Angular.js, one of the famous JavaScript frameworks for developing rich web applications using HTML5 and latest versions of JavaScript (ES5 and above), has evolved over time.
Angular started as a framework aimed to develop rich web applications, faster and better. It provided structure to the not so evolved JavaScript. As browsers got better, new language features were added to JavaScript. To keep up with these changes, Angular has evolved too.
This tutorial is from the DotNetCurry(DNC) Magazine with in-depth tutorials and best practices in JavaScript and .NET. This magazine is aimed at Developers, Architects and Technical Managers and covers TypeScript, Angular, React, C#, Design Patterns, .NET Core, MVC, Azure, DevOps, ALM, and more. Subscribe to this magazine for FREE and receive all previous, current and upcoming editions, right in your Inbox. No Spam Policy.
This article tries to summarize this evolution process. It attempts to look back at the original context of Angular application development and its progress.
This article does not intend to be a tutorial for all the features over time, rather it tries to summarize and provide a bucket list of enhancements over time.
Angular Evolution - v1.x to v6
AngularJS 1.x Context
AngularJS 1.x is seen as MV* framework.
The Model-View-Controller (MVC) and its variations Model-View-Presenter(MVP) and Model-View-View-Model(MVVM) frameworks have been used with applications for a very long time. Be it web applications developed using Java, Microsoft ASP.NET or iOS apps using Objective C, they have all used an architectural pattern. It is a good fit for all applications that have a view to present to the user.
AngularJS’s MV* approach was to support any variations of these famous patterns. In an AngularJS application, that runs in a web browser,
- Controller is a function defined using the API. An object $scope is injected in the function.
- View is defined by HTML templates. Templates can bind data on $scope from the controller.
- Objects on $scope could be either Model or View-Model, depending on overall architecture including server-side.
Figure 1: MV* framework
Even though this is a proven architectural pattern and is the standard solution for an application with view, there are certain disadvantages.
Modern web applications required reusable components or custom HTML elements. One should be able to reuse these elements by adding it in the markup and providing state on the fly. Largely, AngularJS 1.x applications achieved it with Directive, which allowed creating custom HTML elements with AngularJS. They allowed manipulating DOM.
Angular 2 application created a category of directives called Components, which allowed creating custom HTML elements and reusing them.
Angular 2 onwards, there is no explicit controller. The framework is not seen as MV* framework anymore.
An Angular application is now seen as composition of components. See Figure 2 where each component has its own view. It may use HTML elements to render data. It may also have multiple child components providing their own view and functionality.
Figure 2: Composition of components
Unidirectional data flow
With AngularJS 1.x, two-way data binding is a powerful feature. The data on $scope object in the controller is bound to the template/view. If a change happens to this object, it’s updated in the view. If the view changes values, it’s updated on the $scope object, as well as all other view references.
There could be child elements, views or components (AngularJS 1.5 onwards components were introduced) that are tied to the object. When a change occurs in the child component, it’s reflected in the parent and everywhere else that’s maintaining reference to this object.
This incurred high performance cost.
Angular 2 encourages using unidirectional data flow. State propagation (when data changes) always flows from parent component to the child, not the other way around.
If child component has to change parent’s state, it will send an event to the parent. The parent component propagates state change and updates itself as well as all child elements dependent on the state.
Note: There are patterns like Redux that ease maintaining application state and ensure unidirectional data flow between parent and child components. Moving to the component model of developing UI applications makes it easy to use such a pattern. It’s difficult to achieve the same results with MVC and AngularJS 1.x.
Change Detection
Continuing on the data binding topic we discussed in a previous section, it has been one of the important features of AngularJS 1.x.
The Model object is bound to the view in the HTML template. With two-way data binding, changes to the model are reflected in the view and vice-versa. It means model (object structure) is in sync with the DOM (Document Object Model), which is nothing, but the view.
This feature is very beneficial but is also challenging. For the view to be in sync with the model objects, change detection needs to happen.
AngularJS 1.x framework addressed it with dirty checks. The process is asynchronous.
However, the framework looped through the entire DOM structure to identify changes and synchronize view and the model. The process was costly and relatively slow.
