Building front-end apps that Scale - FOSDEM 2014

Post on 10-May-2015

1597 Views

Category:

Technology

1 Downloads

Preview:

Click to see full reader

DESCRIPTION

Developing large apps is difficult. Ensuring that the code is consistent, well structured, tested, and that the architecture encourages maintainability is essential. When it comes to building large server-focused apps the solutions to this problem have been tried and tested. But, how do we achieve this when it comes to HTML5 single page apps? BladeRunnerJS is an open source developer toolkit and lightweight front-end framework that has helped the company I work for (Caplin Systems) ensure that a 200k LoC JavaScript codebase hasn’t become a tangled mess of unstable spaghetti code (with bacon bits). This codebase is then delivered to customers, along with around 50k LoC example functionality for them to build upon, and they're expected not to turn that into a tangled ... you get the idea. In this talk you'll learn about the main concepts we have applied, how we have applied them - and how you can too - to achieve what might sound like the impossible.

Transcript

Building front-end apps that

Scale

Phil @leggetter phil@leggetter.co.uk

Caplin Systems !

!

!

@BladeRunnerJS

Overview

• What is a Large Scale front-end App?

• What are the signs of scaling?

• What are the solutions? (with demos)

What is a large-scale JavaScript app?

–Addy Osmani, Patterns For Large-Scale JavaScript Application Architecture

In my view, large-scale JavaScript apps are non-trivial applications requiring significant

developer effort to maintain, where most heavy lifting of data manipulation and display falls to

the browser.

Great. More detail please!

Large CodebaseMore functionality = More code

Caplin Trader• SDK:

• ~1,000 JavaScript files

• ~131,000 LoC

• ~131 lines per file

• ~650 test files

• ~95,000 test LoC

• Getting Started App:

• ~425 JavaScript files

• ~50,000 LoC

• ~117 lines per file

• ~200 test files

• ~21,000 test LoC

Complexity

Gmail & Caplin Trader• Large Single Page Apps (SPAs)

• Complex functionality

• Complex interactions

• User

• Technology

• Leave open all day

ContributorsThe Human Factor

Who contributes to an app?• Front-end devs

• Back-end devs

• Designers

• QA

• Infrastructure and release engineers

• Technical authors

People who are part of...

• Large teams

• Multiple teams

• Teams spread across an organisation

• Teams spread across multiple organisations

So, how do you ensure an application is maintainable?

1. structure a massive codebase (js, css, html, i18n, images, config etc.)

2. an architecture for complex functionality and interaction within an application codebase and UI

3. make sure that all contributors can work in harmony

4. development must be a productive experience

5. ensure all these compliment each other

Seven signs of scaling?

Dev Setup takes forever• Lots of infrastructure dependencies

• Database

• Auth service

• Realtime server

• One or more Web APIs

• Dev app server

My App won’t load!

• Run full application

• All back-end components running

• Lots of moving parts

My App isn’t working!

• Other contributors breaking functionality

• Code accessed and modified from elsewhere

• Dependency Analysis and bundling out of order

Finding assets is hard

What does this code do?

• Inconsistent coding style and structure

• Side effects

Long App Reloads

Have the tests finished yet?

• Having to run all the tests

• UI-based tests

• Continuous Integration…taking 8 hours!

Scaling Solutions

• Streamlined developer workflow

• Consistency

• Focus on building a single feature (in isolation)

• Scalable loosely-coupled application architecture

• Quality at its core (maintainability)

Goals

Developer Workflow

Building a single feature in isolation

Blades

Blade Demo

Application Architecture

Requires an Architecture that…

• Allows complex interactions

• Allows components to be changed without side effects

• At launch and during runtime

• Is easily extended

• Is easily tested

• Is easily maintained

Blades (again)

MVVM

Services

What is a service?• Use services to access resources

• Persistence Service

• Restful Service

• Realtime Service

• Perform common non-UI tasks e.g.

• LoggingService

• Services registered and accessed via a ServiceRegistry

• Dynamic Service Locator1

1 http://martinfowler.com/articles/injection.html#ADynamicServiceLocator

Why use services?• Used for cross-cutting concern

• Functionality encapsulated behind an interface

• Loose-coupled communication

• Dependencies can be injected at:

• App/Workbench load time

• During runtime

Services Demo

Quality

Manually test in the workbench as part of iterative development cycle

Simple Wins• Consistency

• Architecture

• Coding style and structure

• Loose coupling (MVVM, Services and beyond)

• Facilitates testing

• Can easily swap out implementations

• Avoid testing the DOM

Biggest Win• Testing features in isolation

• Change view model and assert against mocked Service

• Inject mock service, make calls and assert View Model

• Run small suit of Application End-to-End Business Tests

Need Proof?

Our full suite Caplin Trader testing time went from

8 Hours

to

10 minutes

A note on deployment

• The BladeRunnerJS app replicates previous sole target production environment: J2EE

• You can enable dev mode (discussions ongoing)

• You can build to a .WAR frill

• Flat File deployment coming soon (very soon, I hope)

Summary• Streamline developer workflow for productivity

• Build features in isolation (grouping assets together)

• Services e.g. EventHub for loose coupled communication

• Encapsulate using Interfaces

• Test a feature ViewModel < - > Service. Avoid the DOM.

• It all improves Quality and thus maintenance

Phil @leggetter phil@leggetter.co.uk

Caplin Systems !

!

!

@BladeRunnerJS bladerunnerjs.org

top related