Learn Phoenix (Web App) Framework

Learn how to use Phoenix Framework to have fun building real-time web/mobile apps
that are fast for "end-users", reliable, scalable, maintainable and (easily) extensible!

Why?

As web/mobile app developers we need to leverage the work that other (really smart) people have done
instead of constantly building things "from scratch" all the time; that is why we use frameworks to build our apps!

See: "Top 10 Reasons Why Phoenix" (further down this page!)

There are many frameworks to choose from (a few "popular" ones are mentioned below in the "Questions" section).
But if we go by what is "popular" we would still be riding horses (and carts) everywhere and no progress would be made.

Note: all the reasons "Why" for Elixir also apply to Phoenix!
check them out: https://github.com/dwyl/learn-elixir#why

What?

A web application framework without compromise!

Developer Joy

The biggest "constraint" in any technology project is people. An "Entrepreneur"/"Founder" or "Product Owner" can have all the good ideas in the world, if they are unable to convert idea into reality, it's meaningless.

(Should I Care About) Benchmarks?

Obviously, you should run your own benchmarks on your own hardware/cloud and make informed decisions based on the requirements of your app/product, but ... when we read the stats for how many concurrent users a Phoenix App can handle (with live WebSocket connections) we were blown away! It means we can build our real-time app(s) with 90% fewer resources.

Performance Highlights

• considerably lower latency and request response time than anything else! (thanks to Erlang's lightweight processes and ultra-efficient network/message handling)
• 4x more requests-per-second than the equivalent Node.js (Express.js) App.
• 9x more throughput than a Python 2.7 based app. (blocking really sucks!)
• 10x - 100x more requests handled than Ruby-on-Rails (depending on the type of app!)
• Similar performance to Go on a single server, but a much simpler multi-server concurrency model, so horizontal scaling across multiple data centers ("availability zones") is much easier! (Erlang manages the resources for multiple servers/processors as a single app "fleet" and delegates requests/processing to them across clusters/zones!)

All of this means that you spend considerably less money on Hardware/Cloud infrastructure so your app/company can gain a competitive advantage on cost.

If you are in the fortunate position to be considering using something way better for your next project, look no further than Phoenix!
Read more: http://www.phoenixframework.org

Who?

Many people/teams/companies are already using Erlang/Elixir and Phoenix and seeing phenomenal results!
Including: Adobe, BBC, Spotify, Pinterest, Discord (Gamer Chat App), Groupon (Fave), Lonely Planet, Brightcove, Slack ...
See: https://github.com/doomspork/elixir-companies

Who Should Learn Phoenix?

• People who want an easy way of building an app with all the modern conveniences of "generators", built-in Database interface ("ORM") and Firstclass WebSockets for "real-time" while getting legendary reliability.
• Teams who need the developer productivity or Ruby-on-Rails or Django, but want less "magic" (i.e. want to understand how everything works!).
• Companies who want to spend 90% less on both developer and infrastructure costs so you can serve more people with fewer resources.

Who Should NOT Learn Phoenix?

• Anyone who needs to use the "most popular" framework or language (for whatever reason).
• Companies who already have lots of legacy Java/PHP/etc. and cannot invest the time to learn/use something new.
• Developers who want "job security" bug-fixing unreliable code.

How?

Assumptions / Pre-Requisites?

Elixir

You cannot build a Phoenix App without knowing Elixir.
If you are new to Elixir, "star" (bookmark) this repo (so you can return to it tomorrow)
and then go to: github.com/dwyl/learn-elixir learn elixir until you feel like you understand the syntax, then come back and learn Phoenix!

Specifically you should focus on learning the Elixir "Basics":

• types of data
• atoms
• pattern matching
• maps
• function definitions
• modules

Node.js

Phoenix uses Node.js to compile assets like JavaScript and CSS files (using Webpack).
Simply ensure you have Node.js installed. https://nodejs.org
You don't need to know Node to use Phoenix.

