- Schema
- Environment-specific config
- Properties
- Displaying config information
- Config structure
- Sources of configuration information
- Configuring loaders
- Configuring plugins
- Configuration resolution
Intern is configured with a standard JavaScript object. This object may contain
properties applicable to either environment that Intern can run in (Node or
browser). Config properties may be set via a file, the command line, browser
query args, or an environment variable. All of these methods use the same basic
syntax and provide the same capabilities. Assuming Intern is being run with the
default Node runner or browser runner
and without a config
argument, Intern will attempt to load configuration
informatioon from an intern.json
file in the project root.
Wherever config property values come from, the executor will validate and
normalize them into a canonical format ("resolve" them) when the testing process
starts. This allows the executor’s constructor or configure
method to be
flexible in what data it accepts. For example, the canonical form of the
environments
property is an array of objects:
environments: [{ browserName: 'chrome' }];
However, Intern will accept a simple string for the environments
property and
will expand it into an array of a single object where the browserName
property
is the given string.
Intern includes a
JSON schema for
its config format. This schema describes all the properties that might be
included in an intern.json
config file, and includes help text for each
property.
To use the schema with an editor such as VS Code, add a $schema
property to
your intern.json
file that points to the schema, like:
{
"$schema": "./node_modules/intern/schemas/config.json"
}
Tests can run in two basic environments: Node and browsers. By default,
“resource” properties (suites
, plugins
, reporters
, and loader
) in a
config file apply to both environments. This means that if the same config is
used to run tests in a browser and in Node, the same resources will be loaded in
both environments. In some cases this isn’t desirable because tests may load
application code that depends on environment-specific properties or features,
such as the DOM. Intern’s config provides node
and browser
properties for
this use case. These properties specify resources that will only be loaded in
the given environment. The values in these properties will be
shallowly mixed into the base config rather than
replacing it.
⚠️ Note that this is different than themode
property in Intern 3, which had values of “client” or “runner”. Intern 3’s mode indicated whether tests were being run in unit test or functional test mode, but it was sometimes used as an environment check due to the fact that functional tests always run in Node.
A number of config properties are applicable whether Intern is running in Node or directly in a browser. Some of the more common ones are listed below.
💡 See the API docs for a complete list of the available config properties.
Property | Description | Default |
---|---|---|
bail | When true, stop testing after the first failure | false |
baseline | When true, run benchmark tests in baseline mode | false |
benchmark | When true, run benchmark tests (if loaded) | false |
debug | If true, display runtime messages to the console | false |
defaultTimeout | The time, in ms, before an async test times out | 30000 |
filterErrorStack | If true, filter non-application code lines out of stack traces | false |
grep | Regular expression used to filter which suites and tests are run | /.*/ |
loader | An optional loader script and options | { script: 'default' } |
plugins | A list of Intern extensions to load before tests begin | [] |
reporters | A list of reporters to use | [] |
suites | A list of suite paths or globs to load unit tests from | [] |
warnOnUnhandledRejection | If true or a regex matching a unhandled rejection, emit a warning instead of an error |
undefined |
warnOnUncaughtException | If true or a regex matching an uncaught error, emit a warning instead of an error |
undefined |
Some properties are only meaningful for Node or WebDriver tests:
Property | Description | Default |
---|---|---|
capabilities | Default capabilities to be used for WebDriver sessions | { 'idle-timeout': 60 } |
coverage | An array of paths or globs to collect coverage data for | [] |
environments | Browser + OS combinations to be tested using WebDriver | [] |
functionalRetries | rerun failed sessions when at least one other session succeeds | 0 |
functionalSuites | Suites to run in WebDriver mode | [] |
functionalTimeouts | Timeouts used in functional tests | { connectTimeout: 30000 } |
leaveRemoteOpen | If true, leave remote browsers open after testing has finished | false |
maxConcurrency | The maximum number of sessions to drive concurrently | Infinity |
serveOnly | When true, Intern will start its instrumenting web server but not run tests | false |
serverPort | The port the instrumenting server should listen on | 9000 |
serverUrl | A URL a remote executor can use to reach the local Intern | http://localhost:9000 |
tsconfig | Optional path to a tsconfig.json for ts-node. Uses the project's by default. Set to false to explicitly prevent registering tsnode in projects with TypeScript |
`` |
tunnel | The name of a tunnel to use for WebDriver tests | selenium |
tunnelOptions | Options to use for the WebDriver tunnel | { tunnelId: Date.now() } |
The environment-specific properties come into play when Intern is running in that environment:
Property | Description | Default |
---|---|---|
browser | Resources (loader, plugins, reporters, suites) that only apply to browser tests | {} |
node | Resources (loader, plugins, reporters, suites, tsconfig) that only apply to Node tests | {} |
There are also several properties that are handled by the config file processing system that aren’t directly involved in the testing process. These properties are ignored if set programmatically.