Certain other frameworks like ReactJS addressed its Virtual DOM that differed changes and updated only the changed sections in the view.
This change detection process saw a major improvement with Angular 2 and above. Following are some of the improvements in Angular 2 and above.
Improvements in Angular 2
ZoneJS:
Angular included ZoneJS, which provides execution context for a piece of JavaScript code.
Angular uses it for change detection. Angular extended it (forked the library) and called it ngZone.
With the help of this API, Angular identifies asynchronous operations that might have caused a change. The asynchronous operations could include DOM events like button click, text field value change or an XHR/fetch API call to retrieve data from the server. This API has hooks to trigger synchronizing view.
Change Detection Strategy:
Angular 2 and above has two change detection strategies:
Default
Each component in Angular has its own change detector. It is created during the compilation process and hence is specific to each component. Making it specific helps improve performance.
DOM is organized as a tree while it’s rendered on the view. Components primarily render the elements and nodes creating a component tree in an Angular application. As each component has its own change detector, we have a tree of change detectors.
NgZones API, specifically ApplicationRef’s onTurnDone event triggers change detector process. As mentioned earlier, NgZones provides execution context and hence it can identify events that can cause change, be it a button click or successful XHR call response.
The change detectors run through component tree and provide the resultant DOM to update the view. This process is fast primarily due to component specific change detectors. See Figure 3.
onPush
With Angular 2 and above, performance can also be improved with onPush change detection strategy.
Immutable objects: JavaScript objects (non-primitive types) are call by reference. When we compare two objects, in reality we are comparing references, not the data or values.
With mutable objects, data might get updated with a change, but the reference will remain the same. Here a simple comparison will not be enough to detect a change. Hence the Angular framework will have to perform a deep comparison, which is a costly operation. The default change detection strategy incurs this additional cost in performance.
As we use immutable objects in a component, we can instruct the component to check for value change only when object reference changes. It will perform better as it reduces work.
Consider Figure 3. Tab Component with filter has two child components. The child components User Profile Card and History Visualization Card uses filter input from the parent component.
When user changes filter condition, the onPush strategy on the child components and immutable objects are passed-in through input attributes. Angular compares the object references and change detector for just the components invoked. It makes the change detection even more efficient. See Figure 4.
RxJS
Traditionally, Promise API has been very effective with JavaScript applications. They are asynchronous API for long running operations. The long running operation could be an XHR call or iterating through a large result-set in the browser.
Promises help invoke success or failure callback once the long running operation is ready with the results.
However, it’s a one-time operation.
The promise is closed once result or error is obtained. There is no possibility to stream multiple results. Chaining multiple promises is clunky.
An observable is a next step to promises. RxJS is an implementation of Observables and much more. Observable can stream multiple results until it’s closed. It is easy to create an observable from multiple regular operations like looping through an array or converting to an observable from a promise associated with a HTTP call etc.
An observable could also be UI interactions. Actions emitted from a text field as the user is typing-in the text, is a stream, and could be handled effectively with an observable.
Operators - RxJS also comes with ready-made solutions for recurring scenarios working with observables.
Following are some examples.
- Retry - We could attempt to retry a failed HTTP call
- Debounce - Certain observables like UI actions typing in a text field could emit results too quickly, with-in tens of milliseconds. If we are using the data real-time to make a HTTP call, too many calls are triggered. An operator like debounce could be used to wait for certain milliseconds before triggering the next call.
Follow the link for list of RxJS operators by example.
TypeScript
TypeScript is an extension to JavaScript, in other words, it’s a superset to JavaScript. It is an open source project backed by Microsoft. Angular takes advantage of additional features and benefits with the language.
- Data Types - Supports defining data type while creating a variable
- Decorator - Supports decorating additional behavior to a class or a function.
- Generics – Supports building a function or a class that defines data type dynamically. Makes it future proof, as a function parameter’s data type or return type need not be defined while creating the function. Similarly, class field type or parameters for a method on the class need not be defined while creating the class. The function creating an object of generic class or calling generic function can specify the type on the fly.
