On every change detection cycle, Angular synchronously:
Evaluates all template expressions in all components, unless specified otherwise, based on that each component's detection strategy
Executes the
ngDoCheck
,
ngAfterContentChecked
,
ngAfterViewChecked
, and
ngOnChanges
lifecycle hooks.
A single slow computation within a template or a lifecycle hook can slow down the entire change detection process because Angular runs the computations sequentially.
Identifying slow computations
link
You can identify heavy computations with Angular DevTools’ profiler. In the performance timeline, click a bar to preview a particular change detection cycle. This displays a bar chart, which shows how long the framework spent in change detection for each component. When you click a component, you can preview how long Angular spent evaluating its template and lifecycle hooks.
For example, in the preceding screenshot, the second recorded change detection cycle is selected. Angular spent over 573 ms on this cycle, with the most time spent in the
EmployeeListComponent
. In the details panel, you can see that Angular spent over 297 ms evaluating the template of the
EmployeeListComponent
.
Optimizing slow computations
link
Here are several techniques to remove slow computations:
Optimizing the underlying algorithm
. This is the recommended approach. If you can speed up the algorithm that is causing the problem, you can speed up the entire change detection mechanism.
Caching using pure pipes
. You can move the heavy computation to a pure pipe. Angular reevaluates a pure pipe only if it detects that its inputs have changed, compared to the previous time Angular called it.
Using memoization
. Memoization is a similar technique to pure pipes, with the difference that pure pipes preserve only the last result from the computation where memoization could store multiple results.
Avoid repaints/reflows in lifecycle hooks
. Certain operations cause the browser to either synchronously recalculate the layout of the page or re-render it. Since reflows and repaints are generally slow, you want to avoid performing them in every change detection cycle.
Pure pipes and memoization have different trade-offs. Pure pipes are an Angular built-in concept compared to memoization, which is a general software engineering practice for caching function results. The memory overhead of memoization could be significant if you invoke the heavy computation frequently with different arguments.
Zone.js
is a signaling mechanism that Angular uses to detect when an application state might have changed. It captures asynchronous operations like
setTimeout
, network requests, and event listeners. Angular schedules change detection based on signals from Zone.js
In some cases scheduled tasks or microtasks don’t make any changes in the data model, which makes running change detection unnecessary. Common examples are:
requestAnimationFrame
,
setTimeout
or
setInterval
Task or microtask scheduling by third-party libraries
This section covers how to identify such conditions, and how to run code outside the Angular zone to avoid unnecessary change detection calls.
Identifying unnecessary change detection calls
link
You can detect unnecessary change detection calls using Angular DevTools. Often they appear as consecutive bars in the profiler’s timeline with source
setTimeout
,
setInterval
,
requestAnimationFrame
, or an event handler. When you have limited calls within your application of these APIs, the change detection invocation is usually caused by a third-party library.
In the image above, there is a series of change detection calls triggered by event handlers associated with an element. That’s a common challenge when using third-party, non-native Angular components, which do not alter the default behavior of
NgZone
.
Run tasks outside
NgZone
link
In such cases, you can instruct Angular to avoid calling change detection for tasks scheduled by a given piece of code using NgZone.
import{Component,NgZone,OnInit} from '@angular/core';@Component(...)classAppComponentimplementsOnInit{constructor(private ngZone:NgZone){}
ngOnInit(){this.ngZone.runOutsideAngular(()=> setInterval(pollForUpdates),500);}}
The preceding snippet instructs Angular to call
setInterval
outside the Angular Zone and skip running change detection after
pollForUpdates
runs.
Third-party libraries commonly trigger unnecessary change detection cycles because they weren't authored with Zone.js in mind. Avoid these extra cycles by calling library APIs outside the Angular zone:
import{Component,NgZone,OnInit} from '@angular/core';import* as Plotly from 'plotly.js-dist-min';@Component(...)classAppComponentimplementsOnInit{constructor(private ngZone:NgZone){}
ngOnInit(){this.ngZone.runOutsideAngular(()=>{Plotly.newPlot('chart', data);});}}
Running
Plotly.newPlot('chart', data);
within
runOutsideAngular
instructs the framework that it shouldn’t run change detection after the execution of tasks scheduled by the initialization logic.
For example, if
Plotly.newPlot('chart', data)
adds event listeners to a DOM element, Angular does not run change detection after the execution of their handlers.
