• Build complex system by analysis
  
  
• Specifications -> Implementation
  
  
• Build stable and robust System

This was systems design before the Web

like all living networks

• Simple Architecture : URL / HTTP
  
  
• Usage -> Standardization
  
  
• Non stop evolution / endless possibilites
  

So let your project be Web based

One job and some paradigms like Social / Real-Time

Goal : Create Incoming and Outgoing Links

Need : an URI Mapping

That's it

• Structure
  You don't know in which way your project will evolve
  So keep things small and efficient
  
  
• Data storage / persistency
  Iter from more flexible to less
  Memory -> JSON -> NoSQL -> SQL only if needed
  
  
• Hosting
  Be able to move quickly from hosting solutions to another, use PaaS

• What are the ressources of your website ?

  Users = [{username: "Evangenieur"}]

  
  
• What the URI mapping ?

  http://myproject.com/evangenieur

  
  
• How can I serve it ?

  
  @get "/:user": ->
    @send Users.filter (user) =>
      user.username.toLowerCase() is @params.user
              

## Data
Users = [ {
  uri_id: "evangenieur"
  username: "Evangenieur"
  channels: [
    { 
      uri_id: "my_channel"
      name: "My Channel"
      messages: []
    } ] 
  } 
]
## Data Traversal
get_user = (user_uri_id) -> 
  if user = Users.filter((user) -> user.uri_id is user_uri_id).pop()
    user.get_channel = (channel_uri_id) ->
      user.channels.filter((channel) => 
          channel_uri_id is channel.uri_id
        ).pop()
    user
## API
require("zappajs") ->
  @get "/:user/:channel": ->
    chan = get_user(@params.user)?.get_channel(@params.channel)
    if chan and @query.message?
        chan.messages.push @query.message
    @send chan || 404

  @get "/:user": ->
    @send get_user(@params.user) || 404


      

Powered by node.js / zappajs

Goal : 1st Online web service version in 12h to 24h

Dev "stable" iteration every 2 hours

Merge/Deploy iteration every 4 hours past 12h

Do not reinvent the wheel

• Authentification using facebook / twitter / google
  
  
• Data is everywhere build mashups
  
  
• Be resilient use many data sources