Upgrade to PRO for Only $50/Year—Limited-Time Offer! 🔥
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Thames Valley Meetup: Refactoring
Search
Jack Franklin
November 05, 2014
Technology
1
190
Thames Valley Meetup: Refactoring
Jack Franklin
November 05, 2014
Tweet
Share
More Decks by Jack Franklin
See All by Jack Franklin
Advanced React Meetup: Testing JavaScript
jackfranklin
1
240
Components on the Web: Frontend NE
jackfranklin
1
820
ReactiveConf: Lessons Migrating Complex Software
jackfranklin
0
490
Front Trends: Migrating complex software
jackfranklin
1
820
Migrating from Angular to React: Manc React
jackfranklin
1
190
Half Stack Fest: Webpack
jackfranklin
4
560
FullStackFest: Elm for JS Developers
jackfranklin
1
250
Codelicious: Intro to ES2015
jackfranklin
0
390
PolyConf: Elm for JS Developers
jackfranklin
0
280
Other Decks in Technology
See All in Technology
AIエージェント開発と活用を加速するワークフロー自動生成への挑戦
shibuiwilliam
4
830
子育てで想像してなかった「見えないダメージ」 / Unforeseen "hidden burdens" of raising children.
pauli
2
320
「もしもデータ基盤開発で『強くてニューゲーム』ができたなら今の僕はどんなデータ基盤を作っただろう」
aeonpeople
0
230
Bedrock AgentCore Evaluationsで学ぶLLM as a judge入門
shichijoyuhi
2
240
まだ間に合う! Agentic AI on AWSの現在地をやさしく一挙おさらい
minorun365
17
2.6k
20251218_AIを活用した開発生産性向上の全社的な取り組みの進め方について / How to proceed with company-wide initiatives to improve development productivity using AI
yayoi_dd
0
650
NIKKEI Tech Talk #41: セキュア・バイ・デザインからクラウド管理を考える
sekido
PRO
0
200
Oracle Database@AWS:サービス概要のご紹介
oracle4engineer
PRO
1
400
ペアーズにおけるAIエージェント 基盤とText to SQLツールの紹介
hisamouna
2
1.6k
MySQLとPostgreSQLのコレーション / Collation of MySQL and PostgreSQL
tmtms
1
1.2k
モダンデータスタックの理想と現実の間で~1.3億人Vポイントデータ基盤の現在地とこれから~
taromatsui_cccmkhd
2
260
M&Aで拡大し続けるGENDAのデータ活用を促すためのDatabricks権限管理 / AEON TECH HUB #22
genda
0
230
Featured
See All Featured
How GitHub (no longer) Works
holman
316
140k
Design in an AI World
tapps
0
99
The B2B funnel & how to create a winning content strategy
katarinadahlin
PRO
0
190
Visualization
eitanlees
150
16k
Site-Speed That Sticks
csswizardry
13
1k
Unsuck your backbone
ammeep
671
58k
Pawsitive SEO: Lessons from My Dog (and Many Mistakes) on Thriving as a Consultant in the Age of AI
davidcarrasco
0
37
Designing for humans not robots
tammielis
254
26k
Marketing to machines
jonoalderson
1
4.3k
We Are The Robots
honzajavorek
0
120
Agile Leadership in an Agile Organization
kimpetersen
PRO
0
55
Raft: Consensus for Rubyists
vanstee
141
7.3k
Transcript
Refactoring Reafctoirng
@Jack_Franklin
None
changing the design not the behaviour
None
None
None
beware, opinions!
"Any fool can write code for a computer to understand.
Good programmers write code that humans can understand" ! Martin Fowler
None
// create the carousel ! carousel( 400, 500, $('img'), 1000,
2000, true );
function carousel( height, width, images, speed, delay, autoPlay ) {
// code }
// create the carousel ! carousel({ height: 400, width: 500,
… });
var h = 400; var w = 400; var play
= true; var calc = function()… ! for (var key in things)
Name things after their intention
"There are only two hard things in Comp Sci, cache
invalidation and naming things" ! Phil Karlton
var placePin = function(x, y) ! var getLatLon = function(x,
y) ! var user = { coordinates: [x, y] }
var placePin = function(coords) ! var getLatLon = function(coords) !
var user = { coordinates: { x: 1, y: 2 } }
implicit knowledge
if I were to hand the code over to you,
what do I have to explain?
