krisleech

#!/usr/bin/env bash

# manual steps
# Download raw output of master cuke to master-cuke-ci.raw as plain text
# Download raw output of another cuke branch to branch-cuke-js.raw as plain text

set -e

MASTER_RAW_FILE=master-cukes-ci.raw
BRANCH_RAW_FILE=branch-cukes-ci.raw

krisleech

#!/usr/bin/env bash

HOST=localhost:3000

webkit2png -F -o page1 "http://${HOST}/..."
webkit2png -F -o page2 "http://${HOST}/..."
webkit2png -F -o page3 "http://${HOST}/..."

krisleech

<div id="results">
  <div class="result" data-result-id='1'>
    <div data-attr="title">Result 1</div>
  </div>
  
  <div class="result" data-result-id='2'>
    <div data-attr="title">Result 2</div>
  </div>
  
   <div class="result" data-result-id='3'>

krisleech

class Double < Hash
  def initialize(attrs)
    super
    attrs.each { |k,v| store(k,v) }     
  end

  def method_missing(_name, *args, &block)
    has_key?(_name) ? fetch(_name) : super(_name, *args, &block)
  end
end

krisleech

#!/usr/bin/env bash

# quickly find a branch, using the team name as the first argument and autocomplete
# to narrow it down further.

# requires "selecta" (garybernhardt/selecta) installable via brew on MacOS

set -e

if [[ -z "$1" ]]; then

krisleech

## App

require 'bundler'

Bundler.require

### Entity

class Person
  include Virtus.model

krisleech

# Thoughts on Event Sourcing in Ruby. 
# Idea is to have commands which record events in an event store.
# A denormalizer listens to the EventStore for new events and creates highly denormalized record(s) in a relational database.
# Entities are then populated by querying the relational database, since it it totally denormalized no JOIN's are required.
#
# The relational tables serve the same purpose as relation views, the event data might be persisted to multiple tables.
# For example a command might create an event from which the denormalizer will create record in tables which are used for
# persistance of entities and for aggregate reporting.

# The relational data can be recreated using the event store.

krisleech

• Each service represents a bounded context, it is the authority for a business concept and contains the business rules.
• The service has business behaviour and data, exposing one or more resources.
• Services which talk to other services are coupled, instead services should have all the data they need, even if it duplicates of what is contained in other services.
• They must have autonomy, the network and service availability is not reliable.
• Trying to treat services like data sources means using distributed transactions and locks, which are, putting it mildly, difficult.
• Use an event bus to communicate between services:
  • decoupled, no need to know about other services
  • events are async so can be retried
  • events are writtern in past tense
• When a canonical resource has been updated an event is published. The event includes the resource, or an id, or a URL from which it can be retrived. Subscribed services can then update their copy of the data.

krisleech

# initializer
$event_store = EventStore.new
event_store.subscribe(StudyDenormalizer.new, prefix: ‘on’)

# controller

# push a new event (write database)
uuid = UUID.new
event_store.push(aggregate_id: uuid, event: ‘StudyCreated’, data: { … })

krisleech

ssh-rsa AAAAB3NzaC1yc2EAAAABIwAAAQEAq/9WdnAcWFW720c/kK2Kyo21SD0z2U0xm11cLKdPofxwKvb97H0m+URLAZhsbJ232R7sfQuUfgGSM6Spg+iv5qyo2nEY3ksyGYn5xMfkS1tIkmP+Q69XoEmNjBh3Pon0X0AITi9QGbRdQ/pYoYpNTcjePPC/1dOqnl/HWNuUyx/DglnFnUoedQAufTLsnTV3eySc9tYFUEjYUGrX+EDcAD+Fg2VmHLhGItRe4lnCy2MSbDFeGI8N7te4ZnwuXp6fHIPSDn74UdDRWAdcrtu99ZEvieMA1ONf3p21w/T2QdiWBUokXv6kmkq8BMucmGXxtl7tZt3QMYoL8D8PxNcM7Q== kris