Angular - Angular change detection and runtime optimization
Angular change detection and runtime optimization
link
Change detection
is the process through which Angular checks to see whether your application state has changed, and if any DOM needs to be updated. At a high level, Angular walks your components from top to bottom, looking for changes. Angular runs its change detection mechanism periodically so that changes to the data model are reflected in an application’s view. Change detection can be triggered either manually or through an asynchronous event (for example, a user interaction or an XMLHttpRequest completion).
Change detection is a highly optimized performant, but it can still cause slowdowns if the application runs it too frequently.
In this guide, you’ll learn how to control and optimize the change detection mechanism by skipping parts of your application and running change detection only when necessary.
Watch this video if you prefer to learn more about performance optimizations in a media format:
<video #movieplayer …></video><button (click)="movieplayer.play()">
Play
</button>
Creates a local variable
movieplayer
that provides access to the
video
element instance in data-binding and event-binding expressions in the current template.
<p *myUnless="myExpression">
…
</p>
The asterisk (
*
) character turns the current element into an embedded template. Equivalent to:
Binds the presence of CSS classes on the element to the truthiness of the associated map values. The right-hand expression should return
{class-name: true/false}
map.
Allows you to assign styles to an HTML element using CSS. You can use CSS directly, as in the first example, or you can call a method from the component.
Forms
Details
import{FormsModule} from '@angular/forms';
Import
FormsModule
from
@angular/forms
.
<input [(ngModel)]="userName">
Provides two-way data-binding, parsing, and validation for form controls.
Class decorators
Details
import{Directive,…} from '@angular/core';
Import
Directive, …
from
@angular/core';
.
@Component({…})classMyComponent(){}
Declares that a class is a component and provides metadata about the component.
@Directive({…})classMyDirective(){}
Declares that a class is a directive and provides metadata about the directive.
@Pipe({…})classMyPipe(){}
Declares that a class is a pipe and provides metadata about the pipe.
@Injectable()classMyService(){}
Declares that a class can be provided and injected by other classes. Without this decorator, the compiler won't generate enough metadata to allow the class to be created properly when it's injected somewhere.
Directive configuration
Details
@Directive({
property1: value1,…})
Add
property1
property with
value1
value to Directive.
selector:'.cool-button:not(a)'
Specifies a CSS selector that identifies this directive within a template. Supported selectors include
element
,
[attribute]
,
.class
, and
:not()
.
Does not support parent-child relationship selectors.
providers:[MyService,{ provide:…}]
List of dependency injection providers for this directive and its children.
Component configuration
@Component
extends
@Directive
, so the
@Directive
configuration applies to components as well
Details
moduleId: module.id
If set, the
templateUrl
and
styleUrl
are resolved relative to the component.
viewProviders:[MyService,{ provide:…}]
List of dependency injection providers scoped to this component's view.
Creates a link to a different view based on a route instruction consisting of a route path, required and optional parameters, query parameters, and a fragment. To navigate to a root route, use the
/
prefix; for a child route, use the
./
prefix; for a sibling or parent, use the
../
prefix.
The provided classes and
aria-current
attribute are added to the element when the
routerLink
becomes the current active route.
function canActivateGuard:CanActivateFn=(
route:ActivatedRouteSnapshot,
state:RouterStateSnapshot)=>{…}{ path:…, canActivate:[canActivateGuard]}
An interface for defining a function that the router should call first to determine if it should activate this component. Should return a
boolean|UrlTree
or an Observable/Promise that resolves to a
boolean|UrlTree
.
function canDeactivateGuard:CanDeactivateFn<T>=(
component: T,
route:ActivatedRouteSnapshot,
state:RouterStateSnapshot)=>{…}{ path:…, canDeactivate:[canDeactivateGuard]}
An interface for defining a function that the router should call first to determine if it should deactivate this component after a navigation. Should return a
boolean|UrlTree
or an Observable/Promise that resolves to a
boolean|UrlTree
.
function canActivateChildGuard:CanActivateChildFn=(
route:ActivatedRouteSnapshot,
state:RouterStateSnapshot)=>{…}{ path:…, canActivateChild:[canActivateGuard], children:…}
An interface for defining a function that the router should call first to determine if it should activate the child route. Should return a
boolean|UrlTree
or an Observable/Promise that resolves to a
boolean|UrlTree
.
function resolveGuard implementsResolveFn<T>=(
route:ActivatedRouteSnapshot,
state:RouterStateSnapshot)=>{…}{ path:…, resolve:[resolveGuard]}
An interface for defining a function that the router should call first to resolve route data before rendering the route. Should return a value or an Observable/Promise that resolves to a value.
function canLoadGuard:CanLoadFn=(
route:Route)=>{…}{ path:…, canLoad:[canLoadGuard], loadChildren:…}
An interface for defining a function that the router should call first to check if the lazy loaded module should be loaded. Should return a
boolean|UrlTree
or an Observable/Promise that resolves to a
boolean|UrlTree
.