var drawGraph = function(width, height) { width = 160 +
width; height = 172.5 + height; }
None
! var drawGraph = function(width, height) { var graphWidthPad =
160; width = graphWidthPad + width; … }
implicit knowledge is what trips future you up in 6
months time
function someFunc() { doSomething() and.then.something.else(); maybe.even.more(); var x = 2;
var y = 3; keep.on.going(x); and.going.and.going(y); return on.and.on(); }
strive for reusable, composable functions
this makes sure they do one thing and one thing
well
and also makes them much easier to test
easy to test code is usually pretty good
var goToBeginning = function(carousel) { if(carousel.isAtEnd()) { carousel.goToStart(); } }
var goToBeginning = function(carousel) { if(carousel.isAtEnd()) { carousel.goToStart(); } }
carousel goToBeginning
carousel.goToBeginning = function() { if(this.isAtEnd()) { this.goToStart(); } }
Component Component Component Component Component Component Component Component Component Component
components should know little about each other
one thing well
doing this in real life
120 seconds
you will never get this right
you never know less about the problem
premature abstraction is the root of all evil
prefer duplication at first
/users ?created_at[gt]=2014-04-01 &created_at[lte]=2014-05-01 ! if params[:created_at][:gt] users = users.where("created_at >…")
if params[:created_at][:lte] …
if params[:created_at][:gt] users = users.where("created_at >…") ! if params[:created_at][:lte] …
! if params[:created_at][:gte] … ! if params[:created_at][:lt] …
None
filters = params[:created_at] ! map = { lte: '<=', gt:
'>', … } ! filters.reduce(User.all) do |col, (key, val)| sym = map[key] col.where("created_at #{sym} ?", …) end
None
you can be too clever for your own good
filters = params[:created_at] users = User.all ! filters.reduce(users) do |coll,
(key, val)| case key when :lte then coll.where(…) when :gt then coll.where(…) … end
None
prefer clarity of intent over succinct code
if something goes wrong, back out
git commit all the time
you should rebase before pushing anyway
it's a slow, methodical, mechanical process
test driving features
Can a user subscribe? • NO if they are the
owner of the blog • NO if they are an admin of the blog • NO if they are already subscribed to the blog • NO if the blog is private • Else, totally.
given input X I expect output Y
pure function! input X > same output Y no side
effects
you should strive for pure functions
they are easy to test and less prone to causing
large errors
Can a user subscribe? • NO if they are the
owner of the blog • NO if they are an admin of the blog • NO if they are already subscribed to the blog • NO if the blog is private • Else, totally.
it "returns true if the user can" do res =
UserSubscribe.can_subscribe? (user_id, blog) expect(res).to be(true) end
class UserSubscribe def self.can_subscribe? (user_id, blog) true end end
None
it "returns true if the user can" … ! it
"returns false if the blog is private" …
None
class UserSubscribe def self.can_subscribe? (user_id, blog) !blog.private? end end
None
it "returns true if the user can" … ! it
"returns false if the blog is private" … ! it "returns false if the user owns the blog" …
None
class UserSubscribe def self.can_subscribe?(user_id, blog) if blog.private? false elsif user.owns?(blog)
false else true end end end
None
it "returns true if the user can" … ! it
"returns false if the blog is private" … ! it "returns false if the user owns the blog" … ! it "returns false if the user is admin" …
class UserSubscribe def self.can_subscribe?(user_id, blog) if blog.private? false elsif user.owns?(blog)
false elsif user.is_admin?(blog) false else true end end end
it "returns true if the user can" … ! it
"returns false if the blog is private" … ! it "returns false if the user owns the blog" … ! it "returns false if the user is admin" … ! it "returns false if the user is subscribed" …
class UserSubscribe def self.can_subscribe?(user_id, blog) if blog.private? false elsif user.owns?(blog)
false elsif user.is_admin?(blog) false elsif user.is_subscribed?(blog) false else true end end end
None
the first implementation doesn't matter
the first implementation is about understanding the problem
None
class UserSubscribe def self.can_subscribe?(user_id, blog) if blog.private? || user.owns?(blog) ||
user.is_admin?(blog) || user.is_subscribed?(blog) false else true end end end
class UserSubscribe def self.can_subscribe?(user_id, blog) ! !(blog.private? || user.owns?(blog) ||
user.is_admin?(blog) || user.is_subscribed?(blog)) ! end end
class UserSubscribe def self.can_subscribe?(user_id, blog) ! return false if blog.private?
return false if user.owns?(blog) return false if … return false if … true end end
None
prefer clarity of intent over succinct code
TDD doesn't work every time
(but that doesn't mean you shouldn't write tests!)
but sometimes it might make sense to write the tests
afterwards...
#TODOs don't get done
None
49,732,824
Code Review
Code Review
None
Making a difficult change
refactor to make the change easy
make the change
RubyRogues #178 ! http://devchat.tv/ruby- rogues/episode-guide
Refactoring Book ! http://refactoring.com/
Thoughtbot Blog ! http:// robots.thoughtbot.com/
Thank You! ! @Jack_Franklin ! http://javascriptplayground.com/ the-refactoring-tales/