Property | Description |
---|---|
description | A short string describing a test config |
extends | Another config or config file that the config extends |
help | Display a help message |
showConfig | When true, show the resolved configuration and exit |
showConfigs | When true, show information about the currently loaded config file |
Suites may be specified as file paths or using glob expressions. Globbing is handled with the glob Node package.
{
suites: ['tests/unit/**/*.js', 'tests/integration/foo.js']
}
Intern also understands glob-based exclusion using the !
modifier:
{
suites: ['tests/**/*.js', '!tests/functional/**']
}
Note that using globs with the browser client requires that Intern’s own server be used to serve test suites. This is because the browser client on its own has no way to resolve file system globs; it passes the to the Intern server running on Node, which resolves them and returns list of file paths.
If the extends
property is set in a base config, it must be the path to a
different config file. At run time, the properties from the config file with the
extends
value will be mixed into the properties from the config file being
extended.
If the extends
property is set in a child config, it must be the name of a
different child config within the same config file, or an array of such names.
When a child config extends multiple other child configs, properties from the
right-most config being extended will override properties from configs to the
left.
{
configs: {
a: {
/* ... */
},
b: {
/* ... */
},
c: {
/* ... */
},
d: {
extends: ['a', 'c']
/* ... */
}
}
}
Given the config above, and assuming child config “d” is selected, the following resolution process will occur:
- Child “c” will be mixed into child “a”
- Child “d” will be mixed into the result of 1
- The result of 2 will be mixed into the base config
- The result of 3 will be the resolved config
Intern has two config properties that can be used to display configuration
information: showConfig
and showConfigs
.
Setting the showConfig
property to true
will cause Intern to dump the
resolved configuration to the current environment’s console.
$ npx intern showConfig
{
"bail": false,
"baseline": false,
"benchmark": false,
"reporters": [
{
"name": "console"
}
]
}
Setting the showConfigs
property to true
will cause Intern to show
information about a given config file. Intern will print the value of the
current config file’s description
property, and then list all child configs
contained in the config file. For example, with a config file containing the
following data:
{
description: 'Default test suite',
configs: {
webdriver: {
description: 'Run webdriver tests'
},
ci: {
description: 'Run tests on a CI server'
}
}
}
running Intern with the showConfigs
property set would display the following
text:
$ npx intern showConfigs
Default test suite
Configs:
webdriver (Run webdriver tests)
ci (Run tests on a CI server)
The config structure is a simple JSON object, so all of its property values must be serializable (RegExp objects are serialized to strings).
{
// General properties
bail: false,
baseline: false,
suites: ['tests/unit/*.js'],
// Browser and node specific resources
browser: {
suites: ['tests/unit/dom_stuff.js']
},
node: {
suites: ['tests/unit/dom_stuff.js']
},
configs: {
// Child configs have the same structure as the main config
ci: {
bail: true,
'suites+': ['tests/unit/other.js']
}
}
}
There are four general sections to a config:
- General properties: this includes everything but "browser", "configs", and "node"
- Node-specific resources: resource properties ("loader", "plugins", "reporters", "suites") that apply only to Node environments.
- Browser-specific resources: resource properties ("loader", "plugins", "reporters", "suites") that apply only to browser environments.
- Child configs: named configs in the "configs" object; each of these can have any config properties except "configs" (i.e., general properties, Node resources, and browser resources).
Intern takes in configuration data from several sources. In order of increasing precedence, they are:
Multiple configuration methods may be used during a single run of Intern. The configuration will be fully resolved before tests are executed.
An Intern config file is a JSON file specifying config properties, for example:
{
environments: [{ browserName: 'chrome' }],
suites: ['tests/unit/all.js']
}
By default, intern will try to load a file named intern.json
from the project
root directory. A different config file can be specified by passing a config
property to the Node or browser runners.