Use class and style bindings to add and remove CSS class names from an element's
class
attribute and to set styles dynamically.
Prerequisites
link
Property binding
Binding to a single CSS
class
link
To create a single class binding, type the following:
[class.sale]="onSale"
Angular adds the class when the bound expression,
onSale
is truthy, and it removes the class when the expression is falsy—with the exception of
undefined
. See styling delegation for more information.
Binding to multiple CSS classes
link
To bind to multiple classes, type the following:
[class]="classExpression"
The expression can be one of:
A space-delimited string of class names.
An object with class names as the keys and truthy or falsy expressions as the values.
An array of class names.
With the object format, Angular adds a class only if its associated value is truthy.
With any object-like expression—such as
object
,
Array
,
Map
, or
Set
—the identity of the object must change for Angular to update the class list.
Updating the property without changing object identity has no effect.
If there are multiple bindings to the same class name, Angular uses styling precedence to determine which binding to use.
The following table summarizes class binding syntax.
Binding Type
Syntax
Input Type
Example Input Values
Single class binding
[class.sale]="onSale"
boolean | undefined | null
true
,
false
Multi-class binding
[class]="classExpression"
string
"my-class-1 my-class-2 my-class-3"
Multi-class binding
[class]="classExpression"
Record<string, boolean | undefined | null>
{foo: true, bar: false}
Multi-class binding
[class]="classExpression"
Array<string>
['foo', 'bar']
Binding to a single style
link
To create a single style binding, use the prefix
style
followed by a dot and the name of the CSS style.
For example, to set the
width
style, type the following:
[style.width]="width"
Angular sets the property to the value of the bound expression, which is usually a string. Optionally, you can add a unit extension like
em
or
%
, which requires a number type.
To write a style in dash-case, type the following:
<nav [style.background-color]="expression"></nav>
To write a style in camelCase, type the following:
<nav [style.backgroundColor]="expression"></nav>
Binding to multiple styles
link
To toggle multiple styles, bind to the
[style]
attribute—for example,
[style]="styleExpression"
. The
styleExpression
can be one of:
A string list of styles such as
"width: 100px; height: 100px; background-color: cornflowerblue;"
.
An object with style names as the keys and style values as the values, such as
{width: '100px', height: '100px', backgroundColor: 'cornflowerblue'}
.
Note that binding an array to
[style]
is not supported.
When binding
[style]
to an object expression, the identity of the object must change for Angular to update the class list.
Updating the property without changing object identity has no effect.
A number of Angular CLI commands run a complex process on your code, such as linting, building, or testing.
The commands use an internal tool called Architect to run
CLI builders
, which apply another tool to accomplish the wanted task.
With Angular version 8, the CLI Builder API is stable and available to developers who want to customize the Angular CLI by adding or modifying commands.
For example, you could supply a builder to perform an entirely new task, or to change which third-party tool is used by an existing command.
This document explains how CLI builders integrate with the workspace configuration file, and shows how you can create your own builder.
Find the code from the examples used here in this GitHub repository.
CLI builders
link
The internal Architect tool delegates work to handler functions called
builders
.
A builder handler function receives two arguments; a set of input
options
(a JSON object), and a
context
(a
BuilderContext
object).
The separation of concerns here is the same as with schematics, which are used for other CLI commands that touch your code (such as
ng generate
).
The
options
object is provided by the CLI user, while the
context
object is provided by the CLI Builder API
In addition to the contextual information, the
context
object, which is an instance of the
BuilderContext
, also provides access to a scheduling method,
context.scheduleTarget()
.
The scheduler executes the builder handler function with a given target configuration.
The builder handler function can be synchronous (return a value) or asynchronous (return a Promise), or it can watch and return multiple values (return an Observable).
The return value or values must always be of type
BuilderOutput
.
This object contains a Boolean
success
field and an optional
error
field that can contain an error message.
Angular provides some builders that are used by the CLI for commands such as
ng build
and
ng test
.
Default target configurations for these and other built-in CLI builders can be found (and customized) in the "architect" section of the workspace configuration file,
angular.json
.