- Classes - Similar to ES6, classes can be created in TypeScript code. However, it includes future JavaScript features like controlling visibility of a class field or a method with public, private or protected. A “similar” proposal has been made for JavaScript for creating private fields. It is a Stage 3 proposal in ECMA Script at the time of this writing. Click here to view the proposal.
- Interfaces - An interface allows defining structure or shape of an object. It helps define contacts for the API.
- Enum – TypeScript provides an easy way to create named constants with enums.
- IDE friendly – TypeScript with types, compiler and other related features makes it easy for IDEs like Visual Studio Code to provide developer friendly help on APIs and function signatures, navigating to the definitions and references. It will also be possible for IDEs to identify and show errors & warnings better and faster as we type the code.
While Angular was being built, the Angular team announced developing a language extension on top of TypeScript, called AtScript. However, TypeScript team driven by Microsoft chose to implement the additional AtScript features in TypeScript version 1.5. With it, Angular chose to make TypeScript a primary language for Angular. However, there are variations to use Angular with basic JavaScript or Dart.
Changes with each version of Angular
The sections we discussed so far have detailed AngularJS 1.x pros and cons and the purpose of Angular 2.
Upgradation to Angular 2 was a major shift in approach and application architecture. The upgradation process was relatively challenging.
Beyond Angular 2 (v4, v5 and now v6), it has been an incremental update to the framework. Majority of the changes are non-breaking. They are enhancements, new features and bug fixes.
The rest of this article will take a different approach going forward. The latest release will be discussed first, going down to connect with Angular 2 at the end.
Angular 6 – What’s New?
Angular 6 released early May 2018 has considerable improvements to Angular CLI and the Angular eco-system. Following are some of the major improvements.
Support For Angular Elements
Angular Elements is a step in the direction to support Web Components. With the update, an Angular component can be registered as a custom component. A custom component can be used outside Angular context. We may include the script and use the component in a React, Angular or even a simple HTML page.
It enables Angular to be used in a smaller context. It doesn’t have to be all or nothing for Angular applications. Sections of the applications, especially reusable components, could be built in Angular and reused elsewhere.
Angular CLI updates
Angular CLI version has been made consistent with rest of the libraries in Angular Monorepo. The version number has a jump to 6.x from 1.7.x. Some of the important features are as follows.
ng add
Enables adding features to an existing project. For example, a pre-existing Angular application can add material design features with ng add @angular/material command.
ng update
With each version update, this feature enables the project to be in sync. It helps project dependencies synchronize to the latest versions. Third party libraries too can provide this feature for applications using the libraries. It can be done with schematic, which needs to be provided by a third-party library.
New starter components with ng generate
Traditionally ng generate helps add Angular artifacts like components or services to the project. With the update, additional starter components can be added to the project.
Specifically, with Angular Material:
- Use ng add @angular/material to register new starter components
- To add a side nav starter components to the project use ng generate @angular/material:material-nav --name:a-custom-name
Support for workspace and library
A CLI workspace supports more than one project at one place. These projects can be applications or libraries. A library is a special type of project that could readily be published as a package in NPM or Yarn. Applications consume libraries.
Traditionally we install libraries from an NPM registry (into node_modules) and import in the code file using the library. If the library is part of the application we are developing, it helps identify it as a separate library. The Module system will look for the built library in tsconfig, if not found, it will look for it in node_modules
We can build the library using ng build library-name command. Use --prod flag if the intent is to publish the build to a registry.
Ivy Renderer
The Angular team is working on a new renderer that could under the hood decrease the bundle size, render the application faster and boost change detection so much so that it is faster compared to the current view engine. It is expected to be a non-breaking change with no changes required for the current applications once upgraded to a version that supports the new renderer.
The render works in a pipeline. It is designed by keeping tree-shaking in mind, which helps reduce size of the bundles. Ivy also helps debug template code by adding debug points and pointing to the line of HTML code that caused an error. This is a powerful feature.
Ivy renderer is officially not included in Angular 6. However, Ivy runtime is in active development followed by compiler support. New features are being included in beta releases.
Provide with Injectable
Angular 6 now allows providing a service with injectable decorator. A service could be provided at module level or at component level. Angular creates an instance of the service based on it.
A service provided in component tree has a smaller life cycle. As new instances of component are created, the Service is instantiated multiple times. However, providing a Service at module level reuses the same service instances in many components.