Installation

If you've already learned some Elixir and installed Node.js, then the first step to getting started with Phoenix is installation!

The Phoenix documentation is amazing, so we recommend following the official phoenix installation instructions!

You'll also need to install PostgreSQL, there is a tutorial of how to do so linked in the Phoenix installation guide linked above, but you can also check out our learn-postgresql repo for instructions, and raise an issue if you have any trouble!

While a basic understanding of JavaScript can be useful at times, you don't need to know how to use Node to use Phoenix.

If you're curious why they chose Brunch.io over "alternatives",
the short answer is: Simplicity & Speed! see: http://brunch.io/docs/why-brunch
Note: Phoenix v1.4 (unreleased at the time of writing) uses WebPack for asset compilation, see: CHANGELOG.md

Next?

Familiarize yourself with the "Up and Running" (official) guide: https://hexdocs.pm/phoenix/up_and_running.html#content

Practical Examples?

Once you have phoenix installed and followed the official "up and running" guide,
come back and try these beginner-friendly examples:

• Counter (realtime shared counter using Phoenix LiveView): https://github.com/dwyl/phoenix-liveview-counter-tutorial
• Todo List (step-by-step tutorial showing how to build a fully functional TodoMVC clone): https://github.com/dwyl/phoenix-todo-list-tutorial
• Chat (realtime UI/UX using WebSockets): https://github.com/dwyl/phoenix-chat-example
• Encryption (security): https://github.com/dwyl/phoenix-ecto-encryption-example
• Append-only Log (how to store your App's data): https://github.com/dwyl/phoenix-ecto-append-only-log-example

Book?

We recommend that people buy (or borrow) @chrismccord's book: "Programming Phoenix"
see: https://pragprog.com/book/phoenix14/programming-phoenix-1-4

The authors are individually impressive and collectively they comprehenisively cover Phoenix like nobody else can! Chris created Phoenix, José created Elixir and Bruce is ultra-experienced technical author with many successful books to his name!
(i.e: the book is the obvious choice for how to learn Phoenix!)

Video Intro by José Valim (Creator of Elixir)

https://youtu.be/MD3P7Qan3pw

https://youtu.be/srtMWzyqdp8

Resources

• Elixir vs Ruby Showdown - Phoenix vs Rails: https://littlelines.com/blog/2014/07/08/elixir-vs-ruby-showdown-phoenix-vs-rails
• Benchmark: https://github.com/mroth/phoenix-showdown

Our Top 10 Reasons Why Phoenix

"Phoenix provides the productivity of Ruby-on-Rails
with the concurrency and fault-tolerance of Erlang."

0. Beyond all the (fantastic) technical benefits, what attracts us to Phoenix is the Great Community of people around the world who are excited about making Phoenix an amazing tool for building web apps!
   Having welcomming people who will

   • help you when you get stuck, patiently explaining things
   • answer questions (both "noob" and "advanced") and
   • openly discuss (your) ideas for improvements. see: https://elixirforum.com/t/phoenix-v1-3-0-rc-0-released/3947
     
     

1. Phoenix uses the Elixir programming language which means your app is compiled and run on the Erlang Virtual Machine "BEAM".
   Erlang is a battle-tested highly fault-tolerant VM used by many telecommunications companies

2. WebSockets ("channels") are built-in to the framework which means building apps with "real-time" communication and interaction is much easier than virtually any other framework/platform! (no third-party magic module needed! everything you need is already there ready for you to serve millions of people!!)
   see: http://www.phoenixframework.org/docs/channels

3. Easy asyncrhonisity because all programming in Phoenix (Elixir) is Functional! This means it's really simple to abstract useful functionality like request authentication, logging and processing into "piplines" that are easily human-readable! (no third-party async module required! no "promises", "generators" or "observables" to managed!!)