Also, extend and customize Angular by creating your own builders, which you can run using the
ng run
CLI command.
Builder project structure
link
A builder resides in a "project" folder that is similar in structure to an Angular workspace, with global configuration files at the top level, and more specific configuration in a source folder with the code files that define the behavior.
For example, your
myBuilder
folder could contain the following files.
Files
Purpose
src/my-builder.ts
Main source file for the builder definition.
src/my-builder.spec.ts
Source file for tests.
src/schema.json
Definition of builder input options.
builders.json
Builders definition.
package.json
Dependencies. See https://docs.npmjs.com/files/package.json.
tsconfig.json
TypeScript configuration.
Publish the builder to
npm
(see Publishing your Library).
If you publish it as
@example/my-builder
, install it using the following command.
npm install @example/my-builder
Creating a builder
link
As an example, create a builder that copies a file.
To create a builder, use the
createBuilder()
CLI Builder function, and return a
Promise<BuilderOutput>
object.
Now let's add some logic to it.
The following code retrieves the source and destination file paths from user options and copies the file from the source to the destination (using the Promise version of the built-in NodeJS
copyFile()
function).
If the copy operation fails, it returns an error with a message about the underlying problem.
By default,
copyFile()
does not print anything to the process standard output or error.
If an error occurs, it might be difficult to understand exactly what the builder was trying to do when the problem occurred.
Add some additional context by logging additional information using the
Logger
API.
This also lets the builder itself be executed in a separate process, even if the standard output and error are deactivated (as in an Electron app).
You can retrieve a
Logger
instance from the context.
The CLI Builder API includes progress and status reporting tools, which can provide hints for certain functions and interfaces.
To report progress, use the
context.reportProgress()
method, which takes a current value, (optional) total, and status string as arguments.
The total can be any number; for example, if you know how many files you have to process, the total could be the number of files, and current should be the number processed so far.
The status string is unmodified unless you pass in a new string value.
You can see an example of how the
tslint
builder reports progress.
In our example, the copy operation either finishes or is still executing, so there's no need for a progress report, but you can report status so that a parent builder that called our builder would know what's going on.
Use the
context.reportStatus()
method to generate a status string of any length.
NOTE
:
There's no guarantee that a long string will be shown entirely; it could be cut to fit the UI that displays it.
Pass an empty string to remove the status.
src/my-builder.ts (progress reporting)
import{BuilderContext,BuilderOutput, createBuilder }from'@angular-devkit/architect';import{JsonObject}from'@angular-devkit/core';import{ promises as fs }from'fs';interfaceOptionsextendsJsonObject{
source:string;
destination:string;}exportdefault createBuilder(copyFileBuilder);asyncfunction copyFileBuilder(
options:Options,
context:BuilderContext,):Promise<BuilderOutput>{
context.reportStatus(`Copying ${options.source} to ${options.destination}.`);try{await fs.copyFile(options.source, options.destination);}catch(err){
context.logger.error('Failed to copy file.');return{
success:false,
error: err.message,};}
context.reportStatus('Done.');return{ success:true};}
Builder input
link
You can invoke a builder indirectly through a CLI command, or directly with the Angular CLI
ng run
command.
In either case, you must provide required inputs, but can let other inputs default to values that are pre-configured for a specific
target
, provide a pre-defined, named override configuration, and provide further override option values on the command line.
Input validation
link
You define builder inputs in a JSON schema associated with that builder.
The Architect tool collects the resolved input values into an
options
object, and validates their types against the schema before passing them to the builder function.
(The Schematics library does the same kind of validation of user input.)
For our example builder, you expect the
options
value to be a
JsonObject
with two keys:
A
source
and a
destination
, each of which are a string.
You can provide the following schema for type validation of these values.
This is a very simple example, but the use of a schema for validation can be very powerful.
For more information, see the JSON schemas website.
To link our builder implementation with its schema and name, you need to create a
builder definition
file, which you can point to in
package.json
.
Create a file named
builders.json
that looks like this:
builders.json
{"builders":{"copy":{"implementation":"./dist/my-builder.js","schema":"./src/schema.json","description":"Copies a file."}}}
In the
package.json
file, add a
builders
key that tells the Architect tool where to find our builder definition file.
package.json
{"name":"@example/copy-file","version":"1.0.0","description":"Builder for copying files","builders":"builders.json","dependencies":{"@angular-devkit/architect":"~0.1200.0","@angular-devkit/core":"^12.0.0"}}
The official name of our builder is now
@example/copy-file:copy
.
