API Frameworks

Front or back?

Applications often have different layers, including a front end and a back end

• Front end:

  • Visual layer (the layer that users get to see)
  • Made for interaction between user and application
  • Languages: HTML, CSS, JavaScript, …

• Back end:

  • Server side (no direct interaction with the user)
  • Performs (business) logic ‘behind the scenes’
  • Languages: euhm… where should we start?

Framework

In the following slides, several back-end languages and frameworks will be looked at. But first: what is a framework?

• Foundation of existing code
• Used by calling its methods, implementing its interfaces, using its classes, etc.
• Provides default behaviour
• Its 'core code' can usually not be modified
• According to Wikipedia: "a software framework is an abstraction in which software, providing generic functionality, can be selectively changed by additional user-written code, thus providing application-specific software"

Back-end languages

Python

Developed in 1991 by Guido van Rossum

• Simple syntax
• Dynamically typed
• Rich ecosystem with many out-of-the-box libraries (e.g. Numpy, Matplotlib) and frameworks (e.g. Django, Flask)
• Free and open-source

Some webservice frameworks of Python are mentioned below.

Django REST

• Python version 3.5+
• Multiple in-built authentication policies
• Supports automatic URL routing to Django
• Supports testing, caching, throttling, etc.
• Extensive and good documentation

Flask Restful

• Very lightweight module
• Used to access multiple HTTP methods
• Extensive data formatting functionality
• Allows you to add custom fields and inputs based on your data type
• Also supports XML, CSV and HTML amongst others, besides JSON

Falcon

• Python version 2.7 and 3.5+
• Used to build reliable and high-performing large-scale application backends and microservices
• Lightweight with minimal setup
• Majorly used for faster performance
• Integrates with NoSQL easily
• Supports integrated unit tests

FastAPI

• Python 3.6+
• Very high performance (one of the fastest Python frameworks available)
• Designed to be easy to use and learn
• Based on (and fully compatible with) the open standards for APIs: OpenAPI (previously known as Swagger) and JSON Schema
• Integrated security and authentication
• Supports SQL databases, NoSQL databases and GraphQL

Java

Developed in 1995 by Sun Microsystems (acquired by Oracle in 2010)

• Cross-platform: “Write Once, Run Anywhere”
• Multi-threaded processing
• Incredibly secure (compared to other languages)
• Commercial use of Java requires the purchase of a license from Oracle

Java: JAX-RS

All frameworks mentioned in the following slides implement (= make use of) the JAX-RS specification, except for Spring.

JAX-RS:

• Jakarta RESTful Web Services
• Official part since Java Enterprise Edition 6 (Java EE 6)
• Specification developed by the Java Community Process (JCP)
• Standard for developing Java REST applications, it specifies how to create REST web services when using Java

Some webservice frameworks of Java are mentioned below.

Jersey

• Available standalone or as part of the Glassfish application server (and therefore commonly used)
• Provides its own API that extend the JAX-RS toolkit with additional features and utilities
• Very good support for tooling (IDE)

Restlet

• Lightweight core
• Can run as a standalone Java application
• Challenges or manual work involved in de-marshaling the response into a Java object

Spring (MVC framework)

• Does not implement the JAX-RS specification

• Uses the Jackson library to support JSON

• REST service class: @RestController annotation
  

Javascript

Developed in 1995 by Brendan Eich of Netscape under the name Mocha

• Developed as front-end language
• Used as basis for Node.js

Node.js

• Developed in 2009 by Ryan Dahl
• First server-side JavaScript environment
• Lightweight technology, ideal for microservice-based architecture
• Concurrency: single-threaded so that the event loop can execute callback functions

Some webservice frameworks of Node.js are mentioned below.