For better tree-shaking capabilities, Angular 6 allows for providing a service at module level with injectable decorator. Tree shaking is a bundle optimization process that helps reduce the bundle size. Smaller bundles help browser load, bootstrap and run an application faster. Earlier, service could be provided with providers parameter on @ngModule decorator for the module. This approach will continue to work even now.
The decorator injectable is used on the Service class. A parameter providedIn can use the string root or a module name that should be providing the service. As the name suggests, root provides the service with the root module. Providing with a module name works better in lazy load scenarios.
import { Injectable } from '@angular/core';
@Injectable({
providedIn: 'root',
})
export class CustomerSampleService {
constructor() { }
}
Angular 5
Angular 5 was released in Nov 2017. While the version before it (Angular 4) provided features and new syntax for developers, this version focused on performance and reducing bundle size for better bootstrap time. In fact, the first few releases (v4, v5, v6) after a major version change (v2) has considerable improvements to performance and Angular bundle size.
Efficient bundles
Angular 5 made improvements to generating efficient production bundles. The size of the bundle was reduced. Smaller bundle size meant faster load time for an Angular application.
For the projects generated using Angular CLI, a Build Optimizer was included. It achieved efficient bundle size because of the following reasons:
- The tree-shaking was improved with the Build Optimizer in Angular 5. Tree-shaking is a process by which unused and dead code can be removed from a bundle that is being generated. The build optimizer marks parts of the application pure, hence implementing effective tree shaking.
- Angular 5 was updated to use RxJS 5.5. It introduced pipeable operators (earlier called lettable operators), which allowed importing one of many, non-default operators with traditional ES6 syntax. This helped tree shaking capabilities, when used with Angular CLI.
Note: Projects setup to use Web Pack directly need to update build process to include ModuleConcatenationPlugin, which is available for WebPack version 3 and above.
Following need not be done any more:
Import ‘rxjs/add/operator/scan’
Import ‘rxjs/add/operator/filter’
We may use the following syntax,
Import {scan, filter} from ‘rxjs/operators’
Note: Also, it’s plural operators (rxjs/operators) as opposed to singular operator (rxjs/add/operator/xyz) now.
- It is a common practice in Angular to use many decorators. They add dynamic capabilities to the code units. The Compiler would use this code. With Ahead of Time compilation (AoT), one wouldn’t need the decorator to be still part of the bundle. Hence, they were cleaned-up from a generated bundle, thereby reducing size of the bundle.
- Multiple polyfils are not required anymore with Angular 5. Following are a couple of examples:
- There were new pipes added to handle internationalization of numbers, dates and currency values. Before this change, Angular was using browser’s i18n API. It would mean there are subtle changes on how browsers are providing values to the application and it needed polyfills as well.
- Angular 5 updated the pipes with its own implementation instead of depending on browser. It would also mean, we wouldn’t need polyfills to be included.
- Angular had been using Reflective Injector for instantiating and injecting classes. Reflective injector is dynamic, needed additional map files and polyfills. It was replaced by Static Injector, which is performant, primarily because there is no dynamic decision making. It instantiates and injects Angular artifact configured by developer. It also allowed the need of polyfills and map files.
Enhancement to compilation at development time
- ng serve used for live compilation and reloads at the development time had an addition to support Ahead of Time compilation. Use ng serve --aot. This will eliminate variation when publishing a build, which would have been done with AOT.
- Angular compiler is hooked into TypeScript transform improving speed of live compilation and reload at the development time. It will make it easy for developers to see the changes reflected in the app being debugged.
Deprecated @angular/http module
HttpClientModule and HttpClient in @angular/common/http introduced little earlier will become mainstream client API to access services over HTTP. It would mean @angular/http is deprecated. Applications will have to upgrade and remove references to @angular/http.
HttpClient comes with more features,
- Additional progress events while making Http calls.
- Request and response objects are immutable
- JSON response doesn’t required to be parsed. JSON objects are readily available.
- Use interceptors to add middleware logic in the Http calls’ pipeline
Improved Decorator Support
Angular 5 enhancement to decorators allow lambdas and dynamic values to be used in the decorators. Lambdas allow using unnamed functions instead of traditional function syntax for callbacks and values provided in a decorator.