The first part of this is the package name (resolved using node resolution), and the second part is the builder name (resolved using the
builders.json
file).
Using one of our
options
is very straightforward.
You did this in the previous section when you accessed
options.source
and
options.destination
.
src/my-builder.ts (report status)
context.reportStatus(`Copying ${options.source} to ${options.destination}.`);try{await fs.copyFile(options.source, options.destination);}catch(err){
context.logger.error('Failed to copy file.');return{
success:false,
error: err.message,};}
context.reportStatus('Done.');return{ success:true};
Target configuration
link
A builder must have a defined target that associates it with a specific input configuration and project.
Targets are defined in the
angular.json
CLI configuration file.
A target specifies the builder to use, its default options configuration, and named alternative configurations.
The Architect tool uses the target definition to resolve input options for a given run.
The
angular.json
file has a section for each project, and the "architect" section of each project configures targets for builders used by CLI commands such as 'build', 'test', and 'lint'.
By default, for example, the
build
command runs the builder
@angular-devkit/build-angular:browser
to perform the build task, and passes in default option values as specified for the
build
target in
angular.json
.
The command passes the builder the set of default options specified in the "options" section.
If you pass the
--configuration=production
flag, it uses the override values specified in the
production
alternative configuration.
Specify further option overrides individually on the command line.
You might also add more alternative configurations to the
build
target, to define other environments such as
stage
or
qa
.
Target strings
link
The generic
ng run
CLI command takes as its first argument a target string of the following form.
project:target[:configuration]
Details
project
The name of the Angular CLI project that the target is associated with.
target
A named builder configuration from the
architect
section of the
angular.json
file.
configuration
(optional) The name of a specific configuration override for the given target, as defined in the
angular.json
file.
If your builder calls another builder, it might need to read a passed target string.
Parse this string into an object by using the
targetFromTargetString()
utility function from
@angular-devkit/architect
.
Schedule and run
link
Architect runs builders asynchronously.
To invoke a builder, you schedule a task to be run when all configuration resolution is complete.
The builder function is not executed until the scheduler returns a
BuilderRun
control object.
The CLI typically schedules tasks by calling the
context.scheduleTarget()
function, and then resolves input options using the target definition in the
angular.json
file.
Architect resolves input options for a given target by taking the default options object, then overwriting values from the configuration used (if any), then further overwriting values from the overrides object passed to
context.scheduleTarget()
.
For the Angular CLI, the overrides object is built from command line arguments.
Architect validates the resulting options values against the schema of the builder.
If inputs are valid, Architect creates the context and executes the builder.
For more information see Workspace Configuration.
You can also invoke a builder directly from another builder or test by calling
context.scheduleBuilder()
.
You pass an
options
object directly to the method, and those option values are validated against the schema of the builder without further adjustment.
Only the
context.scheduleTarget()
method resolves the configuration and overrides through the
angular.json
file.
Default architect configuration
link
Let's create a simple
angular.json
file that puts target configurations into context.
You can publish the builder to npm (see Publishing your Library), and install it using the following command:
npm install @example/copy-file
If you create a new project with
ng new builder-test
, the generated
angular.json
file looks something like this, with only default builder configurations.
angular.json
{//…"projects":{//…"builder-test":{//…"architect":{//…"build":{"builder":"@angular-devkit/build-angular:browser","options":{//… more options…"outputPath":"dist/builder-test","index":"src/index.html","main":"src/main.ts","polyfills":"src/polyfills.ts","tsConfig":"src/tsconfig.app.json"},"configurations":{"production":{//… more options…"optimization":true,"aot":true,"buildOptimizer":true}}}}}}//…}
Adding a target
link
Add a new target that will run our builder to copy a file.
This target tells the builder to copy the
package.json
file.
You need to update the
angular.json
file to add a target for this builder to the "architect" section of our new project.
We'll add a new target section to the "architect" object for our project
The target named "copy-package" uses our builder, which you published to
@example/copy-file
.
(See Publishing your Library.)
The options object provides default values for the two inputs that you defined;
source
, which is the existing file you are copying, and
destination
, the path you want to copy to
The
configurations
key is optional, we'll leave it out for now
To run our builder with the new target's default configuration, use the following CLI command.
ng run builder-test:copy-package
This copies the
package.json
file to
package-copy.json
.
Use command-line arguments to override the configured defaults.