4. Security & Resilience Mindset is the default. Encryption (SSL) is easy in Phoenix/Elixir and both mitigation of SQL injection, Cross-site Scripting (XSS) and CSRF protection are built-in (enabled by default) so it's virtually impossible for a "novice" programmer to introduce this type of security bug.

5. Concise Code cannot be understated! We can write way fewer lines than in the equivalent Node.js/Java/Rails/Go app, this means developers are more productive and there is less code to maintain!

6. Testability due to functional programming of all controllers!

7. Easy Deployment: https://hexdocs.pm/phoenix/heroku.html

8. Zero-downtime Deployment is free! (again because of Erlang). Erlang manages transitioning "live/active" users from the old to new version of your app without them even noticing that it was upgraded/updated!!

9. Built-in Monitoring/Managment of your app through Erlang supervisors mean that you know exactly how your app is performing, what parts have crashed/restarted and why! This is a feature we pay (a lot) for in other frameworks and here it's free!!

Can you think of another reason why using Phoenix is awesome?!
Please Share your thoughts in this thread: https://github.com/dwyl/learn-phoenix-framework/issues/13

Questions?

Do I need to learn Elixir before trying/using Phoenix?

Yes. See: https://github.com/dwyl/learn-elixir

Do I Need to know Erlang to use Elixir & Phoenix...?

No. You can start learning/using Elixir today and call Erlang functions when required,
but you don't need to know Erlang before you can use Phoenix!

But Phoenix is not "Mainstream" ... Should I/we use it...?

There are many web application frameworks you/we can choose from: https://en.wikipedia.org/wiki/Comparison_of_web_frameworks
So why would anyone select a framework written in a programming language that is not "mainstream"...?

Why are we not using Hapi.js anymore...?

This is missinformation. We are still using Hapi.js for a number of projects where it is appropriate.
This includes several client projects and internal dwyl apps/tools.

We decided to use Phoenix for our new projects for these simple reasons:

• Elixir is a nicer language than JavaScript. #LessIsMore #LessButBetter #SmallIsBeautiful #SyntaxMatters
• JS can be functional, whereas Elixir is (always) Functional! The distinction makes all the difference!
  With "functional" programming, the programs are a lot easier to think about while you are writing/maintaining them!
• Elixir uses the Erlang VM which is way more efficient/powerful than "V8"
• The Erlang VM scales much easier to multi-core multi-server multi-data-center than Node.js
  (or pretty much anything else for that matter!!)

For our new projects we need multi-data-center fault-tolerance!
we get that "for free" by using Erlang -> Elixir -> Phoenix!!

In our opinion Hapi.js is still "the best" Node.js framework and we will continue to use and recommend it
to people that need simple apps that scale and are easy to maintain.
see: https://github.com/dwyl/learn-hapi

Also we still use JavaScript for all our AWS Lambda Micro-Services, that is not going to change.
They are simple, efficient and scale really well!
see: https://github.com/dwyl/learn-aws-lambda

What's "Wrong" with using Rails or Django?

The original "productive" web frameworks were "Ruby-on-Rails" and "Django" (python) back in 2005!
(We used both of these for periods in our "journey" and can speak on the advantages of each of them!)

There is "nothing wrong" with using Rails or Django.
We think there are still plenty of use-cases for both frameworks.
We just know that it's (a lot) easier to build "real-time"
with Phoenix because "Channels" (WebSockets) are baked in,
and the Elixir/Erlang concurrency is a whole different ballgame!
Erlang (and thus Phoenix) can handle millions of concurrent users on a single server,
whereas a Rails/Django server can only handle a few thousand (at best!)
if your app is only serving a few thousand people at once, then you're fine!!

We love the fact that Erlang uses "lighweight long-lived" processes,
which means we can connect millions of (IoT) devices ... For IoT Erlang is (unquestionably) the Answer!
For simpler web apps where you only expect a few users per day, Rails/Django are still viable.