Express (MVC framework)

• #1 Node.js framework, based on Github figures
• Robust and well-tested, backed by a large community
• Focus on performance
• Easy to set up
• Used as basis for many other frameworks:
  • Feathers
  • ItemsAPI
  • KeystoneJS
  • Blueprint
  • …

Sails (MVC framework)

• Emphasis on stability and ease of use
• Out-of-the-box support for authentication
• Own ORM (Object-Relational Mapping) for database interfacing: Waterline
• Uses Blueprints as API generator, which requires minimal manual coding
• Powerful command-line interface
• Used by many large companies => robust enough to support enterprise-level applications

Smooth sailing ensured!

Loopback

• “highly extensible, open-source Node framework based on Express that allows you to quickly create APIs and microservices”
• Add-on support for easy authentication and authorization
• Minimal coding required to create a dynamic API
  

PHP

• Developed in 1995 by Rasmus Lendorf

• Mainly used for web applications

• Strong debugging capabilities

• Free and open-source

• Fast execution times

Some webservice frameworks of PHP are mentioned below.

Laravel (MVC framework)

• Most commonly used PHP framework
• Built-in possibilities for creating REST API’s using the Model-View-Controller architecture
  • Advantage: helps programmers focus on one aspect of development at a time by dividing the API creation process into three parts (Model, View, and Controller)
  • Disadvantage: generally takes more time to develop
• Lots of documentation on the Laravel website and on fora
• Small learning curve for developers who already know Laravel MVC

Lumen

• Based on the Laravel framework
  • Cuts off all Laravel features that are unnecessary for REST APIs
  • Lightweight
• Less configuration necessary (compared to Laravel)
• “Lightning fast micro-services and APIs delivered with the elegance you expect”
  

Guzzle

• Can send both synchronous and asynchronous requests using the same interface
• Middleware system allows you to augment and compose client behavior
• Abstracts away the underlying HTTP transport, allowing you to write environment and transport agnostic code; i.e., no hard dependency on cURL, PHP streams, sockets, or non-blocking event loops

Ruby

• Developed in 1993 by Yukihirio Matsumoto

• Standard structure for common development patterns

• High-quality standards: developer is forced to follow established development patterns

• Very stable codebase, not much maintenance required

• Dynamically typed

• Free and open-source

• Fast execution times

Some webservice frameworks of Ruby are mentioned below.

Roda

• Small framework
• Simple, therefore rather easy to learn
• Plugins available for more complexity
• Routes described like a tree (each ‘branch’ of the tree matches a part of the route)
  

Sinatra

• Slower and more memory-intensive than Roda, but easier to work with
• More of a router than a full-blown framework → plugins necessary for more complex logic
  

Ruby on Rails

• Most popular Ruby framework
• Very complex and complete framework
• Often considered “too bloated” for API development → not suitable for small APIs without much complexity
• In general it greatly influenced web app development, many other frameworks ‘borrowed’ its ideas (seamless database table creations, migrations, and scaffolding of views amongst others):
  • Django (Python)
  • Laravel (PHP)
  • Node.js
  • …

Go (Golang)

• Developed in 2007 (first released in 2011) by Robert Griesemer, Rob Pike and Ken Thompson for Google

• High execution speed

• No interpreter: directly translated into formats that a processor understands

• Built-in testing tool, supports parallel testing

• Scalability: use of GoRoutines (which are functions or methods that run concurrently with other functions or methods) leads to a stable system, particularly when scaling the application upwards

Some webservice frameworks of Go are mentioned below.

Revel

• Comes out of the box with much of its feature-set pre-configured and installed
• Self contained: doesn’t require very many third-party libraries or middleware implementations → relatively reduced complexity for high functionality
• Very appealing for many startups
• Large code base, therefore less lean than other Go frameworks

Martini

• Extremely lean
• Boasts impressive and easy to integrate third party support
• Meant to do a lot of work with a minimal amount of overhead
• Good scalability
• Plenty of documentation and a large install base, but… Martini has not (in its core implementation) been maintained since 2014

Gin

• Very minimalistic, trimmed down framework
• Designed with the disadvantages of Martini in mind → roughly 40 times faster than Martini
• Direct and effective documentation, but support is not as robust
• Easy debugging and issue tracking
• The framework tends to offload a lot of work to the client due to server limitations, limiting potential implementations to what you have