For example, to run with a different
destination
value, use the following CLI command.
ng run builder-test:copy-package--destination=package-other.json
This copies the file to
package-other.json
instead of
package-copy.json
.
Because you did not override the
source
option, it will copy from the
package.json
file (the default value provided for the target).
Testing a builder
link
Use integration testing for your builder, so that you can use the Architect scheduler to create a context, as in this example.
In the builder source directory, you have created a new test file
my-builder.spec.ts
.
The code creates new instances of
JsonSchemaRegistry
(for schema validation),
TestingArchitectHost
(an in-memory implementation of
ArchitectHost
), and
Architect
.
We've added a
builders.json
file next to the builder's
package.json
file, and modified the package file to point to it.
Here's an example of a test that runs the copy file builder.
The test uses the builder to copy the
package.json
file and validates that the copied file's contents are the same as the source.
src/my-builder.spec.ts
import{Architect}from'@angular-devkit/architect';import{TestingArchitectHost}from'@angular-devkit/architect/testing';import{ schema }from'@angular-devkit/core';import{ promises as fs }from'fs';
describe('Copy File Builder',()=>{let architect:Architect;let architectHost:TestingArchitectHost;
beforeEach(async()=>{const registry =new schema.CoreSchemaRegistry();
registry.addPostTransform(schema.transforms.addUndefinedDefaults);// TestingArchitectHost() takes workspace and current directories.// Since we don't use those, both are the same in this case.
architectHost =newTestingArchitectHost(__dirname, __dirname);
architect =newArchitect(architectHost, registry);// This will either take a Node package name, or a path to the directory// for the package.json file.await architectHost.addBuilderFromPackage('..');});
it('can copy files',async()=>{// A "run" can have multiple outputs, and contains progress information.const run =await architect.scheduleBuilder('@example/copy-file:copy',{
source:'package.json',
destination:'package-copy.json',});// The "result" member (of type BuilderOutput) is the next output.const output =await run.result;// Stop the builder from running. This stops Architect from keeping// the builder-associated states in memory, since builders keep waiting// to be scheduled.await run.stop();// Expect that the copied file is the same as its source.const sourceContent =await fs.readFile('package.json','utf8');const destinationContent =await fs.readFile('package-copy.json','utf8');
expect(destinationContent).toBe(sourceContent);});});
When running this test in your repo, you need the
ts-node
package.
You can avoid this by renaming
my-builder.spec.ts
to
my-builder.spec.js
.
Watch mode
link
Architect expects builders to run once (by default) and return.
This behavior is not entirely compatible with a builder that watches for changes (like Webpack, for example).
Architect can support watch mode, but there are some things to look out for.
To be used with watch mode, a builder handler function should return an Observable.
Architect subscribes to the Observable until it completes and might reuse it if the builder is scheduled again with the same arguments.
The builder should always emit a
BuilderOutput
object after each execution.
Once it's been executed, it can enter a watch mode, to be triggered by an external event.
If an event triggers it to restart, the builder should execute the
context.reportRunning()
function to tell Architect that it is running again.
This prevents Architect from stopping the builder if another run is scheduled.
When your builder calls
BuilderRun.stop()
to exit watch mode, Architect unsubscribes from the builder's Observable and calls the builder's teardown logic to clean up.
(This behavior also allows for long-running builds to be stopped and cleaned up.)
In general, if your builder is watching an external event, you should separate your run into three phases.
Phases
Details
Running
For example, webpack compiles. This ends when webpack finishes and your builder emits a
BuilderOutput
object.
Watching
Between two runs, watch an external event stream. For example, webpack watches the file system for any changes. This ends when webpack restarts building, and
context.reportRunning()
is called. This goes back to step 1.
Completion
Either the task is fully completed (for example, webpack was supposed to run a number of times), or the builder run was stopped (using
BuilderRun.stop()
). Your teardown logic is executed, and Architect unsubscribes from your builder's Observable.
Summary
link
The CLI Builder API provides a new way of changing the behavior of the Angular CLI by using builders to execute custom logic.
Builders can be synchronous or asynchronous, execute once or watch for external events, and can schedule other builders or targets
Builders have option defaults specified in the
angular.json
configuration file, which can be overwritten by an alternate configuration for the target, and further overwritten by command line flags
We recommend that you use integration tests to test Architect builders.
Use unit tests to validate the logic that the builder executes.
If your builder returns an Observable, it should clean up in the teardown logic of that Observable