ngAnimate.deps
- Semicolon-separated list of additional angular module files to be loaded, e.g.angular-animate.js
name
- every example should have a name. It should start with the component, e.g directive name, and not contain whitespacemodule
- the name of the app module as defined in the example's JavaScript
Within this tag we provide <file>
tags that specify what files contain the example code.
<example
module="angularAppModule"
name="exampleName"
deps="angular-animate.js;angular-route.js"
animations="true">
...
<file name="index.html">...</file>
<file name="script.js">...</file>
<file name="animations.css">...</file>
<file name="protractor.js">...</file>
...
</example>
You can see an example of a well-defined example in the ngRepeat
documentation.