But why compromise if you don't have to?
If you can have a Tesla for the "price" of a Ford Focus, why wouldn't you?!?
Why settle for good when you can easily have/use the best?

But GitHub Still Uses Rails ... Surely GitHub is "Scalable"?

Yes, GitHub is still using Rails for their Web App/Site.
But ask any of the core team at GitHub if (given the chance to start over) they would chose Rails
to build GitHub in 2017, and see how many of them say "yes, of course" (with a straight face...)!

Also, GitHub does a lot of things to Scale Rails in the background.
And many of their newer (client-side) features are written in JavaScript! see: https://github.com/github

Bottom line is: anything can be made to scale using "DevOps",
but Phoenix is made to scale by default because Erlang (was) invented (to) scale!

Why NOT Use Go?

"There are two kinds of programming languages - those that nobody uses and those that everybody's bitching about" ~ Bjarne Stroustrup (creator of C++)

Go is very popular. Largely due to the fact that Google "sponsors" it.
It was meant to simplify (replace) C++ and Java inside Google ...
and for the most part, it has succeeded!

We really like Go. It was our "number two" choice when deciding what programming language
(after Elixir) in our "post JS stack"... The decision to use elixir instead of anything else was easy:

• Elixir is functional (much simpler to read/write/maintain/extend)
• Elixir functions compose in a really elegant way as "plugs"/"pipelines"
• Data is immutable which makes application state predictable (no guessing!)
• Types are dynamic and inferred (no need to manually declare them) but there are rules which simplify things and the compiler checks them giving an appropriate warning.
• Erlang makes distributed apps much easier and our plan is to build IoT systems that will mean connecting many (millions) devices with persistent connections, Erlang was made for this!
• Phoenix includes many useful things out-of-the box including several security features most people won't even think of.

Further Reading:

• Why Go is Not Good: http://yager.io/programming/go.html
• Go Web Frameworks: https://medium.com/code-zen/why-i-don-t-use-go-web-frameworks-1087e1facfa4
• Why Everyone Hates Go: https://npf.io/2014/10/why-everyone-hates-go/ (click-bait title, some valid points...)
• Discussion: https://www.reddit.com/r/elixir/comments/3c8yfz/how_does_go_compare_to_elixir
  (help wanted expanding this answer...)

Why NOT Use Scala + Play Framework instead...?

If you are already used to writing Java or deploying to the JVM, the Play Framework is a superb choice: https://www.playframework.com
Scala is a good programming language and everyone should learn it! https://www.scala-lang.org/what-is-scala.html

We have used Play a few times (before we adopted Node.js) and really liked it!
But Scala is a "kitchen sink" (multi-paradigm) programming language that allows people to use "all of Java" ...
We think Scala's "Java Interoperability" means it's "too easy" to allow complexity into your codebase. Specifically "OOP Mutation" ...

So why aren't we (DWYL) using "Play" for new projects any more? Firstly, we transitioned to Hapi.js (Node.js) a few years ago because it was more "lightweight" and allowed us to write tiny programs that used only a few Megs of RAM (where our Play apps were very resource-hungry..! have you ever tried running a Scala app on a "basic" laptop like a Chromebook...?)

Summary of "reasons" for Phoenix instead of Play:

• We maintain that Play/Scala still has plenty of valid use-cases.
• Elixir is way simpler than Scala.
  
• The Erlang VM is way better at concurrency than the JVM.
• We love the fact that all data is immutable in Elixir.
• We love how few resources it takes to run a Phoenix app (e.g on a Raspberry Pi!!)

We just think that for our project(s) Erlang's Concurrency model works better and Functional code transforming Immutable data is easier to test and maintain.

If you have evidence that "Scala is Simpler" we would love to hear from you!
Tell us: https://github.com/dwyl/learn-phoenix-web-development/issues

Why not use Haskell?

If you like Functional Programming (FP) so much, why not use Haskell?

Please ask more questions: https://github.com/dwyl/learn-phoenix-framework/issues