Skip to content

Latest commit

 

History

History
232 lines (179 loc) · 3.83 KB

main.md

File metadata and controls

232 lines (179 loc) · 3.83 KB
Raw

What keeps you up at night? 😣

github.com / BrestJS / 2018-codelunch by shimaore

introduction

Promises

The Quest of The Simplest Example

async function() {
  await sleep(10*secondes)
  console.log('10s plus tard')
}

const sleep = (timeout) =>
  new Promise( (resolve) =>
    setTimeout(resolve,timeout)
  )

process.nextTick ≅ (cb) => cb()
// cb is added to "next tick queue"
// wait for event loop completion
// call callbacks in next tick queue

async function () {
  var a = await do_this();
  await nextTick();
  await do_that(a);
}

const nextTick = (...args) =>
  new Promise( (resolve) =>
    process.nextTick(resolve,...args)
  )

const nextTick = () =>
  new Promise( (resolve) =>
    process.nextTick(resolve)
  )

Promises

A Correction

async function (ev) {
  await event(ev,'noël')
  console.log('Enfin!')
  livrer_les_cadeaux()
}
…
ev.emit('noël')

const event = (ev,name) =>
  new Promise( (resolve,reject) => {
    ev.once(name,resolve)
    ev.once('error',reject)
  })

Did You Spot The Memory Leak?

const event = (ev,name) =>
  new Promise( (resolve,reject) => {
    function cleanup() {
      ev.off(name, on_event)
      ev.off('error', on_error)
    }
    const on_event = (...args) => { cleanup(); resolve(...args) }
    const on_error = (error)   => { cleanup(); reject(error)    }
    ev.once('error',on_error)
    ev.once(name,on_event)
  })

And Now For Something

Completely Different

issue #1

build daily reports from an unbounded data stream

Currently

  • Redis Pub/Sub
  • EventEmitter2
  • build aggregates in Redis
  • generate report "manually"

issue #2

distribute realtime events at scale

Node.js streams

npm install -g stream-adventure
&& stream-adventure

src.pipe(dst)

a.pipe(b).pipe(c).pipe(d) ≅ a | b | c | d

Node.js "object streams" objectMode:true

data-in-space vs data-in-time

Single return value Multiple return values
Pull/Sync/Interactive Object Iterables (Array, Set, Map, Object)
Push/Async/Reactive Promise Observable

Promise

one time one sender many recipients

EventEmitter

many times many senders many recipients

Observable (DataStream)

  • objects that inform other objects of the changes they publish
  • Fluent/composable: .map, .concat, .filter, .reduce
  • Observable of Observable: .flatMap, .zip, .take

Discrete vs Continuous Bacon: EventStream / Property Reactive(haskell): Events / Behaviors

A.map(…).filter(…)

A.merge(B)
  .onValue(…)
  .onError(…)
  .onEnd(…)

RxJs, BaconJS, Kefir.JS TC39 Proposal

“Promises are Observables”

Streaming events

inter-machine streaming

Pub/Sub (Redis, RabbitMQ, MQTT, …)

AWS Streaming Data Storm Kafka

de·cen·tra·li·ze!

CERN: 0MQ

Axon github:tj/axon

Issues

  • discovery ("who wants my data?")
  • registration ("how do I connect?")

(Similar to registration & message routing in SIP!)

Côte → Axon + discoverer (but: security?)

Dat → "hyperdiscovery"

Current train of thought

  • lightweight P2P (e.g. Axon, NodeRED's) between my applications
  • at the boundary: "translate" to/from 0MQ, Sock.js, Socket.io, AMQP, SMTP, …

Manipulating

& Composing

Streams

Node-RED nodered.org

Node-RED is very cool, except:

  • difficult to maintain as "code"
  • not "production ready" yet

Amazon Lambda (Step Functions) Apache Flink

Aggregating

is hard

sum, avg, count, min, max, stddev … easy

median deciles / percentiles … harder

(hint: "Bloom filters")

⇒ transducers/transformers

❦ Merci! ❧

github.com / BrestJS / 2018-codelunch github.com / shimaore