Angular 4
Angular 4.0.0 was released in March 2017. It was the first major version after Angular 2 and considering that, many developer focused features and enhancements were included. However, most of these changes were non-breaking.
A major version update between 1.x to 2.x was an architecture shift in building rich UI applications with Angular. Many applications were rewritten in Angular 2. Organizations, developers and projects spent a lot of time and resources in upgrading their apps from Angular 1.x to 2.x.
However Angular 4 was a totally different scenario. It had incremental changes, and improvements to the framework and applications that were written using the framework.
No major version 3.x
MonoRepo: Angular 2 has been a single repository, with individual packages downloadable through npm with the @angular/package-name convention. For example @angular/core, @angular/http, @angular/router so on.
Considering this approach, it was important to have a consistent version numbering among various packages. Hence, the Angular team skipped a major version 3. It was to keep up the framework with Angular Router’s version. Doing so would help avoid confusions with certain parts of the framework on version 4, while the others on version 3.
Following are some more improvements in Angular 4.x
Move animations out of @angular/core
One of the problems Angular had been facing is with size of the bundles. For projects, the framework and all vendor dependencies could grow pretty big. In an effort to reduce the bundle footprint, animations were moved out. Any project migrating from Angular 2 to 4 most likely would install and import animations package explicitly.
import BrowserAnimationsModule from @angular/platform-browser/animations.
Note: reference the module in imports array of @NgModule
Improved View Engine
Under the hood, AOT compilation process improved with version 4. It reduced size of the compiled component by almost half. A smaller bundle size was achieved with this change.
Angular Universal
Server-side rendering capabilities were made mainstream with version 4. Angular Universal allowed server-side rendering for better search engine crawling/indexing capabilities.
Improved conditional statements in the template
ng-if supported else conditions with Angular v4 onwards. The ng-if and else statements supported conditional rendering in the view. It was an enhancement to this feature.
Read the following DotNetCurry article for comprehensive list of Angular 4 features- www.dotnetcurry.com/angular/1385/angular-4-cheat-sheet
Conclusion
Angular brought a major shift with the version 2. However, there are many applications today running on AngularJS 1.x probably due to the cost and effort associated with migrating to Angular 2.
However, considering the improvements, adoption of latest technologies like TypeScript, RxJS etc., I feel it is worth the effort. The migration not only helps with clearing the tech debt, but also helps make the application faster, leaner and better.
References and further reading
Angular Change Detection Explained by Thoughtram
Angular Documentation on RxJS Library
Angular 5 Release Notes and the Blog
Angular 4 Release Notes and Blog
Stackoverflow discussion on differences between Http and HttpClient
Angular 4 development cheat sheet, a DotNetCurry article
Evolution Image reference https://en.wikipedia.org/wiki/File:Human_evolution_scheme.svg
This article was technically reviewed by Ravi Kiran.
This article has been editorially reviewed by Suprotim Agarwal.
C# and .NET have been around for a very long time, but their constant growth means there’s always more to learn.
We at DotNetCurry are very excited to announce The Absolutely Awesome Book on C# and .NET. This is a 500 pages concise technical eBook available in PDF, ePub (iPad), and Mobi (Kindle).
Organized around concepts, this Book aims to provide a concise, yet solid foundation in C# and .NET, covering C# 6.0, C# 7.0 and .NET Core, with chapters on the latest .NET Core 3.0, .NET Standard and C# 8.0 (final release) too. Use these concepts to deepen your existing knowledge of C# and .NET, to have a solid grasp of the latest in C# and .NET OR to crack your next .NET Interview.
Click here to Explore the Table of Contents or Download Sample Chapters!
Was this article worth reading? Share it with fellow developers too. Thanks!
V Keerti Kotaru is an author and a blogger. He has authored two books on Angular and Material Design. He was a Microsoft MVP (2016-2019) and a frequent contributor to the developer community. Subscribe to V Keerti Kotaru's thoughts at his
Twitter account. Checkout his past blogs, books and contributions at
kvkirthy.github.io/showcase.