A child config can be selected by adding @<child>
to the config file name. For
example, to load a child config named “ci” from the default config file, you
could run:
$ npx intern config=@ci
To load a config named “remote” from a config file named “intern-local.json”, run:
$ npx intern config=intern-local.json@remote
In a Node environment, Intern may be configured using an INTERN_ARGS
environment variable. This variable is treated just like a string of command
line arguments. For example, these two executions of Intern are equivalent:
$ npx intern grep='run.*' suites=
export INTERN_ARGS="grep=run.* suites="
$ npx intern
Config properties may be provided directly on the command line when starting
Intern. Properties must be specified using property=value
syntax. For example,
$ npx intern grep='run.*' suites=
Object values may be input as serialized strings (e.g.,
environments='{"browserName":"chrome"}'
). Array values may be set by repeating
a property (e.g., suites="foo.js" suites="bar.js"
).
Query args work very similarly to command line args. They have the same format, but with URL query arg separators, and escaping of special characters as necessary.
http://localhost:8080/node_modules/intern/?grep=run.*&suites=
When creating an executor programmatically it may be configured via its
constructor, and/or via its configure
method.
const intern = new Node({ grep: /run.*/, suites: [] });
or
intern.configure({ grep: /run.*/, suites: [] });
The configure
method may be called any number of times before the testing
process is started.
Intern uses loader scripts to communicate with various loaders so test suites of any module format may be used. Loader scripts are simply standalone scripts (like plugins) that connect Intern with a loader. Intern has a number of loader scripts available to help it work with popular loaders and formats. If no loader script is defined, then Intern will use the default loader script for the environment (CommonJS in Node and <script> tag injection in the browser).
To use one of Intern's pre-defined loader scripts, simply specify it's name. The
loader script will expect the loader package to be installed in node_modules
using NPM.
{
browser: {
loader: 'dojo'
}
}
Additional options may be provided through the options parameter. These options are passed through to the registered loader script.
{
browser: {
loader: {
script: './support/my/custom/loader.js',
options: { basePath: '_build' }
}
}
}
It's useful to think of Intern's loader scriptss as glue code that Intern uses to load configured suites within the environment. When Intern needs to load a module (i.e. a test suite) it hands off a list of these modules to the loader script and waits for the loading process to be handled by the loader. For more information and options look at the loader scripts in the Intern repository.
Plugins are standalone scripts or modules that are loaded by Intern. Standalone
plugins (ones that do not require a loader) are ran in-order early in the Intern
lifecycle before practically anything else. This is to allow plugins a chance to
listen to Intern events and augment the environment as early as possible. Module
plugins have a useLoader
property and are loaded later in Intern's lifecycle
with the configured loader. Any number of plugins may be added to Intern.
{
plugins: [
'node_modules/babel-register/lib/node.js',
{
script: 'tests/support/mongodbAccess.js',
options: { dbUrl: 'https://testdb.local' }
},
{
script: 'tests/support/dojoMocking.js',
useLoader: true
}
]
}
Similar to loader scripts, plugin configurations with options
are passed to
the plugin when it is registered with Intern.
At runtime, the environment-specific resources and any active child configs will be mixed into the resolved config. In general, properties from more specific sources will override properties from lower precedence sources. The order of precedence, from lowest to highest, is:
- A config being extended by the current config
- The current config
- An active child config in the current config
There are a few exceptions:
- The "node" and "browser" properties in a child config are shallowly mixed
into "node" and "browser" in the base config. For example, if "node" in
the base config looks like:
and "node" in a child config looks like:
"node": { "suites": [ "tests/unit/foo.js" ], "plugins": [ "tests/plugins/bar.js" ] }
then the value of node in the resolved config (assuming the child config is active) will be:"node": { "suites": [ "tests/unit/baz.js" ], }
"node": { // node.suites from the child overrides node.suites from the base config "suites": [ "tests/unit/baz.js" ], // node.plugins from the base config remains "plugins": [ "tests/plugins/bar.js" ] }
- Resource arrays in "node" or "browser" ("plugins", "reporters", "suites"),
are added to the corresponding resource arrays in the base config. For
example, if the base config has:
and the "node" section has:
"suites": [ "tests/unit/foo.js" ]
both sets of suites will be loaded when running on Node."suites": [ "tests/unit/bar.js" ]
- Some properties can be extended (rather than replaced) by adding a '+' to
the property name. For example, if the base config has:
and a child config has:
"suites": [ "tests/unit/foo.js" ]
the resolved value of suites will be:"suites+": [ "tests/unit/bar.js" ]
Extendable properties are resources (suites, plugins, reporters), instrumenterOptions, tunnelOptions, and capabilities."suites": [ "tests/unit/foo.js", "tests/unit/bar.js" ]